v>EPA
United States Industrial Environmental Research EPA-600/7-79-038
Environmental Protection Laboratory Fetjruary 1979
Agency Research Triangle Park NC 27711
Proceedings: Third
Workshop on Catalytic
Combustion
(Asheville, NC, October 1978)
Interagency
Energy/Environment
R&D Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of, control technologies for energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.
EPA REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved
for publication. Approval does not signify that the contents necessarily reflect
the views and policies of the Government, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/7-79-038
February 1979
Proceedings: Third Workshop
on Catalytic Combustion
(Asheville, NC, October 1978)
John P. Kesselring, Compiler
Acurex Corporation/Aerotherra Division
485 Clyde Avenue
Mountain View, California 94042
Contract No. 68-02-2611
Task No. 30
Program Element No. EHE624A
EPA Project Officer: G. Blair Martin
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, NC 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
PREFACE
This proceedings document covers the major presentations of the Third
Workshop on Catalytic Combustion held October 3-4, 1978, at the Grove Park
Inn in Asheville, North Carolina. Sponsored by the Combustion Research Branch
of the EPA's Industrial Environmental Research Laboratory — Research Triangle
Park, the workshop served as a forum for the presentation of results of recent
research in the areas of catalyst and catalytic combustion system development.
The first Workshop on Catalytic Combustion was held May 25-26, 1976, at the
Plantation Inn in Raleigh, North Carolina, and the Second Workshop was held
June 21-22, 1977, also at the Plantation Inn. Proceedings documents were not
produced for the first and second workshops, but written Summaries were dis-
tributed to all attendees. Copies of these Summaries are included in this
document as Appendix A. Appendix B lists all attendees of che Third Work-
shop on Catalytic Combustion.
Dr. John P. Kesselring, Acurex Corporation, acted as Workshop Coordi-
nator, and G. Blair Martin, Combustion Research Branch, Environmental Protection
Agency, was the Project Officer.
11
-------
TABLE OF CONTENTS
"Theoretical Analysis of Temperature and Composition in a
Catalytic Monolith Reactor," C. M. Ablow, H. Wise ....
Page
"Progress of Reaction in a Honeycomb Catalyst: CO/Air
Combustion," P. M« Walsh, D. A. Santavicca, B. Kim, F. V. Bracco . . 29
"Application of Catalytic Flame Stabilization for Aircraft
Afterburners," L. C. Angello, T. J. Rosfjord 61
"EPRI View of Catalytic Combustion," A. C. Dolbec 83
"Overview of Criepi Catalytic Combustion Research Program,"
Y. Ishihara, H. Fukuzawa 95
"Catalytic Combustion of No. 6 Fuel Oil," J. Pogson, M. N. Mansour. . HI
"Structural Analysis of a Preliminary Catalytic Combustion Ceramic
Design," S. M. DeCorso, D. E. Carl 139
"Determination of the Rate of Heterogeneous Reaction on Catalytic
Surfaces," P. J. Marteney 157
"Prototype Surface Combustion — Furnace Evaluation," G. B. Martin . 191
"An Analysis of Catalytic Combustion in Monolithic Honeycomb Beds,"
R. M. Kendall, J. T. Kelly, E. K. Chu, J. P. Kesselring 215
"The Development of Catalytic Combustors for Stationary Source
Applications," W. V. Krill, J. P. Kesselring 259
"Fuel NOX Control by Catalytic Combustion," E. K. Chu, J. P.
Kesselring 291
"Prospects for High-Temperature Catalysts," W. C. Pfefferle 331
"Environmental Aspects of Low Btu Gas-Fired Catalytic Combustion,"
B. A. F'olsom, C. Courtney, M. P. Heap 345
"The Advanced Low-Emissions Catalytic-Combustor Program; Phase I -
Description and Status," A. J. Szaniszlo 385
ill
-------
"Catalyst-Coated Flameholder for Lean Stability Improvement,"
E. J. Szetela, J. B. Me Vey 389
"Effect of Inlet Temperature on the Performance of a Catalytic
Reactor," D. N. Anderson ......... 403
"Performance of Multiple-Venturi Fuel Preparation System,"
R. R. Tacina „.....> 427
"Correlations of Catalytic Combustion Performance Parameters,"
D. L. Bulzan 431
"Catalyst Design Studies in Low BTU Gas Combustion," R. Carrubba,
I. T. Osgerby 435
"Catalyzed Combustion of H9/Air Mixtures in a Flat Plate Boundary
Layer," R. W. Schefer, F. S. Robben, R. K. Cheng 437
"Application of Rich Catalytic Combustion to Aircraft Engines,"
G. E. Voecks, D. J. Cerini 477
"Catalytic Combustion in Actual Engines; a Summary of Engine
and Rig Tests," G. E. Enga 491
"High Combustion Efficiency and Low Pollutant Emission by
Catalytic and Other Jfleans," K. C. Salooja 513
Appendix A - Summaries of First and Second Workshops
on Catalytic Combustion . A-l
Appendix B - List of Attendees B-l
IV
-------
THIRD WORKSHOP ON CATALYTIC COMBUSTION
G. Blair Martin, Chairman
John P. Kesselring, Workshop Coordinator
-------
THEORETICAL ANALYSIS OF TEMPERATURE AND COMPOSITION
IN A CATALYTIC MONOLITH REACTOR
By
C. M. Ablow and H. Wise
SRI International
Menlo Park, California 94025
-------
ABSTRACT
A theoretical model of catalytic combustion is developed which
allows calculation of temperature and reactant/product distribution in
a tubular duct with catalytic walls. Under adiabatic conditions, as
prevail in the central ducts of a catalytic monolith combustor, and in
the absence of heat conduction along the reactor walls, the model provides
an analytic solution. It exhibits the existence of multiple steady states
and, for gas mixtures with low Lewis number, temperature excursion in excess
of the adiabatic reaction temperature. Gas-phase reactions are shown to
increase the fuel consumption in a given length of catalytic duct with
corresponding changes in temperature distribution. The model is applied
to a series of experimental results obtained with different fuels in
tubular reactors. Satisfactory agreement is found between theoretical
and experimental data when account is taken of the contribution of heat
loss from the reactor to the environment. The theoretical analysis may
be employed in optimizing the engineering design of monolith reactors
performing over a range of operating conditions.
The research reported here was supported by the U.S. Air Force Office
of Scientific Research under contract number F49620-77-C-0123.
-------
Nomenclature
C specific heat (cal/g-deg)
2
D diffusion coefficient (cm/sec )
F Frank-Kamenetskii factor, Eq. (26)
h mass transfer coefficient (cm/sec)
M ^
2
heat transfer coefficient (cal/cm -sec-deg)
k nondimensional reaction rate constant
k gas phase reaction rate constant (sec )
G
k catalyzed surface reaction rate constant (cm/sec)
S
Le Lewis number
2
m mass flux (g/cm sec)
Nu Nusselt number
N ratio of heat transfer coefficients
w
Pe Peclet number
Q heat released by reaction (cal/gm)
R hydraulic radius (cross sectional area divided by perimeter, cm)
H
R thickness of the duct wall (cm)
w
Sh Sherwood number
St Stanton number
T temperature (°K)
AT temperature rise of adiabatic reaction (°C)
A
v flow velocity (cm/sec)
x distance from duct entry (cm)
Y fuel mass fraction
y ratio of Y to Y at duct entry
\ thermal conductivity (cal/cm-deg-sec)
3
p density (g/cm )
p. viscosity (poise)
| nondimensional distance
T ratio of temperature to AT
A
-------
Subscripts
0 at the duct entry
E exterior of the duct
G in the gas phase
S at the solid surface of catalyst
w in the duct wall
-------
I INTRODUCTION
Catalytic reactors in the form of ceramic monoliths coated with
catalytically active metals have been used for the conversion of low
concentrations of pollutants to oxidation products. The development
of catalytic reactors for the combustion of fuel-air mixtures in thruster
applications is receiving attention because the heterogeneous reaction
can be initiated over a wide range of fuel-air ratios and can be carried
to completion at temperatures sufficiently low to prevent the formation
of such pollutants as nitric oxide. During the operation of such
combustors, the catalytic monolith effectiveness is generally mass
2
transfer limited, i.e. the reactant species have insufficient time
to diffuse to the wall and react before leaving the reactor. Under
these conditions it is desirable to increase the degree of reactant
conversion by homogeneous, gas-phase reactions promoted by the heat
and intermediate chemical species released in a catalyzed reaction.
3 4
Theoretical studies ' of monolith combustion have involved com-
putational investigations of mathematical models of reaction and flow
including temperature dependent transport and kinetic parameters. In
the present work catalyzed combustion is examined by means of a simplified
model.
The analysis includes heterogeneous reaction at the tube wall and
homogeneous, gas phase reaction. For the low fuel/air ratios under
consideration the reaction rates are taken to be of first order in fuel,
and of zero order in oxygen concentration. The reaction kinetics involve
global reaction rates of the Arrhenius type obtained from separate kinetic
studies.
-------
The model includes heat transfer by convection in the gas, by
diffusion from the tube wall to the gas, by conduction to the surroundings,
and by conduction along the tube wall. Conduction to the surroundings
is negligible for ducts in the interior of a monolith where uniform
5
conditions prevail. Ducts near the periphery of the monolith are subject
to heat transfer through the duct wall. Heat conduction along the tube
wall is important near the duct entry where temperature gradients are
high.
If heat conduction along the tube wall is neglected, as contrasted
to the computer modeling employed in reference 6, the model has analytic
solutions in the two cases: (2) temperature independent, diffusion
controlled, heterogeneous reaction with heat transfer to the surroundings
and (b) temperature dependent reaction without external heat transfer.
II DUCT FLOW MODEL
The steady-state temperature and concentration distributions in each
duct cross section are modelled by their average plug flow values in the
gas phase, denoted by subscript G, and their values at the catalyst
surface, subscript S. The heat balance in a cross section of the fluid
reads
d
R — (p v C T ) = -h (T - T ) + R Qk p Y
where, for the fuel-lean case, the nearly constant concentration of
oxidizer is included in the reaction rate constant k . The heat transfer
G
coefficient may be expressed in terms of either the Stanton or Nusselt
numbers:
10
-------
The fuel mass balance in the fluid
RH S V V - -hM(PGYG - PS
The mass transfer coefficient li may be written in terms of the diffusion
coefficient D of the fuel through the gas mixture and the Sherwood
number Sh:
= (D/4 RH) Sh .
The fuel that diffuses to the surface of the catalyst is consumed by
the reaction:
Finally, heat balance in the duct wall may be written as:
Vw I? + VTG - V + Q Vs YS = \ VTS - TE)
where N is the ratio of the heat transfer coefficients for exterior
w
and interior temperature differences.
Ill GOVERNING PARAMETERS
The governing parameters are found by a nondimensionalization of the
equations. The dimensionless distance and temperature, | and T, are
taken to be
5 • x St/R , T = T/AT (5)
n A
n
-------
where the adiabatic temperature rise AT is defined by
AT = Q Y /C . (6)
Since m = p v is a constant, Eq. (1) may be divided by m C St AT to
G. vi A
obtain
- V + V VG Vm CG st ATA
Introduction of the unburned fuel fraction, y, y = Y/Y _, as a
dependent variable and use of the constant pressure relation, p T
G G
constant, reduces the equation to
= -(T-T)+kyA , k = k R p T /m St AT (7)
G S G G G G G H KGO GO A
Similar manipulation of Eq. (2) results in
= -(Sh/Nu Le)
where Le, the Lewis number, is defined by Le = X/p C D. The analogy
G G
between heat and mass transfer, Sh = Nu, gives
Eq. (3) reduces by the same method to
- (PS/PG> yS = (L6) kS ysAS ' ^S = kS PGOTGO/m (St)
12
-------
Finally, Eq. (4) becomes
(10)
2 2
where Pe, the Peclet number, is defined by Pe = R h /R X (St) .
The fuel fraction y at the catalytic surface can be eliminated
S
from Eqs. (7), (8), and (10) by use of Eq. (9). There remain the three
following equations for the three variables y T , and T :
vif G S
dT = -(TG - V
dlf
= VTS - V ' F = 1/(Le + VV
Fraction F may be recognized as the factor that indicates the control of
the process either by reaction kinetics, if k is small so that the
s 7
second term dominates, or by diffusion, if k is large.
S
The differential equations are to be solved under the entry conditions:
= V
For a thin ceramic wall the Peclet number, the ratio of the heat
transfer coefficients for diffusion into the stream to that for conduction
along the wall, will be very large. In this case the term involving Pe
may be omitted from Eq. (13). This approximation lowers the order of the
differential system so that two of the conditions in Eq. (14) need to be
13
-------
dropped. The conditions on T are the ones omitted on the physical'
o
grounds that the temperature of an insulated wall should not be
prescribed. As a result of the omission of the wall heat conduction term,
the model has solutions where the wall temperature jumps discontinuously
from one value to another. The jumps are smoothed over to become intervals
8
of sharp temperature variation if the conduction term is reintroduced.
The equations of the model are then Eqs. (11) and (12) and
TG - TS + F yG ' VTS - V
TG = TGO at
IV CATALYTIC DUCT REACTOR WITHOUT HEAT TRANSFER TO THE SURROUNDINGS
A duct in the interior of the monolith is likely to be in thermal
Librium with adjacent ducts. The
the adiabatic case. Eq. (15) becomes
equilibrium with adjacent ducts. The model is then applicable with N =0,
w
TG '
so that the sum of Eqs. (11) and (12) gives
V + T =0
y G G
Integration of this differential equation gives
yG + TG * X + TGO
where the constant of integration has been fixed by the boundary conditions
in Eq. (16).
14
-------
Equations (17) and (18) relate T T and y to one another and so
G S G
determine the phase space solution curves. Since either y or T is
G G
readily eliminated from Eq. (17) by use of Eq. (18), the phase space
can be represented as a two-dimensional plane. Figure 1 presents such
a phase plane.
The solution curves for positive, non-dimensional reaction rate
constants lie between the extremes of kinetic control where k =0 and
diffusion control where k = °°. Eq. (17) shows that the two conditions
. S
correspond to the lines T = T and T = [T (Le - 1) + 1 + T ]/Le,
S G S G GO
respectively. These lines and the line of the initial condition, T = T,
G GO
form a triangle in the phase plane. Points inside the triangle represent
positive finite reaction rates and therefore are physically accessible.
The vertices of the triangle are points I, K, and D. Point I, T = T =
S G
T + i, corresponds to the end of an infinitely long duct where both gas
GO
and wall have reached the adiabatic reaction temperature. Point K represents
the duct entry temperatures in the absence of exothermic reaction. Point D
represents the duct entry temperatures under conditions of diffusion control.
A point between K and D indicates an entry condition with finite reaction
rate and therefore a wall temperature between those at K and D. At point D,
T = T + l/Le so that temperatures above the adiabatic (T + i) can be
S GO trU
attained if Le < 1. Such a case is indicated in Figure 1 by point D . The
solution curves plotted in Figure 1 are those for propylene/oxygen/nitrogen
mixtures at several inlet temperatures, as described below.
Since the unburned fuel fraction y can only decrease and, by
G
Eq. (18), T can only increase continuously, conditions producing a
G
solution curve with a maximum point cause a discontinuous wall temper-
ature distribution. For a solution curve such as IV on Figure 1, jumps
from the left branch across the middle to the right branch can occur at
any point between the local maximum and minimum on the curve. Stability
considerations suggest that only the two extreme cases can prevail.
15
-------
Four different operating conditions can be identified: One producing
a phase plane solution curve of type IV which gives rise to wall temper-
ature distributions that may jump in either of two locations; another
producing a curve of type III that can cause a jump at the duct entry
or at a location downstream (if the part of the solution curve outside
of the triangle had been drawn in, one would see that a type III curve
is just a type IV curve with its minimum point outside the triangle) ;
a condition (type II) that produces a jump at entry since both the
maximum and minimum of the complete solution curve lie outside the triangle;
and conditions of type I with no maximum or minimum. Since the jumps in
each case are from a branch of the solution curve near side K-I, to one
near D-I, the jump may be regarded as a sharp transition of the wall
reaction from kinetics to diffusion control. For conditions producing
curves of type I without a maximum or minimum, a smooth transition or no
transition at all is obtained.
Equation (11) has been integrated under the assumption of no gas phase
reaction k = 0, for three cases and the results plotted in Figure 2. A
G
comparison of curves II and III shows the effect of inlet temperature on
a given gas mixture. Conditions for curves IA and III differ mainly in
fuel concentration.
An interesting deduction from the above analysis is that the phase is
independent of the homogeneous reaction. In this adiabatic case, the
relation between temperature and concentration is independent of what
fraction of the reaction is homogeneous and what is heterogeneous. The
gas phase reaction enters the calculations when the distance coordinate
for each phase is being obtained, by use of Eq. (11). The importance
of gas phase reaction may therefore be gauged by plotting to k y A
G G G'
a function of T against (T - T ). Since (T - T ) can be read directly
u b Or S G
from Figure 1 as the distance between the solution curve and the line K-I,
a convenient function to plot is it Y—A + T as a function of TC. Where
G FG G G
16
-------
this latter graph comes near or crosses the solution curve, gas phase
reaction makes a major contribution. The homogeneous reaction curve for
case IB is plotted in Figure 1. Since quantitative kinetic data on the
gas phase oxidation of propylene of the temperatures of interest is not
available we have estimated the preexponential factor and activation energy
12 -1
for the homogeneous reaction to be 10 sec and 30 kcal/mol. One sees in
the figure that homogeneous reaction does influence the reactive flow beyond
the point where 60% of the fuel has been consumed.
The parameter plane of the inlet conditions, the temperature and
adiabatic reaction temperature of the gas mixture, has been presented in
Figure 3. The plane is divided to show the regions where conditions
giving rise to the various types of solution curves are found. In region I
the curve of T as a function of T given by Eq. (15) has no inflexion
G S
point with zero or negative slope. In the other regions the corresponding
phase plane solution curves have such inflexion points along with a local
maximum point and a minimum point. Both of these points lie outside the
triangle for parameters of region II, the maximum is in the triangle for
parameter points in region III, and both extreme points are in the triangle of
physically real points for initial conditions in parameter region IV. .
V EFFECT OF HEAT TRANSFER BETWEEN DUCT AND SURROUNDINGS
Ducts near the outer periphery of a monolith combustor lose heat to
their surroundings. This heat transfer appears in the model as a non-
vanishing parameter N in Eq. (15).
w
An analytic solution can be obtained in the diffusion-controlled
fast wall reaction limit in the absence of homogeneous reaction. The
solution reads
N Le/(l + N )
w w
T - T = Av + BV
S E yG tG
A = (1 - Le)/Le[l + N (1 - Le) ]
w
17
-------
B = -A + [1 + Le(T - T )]/Le(l + N ) (19)
0 E w
TG - TE = (TS - V (1 + Nw> * VLe (20)
y =
*
(21)
Equations (19) and (20) present the phase space solution. Elimination
of T from Eq. (20) by use of Eq. (19) gives
s
N Le/(l + N
T -T =Cy + (l+N )By
G E w
C = -1/[1 + N (1 - Le)] . (22)
w
The solution is then represented by two curves, one from Eq. (19) for
T as a function of y and one from Eq. (22) for T
S G
The general solution with finite wall reaction rate and with homo-
geneous reaction may be obtained by numerical integration of the differential
equations of the models, Eqs. (ll) , (12), (15), and (16).
VI APPLICATION
10
The model has been applied to analyze some experimental results
in which the wall and gas temperatures were measured during steady-state
catalytic combustion of fuel/oxygen/nitrogen mixtures in monoliths with
a small number of ducts with catalytic walls, heated externally to the
adiabatic reaction temperatures. Most of the experimental data for
propylene combustion were found to fall into the diffusion-controlled
reaction regime using the transport and reaction-rate parameters applicable
to this system. Consequently the analytic solution given by Eqs. (19),
(20), and (21) was employed. The theoretical curves based on the adiabatic
18
-------
model (no heat transfer) were found to be in good agreement with the
experimental data when account was taken of conductive heat loss from
the reactor to the environment (Figure 4).
With hydrogen as a fuel the reaction system has a small Lewis number
(Le < 1). Under these conditions the theoretical analysis predicts wall
temperatures that exceed the adiabatic reaction temperature (T ). Indeed
A
the experimental data (Figure 5) exhibit a temperature maximum, well in
excess of T . Theoretically, a wall temperature maximum at the reactor
/\
entrance is calculated with a monotonic decline along the reactor tube.
Again the discrepancy in the temperature profile near the entrance is
ascribed to heat conduction along the duct wall.
It is of interest that in calculating the temperature distribution
the available kinetic data for catalytic oxidation of hydrogen have a
higher degree of uncertainty than for the case of propylene oxidation.
4
In our analysis we increased the Arrhenius preexponential factor by 10
over that published in reference 12, to take account of the higher degree
of dispersion of Pt on the monolith-washcoat support. With the reaction
and transport parameters listed in Table I, the experimental data obtained
for this system were found to be in the diffusion-controlled reaction
region.
VII CONCLUSIONS
A theoretical model of reactive flow through a duct with catalytic
walls has been developed that by means of an analytical solution provides
estimates of temperature and fuel product distributions in catalytic
combustion In the absence of heat transfer to the environment. The
model provides a direct way of assessing the importance of the contribute
of the gas phase reaction, and a prediction of the length of duct require*
to reach a specific temperature and fractional fuel conversion.
19
-------
The phase space derived from the model relates the concentrations
and temperatures at the wall and in the stream that can occur together
in each duct cross section. These relations are found to be independent
of the homogeneous reaction and of the flow rate. However the distance
into the duct where a particular phase is reached depends on the flow
rate and degree of homogeneous reaction.
In the adiabatic case, valid for ducts in the interior of a monolith,
the analytic solution is found to be confined to a certain triangle in
the wall temperature-gas temperature phase plane. The vertex of the
triangle is the point where wall and gas reach the adiabatic reaction
temperature. The base of the triangle is on the line where the gas
temperature has its inlet value and extends from the point of equal wall
and gas temperatures to a wall-gas temperature difference inversely
proportional to the Lewis number. Thus high wall temperatures are to be
expected at the duct entry for gas mixtures with low Lewis number. The
surprising fact that temperatures in excess of the adiabatic reaction
temperature can be attained is due to this result.
For an adiabatic reacting system with known initial concentration
of fuel, oxidizer, and inert diluent, the inlet parameters are the gas
inlet temperature and the temperature rise for adiabatic reaction.
Different forms of temperature variation along the duct are found for
different parameter ranges. Solutions with Jump discontinuities down-
stream of the duct entry are obtained from parameters represented by
points in a certain quadrant of the parameter plane (Figure 3). A cold
gas mixture with relatively high fuel content (high adiabatic reaction
temperature) introduced into the reactor, will favor temperature jumps and
multiple steady states. The model also suggests that the contribution of
homogeneous relative to heterogeneous reaction will increase with larger
monolith tube diameter.
20
-------
REFERENCES
1. W. C. Pfefferle, R. W. Carrubba, R. M. Heck, and G. W. Roberts,
"Catathermal Combustion, A New Process for Low-Emissions Fuel
Conversion," ASME paper 75-VA/Fu-l.
2. T. J. Rosfjord, "Catalytic Combustors for Gas Turbine Engines,"
AIAA paper 76-46.
3. A. E. Cerkanowicz, R. B. Cole, and J. G. Stevens, "Catalytic
Combustion Modeling; Comparisons with Experimental Data,"
ASME paper 77-GT-85.
4. J. T. Kelly, R. M. Kendall, E. Chu, and J. P. Kesselring,
"Development and Application of the PROF-HET Catalytic Combustion
Code," paper 77-33, Western States Section/The Combustion
Institute, Stanford, October 1977.
5. S. M. DeCorso, S. Mumford, R. V. Carrubba> and R. Heck, "Catalysts
for Gas Turbine Combustors - Experimental Test Results," J. Eng.
for Power, Vol. 99A, 159-167 (1977).
6. J. Votruba, J. Sinkule, V. Hlavacek, and J. Skrivanek, "Heat and
Mass Transfer in Monolithic Honeycomb Catalysts - I., Chem. Sng.
Sci., Vol. 30, 117-123 (1975).
7. D. A. Frank-Kamenetskii, Diffusion and Heat Transfer in Chemical
Kinetics, Plenum Press, New York (1969).
8. G. Eigenberger, "On the Dynamic Behavior of the Catalytic Fixed-Bed
Reactor in the Region of Multiple Steady States - I. The Influence
of Heat Conduction in Two-Phase Models," Chem. Eng. Sci. , Vol. 27,
1909-1916 (1972).
9. R. A. Svehla, "Estimated Viscosities and Thermal Conductivities
of Gases at High Temperatures," NASA TR R-132 (1961).
10. L. L. Hegedus, "Temperature Excursions in Catalytic Monoliths",
AIChE J., Vol. 21, 849-853 (1975).
11. S. Goldstein, Modern Developments in Fluid Dynamics, Oxford (1938).
12. R. Schefer, R. Cheng, F. Robben, and N. Brown, "Catalyzed Combustion
of H2/Air Mixtures on a Heated Platinum Plate," presented to
Western States Section/The Combustion Institute, Boulder (1978).
21
-------
13. Handbook of Chemistry and Physics (1936).
14. A. Schwartz, L. L. Holbrook, and H. Wise, "Catalytic Oxidation
Studies with Platinum and Palladium," J. Catalysis, Vol. 21,
199-207 (1971).
22
-------
Table I
NUMERICAL VALUES FOR THE CALCULATIONS
Quantity
Fuel/oxygen/nitrogen
Propylene Hydrogen Reference
0.19/5.08/94.73 1.90/4.60/93.50 10
10
10
10
10
10
10
11
9
13
9
ATA (°c)
A
T (°C)
GO
T (°C)
E
Re
Le
R (cm)
R r (cm)
w
Nu
Pr
St = Nu/Re Pr
X (cal/cm-sec-°C)
\ for N_ at 700 °K
G 2
(cal/cm-sec °C)
M. for N at 700 °K
G(poisef
Pe
Preexponential factor
A (cm/sec)
Activation energy
E (kcal/mol)
2
m = JJ-Re/4R (gm/cm sec)
n
PGOTGO g/Cm
kg at 700 °K
119.
417.
536
967.
1.554
0.0368
0.041
6.
0.75
8.27 x 10~3
io-3
1.2 x IO"4
3.2 x 10~6
19.5
8*
10
22.0
0.0210
0.351
254
150.
260
410
1175.
0.326
0.0368
0.041
6.
0.75
6.81 x 10~3
io"3
1.2 x 10~4
3.2 x IO"6
23.7
1.4
3.85
0.0255
0.3395
1.16
12
14,12
Value large enough for diffusion controlled first-order reaction,
although negative order reactions were observed.14
23
-------
ro
250 119
100 300
350 450
200 450
100 450
< 0.5 —
SA-6687-18
FIGURE 1 PHASE PLANE THEORETICAL CURVES FOR PROPYLENE/OXYGEN/NITROGEN MIXTURES;
VF = 1 - (TG - TQQ)'I ^W = ^' homogeneous reaction for Case IB.
-------
1.0
PO
en
0.5
GO
IA 250
AT. (°C)
II
350
III 200
119
450
450
FIGURE 2 THEORETICAL NON-DIMENSIONAL WALL TEMPERATURES PLOTTED AGAINST DISTANCE FROM DUCT ENTRY FOR
A FEW PROPANE/OXYGEN/NITROGEN MIXTURES FLOWING AT Re = 967 IN A DUCT WITH R., = 0.0368 CM
M
-------
1000
900 —
800 —
700 —
600 —
O
< 500 —
400 —
300 —
200 —
100 —
KINETIC CONTROL AT ENTRY
SMOOTH TS
DIFFUSIVE CONTROL AT
DUCT ENTRY
100
200
300
GO
400
500 600
SA-6687-17
FIGURE 3 THEORETICAL REACTIVE FLOW REGIMES FOR PROPYLENE/OXYGEN/NITROGEN
MIXTURES. POINT H FOR DATA REF. 10.
26
-------
550
500 —
U
450 -
400
SA-6687-13
FIGURE 4 PHYSICAL PLANE FOR C3H6/02/N2 MIXTURE. THEORETICAL CURVES WITH
k » 108 exp (-22,000/RT), N... = 0.2.
W
27
-------
640
600 —
500 —
400 —
300 —
240
FIGURE 5 PHYSICAL PLANE FOR H2/02/N2 MIXTURE. CURVES COMPUTED WITH
k = 3.2 x 103 exp (-3850/RT) N... = 0.2.
w
28
-------
PROGRESS OF REACTION IN A HONEYCOMB CATALYST:
CO/Air Combustion
By:
P. M. Walsh,"D\ A. Santavicca, B. Kim, and F. V. Bracco
Department of Mechanical and Aerospace Engineering
Princeton University
Princeton, NJ 08540
29
-------
ABSTRACT
The objective of this program is to collect detailed data within a
honeycomb catalyst for the development and testing of a two-dimensional
model, and to further current understanding of the fundamental sub-
processes which control catalytic combustion. Results of a preliminary
study are presented on the combustion of CO/air in a Pt/Al203/Cordierite
honeycomb catalyst. The test conditions were: inlet temperature,
700 ± 8 K; inlet pressure, 109 ± 4 kPa; reference velocity, 5-31 m/s;
CO/air equivalence ratio, 0.15 - 0.30; and t^O concentration,
1.1 - 3.7 mol%. Efforts were made to obtain uniform profiles of
temperature, velocity, and CO concentration across the inlet gas stream.
Temperature and concentration are uniform within ±1%, but velocity only
within ±10%. The philosophy of the program is that a detailed model of
the catalytic combustor is necessary to be able to draw consistent
conclusions from the experimental data. Since our model has not yet been
adequately developed, the following conclusions from the experiment alone
are only indicative. Unburned CO at the catalyst outlet can be independent
of equivalence ratio in a transitional regime between mass transfer control
and homogeneous reaction domination. A relative maximum or shoulder is
observed in the substrate temperature near the inlet. Substrate
temperature is insensitive to reference velocity at fixed equivalence ratio.
Estimates of the rate coefficients for CO oxidation in the catalytic
combustor exhaust showed good agreement of the activation energy with
published data for homogeneous CO oxidation, indicating that the reaction
in the exhaust is not catalyzed by long-lived species generated solely by
the catalyst.
Other tasks of this same program include: 1. Substrate and exhaust
measurements over a range of inlet temperature and wider ranges of
equivalence ratio and reference velocity during combustion of propane,
higher hydrocarbons, and coal-derived gases. 2. Measurement of conversion
of NH3 to NOx during combustion of coal-derived gases. 3. Measurement of
gas temperature and composition as functions of axial position inside a
monolithic catalyst. All of these tasks are directly related to the
application of catalytic combustion to coal-derived gases and aeronautical
fuels.
31
-------
CONTENTS
List of Figures
Symbols
Acknowledgments
Introduction
Equipment and Procedure
Data and Discussion
Current Programs
Conclusion
References
Figures
32
-------
LIST OF FIGURES
1. Inlet gas temperature vs. radial position.
2. Inlet CO concentration vs. radial position.
3. Inlet gas velocity vs. radial position.
4. Substrate temperature vs. radial position.
5. Substrate temperature, exhaust temperature, and exhaust composition
vs. axial position at different equivalence ratios.
6. Substrate temperature, exhaust temperature, and exhaust composition
vs. axial position at different reference velocities.
7. Ratio of unburned CO to inlet CO vs. reference velocity.
8. Rate coefficient for CO oxidation in the exhaust vs. temperature.
33
-------
ACKNOWLEDGMENTS
This project is supported by the Air Force Office of Scientific
Research under Grant No. AFOSR-76-3052, Dr. B. T. Wolfson, Grant Monitor;
and by the Department of Energy under Contract No. EF-77-S-01-2762,
Mr. H. G. Jacobson, Technical Project Officer.
The authors are indebted to Dr. W. C. Pfefferle for his many valuable
suggestions.
34
-------
SYMBOLS
Cp specific heat at constant pressure (J/kg-K)
D diffusion coefficient (m2/s)
E activation energy (J/mol)
kQ rate constant (cnr/mol's)
Le Lewis number = X/pDCp
P pressure (kPa)
R gas constant =8.31 J/K-mol
t time (s)
T temperature (K)
u gas velocity (m/s)
A thermal conductivity (J/m«s-K)
p density (kg/m^)
equivalence ratio
[ ] concentration (mol/cm^)
Subscripts
ad adiabatic (reaction temperature)
in condition at catalyst inlet
out condition at catalyst outlet
ref reference condition (velocity)
35
-------
INTRODUCTION
Although many important features of catalytic combustion have been
identified, the design of large scale practical devices for heat and power
generation could make use of a model which accurately describes the system
in terms of fundamental chemical and physical processes, and relates
performance to all the relevant design parameters. The goal of our program
is to develop a useful model for combustion in monolithic catalysts and to
obtain experimental data for testing of the model and the predictions of
other investigators. The present work describes experimental results from
a preliminary study of CO combustion in Pt/A^O^/Cordierite.
36
-------
EQUIPMENT AND PROCEDURE
INLET CONDITIONS
Preheated air at a measured flowrate is supplied to a 690 mm long test
section with 25.4 mm square channel. A catalyst is placed with its
downstream end 90 mm from the test section outlet, and insulated from the
wall by Fiberfrax paper. A fuel injector consisting of five 1.6 mm
diameter tubes, each containing five 0.3 mm diameter holes, is located
440 mm from the catalyst inlet. A combination pitot tube and thermocouple
is mounted 200 mm from the catalyst inlet. In addition to measuring gas
velocity and temperature, the pitot tube is used to extract gas samples
which are analyzed to determine equivalence ratio. Pressure is regulated by
a valve in the exhaust pipe, and taps placed up and downstream of the
catalyst are used to measure inlet pressure and pressure drop. The inlet
conditions used in the experiments are summarized below.
Inlet temperature (T. ) = 700 ± 8 K
Inlet pressure (P ) = 109 ± 4 kPa
Reference velocity (u ,.) = 5 - 31 m/s
CO/air equivalence ratio () = 0.15 - 0.30
H20 - 1.1 - 3.7 mol%
A continuous effort has been made to minimize radial gradients in
temperature, velocity, and fuel concentration of the inlet stream. The
entire test section is insulated so that uniform temperature across the
width of the test section is obtained when sufficient time (M hr) has
elapsed after startup of the air preheat system. Figure 1 shows
37
-------
temperature profiles obtained by traversing the width of the test section
with the pitot/thermocouple probe. These measurements were made with CO
being injected and combustion taking place. The uniformity is independent
of air flowrate over the range of velocities used. Sufficient fuel/air
mixing was obtained only by placing baffles downstream of the fuel injector,
and these affect the velocity profile. By trying various configurations,
the velocity uniformity is being improved while maintaining an even
distribution of fuel. The arrangement used in the present experiments
consisted of two screens, each containing four 8 mm diameter holes, placed
30 and 110 mm downstream of the fuel injector. Another screen perforated
with 1.6 mm diameter holes was 190 mm from the injector and 50 mm upstream
of the pitot tube. The resulting fuel distribution is shown in Figure 2.
There is good uniformity over the measured range, however the largest
equivalence ratio shown is a factor of 2 to 5 lower than those at which the
combustor was run. This is due to the limited range of the CO analyzer
(maximum CO = 3 mol%). Velocity profiles, shown in Figure 3, are less
satisfactory (range ± 10%)„ These have recently been improved by replacing
the second screen by one containing five 6 mm diameter holes, which gives a
range of ± 4%. Average reference velocities were determined using the CO
and air flowrates, inlet temperature and pressure, and the cross section
area of the catalyst. The catalyst inlet velocity, taking into account the
fraction of open monolith area, is u. = 1.5 u ,..
in ref
SUBSTRATE AND EXHAUST MEASUREMENTS
Substrate temperatures are measured by a method similar to that
described by Kesselring, Krill, and Kendall (1). Ni-Cr/Ni-Al thermocouples
are fed through the test section wall and into the ends of catalyst
channels. The lengths of wire inside the catalyst are covered by mullite
insulator and both ends of the channel sealed with ceramic adhesive. The
lifetime of these thermocouples under test conditions is short (5 - 20 hr).
A radial temperature distribution in the substrate was measured at the axial
position of highest substrate temperature, to determine the effect of heat
38
-------
conduction to the wall of the test section. The measured profile is shown
in Figure 4. The variation of substrate temperature at this axial position
is ± 20 K over 45% of the sample width.
An expansion quenched, water cooled, stainless steel gas sampling
probe is mounted through an elbow in the exhaust pipe and moved along the
axis of the test section. Gas samples from the catalyst exhaust were taken
at axial positions from 1 to 76 mm downstream of the catalyst outlet. CO
and C02 were determined using infrared absorption, and 02 by a paramagnetic
analyzer. Exhaust gas temperatures are measured using a thermocouple
mounted in the same way as the gas sampling probe. Water concentration is
determined using a solid state humidity sensor mounted in the inlet air
stream before the heating system.
CATALYST AND FUEL
, The catalyst substrate is split cell corrugated Cordierite with
2
approximately 65% open area, and channel cross section area of 1.9 mm .
The washcoat is alumina with specific area 200 m /g. Platinum loading is
o
4.3 kg/m . Catalysts are pretreated to stabilize surface area and
platinum concentration by running for several hours at maximum substrate
temperatures approximately 100 K higher than those at which exhaust
temperature and composition measurements are made. The present data were
obtained using two different samples whose dimensions and pretreatment are
summarized below.
Sample #8: 24.6 x 24.6 x 76.3 ± 0.2 mm
Pretreatment: run 2.3 hr with propane at maximum
substrate temperature 1510 ± 20 K.
Total run time: 9 hr
39
-------
Sample #10: 24.1 x 24.1 x 76.3 ± 0.2 mm
Pretreatment: run 3.1 hr with propane and 2.1 hr
with CO at maximum substrate temperature
1470 ± 20 K.
Total run time: 22 hr
The fuel is carbon monoxide, 99.5 mol%. Natural propane, 96 mol%, is
used for pretreatment.
40
-------
DATA AND DISCUSSION
The combustor was operated over a range of equivalence ratios and
reference velocities at fixed inlet temperature and pressure. No attempt
was made to minimize unburned fuel, rather the objective was to begin
determination of the relationships among operating conditions and
performance characteristics. Figure 5 shows substrate temperature, exhaust
temperature, and exhaust composition as functions of axial position,
obtained at three equivalence ratios with fixed reference velocity. The
substrate and gas temperatures increase with increasing equivalence ratio,
as expected. Unburned CO is consumed by homogeneous reaction in the exhaust
at a rate which increases with increasing average exhaust gas temperature.
Gas temperature increases with distance downstream due to heat release from
the reaction. The amount of unburned CO at the catalyst outlet is
practically independent of inlet equivalence ratio. In a purely mass
transfer limited regime, unburned CO is expected to increase with
increasing equivalence ratio. If conversion is dominated by homogeneous
reaction, unburned CO should decrease as equivalence ratio increases.
Under the operating conditions of Figure 5, the system is apparently in an
intermediate regime. The substrate temperatures in Figure 5 show a
relative maximum at a distance 15-20 mm from the inlet. This may be an
example of the effect explained by Pfefferle (2) as resulting from heat
recirculation by radiation from the catalyst to the inlet gas stream.
Figures 6a-d show data obtained at various reference velocities with
fixed equivalence ratio. The line representing the substrate temperature of
Figure 6a is reproduced in Figures 6b and c to show the similarity of the
data points. Substrate temperature does not exhibit a sensitive dependence
41
-------
on gas velocity, consistent with the concept that in a system with mass
transfer limited surface reaction rate, the substrate temperature depends in
a first approximation only on the adiabatic reaction temperature and the
local value of the Lewis number (Le = A/pDCp). Exhaust composition is shown
in Figure 6d. The concentrations of <>> and CC>2 are nearly the same for the
three runs, but unburned CO increases with increasing reference velocity.
Figure 7 shows the ratio of unburned CO to inlet CO as a function of
reference velocity for the entire series of runs. The amount of unburned CO
is obtained by extrapolating to the catalyst outlet. This figure combines
data taken at various equivalence ratios and 1^0 concentrations.
Rate coefficients for CO oxidation in the exhaust are estimated from
the slopes of the lines through the measured concentrations. The assumed
rate expression for homogeneous CO oxidation is taken from Howard, Williams,
and Fine (3):
= - k0[CO] [02][E20] exp(-E/RT)
An average measured exhaust temperature was used to determine average
exhaust velocity and density. 02 concentration in the exhaust is assumed
constant, and the fi^O concentration is estimated from the measurement in
the inlet air. An Arrhenius plot of the calculated rate coefficients,
k0exp(-E/RT), is shown in Figure 8. The solid line is the fit by Howard
et al. (3) to a large number of homogeneous CO oxidation data over the
range 840-2360 K:
14 3
k0 = 1.3 x 10 cm /mol-s
E = 126 kJ/mol
The agreement of the activation energies of the present data and the
homogeneous reaction indicates that the oxidation of CO in the exhaust is
not being catalyzed by any long-lived species generated solely by the
catalyst.
42
-------
A gradual loss of catalyst activity was observed with increasing total
run time. The possibility of gaseous metal carbonyl formation in the inlet
stream followed by deposition of elements such as iron and nickel on the
catalyst is being investigated.
43
-------
CURRENT PROGRAMS
The techniques for making detailed measurements of substrate
temperature, gas temperature, and gas composition in a catalytic combustor
were developed during the experiments with CO. The programs now being
pursued include study of propane and low and medium-Btu gases. The
measurements will be made over a range of inlet temperature as well as over
wider ranges of equivalence ratio and reference velocity. The work on
propane is the beginning of a series of experiments, on higher hydrocarbons.
The behavior of low and medium-Btu gases in catalytic combustion is of
special interest because these gases contain fuel components with very
different Lewis numbers (H2, CO, CH^). This may produce interesting effects
in the substrate temperature profile. Measurement of the conversion of fuel
nitrogen (NH^) to NQX is also included in the study.
/' •
\ >
A primary objective has been the measurement of gas temperature and
composition as functions of axial position inside a catalyst. This is
accomplished by drilling an 8 mm diameter hole partway through the catalyst
along its axis. The hole is lined with a ceramic sleeve to minimize the
activity of the wall. A combination gas sampling and thermocouple probe is
moved along the catalyst axis from downstream toward the bottom of the hole
If conditions are properly adjusted, the measured gas temperature and
composition at the bottom of the hole are equal to their values at the same
axial position in a catalyst with no hole. The condition which is most
affected by the presence of the hole and probe is the gas velocity in the
catalyst channels opening into the hole. The problem of establishing the
correct velocity has been solved by feeding pressure taps through the side
of the catalyst to measure the pressure at the bottom of the probe hole.
44
-------
The velocity is adjusted until the pressure is equal to its value at the
same axial position in a catalyst without a hole. Measured pressures show
a linear decrease with increasing distance from the inlet during propane
combustion in the Pt/Al203/Cordierite catalyst. Another requirement for
reliable gas measurement inside a catalyst is that the substrate temperature
be unaffected. This condition can also be satisfied despite the changed gas
velocity because substrate temperature does not have a strong dependence on
velocity. It is the constraint of fixed substrate temperature which
distinguishes this experiment from one in which the overall length of
catalyst is varied. In a catalyst of given length, the substrate
temperature profile is different from the profile in an equivalent section
of a longer catalyst. Direct measurements of fuel, oxygen, and product
concentrations and gas temperature as functions of axial position inside a
catalyst provide data for a rigorous test of a model for combustor operation.
Calculations of species concentrations inside monolithic catalytic
combustors have been presented by Kelly, Kendall, Chu, and Kesselring (4)
and by Cerkanowicz, Cole, and Stevens (5), but an experiment with which to
compare these predictions has, to our knowledge, not yet been performed.
45
-------
CONCLUSION
Improved models and more detailed experimental data are needed to
advance the development of catalytic combustors. Preliminary data on CO
combustion in Pt/Al203/Cordierite monoliths exhibit several interesting
features. 1. Substrate temperature is only weakly dependent on gas velocity.
2. A relative maximum or shoulder appears in the substrate temperature
profile near the inlet. A possible explanation is heat recirculation via
radiation (2). 3. The rate of disappearance of unburned CO in the combustor
exhaust is consistent with homogeneous reaction data. Thus the exhaust
reaction is not catalyzed by species generated by the catalyst.
Pressure measurements inside the catalyst during propane combustion
demonstrate a means of compensating for changes in gas velocity caused by
the presence of a probe hole.
46
-------
REFERENCES
1. Kesselring, J. P., W. V. Krill, and R. M. Kendall. Design Criteria for
Stationary Source Catalytic Combustors. Presented at the Second EPA
Workshop on Catalytic Combustion, Raleigh, NC, June 21-22, 1977.
2. Pfefferle, W. C. The Catalytic Combustor: An Approach to Cleaner
Combustion. Journal of Energy, 2(3): 142-146, 1978.
3. Howard, J. B., G. C. Williams, and D. H. Fine. Kinetics of Carbon
Monoxide Oxidation in Postflame Gases. Fourteenth Symposium
(International) on Combustion, The Combustion Institute, Pittsburgh, PA,
1973. pp. 975-986.
4. Kelly, J. T., R. M. Kendall, E. Chu, and J. P. Kesselring. Development
and Application of the PROF-HET Catalytic Combustor Code. Paper 77-33,
Presented at the Fall 1977 Meeting, Western States Section, The
Combustion Institute, Stanford, CA, October 17-18, 1977-
5. Cerkanowicz, A. E., R. B. Cole, and J. G. Stevens. Catalytic
Combustion Modeling: Comparisons with Experimental Data. Presented
at the ASME Gas Turbine Conference and Products Show, Philadelphia, PA,
March 27-31, 1977.
47
-------
0)
}-l
cfl
S-i
CX
s
01
H
710
700
690
710
700
690
^—
-ooo000°°00oor, = °-26
0 0 u = 28 m/s
l 1 l l
_O O O Q _ (j5 = 0.24
O u = 17 m/s
l l i 1
710
700
690
710
700
690
-0oooOooooooo00* = Jo2m/s
1 1 1 1
- 000000°°°0 «)> = 0.24
O O O u = 5 m/s
O
™
5 10 15
Radial position (mm)
20
25
Figure 1. Inlet gas temperature vs. radial position.
P. = 112 kPa
in
48
-------
.07
.06
0
w .05
cfl
0)
0
c
Equival
•
o
•e-
.03
.02
-
5 10 15 20
Radial position (mm)
25
Figure 2. Inlet CO concentration vs. radial position.
P. =112 kPa
in
T. = 700 K
in
u = 60 m/s
49
-------
CO
la
34
32
30
28
26
24
22
20
O O
o 4, = o
3 18
o
iH
CU
> 16
O O
o o
= 0.24
14 h
12
10
8
00
o o
00 = 0.27
oooo°°°oo
J I I I
5 10 15 20 25
Radial position (mm)
= 0.19
Figure 3. Inlet gas velocity vs. radial position.
P. = 112 '
in
T.n - 700 K
50
-------
1500 r—
§ 1400
CO
0)
ex
§
H
1300
8 12 16
Radial position (mm)
20
24
Figure 4. Substrate temperature vs. radial position.
(axial position = 50 mm from catalyst inlet)
51
-------
1 JUU
1300
0)
3 1100
(13
HI
a.
H
900
700
T ,
? 1 H • •'>
-
I
-
V V \;
'VV A A L
A o c
catalyst
outlet
- T.
in i 1
1 i
50 100
Distance from catalyst inlet (mm)
150
trate
•
i
T
gas
0
A
V
0.24 ± 0.03 = <()1
1.10 <)>!
1.22 (bi
Tad ^K^
1470 ± 80
1540 ± 80
1610 ± 80
Figure 5a. Substrate temperature and exhaust temperature vs.
axial position at different equivalence ratios.
P. = 112 ± 2 kPa
in
T. = 700 ± 8 K
in
u = 19 ± 1 m/s
ref
H20 = 3.0 ± 0.2 mol%
Ap/P. = 4.8 ± 0.3 %
in
Catalyst sample #10
52
-------
20 i—
10
8
6
4
S 2
o
£
a 1
° a
•H -0
* .6
i i
§ .4
CJ
C
O
0 .2
.1
.08
.06
.n^
—
v E 5
-".... °
-
.
CO
_
-
-
-
cata
LSV ^
JA
fi
O
V
S
V
0
A
lyst V
outlet
—
-
-
i 1 i
c
1
50 100
Distance from catalyst inlet (mm)
150
0 0.24 ± 0.03 = !
A 1.10 !
V 1.22 ^
Figure 5b. Exhaust composition vs. axial position at
different equivalence ratios.
53
-------
J-i
cd
i~i
cu
ex
g
H
1500,—
1300 -
1100
900
700
catalyst
outlet
I- T
in
I
* substrate
Ogas
50 100
Distance from catalyst inlet (mm)
150
Figure 6a. Substrate temperature and exhaust temperature
vs. axial position.
P. = 109 ± 2 kPa
in
T. = 700 ± 8 K
in
cf> = 0.24 ± 0.02
u = 15 + 1 m/s
ret
H20 - 1.2 ± 0.1 mol%
AP/P. = 3.5 ± 0.1 %
in
Tad = 147° ±.50 K
Catalyst sample #8
54
-------
1500 i—
1300 -
-------
0)
a.
a
0)
H
1500r—
1300 -
1100
900 -
700
-
• substrate
Ogas
catalyst
outlet
- T
in |
1 , 1
50 100
Distance from catalyst inlet (mm)
150
Figure 6c. Substrate temperature and exhaust temperature
vs. axial position.
u = 30 ± 1 m/s
ref
AP/P. = 8.4 ± 0.1 %
in
56
-------
a
0)
u
u
20
10
8
6
O
C -L
° .8
.6
.2
.1
.08
.06
.04
I
CO
catalyst
outlet
50 100
Distance from catalyst inlet (mm)
150
V
0
A
uref (m/s)
15 ± 1
21 ± 1
30 ± 1
Figure 6d. Exhaust composition vs. axial position
at different reference velocities.
57
-------
.5
.4
•H
o
CJ
OJ
C
M
,0
C
O
•H
.3
.1
00
o
10 20 30
Reference velocity (m/s)
Figure 7. Ratio of unburned CO to inlet CO vs. reference velocity.
P. = 109 ± 4 kPa
in
T. = 700 ± 8 K
in
4) = 0.19 - 0.30
H20 = 1.1 - 3.7 mol %
maximum substrate temperature = 1340 ± 90 K
58
-------
o
ro
g
OJ
•H
U
•H
O
O
6
4
1
.8
.6
.4
.2
.1
.08
.06
.04
.02
oo
1400
Temperature (K)
1200
I T
8 9
1/T (K"1 x 10"4)
1000
10
Figure 8. Rate coefficient for CO oxidation in the exhaust
vs. temperature.
59
-------
APPLICATION OF CATALYTIC FLAME
STABILIZATION FOR AIRCRAFT
AFTERBURNERS
By:
LEONARD C. ANGELLO
AIR FORCE AERO PROPULSION LABORATORY
AIR FORCE WRIGHT AERONAUTICAL LABORATORIES
WRIGHT-PATTERSON AIR FORCE BASE, OHIO 45433
and
DR. THOMAS J. ROSFJORD*
UNITED TECHNOLOGIES RESEARCH CENTER
UNITED TECHNOLOGIES CORPORATION
EAST HARTFORD, CONNECTICUT 06108
* Formerly with the Air Force Aero Propulsion Laboratory
61
-------
INTRODUCTION
Catalytic flame stabilization as applied to aircraft afterburners
encompasses the uses of a catalytic surface to initiate, stabilize, and
provide a continuous pilot for flame propagation. Catalytic flame sta-
bilization differs from catalytic combustion (a heterogeneous process) in
that the catalyst is used only to "boot-strap" the fuel-air mixture to
temperature and species conditions such that gas phase reactions can pre-
dominate; the balance of energy release is accomplished through homo-
geneous chemical processes. Previous Air Force Aero Propulsion Lab-
oratory (AFAPL) studies have indicated that an afterburner performance
advantage can be realized by employing catalytic flame stabilization in
aircraft afterburners. Further, because of the accelerated chemical acti-
vity associated with catalysts, an improved ignition capability is expected.
Test programs, however, to verify these indications have as yet not been
performed.
The purpose of this paper is to outline an experimental test pro-
gram currently being performed by the Air Force Aero Propulsion Laboratory
to investigate the applicability of catalytic flame stabilization for
aircraft afterburners. This experimental program shall make use of a
J85-5 afterburning turbojet engine to test concept feasibility. The goal
of the AFAPL program is to employ catalytic flame stabilization in such a
manner that system performance gains result.
63
-------
PROGRAM DESCRIPTION
The AFAPL paper study has considered one application of catalytic
materials for aircraft afterburner flandholders -- namely the use of
catalytically-active honeycomb devices instead of conventional bluff bodies.
The study projects the performance advantages of reduced flameholder pressure
drop and/or increased afterburner combustion efficiency. Other potential
benefits may include improve ignition capability and reduced acoustic instability
(high frequency "screech" and low frequency "rumble").
Based on these indications an experimental test program has been initiated
at the Air Force Aero Propulsion Laboratory to evaluate concept employing
catalytic flandholders in aircraft afterburners. The AFAPL program shall test
catalytically-active flameholder designs in a J85-5 turbojet engine. Candidate
stabilizers shall be examined under a full range of afterburner conditions. A
conventional flameholder design will also be examined as a data baseline. The
performance of the candidate and conventional devices shall be compared to verify
and upgrade analytical model predictions.
Data shall be acquired to evaluate the flame stabilization limits, the
ignition capability, the imposed total pressure drop and the combustion
efficiency associated with each test item. Combustion efficiency shall be
determined by integration of point emission measurements. Pressure drop shall
be determined by measurement of total pressure upstream and downstream of
the flameholder. Flame stabilization limits and ignition capaoility shall be
determined qualitatively.
64
-------
BACKGROUND
The combustion chamber of a jet engine afterburner is a mechanically
simple device: It consists of a straight flow-through duct which contains
one or more flandholders to stabilize a reacting fuel/air mixture. The fuel
is introduced through several spraybars upstream of the flandholders, re-
sulting in a substantially premixed, prevaporized flow. Conventionally, the
flameholders are annular "V" gutters or similar bluff bodies.
The flow conditions of an afterburner differ from the engines main
burner in several respects—it operates hotter, at lower pressure and with
higher velocities.
An afterburner is not subject to the maximum exit temperature limitations
imposed on the main burner. Structual considerations of the turbine blades
of current aircraft jet engines restrict the maximum combustor exit tempera-
ture to approximately 1650K. The combustion gases produced in an after-
burner do not pass through any turbomachinery, and are therefore not similarly
limited. For this reason, at the maximum power condition, afterburner com-
bustion occurs at equivalence ratios near unity, while current main burners
are confined to values below 0.4.
An afterburner operates operates at significantly lower pressure than a
main burner. Typically, the exit nozzle is a variable area, convergent design
which actuates to sustain critical conditions. Thus, the afterburner chamber
pressure is approximately only twice the ambient pressure. In contrast,
current main burner combustors operate up to thirty times the ambient.
A jet engine afterburner functions to augment the maximum thrust that
can be produced by the main engine. It becomes operational only after the
main engine reaches its maximum power output—termed the "military" condition.
In this mode, the engine passes its maximum mass flow. Additionally, the
65
-------
inlet gas temperature is high with typical values of approximately 950K.
These two effects, plus the low pressure, result in a typical average inlet
velocity of 130 m/s. Such values are more than triple the main burner inlet
velocity.
Conventional afterburners do not operate at peak combustion efficiency.
Typically, they release only between 85%-95% of the available chemical poten-
tial while main burners routinely release 99%. The poorer performance is
attributable to the high speed of the flow, which approaches 250 m/s. The
practical restriction of reasonable afterburner lengths results in a short
residence time and thus incomplete combustion. Faster.reaction rates would
enhance the axial rate of energy release as depicted in Figure 1. If the
accelerated change of combustion efficiency is achieved, the designer could
choose either higher performance or decreased hardware length. That is, if
current lengths are acceptable, a performance increase associated with an
improved efficiency would result (Aric). Alternatively, if the current per-
formance level is acceptable, it could be achieved at a shorter length, Lc.
It is not possible to precisely predict the length saving without a measure-
ment of the combustion efficiency. However because of the expected asympotic
approach to high levels of efficiency, the length reduction should be sub-
stantial.
Another implication of the high speed flow in the afterburner is that
the flameholder which pilots and establishes the reactions cannot be wide in
order to avoid unacceptable pressure drop. Therefore, the combustion
originates from a few small sources (one to three flameholder rings) which
again because of the high velocity, spread at angles on the order of 7°.
A minimum afterburner length would be the distance necessary to completely
fill the cross section with reaction - that is, the distance at which the
spreading sources merge. Larger sources would proportionally reduce the
required length as shown in Figure 2 and thus provide the designer the option
of shorter hardware length. It would also be expected that for a given
length, a performance increase would be associated with merging the combustion
zones nearer the flameholder, thus representing the option of increased
system performance.
66
-------
One means to achieve the benefits of a larger flaraeholder, without
imposing unacceptable pressure losses, is to consider using a porous body
instead of a solid body. This design would permit flow through as well as
around the device, producing less pressure drop than a conventional (bluff)
body of equal frontal area. Therefore, the porous body could be made wider,
reaping the benefits discussed above. In essence, this approach could be
termed "toward a flat flame afterburner," for such a design would ensure
maximum energy release in minimum lengths. The ideal case of completely
filling the combustor cross section with a porous flameholder is not
attainable, however, because of the associated pressure loss.
It is recognized that replacing the bluff body stabilizer with a porous
device alone is not acceptable. A conventional flameholder successfully
stabilizes the overall combustion processes by providing a continuous pilot
for the spreading combustion wave. The recirculating wake flow behind the
solid body provides gas residence times the same order as the characteristic
chemical reaction times, and thus establishes the required piloting phenomenon.
The porous stabilizer does not provide a similar piloting zone. The
flow through the body tends to wash out the recirculating, and hence piloting,
character. This deficit can be corrected by making the device catalytically
active. In this case, the reactants passing through it would be substantially
oxidized in a manner similar to a catalytic main burner. Since the after-
burner is always accompanied by a military main burner mode, it always
receives hot inlet gases. This condition removes the low inlet temperature
problem that plagues main burner applications. It also increases the
reaction rates, offsetting the problem of decreased catalyst contact time
due to the high velocities. Note that complete oxidation of the internal
mixture is not a requirement. The degree of reaction would be chosen to pro-
duce a stabilizer wake of hot, combustion species similar to that existing
behind a bluff body device. Thus, the desired flame stabilization should
be achievable using catalytic, porous bodies which would initiate the com-
bustion over a comb ustor cross section greater than a conventional device
of equal pressure loss.
67
-------
FLAMEHOLDER MODELING STUDIES
An in-house modeling effort was initiated at the Air Force Aero Propul-
sion Laboratory by T.J. Rosfjord and B. Eresman to investigate the porous,
catalytically-active flameholder concept. Experienced gained in other cata-
lytic combustion studies indicated that noble metal catalysts and honeycomb
substrates were appropriate flameholder components. The purpose of this
modeling effort was to indicate the trade-offs between flameholder blockage
and imposed pressure loss for various honeycomb catalyst substrates.
The system analyzed is shown in Figure 3. A uniform, constant density
mixture enters the constant area duct, with a portion (Flow II) flowing
through a honeycomb blockage while the remaining flow (Flow I) passes around
it. The honeycomb is characterized by its fractional blockage area, cell
diameter, d, and cell length, H.
The model derived by Rosfjord and Eresman for this system solely incor-
porated fluid mechanic considerations, and not chemical reactions. To
remedy this deficiency, a catalytic reactor model developed at Exxon Research
and Engineering (1) was exercised. A JP-4/air mixture at conditions indi-
cated by the fluid mechanic model were input to the kinetic model which
sought substrate/catalyst combinations to produce a substrate exit flow
temperature of 1350°K: In general, a cell £/d - 40 was required to achieve
this condition.
Results of the fluid mechanic modeling performed by Rosfjord and
Eresman are plotted in Figure 4. This plot illustrates the trade-off of
pressure loss with increasing flameholder blockage, with honeycomb cell
H/d as a parameter.
68
-------
A Conventional J85-5 flameholder represents a 36% blockage and imposes
a 3.2% pressure loss. The model predicts that similarly sized honeycomb
with H/d = 40, would impose only a 2.0% loss. A 45% blockage honeycomb
would impose a loss equivalent to the conventional solid body. Considerations
such as these guided the design of test hardware as shown in Figure 5.
Figure 6 illustrates the fraction of flow through the honeycomb with
increasing flameholder blockage. As expected, the portion of total flow
through the honeycomb is significantly less than its fraction of the total
area. For a fc/d = 40, the 36% blockage device passes 16% of the flow while
a device of 45% blockage—the size which imposes a pressure loss equivalent
to the conventional flameholder design passes 25% of the total flow.
69
-------
PERFORMANCE TRADEOFFS
A study was conducted to quantify performance tradeoffs for the porous,
catalytic flameholder design illustrated in Figure 4. The study considered
the potential gain in combustor efficiency (TI) versus the possible increased
flameholder pressure loss (PL) associated with this catalytic flameholder
design.
The study was carried out using an existing turbojet/turbofan model
(2) capable of predicting the design point performance of a specified system
with parametric variations. The propulsion system investigated in this
study was a low pressure ratio, turbojet engine of approximately 4000 pounds
thrust. Variations in combustion efficiency and flameholder pressure loss
were studied while maintaining thrust rating. The net system gain or loss
was reflected as changes in specific fuel consumption (SFC)-
Figure 7 illustrates the tradeoff as represented by changes in SFC, at
a maximum afterburner power mode. Assume that the baseline burner operates
at n = 90%, PL = 3.2%. Increased efficiency to n = 95% at constant PL
reduces SFC from 1.81 to 1.75, a 3.3% decrease. Alternatively, a decrease to
PL = 2.0% at constant efficiency translates to a 1.7% SFC reduction. The
nearly linear relationship of SFC and n for various PL levels indicates that
approximately a two percentage point improvement in efficiency is required
to offset each percentage point increase in pressure loss.
The tradeoff depicted in Figure 7 implies that a system gain is
achieved for flameholders which result in ri improvements which outweigh
increased PL. That is, higher pressure losses may be acceptable. This is
not true, however, if one considers the total engine operating envelope.
The afterburner is utilized for only a small fraction of the total engine
life. During AB operation, its efficiency is an important consideration.
70
-------
However, the flameholder imposes pressure losses at all times.
This situation is depicted in Figure 8. The SFC for various degrees
of afterburner operation is plotted with combinations of n and PL as a
parameter. The baseline case is n = 90%, PL = 3.2%. As shown, increases in
PL can be offset with TI gains at afterburner modes, but result in an SFC
penalty without afterburner operation. As well, decreases in PL will
benefit SFC at all power settings.
71
-------
SUMMARY
In summary, the concept of employing catalytic flame stabilization
to the design of flameholder for aircraft afterburners has been developed.
Flameholder modeling studies have been presented which show this concept
to offer performance benefits in terms of increased efficiency and reduced
pressure drop losses. Further benefits may also include increased ignition
capability as well as reduced acoustic instability. The feasibility of
achieving these advantages can only be evaluated by experimental test
programs such as those being currently performed at the Air Force Aero
Propulsion Laboratory.
72
-------
REFERENCES
1. Cerkanowicz, A.E., Cole, R.B., and Stevens, J.G., "Catalytic Combustor
Modeling: Comparisons with Experimental Data," ASME Paper 77-GT-85,
March 1977.
2. Witherell, Capt. R.E., Design Point Turbine Engine Performance Program,
AFAPL TR-68-88. 1968.
73
-------
c
C'~< ->
O
u
1.0
o
-------
NARROW SOURCES
-*4
tn
WIDE SOURCES
FIGURE 2
-------
r t t i \.ii.i\-. 1.1; i.i
""V V 1 V ' I I -. i I i i.
U"n>. ^: A v\ v -^ t \
CTl
UNIFORM PROPERTIES AT a
FLOW I -- NOZZLE
FLOW II -- BUNDLE OF ROUGH PIPES
WITH HEAT ADDITION
FIGURE 3
-------
0.12
0.08-
0.04-
0.00
0.00
SOLID
0.25 0.50 0.75
FRACTIONAL BLOCKAGE AREA
1.00
FIGURE 4
-------
FIGURE 5
78
-------
1.00
vo
0.75
1.00
FRACTIONAL BLOCKAGE AREA
FIGURE 6
-------
00
o
u
u.
94
EFFICIENCY
FIGURE 7
-------
1.80
ETA
ETA
ETA
ETA
•90, PL
'90, PL
90, PL
97, PL
1.50'
-------
EPRI VIEW OF CATALYTIC COMBUSTION
by
A. C. Dolbec
Program Manager - Power Generation
Advanced Fossil Power Systems
Electric Power Research Institute,
Palo Alto, California 94303
83
-------
ABSTRACT
As current EPRI efforts focus on combustion turbine applications of catalytic
combustors/ this paper presents the specific turbine duty cycles and general
usage. Peaking and baseload applications are discussed.
The expected problems that are apparent at this time are discussed even though
the known list is not complete. Apparent catalytic system advantages are
discussed in length insofar as they are apparent at this time. Some of these
will turn out to be speculative but possibly others real, offering a hope of
considerable flexibility for turbine designs in the future. Current work at
EPRI regarding catalytic combustion for turbines is outlined, along with those
things expected to be done during 1979. Additional work, as yet undefined, is
planned.
85
-------
Introduction
Although this paper is titled "EPRI View of Catalytic Combustion," the program
and plans presented here represent only the Power Generation/Advanced Fossil
Power portion of EPRI. Other portions of EPRI, such as the Fuel Cell Program
and Fossil Fuel programs, have followed progress in the catalytic combustion
area, and while no major work has been committed, there is a continuous evalu-
ation of the potential of catalytic combustion toward applications other than
combustion turbines. Specifically, this paper describes work and plans for
the use of catalytic combustion techniques on combustion turbines.
The combustion turbine is considered a vital portion of most advanced fossil
power schemes now receiving considerable R&D attention. Several EPRI studies
(references 1 and. 2) indicate that combustion turbines will figure prominently
in intermediate and probably baseload power plant applications in the late
1980's and 1990's. At those times it is expected that environmental require-
ments will demand considerably lower emission levels than possible with
today's technologies. For these reasons certain generation schemes which
allow much improved emissions have been favored by EPRI over others that show
less promise for great improvement.
Combustion Turbine Applications^
Catalytic combustion techniques for combustion turbines need to be explored in
two general areas. As there are now some 2000 gas turbines in the United
States used for power generation, a "retrofit" market exists for improving the
emissions of current turbines where these are located in such areas that
improvements would contribute to the overall air quality of the locale. The
other basic direction for catalytic combustion is in the design of an improved
and more reliable power plant. These prospects appear at this'time to be
quite exciting. On a retrofit basis, combustion turbines would have to be
modified in the field to include a catalytic element within the basic struc-
ture of combustion liner and transition piece, as shown in Fig. 1. The
objective of this work would be to design a system that could be furnished at
minimum cost, which usually means changing the fewest number of existing
I
parts. Most of the some 2000 turbines in commercial use generating power
86
-------
today are used in "peaking" service. This indicates a cycle of some 1500
hours a year with possibly up to 200 starts per year. Thus, catalytic
elements in the peaking application need not have an extremely long life but
should be able to take significant cycling. As many of these turbines would
be at least 10 years old by the time catalytic equipment was available for the
market, the cost of such a retrofit would be extremely important.
>
As the peaking application offers many opportunities for inspection, the
ability of the system to be quickly inspected and assure its integrity is
highly desirable. These are the general design considerations for a retro-
fitted system usable on present peaking applications.
New designs of combustion turbines would be used in peaking applications but
also in intermediate to baseload duty cycles as new plant technologies become
available. This means that for intermediate load use, operation would be
between 2000 and 5000 hours per year, whereas baseload applications (probably
not significant before the 1990*s) would be in excess of 5000 hours a year up
to some 7500 to 8000 hours per year. Intermediate and baseload applications
would then require a once-a-year minimum time for replacement, and hopefully
inspection could be once a year with replacement every three years. This
could mean that some 24,000 hours of usage would be required for catalytic
elements before they should be replaced. Newly designed machines that might
be exposed to this service could benefit in several ways from a catalytic
combustion process.
Some Advantages of Catalytic Combustion
Of course, the basic motive of catalytic combustion as perceived now is the
potential for emissions reduction. In addition, other features of the cata-
lytic combustor are possible, although at this time very little has been done
to analyze how realistic they might be.
Because of the great reduction in peak flame temperature, the amount of heat
radiated to metal parts of the machine is greatly reduced. This offers a
fewer number of hot parts which should lead towards improved machine relia-
bility. If ways can be found to initiate combustion by catalytic means, then
87
-------
--the machine can be lit off in what could only be described as a "soft" manner,
taking many seconds to achieve peak flame temperature as contrasted to the
present spark ignition and immediate (within a few seconds) light off. The
advantage of a soft ignition is that it much reduces the thermal shock to
-machine parts, particularly when a cold machine is being started. For new
turbine designs, the pattern of temperature across the combustor outlet with a
catalytic scheme may be considerably less than with the present flame and air
supply arrangements. This offers the possibility of reducing the maximum
-^temperature exposure for metal parts while keeping the average temperature the
same. In other words, hot spot temperature reduction could offer prolonged
life of hot gas path parts downstream of the combustion area. The design of
present day combustors is largely an empirical science. Many data points are
developed by use of combustion test stands, and these describe the overall
temperature, stress, and emission levels that can be expected from a given
configuration. An indicated advantage of the catalytic schemes is the ability
to predict the temperature and emissions with much more accuracy when scaling
from one size of machine to another or one design to another.
This offers the promise that many expensive and time-consuming combustion
tests can be eliminated once a secure data base is established that defines
the catalytic combustion design approach. By the way, if improving metal
temperatures can be turned into a design feature, the amount of cooling air
now necessary to keep metal temperatures below roughly 1500°F may be diverted
to more efficient use of the cycle.
Problems to be Solved
The turn-down ratio, or region of stable control for a conventional combustor,
is roughly 4.5 to 1. Catalytic combustion experiments at this time have
demonstrated a turn-down ratio of less than 2 to 1 and potential techniques
for improving this characteristic have not been identified. This means that a
current design would have to have a conventional combustor for ignition and
some portion of the speed and load profile. Such a system is more expensive
but, even worse, might have a significant reduction in reliability.
88
-------
Of great concern to turbine designers is the possibility that a catalytic
monolith structure would fracture and be ingested by the turbine. Designs of
such structures must be demonstrated that will pass through the turbine
without causing damage to the blading. There is a strong feeling among many
people in the industry that this can be accomplished.
The premixing required for stable catalytic combustion is not expected to be
greatly different from the premixing required for multi-stage combustion now
under development by most gas turbine manufacturers. The difficulty with any
large volume premixing area is the potential for explosion and the consequent
safety hazard it poses. A considerable amount of research should be applied
to the area of premixing gas turbine fuel and air to minimize the possibility
of an explosion. This data needs to be characterized for a wide range of
liquid and some future gas fuels, as each manufacturer will have to have
particular design features that address this concern.
One of the design objectives of any combustion system, particularly a cata-
lytic system, is a satisfactory life of the parts. Practical life
expectancies must be established for various catalysts as well as system
designs that would allow easy access to change the catalyst should the life be
shorter than the application requirements. For instance, the peaking applica-
tion offers opportunities to change parts at frequent intervals. Base and
intermediate load operation is much more restrictive and would permit a
change-out probably no more often than once every 8000 hours. Catalyst life
at the higher turbine temperatures contemplated in the future needs to be
established.
The progression of turbine inlet temperatures in the past and the estimated
future is shown in Fig. 2. While catalysts can be demonstrated for today's
temperatures, work needs to be started now to demonstrate their applicability
to the higher temperatures required for gas turbines in the future .
Within the next ten years it is expected that coal-derived liquids will be
available for gas turbine fuels in at least a limited sense. The type of
coal-derived liquid may vary with the type of coal and change considerably
with the process used for conversion. If catalytic elements applied today
89
-------
would not be compatible with future coal-derived liquids, then they would have
a very limited application lifetime. Demonstration of the types of
contaminants in coal-derived liquids that could affect catalyst lifetime must
be shown in order to confidently plan new combustion systems for the future.
Coal-derived gases have the same variability, depending on the process, and
also must be tested for catalyst compatability.
The potential problems mentioned above are not intended to be a complete list
but only some of the more significant problems that have been identified at
this time. Many of these are receiving attention at the moment, and it is
expected that all will be on some path to solution before the end of 1979.
Establishing a data base that will allow people to proceed with the design of
catalytic combustion systems is an immediate priority and is being addressed
in various areas.
Current EPRI Work
Several groups are working on catalytic combustion, including DOE, EPA,
combustion turbine manufacturers, several R&D companies, component manufac-
turers, and EPRI. Current EPRI concerns and work are directed at several
areas. The first, of course, is to develop an overall plan for establishing
catalytic combustion in gas turbines. As mentioned previously, this involves
one solution to retrofit existing turbines in the field and another solution
to create a data base for designing new machines. Since it is considered a
significant problem to establish future priorities, a test for catalyst
poisoning with coal-derived liquid fuels is being pursued immediately. If
special work is to be required for the use of coal-derived liquids, it must be
identified now, as a solution could be quite lengthy. Since some future fuels
could have significant fuel-bound nitrogen, it is important to establish the
ability to convert fuel-bound nitrogen in a combustion turbine using catalytic
systems. Work will commence in 1979 to identify the most significant
combustion schemes that would enable high fuel-bound nitrogen to be converted
to free nitrogen in a catalytic system. The amount of this conversion is
important in order to establish the future of fuel economics. Other projects
expected to start in 1979 as part of the EPRI program will be a characteriza-
tion of the premix requirements, with an emphasis on necessary data to
90
-------
minimize explosion potential, and data to establish possible lifetimes of
various catalytic elements.
Owing to the great potential exhibited by catalytic combustion experiments to
date, a system of this type applied to combustion turbines must be pursued
with all possible speed. Since catalytic combustion is the first new type of
combustion system applied to turbines since they have been used for power
generation, a considerable amount of work needs to be accomplished in order to
properly establish a design data base. It is difficult to say when all this
work will be done so that turbine designers can confidently proceed to design
reliable combustion systems. Hopefully, in five years we would start to see
combustion systems being offered for units in the field and the next genera-
tion of gas turbine designs that should be commercialized within 10 years
would use some form of catalytic combustion, either as a basic combustion
system or as an option for the domestic U.S. market.
91
-------
References
1
Spencer, D., Gildersleeve, O., "Market Potential for New Coal
Technologies," EPRI Journal, May 1978, p. 19.
2. "Clean Coal - what Does it Cost at the Busbar?" EPRI Journal, November
1976, p. 6.
92
-------
Compressed
air
Fuel [
Combustor
First-stage nozzle
Figure 1. Gas Turbine Hot Gas Path
93
-------
Firing Temperature (°F)
3000
2600 —
2200 r—
VO
1800 \—
1400
1955
1960
1965
1970
1975
1980
1985
1990
Figure 2. Estimated Turbine Inlet Temperature Versus Years
-------
OVERVIEW OF CRIEPI CATALYTIC COMBUSTION RESEARCH PROGRAM
By:
Yoshimi Ishihara and Hisashi Fukuzawa
Energy and Environment Laboratory
Central Research Institute of Electric Power Industry
Tokyo, Japan 182
95'
-------
ABSTRACT
The objective of this research program is the development of new types of
low-NOx-emission utility boilers and high-efficiency gas turbines through the
catalytic combustion of light fuel oils, such as naphtha and NGL, or fuel gases.
The steam temperature used in steam power plants is limited to 538°C or
566°C by the thermal fatigue characteristics of the metals used for the steam
pipes. Steam at this temperature can easily be obtained by heat transfer with
gas at LQOO°C, without the need for higher-temperature gas. Fuel can be
completely burned at IOOO°C through the application of catalytic combustion.
Moreover, the catalyst bed can be ;cooled and the temperature maintained at
approximately IOOO°C by the use of immersed water tubes and steam tubes. We
anticipate that the use of this method of combustion will enable reduction of
the NOx concentration to below the level obtainable by flue gas treatment for
NOx control (30 p-pm or less).
Gas turbine efficiency can be improved by raising the combustion gas
temperature, but this has the negative effect of increasing the NOx concentra-
tion.
CRIEPI is planning to investigate combustion methods combining two-stage
combustion, flue gas recirculation, and catalytic combustion for the maximum
reduction of NOx formation.
97
-------
INTRODUCTION
Control of NOx emissions from stationary sources in Japan is prescribed
by type of fuel and type and capacity of equipment as shown in Table I. In
metropolitan and industrial areas where the environmental standards cannot be
met even if the emissions regulations laid down by the government are observed,
stationary sources are placed under even tighter controls by agreement with
local governments. To restrict total emissions to below current levels, it is
necessary /to strengthen emission control measures for existing plants. More-
over, the maximum possible reduction is required for the construction of new
plants. Thus, for the natural gas boilers in the Chita Thermal Power Plant of
the Chubu Electric Power Company, which was constructed in 1977, the NOx
concentration was reduced to approximately 50 ppm by combustion control. Then,
further reduction to approximately 10 ppm was achieved by the installation of
NH3 catalytic reduction equipment.
Fuels which can be easily gasified are currently being used in Japan for
power generation. In 1977, 3,805 x I03 kA of naphtha, 5,586 x I03 tons of
liquid natural gas (LNG), and 2,990 x I03 kJL of natural gas liquid (NGL) were
used. Consumption of LNG and NGL is tending to increase.
In boilers burning these fuels, NOx in flue gas can be easily reduced to
100 ppm or less by improved combustion control to meet the regulation values
set by the government. However, in the installation of new utility boilers,
cases:jn which it is necessary to reduce NOx to the maximum extent possible by
flue gas treatment (FGT) for NOx control are anticipated to increase.
Catalytic reduction with NH5 as an additive is already capable of practical
application as an FGT process for NOx control. This process yields clean flue
gas containing almost no S02 or particulates, equivalent to the combustion gas
from the fuels mentioned above. However, consumption of a large quantity of
98
-------
NH3, which in itself requires a considerable input/of energy, is not desirable
from, the viewpoint of energy conservation.
If the combustion temperature in the boiler is reduced to approximately
IOOO°C, it is thought possible to reduce the concentration of NOx to the same
level as that enabled by FGT technology. In other words, it is considered
possible to develop an energy-conserving, low-NOx boiler.
With this concept, CRIEPI has initiated research in catalytic combustion
(.
and will set up experimental equipment this year.
In addition, CRIEPI is cooperating in a government-sponsored project to
*
develop a combined-cycle system employing a steam turbine and a high-temperature
gas turbine burning clean fuels such as natural gas, naphtha, or kerosine. It
is well known that the NOx concentration is increased by raising the combustion
gas temperature in a gas turbine. Thus, NOx counter-measures are required.
Although catalytic combustion does not come within the purview of this project,
CRIEPI research into catalytic combustion includes the application of NOx
control to: .the..burner of highrtemperature gas turbines.
This paper describes the current status of catalytic combustion technology
in use in Japan and introduces the CRIEPI R&D program on both catalysts and
catalytic combustion systems. The aim of this program is the application of
catalytic combustion technology to low-NOx, high-efficiency power plants.
CURRENT STATUS OF CATALYTIC COMBUSTION TECHNOLOGY IN JAPAN
Catalytic combustion has been in use for several decades in "platinum
pocket heaters" (portable warmers) for the advantages of low temperature in
comparison with flame combustion as well as the property of combustabiIity
without flame. With this heater, cotton wool in a container is soaked with
petroleum benzine or ethyl alcohol as a fuel. This is then covered by glass
wool with platinum catalyst. As fuel vapor diffuses to the outside through the
platinum catalyst, it gradually oxidizes and generates heat. This is used in
cold areas to warm the hands or a part of the body when fishing or playing golf.
99
-------
Catalytic combustion equipment has recently been installed for deodorization
of waste gas from plastic fabrication, printing, and paint factories. Table II
shows some examples of major installations (Reference I).
CRIEPI R&D PROGRAM ON CATALYTIC COMBUSTION
BACKGROUND
The steam temperature of recently-constructed steam power plants has
frequently been limited to 566°C (I005°F) or 538°C (IOOO°F) due to such factors
as the strength of the pipe steel at high temperatures, oxidation corrosion,
and the cost of materials. Thus, improvement of generation efficiency is being
sought by means of raising the steam pressure. To further raise generation
efficiency, R&D on a combined-cycle system incorporating gas turbines is being
advanced. Steam at 538 - 566°C is easily obtained by heat exchange with gas
at IOOO°C.
NO formed by the thermal reaction between N£ and 02 in the process of
combustion, i.e., thermal NO, is controlled by the combustion temperature,
fuel/air ratio, and residence time in the combustion zone, The effect of the
combustion temperature is particularly large, Hence, if the combustion tem-
perature is reduced to IOOO°C, it is thought to be possible to reduce the
concentration of NOx to the level obtainable by FGT for NOx control.
Catalytic combustion enables the complete burning of fuel at IOOO°C. Siince
f
the operating conditions of a catalytic combustor are limited by the critical
temperature of the catalyst, development of a catalyst which can be used at
IOOO°C and a catalyst bed cooling system is necessary.
RESEARCH PROGRAM
Figure I shows the boiler being considered by CRIEPI for catalytic ""
combustion at IOOO°C. Fuel gas consisting of vaporized fuel and flue gas
recirculated from the inlet or outlet of the air preheater is premixed with air
and introduced to the catalyst bed and burned. The heat generated by this
combustion is cooled and removed by the water tubes of the steam generator, which
100
-------
are arranged over the catalyst bed,and the steam tubes of the steam superheater.
The temperature in the catalyst bed is thereby maintained at approximately
IOOO°C. Exhausting of flue gas which has passed through the catalyst bed from
the stack through a heat exchanger such as a steam reheater, economizer, or
air preheater is the same as with conventional boilers.
Figure I shows catalytic combustion of fuel in a single stage. However,
to maintain a uniform temperature distribution in the direction of the gas flow
in the catalyst bed, combustion which introduces the fuel/air mixture in
multiple stages is required.
Promotion of R&D on a high-efficiency gas turbine by the Agency of
Industrial Science Technology of the Ministry of International Trade and
Industry began this year. CRIEPI is participating in this project by conducting
research on combined-cycle power plant systems and heat-resistant materials
and cool ing .systems.
Although the CRIEPI research on the application of catalytic combustion to
gas turbine combustors is separate from the MITI research project, R&D on a
low-NOx combustor is being advanced along the lines of the concept shown in
Fig. 2.
CRIEPI shaped its approach to research on the control of NOx for power
plant boilers employing catalytic combustion three years ago. Research wi I L
be promoted after the installation of experimental equipment this year. We are
aware that EPA is carrying out research with the same objective, Several of
the reports issued by EPA and the Acurex Corporation have been interesting and
instructive to us (References 2, 3, .4, 5).
CRIEPI has initiated research in the following three areas.
(I) Development and selection of catalysts
(2) NOx emission control by combustor modifications and manipulation of
operating conditions
(3) Cooling systems for catalyst bed temperature control
101
-------
DEVELOPMENT AND SELECTION OF CATALYSTS
Two types of catalyst are being investigated: catalyst for utility boilers
to be used at approximately IOOO°C and catalyst for the combustor of
high-efficiency gas turbines which can be used at 1300 - I400°C.
The catalyst for utility boilers must have sufficient catalyst life under
operating conditions which not only permit complete combustion of the fuel but
also minimize NOx formation. In addition, to improve the heat transfer effi-
ciency of the steam generator or steam superheater installed in the catalyst
bed, it is expedient to maintain a uniform temperature distribution in the
direction of the gas flow in the catalyst bed. For this purpose, it is necessary
to select a catalyst having moderate activity.
Since it is necessary to raise the inlet temperature of a gas turbine to
1300 - I400°C, the catalyst for a high-temperature gas turbine must be one
which can be used at these temperatures. Moreover, since it is desirable to
make the catalyst bed small in order to make the combustor compact, the
catalyst must have high activity.
CRIEPI has been conducting research on NOx control by catalytic reduction
for the past five years, and various types of catalyst have been investigated.
In applying these investigations to research on a catalyst for catalytic
combustion, the following items were studied in order to clarify the character-
istics required of a suitable catalyst.
o High operational temperature capability
o Combustion rate of fuels with each catalyst
o Catalyst life at high temperature
NOx EMISSION CONTROL BY COMBUSTOR MODIFICATIONS AND MANIPULATION OF OPERATING
CONDITIONS
NOx formed by combustion of fuels which do not contain nitrogen compounds,
such as LNG, NGL, and naphtha, is termed "thermal NOx," Formation of thermal
NOx increases as the combustion temperature is raised. In addition, the effects
of the fuel/air ratio and residence time in the combustion area are particularly
significant. If a fuel rich condition occurs due to an increase in the fuel/air
102
-------
ratio, CO and unburned hydrocarbons will form in the flame, resulting in a
reducive atmosphere. Consequently, formation of NOx will reduced. In staged
combustion, fuel rich combustion is performed in the first stage and complete
combustion can be performed by adding air in the second stage. A remarkable
reduction in NOx formation can thereby be achieved. Moreover, since the flame
zone is larger in staged combustion and the maximum flame temperature may be
lower, NOx formation is reduced. Flue gas recircuI at ion permits reduction of
the NOx concentration both by a lower combustion rate due to the reduced oxygen
concentration as w,el I as the lower combustion temperature resulting from the
addition of low-temperature gas. A remarkable reduction of the NOx concentration,
equivalent to that possible with conventional combustion, can be achieved by the
application of staged combustion and flue gas recircuI ation to catalytic combus-
tion.
CRIEPI plans to apply staged combustion and flue gas recirculation to
catalytic combustors with the objective of using this technology in both utility
boilers and gas turbine combustors.
COOLING SYSTEMS FOR CATALYST BED TEMPERATURE CONTROL
When catalytic combustion takes place at temperatures above IOOO°C, the
fuel combustion reaction proceeds not only on the surface of the catalyst but
also in the gas itself. In order to maintain the temperature of the catalyst
bed below the critical temperature of the catalyst, it is necessary to provide
for cooling of the catalyst bed.
For catalytic combustion applied to utility boilers, methods of cooling
the catalyst bed by the installation of water tubes or steam tubes in the bed
are being studied.
Thermal conduction in the catalyst bed is complex, comprising not only
conductive transfer between the catalyst and the metal tubes but also convective
and radiative heat transfer.
CRIEPI is conducting preparatory experiments with the objective of obtaining
basic data for the design of experimental equipment for catalytic combustion.
These experiments employ buHt-in cooling pipes and a device which circulates
103
-------
high-temperature flue gas in a duct filled with catalyst.
Since it is thought that a prototype catalytic combustor is required for
the precise investigation of a cooling system for the catalyst bed, this
research is concentrated mainly on an investigation of cooling systems which
maintain a uniform temperature of the experimental equipment when-oper-at-i-FH§-at
high temperatures. It is expected that the results of this investigation will
provide useful information for the next step of development.
Both the addition of excess air and water injection are being used to
lower the gas temperature at the inlet of gas turbines. However, both of these
measures are undesirable in that they degrade thermal efficiency.
Our research indicates that it is possible to reduce the combustion tem-
perature to 1300 - I400°C by flue gas reci rcu I at ion if the excess a,i r is reduced
to near the theoretical values. Further, it appears possible to reduce NOx by
means of a lower combustion temperature in combination with staged combustion.
Various problems in connection with the application of catalytic combustion
to boilers and gas turbines will be studied this year and next. In addition to
model experiments for utility boilers using small-scale catalytic combustion
experimental equipment with a built-in cooling system for the catalyst bed,
model experiments for gas turbine combustors using pressurized~combustion
experimental equipment are planned.
104
-------
REFERENCES
I. Nakayama, T, Catalytic Combustion Reaction. J, Japan Petroleum Institute,
19(3): 247-251, 1976
2. Thompson, R. E., Pershing, -Di:.W,, Berkau, E, E. Catalytic Combustion, A
Pol Iution-Free Means of Energy Conversion ? EPA 650/2-73-081, U.S.
Environmental Protection Agency, NTIS No,PB 223-0020, August 1973, 51 pp.
3. Kesselring, J. P. Catalytic Oxidation of Fuel for NOx Control from Area
Sources. EPA 600/2-76-037, U.S. Environmental Protection Agency, February
1976, 194 pp.
4. Martin, G. B. Evaluation of a Prototype Surface Combustion Furnace. In:
Proceedings of the Second Stationary Sources Combustion Symposium, Volume IK
New Orleans, LA, August 29 - September I, 1977, pp. 255 - 278
5. Kesselring, J- P., Krill, W. V-, Kendall, R. M. Design Criteria for
Stationary Sources Catalytic Combustors. In; Proceedings of the Second
Stationary Sources Combustion Symposium, Volume H, New Orleans, LA, August
29 - September I, 1977. pp. 193 - 228
105
-------
TABLE 1.. EMISSION STANDARDS FOR NITROGEN OXIDES FROM
EXISTING AND NEWLY INSTALLED FACILITIES
Kind of faci 1 ity
Boi ler
Gas firing
Coa 1 firing
Sol i d-f ue 1 f i ri ng
Oi l-tar f i ri ng
Other types
Meta l-heati ng
Oi l-heati ng
furnaces ^
Ceme nt- s i n te r i ng
furnaces
Coking coal furnaces
Note: *l (a) 100,
Flue
gases*1
(a)
(b)
(c)
(a)
(b)
(c)
(a)
(b)
(c)
(a)
(b)
(c)
(a)
(b)
(c)
(a)
(b)
(c)
(a)
(b)
(c)
(a)
(a)
000 normal
Standard values
Exi sti ng
faci 1 ities
(ppm)
130
130
150
750
750
750
600
600
600
280
280
280
230
I90*2
—
220*3
220*3
220*3
210
210
180
—
—
m3/h and over.
Newly insta 1 led
faci 1 ities
(ppm)
100
130
—
—
—
480
480
480
—
—
150
150
150
100
I50*3
I50*3
100
100
150
250
200
(b) 400,000 normal m3/!
less than 100,000 normal m3/h. (c) 100,000 normal m3/h to less
than 40,000 normal m3/h.
*2 Excluding those incorporated in (or under construction) sulfur
oxide treatment facilities.
*3 Heating furnaces for welded steel pipes are excluded.
*4 Excluding cracking furnaces and independent heating furnaces
for ammonia production.
106
-------
TABLE II. EXAMPLES OF CATALYTIC COMBUSTION IN OPERATION
Industry
Photogravure
printing
Photogravure
printing
Magnetic
printjng
Ch
Metal
printing
Synthetic
resins
Synthetic
resins
Fl ue .Gas n . , .
Components Concent rat, on
p as GS (ppm)
To 1 uene
c+u , IPA * 4. 30° - 1*000
Ethyl acetate '
Butyl acetate
Toluene 80Q _ , OQ
n-Hexane
Tol uene
SK ''20° - 2'500
lorine compounds
To 1 uene
XV|« 500 -,,500
Butanol
Methanol
Formalin l-,200 -.1,500
Phenol
Phenol
Forma 1 in
Toluene 1,200 - 1,500
Methanol
Acetone
Flow Rate
of Flue Gas
(Nm-Vmin)
30
50
60
30
1 00-1 30
100-130
350
Catalyst Bed
Temperature
Ini-et (°C)
270
305
270
270
200
200-250
250
Gas
Concentration
as C| (ppm)
7
10
8-10
7-10
1-3
10-20
3-4
-------
o
CO
Steam
Reheater
Stack
Super Heater
Economizer
Air Preheater
Light Fuel Oil
or Fuel Gas
Figure I Schematic Figure of Catalytic Combustion
Utility Boiler
-------
Air
Fuel
Compressor
, Catalyst
Combustor
I400°C
Gas Turbine
600°C
200°C
I30°C
Steam
Turbine
Drum
Stack
Electricity
Generator
Steam
Condenser
Air Preheater
300°c| I I20°C
Waste Heat Boiler
Figure 2 Schematic Figure of a High Efficiency Gas Turbine System Having
Low NOx Combustor
-------
#6 FUEL OIL CATALYTIC COMBUSTION
By
J. T. Pogson*
and
M. N. Mansour**
October 1978
*Project Manager, Energy & Environmental Division, Acurex
Corporation
**Project Manager, Research & Development Department,
Southern California Edison Company
111
-------
ABSTRACT
A test program to demonstrate the feasibility of burning #6 fuel oil
in a catalytic combustor has been completed. Sixteen hours of experimental
testing were accomplished, demonstrating that stable catalytic combustion
of heavy fuel oils is feasible. Test parameters included an air preheat
temperature of 800°F, air/fuel ratios from 11.6 to 15 and 25 to 45, bed
heat release rates from 47,500 to 116,675 Btu/hr, and bed temperatures
from 1950°F to 2310°F.
While the feasibility test program was successful, considerable design
and development testing is required to adapt the catalytic combustor to
existing systems. Improved fuel oil delivery and ignition systems and the
elimination of flashback potentials must be addressed. At the same time,
catalyst and bed materials resistant to thermal shock and capable of long
life operation at near stoichiometric air/fuel ratio without the use of
large quantities of diluent to moderate flame temperature must also be
developed.
113
-------
ACKNOWLEDGEMENTS
This paper describes the work accomplished under Contract V3257907
for the Southern California Edison Company. The authors are grateful to
the many individuals of the Acurex/Aerotherm staff who were of assistance
during this program. We would particuldrly like to thank Messrs. Ed Chu,
Scott Hagaman, Kerry Seifert, and Patrick Cawley for their operation of
the test facility and assistance in the data acquisition.
114
-------
INTRODUCTION
The primary purpose of the project described in this paper was to
demonstrate the feasibility for catalytic combustion of #6 fuel oil.
Catalytic combustion offers the potential for an alternate technology to
reduce NOX emissions from the nation's stationary sources. This technology,
while still in its research stage, has been recognized and demonstrated in
previous programs with gaseous and light distillate fuels oils, (1)*, (2)
and (3). Catalytic combustion of very low nitrogen content fuels produces
low NOV, CO, and unburned hydrocarbons due to its inherent temperature
A
control capability and a concurrent promotion of oxidation reactions.
Typically, CO and unburned hydrocarbon emissions are less than 5 ppm and
thermal NOV emissions are 10 ppm corrected to 3 percent excess oxygen.
A
Preliminary test results have also indicated that a reduction in the con-
version of fuel nitrogen to NO may be achieved in a catalytic combustor
A
when gaseous and light fuel oils are doped with nitrogen containing com-
pounds (3). Conversion rates of organic bound nitrogen to NO of less than
A
20 percent have been demonstrated with two-stage rich/lean catalytic burners.
The objective of the present program was to demonstate for a minimum
of 10 hours both rich and lean combustion of #6 fuel oil in a graded cell
catalytic combustor. During this investigation, preliminary data were
acquired to determine the combustor system performance and emissions. The
obtained test results were intended to provide a baseline data for further
development activities.
Prior to this project catalytic combustion testing had used gaseous
or light distillate oil fuels. These fuels require less complex fuel
*Numbers in parentheses designated references.
115
-------
delivery systems and permit the investigator to place his primary attention
on the study of the combustion process. Therefore, to examine the type of
experimental problems associated with catalytic combustion of #6^fuel oil,
several pre-project experiments were conducted. Specific objectives of
these pre-project experiments were:
1. Develop a #6 fuel oil delivery system that could supply the low
flow rates required for this test program
2. Select a fuel oil atomizing nozzle that would not plug or be car-
bonized by the catalytic combustor radiant heat load
3. Determine the potential for fuel oil build-up on the duct walls
upstream of the catalytic combustor
4. Determine the potential for flashbacks and flame holding due to
the presence of atomized fuel mist upstream of the combustor bed
5. Test and inspect the catalytic combustor for fuel oil fouling at
the combustion surfaces
6. Identify a safe light-off technique.
The initial experimentation was conducted during November 1977 in the Acurex/
EPA high pressure catalytic combustion facility. This facility was used only
during these pre-project tests because the Acurex-owned atmospheric test
facility had not yet been completed.
DESCRIPTION OF HARDWARE
Catalytic Combustor
The demonstration of #6 fuel oil was conducted in a small-scale Acurex
experimental test facility. The operational portion of the test facility is
shown in Figure 1. The entire combustion apparatus is enclosed within a
fume hood to keep all combustion products from the surrounding room. An
overview of the lab is shown in Figure 2. Premixed fuel/oxidizer flows ver-
tically downward through an expansion chamber and is encompassed by a chamber
heater (Figure 3) that maintains the temperature of the gases entering the
system at the desired level.
116
-------
The gaseous fuel and oxidizer flow is monitored through an external
control panel shown in Figure 4. Rotameters are used to set gas flow to
the desired rates. Gas pressures are both regulated and monitored from
the front panel.
A 36-kilowatt heater is installed on the oxidizer flow line. This
heater is designed to allow the mixture of fuel and oxidizer to enter the
catalyst bed at temperatures up to 1000°F. The temperatures of both the
oxidizer heater and the maintaining heater around the expansion chamber
are controlled with solid state controllers. These temperatures are read
out with the heater controllers on the front panel.
Twenty-four channels of thermocouple readout are available for instru-
menting various combustion systems. The thermocouples (Type K) are fed
from the interior of the combustion chamber fume hood through a heavy duty
thermocouple junction box (Figure 1, No. 11). These thermocouple readouts
are shown in Figure 4 at the bottom of the right-hand console.
Fuel and oxidizer are mixed downstream of the oxidizer heater immedi-
ately before entering the expansion chamber. Flame guards are placed in
the fuel line to prevent accidental flashbacks. A rupture disk is located
in the top of the expansion chamber to allow venting should a major over-
pressure occur.
Catalyst beds are placed within a quartz reaction chamber (Figure 1,
No. 3) so that a window could be cut through the surrounding insulation to
View the catalytic reactor during the experiments. Combustion gases move
vertically downwards, enter the catalytic reactor, and after reaction enter
a watercooled manifold where they make two 90° turns and are exhausted into
the top of the fume hood.
A water-cooled sampling probe is available to monitor combustion pro-
ducts. This probe extracts a gas sample from the combustion gas flow and
passes it into^a heated sample line where it is transmitted to the envissions
bench. The emissions bench consists of instrumentation capable of analyzing
the concentrations of NO, NOX, CO, C02» 02 and unburned hydrocarbons in the
exhaust system utilized in the facility. Figure 5 illustrates the
emissions sampling system utilized in the facility.
117
-------
At the bottom of the water-cooled manifold there is a viewing port
attached to a standoff tube. Through this viewing port, an observer can
readily see the catalytic combustor along its axis during operation. The
viewing port also allows an optical pyrometer to be used to determine bed
temperature.
Fuel Delivery Systems
Since the control panel associated with the Atmospheric Catalytic Com-
bustion Facility is capable of providing only gaseous fuel control, and the
program conducted for SCE used #6 fuel oil, a separate fuel delivery system
was required to introduce the liquid fuel into the combustion chamber.
Figure 6 is a schematic diagram of the fuel delivery system that was fabri-
cated to deliver #6 fuel oil into the preheated oxidizer gases in the ex-
pansion chamber of the facility. A 55-gallon drum of oil served as a
reservoir. This drum was heated electrically by the use of both blanket
heaters and an immersion heater. These heaters were able to provide close
temperature control to 200°F, the temperature at which most tests were
conducted.
To obtain the degree of fuel atomization required, a gas atomized
fuel nozzle was used during the test program. As shown in Figure 6, the
atomizing gas line was passed through the drum of oil to be heated to about
the same temperature as the oil it was to atomize. This was done to pre-
vent potential oil viscosity increases due to an abrupt temperature drop
when the oil came in contact with the atomizing gas as it left the nozzle.
Fuel flow was measured by a meter designed to transmit a pulse with each
revolution, thus signaling the passage of a discrete volume of liquid.
A pulsed digitizer was attached to the output of the flow meter, and the
flow rate was determined by averaging the number of pulses over a given
period of time, using the calibration factor of the meter. A calibration
was conducted on this meter using #6 fuel oil under temperatures and
pressures that were encountered during the tests and the volume per pulse
was measured to be 5.1 cm3.
118
-------
RESULTS AND DISCUSSION
Thirty-two tests were conducted under this project. Air/fuel ratios
ranged from 11.6 to 15 and from 25 to 45, with heat release rates from
47,500 to 116,700 Btu/hr, and air preheat held nearly constant at approxi-
mately 800°F. Fuel oil was atomized with air, nitrogen, and argon (1 to
1.4 scfm) during the various tests. A total of 16 hours of catalytic com-
bustion demonstration tests were performed.
During the tests, the combustor ceramic surface temperatures varied
from 1950°F to 2310°F. Temperatures were limited to less than 2400°F by
the quartz enclosure. Temperature control was achieved by the use of
either nitrogen or argon diluent to moderate the flame temperature. Diluent
volumetric flow rates ranged from zero to twice the air flow rate, depending
upon the air/fuel mixutre. As has been shown in previous studies, catalytic
combustion at temperatures below 2300°F does not produce significant thermal
NOV. Therefore, the data was examined without parameterizing the combustor
A
bed temperatures to quantify its affect on NOV formation.
A
Four different fuel delivery set-ups were used during the test series.
Test facility modifications were made following each test series to improve
combustion stability. The initial configuration utilized a venturi section
with a fuel nozzle mounted approximately 8 inches above the catalyst bed.
The oil entrance port was oriented horizontally in the venturi restriction,
and an adapter/holder for a Delavan nozzle was used to effect a 90° turn.
Tests conducted on this configuration were not stable. Build up of car-
bonaceous and gummy deposits were experienced on the surface of the catalyst
bed, and a significant amount of unburned fuel droplets were noted passing
through the bed. Additionally, flashbacks to the tip of the nozzle occurred
with such frequency that useful data could not be obtained.
Observation of the combustion process indicated that there was insuffi-
cient distance from the nozzle to the catalyst bed for significant vaporiza-
tion of the oil. Initial calculations had indicated that a distance of
only 4 inches should have been sufficient; however, these calculations were
based on idealized conditions that were not being met by the configuration.
119
-------
The distance between the fuel nozzle and the catalytic combustor was
then increased. This was accomplished by positioning the nozzle within
two coaxial tubes. The largest diameter of the outer tube was three-quarters
of an inch, which allowed it to fit down from the top of the expansion cham-
ber- For these tests, the nozzle was positioned about 14 inches from the
entrance to the catalyst bed. Operationally, this configuration signifi-
cantly improved the performance of the system. Although flashbacks still
occurred, the frequency of these occurrences was reduced to the extent that
data could be easily collected between the flashbacks. A great number of
ash particles were generated during this experiment. Subsequent analysis
showed that the ash particles were approximately 95 percent carbon or
hydrocarbon.
This set of experiments was discontinued when severe catalyst bed
degradation was noted. Post-test observation of the catalyst bed indicated
that the zircon mullite section had partially melted and the exterior of
the fuel delivery nozzle was coated with oil and carbon. This buildup on
the exterior of the nozzle indicated that eddies were being formed around
the end of the nozzle/fuel delivery tube. Material was being swept up into
these eddies and deposited on the nozzle. Configuration modifications to
Prevent a recurrence of this problem were made. To overcome buildup on the
outside of the atomization nozzle, combustion air was forced over the sur-
face by fabricating a flow director that attached around the nozzle. For a
test series with this modification, the nozzle was placed 18 inches from
the entrance to the catalyst bed. This location allowed a spacing of 1/2
inch between the edge of the flow director and the walls of the conical
expansion chamber.
Performance of the system was dramatically improved after the addition
of the flow director section to the nozzle. Although flashbacks did occur
occasionally, they were much less intense than previous flashbacks. The
majority of the data taken on the lean side utilized this configuration and
catalyst. Post-test examination of the catalyst indicated that the catalyst
was in good shape and that there was no fouling due to oil or ash.
120
-------
The next test series was directed towards rich combustion in a catalytic
bed. Combustion under lean conditions allows high volumes of air to flow
through the system. Under rich conditions, however, the amount of air flow-
ing through the system is reduced. It was found that with the nozzle and
flow director placed at the location 18 inches from the entrance to the bed,
there was insufficient air flow forced through the flow director. Flash-
backs again became a problem. Additionally, the presence of the lead-out
wires from the thermocouples in the slowly moving, highly loaded rich mix-
ture caused the formation of bridges of carbonaceous material across the
wires. This bridging became so severe that the entire 90 mm diameter of
the quartz reactor was covered. The combustion taking place in the bed was
extremely irregular and unstable. The experiment ended with no data collected.
To remedy these problems, the entire nozzle assembly was moved further into
the conical expansion chamber to direct the majority of the air flow across
the surface of the nozzle. This was accomplished by raising the fuel de-
livery assembly to 22 inches from the entrance to the catalyst bed. At this
location, there was an annul us of 1/4 inch between the walls of the expan-
sion chamber and the flow director.
Hith this modification, the rich combustion tests were run with little
difficulty. The system behaved in a stable manner with few flashbacks. Sub-
sequent examination of the system after the rich combustion test series
showed that there was negligible fouling of the walls, or nozzle. Figures
7 and 8 show the catalyst after completing the last series of tests. There
also was no visual evidence of oil or ash fouling.
The NOV levels and nitrogen conversion to NOV obtained during the test
A A
program are shown in Figure 9 as a function of the air/fuel ratio. As
indicated in Figure 9, NO emission levels, increased from a value of
A
208 ppm at an air/fuel ratio of 11.8 (A/F = 14 at <|> = 1) to 1099 ppm at
an air/fuel ratio of 30. This corresponded to nitrogen conversion effi-
ciencies of 18 percent and 95 percent, respectively, at the aforementioned
air/fuel ratios. Due to the low-catalytic burner flame temperature all
measured NOX was considered to be fuel-nitrogen generated.
121
-------
The data shown in Figure 9 illustrate that during combustion of rich
air/fuel ratios, the conversion of fuel nitrogen is low, while during lean
combustion the catalytic burner approaches 100 percent conversion of fuel
nitrogen to NOX- This conversion rate is much greater than that experi-
enced with conventional atomized fuel burners. A possible explanation of
this difference may be the fuel stratification which occurs in conventional
burners, leading to local rich combustion zones and hence lower nitrogen
conversion rates to NO .
A
Also shown in Figure 9 are four data points taken with a separated
combustor, which occurred when the zircon mullite test section melted. As
illustrated, these correlated data correspond to NOV levels in the 50 to
A
100 ppm range, or an order of magnitude lower than that attained with the
nonseparated combustor. Hhile several reasons may be advanced to explain
this observation, it was felt that the gas phase reduction of NOV occurred
A
as a result of the bed separation. It was postulated that as the air/fuel
mixture passes through the initial section of the combustor there is a
partial oxidation of the fuel and a corresponding increase in the mixture
temperature and the formation of NO precursors (NH9, HCN, NO). In the sub-
A £
sequent open volume a partial homogeneous reaction occurs at a temperature
in the 1000°F to 1500°F range which allows the self-destruction of the NOY
A
precursors to Ng. Thus, in the remaining section of the catalytic bed,
most of the fuel nitrogen has already been destroyed and thus cannot be
converted to NO ,
A
Figure 10 compares the corrected C02 measurements with those predicted
from theoretical calculations. The data have been corrected to zero diluent.
As shown, fair agreement exists between the data and theory.
The calculated adiabatic flame temperature is compared to the measured
thermocouple readings in Figure 11. The measured temperatures were corrected
to zero diluent for comparison purposes. Again, the corrected data agree
with the predicted temperatures, which indicates that the performance of
the combustor system was satisfactory and that the data obtained are
reasonable.
Attempts were not made to achieve ignition of heavy fuel oil-air
mixtures without the use of gaseous fuels. This decision was made when
previous experimentation showed that when the bed temperature dropped
122
-------
below 1800°F the combustor appeared to extinguish itself. That is, the
combustor temperature would rapidly drop to the preheat temperature and
smoke emissions were visible from the stack. Thus, it was concluded that
the 75Q°F to 800°F preheat would not vaporize the fuel oil sufficiently to
permit lightoff.
Table I contains the corrected data for all the data points shown in the
previous figures. The NO and CO data are corrected to 3 percent excess Og.
Data for ^» C^' anc* temPerature are corrected to zero diluent. Air/fuel
and heat release values are calculated from the measured rotometer and fuel
flowrate data.
CONCLUSIONS
The following conclusions may be drawn from the experiments and test
data.
Stable catalytic combustion of heavy fuel oils is feasible using
a gaseous fuel lightoff technique and was demonstrated by the
series of tests covering 16 hours of experiments.
Significant vaporization is attained during lean operation with an
850°F preheated air/diluent mixture to provide continuous catalytic
combustion of heavy fuel oils without the deposition of significant
unburned fuel or ash deposits on the burner.
Additional design and development of the catalytic burner is
required to eliminate unstable combustor operation. Flashbacks
occurred when the average velocity of the air/diluent mixture was
less than 8 to 9 feet per second.
Stable burner operation was achieved at air/fuel ratios ranging
from 11.6 to 15 and 25 to 45 with diluent to moderate the bed
temperature. Even at the lower air/fuel ratios, > 1.0, visible
smoke emissions were not detected.
Air/fuel ratios below 15 resulted in low conversion rates of the
fuel bound nitrogen to NOV.
A
Catalytic lightoff could not be attained at the lower air/fuel
ratios (A/F = 11.6 to 15) because of insufficient fuel vaporiza-
tion and the tendency for flashbacks to occur.
123
-------
• Separation of the catalytic combustion support elements, which
occurred when the zircon mullite element partially melted, appears
to offer a means of reducing nitrogen conversion rates by an order
of magnitude below those attained with the standard close packed
graded cell combustor.
RECOMMENDATIONS
It is apparent from the tests that additional development efforts are
required to provide a reliable heavy fuel oil catalytic burner. One of
these activities must be concerned with the development of a higher thermal
shock resistant support material.
Other required development efforts are:
• The determination of a fuel atomization system which will require
less atomization fluid and permit sufficient vaporization to allow
more stable operation at rich conditions. A steam atomization
nozzle or a fuel oil prevaporizer may provide such a system.
t Development of a positive heavy fuel oil ignition system to reduce
the dependence upon high preheat temperatures.
• Testing of a separated bed combustor to explore the potential for
reduced fuel bound nitrogen conversion to NO .
A
• Development of catalytic burner systems which do not require large
volumetric flow rates of diluent for temperature control purposes.
• Combined burner system life testing on a subscale combustion rig
to establish system reliability and heavy fuel oil contaminate
effects.
• Further system characterization to provide scale-up data for
greater heat release rates.
124
-------
REFERENCES
1. Roessler, W. U., et al., "Investigation of Surface Combustion Concepts
for NOX Control in Utility Boilers and Stationary Gas Turbines," Environ-
mental Protection Technology Series Report EPA-650/2-73-014, August 1973.
2. Kesselring, J. P., et al,,"Catalytic Oxidation of Fuels for NOX Control
from Area Sources," Environmental Protection Technology Series Report
EPA-600/2/76-037, February 1976.
3. Kesselring, J. P., et al., "Design Criteria for Stationary Source Catalytic
Combustion Systems," Acurex Final Report 78-278, March 1978.
125
-------
TABLE I. DATA CORRECTED TO 3 PERCENT EXCESS 02
Data PPM NO UHC Corrected
Pt. C02 °2 CO PPM PPM A/F Temp
No. 0% Diluent 0% Diluent 3* Excess 02 3% Excess 02 3% Excess 02 14. Q °F
3-1
3-2
3-3
3-5
3-6
3-7
3-8
3-9
2-1
2-2
2-3
2-4
2-5
2-6
2-7
2-8
2-9
2-JO
2-11
2-12
2-13
2-14
2-15
2-16
2-17
2-18
I-I
1-2
1-3
1-4
1-5
1-6
13.6%
16.6%
12.1%
—
14.7%
10.2%
11.7%
7.7%
5.0%
5.4%
4.9%
6.7%
4.4%
4.8%
4.5%
4.9%
4.9%
4.6%
5.5%
5.8%
4.8%
5.0%
5.9%
6.4%
5.1%
6.2%
7.0%
7.8%
7.7%
7.5%
7.0%
8.0%
0.2%
0%
0.6%
0.03%
0%
9.1%
0.9%
10.4%
14.0%
13.2%
14.1%
12.6%
15.3%
14.7%
15.5%
14.9%
15.0%
15.7%
14.0%
13.3%
14.7%
14. 3%
13.8%
13.2%
14.7%
14.1%
16.0%
11.6%
12.6%
13.0%
13.5%
13.0%
1360
0
--
--
2708
0
—
0
54
47
94
54
79
89
98
92
72
122
67
63
57
57
62
61
115
92
Unstable
64
70.4
90
113
113
208
221
194
473
596
749
212
776
797
823
907
894
884
890
906
957
969
977
978
935
943
1014
1011
1015
956
1099
Unstable
1052
32
157.5
156
126
—
--
—
—
-
-
—
-
-
-
-
-
-
-
--
—
-
-
-
-
—
-
—
.
—
--
--
--
0
0
0
0
1.03
1.0
0.88
0.94
1.35
1.8
0.83
2.1
2.95
3.05
2.63
2.2
3.2
2.8
2.2
2.5
2.6
2.9
2.5
2.5
3.1
2.8
2.4
2.2
2.6
2.0
2.5
2.2
2.3
2.5
2.6
2.1
49275
49275
49275
49275
49275
49275
49275
49275
109217
111232
101887
118562.
69452
64504
54334
47516
48760
44041
47473
104669
107528
84221
84936
849,36
47473
47759
116675
101374
82078
82500
80264
64504
3906
4075
3580
4266
3614
2838
3297
2900
2080
NC
NC
NC
1983
NC
NC
NC
NC
NC
NC
NC
2089
2176
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
Avg
VEL
Ft/Sec
14.0
12.3
8.6
12.0
13.6
13.3
7.7
15.1
31.6'
33.9
29.5
29.9
22.2
18.2
13.1
13.7
14.9
14.9
15.4
27.6
33.3
23.2
23.3
22.9
13.8
13.2
28.5
24.1
19.9
20.6
20.3
13.6
— meter inoperable
NC not calculated
126
-------
no
-vl
Figure 1. Atmospheric catalytic combustion facility; view of reactor showing
catalyst bed in place within quartz tube.
-------
-B
ro
oo
IB-IB
Figure 2. Layout drawing of atmospheric catalytic combustion facility.
-------
ro
to
Figure 3. Expansion chamber and other features of the catalytic combustion facility.
-------
Figure 4. Control panel for atmospheric catalytic combustion facility.
130
-------
Catalytic
Reactor
Water
Cooled
Sampling
Probe
Heated sample line
Filters
CO
co2
vS
NDIR ±
o
N
Chemiluminescent
•G*-l
Pumps Cold
Filters
NO/NO,
Figure 5. Schematic diagram of emission analysis equipment used
with atmospheric catalytic combustion facility.
-------
Hand
Valve
Pressure
Guage
co
ro
Rotameter
Air Flow
Guage
Delevan Nozzle
and Holder
55 Gal #6 Fuel Oil
Figure 6. No. 6 fuel oil delivery system.
-------
o
I
Figure 7. Post-test photograph of catalyst.
133
-------
(J
00
6^C:
Figure 8. Expanded post-test photograph of catalyst.
-------
CJ
01
CM
o
1200
1100
1000
900
800
700
600
OJ
i.
o 500
E 400
O-
§ 300
200
100
100
90
80
70
60
50
40
30
20
10
7
'100% conversion
of fuel nitrogen
separated bed data
CM
i
i I
8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Air-Fuel Ratio
Figure 9. NO and nitrogen conversion -- 3 percent excess
-------
CM
o
18
16
14
12
o>
o
en 10
O-
6
4
Excess Diluent
Theoretical CO,
i I
j I
10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44
Air-Fuel Ratio
Figure 10. Comparison of corrected C0? measurements with theory.
-------
01
a.
i-1
u>
ID
-O
10
5000
4000
3000
2000
1000
s
TOO
200
300
400
Percent Theoretical Air
Figure 11. Comparison of combustion product temperature
with the adiabatic flame temperature.
-------
STRUCTURAL ANALYSIS OF A PRELIMINARY CATALYTIC CERAMIC DESIGN
By:
S.M. DeCorso
D.E. Carl
Combustion Turbine Systems Division,
Westinghouse Electric Corporation
Lester Branch Box 9175, A-603
Philadelphia, Pa. 19113
139
-------
ABSTRACT
Structural analysis of a preliminary design catalytic ceramic element and
its support structure for ultimate use in a low emission application has been
performed. A thermo-mechanical analysis of the major components in this con-
ceptual design has been performed for both steady-state and transient (shut-
down) situations. Consequently, an arrangement that is thought to be a viable
catalytic ceramic combustor element has been identified. Further development
is necessary before full scale application.
141
-------
ACKNOWLEDGMENT
This work was performed with the support of the Electric Power Research
Institute under Contract No. RP 421-1 with Dr. A. Cohn as Program Manager.
The authors wish to thank R. G. Glenn, E. W. Tobery and W. S. Y. Hung for
their contributions in performing this task.
142
-------
INTRODUCTION
In the development of a catalytic combustor for large industrial combus-
tion turbine applications, the problems relating to the structural design of
the full size catalytic element and its support structures have to be identi-
fied and solved. At this stage of the development, it is necessary to gener-
ate a conceptual design of the full size catalytic element and its supporting
metal structure. Then thermal, mechanical, and stress analyses must be made
to determine its expected structural performance in the combustion turbine
transient and steady-state operating environments.
The objectives of this study are to establish a conceptual reference
design arrangement for a preliminary catalytic ceramic combustor element to
be used in large industrial combustion turbines for low emission application
and to evaluate the catalytic reactor structural performance through thermal
and mechanical analyses of the catalyst ceramic elements and supporting metal
structures over the operating range of the combustion turbine.
CONCEPTUAL DESIGN
Total Combustion System
Based on the small scale test experience in the past (Reference 1), a
conceptual design arrangement of a catalytic combustor in an industrial com-
bustion turbine has been envisioned as shown in Figure 1. Adequate mixing in
the fuel preparation zone should be provided to produce a faily uniform fuel-
air mixture prior to entering the catalytic element. The catalytic element
should be sized to provide sufficient contact time for good combustion effi-
ciency and a length short enough to minimize the pressure drop at an acceptable
level. Preliminary analyses have indicated that the diameter of the catalytic
ceramic element will be larger than that of the conventional combustor currently
in production. Downstream of the catalytic element, the reacted products are
mixed with dilution air in the "transition section" which leads to the turbine
inlet.
Catalytic Ceramic Element
From a preliminary analysis of the available information on contact time,
reference velocity, pressure loss, combustor operating conditions and space
143
-------
limitation, including a consideration of the structural integrity, a 7 inch
(178 mm) long by 15 inch (381 mm) diameter ceramic substrate has been selected
for the present study.
The substrate is to be supported by a metallic cylinder. Preliminary
analyses have shown that, if unprotected, the metal cylinder will reach tem-
peratures higher than those allowed for materials currently used. Hence, an
intermediate layer between the substrate and the support metal cylinder is
required to provide cooling and/or insulation. Further analysis showed that
the difference in thermal growth between the ceramic substrate and the metallic
cylinder was sufficiently large that some provision must be included in the
design to absorb these movements and to hold the substrate.
The recommended conceptual design utilizes an air-cooled coiled wire to
provide the flexibility and conformity as shown in Figure 2. The coil is com-
pressed in assembly. A layer of castable ceramic on the outer cylindrical
surface of the substrate is required to provide a smooth surface for contact
with the coil and to act as a thermal barrier. In the design, two outer
layers of cells in the substrate has been made inactive to provide the desired
temperature gradient and structural stiffness.
THERMO-MECHANICAL ANALYSIS
The thermo-mechanical analysis was performed for both the steady-state
and the transient conditions. One dimensional analogs are utilized in the
preliminary steady-state analysis to allow a quick evaluation of the mechani-
cal performance of the catalytic element. The analysis was extended to con-
sider the transient condition in order to determine the effects of different
loadings on the structural components over the entire operating cycle. The
effects of two dimensionality was introduced in the transient solution which
also included the steady-state case.
Steady-State Analysis
In performing the analysis, the catalytic element is divided into its
three components: the ceramic substrate, its metal support cylinder and an
intermediate compliant layer between them.
144
-------
Substrate—
Since the conceptual design of the total combustion system is still pre-
liminary at this stage, it is sufficient to select a typical material for
analysis. A survey shows that cordierite (2 MgO2Al20 -5 SiO.) is the sub-
strate material used in catalytic converters on automobiles. Most of its
material properties are readily available. On this basis, it was selected as
the substrate material in the present study.
The ceramic material when used as a support for catalyst takes the form
of cellular tubes. It follows that the bulk mechanical properties will depend
on such features as cell shape, cell size, cell wall thickness and manufactur-
ing techniques in addition to the mechanical properties due to the pure mate-
rial. For instance, depending upon cell count, proportions and load direction,
the allowable compressive loading on one square inch of bulk surface of the
substrate varies considerably (2).
Like most ceramics, cordierite has the capability to withstand thermal
shock mainly due to its very low coefficient of thermal expansion (Reference 2)
and a low modulus of elasticity, 4 - 18 x 10 psi (28 - 124 G Pa) depending upon
porosity, (3). According to Ref. (2), it can also withstand continuous duties
at temperatures of 1350°C (2462°F) and transients as high as 1500°C (2738°F).
For the present analysis, a uniform temperature of 750°F at the inlet and
a nominal temperature of 2300°F (1260 C) at the exit of the reactor are
assumed. The temperature rise from inlet to discharge is considered linear.
An analysis of the thermal expansions showed that the temperature gradi-
ent will cause a thermal distortion to occur in the substrate disk. Hence, it
is important that no axial restraints be applied across the face of the disk.
The pressure drop across the substrate is influenced by the size and
shape of the cells and the wall thickness used.
For a typical substrate structure such as that used in (Reference 1), the
pressure loading is well within the allowable limit.
Support Cylinder—
The metal support system of a catalytic combustor will experience similar
pressure and thermal loadings as the secondary dilution zone of a conventional
145
-------
turbine combustor. It is convenient, therefore, to select the material used
for the conventional combustor as the material for the support system. There
has been extensive experience in the use of Hastelloy X, a nickel base alloy,
as the material for combustor components. For the purpose of the present
analysis, Hastelloy X is selected as the candidate material for the metal sup-
port system. Its mechanical properties are readily available from Ref. (4).
The metals are film cooled by air at the compressor discharge temperature.
In the design, adequate film cooling will be provided such that the catalytic
combustor shell can be expected to perform its duty, namely, withstand the
external loads caused by gravity, airflow, fluid pressure and temperature.
The fundamental stress in the cylindrical elements is the hoop stress due
to the pressure differential across the wall. The maximum pressure loading is
associated with the pressure drop due to flow losses in the combustor.
Assuming a pressure loss of 5%, the calculated hoop stress for a 0.050 in
(1.27 mm) thick wall is at a reasonable stress level. A thicker wall can be
used to reduce this stress, if experience deems it necessary.
The cylinder will experience dimensional changes resulting from tempera-
ture changes. The differential thermal growth between the cylinder and the
substrate will likely be the parameter that defines the design. These differ-
entials must be accommodated if high stresses are to be avoided. The accommo-
dations for these relative movements, either completely or partially, suggest
an intermediate layer of high compliancy.
Compliant Layer—
Its primary function is to absorb the relative movements across the layer
between the cylinder and the substrate. An additional requirement is that it
must distribute the restraints it applies to the substrate uniformly and
lightly. This requirement stems from the allowable loading of the substrate,
particularly in transverse directions (2).
One promising scheme is to use a metallic mesh layer as possible support
of the substrate (see Figure 3). With cooling flow through the wire mesh, the
metal temperatures will be low enough for the coils to survive. The wire
mesh, of proper wire size and density, could provide flexibility, good
146
-------
capability for conformity and a large number of support points. Therefore, a
metallic coiled wire compliant layer will be capable of following the differ-
ential growths, conforming to the surface of the substrate, having a multitude
of contact points, resisting shearing and operating at high temperatures.
Transient Analysis
In reviewing the operating cycle of a combustion turbine, it has been
determined that the most severe transient loading occurs during the shut-down
process. Hence, the present analysis is aimed at evaluating the stresses that
can be developed during this critical period.
A one-dimensional time dependent analysis of the heat transfer at various
stations along the combustor was performed. The change in pressure with time
after shut-down was determined based on actual engine data. These calculated
temperatures and pressures as a function of time after shut-down are then the
input to an existing computer code to perform the stress analysis. In this
way, a time history of stresses was developed with temperatures along the
combustor at simulated operating conditions at various times following shut-
off of 'the device.
Thermal Analysis of Support Cylinder—
The metal support system, must be kept at temperature levels which will
result in acceptable life. This temperature was taken to be 1400°F (760°C).
The steady-state temperatures at various stations along the entire metal
support structure were calculated. Starting at these levels, transient tem-
peratures for the various stations were calculated to simulate the shut-down
process. Actual engine data were used to determine airflows and combustor
shell temperatures as a function of time. This information was then used to
calculate internal convective heat transfer coefficients as a function of
time. The heat transfer coefficients and air temperature were then used to
determine the metal temperature at the various stations through the use of an
existing transient heat transfer computer code.
Stress Analysis of Support Cylinder—
In order to perform a finite element stress analysis of the metal support
cylinder, the structure has been divided into five elements with each element
further subdivided into nodes as shown in Figure 4.
147,
-------
The transient stress response at three selected locations are plotted as
shown in Figures 5-7. The typical decay of stresses in an initially low
stress region is illustrated in Figure 5. In all transient stress plots, it
was observed that all initial high stresses tend to decay rapidly to low val-
ues with the cool-down as illustrated by the hoop stresses shown in Figure 6.
The only stresses that show a temporary increase are those in shell elements
C and D. The meridional stresses at node 9, element C increase for about
30 seconds and then start a gradual reduction. The inner hoop stress at node
element D increases for about 5 seconds and then decreases in a typical manner
as shown in Figure 7. The maximum value reached is less than 30,000 psi
(207 MPa) which is low in comparison with the steady-state hoop stresses.
Substrate—
In comparison with the thin walled, flexible shell, the substrate is a
three-dimensional device, which can be considered a solid in some aspects.
The most refined elements of a finite element program, namely iso-
parametric three dimensional solid elements, are recommended for use in the
calculation of stresses in bulky structures such as the substrate. There is
favorable experience with these elements, but their limitations have to be
recognized. The first restraint is associated with modelling which involves
a design of the elements to represent the substrate. Since the substrate is
supported along the outer surface by the compliant layer, a set of reaction
nodes (discrete load points along the outer edges of the outer elements by
which loading and/or attachment to adjacent elements is secured) must be spec-
ified. The calculation procedure and the characteristics of this element are
such that when the reactions at these nodes are developed during the calcula-
tion, the stress fields are inaccurate through two or more layers of elements
in the model. This makes it necessary to use a larger number of elements and
layers. Also, the representation of the contact of the compliant layer with
the substrate makes a large number of surface nodes desirable.
148
-------
In addition, a problem exists in defining the properties of the substrate;
usually, when material properties are quoted, such as Young's modulus, density,
thermal coefficients of expansion, etc, the values do not recognize the exis-
tence of the details, such as holes and cell structure. The results of the
stress and other mechanical calculations, therefore, are bulk stresses and
deflections, good enough to see if the substrate will fit the space, or be
crushed due to the distributed loading on the surface, but not fine enough to
determine stresses through or along cell walls. With brittle material, it is
these small, local stress details that can cause failure much as flaw details
in metallic parts affect performance. Thus, modelling of the cellular con-
struction may be required.
The stress analyses presented thus far have beeni somewhat simplified.
For example, only the pressure loading on the outer contours of the substrate
has been considered.
While the substrate is simple in shape and some simple calculations indi-
cate acceptable performance, the device, when considered in more detail,
rapidly becomes very complex for the available methods of handling it.
Because of these limitations, the future designs can be better judged
only after operational and test procedures are applied.
In summary, detailed stress analysis of a substrate/compliant layer/shell
system under the expected thermal and mechanical load situations in a catalytic
combustor will be an extremely time consuming task. Consequently, the most
cost-effective approach appears to be the testing of the concept described in
the paper and the use of the sophisticated finite element methods to investi-
gate the design after the tests provide the necessary thermal and flow input.
From the present study, the following conclusions have been reached:
(1) This preliminary design as currently conceived is expected to per-
form its functions from the mechanical standpoint. Considerable
design & development will be required before full scale application
is possible.
(2) At this stage, additional experimental input is needed so that more
sophisticated computer analysis can be applied for future designs.
(3) The analyses suggested that a reasonable amount of experimental work
in conjunction with analytical methods discussed in the report will
provide the necessary design information.
149
-------
REFERENCES
1. DeCorso, S.M. , et al, "Catalysts for Gas Turbine Combustors— Experi-
mental Test Results", ASME Paper No. 76-GT-4, March 1976.
2. Anon., "CELCOR Cellular Ceramics", Corning Glass Works, Corning, N.Y.
3. Boyer, W.A., Corning Glass Works, private communication, December, 1977.
4. Anon., "Hastealloy Alloy X", Haynes Stellite Co., Kokomo, Indiana,
August, 1961.
150
-------
PRIMARY AIR
en
CATALYTIC CERAMIC ELEMENT
MULTIPLE FUEL NOZZLES
CONVENTIONAL FUEL NOZZLE
TRANSITION
SECTION
FUEL PREPARATION ZONE
Figure 1. A Conceptual Design Arrangement of a Catalytic Ceramic Element in an Industrial Combustion
Turbine
-------
SUBSTRATE RETAINER
COOLING FLOW!
AIR FILM
COOLING FLOW
PROCESS
FLOW
Figure 2. A Conceptual Design with Film Cooling of Metal Support Structure
152
-------
COMPRESSOR DISCHARGE
AIR FLOW
SUBSTRATE RETAINER ,
COOLING FLOW f
FLOW
PROCESS
FLOW
Figure 3. A Conceptual Coil Wire Design With Cooling
-M*-
ch»
so
38
48
FLOW DIRECTION
CATALYTIC ELEMENT
CENTERLINE
METAL CYLINDER
WITH 0.050 (1.27 mm)
THICK WALL
25
Figure 4. Catalytic Combustor Support Cylinder Model for Finite
Element Analysis
153
-------
-H*-
METAL CYLINDER
WITH 0.050 (1.27 mm)
THICK WALL
(MPa)
0 -
•1 -
-2 —
- -3 —
-4 -
-5 -
-6 -
•7 -
ut
K
-800
-1,000
O MERIDIONAL (INNER & OUTER)
A HOOP (INNER & OUTER)
10
20
30
TIME (SEC.)
40
50
25
J
60
Figure 5. Transient Stress Response - Shell Element A, Node 1
154
-------
FLOW DIRECTION
CATALYTIC ELEMENT
CENTERLINE
METAL CYLINDER
WITH 0.060 (1.27 mm)
THICK WALL
25
O MERIDIONAL
A HOOP
INNER
- OUTER
•600 -
-700 -1
I
I
10
50
20 30 40
TIME (SEC.)
Figure 6. Transient Stress Response - Shell Element C, Node 9
60
155
-------
-»+*-
50
JF
38
FLOW DIRECTION
CATALYTIC ELEMENT
CENTERLINE
METAL CYLINDER
/WITH 0.050 (1.27 mm)
THICK WALL
(MPa) (PSD
300 —i
60
00
HI
oc
co
200 -J
100 —
0 -
-100 —
-200 -
-300 —'
40,000 f—
20,000 —j
O MERIDIONAL
A HOOP
-20,000
-40,000
?
/
r
i
OUTER
1 1 1 1 1
10
20
30
TIME (SEC.)
40
50
60
Figure 7. Transient Stress Response - Shell Element D, Node 44
156
-------
DETERMINATION OF THE RATE OF HETEROGENEOUS REACTION
ON CATALYTIC SURFACES
Effects of Surface and Fuel on Observed Performance
By:
Pierre J Marteney
United Technologies Research Center
East Hartford, Conn 06l08
157
-------
ABSTRACT
A program concerned with the fundamentals of catalytic combustion is in
progress at UTRC. The ultimate goal is to assess the practicality of catalytic
combustion using multicomponent liquid fuels. Experimental observations of
catalyst performance are examined using a computer code which considers chemical
reaction in the gas and on the surface and heat and mass transport. The reaction
rate on the surface is found by estimation, all other processes being known
or modelled.
Initial experiments were conducted to verify the approach. A mixture of
argon-diluted propane and air in the ratio of 20/1 was used to restrict
temperature rise, eliminate or inhibit homogeneous reactions, and expedite
chemical analysis of reactants and products with a time of flight mass spectrometer.
It has been found that the analytical program correctly predicts ignition temp-
erature, temperature rise and combustion efficiency with a heterogeneous activation
energy of 10 to 11 kcal/mole.
Further experiments in the fuel-rich regime have suggested that a two-step
reaction model is required for better match of the data. C02/CO ratios were
found to approach equilibrium with increasing length, indicating that C02 is not
initially formed. This mechanism is being incorporated into the analytical program.
The effect of fuel and surface variations has also been investigated.
Specialty catalyst, having two values of active-surface loading on each of
several substrates, have been used with both pure and industrial-grade propane,
which contains approximately 35 percent propylene. In addition , the effect of
catalyst aging has been examined using samples successively heated in air at
higher temperatures. In general, the impure propane is more reactive, but under
certain conditions, especially in the fuel-rich regime, the industrial-grade propane
appears to inhibit reaction. The variation of reaction rate under these conditions is
now being determined.
159
-------
NOMENCLATURE
A0 Surface area of reactor structure per unit volume of reactor, m
S
2
Ax Total cross-sectional area of reactor, m
C Specific heat of reactor structure, cal/kg-deg K
s .
G Green's function or, mass flow rate (based on superficial area),
kg/m -sec
h Enthalpy, cal/kg
-^ Heat convection coefficient for heat transfer between the interstitial
fluid and the reactor bed, cal/m2-sec-deg K
~fl a Heat convection coefficient for heat transfer between the interstitial
fluid and the surroundings, cal/m -sec-deg K
H Heat of reaction, cal/kg (negative for exothermic reaction)
k Thermal conductivity of reactor structure, cal/m-sec-deg K
k Mass transfer coefficient for species J, m/sec
C
L Length of reactor, m
Mj Molecular weight of species J, kg/kg-mole
M Average molecular weight, kg/kg-mole
p Perimeter of reactor, m
P Pressure, Newtons/m
Rate of heterogeneous chemical reaction on catalytic surfaces,
kg/m -sec
rhom Rate of homogeneous chemical reaction in the interstitial fluid,
kg/m -sec
t Time, sec
T Temperature, deg K
Wj Weight fraction of species in interstitial fluid
160
-------
z Axial distance, m
Yj. Moles of species J reacted or formed per mole of fuel reacted
6 Void fraction of reactor bed
| Dummy variable for z
p Density of interstitial fluid, kg/m^
p Species concentration, kg/nr
j
PS Density of reactor structure, kg/m?
Subscripts
a Refers to surroundings
i Refers to interstitial fluid
o Refers to reactor inlet condition, or, reference condition
s Refers to catalytic surface
161
-------
INTRODUCTION
In hydrocarbon-air combustion systems, off stoichiometric combustion can
be used to control pollutants, particularly NO, by limiting the flame tempera-
ture rise. Fuel lean combustion can be used effectively with gaseous or
prevaporized fuel premixed with excess air. In theory, reduction of emissions
would accompany reductions in equivalence ratio. In practice, however, the
lean flammability limit must be considered; desired control measures may
necessitate a mixture which is nonflammable. The lean mixtures may also
require reaction times longer than available residence times to product
hydrocarbon burnout. Catalytic combustion provides a means of extending
the flammability limit well below the gas phase value, at the same time
providing enhancement of the overall reaction rate.
Catalytic combustion may also be applied to the fuel rich regime. Fuel
rich combustion may be of value with certain liquid fuels, especially those
containing bound nitrogen and sulfur compounds. Rich combustion could be
employed with vaporized or partly vaporized residual fuel in order to
minimize the production of nitrogen or sulfur oxides; however, the formation
of soot or long chain hydrocarbons could limit fuel rich operation. Catalytic
combustion may be able to promote oxidation of long chain hydrocarbons and
soot precursors. If catalyzed surface reactors are found to be tolerant of
liquid fuel, the fuel rich combustion could occur entirely within a catalyst
bed.
In assessing the value of catalyzed reactors, it is necessary to be able
to determine the kinetics of the heterogeneous reactions. In most cases,
however, homogeneous reactions also occur, and often dominate. In the
present studies, the gas phase reactions have been eliminated or largely
suppressed by dilution of fuel/oxygen mixtures with inert gas, thus re-
ducing the net temperature rise. An analytical program treating both
homogenous and heterogeneous processes has been developed; using established
values for homogeneous kinetics, coupled with treatment of heat and mass
transfer within the catalyst bed, unique solutions of heterogeneous rates
are obtained.
A global, Arrhenius-type rate expression, first order with respect to
reactants, is used. Details of adsorption, activation, etc. are not explicitly
stated in the model. However, changes in physical processes would be evi-
denced and might appear as changes in the overall rate, for example, as
functions of equivalence ratio. In this study, experiments were conducted
over a fuel/oxygen equivalence ratio of 1.0 to 3.0 to determine whether
apparent changes in mechanism occur, and to determine the relative importance
of homogeneous and heterogeneous reactions in practical systems.
162
-------
During the time in which the analytical model was under development,
initial experiments on propane/air ignition and stability were conducted to
establish the general features of catalytic combustion. A second line of
tests established that reaction in the diluted mixtures was initiated at
approximately the same temperatures. The third series of tests focused
on obtaining data relating the experimental conditions to combustion efficiency,
reaction rate, etc., by means of exhaust gas analysis.
Tests concerning the interaction of fuel and surface have also been con-
ducted. In one series of tests, both surface loading (active surface weight
per unit volume of catalyst) and type of catalyst substrate (honeycomb and
split-cell ceramic or spiral-rolled, thin-wall metal) were varied. In
addition, samples of one batch of catalyst, which had been subject to progres-
sively more severe heating in air, were tested.
EXPERIMENTAL APPARATUS AND PROCEDURE
A general view of the experimental system is shown in Figure 1; a
photograph of the burner unit is shown in Figure 2. A cross section of
the burner is given in Figure 3. The burner unit was fabricated from 2
inch OD threaded stainless steel pipe and pipe caps. Within the body,
several screens were located to promote mixing and prevent flashback.
The upper cap was bored to the diameter of the threads; catalyst samples
rested on the lip of the body and were retained by the upper cap. Typical
mounting of samples is shown in Figure 4. In this instance, a 1 inch
diameter sample was mounted in a cored section of untreated substrate.
Gases were taken from cylinders and metered through calibrated flow-
meters. Preheating of the system was accomplished by passing the gases through
hot watt tubular heaters; the gas/burner system could be preheated to approxi-
mately 450°C. Temperatures were monitored by sheathed thermocouples in
the gas stream and by the fine wire thermocouples within the bed.
A Bendix time-of-flight mass spectrometer was used for analysis of gases.
A fine-tip quartz probe and heated stainless-steel manifold conducted the
samples to the TOF. Sample line pressure was maintained at 10 to 15 torr. The
mass spectrum from m/e 12 to 44 was scanned for each analysis; concentrations
of the species Ar, CsHs/ O2, CO2/ CO and H2O were determined using a reduction
program developed at UTRC. Certified standard gases were purchased and used
for reference. The prior calibrations of flowmeters and TOF fragmentation
and sensitivity factors were validated by the coincidence of stated and
analyzed composition of the reference samples.
All experiments were conducted using samples from a single batch of
catalyst.* The refractory substrate is in a honeycomb configuration, having
eight cells per inch, with a minimum cell wall thickness of .008 inch. The
Matthey-Bishop, Malvern, PA.
163
-------
active surface is platinum, and represents approximately 0.1 percent by
weight. Samples for test were cut to shape and thickness from 1 inch thick
material supplied by the manufacturer.
For determination of ignition temperature, the system temperature was
slowly raised until a measurable temperature difference developed between
inlet and outlet gases. For measurements of combustion efficiency, reaction
rate, or gas composition, the system was brought to a specific initial temper-
ature with no fuel flowing, after which propane flow was initiated. Preheat
then remained unchanged through the run; the data were taken after all temper-
atures had stablized.
IGNITION TEMPERATURE
The dependence of ignition temperature on stoichiometry in the propane/
air system is shown in Figure 5 for reference velocities of 40 and 180 cm/sec.
Minimum ignition temperatures of approximately 450K were observed. A
moderate effect of equivalence ratio is evident, although the dependence
is much weaker than predicted for "normal" ignition. Figure 6 illustrates
the relationship between equivalence ratio and ignition temperature given by
Burgess and Hertzberg (Reference 1). At 300K the lean limit is & = 0.58; for
0 = 0.2 a minimum ignition temperature of 1223K is indicated. At the lower
equivalence ratios, therefore, ignition must be due to heterogenous
processes only, since gas phase reactions will not propagate a flame or lead
to temperature rise.
Decreased residence time (or increased gas flow) is reflected in
increased ignition temperature. A factor of 4.5 in velocity is responsible
for a change in ignition temperature of approximately 5OK over a broad
range of stoichiometries. The space velocities in these tests varied from
_approximately 30,000 to 350,000 hr"1- Data such as shown in Figure 6, re-
stated in terms of space velocity, are shown in Figure 7- Increase in space
velocity simultaneously reduces residence~~time and promotes transition from
laminar to turbulent flow, thus enhancing surface contact. The net effect
is a slow change in ignition temperature. At higher space velocities,
however, the onset of reaction could be at considerably higher temperatures,
since further changes would produce only small increases in turbulence.
The ignition temperature for nitrogen-diluted mixtures was found to be
essentially the same as for the normal mixtures, as shown in Figure 8. For
the extreme dilution, a factor of 40 to 1, the ignition temperature differed
by less than 10K. The substitution of argon for nitrogen produced a slightly
lower ignition temperature, with a minimum value of approximately 425K. This is
due to the lower heat capacity of argon, allowing greater local retention
of heat and quicker bootstrapping into reaction. Except for the higher
exhaust temperature however, all features, of the argon-diluted mixtures were
the same.
164
-------
REACTION STUDIES
Based upon observations that dilution does not significantly alter the
onset or extent of reaction, studies to determine the rate and general
course of reaction were undertaken. In order to permit estimation of the
heterogenous reaction rate, however, it is necessary to consider the effects
of homogenous reaction and heat and mass transfer. An analytical program
was developed for this purpose. In brief, for given entrance conditions, the
program proceeds stepwise along reactor length, determining the bed tempera-
ture, in terms of surface concentrations and interstitial temperature.
Iteration proceeds until assumed and calculated profiles agree within speci-
fied tolerance. A fuller description of the program is given in the Appendix.
The heterogeneous reaction parameters a and E, used in the Arrhenius rate
expression k-^e^. = a exp-E/RT, can be found by fitting conversion data at
different temperatures. These parameters, once fixed, should apply to all
situations in which similar kinetics are obeyed. The homogeneous reaction
rate is taken from data of Clarke et al. (Reference 3). As illustrated in
Figure 9, a large temperature difference is not necessary to capitalize on the
temperature sensitivity of the Arrhenius expression for activation energies
in the range of 5 to 10 kcal/mole. For example, if conversion at 525K is used
as a reference point, predicted conversion at 600K would be twice as great
for an activation energy of 10 kcal/mole as for 5 kcal/mole. Thus, temper-
ature differences of only several hundred degrees are adequate to specify
activation energy.
In the undiluted propane/oxygen and propane/air systems, however, a
temperature rise of 200K would be obtained only for very lean mixtures.
Through dilution with argon, the temperature rise can be tailored to desired
levels. For example, a stoichiometric mixture of propane in oxygen diluted
20 to 1 by argon has a net temperature rise of only 200K to 250K, depending
upon inlet temperature, bed length and velocity. Resulting conditions are
thus optimum for determination of activation energy; the parameter a can
be determined from the same data once activation energy is known. The result
of applying this approach to experimental data is shown in Figure 9, in which
the fit to experimental data is shown for an activation energy of 10 kcal/
mole.
In addition to providing a reasonable fit to the conversion efficiency,
the analytical program also correctly predicts the iginition temperature and
temperature rise. However, an unexpected result of the computations is that
the system becomes transport-limited rather early, with steady-state fuel
concentration on the surface dropping to values orders of magnitude below
the oxygen concentration. In the case of propane, the net transport of fuel
to the surface, evidenced by the ratio of mass transport coefficient of propane
165
-------
to that of oxygen, is smaller by approximately thirty percent. Especially
in the case of fuel lean mixtures, therefore, low surface concentration of
fuel restricts the net reaction.
CHANGE IN MECHANISM WITH EQUIVALENCE RATIO
In the development of the reaction model and laboratory experiments,
principal emphasis was on fuel lean conditions, since the ability of catalytic
combustion to produce low emissions at low equivalence ratios has been
demonstrated. However, catalytic combustion schemes using fuel rich
mixtures have also been advanced. For instance, a fuel rich preburner
operating at an equivalence ratio of 1.4 to 1.8 is attractive if carbon
formation can be suppressed. The catalytic combustor, if promoting reaction
of fuel to CO or CO2, could be of great value. To determine the result of
rich-mixture operation, tests were conducted at equivalence ratios between
1.0 and 3.0. The results indicate that a two step reaction model may be
required.
The concentrations of CO and C02 in the reactor exhaust are shown in
Figures 11 and 12 for bed lengths of 1 and 0.3 inch, respectively. Inlet
conditions were identical for both cases, and exhaust temperatures from
the 0.3 inch bed were slightly lower. Consistent with the higher exhaust
temperature in the 1 inch reactor, greater conversion to CO2 was observed.
However, the sum of CO2 and CO produced was essentially constant, as shown
in Figure 13. This indicates that the governing step in conversion is an
initial reaction on the surface, forming CO. Conversion of CO to CO2
could then take place either on the surface or in the gas phase.
The effect of bed length or residence time is more strongly illustrated
in Figure 14. The ratio of CO2 to CO is taken from the data of Figures 11
and 12. The curves on Figure 14 represent equilibrium CO2 to CO ratios at
800K and 900K (References 4 and 5). The concentration of C02 in the 1 inch
bed is greater by a factor of approximately 1.5 at all equivalence ratios.
The CO2 to CO ratio in the 1 inch bed corresponds to equilibrium at a tempera-
ture of approximately 850K which is the measured bed temperature. The CO2
to CO ratio for the 0.3 in bed would correspond to an equilibrium temperature
higher than observed in the bed. In terms of kinetics or equilibrium, no
argument could be advanced for destruction of CO2- The governing step,
therefore, must be formation of CO on the surface. The apparent excess of
CO in the 0.3 in bed reflects the time-dependent conversion of CO to CO2,
either on the surface or in the gas phase.
Modification of the analytical model to include the two reactions, fuel
to CO and CO to C02, is in progress. Sufficient data exist for homogeneous
rates; the rate for conversion of CO to CO2 will be determined using CO/
oxygen mixtures and ignoring the first reaction step. The rate of the first
heterogeneous step can then be found when propane/oxygen mixtures are utilized.
166
-------
EFFECT OF FUEL AND CATALYST VARIATIONS ON PERFORMANCE
In planning for catalytic cpmbustors, it is desirable to understand the
interaction of active surface, configuration, and fuel on the performance.
A test matrix, in which these paramters were varied, was developed. Partial
results of this series of the test are shown in Table 1. The designations T,
Al, and M indicate substrates of Torvex (honeycomb), Alsimag (split-cell)
and metal (split-cell) respectively; the second entry indicates the loading
of the active surface, platinum, in grams per cubic foot. A third entry, I,
indicates the use of industrial rather than pure propane.
Within the precision of the temperature measurements, little difference
was found between the two loadings for a given substrate, when pure propane
fuel was used. The minimum ignition temperature, although not shown, varied
only slightly. Differences between Torvex and Alsimag may be ascribed to the
total surface per unit volume; the ratio is aproximately 2 to 1. The gain in
performance due to added active metal is essentially negligible. In the case
of metal substrate, variation in loading from 20 to 40 g/ft3 did produce
significant differences, probably because the gross surface area is still
greater than in the two preceding cases, and additional metal can result in
additional active surface.
Variation of fuel produced minor differences; combustion temperatures
using industrial propane were slightly higher. However, the minimum ignition
temperatures were lower by 25K to 50K, and the ignition was more rapid. Dis-
appearance of propane also tended to be greater; at the same time, the frac-
tion of CO in product gases tended to be larger.
The effect of high-temperature exposure of catalyst to air was also
investigated. Samples which had been stabilized in air for 24 hours at
1000 C (OC-1), then treated for an additional 24 hours at 1100 C (OC-2) and
an additional 24 hours at 1200 C (OC-3) were tested on the two grades of
propane. Results are shown in Table 2.
The effect of air-aging is strong in two important respects. The decree
of reaction, as shown in Table 2, is reduced by the successive treatments.
Not shown, but a pronounced effect, is the rate at which reaction propagates.
Response of sample OC-1 to fuel was rapid; response of sample OC-3 was very
sluggish. This was especially true for pure propane. Industrial propane
produced a more uniform response time, although performance also decreased
with aging.
167
-------
ACKNOWLEDGEMENT
Thanks are due to Ms. C. J. Hewitt for programming and running the analytical
model and to Dr. M. F. Zabielski for assistance in mass spectrometric analysis.
REFERENCES
1. Burgess, D. and M. Hertzberg: The Flammability Limits of Lean Fuel-Air Mixtures.
ISA Transactions, Vol. lht pp. 129-136, (1975).
2. Coward, H. F. and G. W. Jones: Limits of Flammability of Gases and Vapors.
Bulletin 503, Bureau of Mines, U.S. Dept. of Interior, Washington, B.C. (1952).
3. Clarke, A. E., A. J. Harrison and J. Odgers: Combustion Stability in a Spherical
Combustor. Seventh Symposium (international) on Combustion, pp. 66^-673, (1958).
1^. Gordon, s. and S, J. McBride: Computer Program for Calculation of Complex
Chemical Equilibrium Compositions Rocket Performance, Incident and Reflected
Shocks and Chapman-Jouget Detonations. NAS SP-273, (1971).
5. Stull, D. R., et al.: JANAF Thermochemical Tables. Dow Chemical Company,
Midland, Michigan, I960 et seq.
168
-------
TABLE 1
Combustion Temperatures as a
Function of Substrate and Active Surface
All runs at Tin = 400 C
T-19 T-41 Al-41 AT-74 A174I
.4
1.0
1.8
625
785
770
570
800
760
665
930
825
695
800
815
650
940
830
0 M-20 M-20I M-40 M-40I
.4
1.0
1.8
530
680
695
540
750
_
645
885
815
660
890
825
169
-------
TABLE 2
Effect of Heating on Catalyst Activity
All runs at Tin = ^00 C
OC-1 OC-2 OC-3
0 T,K T,K T,K
O.lj- 605 1(25
1.0 755 415
1.8 710 14.30
( C.P. Fuel)
oA 1+65 hko
l.o 685 535
1.8 700 590 490
(industrial Propane)
170
-------
FIG. 1
UTRC CATALYTIC COMBUSTION LABORATORY
BURMER ASSEMBLY
FLOWMETERS
Bl NDI K :
MASS SPECTOM" Ml.'
171
-------
FIG. 2
UTRC CATALYTIC BURNER
SAMPLE PROBE
CATALYST
BURNER
ASSEMBLY
172
-------
FIG. 3
SECTIONAL VIEW OF CATALYTIC BURNER
T/C- OUTLET-
PROBE TO BENDIX'
MASS SPECTROMETER
T/C-BED-
SPACER,
T/C - FURNACE-
HONEYCOMB
CATALYST
HEATER
•FLASHBACK
SCREENS (2)
ARGON FUEL/O2
173
-------
FIG. 4
TYPICAL MOUNTING OF CATALYST SAMPLE
174
-------
FIG. 5
550
ai
D
I
QC
LU
Q.
500
450
IGNITION TEMPERATURE OF PROPANE/AIR MIXTURE
C.P. PROPANE
MATTHEY BISHOP CATALYST, 1 IN. THICK
I
I
I
I
I
V = 40'CM/SEC
I
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
EQUIVALENCE RATIO -0
175
-------
FIG. 6
EFFECT OF TEMPERATURE ON LEAN F LAMM ABILITY LIMIT
OF PROPANE/AIR MIXTURES
REF: BURGESS AND HERTZBERG
CO
<
<
LU
1.0
0.8
0.6
0.4
0.2
200
400
600
800
1000
1200
MIXTURE TEMPERATURE, T-deg C
176
-------
FIG. 7
EFFECT OF GAS SPACE VELOCITY ON IGNITION TEMPERATURE OF PROPAIME-AIR MIXTURES
C.P. PROPANE
MATTHEY-BISHOP CATALYST, 1 IN. THICK
550
O
LLJ
Q
I
~ 500
LU
DC
QC
LU
CL
5
LU
450
3x 104
105
SPACE VELOCITY. GHSV - HR
-1
177
-------
FIG. 8
EFFECT OF NITROGEN ADDITION ON IGNITION TEMPERATURE
OF PROPANE-AIR MIXTURES
C.P. PROPANE
MATTHEY-BISHOP CATALYST, 1 IN. THICK
EQUIVALENCE RATIO, 0= 1.0
NITROGEN ADDED:
O 25 PERCENT
0 50
D 100
500
LLJ
Q
I
DC
111
0.
450
400
0
a 0
o
a>
on o a
3x 104
105
SPACE VELOCITY, GSHV - HR~1
178
-------
FIG. 9
VARIATION OF HETEROGENEOUS RATE WITH TEMPERATURE
K=aexp_E/RT
o
LU
co
I
0.5
E,a= 10,960
400 500
600
T-DEGK
700 800
179
-------
FIG. 10
COMBUSTION EFFICIENCY IN ISOTHERMAL TESTS
FUEL- C.P. PROPANE
CATALYST-MATTHEY BISHOP HONEYCOMB, 11N. THICK
EQUIVALENCE RATIO - 1.0
DILUENT ARGON 20/1
100
(-
z 80
LLJ
Q
tr
LLJ
Q.
>
£ 60
LLJ
y
LL
LL
LLJ
z 40
O
C/5
D
CO
^
8 20
0
O EXPERIMENTAL
— — ANALYTICAL, E=!10 KCAL/MOLE
— ^ i
^^^^^
0 ^^
•^^^
^r
— .,
s
^\ f
\^J f
— f
.
/
I
'
V
, /
^ 1 r» 1
300 ^ 400 500 60
TEMPERATURE (°K)
180
-------
FIG 11
DEPENDENCE OF CO2 AND CO CONCENTRATION ON EQUIVALENCE RATIO
BURNER INLET MIXTURE: Pr/Oj/Ar 20/1 DILUTION
INLET TEMPERATURE: 673 K
OUTLET TEMPERATURE: 925 K
>
a
a
tr.
t-
z
LLJ
o
z
o
o
104
103
L = 1 INCH
CO
102
1.0
2.0
EQUIVALENCE RATIO, i
3.0
181
-------
FIG 12
DEPENDENCE OF MEASURED CONCENTRATIONS OF CO2 AND CO ON
EQUIVALENCE RATIO
BURNER INLET MIXTURE: Pr/O2/Ar, 20/1 DILUTION
INLET TEMPERATURE: 673 K
EXHAUST TEMPERATURE: 875K
>
E
cc
H-
01
o
8
104
102
1.0
I
2.0
EQUIVALENCE RATIO,
L = 0.3 INCH
CO
3.0
182
-------
FIG 13
VARIATION IN CO2 AND CO PRODUCTION WITH BED LENGTH
1_ = 1 INCH
L = 0.3 INCH
O
Sc
DC
UJ
O
10*
103
102
XV YYYYWW
1.0
I
2.0
EQUIVALENCE RATIO,
SUM
>co
>C02
3.0
183
-------
FIG 14
EFFECT OF STOICHIOMETRY ON MEASURED AND CALCULATED RATIO CO2/CO
EQUI LI BRI DM, 800 K
•EQUILIBRIUM, 900 K
O L= 1 INCH
A L = 0.3 INCH
50
o
u
OJ
o
o
20
10
0.5
0.2
1.0
\\
\\
\
800 K
0
A 4
900 K
2.0
EQUIVALENCE RATIO,
3.0
184
-------
APPENDIX
STEADY-STATE ANALYSIS OF THE USE OF CATALYZED
SURFACE REACTORS TO PROMOTE COMBUSTION
The analysis of a catalyzed surface reactor to promote hydrocarbon
combustion pertains to a monolithic structure containing uniformly distributed
parallel flow passages of arbitrary cross section. The walls of the flow
passages are impregnated with catalytic materials. Fuel and air flowing
through the passages diffuse normal to the flow and are taken to react at
the catalyzed surface.* Combustion products and the heat generated by
catalytic reaction are transported from the walls to the interstitial fluid.
As the fluid moves downstream it is also heated by homogeneous gas phase
reaction to a temperature which exceeds the temperature of the structure.
From this point to the end of the reactor the fluid loses heat to the surface
by convection. The heat gained by the structure in the downstream end of
the reaction chamber is conducted back upstream.
In developing a steady-state model of this system, the bed temperature
and the temperature and reactant concentrations in the interstitial fluid
are assumed to vary only with axial distance along the reactor. The axial
diffusion of heat and mass in the fluid is neglected. In applying the model
to the analysis of low temperature experiments designed to extract informa-
tion on rates of catalytic reaction, radiation heat transfer can be neglected.
Mass transfer coefficients and heat convection coefficients relating heat
transfer between the interstitial fluid stream and either the reactor bed
or surroundings are permitted to vary with axial distance along the reactor.
The general equations describing the rates of change with axial dis-
tance of the weight fractions of each of the chemical species in the inter-
stitial phase are:
(2)
*
Diffusion of reactant molecules within the pores of the substrate is not
considered.
185
-------
dw
H2o
dz
MH,
H2o
(3)
dw(
cos
dz
G
C02
r
As |/C02 -
dz
= o
(5)
The rates of change of species concentrations with axial distance are then given by
dz
= p -7-* + w
r dz J
dz
(6)
where
dP
dM
dz P dz M dz
dz
(7)
and
i dM
M dz
(8)
Empirical correlations used to represent heat and mass transfer coefficients and
pressure drop as functions of the flow regime in the reaction chamber.
Rates of homogeneous gas phase reaction can normally be written as global
functions of reactant concentrations and temperature
(9)
186
-------
Surface concentration of reactants and products can be related to the rate of
catalytic reaction by
. 0?
. F
F • C
IVI ij ^ v/, i
where
The change in enthalpy of the interstitial fluid with position is related to
the temperature of the structure by
„ *hi
Assuming structural thicknesses normal to the flow direction are small enough so
that the temperature at all points of a cross section can be taken as the same,
the equation describing heat conduction in the bed may be written as
=K(VTi)+AsHr
she»
where the thermal conductivity of the monolithic structure is taken as constant.
If the heat transfer coefficient in Eq. (13) is a function of axial position,
it is often desirable to express it in the form
A(z) = A) + a(z).
187
-------
The boundary conditions for Eq. (13) are
= 0 at z = 0
z = L
(16)
providing that heat loss from the ends of the structure is negligible. It should
be noted that this condition also implies that all heat transferred from the fluid
to the bed in the downstream section of the chamber is transferred back to the
fluid in the upstream section, i.e.,
T(Z)-Ti(z
= 0. U?)
With the knowledge of variation in fluid properties with temperature and extent
of reaction, the system is completely defined. The simultaneous solution of these
equations can be simplified greatly by converting Eq. (l3)j together with the
boundary conditions given by Eqs. (15) and (l6), into integral form (Refs. A-l
and A-2) as
V1'-
(18)
where
Equation (l8) is an implicit integral equation which can be solved numerically to
determine the temperature at any point in the bed in terms of the surface concen-
trations and the temperature in the interstitial fluid. By solving this equation
simultaneously with Eqs. (l) through (12), both the fluid and the surface
concentration and temperature profiles can be determined. Numerical methods have
188
-------
been developed to accomplish this. The computational scheme consists generally
of assuming a bed temperature profile, computing the fluid temperatures and con-
centrations corresponding to this assumed profile using finite difference repre-
sentations of Eqs. (l) through (12), and then using Eq. (18) to compute a new
bed temperature profile,. The next assumed bed temperature profile is then computed
from the calculated profile by adjusting the latter to satisfy Eq. (l?)> i.e.,
(19)
IQ Ji(z}dz 'calculated
Iteration continues until successive assumed and calculated profiles converge within
a fixed tolerance.
189
-------
REFERENCES
A-l. Kesten, A. S.: An Integral Equation Method for Evaluating the Effects of Film
and Pore Diffusion of Heat and Mass on Reaction Rates in Porous Catalyst
Particles. AIChE Journal 15, pp. 128-131, January 1969.
A-2. Kesten, A. S., J. J. Sangiovanni and L. S. Bender: The Use of Axial Heat
, Combustion as a Mechanism for Promoting Exothermic Chemical Reactions in
Packed-Bed Reactors. Proceedings of the Fourth International Heat Transfer
Conference, Vol. VII, Paris, France, September 1970.
190
-------
PROTOTYPE SURFACE COMBUSTION
FURNACE EVALUATION
by
G. Blair Martin
Combustion Research Branch
Energy Assessment and Control Division
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
191
-------
ABSTRACT
Prototype Surface Combustion Furnace Evaluation
by G. Blair Martin
The emissions characteristics of a prototype surface combustion
residential furnace have been evaluated using both propane and natural
gas. The combustion of premixed fuel and air takes place on a refractory
surface without a visible flame. Heat is transferred from the surface
to an air cooled firebox wall by radiation. This maintains a relatively
low surface temperature and reduces the oxides of nitrogen (NOV).
A
The furnace was operated over a range of excess air from 5 to 45%
and with heat input from 16,000 to 24,000 watts. For a nominal operating
point for natural gas at 10% excess air, NOV emissions were less than
A
15 ppm (as measured). CO and HC emissions were also low. Furnace effi-
ciency calculated from flue losses was greater than 80%. Performance
on propane was similar.
193
-------
INTRODUCTION
The Combustion Research Branch (CRB) of the EPA's Industrial Environ-
mental Research Laboratory at Research Triangle Park, N.C., has the respon-
sibility for carrying out a combustion modification R&D program directed
toward control of nitrogen oxides and other pollutants from stationary
combustion sources. In addition to pollutant emission control it appears
that the technology can also lead to equal or improved efficiency of energy
utilization compared to current practice. The majority of the R&D is
carried out under EPA contracts initiated and directed by CRB and performed
by private organizations. However, CRB also maintains an in-house research
program in a number of areas. This in-house activity provides direct
project officer expertise in combustion research and provides the capability
for initial evaluation of potential control techniques or promising novel
concepts applicable to combustion systems. The results of many of these
studies are well documented.
A continuing activity is the evaluation of novel combustion devices
which may have the potential for very low pollutant emissions. One such
device is a prototype of a residential gas fired furnace using surface
combustion to promote fuel oxidation. The purpose of this paper is to
describe the performance of that prototype.
BACKGROUND
Stationary combustion equipment used in residential heating can be
classified as area sources. Pollutant emissions from an individual unit
are relatively small, the discharge into the atmosphere is near ground
level in populated areas, the emissions are concentrated in the heating
season and the number of sources is large (i.e., about 55 X 10 units).
195
-------
Although the residential sources contribute only about 3% of the mass
emissions of nitrogen oxides (NO ), the environmental impact may be much
A
more significant. In addition, these sources require relatively clean
fuels (e.g., natural gas and distillate oil) and account for over 10%
of the stationary source energy usage (ref. 1). In 1974, residential
*| ^
warm air furnaces used over 4.5 X 10 kj of fuel with natural gas and
propane providing about 60% and fuel oil providing the balance. Based
on these factors, it appears that both emission control and energy con-
servation in this source class are required. The emission levels for
distillate oil fired residential furnaces have been characterized
(ref. 2) and there has been some technology development for reducing
emissions and increasing efficiencies (ref. 3). Much less work has
been done on gas fired residential furnaces because of less flexibility
in adjusting operating conditions and relatively simpler equipment
design. The normal residential gas furnace does not have a "burner"
as such, but rather relies on natural draft to provide combustion air
and to accomplish fuel and air mixing. The available gas pressure is
used to entrain some air and to form a primary air and fuel mixture that
is admitted to the combustion chamber through one or more log manifolds.
Secondary air to complete combustion is admitted to the combustion chamber
around the manifolds. Although the primary air can be controlled to 'a
degree, there is essentially no way to control secondary air. Therefore,
the overall excess air cannot be controlled to a significant degree.
The available emission data on gas fired furnaces are very limited.
Hall (ref. 3) reported data for three gas furnaces with NO emissions
ranging from 0.084 to 0.115 gm of NOV (as NO) per 106 cal [i.e., 0.32 to
in
0.42 kg NOX (as N02) per 10 J]. Any given furnace could only be
operated over a narrow range of excess air; however, the three different
designs operated at levels in the range of 20 to 60% excess air. The
emissions of carbon monoxide (CO) and hydrocarbon (HC) were low for all
three furnaces at the normal operating point. DeWerth (ref. 4) measured
emissions from 38 forced warm air gas furnaces in the AGA laboratory
and reports an average NO value of 0.097 Ib NOV (as NO.,) per 10 Btu
A At
196
-------
[i.e., 0.415 kg NOX (as N02) per 10 J] with a small effect of furnace
design. Decreasing the primary air setting to the point of yellow flame
operation decreased the NOX level by about 10%; however, CO emissions
increased by a factor of 20 and the N02 component increased from about
5% of the total NOY to the range of 10 to 17%. Data for other gas
A
appliances and for modified furnace design also showed high fractions of
NOg associated with higher CO levels. Several modifications of furnace
design were examined with the most successful one achieving 37 ppm
of NOV [i.e., 0.172 kg NOV (as NO,) per 1010 J] at low CO levels by
X X C-
putting a screen in the flame zone to increase radiant heat transfer.
Finally, Brookman (ref. 5) performed field measurements on 50 residential
gas fired furnaces and boilers and reported average NO emissions of
112 Ib (as N09) per 106 ft3 of gas [i.e., about 0.455 kg NOV (as N09)
ln C. X £
per 10IU J].
The efficiency of the furnace is dependent on the extent of heat
loss from the system by various avenues. The predominant energy losses
are the following:
1. Latent heat of vaporization of the water formed during
combustion which cannot be recovered without a condensing
heat exchanger. About 10% of the gross heat input of natural
gas is irretrievably lost in this form using current technology.
2. Sensible heat in the combustion products which is dependent
on the amount of excess air used for combustion, the amount
of dilution air and the temperature of the gas in the flue.
This loss can be reduced by low excess air operation, elimina-
tion of dilution air and/or reduction of stack temperature.
3. Draft losses, which remove residual heat from the firebox
during the off-cycle. This loss can be reduced by minimizing
or eliminating air flow through the furnace during off-cycle.
4. Miscellaneous casing losses. These losses are normally held
to a minimum by proper insulation of the furnace enclosure.
197
-------
The significant features of these losses are discussed for fuel
oil in reference 3. The general discussion also holds for natural gas,
although differences in latent heat losses and system design will change
the quantitative levels. DeWerth (ref. 4) reported that the average of
furnaces tested in the AGA Lab had 5.8% C02 in the flue products (approxi-
mately 100% excess air); however, no flue temperatures were given. Flue
losses for two units tested were reported to be 20.1 and 26.6%, based on
steady state operation. The cycle average efficiency will be further
reduced by the draft and casing losses.
EXPERIMENTAL APPROACH
This section provides information on several aspects of the experi-
mental approach including: 1) the fuel supply system; 2) the Bratko* proto-
type furnace; 3) the analytical instrumentation; and 4) the range of variables
examined.
Fuel Supply
Natural gas was provided from a laboratory gas line at a pressure of
3.45 X 10 pascal gage [5 psig]. Propane was provided from a pressurized
4
gas bottle and the line pressure was controlled at 4.83 X 10 pascal gage
[7 psig]. The gas supply was passed through a dry gas meter to obtain a
volumetric flow rate. Following this the pressure was regulated to 3.45 X
o
10 pascal gage [0.5 psig] and the gas passed through an ASTM orifice meter.
The pressure drop across the orifice was used to adjust the gas flow to a
constant value during a test series. The gas supply was connected to the
fuel inlet system of the furnace.
Bratko Prototype Furnace
The furnace tested was a prototype of a surface combustion concept
for use in residential gas furnaces. The concept is based on combustion
of premixed fuel and air on the surface of a refractory matrix without
(*) Bratko Corp., 10714 Harvard Ave., Cleveland, OH 44105
198
-------
visible flame. The glowing refractory surface radiates to a surrounding
air cooled combustion chamber and, thereby, maintains the surface temperature
at a relatively low value (i.e., below about 1250 K). A schematic of the
surface combustor and firebox is shown in Figure 1. The combustion chamber
was enclosed in a conventional upright furnace body and used a three speed
circulating air fan to pass air over the heat exchange surface. The speed
of the fan increased stepwise as the pressure drop across the system increased.
The combustion air was supplied by an external centrifugal fan that
was considerably oversized in capacity, but was necessary to supply the
required pressure. [This arrangement would not be practical in a production
furnace; however, Bratko stated that a small fan with the necessary pressure
capability has been developed (ref. 6)]. A regulator drops the natural gas
supply pressure to 3450 pascals gage (0.5 psig), then the gas is piped
through solenoid valves to both a pilot burner and the main furnace.
The pilot burner premixes the gas with part of the combustion air and the
mixture is ignited with a spark. A flame rod is used to prove the pilot
flame prior to the burner gas solenoid valve opening. The burner gas supply
is premixed with combustion air with a mixing device (in the prototype
furnace the amounts of both air and fuel could be controlled independently,
although in practice, relatively fixed flow rates would be expected). The
premixed gas and air is fed to a plenum inside the surface combustor at a
Pressure of about 895 pascals gage (0.13 psig) and is forced through the
refractory pad to burn on the surface. The refractory is selected to be
sufficiently insulating that the plenum temperature remains essentially
at ambient, thereby eliminating preignition due to heat feedback.
Analytical Instrumentation
The gas sample for analysis was withdrawn from the flue after the gas
had been cooled to 422-475 K in the convective pass of the heat exchanger.
The samples are conditioned and analyzed with a similar system to that
199
-------
described earlier (ref. 7). To summarize, the analytical instruments are:
1) oxygen by paramagnetic resonance; 2) CO and C02 by non-dispersive
infrared (NDIR); 3) hydrocarbon by flame ionization detector; 4) NO by a
long path NDIR; and 5) NO and NO by chemiluminescence analyzer. The
n
temperature of the flue gas was measured with a dial thermometer and cross
checked with a thermocouple.
Range of Variables
The tests were planned to screen the furnace combustion and emission
performance over the permissible range of operation. The two variables
were fuel flow and excess air level. Since there was some interaction
between fuel and air flow in the mixer it was not possible to attain
completely comparable input levels. Therefore, three nominal values were
used: 16,000, 19,000 and 22,000 watts (55,000, 65,000, and 75,000 Btu/hr,
respectively). The excess air range for the first two fuel flows was from
45% down to the point where excessive CO (i.e., > 200 ppm) was formed. For
the highest rate the maximum excess air was 25%.
In addition a limited number of tests were run with propane to establish
the multifuel capability of the unit.
EXPERIMENTAL RESULTS
This section presents a discussion of two specific aspects of the
results: 1) emission; and 2) efficiency.
Emissions Measurements
The primary evaluation of the furnace performance was based on measure-
ments of pollutant species in the flue gases as a function of operating
condition. To fully characterize the operating range the excess air level
was varied from a high value (i.e., > 25%) down to the point that excessive
200
-------
CO (i.e., > 200 ppm) was being emitted. No hydrocarbons or smoke were
observed during the test. The effects of excess air and fuel input are
discussed below.
Excess Air. The effect of excess air level on emissions is shown in
Figure 2 for a thermal input of 16»000 watts (55,000 Btu/hr). The data shown
here were obtained during five different experimental periods covering 4
months and the small scatter in the points shows that the performance was
quite repeatable. At 40% excess air, the CO emissions are quite low (i.e.,
20-30 ppm) and remain low as the excess air is reduced to 10%. When the
excess air is reduced below 10% the CO begins to rise sharply and at 5%
excess air is greater than 1500 ppm. The results for oxides of nitrogen
are shown as measured for both NO and NOX and corrected to zero % excess
air for both NO and NOX. The NOX concentration to zero % excess air
gives a direct indication of the trend of mass emissions as a function of
excess air and will be used as the primary basis of discussion. The NOX
is low (i.e., 7.5 ppm) at 40% excess air and rises slowly to 10 ppm as
excess air is decreased to about 15%. Then as excess air is reduced
further, the NO rises sharply to about 19 ppm at 7.5% excess air. Compari-
A
son of the as measured values of NO and NO (where the difference is inter-
A
preted as being N02) shows that the amount of N0« increases as the excess
air is decreased; however, the ratio of N02 to NO is approximately constant.
This holds true even when high CO levels are observed.
Fuel Input. The effect of fuel input on emission performance was
examined. The furnace design condition was taken as a nominal 16,000
watts (55,000 Btu/hr) and two higher levels were evaluated: 19,000 and
22,000 watts (65,000 and 75,000 Btu/hr, respectively). The data for
these higher levels are shown in Figures 3 and 4. The emission trends are
essentially the same as the baseline case with only small changes. The
point at which CO emissions increase sharply is moved down to 5% excess
ai> for both higher heat inputs. The NOX emissions are increased slightly
as heat input increases for excess air levels of 10 to 30%; however, the
ends of the range do show minor differences. Finally, the ratio of N02
to NO decreases as the heat input increases.
201
-------
Fuel Type. A limited series of runs were performed using propane
as the fuel. In general, the effect of excess air on emissions was similar
to that for natural gas, except that the onset of excessive CO occurred
at somewhat higher excess air (i.e., 10-15%). A run with propane at a
nominal 26,000 watts (90,000 Btu/hr) heat input gave a different trend
as shown in Figure 5. The onset of CO formation occurs at a higher excess
air and, of more interest, the ratio of N02 to NO increases significantly
as the CO begins to increase.
Thermal Efficiency
The data gathered for evaluation of the furnace emission performance
can also be used to gain some insight into the thermal efficiency of the
furnace. In addition, several measurements of furnace warm air output
were made. These two indications of the furnace thermal performance are
discussed.
Stack Heat Loss. The main energy loss from a furnace is in the
sensible and latent heat associated with the flue gas components, princi-
pally N2, C02, 02 and H20. The latent heat loss is determined by the
carbon to hydrogen ratio of the fuel and is irretrievable unless a condens-
ing heat exchanger is used. It should be noted that condensing heat
exchangers for this application are not currently available. For natural
gas the latent heat is equivalent to about 10% of the gross heating value.
The sensible heat losses can be minimized by low excess air operation and
the lowest stack temperature consistent with avoiding condensation (i.e.,
about 180 K net). A generalized plot of efficiency versus excess air
with stack temperature as a parameter is shown in Figure 6 for natural
gas.
202
-------
The flue gas temperature taken during the experimental runs was
relatively independent of excess air for a given firing rate; however,
the temperatures increased with increased firing rate. The average
flue gas temperatures were 447, 475 and 498 K, at 16,000, 19,000 and
22,000 watts, respectively1. An operating excess air of 15% 1s selected
based on the emission curves, giving steady state furnace efficiencies
for the three firing rates of 83.4, 82, and 81%, respectively. When the
casing losses are taken into consideration, the actual heat loss would
be about 2% greater. At the low firing rate, the stack temperature was
below the value required to prevent condensation and a production furnace
would require less convective heat transfer surface. For the high firing
rate, the stack temperature potentially could be reduced by about 22 K.
If heat exchanger design was changed to achieve a 475 K stack temperature
for each firing rate, the theoretical efficiencies would be the same
as that for the 19,000-watt heat input. However, a given furnace is
normally designed to be capable of a range of firing rates, and the
existing design represents a reasonable compromise for input over 19,000
watts.
Some measurements of output air velocity and temperature rise were
made at the output of the plenum on top of the furnace. As an extended
duct run was not used to smooth the flow, the stratification of flow
due to the heat exchanger was significant. Measured exit velocities
were 1.25 to 11.5 m per sec (250 to 2300 ft/min) across the exit face of
the plenum, and low temperature rises were normally associated with low
velocity areas. These stratifications also strongly influence the
accuracy of the measurements; however, they can be taken as semi-
o
quantitative. The warm air flow was about 0.45 m /sec for two firing
rates, with average temperature rises of 24 and 31 K for heat inputs
of 19,000 and 24,000 watts, respectively.
Discussion of Results
Examination of the experimental results raises several points that
merit further discussion. It 1s obvious that some additional measurements
within the combustion chamber would greatly assist in explanation of the
203
-------
phenomena involved; however, the fact that the furnace was a prototype
inhibited making the modifications required for sampling.
Operating Principle. The inventor of the furnace stated (ref. 6)
that the furnace was designed to operate with a hot refractory surface as
a means of increasing the radiative heat transfer to the heat exchanger.
The use of radiative heat transfer for combustion zone heat removal is
a common feature of large conventional boilers with diffusion flame
burners (especially the water-tube type); however, in general residen-
tial systems do not have heat transfer surface in the combustion zone.
It appears that the key features of the design are: 1) the insulating
properties of the refractory used to form the surface combustor; and
2) the use of radiative heat transfer to reduce the refractory surface
temperature to a level where degradation does not occur. The refractory
material was selected to have a low thermal conductivity and, therefore,
a very slow soakback of heat from the reacting surface to the plenum
containing the premixed fuel and air. It appears that this has a two-
fold benefit: 1) the heat release is maintained in a thin zone close to
the outside of the refractory surface; and 2) radiative heat transfer
reduces the refractory surface temperature to a level where degradation
does not occur. The degradation of the refractory with time, and the
attendant increase in probability of autoignition, needs to be evaluated.
The use of radiative heat transfer allows a high rate of energy
removal and apparently maintains the surface temperature well below the
adiabatic temperature for the fuel and air stoichiometry being used.
This allows the use of fuel/air ratios near stoichiometric where the
adiabatic temperature is about 2200 K (3500° F), and yet the reaction
zone temperature is maintained below the level where degradation of the
refractory would occur. In addition, the temperature is low enough
to reduce thermal NO formation.
The observations of the effect of heat input on combustor performance
indicate that some increase in surface temperature may occur with increased
fuel input as indicated by the lower excess air capability. This is a
204
-------
relatively small effect that may be associated with either the temperature
or the thickness of the heat release zone. The small increase in NO level
A
suggests that the change is relatively small. It appears that the increased
heat release is balanced by increased radiative transfer and/or an increased
convective removal of energy due to the larger volume of combustion gases at
a given excess air level. The effects could be explained either on the
basis of complete reaction of the fuel within the surface or on the
basis of partial reaction in the surface with oxidation being completed
by homogeneous reactions in the space between the surface and the firebox
wall. To answer these questions fully would require measurement of:
1) surface temperatures for the heat exchanger and the refractory material;
2) gas temperature at the firebox exit; and 3) gas phase species measure-
ments within the firebox.
Based on the foregoing discussion and operating experience with two
different combustor and firebox configurations in Bratko prototype furnaces,
it appears that proper matching of the combustor and radiative heat transfer
surface does significantly influence the performance.
NOg Emissions. As noted in the background section, above, instances
where the N09 is greater than 25% of the NO have been observed by several
£ /\
investigators for natural gas combustion under conditions where CO levels
are high. This same trend was observed for an earlier version of the Bratko
prototype tested for both methane and propane at low excess air (i.e., < 20%)
Due to the fuel/air mixer configuration of the earlier furnace, excess air
was changed by increasing fuel flow at a relatively constant air flow. With
the new furnace, propane at a high fuel rate exhibited a high N02 to NO
ratio at low excess air. This effect was not noted with the new furnace
burning methane even at relatively high firing rates; however, the excess
air level at which CO increased was significantly lower (i.e., < 6%).
Although there is some disagreement with his interpretation, Merryman (ref.
8) has reported "flame front N02" for flat flames. His data show conditions
within the reaction zone where N02 is a large percentage of the NOX< This
is also the zone in which significant amounts of CO were present. Although
a flat flame is difficult to relate directly to a surface combustor,
205
-------
these results give a possible basis for explaning the furnace data. It
can be postulated that, for a given combustor surface and heat exchanger
area, total oxidation occurs in or very near the surface until some
critical fuel flow rate is reached and the final oxidation of fuel fragments
(e.g., CO) begins to occur in the gas phase above the surface. At low
excess air levels, the combustion may not.be completed fast enough by
these gas phase reactions, thereby allowing some pockets of gases containing
CO and high N02 to NO ratios to escape and be quenched before the N02 can
relax back to NO. If this is the case, the data would tend to indicate that
this phenomenon is sensitive to the relationship between the surface combus-
tor and the radiant heat exchanger firebox.
Efficiency. Although the furnace efficiency is good for all the cases
tested, the potential exists for optimizing the furnace for each heat input
value by fine tuning the convective section of the heat exchanger to get
minimum allowable stack temperature. Other features of the furnace may
provide an increase of cycle average efficiency, particularly by reducing
or eliminating off-cycle losses. One source of such loss is the continued
flow of draft air through the firebox during the off-cycle. This not only
removes accumulated heat from the furnace components, but may also cause a
loss of heated air from the furnace space. The Bratko design provides a
positive air shut off, which should reduce off-cycle losses significantly.
CONCLUSIONS
Based on the furnace evaluation the following conclusions can be drawn:
1. The surface combustion concept shows clear potential for NO
emissions significantly below previously reported values,
while also giving low levels of CO and HC.
2. The Bratko furnace represents an application of surface combus-
tion to a practical system once a satisfactory combustion air
source is incorporated.
206
-------
The low excess air capability gives the potential for high
efficiency. Other furnace features which limit off-cycle
losses also have the potential for increasing cycle average
efficiencies.
The concept appears to lend itself to a wide range of firing
rates. In fact, performance improved with natural gas at
the higher firing rates tested.
207
-------
REFERENCES
1. Barrett, R. E., S. E. Miller, and D. W. Locklin, "Field Investigation of
Emissions from Combustion Equipment for Space Heating," Battelle Columbus
Laboratories, EPA-R2-73-084a, NTIS No. PB 223-148, June 1973.
2. Combs, L. P., and A. S. Okuda, "Residential Oil Furnace System Optimi-
zation - Phase II," Rockwell International, EPA-600/2-77-028,
NTIS No. PB 264-202/AS, January 1977.
3. Hall, R. E., J. H. Wasser, and E. E. Berkau, "A Study of Air Pollutant
Emissions from Residential Heating Systems," U. S. EPA, ORD, Control
Systems Laboratory, EPA-650/2-74-003, NTIS No. PB 229-697/AS, January 1974.
4. DeWerth, D. W., and R. L. Himmel, "An Investigation of Emissions from
Domestic Natural Gas-Fired Appliances." Presented at the 67th APCA Annual
Meeting, June 14, 1974, Denver, Colorado.
5. Brookman, G. T., and P. W. Kalika, "Measuring the Environmental Impact
of Domestic Gas-Fired Heating Systems." Presented at the 67th APCA
Annual Meeting, June 14, 1974, Denver, Colorado.
6. R. Bratko, Bratko Corporation, Cleveland, Ohio, private communication,
August 1976.
7. Pershing, D. W., J. W. Brown, and E. E. Berkau, "Relationship of Burner
Design to the Control of NOX Emissions Through Combustion Modification."
In Proceedings, Coal Combustion Seminar. EPA-650/2-73-021, NTIS No.
PB 224-210/AS, September 1973.
8. Merryman, E. L., and A. Levy, "Nitrogen Oxide Formation in Flames:
The Roles of N02 and Fuel Nitrogen," 15th Symposium (International)
on Combustion, Tokyo, Japan, August 1974.
208
-------
SECTIONAL VIEW OF SURFACE COMBUSTOR AND FIREBOX
PREMIXED
AIR AND
FUEL
TO CONVECTIVE HEAT EXCHANGER
0.10m
kVWWV
MINI
COOLING AIR
Figure 1. Sectional view of surface combustor and firebox.
209
-------
EMISSIONS CHARACTERISTICS AT A NOMINAL HEAT INPUT
OF 16,000 WATTS OF NATURAL GAS
O NO (AS MEASURED)
D NOX (AS MEASURED)
A CO (AS MEASURED)
NOX (CORRECTED TO 0% EXCESS AIR)
15 20 25 30
EXCESS AIR, percent
Figure 2. Emissions characteristics at a nominal heat input of 16,000 watts of natural gas.
1100
1000
900
800
210
-------
EMISSION CHARACTERISTICS AT A NOMINAL HEAT INPUT
OF 19,000 WATTS OF NATURAL GAS
25
20
K
8 15
LU
C9
10
ONO
DNOX
A CO
.A.
I
T
10
30
I
A I
T
35
40
1100
1000
900
800
700
600
500
LU
1
oc
x
o
400
cc
u
300
200
100
45
15 20 25
EXCESS AIR, percent
Fiaure 3. Emission characteristics at a nominal heat input of 19.000 watts of natural gas.
211
-------
EMISSION CHARACTERISTICS AT A NOMINAL HEAT INPUT
OF 22,000 WATTS OF NATURAL GAS
20
> 15
oc
a
a.
a.
X
o
a 10
o
EC
T_ .
\
0
I
ONO
DNOX
AGO
10
1100
1000
900
800
00
ee
o.
00 iu
a
x
o
400
100
200
00
40
45
15 20 25 30 35
EXCESS AIR, percent
Figure 4. Emission characteristics at a nominal heat input of 22,000 watts of natural gas.
212
-------
EMISSION CHARACTERISTICS AT A NOMINAL HEAT INPUT
OF 26,000 WATTS OF PROPANE
15
10
>
K
E
a.
V)
x
o
CD
ONO
DNOX
AGO
a
1600
1400
1200
1000!
S
c/i
800 ~
UJ
O
X
600
o
CO
DC
400
200
0 10 20 30 40 50
EXCESS AIR, percent
Figure 5. Emission characteristics at a nominal heat input of 26,000 watts of propane.
213
-------
GROSS EFFICIENCY AS A FUNCTION OF OPERATING CONDITIONS FOR NATURAL GAS
86
10
20
60
70
30 40 60
EXCESS AIR, percent
Figure 6. Gross efficiency as a function of operating conditions for natural gas.
80
214
-------
AN ANALYSIS OF CATALYTIC COMBUSTION
IN MONOLITHIC HONEYCOMB BEDS
By:
Robert M. Kendall, John T. Kelly, Edward K. Chu, and
John P. Kesselring
Acurex Corporation/Energy & Environmental Division
485 Clyde Avenue
Mountain View, California 94042
215
-------
ABSTRACT
An analysis of catalytic combustion in monolithic catalyst beds is
developed. This analysis serves as a basis for the PROF-HET computer code
set. The MET code treats the critical phenomena associated with "blowout"
of catalytic combustors by performing complex energy and species mass
balances along the catalyst surfaces. The PROF code treats the phenomena
associated with ignition and sustenance of the homogenous gas phase reactions.
The code accurately treats the axial feedback of chain carrying species and
energy. This feedback is required to maintain the flame-like gas phase
combustion process, and thus avoid "breakthrough".
'Based on a series of MET computer simulations of catalytic combustor
performances the benefits of large cells, high preheat, and high adiabatic
flame temperature are qualitatively related to increases in the maximum
mass throughput at blowout. In a similar manner simulations with PROF in-
dicate the degree of surface combustion required before the system will
sustain homogeneous combustion reactions. These simulations have led to
catalytic combustor designs with greatly improved performance.
217
-------
LIST OF SYMBOLS
A cross sectional area
CM Nusselt number divided by Reynolds and Prandtl numbers
(Stanton number)
C dimensionless mass transfer coefficient defined by Equation (8)
m (Sherwood number)
C specific heat
C circumference of bounding tube
w
D.. binary diffusion coefficient
' U
D diffusion constant defined by Equation (17)
E activation energy for kinetic reaction
F. diffusion factor of species i
h enthalpy ,
J species flux
k thermal conductivity
K channel segment radiative heat transfer view factors
K equilibrium constant
Le Lewis number
m mass rate of gas
M molecular weight
p pressure
Pr Prandtl number
q heat flux
R gas constant
Re Reynolds number
s distance along flame axis
218
-------
LIST OF SYMBOLS (Concluded)
T temperature
W chemcial production rate
X mole fraction
Y mass fraction
e wall emissivity
p density
a Stefan-Boitzmann constant
Superscripts
P reaction products
R reaction reactants
Subscripts
i denotes species
m denotes reaction
o inlet conditions
r radiation
rl upstream reservoir
r2 downstream reservoir
s solid bed material
w wall
219
-------
INTRODUCTION
Catalytic combustion in a honeycomb monolith is a complex process
which involves the interaction of several physical and chemical phenomena.
Of primary importance are (1) radial heat and mass transport between the gas
and wall, (2) axial heat and mass transport in the gas, (3) axial radiative
and conductive wall heat transfer, (4) heterogeneous surface and bulk gas
phase chemical kinetic reactions. The interaction of these phenomena de-
termines the maximum mass throughput and fuel conversion efficiency of the
catalytic bed.
In this paper, the analysis of a catalytic combustor is discussed
with regard to:
• Fundamentals of operation, where a simplified version of the
catalytic combustion process is described and the important sys-
tem implications are introduced,
• The PROF-HET computer code, which models all of the important
physical phenomena occurring within monolithic catalytic com-
bustors and verifies some of the conclusions of the simplified
model,
• Conclusions and recommendations regarding major system impacts
and further analytical requirements.
The following discussion provides the information needed to understand the
operation of the catalytic combustion system.
FUNDAMENTALS OF OPERATION
Catalytic combustion in a monolith bed includes the interaction of
chemical reactions (surface and gas phase), diffusive heat and mass transport
221
-------
(laminar or turbulent), convection, bed conduction, and radiation. These
phenomena are depicted schematically in Figure 1. During steady operation,
the catalytic combustion process can be described as follows:
• Premixed fuel and air are introduced into the combustor.
o These gases diffuse to the catalyst-coated surface of the com-
bustor and react on the active sites at and within the surface.
Near the cell entrance, where most of the gas is at low tempera-
ture, gas-phase chemical reactions are unimportant.
• At the entrance, heat release is controlled by catalytic wall
chemical reactions. This heat is transferred by conduction,
radiation, and convection. Further down the channel, where the
gas has been preheated to a high temperature, gas-phase reactions
become active. In this region fuel is rapidly consumed by a
"flame type" phenomenon which controls the amount of unburned
hydrocarbon emissions that escape the system.
• Surface reaction products diffuse back to the main flow of gases
and are carried downstream.
Under normal operating conditions, wall and gas phase reactions are
active and very little unburned hydrocarbon escapes the bed for lean and
stoichlometric initial mixture ratios. However, it has been experimentally
observed that above a certain mass flow limit, small increases in flow rate
cause an abrupt rise in unburned hydrocarbon emissions. The abruptness of
the increase indicates that a "flame type" phenomenon has been extinguished.
This condition, called breakthrough, represents an upper mass throughput
for low unburned hydrocarbon emissions.
Increasing the mass throughput in a catalytic bed to levels much
above the breakthrough point can cause the front of the bed to become cool.
It has been experimentally found that small increases in mass throughput,
once the front end of the bed has become cool, can cause the cool region to
spread downstream. At this point, all wall reactions are extinguished and
the entire bed becomes cold. This condition, called blowout, represents
the maximum mass throughput for hot bed operation. It is very important to
know when this blowout condition occurs for a given catalyst system.
222
-------
, For the purpose of understanding system characteristics, considerable
simplification of the catalytic combustion process can be made if it is as-
sumed that:
t No conductive or radiative heat transfer occurs
• The Lewis number is unity for all species
t The combustion reaction is a single global reaction described
by an Arrhenius law equation
Once these assumptions are made, it is possible to perform a simple
mass balance on the lean reactant at the wall of the monolith bed; that is,
the mass of lean reactant transported to the wall is equal to the mass of
lean reactant consumed at the wall. In equation form this can be written as
m = Nu ff (KE - Kw) =
where m = mass of lean reactant transported to and consumed at the wall,
per unit area
Nu = Nusselt number for mass transfer (Sherwood number)
p = gas density
P = diffusion coefficient
D = diameter of one channel of monolith bed
KE = mass fraction of lean reactant at boundary layer edge <
K,, = mass fraction of lean reactant at monolith wall
A = preexponential factor
AE = activation energy
R = universal gas constant
T. = adiabatic flame temperature of fuel/oxidizer
T.J =.temperature at monolith wall*
Tp = preheat temperature of fuel/oxidizer
*
Based on the assumptions, this temperature can be related to
the residual concentration of lean reactant at the surface, Ku.
KW
TW = TP + K (TA " V
223
-------
Each of these expressions for the mass flux can be shown graphically
by plotting the mass flux vs. the mass fraction of lean reactant at the mono-
lith wall (i.e., plot m vs. K.,). This is shown in Figure 2.
r- m - AKwrAE/RTW
m
m - Nu
-------
The simplest of these to implement is the use of large diameter cells.
These large cells operate at a low transfer coefficient and should be
effective for both lightoff and sustained operation.
However, using large cells throughout the bed would result in poor
surface conversion of combustibles to products. The amount of surface con-
version is directly related to the number of transfer units in the bed, where
the length of each transfer unit is equal to
Pr-Re n
4 Nu ' u'
where P s Prandtl number
r
Re = Reynolds number
Nu = Nusselt number
D = cell diameter
Thus, to get complete conversion on the catalyst surface, it is necessary to
have many transfer units available by minimizing the length of each transfer
unit. This suggests use of small diameter cells. The small cells will"also'
accelerate gas phase reactions, which are helpful for full conversion.
As a consequence of these operational fundamentals, it appears that a
catalytic monolith bed used for the purpose of combustion should use large
diameter cells at the front of the bed to prevent blowout, and small diameter
cells at the back of the bed to maximize the number of transfer units in a
given length of bed. Therefore, for a given catalyst, it was postulated that
superior performance can be obtained by using the catalyst in a graded cell
configuration, with large cells at the front end, small cells at the back
end, and perhaps one or more intermediate sized cells between. A complete
discussion of the model used to support these conclusions is given in the
following section.
THE PROF-HET COMPUTER CODE
The PROF-HET computer code models the important physical phenomena
occurring within monolithic catalytic combustors. An efficient numerical
technique which includes axial and radial heat and mass transport, axial
225
-------
radiative and conductive wall heat transfer, and heterogeneous surface and
bulk gas phase chemical kinetic reactions has been developed to establish
how performance varies with bed operating and design parameters. In this
technique, matrix procedures are used to solve the finite difference form of
the governing differential equations. The axial distribution of both wall
and bulk gas properties, as well as wall temperature, are output by the code.
The solution procedure is reliable and stable for the range of input param-
eters used to date.
The PROF-HET model differs from previous models in that it can handle
the high-temperature effects of catalytic combustion, where bed'radiative
heat transfer and "flame type" phenomena are important. Most of the models
constructed to date have focused on catalytic cleanup devices (e.g., automo-
tive catalytic mufflers, and industrial process exhaust cleanup devices for
sludge drying, PVC processing, foodstuff processing, etc.) where the temper-
ature rise due to catalytic oxidation is small compared to that which occurs
during combustion. For example, Votruba et al. (Reference 1) published an
analytical study on the heat and mass transfer in monolithic catalysts. In
their model, the heat and mass transfer normal to the channel walls was
treated by a transfer coefficient approach where the detailed distribution
of properties across the channels need not be known. This reduced an essen-
tially two- or three-dimensional (for noncircular channels) problem to one
dimension. Equations for the variation of wall temperature and gas condi-
tions as a function of distance along the channel were developed. Calcula-
tions made using this model are more economical than higher dimensional
models. Illustrative predictions for catalytic monoliths where the wall
temperature rise was both low and high were given. These predictions showed
that monolithic structures give more stable operation and less pressure drop
than packed beds. However, since radiative heat transfer and gas phase re-
actions were ignored and only fully developed heat transfer parameters uti-
lized, the model cannot be accurately applied to catalytic combustors.
Cerkanowicz et al. (Reference 2) presented results generated using
a catalytic combustor model similar to Votruba's model, except for the in-
clusion of a gas phase chemical reaction. As with Votruba, only constant
226
-------
fully developed heat transfer coefficient values were utilized in the model.
Also, the gas phase chemical reaction was assumed to be a one-step process.
This approach is useful for illustrative predictions to show global effects
due to gas phase reactions. However, the method is not sufficiently funda-
mental for exploring the types of "flame" phenomena and pollutant formation
processes which occur in catalytic combustors.
Young and Finlayson (Reference 3) developed one-, two-, and three-
dimensional models for monolithic catalytic converters. Their two- and
three-dimensional models treated heat and mass transfer perpendicular to
the wall by an orthogonal collocation method which determines the detailed
distribution of properties across the tube. As in Votruba's model, bed
radiative transfer and gas phase reactions were riot included, making 'appli-
cations to catalytic combustors limited^ Young and Finlayson's predictions
showed that three-dimensional effects associated with peripheral temperature
and concentration variations are not important to overall bed operation.
They also showed that holding transfer coefficients constant over the bed
length is not an adequate approach.
Heck (Reference 4), using a finite difference procedure, compared
one- and two-dimensional catalytic converter model predictions. Like
Finlayson, Heck concluded that one-dimensional constant heat transfer solu-
tions were not adequate. However, he showed that if the transfer coefficients
are allowed to be functions of distance and wall temperature, characteristic
of developing boundary layers, predictions comparable to the two-dimensional
results could be achieved at a fraction of the cost.
Following Heck, this study employs a one-dimensional model with trans-
fer coefficients given as functions of distance, wall temperature, initial
gas temperature, and inlet flow conditions. In addition to including wall
reactions and bed heat conduction, the present model also includes bed radi-
ative heat transfer and gas phase chemical reactions and axial diffusion.
The important bed operating characteristic of blowout is controlled
by events occurring near the combustor inlet. At this location, the bulk
gases are relatively cool and heat release due to gas phase chemical
227
-------
reactions is small. Heat release in this zone is solely due to wall surface
chemical reactions. The heat produced by these reactions is carried away by
radial transport of heat to the bulk gases, heat conduction towards the front
of the bed, and radiative transfer towards the front of the bed and into the
upstream reservoir. Because the thermal boundary layer is developing in this
region, the heat transfer coefficient is large. Above a certain limiting
mass flow, the heat transfer coefficient for the developing boundary layer
becomes so large that the radial convective heat loss plus the radiative heat
loss exceeds the heat produced by wall reactions. The wall reactions are
then extinguished and the bed becomes cold. To model this blowout state,
the following phenomena must be treated:
• Heterogeneous surface chemical reactions
t Radial heat and mass transport
• Axial convection
• Axial bed conductive and radiative heat transfer
Unlike blowout, breakthrough and certain emissions phenomena are con-
trolled by processes which occur away from the front of the bed where the
bulk gases are hot. Breakthrough occurs when there is insufficient preheat
of the bulk gases by the wall reactions to "light off" gas phase "flame type"
phenomena. Since breakthrough is believed to be flamelike in nature, in addi-
tion to the above phenomena, the following phenomena must be treated:
• Homogeneous gas phase chemical reactions
• Axial gas phase heat and mass transport
Since the controlling regions for blowout and breakthrough are spa-
tially separated, two compatible models, optimized for their respective re-
gions, have been developed to treat blowout and breakthrough. The HET
numerical model is employed to determine blowout and the PROF code is used
to predict breakthrough and emissions. Both codes employ compatible thermal,
transport, and chemical reaction data and formulations. They differ primar-
ily in the amount of detail included in their respective regions of applica-
tion.
228
-------
The HET code governing equations are developed by integrating the
steady two-dimensional gas phase species, mass, and energy equations across
a plane perpendicular to the axis of a channel. This results in a set of
quasi-one-dimensional equations which can be written in terms of bulk gas
properties and wall heat and mass fluxes. As previously indicated, wall
fluxes are treated by a transfer coefficient approach where the fluxes are
directly .proportional to bulk and wall gas states. The proportionality
factor varies over the bed length and is a function of channel diameter,
distance down the channel, inlet gas temperature, wall temperature distri-
bution, and flow Reynolds and Prandtl numbers. The use of transfer coeffi-
cients permits the application of efficient one-dimensional solution proce-
dures to a two-dimensional problem. Heat transport in the bed is determined
by heat conduction in the solid and radiative transfer within the channel
and outside into the upstream and downstream reservoirs. The radiative
transfer is modeled through a view factor approach, in which all sections of
the channel are able to radiatively communicate with each other and with the
upstream and downstream reservoirs.
The quasi-one-dimensional governing equations are:
Species balance in the gas phase
dY-
* *T = AWi - CwJW (1)
Species balance at the wall
Energy balance in the gas phase
i • -CA <3>
Overall energy balance
' dh .
mdT= '
229
-------
Global continuity has been incorporated into these equations, and the momen-
tum equation has been replaced by an assignment of fixed pressure. This
assignment is usually reasonable for the flow rates and channel dimensions
of interest.
Applying the transfer coefficient approach, wall species flux, JWl- ,
becomes
Similarly, wall heat flux, q.., including a term due to chemical reaction be-
w
tween the bulk and wall gas, becomes
m
_ Ml <->
qw = A CH
(h -
(6)
The heat transfer coefficient formulation, taken from Kays (Reference 5),
is applicable to circular channels with variable surface temperature, and is
given by:
CH = ReTr
where
b - 8b
V
dT.
Y+ - 2x/D h _ 'w . _ , T
X - ReTF ' b - —f ' a - Tw - Tc
(1}
and Gn and Ap are constants and eigenvalues whose magnitudes (given in Ref-
erence 5) depend on whether the flow is laminar or turbulent. This expres-
sion is valid for flows with fully developed velocity profiles and developing
thermal profiles and is assumed to be adequate for the problem of interest.
Entrance effects due to channel web thickness and developing velocity pro-
files are not considered in the code, except to define a limiting heat
230
-------
transfer coefficient at the channel entrance. The limiting value is based
on stagnating flow on the channel web.
The mass transfer coefficient Cm. is developed from the heat transfer
coefficient by:
= C
Wall reactions are given by
-EW/RT
(8)
:
(9)
where kw = Awe
oxidizer concentrations at the wall.
and £ and m are arbitrary exponents on fuel and
s at the wall.
Bulk gas phase reactions are given by
(10)
m
where y. are the stoichiometric coefficients of the reaction m, and Rm
is given by
Rm = k
fm
m
- e
(11)
and kf has the Arrhenius form kf = are"
Tm Tm
The wall radiative heat flux q at station j is given by:
r.
J
- E Kkjr: - KJ
(12)
where K is the channel segment view factor, k denotes all other stations
except j and rl, r2 denote upstream and downstream reservoirs.
Completing the definition of the problem, the boundary conditions
for equations (1) through (4) are:
231
-------
s = 0 Yi = Y0., h = h0, w
ds o
= 0, T
ds
L = °> \2
Applying a straightforward linear finite differencing technique, the
differential equations 1 through 4 are reduced to algebraic form. The
resulting algebraic equations are solved by a Newton-Raphson matrix pro-
cedure which includes a predictor-linearized corrector step. Very briefly:
1. Initial values are guessed for TW at all grid points
2. By applying known upstream conditions and the initial guessed
T's, grid point values for Y . , T, h, Yw., T h , etc. are
W i "i W W
found through Newton-Raphson solution procedures
3. Using the derivatives obtained from the solutions at each
grid point, the rate of change of wall temperatures with
respect to initially guessed wall temperatures at each
grid point is constructed
4. Assuming the system is linear, corrections to all T's are
made by applying the derivatives from step 3
5. Using the corrected TW'S as new guesses, steps 2 through
4 are repeated until guessed T 's equal corrected T 's
the ^Remixed One-dimensional Flame (PROF) code has been described
in detail elsewhere (Reference 6). Very briefly, the two-dimensional
governing equations are integrated across the channel, producing a set of
quasi-one-dimensional equations similar to the HET model formulation.
These equations are:
Species
dY.
.
' dT (AJi ) - Vw. 04)
232
-------
They differ from the HET equations by the addition of axial gas phase dif-
fusion, J.j, and heat conduction i J^h.. + k (dT/ds) terms, where 0. is given
by 1
!!!i+!i(*.F
. ds M I Hs ri
(16)
where the binary diffusion coefficient, Dj., is replaced by the bifurcation
approximation (Reference §)
and
The development of Equation (16) is given in Reference 6.
The inclusion of the axial diffusion terms makes the PROF model a
multivariable (Y.,T) boundary value problem. A difference between the PROF
and HET models is the assignment of wall state (YWl- , Tw) in the PROF model,
whereas in the HET model the wall state is calculated as part of the solution.
For PROF calculations, the wall state is found by running the HET code for
the same inlet flow conditions. The predicted wall state is then input as
a boundary condition into the PROF code. Additional boundary conditions for
PROF calculations are the initial gas composition, pressure, arid temperature;
and the condition of no heat and mass diffusion at the downstream boundary.
The PROF code can handle many chemical species, including those which
model detailed combustion and pollutant formation mechanisms. As discussed
in' Reference 6, solving the species equations, including chemical production
terms, Rm, is difficult. The PROF code has been optimized so that chemical
reaction rates -- from nearly equilibrated to inactive -- can be handled
reliably, accurately, and efficiently. This has been demonstrated by compari-
son of detailed free flame species predictions and data in Reference 9.
233
-------
Also, a favorable comparison in Reference 9 of PROF predictions of flame
quench in small diameter tubes and data demonstrates the accuracy of the
code when applied to confined flame problems such as catalytic combustors.
To reduce the PROF differential equations to algebraic form, straight-
forward linear finite differencing is used. The resulting system of simul-
taneous equations are solved by a predictor-linearized corrector solution
procedure which consists of the following steps:
1. Initial values are selected for YI , T, h at all grid points.
These may be output from a prior run or may be generated by a
linear interpolation between initial and guessed final values.
2. By applying known upstream conditions and the initial guessed
downstream values, grid point values for Y., h, T, etc., are
found through matrix solution of the equation set
3. When the downstream boundary is reached, the no-diffusion bound-
ary condition is applied.
4. Using the derivatives of YI obtained from chemistry solutions at
all grid points, the rates of change of all Y.'s with respect to
initially guessed Y.'s at each grid point are constructed
5. Assuming the system is linear, corrections to all Y^'s are made
by applying the derivatives from Step 4
6. Using the corrected Y.'s as new guesses, Steps 2 through 5 are
repeated until the guessed Y. equals the corrected Y.
HET code predictions illustrating the effect of bed channel diameter,
preheat gas temperature, mixture ratio, conductivity (and/or void fraction)
and surface activity on the important bed operating characteristics of
blowout are presented. Also, PROF code predictions which show the effect
of channel diameter and preheat gas temperature on the important bed operat-
ing characteristic of breakthrough are given. These results have signifi-
cant system design implications for achieving catalytic combustors with high
heat release (high blowout limit) and low emissions (high breakthrough limit),
234
-------
Numerical results presented in Figures 4 through 9 use a surface
reaction rate based on an assumed activation energy and an experimentally
found blowout condition. Methane fuel and air are assigned as initial re-
actants in all of the calculations. For methane fuel on a platinum catalyst,
Anderson (Reference 7) experimentally found an activation energy of 96
kj/mole. This value of activation energy was used in all of the calcula-
tions. The pre-exponential factor was obtained by matching predicted blow-
out mass flowrates to experimentally found values for a bed operating on
methane fuel with 0.00635 m diameter channels, initial gas preheat of 672 K
and 193 percent excess air. The preexponential factor, AW, found using this
approach is 6.5 * 106 mole/m2/sec for 1 and m in Equation 9 assumed to be
one and two, respectively. This preexponential rate factor incorporates
effects of surface area, catalyst dispersion, catalyst activity, etc. and
is a global rate for the experimental support/catalyst system. All calcula-
tions presented below are for a pressure of 101.325 KPa.
Figure 4 gives the MET code predicted distribution of wall and bulk
gas fuel concentration and temperature through a monolith bed. Initial
gas conditions and bed geometrical and material properties are listed on
the figure. These results are typical of catalytic combustor operation at
high mass throughput and graphically illustrate how blowout occurs.
Close to the channel entrance radial heat transport and radiative
heat losses are large, causing the surface temperature to be much lower than
the adiabatic flame temperature. Due to the low wall temperature, surface re-
actions are much slower than the radial transport of fuel and oxidizer to
the wall, and fuel concentration at the wall is a substantial fraction of
the bulk gas value. The wall reactions are controlling heat release in this
case and the front of the bed is said to be kinetically controlled. Further
down the channel, heat losses decrease and the wall temperature rises. This
drives the wall reaction rate to much higher levels than the radial mass
transport which is now controlling heat release. This region of the bed is
said to be mass transfer controlled and any fuel reaching the wall is
rapidly consumed giving low values of fuel concentration at the wall. As
235 ,
-------
the mass flow through the channel is increased, the heat transfer coefficient
and radial heat transport away from the wall is increased and a greater por-
tion of the front of the bed becomes kinetically controlled. This is illus-
trated in Figure 5 where wall fuel concentration distributions are given
for several bed mass throughputs. These results show that the kinetically
controlled region spreads downstream as mass throughput increases. The
wall then becomes cooler and the surface chemical reactions are extinguished.
In this study, blowout is defined as the condition where the kineti-
cally controlled region sweeps down the bed and the wall reactions are ex-
tinguished. It should be noted that the movement of the kinetically
controlled region to locations downstream where the channel flow is more
hydrodynamically developed does not ease the problem of the extinguishing
of wall chemical reactions. This is due to the thermal boundary layer
initiation point moving concurrently with the kinetically controlled region,
resulting in locally high values of heat transfer coefficient which can
extinguish wall reactions.
Predictions of blowout mass throughput as a function of channel
diameter are presented in Figure 6. The upper curve in Figure 6 is for
a preheat temperature of 672 K and the lower curve is for a preheat tempera-
ture of 550 K. The parameters held constant for these calculations are
listed on the figure. These include mixture ratio, flow area, wall conduc-
tivity (or product of conductivity and wall solid cross sectional area) and
surface emissivity. The dashed curve at the top of the figure is a constant
Reynolds number of 2,000 line. For fully developed pipe flow, this curve
represents an approximate upper limit for purely laminar flow. Between
Reynolds numbers of roughly 2,000 and 10,000 is a transitional flow regime
where part of the tube flow is laminar and part is turbulent. If the length-
to-diameter ratio of the channel is less than 50, entrance effects and the
transition of laminar to turbulent flow in the developing boundary layer
must be considered.
The point of transition to turbulent flow within the channel depends
on the entering freestream turbulence level, disturbances due to entrance
geometry, rate of wall heating and roughness of tube wall. If the channel
length is short and disturbances due to entrance effects and roughness are
236
-------
not severe, laminar flow can be maintained within the entire channel for
values of Reynolds numbers based on channel diameter much above the fully
developed 2,000 value. For the geometries and flowrates of interest, fully
developed flow is never achieved in the channel and Reynolds numbers do not
exceed the 2,000 limit by a large amount. Therefore, laminar flow should
prevail for most of our cases of interest.
The mass throughput curves in Figure 6 show that blowout increases
almost linearly as tube diameter increases for both the 550 K and 672 K pre-
heat cases. This is primarily the result of the heat transfer coefficient,
for fixed mass throughput, decreasing with increases in diameter of the cells
and in thickness of the web. Increasing the channel diameter permits more
mass to pass through the channel before blowout occurs. However, increasing
diameter also decreases the mass transfer coefficient which reduces fuel con-
version efficiency. A longer bed is then required to convert all of the fuel
to combustion products. Comparison of the two curves shows that preheat has
a very strong influence on blowout. This is even more dramatically demon-
strated in Figure 7 where the channel diameter is held constant at 0.003175
meters along with all the other parameters and gas preheat temperature is
varied. These predictions show that preheat has a very strong influence on
blowout with higher preheat producing more than a proportionate increase in
blowout mass throughput. This is due to the "activation" nature of the wall
reaction rate, which is an exponential function of wall temperature. These
results indicate that, for maximum heat release, beds should be operated at
as high a temperature as is compatible with the degradation of the catalyst
or is acceptable from an NO emissions point of view.
Figure 8 shows the effect of mixture ratio on blowout. The upper
curve is for a preheat temperature of 672 K and the lower a preheat of 550 K.
Numerical values of the other parameters held fixed during the calculations
are listed on the figure. Both curves show a rapid rise in blowout mass flow-
rate as the amount of excess air is decreased. In Figure 8, the parameter
controlling the high blowout mass throughput rates is surface temperature.
For low excess air levels and no wall cooling, surface temperatures (~2200 K
at 0 percent excess air) are very high, exceeding present bed material
operating limits. These high temperatures drive the surface chemical
237
-------
reaction rates to very high levels which far exceed radial (and radiative)
heat losses under laminar flow conditions. Large mass flowrates are needed
to produce blowout at these conditions. Once again, the strong influence
of wall temperature on blowout is evident.
For the 50 percent excess air, 550 K preheat case, the blowout
Reynolds number based on tube diameter is in the transitional flow regime.
To investigate the effect of fully turbulent flow on blowout, the MET tur-
bulent developing boundary layer heat transfer coefficient model was acti-
vated for the 50 percent excess air case. These predictions show that the
high turbulent heat transfer coefficient forces the blowout mass throughput
down to very low values where laminar flow would prevail. From these re-
sults, it may be conjectured that as mass throughput approaches a value such
that the developing channel boundary layer transists from laminar to tur-
bulent flow within the channel, the surface reactions in the downstream
turbulent flow portion of the tube could be extinguished.
Figure 9 gives the effect of bed material conductivity on blowout.
Since the solid cross sectional area enters the governing equations coupled
with conductivity, the variation of blowout with conductivity can also be
interpreted as blowout variation with solid cross sectional area for fixed
conductivity. The blowout trends in Figure 9 show that the blowout mass
throughput varies almost linearly with conductivity. However, at low values
of conductivity, the blowout limit levels off and reaches a constant value
for no wall conductivity. It should be noted that radiative heat transfer
is included in these calculations and, therefore, the zero conductivity
calculation does not represent adiabatic conditions.
Figure 10 gives the effect of wall activity on blowout. Variations
in wall activity model the effect of increasing surface area, catalyst load-
ing and dispersion on blowout. Since the exponential factor in the catalyst
rate expression was held fixed during these calculations, the results repre-
sent a single catalyst whose amount and distribution on a monolith bed has
been varied. Results in Figure 10 show that the effect of surface activity
on blowout is nearly linear.
238
-------
The results of the parametric blowout calculations are summarized in
Table 1. These results indicate that for maximum mass throughput, surface
temperature should be as high as is compatible with the support/catalyst ma-
terial combination and the channel diameter should be large. However, large
diameter channels also have poor fuel conversion performance and long beds
are needed to convert all the fuel to combustion products by wall chemical
reactions. This dilemma of high mass throughput but poor fuel conversion
for large diameter channels can be solved by adding additional beds behind
the first bed to efficiently convert the remaining fuel. Blowout should not
be a severe problem for these additional beds because the entering gases are
highly preheated. The next section addresses .the fuel conversijon or break-
through problem.
TABLE 1. EFFECT OF PARAMETER CHANGES ON BLOWOUT
Parameter
Effect of Increase on Blowout
Comments
Channel diameter
Gas inlet
temperature
Initial fuel/air
mixture ratio
Conductivity
Surface
activity
Linear increase
Exponential increase
Exponential increase
in lean systems
Linear increase
Linear increase
Same type of
behavior as temper-
ature
No variation as
conductivity goes
to zero
Breakthrough
As indicated previously, large diameter channels increase blowout
mass throughput but decrease fuel conversion by wall chemical reactions.
Therefore, small diameter channels should be used to minimize unburned fuel
emissions. However, blowout, channel mechanical forming, and pressure drop
considerations limit the minimum channel diameter that can be applied.
239
-------
Complete fuel conversion by wall chemical reactions in channels of practical
size requires long beds. To minimize bed length, homogeneous chemical re-
actions must be activated to rapidly consume any fuel remaining in the bed.
As discussed previously, homogeneous gas phase reduction of fuel is
postulated as "flame-like" in nature. The PROF predictions in Figures 11 and
12 support this postulate by illustrating the importance of upstream diffusion
of reactive chemical species and heat on homogeneous reactions. Boundary con-
ditions for these calculations are listed on the figures, and Table II gives
the elementary chemical kinetic reactions and associated rates applied in the
calculations. The prediction represented by the dashed curve in Figure 11 in-
cludes both diffusion and chemical reactions. This prediction shows a rapid
decay of fuel concentration, indicative of "flame-type" phenomena. In Figure 11
predictions represented by the solid line and circular symbols include only
chemical reactions or diffusion, respectively. These predictions show the less
rapid fuel decay characteristic of wall reactions only. The effect of includ-
ing both diffusion and chemical reactions can also be seen in Figure 12. This
figure shows that predictions which include both diffusion and chemical
reaction have a rapid rise in temperature indicative of "flame-type" phe-
nomena whereas other calculations show a much slower rise to the final
temperature, indicative of wall reactions. The "flame-type" nature of the
homogeneous fuel concentration reduction is further shown in Figure 13,
which presents detailed species concentrations through the channel. From
this figure it can be seen that the rapid decay of Cfy is accompanied by an
increase in free radicals (0 atoms for example) and production of CO and
H£. The CO and \\% are subsequently oxidized to C02 and 1^0. This sequence
of events is very similar to those which occur in free methane/air flames,
and demonstrates the "flame-type" nature of the processes occurring within
che catalytic combustor. It may be concluded that, to accurately predict
breakthrough, the analytical model must include the effects of axial heat
and mass diffusion, as well as homogeneous chemical kinetic reactions.
PROF code predictions demonstrating the effect of channel diameter
on breakthrough are given in Figure 14. Axial heat and mass diffusion,
as well as chemical kinetic reactions, are included in these and all sub-
sequent calculations. The boundary conditions for this case are listed on
240
-------
the figure and chemical kinetic reactions and rates are given in Table II.
Figure 14 shows that, as channel diameter is decreased, the rapid fuel
decay region associated with ''flame-type" phenomena moves towards the front
of the bed. This is due to the acceleration of bulk gas heating through
the increase in heat transfer coefficient which accompanies reductions in
channel diameter. Examining detailed computer printout shows that for this
100-percent excess air case, the "flame-type" phenomena is initiated at a
channel bulk gas temperature of approximately 1400 K. These results show
that wall reactions play an important role in preheating the gases to tem-
peratures sufficiently high to light off the "flame-type" phenomenon. The
importance of gas preheat is shown in Figure 15, where predictions of fuel
concentrations for a fixed channel diameter and several inlet bulk gas mix-
ture temperatures are presented. As can be seen, preheating the inlet gas
to higher temperatures causes the "flame-type" rapid fuel decay region to
approach the front of the bed. For these cases the "flame-type" phenomenon
is initiated at a bulk gas temperature of approximately 1400 K.
In summary, the parametric breakthrough calculations show that initia-
tion of "flame-type" phenomena in catalytic combustors requires high channel
bulk gas temperatures. These temperatures can be achieved by a combination
of wall reaction heating, which is a function of channel diameter, and/or
inlet gas preheat. Once the "flame-type" phenomena are active, rapid decay
of fuel and fuel fragments, characteristic of free flame behavior, is achieved
If "flame-type" phenomena are active, only short bed lengths are needed to
reduce unburned fuel and fuel fragments to extremely low concentrations.
CONCLUSIONS
The PROF and MET codes have been successfully used to characterize
catalytic combustor performance. HET code predictions indicate that blow-
out mass throughput increases roughly linearly with increases in channel
diameter, conductivity, and catalyst/support surface activity. Also, blow-
out increases roughly exponentially with increases in inlet mixture pre-
heat temperature and fuel/air ratio for lean operation. Therefore, for
maximum catalytic combustor mass throughput, surface temperature should be
as high as is compatible with the support/catalyst material combination and
241
-------
channel diameter should be large. Maximum channel diameter, however, is
limited by fuel conversion requirements.
PROF code predictions show that homogeneous chemical kinetic phe-
nomena in catalytic combustors are "flame-like" in nature and proper model-
ing of this effect requires treatment of axial heat and mass diffusion as
well as homogeneous chemical kinetic reactions. Predictions which include
axial diffusion indicate that the rapid decay of fuel, associated with
"flame-type" phendmena, moves towards the front of the bed as channel diam-
eter is decreased and initial preheat temperature is increased. These re-
sults show that high gas temperatures, produced by either wall reaction
heating or high preheat, are needed to "light off" the "flame-type" phe-
nomena in catalytic combustors.
Predictions suggest that a catalytic combustor system which has high
mass throughput (high blowout limit) and low emissions (no breakthrough)
could be constructed by joining two or more bed segments in series. The
first segment would have channels large enough to prevent blowout and yet
small enough to convert sufficient fuel to meet the preheat/blowout re-
quirement of the second bed segment. The second segment would have smaller
diameter channels to convert more of the fuel to products and further heat
the gases. The last segment would have very small diameter channels and the
entering gas preheat would be sufficient to "light off" homogeneous "flame-
type" phenomena. Any fuel remaining in this segment would be rapidly con-
sumed by homogeneous reactions. This system design, called the graded cell
concept, is described in Reference 8, where tests have shown this system to
have very high mass throughputs and heat release while maintaining very low
unburned hydrocarbon, CO, and NO emissions.
/\
242
-------
REFERENCES
1. Votruba, J., Sinkule, J., Hlavacek, V. and Skrivanek, J., "Heat and
Mass Transfer in Monolithic Honeycomb Catalysts-I.", Chemical Engi-
neering Science, 1975, Vol. 30, pp. 117-123, Pergamon Press,
Great Britain.
2. Cerkanowicz, A. E., Cole, R. B. and Stevens, J. G. "Catalytic Com-
bustion Modeling; Comparisons with Experimental Data," ASME paper
77-GT-85, presented at the ASME Gas Turbine Conference, Philadelphia,
Pennsylvania, March 27-31, 1977.
3. Young, L. C., and Finlayson, B. A., "Mathematical Models of the
Monolith Catalytic Converter; Part II. Application to Automobile
Exhaust," AIChE Journal, Vol. 22, No. 2., pp. 343-353, March 1976.
4. Heck, R. H., Wei, J., and Katzer, J. R., "Mathematical Modeling of
Monolithic Catalysts," AIChE Journal, Vol. 22, No. 3, pp. 477-484.
5. Kays, W. M., Convective Heat and Mass Transfer, McGraw Hill, New
York, 1966.
6. Kelly, J. T., and Kendall, R. M., "Premixed One-Dimensional Flame
(PROF) Code Development and Application," Proceedings of the 2nd EPA
Stationary Source Combustion Symposium, Volume IV, EPA-600/7-77-073d,
July 1977.
7. Anderson, R. B., Stein, K. C., Feenan, J. J. and Hofer, L.J.E.,
"Catalytic Oxidation of Methane," Industrial and Engineering Chemistry,
Vol. 53, No. 10, pp. 809-812, October 1961.
>8. Kesselring, J. P., Krill, W. V., and Kendall, R. M., "Design Criteria
for Stationary Source Catalytic Combtistors," Western States Section/
The Combustion Institute Fall Meeting on Catalytic and Fluidized Bed
Combustion, Paper Number 77-32, Stanford, California, 17-18 October
1977.
9. Kendall, R. M., and Kelly, J. T., "Premixed One-Dimensional Flame
Code (PROF) - Its Formulation, Manipulation, and Evaluation,"
Aerotherm Report TR-75-158, July 1975.
243,
-------
TABLE 2.
CH4 COMBUSTION CHEMICAL KINETIC REACTIONS AND RATES
ro
KINETIC REACTION OAT*
M/WBCR OF REACTIONS*
REACTION
1
?
3
4
5
6
7
A
9
10
11
1?
13
14
:5
16
17
10
19
20
y\
??
?S
20
25
?6
27
2*
CH4 +OH + • «>
CH4 +H 4 -->
CH4 4Q 4 -->
CH3 +0 4 -->
CH3 +02 4 -->
CH20+ +« — >
CH20+OH 4 ->>
CH20+0 4 -->
CM204H 4 -•>
CMO 402 4 -->
CMO + OM + -->
CHO +0 + -->
CMO + +1 -->
CO +OM + -->
CO +0 +H — >
MO? +0 + -->
M02 +OH + -->
HO? +H + -->
HO? 4H 4 -->
H 4Q2 41 -->
H +02 + -->
0 ' 4H2 4 -->
OH 4H2 4 -->
OH 40H 4 -->
H 40H 4M >->
0 4H 4« -->
M 4H 4f* •->
0 40 4*1 -->
CH3 4H20
CH3 4H2
CH3 +OH
Cfi20+H
CM20+OH
CO +H2
CHO +H20
CHO 40H
CMO +M?
CO +M02
CO +H20
CO +OH
CO +H
C02 +H
C02 +
C? +OH
02 +M20
OH +f*H
02 +H2
MO? +
OH +o
OH +H
H20 +H
H20 +0
H20 «
OH +
H2 +
P2 +
PRE FXP FACTOR
.1000+lH
.3000+15
.2000+1*
.iono+i3
.?000417
.2500+1M
.1700+14
.3UOO+14
.1000*11
.2000+13
.36PO+19
.2500414
.25PP+15
.?oon+i6
.2200+15
.1700+14
.??no+l4
.f.oon+13
.7000*20
.••ono+19
.?000+20
.4000*19
TEMP EXP ACTIVATION ENERGY
INDIVIDUAL THIRD BOOT EFFIC.
.000
.000
.000
.000
.000
.ono
.000
.000
.000
.000
.000
.500
.900
.000
-1.000
.000
.000
.000
.000
.000
.000
.000
.000
.ono
-1.000
•1.000
-1.000
-1.000
6.0000
11.9000
6.9000
3.3000
15.0000
35.0000
1.0000
.0000
3.0000
.0000
.0000
.0000
20.6000
l.OBOO
2.5000
.0000
.0000
2.0000
.0000
.8700
16.0000
9.4600
S.2000
.7AOO
.0000
.oono
.0900
.0000
H90
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
20.000
.000
.000
.000
.000
,.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
".noo
.030
.noo
.000
.000
.000
.000
.000
.noo
.000
.000
.noo
.000
.000
.noo
.000
.noo
-------
Surface Reaction
Controlled
Gas Phase Reaction
Control ted
ro
4s>
en
m
\ KSS$^^^m^^^
^radiation
K.|W,TW Wall reaction
^conduction
Figure 1. Physical events in a monolith cell.
-------
ro
-p»
en
M
CoM stable
solution
Figure 3. Simplified mass balance solution for catalytic combustion
in a monolith bed.
-------
ro
o>
CL
0>
•"
1300
1200
1100
1000
900
800
700
650
££
o
5
4
£ 3
|
o
O
"S
Wall temperature
0.003175 m Channel Diameter
0.8 Void Fraction
0.8661 W/m/K Conductivity
100 Percent Excess Air
0.0929 m Diameter
0.8 Wall Emissivity
0.04 kgm/s Flowrate
Bulk Fuel Concentration
Bulk Temperature
Wail Fuel Concentration
0.1 0.2 0.3 0.4 0.5 0.6
Distance (m x 102)
0.7 0.8 0.9 1.0
Figure 4. Wall and bulk gas temperature and fuel concentration through bed.
-------
CO
4
I 3
2
2
•S 1
o
O
0.003175 m Channel Diameter
0.8 Void Fraction
0.8661 W/m/K Conductivity
100 Percent Excess Air
0.0929 m Diameter
0.8 Wall Emissivity
.075 kgm/s Mass Flow
0.1 0.2 0.3 0.4 0.5 0.6 0.7
Distance (m x 102)
0.8 0.9 1.0
Figure 5. Wall fuel concentration distributions for several flowrates.
-------
ro
0.14
0.12
i 0.10
O)
CO
0.08
0.06
0.04
0.02
0.8 Void Fraction
\ 0.8661 W/m/K
\ 193 Percent Excess Air
\ 0.0929 m Diameter
0.8 Wall Emissivity
\
\
\
\
\
672K
Re = 2000
550K Tinlet
0.2 0.4 0.6 0.8 1.0
Channel Diameter (m x 102)
1.2
Figure 6. Blowout mass throughput for various channel diameters.
-------
ro
en
o
0.14
0.12
O.10
o
*
Q.
0)
O
CA
CO
0.08
0.06
0.04
0.02
300
0.003175 m Diameter Channel
0.8 Void Fraction
0.8661 W/m/K Conductivity
193 Percent Excess Air
0.0929 m Diameter
0.8 Wall Emissivity
400
500 600
Inlet Temperature (K)
700
800
Figure 7. Blowout mass throughput for various gas preheats.
-------
0.3
O
CD
CO
I 0.2
Q.
TO
ro
tn
0.1
-------
ro
tn
ro
14
12
10
O) Q
:* o
O.
O)
I
H
i
6
0.003175 m Diameter Channel
0.8 Void Fraction
193 Percent Excess Air
0.0929 m Diameter Bed
0.8 Wall Emissivity
550K Inlet Temperature
0.2 0.4 0.6 0.8 1.0
Conductivity (W/m/K)
1.2 1.4
Figure 9. Blowout mass throughput for various bed
conductivities.
-------
ro
en
CO
^ ^
§
o
X
u
12
10
8
O)
3 6
Q.
0)
14
£
$
co 2
200 400 600 800 1000 1200 1400
(mole/m2/sx104)
Figure 10. Blowout mass throughput for various surface reaction activities.
-------
o
I
ro
Gas Phase Reaction and Diffusion
No Diffusion-Gas Phase Reaction
0 No Gas Phase Reaction-Diffusion
0.00325 kg/sec Mass Flow
0.003175 m Channel Diameter
550K Preheat Temperature
100 Percent Excess Air
7 8 9 10
Distance (mx 100)
Figure 11. Bulk gas fuel concentration through bed.
-------
ro
en
01
1600
1500
1400
1300
^ 1200
? 1100
3 1000
K-W 900
800
700
600
500
Gas Phase Reaction and Diffusion
o No Gas Phase Reaction-Diffusion
No Diffusion-Gas Phase Reaction
0.00325 kg/sec Mass Flow
0.003175 m Channel Diameter
550K Preheat Temperature
100 Percent Excess Air
2345
Distance(mx 100)
Figure 12. Bulk gas temperature through bed.
-------
0.1
ro
CTJ
01
0.01
CO
,2
1 0.001
o>
u
o
O
0.0001
0.00001
CO,
0.00325 kg/sec Mass Flow
0.003175 m Channel Diameter
550K Preheat Temperature
100 Percent Excess Air
12345
Distance (mx 100)
Figure 13. Detailed species concentrations
through bed.
-------
mo
OB
ro
en
10
-1
10
-2
| 10
8
o
O
"o>
u.
-3
10
-5
10
-6
0.0064 m
0.00325 kg/s Mass Flow
550K Preheat Temperature
100 Percent Excess Air
d = 0.0046 m
0.0032 m
01 23456
Distance (mx 100)
Figure 14. Effect of channel diameter on
breakthrough.
-------
rv>
tn
00
10-2
(Q
| 10'3
o>
o
c
o
^ 10'4
0)
u.
10-5
ID'6
0.00325 kg/s Mass Flow
100 Percent Excess Air
0.0046 m Channel Diameter
Tjnlet = 530K
1234567
Distance (mx 100)
Figure 15. Effect of gas preheat temperature on
breakthrough.
-------
THE DEVELOPMENT OF CATALYTIC COMBUSTORS
FOR STATIONARY SOURCE APPLICATIONS
By:
W.V. Krill and J.P. Kesselring
Acurex Corporation
Mountain View, California 94042
259
-------
ABSTRACT
An experimental program has been conducted for the Environmental Protec-
tion Agency to develop design criteria for catalytic combustors as applied to
stationary systems. The program included catalyst screening tests from which
the graded cell concept was developed. The graded cell catalyst exhibits
greatly enhanced combustion characteristics in terms of increased maximum
throughput. Advanced testing of the graded cell catalysts showed high heat
release rate capabilities with low emissions of CO, HC, and thermal NOX. Opera-
tion of the catalysts under fuel-rich conditions showed capability to control
fuel NOx emissions. Additional criteria for system scaleup and operation under
varying preheat and pressure conditions were also generated.
Catalysts developed during the program were incorporated into three
small-scale systems with heat extraction. A radiative catalyst/watertube
system, utilizing direct heat removal from the catalyst, was devised and tested.
The concept has potential application to watertube boilers. A model gas turbine
combustor was tested at pressures between .101 and 1.01 MPa (1 to 10 atmospheres)
to investigate operating characteristics and fuel nitrogen conversion to NOX.
The final system, a two-stage combustor, was constructed to utilize fuel-rich
first stage combustion for fuel NOx control. Measured emission results make
the concept attractive for a variety of future system applications.
261
-------
INTRODUCTION
Catalytic combustors have been shown to reduce the levels of CO, HC, and
emissions over those of conventional burners in laboratory tests with both
clean and nitrogen-doped fuels. The operating conditions for these catalytic
combustors are limited by the catalyst temperature capability. Since the tem-
perature obtained through normal one-stage, low excess air operation (necessary
for high system efficiency) is outside the temperature capability of current
catalyst systems, system techniques to control temperature are required. The
development of combustors for boiler, furnace, and gas turbine systems depends
on establishing design criteria for system temperature control and related sys-
tem parameters. The EPA catalytic combustion program has focused on determining
these criteria for stationary source combustors.
System design criteria include information on the catalyst/washcoat/
substrate combination and its operational life, preheat requirement, heat
release capability, and pressure and temperature limitations. This information
was developed through the preparation of varying catalyst combinations and com-
parison of the catalysts through combustion screening. Developed catalysts were
then incorporated into small-scale systems for demonstration of energy release
and heat extraction concepts. Additional criteria were developed to operate
these systems at low emission levels with the highest achievable efficiency.
The following summarizes results for the catalyst screening and system config-
uration tests conducted under EPA Contract 68-02-2116.
CATALYST DEVELOPMENT
Preliminary Screening Tests
An initial combustion test series was conducted at the Jet Propulsion
Laboratory in Pasadena, California on systems with monolithic catalyst supports.
This test series identified the most suitable combustion characteristics for
catalysts as follows:
263
-------
• Low ignition temperature
• Low preheat requirements for sustained combustion
• Combustion uniformity throughout the bed
• High heat release capability
a High combustion efficiency
• Low pollutant emissions
• High operating temperature
• Fuel flexibility
• Long life.
Catalyst models tested included variations in substrate, washcoat, and catalyst
materials — as well as substrate geometry and preparation techniques.
Catalysts developed in the screening test series showed greatly enhanced
combustion characteristics over those initially tested. Specifically, perfor-
mance was improved by:
1. Increased catalyst loading, resulting in lower initial lightoff
temperatures, higher mass throughputs, and increased lifetime
at 1367K (2000°F).
2. Increased cell size, allowing higher possible mass throughputs
at the expense of increased hydrocarbon emissions.
4. Heavier hydrocarbon fuels which promote lightoff at lower ignition
temperatures.
5. Hydrogen sulfide G^S) fixation of platinum catalysts to promote
retention of platinum surface area.
6. Presintering of catalyst washcoats which reduces burying of
active catalyst below the surface during combustion.
7. Stabilization of Y-Al2C>3 washcoats with cesium oxide (Cs20) up
to 5 weight percent increased surface area. Stabilization of
alumina washcoats with ceria up to 5 weight percent had a
negative effect on surface area.
264
-------
8. Decreased cell size to significantly reduce unburned hydrocarbon
and carbon monoxide emissions.
9. Catalyst beds of combined large cell and small cell monoliths
also significantly increase throughput (at a given preheat
temperature) and overall catalyst life with low emissions.
10. Bed temperature uniformity was increased by operation at higher
temperatures.
Catalysts with combined large and small cell segments represent the best
concept developed by the test series. This graded cell concept (three segments
shown in Figure 1) was further developed through predictions of the PROF-HET
catalytic combustion computer code (Reference 1). Finally, preliminary data
with ammonia-doped natural gas indicated a potential for control of fuel nitro-
gen compounds to NOX under lean conditions.
Graded Cell Catalyst Tests
Additional catalyst screening tests were conducted at Acurex on the
graded cell configuration. The preliminary objective of these tests was to
identify the best catalysts for system application. Catalysts were also tested
for high temperature capability, system scaleup design criteria, and conversion
data of fuel nitrogen to nitrogen oxides under varying operating conditions.
Screening catalysts were obtained from six sources, including W.R. Grace
and Company, Universal Oil Products, Inc., William Pfefferle (a private con-
sultant), Matthey Bishop, Inc., Johnson Matthey, and Acurex. Substrate mate-
rials were either DuPont alumina or Corning zirconia spinel. Washcoats varied
from proprietary preparations with high pretest surface area to no washcoat
and a subsequent low pretest surface area. Catalysts were either precious
metal, metal oxide, or mixtures. The catalyst loadings and fuels used are
listed in Table I.
To support combustion test results, pre- and post-test catalyst physical
measurements were made. These measurements included catalyst surface area and
dispersion performed in the Acurex catalyst characterization laboratory.
Additional scanning electron microscopy (SEM) and energy dispersive analysis
by X-ray (EDAX) tests were performed at the Jet Propulsion Laboratory as required.
265
-------
Table II is presented for reference, and summarizes all surface area and dis
persion measurements.
Catalysts were compared by limited aging, maximum throughput, varying
stoichiometry, and minimum preheat tests. Some catalysts, primarily precious
metals with low loadings, exhibited lifetimes of less than 10 hours, precluding
extensive data evaluation. Others were tested to completion in a 20 to 30 hour
test sequence. In addition, some catalysts showed preferential activity for
either fuel-rich or fuel-lean combustion or for operation only at higher pre-
heats of 700 to 811K (800 to 1000°F).
The maximum throughput (volumetric heat release rate) capabilities were
compared for various graded cell catalysts to indicate relative activity.
Catalyst data selected from Table I are compared below.
MAXIMUM THROUGHPUT COMPARISON DATA**
J-_ - J -D j
Catalyst Type
UOP Proprietary
Pt-Ir/Al203
Stab. Pt/Al203
JM Proprietary
Co203/Al20s
Co203/Zirconia
Manufacturer
UOP 2
W.R. Grace
Matthey
Bishop
Johnson
Matthey
Pfefferle 3
Acurex 3
tsea
K
1522
1256-
1589
1611
1589
1617
1644
SV, 1/Hr
343,900
327,400
348,000
603,300
443,100
661,400
a' ' ' 3
q ' Hr-Pa-m
3.9 x 106
2.8 x 106
5.2 x 106
7.2 x 106*
5.1 x 106*
7.8 x 106*
Dea
Uniformity
Excellent
Ragged
Ragged
Excellent
Excellent
Excellent
Spinel
**
All results with natural gas/methane fuel
*
Bed not blown out
The first three catalysts, either precious metal or proprietary materials,
exhibited volumetric heat release rates (q"1) of approximately 3.7 x 106
Q f ' o
J/hr-Pa-m (10 x 10 Btu-hr-Atm-ft ) at blowout conditions. The latter three,
either proprietary or cobalt oxide catalysts, reached approximately 7.4 x 10
36 3
J/hr-Pa-m (20 x 10 Btu/hr-Atm-ft ) without blowing out. This heat release
266
-------
rate represented the maximum facility capability in the screening test configura-
tion. It is apparent that the cobalt oxide catalysts are capable of higher heat
,release rates than the precious metals in the graded cell configuration. The
oxides required operation at higher temperature, however, to produce combustion
efficiencies comparable to the precious metals at lower temperatures.
During graded cell catalyst testing, operating temperatures were varied
from 1367K to over 1978K (2000 to 3100°F). Thermal NOx data were obtained with
natural gas as the primary test fuel. A number of the test models are compared
in Figure 2. From 1367 to 1644K (2000 to 2500°F), little variation among the NOx
emissions with catalyst type is apparent with NO levels below 20 ppm. At 1756 to
1867K (2700 to 2900°F), the NOx production rate begins to increase, as with
conventional flame combustion but at a less rapid rate.
In Figure 2, two different catalyst geometries are shown for fuel-lean
operation in the 1644 to 1978K (2500 to 3100°F) range. The upper curve is for
a two-segment graded cell model (A-029) that acted primarily as a flame holder
for downstream gas phase reactions. The lower curve represents data for a
three-segment catalyst (A-030) where significantly increased combustion occurs
on the surface of the catalyst bed. It is apparent that maximizing the amount
of surface reaction occurring in a catalytic combustor minimizes the production
of thermal NOX. These results indicate the importance of the graded cell con-
figuration, particularly for thermal NOX control in high excess air gas turbine
applications.
In addition to baseline catalyst screening tests, selected catalysts
were tested to characterize conversion of fuel-bound nitrogen to nitrogen
oxides with catalytic combustion. Ammonia was added to natural gas to simulate
fuels of varying nitrogen content. Exhaust gas analyses for NOX by chemilumi-
nescent analyzer and for ammonia (NH3) and cyanide (HCN) by specific ion
electrode were performed.
A nickel oxide/platinum catalyst was prepared at Acurex and tested over
a range of stoichiometries from 55 to 200 percent theoretical air at a nominal
bed temperature of 2900°F. Fuel dopant concentration ranged from 2500 to
10,000 ppmv NH3 in the fuel. The results are shown in Figure 3 as the percent-
age of the incoming NH3 converted to NH3, HCN, and NO. The NH3 conversion to
NO increased from zero levels under very fuel-rich conditions to better than
267
-------
90 percent on the fuel-lean side., NH3 conversion to HCN showed the opposite
trend — high under rich conditions and decreasing to zero on the lean side.
Unconverted ammonia was highest below 70 percent theoretical air and remained
at low levels under lean combustion.
The total of these three curves (dashed line and cross symbols)
represents all NOX precursor species for the NiO/Pt catalytic combustor. A
distinct minimum occurs between 70 and 80 percent theoretical air, where only
20 percent of the fuel nitrogen is converted to NOX precursors.
A second graded cell catalyst (cobalt oxide/platinum) was tested with
simulated fuel nitrogen. The fuel nitrogen conversion is shown in Figure 4.
The ammonia conversion to nitric oxide provided the same characteristic curve
as the nickel oxide/platinum catalyst. Differences in the HCN and NH3 species
measured, however, resulted in lower total NOX precursor (NO + NH3 + HCN)
levels under fuel-rich conditions. The minimum occurred at a lower value of
theoretical air (60 percent) than that of the previous nickel oxide catalyst
(75 percent). The cobalt oxide catalyst could thus be operated fuel rich
without dilution to achieve low conversion of fuel-bound nitrogen to nitrogen
oxides.
The low fuel nitrogen conversion of these two catalysts at 60 to 80
percent theoretical air has important system implications. Combustors which
could operate fuel-rich, possibly with staging, have potential for fuel NOX
control. This characteristic of the catalytic combustor under fuel-rich
conditions is the basis for the two-stage catalytic combustor demonstrated
under this program.
Design criteria for the graded cell catalyst concept also included
scaleup, pressure, and preheat characteristics. Catalyst blowout was again
selected to compare activity. Based on the results of small-scale testing, a
Universal Oil Products catalyst was selected for scaleup. Analytical modeling
predicted that combustion throughput capability would scale proportionately
to bed frontal area. Therefore, the bed diameter was increased to provide a
2.7 increase in frontal area over the small-scale model. Bed length remained
fixed at 7.6 cm (3.0 inches).
263
-------
Test data showed that maximum throughput (at blowout) did scale approx-
imately with frontal area. Maximum throughput reported for the small-scale
catalyst was 258.5 MJ/hr (245,000 Btu/hr) and 3.42 x 106 J/hr-Pa-m3 (9.3 x
10" Btu/hr-atm-ft ) volumetric heat release rate. This compares to a volumetric
heat release of 4.38 x 106 J/hr-Pa-m3 (11.9 x 106 Btu/hr-atm-ft3) at 926.3 MJ/hr
(878,000 Btu/hr) for the scaleup catalyst at 672K preheat.
A series of blowout tests were then conducted to determine the operational
mass throughput limit of the catalyst for varying preheat and pressure conditions.
The blowout points used are shown below.
BLOWOUT DATA - CATALYST A-041
Data
Point
1
2
3
4
Bed Temp
K (°F)
1588 (2400)
1588 (2400)
1588 (2400)
1588 (2400)
Preheat Temp
K (°F)
608 (635)
478 (400)
389 (240)
603 (625)
Max. Fuel Flowrate
Kg/hr (Ibm/hr)
15.0 (33.0)
10.5 (23.1)
8.4 (18.5)
22.2 (49.0)
Pressure
MPa (atm)
.195 (1.93)
.140 (1.39)
.134 (1.33)
.301 (2.98)
Two things are evident from the data:
• Blowout scales linearly with pressure
• •
(mfuel max atm X mfuel max, 1 atm)
• Blowout for catalyst A-041 was exponential in preheat temperature,
although a relatively weak exponential factor is shown.
Operation of the catalyst was, of course, possible at any combination of
preheat and fuel flowrate below the blowout limits.
The exponential behavior of preheat temperature and its effect on blowout
was shown in PROF-HET code predictions (Reference 1). The blowout data obtained
support this prediction and can be employed as design criteria for operation of
catalyst A-041 under varying conditions.
The graded cell catalyst screening tests identified the design criteria
required for incorporating the graded cell configuration into system applications,
269
-------
Specifically, these criteria included mass throughput and heat release capa-
bilities, emissions under varying operating conditions and with nitrogen-
containing fuels, lightoff requirements, and current lifetime capabilities.
In addition, the state of the art in catalyst development has been evaluated,
including an understanding of preparation techniques, material capabilities,
and material interactions.
SYSTEM CONCEPT TESTING
The design criteria generated for graded cell catalyst configurations
were used in the specification of three small-scale systems incorporating heat
extraction techniques. A radiative catalyst/watertube system demonstrated a
stoichiometric, water-cooled combustor for boiler application. A model gas
turbine combustor utilized high excess air for catalyst and exhaust gas temper-
ature control. Finally, a two-stage combustor was constructed to demonstrate
fuel nitrogen control with application to either boiler, furnace, or turbine
equipment.
Radiative Catalyst/Watertube System
The radiative catalyst/watertube concept is shown in Figure 5. A
stoichiometric fuel/air mixture is fed to the radiative section which contains
a close-packed array of catalyst elements and watertubes. The mixture is par-
tially combusted by the catalyst which is kept at a low surface temperature by
radiation heat loss to the watertubes. The combustion products and remaining
unburned fuel and air are then passed to a downstream catalytic adiabatic com-
bustor to complete combustion reactions. A final convective section extracts
energy from the fully combusted gases. Both catalyst sections operate well
below the maximum use temperature of the catalyst supports — the radiative
section by radiative cooling and the adiabatic section by dilution of the fuel/
air mixture with exhaust products from the radiative section. The radiative
section was constructed and tested independently of the downstream adiabatic
combustor and convective sections.
An initial test series was conducted using a platinum catalyst on Coors
alumina cylinders. Stoichiometry was varied from 40 to 219 percent theoretical
air and fuel mass rate from 2.1 to 6.7 kg/hr (4.7 to 14.8 Ibm/hr) of natural gas.
270
-------
Preheat conditions were also varied. Figure 6 shows the energy extracted by the
cooling tubes out of the total available energy at the bed inlet as a function
of theoretical air. The total available energy includes the latent heat of the
fuel and the sensible preheat energy. The energy extracted by the watertubes is
essentially the same as the fuel energy release since the bed inlet and outlet
temperatures were approximately equal. At low theoretical air, the heat extrac-
tion is controlled primarily by the catalyst surface temperature, peaking at
approximately 100 percent theoretical air. As theoretical air further increases
above 100 percent, surface temperature again begins to decrease, decreasing the
radiant exchange. The higher mass throughput, however, also increases convec-
tive heating of the watertubes such that the energy exchange does not fall off
rapidly. Overall combustion efficiency at 100 percent theoretical air was cal-
culated as approximately 37 percent from the data. Significant emissions of CO
and HC were measured for the radiative section due to the incomplete combustion.
NOX levels were consistently below 2 ppm as measured.
The radiative section was also tested to evaluate fuel nitrogen conversion
characteristics of the system. For those tests, natural gas was doped with
ammonia and stoichiometry was varied from 52 to 120 percent theoretical air.
Figure 7 shows the fuel nitrogen conversion characteristics of the
radiative system for natural gas doped with 2000 ppm of ammonia. Low NOX and
high NH3 values above 100 percent theoretical air are consistent with the
incomplete combustion characteristics of the radiative section. The low point
in the total NOX precursor curve (NH3 + HCN + NOX) at 60 percent theoretical
air is similar to those obtained for metal oxide graded cell catalysts.
The actual measured heat release of the radiative catalyst/watertube
system was not as high as that predicted at the nominal 4.3 Kg/hr (9.5 Ibm/hr)
methane flowrate design condition. Therefore, the radiant section as tested
is not fully suited for complete system development. The addition of the down-
stream adiabatic catalytic combustor results in too high a temperature in that
region at stoichiometric conditions since combustion efficiency in the first
stage is not as high as expected. The radiative catalyst/watertube section did
exhibit excellent performance at stoichiometric conditions with very low levels
of NOX. The potential for control of fuel nitrogen conversion and the extremely
stable operation experienced under all test conditions make it attractive for
future boiler applications.
271
-------
Model Gas Turbine Combustor
Since the graded cell catalyst was demonstrated to have the low preheat,
high heat release, and pressure capabilities required for gas turbine applica-
tions, a 1055.1 MJ/hr (106 Btu/hr) model turbine can and fuel injection system
was constructed. The system and catalyst are shown in Figure 8. Testing was
performed at Acurex and Pratt and Whitney Aircraft (West Palm Beach, Florida)
facilities. Acurex and UOP catalysts were prepared on either DuPont alumina or
Corning zirconia spinel support materials of varying configurations.
The model gas turbine combustor was first tested at Acurex with propane
at pressures between .101 and .354 MPa (1 and 3.5 atmospheres). A heat release
rate of 263.8 MJ/hr (250,000 Btu/hr) at 1478K (2200°F) bed temperature was run
as the nominal test condition. No significant emissions of NOX or CO were
obtained,
Pratt and Whitney test data were obtained with propane, No. 2 oil, and
No. 2 oil with 0.5 weight percent nitrogen as fuels. Heat release rates to
844 MJ/hr (800,000 Btu/hr) were achieved with low NOX emissions for both propane
and No. 2 oil. Some difficulty was encountered with flashback and flameholding
on the fuel nozzles when running No. 2 oil. High CO and unburned hydrocarbon
emissions resulted from operating at low bed temperatures (near the breakthrough
limit) to avoid flameholding. Variations in pressure were not found to affect
emission levels. Tests run with pyridine-doped No. 2 fuel oil, however,
increased the NOX emission levels, representing percentage conversions of fuel
nitrogen to NOX of 100, 61, and 55 percent for test pressures of .303, .505,
and .707 MPa, respectively.
Subsequent inspection of the test hardware showed that low liquid fuel
inlet velocities due to a fabrication error were responsible for the flashback
and flameholding.
A final test series was conducted at Acurex (after the test hardware
had been reworked) with natural gas, natural gas doped with ammonia, and No. 2
oil. Pressures from .101 to .808 MPa (1 to 8 atm) with natural gas and .101 to
.505 MPa (1 to 5 atm) with fuel oil were included. A decrease in ammonia con-
verted to NOX with pressure was observed. These results are consistent with
those obtained at Pratt and Whitney with pyridine-doped No. 2 fuel oil. Maximum
throughput for the catalyst at .303 MPa, 1422K (3 atm, 2100°F) pressure and bed
272
-------
temperature, and 561K (550 F) preheat temperature was also Investigated. At
space velocities near 200,000 per hour, the catalyst began to break through with
increasing CO and unburned hydrocarbon emissions. Nitrogen oxide emissions
remained at near zero levels throughout the test. Full blowout was not achieved
as control of catalyst temperature during breakthrough produced difficulties in
system control. The maximum heat release obtained was 615 MJ/hr (583,000 Btu/hr).
Final tests were conducted with diesel fuel to compare emissions with
those from natural gas. An increased bed temperature was maintained for the oil
tests to maintain uniform bed conditions and suppress soot formation. The NOX
levels were higher than for natural gas (15 ppm compared to 3 ppm) due primarily
to higher average bed temperatures.
The results of the model gas turbine testing demonstrated the application
of the graded cell concept in a system similar to current turbine combustor designs,
High mass flowrates were achieved in a relatively small volume combustor. Overall
pressure drop for the combustor and fuel injector were measured at less than one
percent at .303 MPa (3 atmospheres) test pressure.
Two-Stage Combustor
The two-stage catalytic combustor is attractive for two reasons. First,
it allows control of bed temperatures to those compatible with the support mate-
rial without large excess air requirements. Second, the first stage can be
operated fuel-rich, which has been shown to be advantageous for reduced conver-
sion of fuel nitrogen to nitrogen oxides. A two-stage combustor was designed
and constructed to demonstrate these concepts.
The two-stage combustor is shown schematically in Figure 9. A fuel-rich
mixture is introduced into the primary stage which contains a graded cell cata-
lyst bed. The fuel is partially combusted, and the released energy is removed
by an interstage heat exchanger. Sufficient secondary air is then injected into
the combustion products to complete combustion of the remaining fuel in the
second stage. The full system combustor would also include a second heat
exchanger to remove the combustion energy released in the second stage.
The two-stage combustor containing two cobalt oxide catalysts was tested
with natural gas at .101 MPa and .202 MPa pressures (1 and 2 atmospheres).
273
-------
Lightoff and steady-state operation presented no unusual control problems. Some
thermal and fluid dynamic interactions between stages were apparent, however.
The combustor was tested at an overall stoichiometry varying from 70 to 150
percent theoretical air at a nominal fuel flowrate equivalent to 211 MJ/hr
(200,000 Btu/hr) heat release rate. The first stage stoichiometry was varied
from 40 to 70 percent theoretical air. Ammonia was added to the natural gas
fuel at a rate of 0.2 to 0.4 percent.
Bed temperatures ranged from 1256K to 1660K for a relatively constant
preheat of 617K (650°F). The energy extracted in the interstage heat exchanger
represents 50 to 60 percent of the combustion energy generated in the first stage.
The results of the fuel nitrogen conversion data are shown in Figure 10
as a function of overall combustor stoichiometry. The data show that when
operating above 100 percent theoretical air, only nitrogen oxides are normally
present. Under overall fuel-rich conditions, fractions of ammonia and cyanide
were also present. These results are consistent with fuel nitrogen data obtained
on the single cobalt oxide catalyst (model A-037, Figure 4), The data in
Figure 10 show a nominal 30 percent conversion rate of fuel nitrogen to NOX
precursors with a value of approximately 27 percent near overall stoichiometric
conditions, A slight decrease in conversion was noted at .202 MPa 02 atm)
pressure.
The data shown in Figure 10 at approximately 10 percent conversion levels
varied in test conditions from the other data in two respects:
1. The first stage was operated at higher values of theoretical air
(75 percent) compared with 50 percent for the initial data, and
2. The first stage catalyst had experienced some sooting by later
test times when the data were taken, causing the catalyst to
operate at lower temperatures with incomplete combustion.
The first stage sooting of the cobalt catalyst proved to be a limiting factor in
the test life of the system. The incomplete combustion occurring at later test
times was evident by increasing measured carbon monoxide levels.
The demonstration of the two-stage catalytic combustor showed a number
of important results:
274
-------
1. The two-stage combustor is effective in controlling conversion
of fuel nitrogen to nitrogen oxides under stoichiometric and
fuel-lean conditions.
2. A slight decrease in nitrogen conversion was found at .202 MPa
(2 atmospheres) pressure.
3. The variation of first stage stoichiometry impacts overall fuel
nitrogen conversion.
4, First stage sooting of the cobalt oxide catalyst was a limiting
factor in combustor operating life.
Application of the concept to both boiler and turbine systems is possible. Con-
vective heat exchangers downstream of each catalyst stage would provide for
steam raising in boiler applications. Interstage cooling would not be required
for gas turbine systems where high excess air in the second stage could be used
to control the catalyst and exhaust gas temperatures. Further work is required
for optimization of the system and catalyst elements for specific applications.
CONCLUSIONS
As a result of this research and development program, significant progress
has been made toward developing a practical catalytic combustion system. Before
the step to demonstration can be taken, however, additional work relating to the
integration of the catalytic combustor into the total combustion system must be
performed.
Based upon the analysis and test results of this program, the design,
fabrication, and operation of catalytic combustors with high volumetric heat
release rates and low emissions have been demonstrated. Both precious metal
and oxide catalysts have been tested over a wide operating temperature range.
The precious metal catalysts should be limited to temperatures below 1589K
(2400°F) for catalyst life considerations, while oxide catalysts can be oper-
ated for long periods at temperatures above 1644K (2500°F). Catalyst perfor-
mance has been greatly enhanced through the use of graded cell monoliths and
higher catalyst loadings.
275
-------
Catalytic combustors have been shown to be effective in controlling
both thermal and fuel NOx emissions. The thermal NOX control appears to result
from maximizing surface reactions in the combustor, while fuel NOX can be
minimized by operating at a rich fuel/air ratio which minimizes the formation
of NH3, HCN, and NO, with complete combustion of CO and HC at a later time.
The maximum throughput of a catalytic combustor is a linear function of
pressure and an exponential function of preheat. Thus, for a given preheat,
the catalyst is face velocity limited in throughput ability.
Small-scale catalytic combustion system configurations have been tested
and demonstrate the feasibility of direct radiative removal of bed heat for
temperature control, two-stage catalytic combustion for temperature and fuel
SOX control, simulated exhaust gas recirculation through the use of nitrogen
iiluent for temperature control, and high excess air operation. The combustion
system concepts that have been operated show that it is possible to operate near
stoichiometric conditions with less than 10 ppm NOX and CO in a natural gas-
Eired catalytic combustor.
A number of areas in catalytic combustion need to be addressed to capital-
ize on the progress to date. Additional testing of simple and mixed oxide cata-
lysts for combustion and fuel nitrogen conversion abilities is needed, along
ffith life testing of selected catalysts to 1000 hours at various pressures.
Exploratory work with heavy fuel oils (Nos. 4, 5 and 6) and pulverized
:oal should be conducted to determine system feasibility and fuel preparation
problems. The potential of catalytic combustion in controlling NOX emissions
Erom the combustion of these fuels is great and needs early experimental veri-
fication.
Development of auxiliary systems required to interface with the catalytic
:ombustor is also needed. This includes lightoff systems, temperature control
systems, and fuel and air introduction systems. In addition, further testing
jf the radiative catalyst/watertube, two-stage combustor, and gas turbine
:ombustor systems is needed to more thoroughly define operating ranges with a
variety of fuels. .
276
-------
Finally, the design, fabrication, and operation of a demonstration unit
should be undertaken when the above work is completed. The demonstration unit
would be operated as a laboratory device for several months prior to the initi-
ation of field demonstration tests.
277
-------
REFERENCES
1. Kelly, J.T., £t al., "Development and Application of the PROF-HET
Catalytic Combustor Code," Paper No. 77-33 presented at the Western
States Section of the Combustion Institute, October 1977.
2. Kesselring, J.P., et_ al., "Design Criteria for Stationary Source Catalytic
Combustion Systems," Acurex Final Report 78-278, March 1978.
278
-------
TABLE I. GRADED CELL CATALYST MODEL S
Sample No.
AERO-025
AERO- 026
AERO-027
AERO-028
AERO-029
AERO- 030
AERO-031
AERO- 032
AERO-033
AERO- 034
AERO- 035
AERO-036
AERO-037
AERO-038
AERO-040
AERO-041
No. of
TC
9
6
6
-
6
6
E
6
7
7
7
7
4
5
5
4
Substrate
Nanii. Type
DuPont
DuPont
DuPont
DuPont
Corning
Corning
OuPont
DuPont
Coming
Corning
DuPont
Corning
Coming
DuPont
DuPont
DuPont
Alumina
Alumina
Thorla Im-
pregnated
alumina
Zlrconla Im-
pregnated
alumina
Zlrconla
spinel
Zlrconla
spinel
Alumina
Alumina
Zlrconla
spinel
Zlrconla
spinel
Alumina
Zirconia
spinel
Zlrconla
spinel
Alumina
Alumina
Alumina
Mashcoat
Kami. Type
Grace
UOP
None
None
None
None
Hatthey
Bishop
Natthey
Bishop
None
Oxy-
Catalyst
Matthey
Bishop
Aero
None
None
Johnson
Matthey
UOP
Rare earth
stabilized
alumina
10-18 Wt. I
Proprietary
_
_
_
_
Proprietary
Proprietary
_
Y-Alumina
4 Ht. t
Proprietary
Zlrconla
Magnesia
.8-1.0 Wt. t
_
_
Proprietary
Propri etary
Catalyst
Nlnu. Type
Grace
UOP
H. Pfefferle
«. Pfefferle
M. Pfefferle
Aero
Natthey
Bishop A
Natthey
Bishop B
Aero
Aero
Hatthey
Bishop C
Aero
Aero
H.Pfefferle
Johnson
Hatthey
UOP
Pt/Ir
Proprietary
Pt/Ir/Os
Pt/Ir/Os
NIO/Pt
C0203/Pt
Stab. Pt
Stab. Pt
N10/Pt-Pd
CojOa/Pt
Stab. Pt
NIO/Pt
Co,0,/Pt
C O
Co203
'ropri etary
Proprietary
Ht. I
0.6-1.0
-
0.29
0.29
0.29-0 t
.29/.46f
2.9 N10
2.4-0 Pt
5.4-7.8
Co203
.33-. 76
.85-1.09
.67-0 Pt
2.2X.94/
2.0 N10
2.2-0 Pt
9.3/S.6/
4.3 C0203
1.96/1.2/
0.9
1.6/2.V
2.1 N10
1.2/0.9/
0 Pt
2.7/2.S/
3.4 Co,0,
4.0/0/0 Pt
15.9/4.1/5.;
-
-
OK
.47/.76/.80
-
.36/.24/.17
.52/.08/.08
O.S/.31/0
.5/. 8/5.0
4.1/0/0
_
-
15/0.7/0 Pt
5.0/2.1/
4.5 N10
3.0/0/0 Pt
7.7/9.9/
20.8 Co203
_
3.0/3.2/
4.0 N10
2.0/1.4/
0 Pt
4.5/4.S/
10.1 Co.O,
6.7/0/O'Pl
10.3/2.9/8.8
-
-
Dates
Tested
4/25 - 4/29/77
6/24 - 6/29/77
6/30 - 7/1/77
_
7/10 - 7/11/77
7/14 - 7/21/77
8/2 - 8/10/77
7/26 - 7/29/77
9/14 - 9/20/77
9/22 - 9/27/77
10/1 - 10/6/77
10/10 - 10/12/7!
12/5-12/23/77
12/28 - 12/29/77
1/25 - 2/6/78
12/30/77-1/3/78
Fuel
Nat. Gas
Nat. Gas
Nat. Gas
-
Nat. Gas
Nat. Gas
Nat. Gas
Nat. Gas
Nat. Gas
Nat. Gas
Nat. Gas
Nat. Gas
Nat. Gas
a
Methane
Nat. Gas'
Nat. Gas
Nat. Gas
Test Purpose
Screen Grace Pt/Ir catalyst
Screen UOP catalyst
Determine effects of high melting point precious metals and
catalyst without washcoat
Not to be tested based on results of A-027.
Investigate metal oxide catalyst capabilities (no
washcoat). Perform high temp, operation (3100°F).
Metal oxide comparison, extensive evaluation
Compare to A-032
Screen Hatthey Bishop catalyst
1
Fuel nitrogen and pressure testing 1 Test difficulties
\ precluded data
1 results
Comparison to A-033 )
Screening comparison
Fuel nitrogen testing
Fuel nitrogen and pressure testing
Investigate catalyst-support interactions
Screen Johnson Matthey catalyst
Catalyst scaleup, 6. 06-inch diameter
ro
-------
TABLE II. SURFACE AREA AND DISPERSION MEASUREMENTS ON GRADED CELL CATALYSTS
ro
CO
o
Sample No.
AERO-025
AERO-026
AERO- 027
AERO-028
AERO-029
AERO-030
AERO-031
AERO- 03 2
AERO-033
AERO-034
AERO-035
AERO-036
AERO-037
AERO-038
AERO- 040
AERO-041
Surface Area m2/g
Pretest Post-test
1.55-2.49
5.94
0.44
0.06
0.60
0.15
24.38
11.99
-
-
5.17
-
-
0
4.00
6.37
0
0
-
-
0
.01
0
0
-
-
0.09
—
-
-
0
0.50
Dispersion %
Pretest Post-test
1.5-4.9
20.64
8.33
-
-
36.16
9.11
-
-
4.09
-
-
-
-
0
-
-
—
-
0
0.20
-
-
0
—
-
-
—
Catalyst
Pt/Ir
Proprietary
Pt/Ir/Os
Pt/Ir/Os
NiO/Pt
Co203/Pt
Stab. Pt
Stab. Pt
NiO/Pt-Pd
Co203/Pt
Stab. Pt
NiO/Pt
Co203/Pt
Co203
Proprietary
Proprietary
SEM/EDAX Results and Comments
No Pt or Ir found on back two
segments by SEM/EDAX
Not combustion tested
Invalid test data
Invalid test data
Zero surface area on each segment
-------
rv>
CO
ummmmmw
• •••»*!"
Figure 1. Corning square-celled extruded monolith structures.
-------
Test Model
350
300
250
o 200
150 -
-------
lOOi-
80
ro
co
co
o>
u
tu
Q.
X
o
-------
100 r
ro
o>
-P=.
40
60
Catalyst A-037 (Co203/Pt)
80 100 120 140
Theoretical air, percent
160
Figure 4. NHs conversion characteristics, natural gas doped with ammonia
Co203/Pt catalyst.
2 atm
3 atm
180
-------
• Steam
drum
-Refractory lining
ro
CO
01
Catalyst
coated cylinder
Mud
drum
Radiative heat
transfer section
_ Monolith bed
- Adiabatic
Combustor
Convective heat exchanger
Watertube
Figure 5. Radiative watertube boiler concept.
-------
1 OU
125
100
± 75
•-3
ro s:
CO
en
50
25
n_
140
120
100
-
i
o
x 80
s_
.c
3
CD
1 60
OT
S-
O)
c:
* UJ
40 1
20
I *
-
— - — Q
^o—Q ~a """""
PJ Total available energy
GJ
/
i-^
/"^
/
/ Fuel mass flowrate =2.1 Kg/hr
/ ^
£• 1 iw
x^ l^
/ "7
/ <
/
•
/
\
^ Cooling tubes
O~
.x^ '^ " — ~~ Q
>^^°
40 60
80
180 200 220
160
Theoretical air, percent
Figure 6. Radiative catalyst/watertube system energy release vs. theoretical air.
-------
l\3
00
100
80
O)
u
0)
O-
X
O
60
O)
c 40
O
O
CO
20
50
/
- /*
/
/ 0
0
0
0
60
70
80 90 100
Theoretical air, percent
NH, + HCN + N0
NH,
O'
i
Figure 7. Radiative catalyst/watertube system, fuel nitrogen conversion
-------
o
Figure 8. Model gas turbine combustor.
288
-------
ro
oo
FX-OW
PIEECTION
IB" eer-
o
SECTION!
(?)
INLET \ OUTLET 1
IB" ESF
- SECT1OM
(?)
sl^ W&TEE OUT J
CATALYTIC.
Figure 9. Two stage catalytic arrangement.
-------
lOOr
QJ
S-
CL
X
O
80
60
A All NOX
©All NH3
Q NH3 + HCN
NO
ro
«3
o
t.
O)
>
c:
O
u
CO
40
20
., A__^L--^'
•.202 MPa
First stage
stoichiometry
50% T.A.
50
TOO
Overall theoretical air, percent
150
Figure 10. Two-stage combustor fuel nitrogen conversion.
-------
FUEL NOV CONTROL BY CATALYTIC COMBUSTION
A
BY
Edward K. Chu and John P. Kesselrlng
Acurex Corporation/Energy & Environmental Division
485 Clyde Avenue
Mountain View, California 94042
Work performed under EPA Contract 68-02-2611, Task 11
-------
ABSTRACT
The use of catalytic combustion in controlling fuel NOX was evaluated
experimentally. Screening tests of three different catalysts were conducted
over a range of stoichiometries from rich to lean, to determine the effect
of theoretical air, bed temperature, mass throughput, fuel type, and fuel
bound nitrogen type and concentration on the conversion of fuel bound nitro-
gen to NOV or NOV precursors. Minimum fuel NOV conditions were then deter-
A A A
mined. Since these minima occur at fuel-rich conditions, the use of a two-
stage catalytic combustor for control of fuel NO appears promising.
A
Detailed studies of the chemical reactions occurring in catalytic com-
bustion were made to determine what nitrogenous species are formed. The
experiments were performed using an oxygen/argon mixture as the oxidant in
order to eliminate thermal NOX reactions from the system. The results showed
that N£, NO, NH3, and HCN are the major products of the fuel bound nitrogen
conversion process. Further, the data tended to agree with the reaction
mechanism proposed for homogeneous reactions; i.e.,
HCN -»• NH-
The reaction, however, takes place on the catalytic surface rather than in
the gas phase.
293
-------
INTRODUCTION
An emerging technology which shows promise in controlling both thermal
and fuel NOV emissions is that of catalytic combustion. As reported in Ref-
A
erence 1, catalytic combustors have been tested at temperatures to 2000 K
(3140°F) with thermal NOX emissions less than 90 ppm corrected to 0 percent
excess air. Fuel NOV control through the use of a two-stage catalytic com-
A
bustion system is also reported in Reference 1, with NH^-doped natural gas
used to simulate a nitrogen-containing fuel. Conversions of bound nitrogen
to NOV of only 11 percent were measured for this system.
A
The subject of fuel nitrogen conversion in catalytic combustion sys-
tems has been examined by Matthews and Sawyer (Reference 2). Trace amounts
of either nitric oxide or ammonia were added to propane/oxygen/argon mixtures
and combusted on a platinum catalyst under rich, stoichiometric, and lean
combustion conditions. Conversion of the two model fuel nitrogen compounds
to nitric oxide was measured as a function of equivalence ratio, adiabatic
flame temperature, and fuel nitrogen concentration. Since the monolithic
catalyst bed was uninsulated during operation, the measured bed temperatures
were much lower than the corresponding adiabatic flame temperatures. For
this operational mode, the conversion of fuel nitrogen to NOV was found to be
A
strongly dependent on equivalence ratio, weakly dependent on calculated adi-
abatic flame temperature, and moderately dependent on fuel nitrogen concen-
tration. The precise fate of the fuel nitrogen was not determined.
The current study was conducted under EPA Contract 68-02-2611 to
achieve two objectives. First, the study was to define operating conditions
for catalytic combustors giving low levels of NOV emissions. Second, the
X
study was to conduct a systematic experimental determination of the fate of
fuel nitrogen during catalytic combustion.
PROGRAM OBJECTIVES
This study had two primary objectives. The first objective, discussed
in Screening Test Results, was to evaluate the parameters which affect the
conversion of fuel bound nitrogen to NOV during catalytic combustion, and to
A
295
-------
define conditions yielding low NOY emissions. Specifically, the following
J\
parameters were examined:
• Catalyst type
• Percent theoretical air
• Mass throughput (or heating rate)
• Reaction temperature
• Fuel bound nitrogen type and content
The second objective, described in Detailed Test Results, was to de-
termine the fate of fuel bound nitrogen during catalytic combustion with and
without the coexistence of fuel sulfur. Nitrogenous species measured include:
NO, NOX, N2, N20, NH3, and HCN.
EXPERIMENTAL APPARATUS
Figure 1 illustrates the experimental apparatus. For simplicity, the
equipment can be divided into the following five categories:
• Flow metering and measurement systems
• Preheating system
• Combustor and housing
• Probe and sample transfer system
• Analytical equipment
As illustrated in Figure 1, the following components of the reactive
mixtures can be metered in the system:
C3H8/CH4
NH3/CH3NH2
H2S
Air/Argon
N2/02
Fuel
Oxidant
296
-------
The oxidant is allowed to pass through the main heater, which consists of a
series of six 6000-Watt exposed-element heaters (GTE Sylvania Model 138825,
Style A). Within this main heater, the gas mixture can be heated to as high
as 811°K (1000°F) depending on the preselected preheat temperature.
The fuel, however, bypasses the main heater to avoid preignition
hazards within the main heater, and passes through a flame arrester before
mixing with the oxidant. The fuel/oxidant mixture then passes through the
mixing chamber and flow straightener before arriving at the catalytic com-
bustor. During this time, the preheat temperature of the gas mixture is
maintained by an auxiliary heater jacketted around the exterior of the mixing
chamber.
The combustor consisted of three 0.025m (1 inch) segments of graded
cell catalyst. Table 1 lists the four different catalysts tested and gives
information on the cell sizes used. Two UOP catalysts were required when
performance changes were noted on UOP-1 after several hours of testing. As
noted in Table 1, UOP-2 used only intermediate and small cells. The combus-
tor was mounted inside a 0.0457m (1.8 inch) I.D. fused quartz tube. A thick-
ness of six inches of insulation was applied to the exterior of the quartz
tube housing to minimize heat losses from the combustor.
TABLE 1. CATALYST TYPE AND GEOMETRIC CONFIGURATIONS
Catalyst Type
UOP-1
Proprietary catalyst
on DuPont alumina
UOP-2
Proprietary catalyst
on DuPont alumina
Acurex Pt/NiO
2.5 wt % Pt on large
cells; 6 wt % NiO on
DuPont alumina
Acurex Pt
5 wt % Pt on DuPont
alumina
Large Cell
Diam, m (in)
6.35 x TO'3
(0.25)
4.76 x 10'3
(0.1875)
6.35 x 10-3
(0.25)
6.35 x 10-3
(0.25)
Intermediate Cell
Diam, m (in)
4.76 x 10-3
(0.1875)
4.76 x 10-3
(0.1875)
4.76 x 10-3
(0.1875)
4.76 x 10-3
(0.1875)
Small Cell
Diam, m (in)
1.59 x 10-3
(0.0625)
1.59 x 10-3
(0.0625)
1.59 x 10-3
(0.0625)
1.59 x 10-3
(0.0625)
297
-------
Five K-type thermocouples were placed in the combustor at various lo-
cations to monitor in-depth combustion temperatures. In addition to these
five thermocouples, a thermocouple located at the combustor exit measured
the exhaust gas temperature.
A water-cooled stainless steel sample probe was used in this experi-
ment. This probe was selected because of its durability in a hostile environ-
ment. The probe, however, is capable of reducing N02 to NO catalytically
unless the probe is sufficiently quenched by water. For this reason, the
actual measurement of NO and N02 may not be completely reliable. Neverthe-
less, from studies of homogeneous combustion (Reference 3), N02 exists only
in the flame zone. In the near-postflame region the concentration of N02 is
generally low as a result of rapid conversion from N02 to NO. If the same
analogy were drawn, N02 concentration should also be low in this study.
Chemical species measured in this experiment include: CO, C02, 02,
UHC, NO, NOX, N2, N20, NH3> and HCN. The list of the analytical equipment
used and the corresponding chemical species measured is shown in Table,2.
SCREENING TEST RESULTS
In this section, the results of the screening tests are presented and
discussed. For convenience of discussion, the emission characteristics of
each catalyst will be discussed independently prior to comparison with each
other.
Figures 2 to 5 show the effect of bed temperatures on the conversion
of chemically bound nitrogen for the UOP catalyst. In Figures 2 and 3,
natural gas was fired at a heating rate of 42.2 MJ/hr (40,000 Btu/hr) and
with a chemically bound nitrogen content of 0.3 percent by weight of the fuel.
The bed temperatures were 1367 K (2000°F) and 1478 K (2200°F), respectively.
As may be seen from the figures, an increase in NOV yield over the entire
A
range of theoretical airs was observed with increased bed temperature. The
same effect was observed in Figures 4 and 5, although the heating rate was
reduced to 25.32 MJ/hr (24,000 Btu/hr) and the chemically bound nitrogen
content was increased to 0.5 percent by weight of the fuel.
298
-------
Figures 4 and 5 also show the percent of NH3 and HCN yield under
fuel-rich conditions. Both figures show that these two species rapidly
diminish as stoichiometric proportions are approached. The disappearance of
NHg and HCN is largely a result of these species being oxidized to NOX, as
evidenced by the increasing level of NOV. In the fuel-rich condition, however,
A
some of these species were presumably converted to N2» as is the case in
homogeneous combustion. By summing the percent yields of NH3, HCN, and NOX
in fuel-rich conditions and plotting them on the figure, the condition which
yields the highest percent of N2 can be determined. Figures 4 and 5 show
that this condition occurs somewhere between 85 and 95 percent theoretical
air with the UOP catalyst.
The effect of heating rate (or mass throughput) on chemically bound
nitrogen conversion can be assessed by comparing both Figures 2 and 4 with
Figures 3 and 5. In the case of Figures 2 and 4, the bed temperature of the
combustor was held at 1376 K (2000°F). At this bed temperature, the low
heating rate yielded slightly lower NOV emissions than the higher heating
A
rate for both rich and lean conditions. When the bed temperature was increased
to 1478 K (2200°F) (see Figures 3 and 5), this behavior was observed only for
fuel-rich conditions. For fuel-lean conditions, the opposite behavior was
observed. A similar NOX emission characteristic with respect to heating
rates at a bed temperature of 1478 K was also observed in Figures 6 and 7
when propane was used as fuel instead of natural gas. Duplicating this result
substantiated that a higher heating rate can in fact suppress fuel NO forma-
A
tion in the fuel-lean condition at a bed temperature of 1478 K.
An explanation of the above phenomenon is not available at this time.
However, it is believed that this phenomenon was closely associated with the
occurrence of homogeneous combustion in the combustor. In the case of the
lower bed temperature, i.e., 1367 K, the contribution of homogeneous combus-
tion to the overall combustion was expected to be insignificant at both heating
rates, since the bed temperature was not hot enough to initiate significant
homogeneous reactions. On the other hand, the mass transfer rate of NH3 in-
creases as the heating rate (or mass throughput) increases. Thus, more fuel
NOX was formed at the higher heating rate than at the lower one.
299
-------
In the case of higher bed temperature, i.e., 1478 K, the bed tempera-
ture was insufficient to initiate homogeneous combustion under fuel-rich
conditions, and the same behavior as the lower bed temperature cases is
noted. However, for fuel-lean conditions, homogeneous combustion may become
significant. Since the lower heating rate case has a longer residence time
than the higher heating rate case, the contribution of homogeneous combustion
to the lower heating rate was therefore more important. As a result, the
lower heating rate yielded a higher NOV yield than the higher heating rate.
X
The effect of fuel type on the chemically bound nitrogen conversion
can be assessed by comparing Figure 3 with Figure 7 and Figure 5 with Figure
6. These figures show that the emission characteristics for both fuels doped
with NH3 are very similar qualitatively as well as quantitatively. These
results indicated that the combustion process was weakly coupled to the
chemically bound nitrogen conversion at the bed temperature of 1478 K (2200°F),
The effect of chemically bound nitrogen content is shown in Fiigure 8.
Similar to homogeneous combustion, the NOY yield decreases as the chemically
A
bound nitrogen content increases. This result indicates that the following
self-destruction reactions also occur during catalytic combustion when the
chemically bound nitrogen concentration is abundant in the reactive gas mix-
ture.
This result, however, contradicts the results of Matthews and Sawyer (Ref-
erence 2). Their results show the opposite behavior. It is believed that
the discrepancy is due to the difference in bed operating temperatures. The
data for this study were obtained at bed temperatures of 1367 and 1478 K,
whereas the data of Matthews and Sawyer were obtained at much lower tempera-
tures.
300
-------
The effect of chemically bound nitrogen type is shown in Figure 9.
Methylamine and ammonia were used as the model chemically bound nitrogen com-
pounds in this study. The results shows almost no discrepancy between emis-
sions for the two compounds. This indicates that the pyrolysis kinetics of
chemically bound nitrogen is not the controlling process in the conversion
of chemically bound nitrogen to NOY.
i x
Figure 9 also shows the effect of bed temperature on the chemically
bound nitrogen conversion. The NOV yield was weakly dependent on bed tern-
/\
perature when the bed temperature was higher than 1400 K. As the bed tem-
perature was reduced below 1400 K, the NOV yield gradually became strongly
. . /\
dependent on bed temperature.
The above results confirm that the chemically bound nitrogen conver-
sion process is likely to occur at the surface rather than in the gas phase.
At bed temperatures above 1400 K, the reaction rate of the conversion pro-
cess was so fast that it was controlled by diffusion. Thus, the NOV yield
A
was weakly dependent on the bed temperature. As the bed temperature was
reduced below 1400°K, the reaction rate was no longer controlled by diffusion,
but instead controlled by the surface kinetics. Since the reaction rate of
the conversion process was slower at these temperatures, it allowed the self
destruction reaction, as described previously, to become more significant
and thereby convert more NO to ^.
Screening test results of the Acurex platinum catalyst, showing the
effect of bed temperatures, is presented in Figures 10 and 11. The bed
temperatures were 1367 K (2000°F) and 1478 K (2200°F). Both results show
that 100 percent NOV yields were observed under fuel-lean conditions.
/\
Under fuel-rich and near-stoichiometric conditions, significant dif-
ferences were noted for the two temperatures. The NOX yield was higher for
the lower bed temperature than the higher one, and the opposite effect was
noted for the NH3 and HCN yields. These results indicate that the fuel oxi-
dation reaction and the chemically bound nitrogen oxidation reaction were
competing with each other for oxygen at these conditions. Apparently, the
chemically bound nitrogen oxidation reaction was dominant at 1367 K, resulting
in a higher NOX yield and lower NH3 and HCN yields. When the bed temperature
301
-------
was increased to 1478 K, most of the free oxygen was consumed by the fuel
and hence was unavailable for the chemically bound nitrogen.
The screening test results of the Acurex NiO/Pt catalyst, indicating
the effect of heating rates, are shown in Figures 12 and 13. Since NiO is
a base-metal catalyst, it requires a minimum of 1589 K (2400°F) bed tempera-
ture for satisfactory combustion efficiency. At this bed temperature, the
NOX yield was almost independent of heating rates (see Figures 12 and 13).
However, significant variations in terms of NH3 and HCN yields were
observed under fuel-rich conditions. The NH3 yield was higher at the lower
heating rate than the higher one, and the opposite was true for the HCN
yield. It is of interest to note that the same variation was also observed
with the UOP catalyst (see Figures 4 and 5).
Since the conversion of chemically bound nitrogen is likely to occur
at the surface rather than in the gas phase, as discussed previously, the
type of catalyst can have significant effects on the conversion of chemi-
cally bound nitrogen. Some of these effects may be seen in Figures 11 and
13, where the emission characteristics of the Acurex Pt catalyst and the
Acurex NiO/Pt catalyst are shown, respectively. The Acurex Pt catalyst has
a highly reactive surface, and as a result it not only completely oxidized
the fuel but also converted all NH3 to NOX under fuel-lean conditions. At
the temperature tested (1478 K), no selective catalytic reaction was ex-
hibited by the Acurex Pt catalyst. On the other hand, the Acurex NiO/Pt
catalyst has a less reactive surface. Consequently, a higher operating tem-
perature (1589 K) was required. For fuel-lean conditions, some selective
catalytic reaction capabilities were demonstrated by the NiO/Pt catalyst.
It exhibited good combustion efficiency (though not as good as the Pt cata-
lyst), and concurrently lowered the NH3 conversion from 100 percent to 80
percent.
The UOP catalyst showed better performance than the NiO/Pt catalyst
(see Figure 7). It had a combustion efficiency comparable to the Pt cata-
lyst and yet yielded only 75 percent NOX from NH3, which was 25 percent
better than the Pt catalyst. However, it should be noted that the chemically
bound nitrogen contents were not the same in both cases. The chemically
bound nitrogen content in the UOP catalyst case was lower than the other
302
-------
two cases. Nevertheless, it was also found that the NOV yield decreases as
A
the chemically bound nitrogen content increases (see Figure 8) for a percent
theoretical air of 310. If the same behavior held true in all fuel-lean
conditions, as is the case in homogeneous combustion, the UOP catalyst
should yield even less NOY when the chemically bound nitrogen content in-
/\
creases.
DETAILED TEST RESULTS
The results of detailed tests are presented and discussed in this
section. The results include the determination of the fate of chemically
bound nitrogen during combustion in the UOP catalyst under the following
conditions:
0 Variable percent theoretical air under fuel-rich and stoichio-
metric conditions at a fuel nitrogen content of 3 percent by
weight of the fuel
• Same conditions as above but with variable NHg concentration
• Same conditions and the same NHg concentration as above but
tested with an aged UOP catalyst
The Acurex Pt and the Acurex NiO/Pt catalysts were also tested in
this experimental series, but on a less extensive scale. Only two test con-
ditions were performed with these two catalysts. The first condition was
selected at 75 percent theoretical air where N2 is presumably the major pro-
duct from the chemically bound nitrogen conversion process. For the other
condition, the Acurex Pt catalyst was tested near the stoichiometric con-
dition where the competing reactions between the hydrocarbon oxidation and
the chemically bound nitrogen conversion occurred; and the Acurex NiO/Pt
catalyst was tested at 200 percent theoretical air to verify that the selec-
tive catalytic reaction had in fact occurred in the fuel-lean condition.
To assure that the nitrogen species NOX, N2, N20, HCN, and NH3 were
the dominant products of the chemically bound nitrogen conversion process,
a nitrogen mass balance was performed for each test condition. Since the
emission data were taken on a dry basis, the following equation was used to
account for the water content in the combustion products:
303
-------
\
where YN is the total mass fraction of elemental nitrogen in the combus-
'"o
tion products, Nj is the ppm of the jth nitrogen species, K^Q is the mole
fraction of water vapor, and MW is the mean molecular weight of the combustion
products. KH2o and MW are estimated from chemical equilibrium calculations.
The results of the chemically bound nitrogen mass balance are shown
in Table 3. As may be seen, a maximum deviation of ±12 percent between the
input fuel nitrogen content and the measured nitrogen species in the combus-
tion products was obtained, except for two data points. N20 was not detetted
for any test condition. Under fuel-rich conditions, as was expected, N2 was
the dominant product of the chemically bound nitrogen conversion process.
Under a fuel-lean condition of 200 percent theoretical air, a conversion of
13 percent NH3 to N2 was noted for the NiO/Pt catalyst.
Figure 14 shows the variations of the nitrogen species concentrations
as a function of percent theoretical air. A reaction mechanism similar to
that for homogeneous reactions, i.e.,
HCN
can be used to interpret the data presented in Figure 14. At 75 percent
theoretical air or below, the N2 formation reaction apparently was the domi-
nant reaction, and consequently very low levels of NO, HCN, and NH3 concen-
trations were observed. Above 75 percent theoretical air, the N2 formation
reaction became less important. The slow-down of the N2 formation reaction
combined with the unavailability of oxygen led to a rapid buildup of NH3 and
a gradual increase of NO at conditions between 75 and 90 percent theoretical
304
-------
air. However, as the precent theoretical air exceeded 90s NO was quickly
formed with a simultaneous decrease of NH3.
Figure 15 shows the effect of the chemically bound nitrogen content
on N£ yield under fuel-rich conditions. It was shown that the percent N2
yield increases as the chemically bound nitrogen content increases, which is
the same behavior noted in the screening test results under fuel-lean con-
ditions (see Figure 8). Also of interest is that all but the 75 percent
theoretical air data exhibited the same slope. However, it should be noted
that one of the 75 percent theoretical air data points was associated with
a significant error in the nitrogen mass balance (see Table 3). If the N2
concentration were corrected for this error, a similar slope was observed
in comparison with the other data (shown by the dashed line in Figure 15).
Two of the test conditions under the 3 percent chemically bound
nitrogen content test series were repeated to evaluate the effect of an
aged catalyst on the conversion of chemically bound nitrogen. As may be
seen from Table 3, the aging of a catalyst was demonstrated by the lesser
requirement of diluent argon than that of the active catalyst for the same
bed operating temperature. As for the nitrogen species emissions, the aged
UOP catalyst yielded 30 percent or more less N2 than that of the active UOP
catalyst, depending on the theoretical air level. In return, higher con-
centrations of NO, NHg, and HCN were emitted by the aged catalyst than the
active one. This result, again, confirmed that the chemically bound nitrogen
conversion process is a surface event.
H2S EFFECTS ON CHEMICALLY BOUND NITROGEN CONVERSION
Current flat flame studies show that under fuel-rich conditions, the
presence of fuel sulfur can have a radical effect on the fuel NO formation
mechanism. Both enhancement and inhibition of fuel NO have been observed,
depending on the stoichiometric ratio, the residence time, and (probably)
the model fuel-bound nitrogen compound. DeSoete (Reference 4) observed sig-
nificant enhancement of NHo conversion to NO due to the presence of H2S in
his experiment. Wendt et al., ("sference 5), on the other hand, observed
both enhancement and inhibition of fuel NO due to the presence of H2S, de-
pending on the stoichiometric ratio. However, measurements made in the work
305
-------
of Wendt et al., were over much longer residence times, and these experi-
ments involved both lower nitrogen and sulfur levels and higher flame tem-
peratures than those of DeSoete. Furthermore, DeSoete used NH3 as the model
fuel-bound nitrogen compound, while Wendt et al. used C2N2.
As part of the detailed tests, the effect of H2S on the chemically
bound nitrogen conversion during catalytic combustion under fuel-rich con-
ditions was evaluated. The results of this study are shown in Figure 16 in
terms of percent NO and N2 yields as a function of percent theoretical air.
Also shown in the same figure are the results with the same test conditions
but without H2S dopant. Figure 16 shows that the addition of H2S enhanced
NO formation and suppressed N2 formation under fuel-rich conditions, which
is in agreement with DeSoete's results. However, at the stoichiometric con-
dition, the addition of H2S inhibited NO formation. A similar trend was
also observed by Wendt et al., although their data were taken at conditions
between 46 and 70 percent theoretical air.
Additional evidence showing that the addition of H2S can inhibit N2
formation can be seen from the gas chromatograph measurements taken before
and after the H2S flowrate was turned off. These two measurements are shown
in Figure 17. As may be seen, the N2 concentration in the case with H2$
dopant is significantly lower than the case without. Also, it is of interest
to note that the combustor bed temperature increased more than 367 K after
the H2$ flowrate was turned off, indicating that the existence of H2S also
inhibited combustion.
Attempts to perform nitrogen mass balances with the H2S dopant data
have not been successful (see Table 3). With the measured nitrogen species,
i.e., NO, N20, N2, HCN, and NH3, it was found that the sum of these species
was consistently lower than the amount of nitrogen input to the system.
Also, there was a trend showing that the percent error decreases as the
percent theoretical air increases. This trend implies that either there may
be some nitrogen species existing only under fuel-rich conditions, which
were not accounted for in the current study, or erroneous measurements were
induced by the existence of H2S under fuel-rich conditions.
306
-------
CONCLUSIONS
Based upon the results of the screening tests, the bed operating
temperature was shown to affect the fuel NOX formation, depending on the
temperature range and the reactivity of the catalyst. For the UOP catalyst,
at bed temperatures below 1478 K the amount of fuel NOV formed decreases as
J\
the bed temperature decreases. At bed temperatures above 1478 K, the for-
mation of fuel NOV was found to be insensitive to bed temperatures. For
A
the Pt catalyst, fuel NOX was insensitive to bed temperatures, even at
1367 K. Based on this data, the formation of fuel NOX is a surface event.
Depending on whether the surface reaction is in the kinetic regime or in
the diffusion regime, fuel NO formation may be sensitive to bed tempera-
J\
tures. In terms of low NOX operating conditions for the catalytic combustor,
a low bed temperature is recommended within the constraint of satisfactory
combustion efficiency. This approach seems ideal for gas turbine applica-
tions. However, for boiler applications, this approach is unacceptable
because boilers require a high flame temperature and hence a different
approach should be used.
The mass throughput (or heating rate) appears to be an effective
approach for controlling fuel N0¥ for the UOP catalyst. Increasing the mass
J\
throughput tends to maximize the surface reactions and minimize the homo-
geneous reactions within the combustor, and therefore lowers the fuel NOX
formation. For the NiO/Pt catalyst, since it is a less reactive catalyst,
the bed operating temperature remains the dominant parameter, and therefore
fuel NOX formation was unaffected by the variation of mass throughput.
However, if the bed temperature were increased further, a similar result to
the UOP catalyst is anticipated.
Under fuel-rich conditions, the fuel NOY emission levels were low for
rt
all three types of catalyst. Also, there appeared to be a condition where
NO, HCN, and NH3 concentrations reach minimum levels. However, this con-
dition varies with the mass throughput. It is vital that this condition be
determined as a function of mass throughput prior to the development of the
two-stage catalytic combustion system.
307
-------
The molecular structure of both fuel and chemically bound nitrogen
compounds have been found to be insensitive to fuel NOX formation. These
results indicate that the results of the current study can be extrapolated
to the combustion of heavier fuels, such as residual oil, with reasonable
confidence provided that the fuel is in the vapor state prior to entering
the combustor.
The chemically bound nitrogen content has been found to inversely
affect the formation of fuel NOV. This is because of the self-destruction
A
reaction prompted by the excess NHg.
Based upon the results of the detailed tests, it was verified that
NO, N2, NH3, and HCN are the dominant products of the chemically bound nitro-
gen conversion process. N20 was not detected in any condition tested.
With the addition of H2S to the reactive mixture, the formation of
fuel NO has been enhanced under fuel-rich conditions and inhibited at the
stoichiometric condition. A nitrogen balance within reasonable accuracy was
not obtained with H2S addition. This indicated that either additional
nitrogen species should be included in the experiment or the measurement
techniques employed were adversely affected by the existence of H2S. The
actual cause of the imbalance, however, remains to be determined in future
efforts.
In summary, there is a striking similarity between catalytic combus-
tion and homogeneous combustion in terms of fuel NOV formation. However,
X
the similarity is somewhat expected since the role of the catalyst is to
accelerate the reaction rate towards equilibrium rather than to alter the
chemical nature of a given reactive mixture. Current results show that
there are great potentials for catalytic combustion in controlling fuel NOY,
/\
one of the proven concepts being the two-stage catalytic combustion system.
The use of a one-stage catalyst for fuel NOY control, however, requires fur-
/\
ther development, since a "real" selective oxidation catalyst has not yet
been found. However, such a system may be possible through use of the
graded cell concept (see Reference 1). For example, one can easily design
the first segment of the graded cell catalyst to achieve the condition where
308
-------
the NOV self-destruction reaction is maximized. Then, by using a reduction
A
catalyst as the second segment, fuel NOX can be eliminated in this region.
The last segment of the graded cell catalyst would again be an oxidative
catalyst designed primarily to complete the combustion.
309
-------
REFERENCES
1. Kesselring, J. P., et.al., "Design Criteria for Stationary Source
Catalytic Combustion Systems," Acurex Final Report 78-278, March, 1978.
2. Matthews, R. D., and Sawyer, R. P., "Fuel Nitrogen Conversion and Cata-
lytic Combustion," University of California/Berkeley, Energy and Environ-
ment Division Report LBL-6396, October, 1977.
3. Bowman, C. T., "Kinetics of Pollutant Formation and Destruction in Com-
bustion," Prog. Energy Combustion Science, Vol. 1, 1975.
4. DeSoete, G., "La Formation des Exyde D'Azote Dans La Zone d'Oxydation
des Flammes d'Hydrocarbures," Institute Francais du Petrole, Final
Report No. 23309, Ruiell Malmaison, France, June, 1975.
5. Wendt, J. 0. L., Corley, T. L., and Morcomb, 0. T., "Interactions
Between Sulfur Oxides and Nitrogen Oxides in Combustion Process," Pro-
ceedings of the Second Stationary Source Combustion Symposium, Vol. IV,
EPA-600/7-77-073d, July, 1977-
310
-------
TABLE 2. ANALYTICAL EQUIPMENT USED
Chemical
Species
Method
Detection
Range
Equipment Model
NH.
HCN
CO
co2
°2
UHC
NO, NO.
Absorbed In 0.005 N H2S04
Ammonia — Sensing Electrode
Direct Nesslerization (Colorimetric)
Absorbed in 0.2 N NaOH
Cyanide Ion-Selective Electrode
Titrimetric -- Silver Nitrate Solution
6.C: Silica Gel & Molecular Sieve
5A Columns
G.C: Poropac Q Column
Infrared Absorption
Infrared Absorption
Paramagnetic Detection
Flame lonization
Optical Detector (Chemiluminescent)
2-10 ppm
>10 ppm
<10 ppm
>10 ppm
>50 ppm
>50 ppm
0-2000 ppm
0-20%
0-21%
0-100,000 ppm
0-10,000 ppm
Orion 951000
Chemtrix Cnl015 M
Carle G.C. 8700
Carle G.C. 8700
Hartman & Braun Uras 2T/CO
Hartman & Braun Uras 2T/C02
Hartman & Braun Magnos 5T
Beckman Model 400 HC Analyzer
Air Monitoring Inc. Model 32 C
-------
TABLE 3. FUEL NITROGEN BALANCE
oo
ro
Run
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Catalyst
Type
Pt
Pt
NiO/Pt
NiO/Pt
UOP
H2S
Additive
No
Yes
Fuel-N
Content
(Wt. % of Fuel)
2
3
2
2
3
3
3
3
4
3
3
3
0
0
4
0
6
0
0
0
0
0
0
0
Theoretical
Air
(%)
98
77
200
75
100
90
80
75
70
100
75
90
80
70
100
90
80
75
70
Percent Conversion of Nitrogenous
Species In the Combustion Products
NOX
76.00
0.83
80.57
9.63
34.13
1.06
0.58
0.29
1.00
23.77
0.25
0.68
3.28
2.47
22.97
7.55
2.44
0.89
0.21
N2
15.97
79.05
13.19
58.94
71.54
58.93
79.29
98.68
74.86
68.20
74.01
61.50
54.68
45.60
34.46
44.44
31.76
18.59
8.98
NH3
8.12
18.07
1.81
41.20
4.77
44.11
25.17
4.49
4.29
5.90
5.36
42.34
37.92
41.91
8.53
31.12
23.80
27.99
20.24
HCN
1.22
1.51
1.21
3.09
1.52
2.96
2.03
2.59
2.95
0.72
2.33
4.71
16.79
7.00
8.50
7.01
9.28
10.14
11.68
N20
0
Error
(%)
1.00
-0.56
-3.21
12.86
11.97
6.81
7.07
6.05
-16.91
-1.43
-18.02
9.23
12,66
-3.02
-25.54
-9.88
-32.72
-42.39
-58.89
-------
C3H8/CH4
CJ
Flow neterlng
and measurement
systems
Preheating
system
Flame 1on1zat1on
HC analyzer
Paramagnetic
analyzer
Infrared
analyzer
CLA
GC
chemistry
UHC
CO, CO-
NO, NO,
• NH-
Analytical
equipment
Figure 1. Experimental apparatus.
-------
CO
100
80
0*
>
o
60
40 ~
20 -
60
80
UOP CATALYST
Nat. Gas/Air, YN = .003
Heating Rate = 42.2 MJ/hr (40,000 Btu/hr)
Bed Temp. -=1367 K (2000°F)
120
140 160 180
% Theoretical A1r
200
220
Figure 2. NH3 conversion at 1367 K - UOP catalyst.
-------
CO
tn
100
I/I
41
O
O
80
60
40
20
60
UOP CATALYST
Nat. Gas/Air, YN = .003
Heating Rate = 42.2 MJ/hr (40,000 Btu/hr)
Bed Temp. = 1478 K (2200°F)
80
I
100
120 140 160
% Theoretical Air
180
200
220
Figure 3. NH3 conversion at 1478 K - UOP catalyst.
-------
UOP CATALYST
Nat. Gas/Air, YN = .005
Heating Rate = 25.3 MJ/hr (24,000 Btu/hr)
Bed Temp. = 1367 K (2000°F)
% Theoretical Air
Figure 4. NH3 conversion at 1367 K - UOP catalyst.
316
-------
UOP CATALYST
Nat. Gas/Air, YN » .005
Heating Rate = 25.3 MJ/hr (24,000 Btu/hr)
Bed Temp. = 1478 K (2200°F)
91
O
CJ
100
80
60
40
\
20
60
80
«0
?
I
I
100 120 140
% Theoretical Air
160 180
Figure 5. NH3 conversion at 1478 K - UOP catalyst.
317
-------
o
0>
>
o
o
100
80
60
40
20
UOP CATALYST
C3Hg/A1r, YN = .005
Heating Rate = 25.3 MJ/hr (24,000 Btu/hr)
Bed Temp = 1478 K (2200°F)
60
80 100 120 140
% Theoretical Air
160
180
Figure 6. NH3 conversion at 1478 K - UOP catalyst.
318
-------
o
0)
>
o
o
,20 .
60
UOP CATALYST
C3H8/Air, YN
.003
Heating Rate - 42.2 MJ/hr (40,000 Btu/hr)
Bed Temp. = 1478 K (2200°F)
80
100 120 140
% Theoretical Air
160
180
Figure 7. NH3 conversion at 1478 K -UOP catalyst.
319
-------
120
no
100
90
80
70
60
50
UOP CATALYST
C3HQ/A1r/NH3
Percent Theoretical A1r = 310
Heating Rate - 25.3 MJ/hr (24,000 Btu/hr)
Bed Temp. = 1478 K (2200°F)
I
I
I
I
I
I
I
N
9
0
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Chemically Bound Nitrogen Content, Wt. % of Fuel
Figure 8. Effect of nitrogen content on NOV yield - UOP catalyst.
A
320
-------
100
80
"J 60
40
20
1000
800
1250
UOP CATALYST
C3H8/Air/N-Dopant
Fuel-Lean Combustion
Heating Rate =42.2 MJ/hr
I
I
1500 1750
Bed Temp. , F
_ I
2000
2250 2500
1000
1200
Bed Temp., K
1400
1600
Figure 9. Effect of dopant type on NOV emission - UOP catalyst.
A
321
-------
100
80
- 60
in
ro 40
20
AEROTHERM Pt. CATALYST
C3Hg/Air, YN = .02
Heating Rate = 42.2 MJ/hr (40,000 Btu/hr)
Bed Temp. = 1367 K (2000°F)
60 80 TOO 120 140
% Theoretical Air
160
180
Figure 10. NH3 conversion at 1367 K - Pt catalyst.
322
-------
60
AEROTHERM Pt. CATALYST
C3H8/A1r, YN = .02
Heating Rate = 42.2 MJ/yr (40,000 Btu/hr)
Bed Temp. = 1478 K (2200°F)
80
TOO 120 140 160
% Theoretical A1r
180
200
Figure 11. NHg conversion at 1478 K - Pt catalyst.
323
-------
100
80 -
o
•1—
fc.
c
o
o
60
40 -
20 -
NiO/Pt CATALYST
C3H8/Air, YN = .02
Heating Rate = 25.3 MJ/yr (24,000 Btu/hr)
Bed Temp. = 1589 K (2400°F)
80
100 120 140
% Theoretical Air
160
180
200
Figure 12. NH3 conversion at 1589 K - NiO/Pt catalyst.
324
-------
100
N10/Pt CATALYST
C3H8/Air, YN = .02
Heating Rate = 42.2 MJ/hr (40,000 Btu/hr)
Bed Temp. = 1589 K (2400°F)
o
•r"
&.
o
o
80
60
40
20
60
80
100
N0x
NH3
HCN
I
NO.
O
<
120 140 160
% Theoretical Air
180
200
Figure 13. NH3 conversion at 1589 K - NiO/Pt catalyst.
325
-------
120
100
80
60
O)
u
O
O
40
20
0
60
C3Hg/02/Ar/NH.
Fuel-N content = 3%
'3
365 ppm
70
80 90
% Theoretical air
100
O
V
no
Figure 14. Effects of percent theoretical air on the emissions of
nitrogen species under fuel-rich conditions.
326
-------
100
C3H8/02/Ar/NH.
90 „
80 .
o
oo
70 -
60 -
50
2.0
- 75% TA
- 90% TA
-100% TA
2.5 3.0 3.5
Fuel nitrogen content, % by wt. of fuel
vt
0
t»
I
4.0
Figure 15. Effects of fuel-N content on No yield under
fuel-rich conditions.
327
-------
C3Hg/02/Ar/NH3
0)
•r-
U
V
a.
a
o
O
% Theoretical A1r
Figure 16. Effects of S on the conversion of chemically bound nitrogen.
328
-------
CO
ro
10
With H2S dopant
Without H2S dopant
Figure 17. Comparisons of No concentrations with and without H^S dopant.
-------
PROSPECTS FOR HIGH-TEMPERATURE CATALYSTS
By
William C» Pfefferle
Consultant
Middletown, N. J. 07748
This work was supported by EPA Contract 68-02-2611, Task 30.
331
-------
ABSTRACT
The development of the catalytic combustor, a device that not only
yields low emissions but fuel savings as well, has created a need for cata-
lysts capable of operating at 1600-1700C for prolonged periods. A high
temperature combustor catalyst must be able to withstand the thermal stresses
associated with cyclic operation. Materials to be suitable for use in a
high temperature combustion catalyst must have a high melting point (above
2000C), a very low vapor pressure at the use conditions, and catalytic
activity or compatibility with a suitable high temperature catalytic com-
ponent. Good thermal shock properties are required but need not be an
intrinsic quality of the selected material. Suitable candidate materials
include, among others, catalytic alumina spinels such as magnesium cobalt
aluminate and stabilized zirconia doped with a catalytic metal such as
nickel. Prospects for a durable catalyst capable of 1700C operation
appeared to be excellent. Fabrication of a monolithic structure directly
from a material incorporating the catalytic component appears to be the
best route to a successful high temperature catalyst.
333
-------
INTRODUCTION
In contrast to other catalytic systems, the catalytic combustor re-
quires catalysts capable of both continuous and cyclic operation at very
high temperatures, ideally as high as 1600 or 1700C, without degradation of
either catalytic or physical properties. It is obvious that this requires
the use of high-melting point, low volatility materials. Similarly, very
low surface area catalysts are needed, since sintering in use would alter
mechanical properties. It is important to recognize that high surface area
materials, even a high purity alumina (MP«^2000C), such as is used in re-
forming catalysts, will sinter badly at temperatures below 600C.
As is the case with any catalytic system, the performance cf a catalyst
for combustion of fuels depends on many factors besides the intrinsic cata-
lytic activity. Structural, thermal and chemical properties of both the
catalytic agents and the support material are important. The important
factors include:
Selectivity (No NOV formation)
X
Resistance to Poisoning
Mechanical Strength
Thermal Shock Resistance
Attrition Resistance
Chemical Inertness
With combustion catalysts, no selectivity problems are encountered,
even at the highest temperatures, with respect to the fixation of atmos-
pheric nitrogen. However, combustion catalysts can convert fuel nitrogen
to nitrogen oxides.
Precious metal combustion catalysts are normally resistant to most
*'' v
catalyst poisons at the temperatures of catalytic combustion. Even with
335
-------
base metal oxide catalysts, catalyst poisoning is not a significant problem
under catalytic combustor operating conditions since all base metal salts
decompose to the corresponding oxides at the temperatures involved.
Mechanical strength is not usually a problem with catalysts since the
mechanical load is typically just the weight of the catalyst itself. How-
ever, it is important that a catalyst not degrade in use to the point of
crumbling. Unlike most catalysts, gas turbine combustor catalysts must
maintain structural integrity at very high flow velocities and resist both
attrition and thermal shock.
Thermal shock resistance is especially important for combustion cata-
lysts since it is often desirable to light offa combustor as rapidly as
possible, A high thermal conductivity and high strength contribute to good
thermal shock resistance. Systems with high volume change phase transi-
tions should be avoided.
A good attrition resistance is important for combustor catalysts be-
cause of the high flow velocities encountered in many applications. Proper
firing conditions are critical. Slip coatings must be well bonded.
Chemical inertness of a catalyst to the process stream is important in
many applications. In the case of platinum combustion catalysts, halides
in the fuel can cause rapid loss of platinum. The problem is the volatil-
ity of the platinum oxyhalides. Similarly, chromia can be stripped off a
catalyst at even lower temperatures than can platinum. Chromium oxychlor-
ide has a vapor pressure of one atmosphere at 390C. At high temperatures
one must also consider the interaction between the support and the cataly-
tic material. For example, cobalt oxide reacts readily with alumina to
form the less active cobalt aluminate (Reference 1). Slip Coatings can
react with the support and degrade mechanical strength and thermal shock
resistance.
336
-------
BASIS FOR DEVELOPMENT OF HIGH TEMPERATURE CATALYSTS
Prior to the development of a catalytic combustor suitable for gas
turbine use, combustion catalysts were limited to an operating temperature
of about 800C. Above this temperature catalysts degraded very rapidly. At
the second workshop on catalytic combustion in June 1977 catalysts operable
at temperatures between 1200 and 1500C were reported. It is believed that
higher temperature catalysts capable of operating as high as 1700C are
feasible. A basis for the selection of such a high temperature catalyst
follows.
1. Only materials with melting points above about 2000C should be con-
sidered. Even with low surface area catalysts, grain growth of the struc-
tural material must be limited. Ideally, this means the melting point of
the material must be about 1.5 times the catalyst operating temperature.
For a 1700C operating temperature one would prefer a material that melts
no lower than 2700C. Zirconia (MP 2760C) meets this requirement. For a
1500C operating temperature the corresponding requirement for the melting
point is about 2385C. Based on data for alumina catalysts, somewhat lower
melting points should be allowable. Kesselring (Reference 2) has reported
the operation of an alumina (MP 2050C) based on a catalyst at 1500C with-
out apparent catalyst degradation.
2. Low surface area catalysts are preferred. As already noted,
sintering of high surface area substrates could lead to degradation of
mechanical properties and to undesired solid state interactions. Further,
high surface area has already been shown to be unnecessary to good low
temperature light off activity. Kesselring (Reference 2) has also shown
2
that a catalyst can give good combustion results with a surface of 0.02M /g
pr less.
337
-------
3. Very low vapor pressure materials are required. Because of the
high volumetric flows which a combustor must handle, many refractories
which might be suitable in a non-flow environment will have an unacceptably
Large weight loss in a combustion system. Beryllia, an excellent ceramic
material with very good thermal shock resistance, is unacceptable in a com-
bustor environment because beryllia suffers severe weight loss in air at
temperatures over 1100C if as little as 0.5% water vapor is present.
Obviously, materials must be selected with care with this consideration in
mind.
4. Base metal oxide catalysts are required. This is because precious
metals are much too volatile under catalytic combustor operating conditions.
At 1400C, platinum (MP 1773C) exhibits a vapor pressure in air equivalent
to .a weight loss of about 56 grams of platinum per 100,000 cubic feet of
air (Reference 3). For a combustor with a throughput of 1.1 pounds per
second, this would represent less than two hours operation. It is obvious
why platinum catalysts can degrade so rapidly at high temperatures. On the
other hand, many high melting, low vapor pressure base metal oxide com-
pounds exist. Many are catalytically active or can be made so by proper
doping. Cobalt aluminate is an example of a catalytically active oxide
compound (Reference 1).
5. Slip coated catalysts should be avoided. Although in principle it
should be possible to make a high temperature slip coated catalyst the re-
quirement for tight, reliable bonding of the slip to the substrate would
appear to require both a very high firing temperature and careful selection
of slip and substrate materials for compatibility. Inasmuch as even
conventional auto exhaust catalysts lose significant amounts of slip in use,
slip coated catalysts would not appear to be the preferred choice, at least
for gas turbine applications. Preferrably, a high temperature combustion
catalyst should be made by forming a catalytic composite into a desired
structure and then firing, a procedure already used with many low tempera-
ture commercial catalysts.
338
-------
MATERIALS OF INTEREST
As previously noted, a number of oxide compounds exist which meet the
basic requirements for a high temperature combustion catalyst. It is im-
portant to note that refractories are presently used at even higher temper-
atures than those contemplated for combustor catalysts. Alumina has even
been used at temperatures as high as 1900C (Reference A). Of course, an
alumina catalyst for operation at even 1500 or 1600C can not be expected to
maintain much surface area. It is fortunate that high surface area is not
required for combustor combustor catalysts.
Of the materials which might be considered for combustor catalysts,
those based on zirconia or alumina are of most interest. This is primar-
ily because there is more data available for these materials and because
systems based on these materials could be readily produced commercially.
The highest melting oxides available are thoria (MP 3220C), urania
(MP 2876C), hafnia (MP 2837) and zirconia (MP 2760). Of these thoria is
reported to be one of the most stable of the refractory oxides (Reference 5).
However, it is radioactive and therefore is not a prime candidate material.
Zirconia, on the other hand, is readily available and there is a great deal
of data on its use.
ZIRCONIAS
Although zirconia is not catalytically active, it should be readily
possible to dope zirconia with say cobalt or nickel. A calcium stabilized
zirconia with nickel replacing some of the calcium should make a very good
candidate material.
Another zirconia candidate is zirconia-spinel. As reported by Corning
at the last Catalytic Combustion Workshop, a zirconia-spinel honeycomb is
suitable for operation at 1700C. If some of the magnesium in the spinel is
339
-------
replaced by nickel, one has suitable catalytic honeycomb. Many other cata-
lytic materials could be utilized with zirconia. Any spinel with catalytic
properties would be a good possibility since most spinels melt at about the
same temperature.
The major drawback to zirconias is poor thermal shock resistance.
However, considerable toughening can be achieved by various techniques such
as the introduction of microcracks. Nevertheless, a flexible structure such
as Coming's flexible cell design would seem to be important.
ALUMINAS
Alumina is perhaps the most widely used of the refractory oxides.
Although alumina has a melting point of only 2000C, most alumina spinels
melt somewhat higher (^ 2100C). Inasmuch as alumina can be used at least
for short periods at 1900C, 1600 or 1700C does not appear out of reach for
alumina spinel type catalysts. As is the case with zirconia, pure alumina
is not a good catalyst. However, a number of catalytically active spinels
are known, e.g. cobalt aluminate. Many doping possibilities exist.
Although better than zirconia, alumina has relatively poor thermal
shock properties. Also as with zirconia, toughening techniques exist.
Chromium alumina composites,1 for example, can survive many hundreds of
thermal shock cycles of a severity which destroys pure alumina samples in
the first cycle. Further, the resistance to oxidation at 1500C of a
30%Cr - 70%A1203 composite is reported to be excellent (Reference 6).
Chromium of course is catalytically active.
CATALYTIC COMPONENTS
Lanthanum chrome is not only catalytically active (Reference 7), but
is also a good high melting refractory with good mechanical properties.
This material is presently being studied for use in MHD electrodes.
A large number of catalytically active oxide compounds exist (Refer-
ence 7). Although many have relatively low melting points, others such as
lanthanum chrome have high melting points. Even the lower melting compounds
could be used as dopants in higher melting systems.
340
-------
Ideally, however, the catalytic material should have good structural
i
as well as good catalytic properties, as is the case with lanthanum1chrome.
Another such material is magnesium chrome spinel. As with lanthanum chrome,
magnesium chrome spinel is being considered for use as a high temperature
MHD electrode material (Reference 8).
341
-------
CONCLUSIONS
Prospects for the development of durable combustion catalysts for
operation at 1600-1700C appear to be excellent. A number of readily avail-
able materials meet the basic requirements for such a catalyst. Fabrication
into a suitable structure of a selected matrix material incorporating the
catalytic component avoids the difficulties associated with slip coated
catalysts. Such an approach permits catalysts with optimal mechanical as
well as catalytic properties.
Base metal oxides are required as the catalytic component inasmuch as
platinum metals have a high vapor pressure at combustor operating conditions.
In general, the choice of materials is dictated much more by considerations
such as melting point, chemical stability, and vapor pressure than by cata-
lytic properties. At the high temperature conditions contemplated for cata-
lytic combustor .operation,- solid state reactions are rapid and materials
must be chosen on the basis of their behavior as ceramics. Suitable materi-
als include stabilized zirconias, spinels, zirconia spinels and lanthanum
chrome,
Although suitable materials can now be specified, much work remains in
developing formulations which optimize both structural and catalytic proper-
ties. A program to develop thermal shock resistant high temperature cata-
lytic structures for combustion catalysts is warranted.
342
-------
REFERENCES
1. Schachner, H. Cobalt Oxides as Catalysts. COBALT, December 1960.
2. Kesselring, J. P., W. V. Krill, and R. M. Kendall. Design Criteria for
Stationary'Source Catalytic Combustors. Second Workshop on Catalytic
Combustion, Raleigh, North Carolina, 1977.
3. Alcock, C. B., and G. W. Hooper. Thermodynamics of the Gaseous Oxides
of the Platinum - Group Metals. Proc. Royal Society (London) A, 254:
551-561, 1960.
4. Campbell, I. E. High Temperature Technology. John Wiley and Sons,
New York, 1957.
5. Ryshkewitch, E. Oxide Ceramics: Physical Chemistry and Technology.
Academic Press, New York, 1960.
6. Shev'lin, Thomas S. Oxide-Base Cermets - Chromium-Alumina.
In: Cermets, J. R. Tinklebaugh and W. B. Crandall, eds. Reinhold, New
York, 1960. Chapter V pp. 97-109.
7. Voorhoeve, R. J. H., J. P. Remeika, and L. E. Trimble. Defect Chemistry
and Catalysis in Oxidation and Reduction over Perovskite-type Oxides.
Annals of the N. Y. Acad, of Sci., 272: 3-21. 1976.
7. Negas, T. Private communication.
343
-------
ENVIRONMENTAL ASPECTS
OF
LOW BTU GAS-FIRED
CATALYTIC COMBUSTION
By:
B.A. Folsom, C.Courtney
and M.P. Heap
Energy and Environmental Research Corporation
8001 Irvine Boulevard
Santa Ana, California 92705
345
-------
ABSTRACT
One alternative to the economic and efficiency penalties of stack gas
sulfur removal from conventional coal-fired power plants is to gasify the
coal, remove the sulfur products from the fuel gas and fire the essentially
sulfur-free gas in a high-efficiency combined gas turbine-steam turbine
power cycle.
The low Btu gases produced in these systems can contain high levels of
ammonia and nitrogen oxide emissions could be a problem if the ammonia is
oxidized to NO in the combustion process.
This paper presents some preliminary data on the processing of nitroge-
nous compounds in low Btu gas-fired catalytic combustors. Three experiments
are discussed. In the first experiment an ammonia-containing low Btu gas
was fired in a catalytic combustor. Nearly all of the NH3 was oxidized to
NO under lean conditions, but approximately half was reduced to N2 under
rich conditions. The second experiment was designed to simulate the opera-
tion of a reheat combustor firing low Btu gas in an oxidant-containing NO.
Under rich conditions up to 85 percent of the NO was reduced to N2« Methane
doped with ammonia was the fuel in the third experiment and the results
obtained here were found to be similar to results obtained by other
Investigators.
347
-------
ACKNOWLEDGMENT
The work upon which this publication is based was prepared under Contract
No. 68-02-2196 with the Environmental Protection Agency. The authors wish to
express their appreciation to Messrs. J. Johnson and J. Keene of Energy and
Environmental Research Corporation, Dr. J. Kesselring of Accurex Corporation,
and Mr. G.B. Martin for their assistance in various portions of the work.
349
-------
SECTION 1
INTRODUCTION
This paper addresses the catalytic combustion of low Btu gas (LEG) in
combined gas turbine-steam turbine power cycles. The rising demand for
electrical power coupled with the limited availability of petroleum mandates
the construction of many alternate fuel-based power plants by 1985-1990.
These power plants must have high overall efficiency, low net cost of elec-
tricity and must be environmentally acceptable. Since these plants will
commence operation in the future when environmental restrictions are pro-
jected to be much more restrictive than at present, the environmental design
goals should be to minimize emissions rather than to meet current New Source
Performance Standards.
Several advanced coal-based power cycles are currently being developed
for this application (1,2,3,4). The integrated gasifier LBG-fired combined
gas turbine-steam turbine cycle is one of these alternatives. The primary
advantage of this cycle is the potential for low emission of sulfur products
without the economic and efficiency penalties associated with stack gas sul-
fur removal (5,6). Since the coal gasification process operates fuel-rich,
the sulfur in the coal is converted mainly to t^S in the low Btu offgas
which can be removed (potentially) more easily than S02- Furthermore, the
H2S concentration is high and the total volume of gas to be processed is
small in comparison with total stack gas. Energy losses associated with the
gasification and cleanup processes can amount to as much as 30 percent of the
coal's heating value. However, if the clean LEG is fired in a gas turbine
combustor as part of a combined cycle power plant the gains in overall
COS is the second principal LEG sulfur product.
351
-------
power plant performance, when compared to conventional steam cycles, may
more than.offset the losses in the gasification and cleanup processes.
Figure 1 is a simplified schematic diagram of a LBG-fired gas turbine-
steam turbine combined cycle power plant with integrated gasifier, Only
the major heat, work and mass flow paths have been shown for clarity. The
system consists of a gas turbine topping cycle with exhaust heat transferred
to a steam turbine bottoming cycle through a waste heat recovery boiler.
Low Btu gas is produced in a gasifier supplied with coal, compressed air
from the gas turbine air compressor, and steam from the bottoming cycle.
This "integration" with other cycle components significantly reduces the
energy losses attributable to the gasification process.
"Raw" LEG exits the gasifier at temperatures as high as 1370 K depending
upon gasifier design and operation containing three pollutant precursors:
• Sulfur compounds which can be oxidized to SOX in the combustor,
• Particulates including carbon, tars and ash which may damage
turbine blades and may be emitted to the atmosphere,
• Ammonia which may be oxidized to NOX in the combustion process.
Raw LEG is processed through a gas cleanup system to reduce the concen-
trations of these materials prior to combustion. Several gas cleanup systems
have been developed or are under development for this application. Hot gas
cleanup systems process the LEG without significantly reducing its tempera-
ture and remove a majority of sulfur products and particulates. Cold gas
cleanup systems require cooling the LEG to near ambient temperature (400 K),
and they remove a majority of the sulfur products and particulates and may
also remove a substantial portion of the ammonia. The sensible heat removed
from the LEG may be used for a variety of purposes, but it effectively
bypasses the gas turbine topping cycle and contributes to reduced efficiency.
The thermodynamic trade-offs in hot and cold gas cleanup processes have
recently been investigated (7,8,9). Figure 2 shows the effects of LEG inter-
cooling on the efficiencies of the individual gas and steam turbine subcycles
and the overall cycle. The configuration considered was a. 1366 K (2000 F)
352
-------
turbine inlet temperature, 12.0 pressure ratio gas turbine with integrated
gasifier and a nonreheat steam bottoming cycle optimized for the gas tur-
bine waste heat (other cycle parameters are listed in Ref. 8). Three uses
for the LBG sensible heat are considered: dumping the heat into the environ-
ment, raising steam and injecting it into the gasifier, and raising steam to
operate a separate high-efficiency steam cycle. Since LBG gasifier offgas
temperatures are usually well in excess of 900 K, dumping this sensible heat
is clearly wasteful. A more economical alternative is to generate steam
either for gasifier injection or to operate a separate steam cycle. Figure 2
shows that all three alternatives reduce overall cycle efficiency. Depending •
upon gasifier design and operating parameters, the sensible heat in LBG off-
gas can amount to as much as 20 percent of the total heat release and the
losses due to intercooling the LBG with a cold gas cleanup system can be
substantial.
From a combustor design point of view, the choice of a hot gas cleanup
system as opposed to a cold system could well have two important effects:
• Higher adiabatic flame temperature
• Higher NH^ concentration
The sensible heat in the LBG contributes directly to the adiabatic flame
temperature. For typical LBG compositions, cooling the gas by 400 K will
decrease the adiabatic flame temperature at stoichiometric conditions by
about 200 K thus reducing the potential for thermal NOX formation. Cooling
the LBG will also require an increased LBG flow rate and reduced stoichi-
ometry (percent theoretical air) to achieve the same turbine inlet temperature.
The concentration of NH3 in raw LBG depends upon gasifier design and
operating parameters and may be as high as 0.38 percent (6). With hot gas
cleanup systems no NH3 is removed, and the entire amount will enter the gas
turbine combustor and could be oxidized to NOX. A concentration of 0.38 per-
cent NH3 in the LBG will produce approximately 3.2 Ib N02/106 Btu if all NH3
is converted to N02. This is far in excess of the current New Source Per-
formance Standard, 0.7 Ib N02/106 Btu.* Thus, one important challenge to
This is the current New Source Performance Standard for solid fossil fuel-
fired steam generators with greater than 250 x 10& Btu/hr heat input. The
comparable standard for a gaseous fuel is 0.2 Ib N02/10^ Btu.
353
-------
the combustor designer is to minimize the conversion of NH3 to NOX.
If a low temperature cleanup process is employed, a substantial por-
tion of the NH3 may be removed — NH3 levels from 100 to 400 ppm have been
reported (3,4). While these levels are sufficiently low to meet current
New Source Performance Standards even with full conversion, further NOX
control may be necessary to meet future standards.
Most proposed LBG-fired combined cycles utilize the basic cycle con-
figuration illustrated in Figure 1. However, a recent study of several
alternative cycle arrangements has shown the potential for significantly
improved performance (7,8). One very promising alternative cycle is the
reheat cycle shown in Figure 3. In this cycle the hot combustion products
from the main gas turbine combustor are part:ally expanded in a high pres-
sure turbine and then reheated in a second combustor before being expanded
to atmosphere. The reheat combustor could be fired with LEG from the main
high pressure gasifier throttled to the proper pressure, or (more efficiently)
from a second midpressure gasifier as shown in Figure 3. The performances
of a reheat cycle with 1366 K (2000 F) turbine inlet temperatures and a
basic cycle with the same turbine inlet temperature are shown in Figure 4.
Reheating increases the gas turbine subcycle efficiency, particularly at high
pressure ratios. It also increases the steam bottoming cycle efficiency by
increasing the turbine exhaust temperature, thus allowing higher temperature
steam to be generated. The result is a substantial increase in overall cycle
performance. The reheat combustor also provides the opportunity to reprocess
the NOX formed in the main combustor through a second reaction zone.
The design and operation of low NOX combustors for LBG-fired combined
cycle power plants has been considered by several investigators (3,4,7,8,11,
12). The development of low NOX combustors for LBG-fired combined cycle power
plants is the subject of EPA Contract 68-02-2196. Several alternative com-
bustor designs are being investigated analytically and experimentally in a
search for the lower bounds of pollutant emissions limited by chemistry and
practical mixing constraints (12,13). Catalytically supported combustion is
one of the concepts being considered.
354
-------
The remainder of this paper discusses some recent measurements of
nitrogenous specie processing in an atmospheric pressure LBG-fired catalytic
reactor. Three experiments are discussed. In the first experiment a LBG
with NH3 concentration comparable to that obtained from a cold gas cleanup
system was fired over a range of stoichiometries. The oxidant was air
diluted with N£ to maintain the equilibrium adiabatic flame temperature at
1473 K. In the second experiment the NH3 was replaced by an equal amount
of NO to simulate a reheat combustor reprocessing NOX from a high pressure
combustor. The third experiment utilized CH4 doped with NH3 as the fuel,
and was included to compare results from this catalytic reactor with those
of prevoius investigations firing CH4 in other reactor designs.
355
-------
SECTION 2
EXPERIMENTAL APPARATUS
CATALYTIC REACTOR
The catalyst used in these tests was a platinum coated graded cell
monolith supplied by the Acurex Corporation. The desirable features of
the graded cell catalyst have been discussed previously (14) and include:
• Consistent low temperature light-off
• High throughput without blowout
• Low CO emissions
The catalyst used for these tests was 25 mm (1.0 in.) diameter and
included three 25 mm (1.0 in.) long segments of varying cell sizes. Each
segment consisted of a hexagonal cell Du Pont alumina monolith washcoated
with Du Pont alpha alumina and subsequently loaded with platinum. The cell
sizes and platinum loadings were as follows:
Axial
Position
Upstream
Center
Downstream
Cellsize
mm
7
5
3
Platinum
Loading
wt %
6.3 - 6.4
5.2 - 5.4
2.1
The maximum recommended temperature for this catalyst is 1588 K (2400 F).
Type K thermocouples were cemented in two of the cells in the downstream
segment to monitor maximum monolith temperature.
The 25 mm (1.0 in.) monolith diameter was chosen to be compatible with
the characteristics of the LEG synthesizing system described in the next
356
-------
section. This relatively small diameter raises two important questions:
• Is the system dominated by edge effects?
• Will heat loss from the periphery be significant and prevent
the attainment.of near-adiabatic conditions?
The gas passages near the center are bounded on all sides by platinum-
coated monolith surfaces. In contrast, the outermost gas passages are
partially bounded by the surrounding supporting material. The reactants
passing through the outer passages will thus encounter a reduced amount of
catalytically active surface area and may process the reactants differently.
This "edge effect" is aggravated by large cell sizes and small overall mono-
lith diameters such as the upstream segment of the monolith tested here.
The ratio of surrounding tube surface area to monolith surface area is an
index of the magnitude of edge effects. It can be shown that this ratio is
approximately 0.5 d/D where D is the monolith outside diameter and d is the
hexagonal cell size. For the catalyst segments utilized here, these ratios
were 0.14, 0.10 and 0.06. Thus, the surface area presented by the surround-
ing tube surface was small compared to monolith surface area.
To minimize heat losses and approximate adiabatic conditions, the cata-
lyst was mounted in a refractory tube and electrically backheated. The
catalyst and its housing are shown schematically in Figure 5. The tempera-
ture between the refractory tube and the heating element was measured with
a thermocouple and the heating power was controlled to maintain this tem-
perature near the equilibrium adiabatic flame temperature of the reactants.
The reactants passed through a sintered stainless steel disc immediately
upstream of the catalyst. This disc had three functions:
• Flame arresting in the event of flashback
• Velocity profile smoothing
• Temperature profile smoothing
A thermocouple downstream of the disc measured reactant temperature. Flash-
back was observed as a rapid rise in this temperature as a flame front moved
toward the disc.
357
-------
The combustion products discharged into a 100 mm (4.0 in.) exhaust
stack and a water-cooled stainless steel sample probe was installed to mea-
sure combustion product concentrations. The outside diameter of the probe
was 9.4 mm (3/8 in.) and the probe tip was 44 mm (1-3/4 in.) from the down-
stream catalyst face.
FLOW SYSTEM AND INSTRUMENTATION
The flow system and instrumentation utilized in these experiments are
shown in Figure 6. LEG was synthesized by blending together high purity
gases from cylinders. All gases were high purity grade (99.97 percent or
better) with the exception of CO (99.0 percent). NH3 and NO were supplied
as custom grade mixtures (±2 percent accuracy) to facilitate metering:
1.0 percent NH3 in Ofy and 8570 ppm NO in N£. Oxidant was supplied as dry
air in pressurized cylinders. Oxygen concentration ranged from 19.9 to
23.1 percent and was adjusted by nitrogen addition to the concentration
required' to achieve the desired flame temperature.
All gases were metered with sapphire jewel critical orifices. Pres-
sures upstream of the orifices were measured with high accuracy (iO.ll per-
cent full-scale) variable capacitance pressure transducers and the down-
stream pressures were maintained constant. All transducers were calibrated
weekly against a common laboratory reference. The flow rate through each
orifice was calibrated by filling an evacuated tank. The total inaccuracy
in each gas flow rate was less than 0.5 percent.
Reactants were preheated electrically to a maximum of 773 K. The
reactant temperature was monitored with the thermocouple downstream of the
sintered disc and the preheat power was adjusted to achieve the desired
temperature.
Combustion products were sampled.with a water-cooled stainless steel
probe and analyzed for 02, C02, CO, NO, NOX, NH3 and HCN. The sample train
components were constructed entirely of stainless steel, glass and Teflon.
NH3 and HCN were measured by bubbling known volumes of combustion products
through three water baths in series. The NH3 and HCN were absorbed in the
water and measured with Orion specific ion instrumentation. The accuracy
358
-------
of this procedure was checked by synthesizing known concentrations of NH3
and HCN in air and processing the mixtures as combustion products. Agree-
ment between calculated and measured concentrations was consistently within
10 percent.
The other product species were monitored continuously. The sample was
dried to 273 K dewpoint in a cyclone water trap located as close as possible
to the sample probe. Residence time between probe and trap was less than
0.5 sec. Residence time between trap and instruments was less than 2.0 sec.
The chemiluminescent analyzer was a Thermo Electron Model 10 A equipped
with a stainless steel NOX converter operated at 1073 K. CO and C02 were
measured with Beckman NDIR analyzers and the 02 was measured with a Taylor
paramagnetic instrument. These readings were used to calculate combustion
efficiency and close the carbon balance.
359
-------
SECTION 3
EXPERIMENTAL RESULTS
LEG TESTS
The term low Btu gas does not refer to a specific gas composition, but
rather to a family of fuel gases produced by reforming coal with air and
steam. The composition of LEG depend upon:
• Coal composition
• Gasifier design and operating conditions
• Air and steam flow rates
• Product gas cleanup system
The heating value of LEG may range from'16,000 to 41,000 J/m3 (80 to 200 Btu/
ft3) and the primary combustible specie are CO and H2- The ratio of CO to
H2 concentrations ranges from 0.5 to 2. Hydrocarbon fuel gases (normally
CH4) may also be present in quantities up to 10 percent, and nitrogen is the
primary diluent comprising 35 to 55 percent of the fuel gas with the remain-
ing diluents being C02 and H2°- Trace amounts of H^S, COS and NH3 may also
be present, since these specie are not entirely removed from the product
gas by the- gas cleanup system.
The relationship between LEG composition and combustor performance is
currently under investigation as part of EPA Contract 68-02-2196 (13). For
the experiments reported in this paper the following LEG composition was
synthesized using the metering system described earlier:
CO = 20 percent
H2 = 20 percent
CH4 = 5 percent
N2 = 55 percent
360
-------
The higher heating value of this synthesized LEG is 37,200 J/m3,
(182 Btu/ft3) and the adiabatic equilibrium flame temperature at stoichio-
metric conditions is 2135 K. Since this temperature is higher than the
maximum acceptable catalyst temperature of 1588 K, it was necessary to dilute
the reactants with nitrogen to ensure that this temperature was never exceeded
within the catalyst bed.
An initial test series was carried out to select suitable catalyst
operating parameters for troublefree operation. This LEG composition was
found to light-off consistently at room temperatures under a wide range of
both fuel-rich and fuel-lean conditions. Under fuel-lean conditions with
adiabatic equilibrium temperatures in excess of 1273 K and back heating the
catalytic reactor to control heat loss, the catalyst temperatures rose to
essentially adiabatic equilibrium flame temperatures. Measured exhaust CO
concentrations were less than 50 ppm (often less than 5 ppm). Measured
concentrations of C02 and 02 confirmed complete combustion within the limits
of experimental error. Under similar but fuel-rich conditions the catalyst
temperature also reached adiabatic equilibrium flame temperature. When
excess catalyst heat loss or low adiabatic equilibrium flame temperatures
prevented the attainment of 1273 K exhaust CO concentrations rose substan-
tially, indicating incomplete combustion.
Maximum turbine inlet temperatures (TIT) for state-of-the-art stationary
gas turbines are in the range of 1400 to 1500 K (2000 to 2200 F). As turbine
cooling technology improves, the maximum TIT is expected to increase and
1700 K (2600 F) has been projected for the next generation of gas turbines
for utility use (6). An adiabatic equilibrium flame temperature of 1473 K
(2200 F) was selected for the experimental investigation based upon the
information on TITs and the operating'characteristics of the catalyst.
The catalytic reactor operates with premixed reactants, and it is possible
for a flame front to propagate upstream through the reactants from an ignition
source on the catalyst under certain conditions. It is not sufficient that the
average reactant velocity be greater than the mixture flame speed since flash-
back can occur in the boundary layer along the tube wall or in other low
velocity regions. In these experiments the average reactant velocity was
maintained at 3.0 m/sec (10.0 ft/sec) and this was sufficient to prevent
361
-------
flashback. This corresponds to a space velocity of 1.4 x 10 /hr and an
average catalyst residence time of 13 msec. The heat release rate was
3.9 x 106 J/hr (3700 Btu/hr) or 1.0 x 1011 J/m3 (2.7 x 10& Btu/ft3).
Ammonia Processing
An LEG cleaned with a cold sulfur removal process contains low con-
centrations of ammonia typically in the range of 100 to 500 ppm. The LEG
composition discussed above was doped with 471 ppm NH3 and mixed with air
and nitrogen to adjust the adiabatic equilibrium flame temperature before
passing to the catalytic reactor. Figure 7 shows the measured concentra-
tions of NO, NOX, NH3 and HCN as ppm dry and the calculated percentage con-
version of the input ammonia to the various XN specie. For lean mixtures
complete conversion of the input ammonia is constant due to the nitrogen
dilution, but under rich conditions complete conversion increases because
the LEG fraction of the reactants increases.
Under fuel-lean conditions approximately 90 percent of the input ammonia
is converted to NOX. As the stoichiometry is reduced below 100 percent of
the theoretical air required for complete combustion, the fraction of NH3
converted to NO decreases rapidly to essentially zero at 52 percent theo-
/
retical air. As the stoichiometry is reduced, the fraction of ammonia
passing through the reactor increases from near zero at 100 percent theo-
retical air to over 50 percent at 44 percent theoretical air. HCN is pro-
duced at stoichiometries less than 80 percent theoretical air, and at 44 per-
cent theoretical air the HCN concentration corresponds to almost 50 percent
conversion of the NH3 in the original reactants.
A low NOX combustor might well consist of a staged heat release system
in which the products of a fuel-rich primary zone are burned out in a lean
secondary zone. The challenge to the combustor designer is to minimize the
conversion or retention of the NO, NH3 or HCN in the primary zone products
to NOX in the lean burnout zone. It has been shown that under certain con-
ditions large portions of these nitrogenous compounds can be converted to
NOX in lean combustion zones (13). Consequently, it is of considerable
interest to minimize the total XN specie exiting the rich primary zone.
The percentage,of input ammonia converted to ZXN specie is plotted in Fig-
ure 8 as a function of percentage theoretical air. Under rich conditions
362
-------
ZXN was calculated as the sum of NO, NH3 and HCN, and under fuel-lean
conditions it was assumed to be entirely NOX. It can be seen that for a
range of stoichiometries between 60 and 100 percent theoretical air the ZXN
is minimized. It should be noted that the operation of the catalyst in this
range requires the addition of substantial amounts of diluents (in this
instance nitrogen) to reduce flame temperatures. In practical systems
flame temperatures could be limited either by the recirculation of combus-
tion products or by radiantly cooling the primary 'section in a similar manner
to that proposed by Kesselring (14). Adiabatic combustion of this LEG in
air without temperature control requires stoichiometries richer than 40
or leaner than 230 percent theoretical air to avoid, temperatures in excess
of 1473 K. Under both conditions ZXN is essentially 100 percent.
Nitric Oxide Processing
A reheat combustor incorporated in a combined cycle power plant would
utilize the partially vitiated lean combustion products exiting the high
pressure turbine as oxidant and additional LEG as fuel. The NO produced in
the high pressure combustor would be reprocessed through a second reaction
zone and could be partially reduced to N£. Thus, the reheat combustor offers
the potential for both NOX control as well as a gain in overall cycle
efficiency. As a first order simulation of an LBG-fired reheat combustor
the LEG composition referred to previously was doped with 471 ppm NO and
fired in the catalytic reactor under the same conditions as those described
for the ammonia doping tests. Since no NH^ was present in the reactants,
this experiment simulates the combustion of an LEG with all the NH-j removed.
It could be speculated that if the premixing occurred under the correct
temperature, conditions that the ammonia present in an LEG gas could homo-
geneously reduce the NO produced in the high pressure combustor.
Figure 9 presents the results of the experiments involving NO doping
in the same format as Figure 7. Combustion products were analyzed for
ammonia and HCN at all stoichiometries richer than 120 percent theoretical
air. However, concentrations were found to be zero within in the limits of
the experimental techniques. Under fuel-lean conditions essentially all of ,
the NO entering with the reactants exited as NOX. However, under rich
363
-------
conditions the catalytic reactor was an effective method of reducing NO.
Between 60 and 80 percent theoretical air 78 to 85 percent of the NO was
removed and at stoichiometric conditions only 50 percent of the input NO
was retained in the exit stream.
METHANE TESTS
Several investigators have tested catalytic reactors firing hydrocarbon
fuels doped with NH3 (14,15), and it is of interest to operate the catalytic
reactor tested here with similar fuels to provide a comparison with other
investigations. Consequently, the catalytic reactor was fired with CH^
doped with NH3 to give a concentration of 9417 ppm NH3 in City, and the
reactants were diluted with N£ as before to adjust the adiabatic equilibrium
flame temperature. Similar test runs were conducted with methane as those
described earlier for LEG to identify suitable catalyst operating parameters.
With LEG light-off occurred consistently at room temperatures. However, it
was found necessary to preheat the methane mixtures to between 573 and 673 K
in order to sustain ignition. With LEG the reactant velocity (space velocity)
was required to be high to avoid flashback. If the catalytic reactor was
operated at similar velocities with methane, blowout occurred and it was
found necessary to reduce the space velocity to 4.7 x 10 /hr for sustained
ignition. Flashback was not a problem with the methane mixtures.
Measured nitrogenous specie concentrations are presented in Figures 10
and 11 in the same format as those described earlier. In comparison with
the LEG data, the actual concentrations are much higher due to the higher
concentration of ammonia in the reactants. However, the percentage of ammonia
converted to the various XN specie to the various nitrogenous specie are
similar. The ZXN also exhibit the same broad minimum over the range 70 to
100 percent theoretical air.
364
-------
SECTION 4
DISCUSSION
The results presented in this paper are part of a more general study
to develop low emission combustors for application to advanced power sys-
tems (8). The approach being taken in this study involves both numerical
and experimental investigations to define combustor design parameters to
minimize NOX production from simulated LEG gases containing trace quantities
of NH3 and H2S. It is generally recognized that the most cost-effective
technique for controlling NOX emissions from stationary sources is to modify
the combustion process in such a way as to prevent NO formation. For fuels
containing nitrogen species (other than N2) this normally involves dividing
the total heat release process into two zones. Since it is known that fuel
nitrogen conversion to NO is minimized under fuel-rich conditions, the fuel
is caused to react in the first zone with less than the stoichiometric air
requirement. The remainder of the air is provided in the second zone to
complete the heat release process. Minimum NOX emissions require that:
• The rich combustion products formed in the initial zone contain
minimum quantities of nitrogen specie such as NO, NH3 or HCN.
• The rich combustion products are burned in such a way to prevent
thermal NO production (oxidation of molecular nitrogen) and to
maximize N2 production from the NO, NH3 and HCN which might be
present in the reactants exiting from the first zone prior to
second stage burnout.
Thermal NO production can be essentially eliminated in the second heat
release zone if peak temperatures never exceed 1800 K (2800 F), -Consequently,
the investigation to define combustor design parameters for minimum NOX
365
-------
emissions are concentrating upon the minimization of XN specie in rich
combustion products and those factors which maximize N2 production during
the burnout of these products.
Thermodynamic calculations indicate that equilibrium ZXN concentration
in rich combustion products of fossil fuels and air is less than 10 ppm under
optimum conditions. This provides a target emission level for advanced com-
bustors; however, typical emission levels far exceed this level suggesting
that the process of N2 production is kinetically limited. Two competing
reaction paths (16) occur under combustion conditions which account for the
production of N2 or NO. These are:
NX + Oxidant ->• NO + A
NX + NY -»• N£ + Z
The precise details of Path B which predominates under fuel-rich conditions.
are open to question. When ammonia is the initial fuel nitrogen specie, it
will break down to give NH£ which will then produce N2 via Path B. For high
temperature flames Morley (17) has shown that regardless of their nature,
nitrogen compounds are quantatively converted to HCN in the reaction zone of
premixed rich hydrocarbon flames. Two possible mechanisms for HCN formation
are:
CH3 + NO -»• NH3NO -»• CH2NOH •*• HCN (1)
or
CH + NO -* HCN + 0 (2)
The subsequent oxidation of HCN via NCO as a possible intermediate would then
allow interconversion of the three major stable nitrogenous specie via
HCN + OH J CN + H20 (3)
CN + OH £ NCO + H (4)
NCO + H £ NH + CO (5)
NHi + H J NH1_1 + H2 (6)
N + OH £ NO + H (7)
N + NO £ N2 + 0 (8)
366
-------
If hydrocarbon specie are present other reactions are possible sources of
HCN, such as
CH + N£ £ HCN + N (9)
or
CH2 + N2 2 HCN + NH (10)
Fuels containing bound nitrogen normally contain sulfur compounds and Wendt
(18) and co-workers have shown that the presence of sulfur specie may affect
the interconversion of nitrogenous specie in rich combustion products.
The combustor designer is faced with practical constraints such as
size, pressure drop, construction cost and maintainability, but the basic
design parameters controlling ZXN emission from the rich first stage are
the method of fuel/air contacting, and the time, temperature, and stoichio-
metric history of the reactants. Numerical investigations (19) associated
with a search for the lower bounds of fuel nitrogen conversion involving
ideal fuel/air contacting have indicated several potentially promising com-
bustor concepts. Two of these concepts are illustrated in Figure 12. If
the conversion of NH3, NO and HCN to N2 in rich products is kinetically
limited, then the rate of this conversion can be accelerated by increasing
the temperature of the reactants (20). An idealized combustor design (see
Figure 12a) which relies upon heat transfer feedback in the first stage zone
is one method of achieving this effect while ensuring that the second stage
temperatures are too low to allow thermal NO production. The rate of ZXN
decay in rich combustion products is not only dependent upon temperature, but
also upon stoichiometry and the optimum primary reactor might well include
distributed air addition to vary stoichiometry as a function of time (see
Figure 12b).
The idealized combustor concepts illustrated in Figure 12 do not indicate
how the fuel and air are mixed or how ignition stability is achieved. The
experimental results presented in this paper concern one method of fuel/air
contacting — the use of a catalytic reactor. Results are presented which
show the stable nitrogenous specie produced as a function of reactant stoi-
chiometry for a constant adiabatic flame temperature when a simulated LEG
'367
-------
and methane were doped with NH3 and when the LEG was doped with NO. The
results with ammonia doping are consistent with the kinetic mechanisms
described above and are similar to results obtained with premixed flat
flames. However, it is somewhat surprising that no HCN or ammonia was
observed with NO doping.
Under fuel-rich conditions Path B predominates and N£ is produced,
resulting in low NO emissions. However, if the reactants are too rich those
reactions involving HCN oxidation predominate and although NO is no longer
produced, the NH3 present in the original reactants either does not decompose
or produces HCN. The ratio of NH3 to HCN will depend upon reactant tempera-
ture as well as stoichiometry. It can be speculated that the failure to form
either HCN or NH3 when LEG is doped with NO may be associated with the low
reactant .temperature or the influence of the catalyzed heterogenous
reactions.
Recognizing the preliminary nature of the results, it appears that a
catalytic reactor offers considerable promise for utilization in low emis-
sion combustors firing LEG. Future effort will concentrate upon determining
the effect of LEG composition on NH3 and NO processing in catalytic reactors
since it has been shown that fuel composition has a significant effect upon
fuel NO production in diffusion flames (13). Investigations are currently
underway to couple a catalytic reactor to a second stage burnout reactor to
investigate its potential for application in the combustor concepts shown in
Figure 12.
368
-------
REFERENCES
1. Evaluation of Phase 2 Conceptual Designs and Implementation Assessment
Resulting from the Energy Conversion Alternatives Study (EGAS), prepared
under Interagency Agreement E(49-18)-1751, NASA Report No. TM X-73515.
2. Shaw, H., Cerkanowicz, A.E., and Tung, S.E. Environmental Assessment of
Advanced Energy Conversion Technologies - Interim Report, Vbl. 1, State-
of-the-art. Contract No. 68-02-2146, U.S. Environmental Protection
Agency, Cincinnati, Ohio, 1977.
3. -Harris, L.P., and Shah, R.P. Energy Conversion Alternatives Study
(EGAS), General Electric Phase II Final Report: Vol. II, Advanced
Energy Conversion Systems - Conceptual Designs: Part 3, Open Cycle Gas
Turbines and Open Cycle MHD, General Electric Report No. SRD-76-064-2,
NASA Report No. NASA CR-134949, 1976.
4. Beecher, D.T., et. al. Energy Conversion Alternative Study (EGAS),
Westinghouse Phase II Final Report: Vol. II - Combined Gas-Steam
Turbine Plant Using Coal-Derived Fuel, Westinghouse Report No. 76-
9E9-ECAS-R2v.2, NASA Report No. NASA CR-134942, 1976.
5. Fluor Engineers and Constructors, Inc. Gasification-Combined-Cycle
Power Plants. EPRI Journal, July/August, 1978. p. 43
6. Robson, F.L., Blecher, W.A., and Giramonti, A.J. Combined-Cycle Power
Systems. EPA-600/2-76-149, U.S. Environmental Protection Agency,
Washington, DC, 1976. p. 359
7. Tyson, T.J., Heap, M,P., ,Kau, C.J., Folsom, B.A., and Brown, N.D. Low
NO Combustion Concepts for Advanced Power Generation 'Systems Firing Low-
Btu Gas. EPA-600/2-77-235, U.S. Environmental Protection Agency,
Washington, D.C., 1977.
8. Folsom, B.A., Corley, T.L<, Lobell, M.H., Kau, C.J., Heap, M.P., and
Tyson, T.J. Evaluation of Combustor Design Concepts for Advanced Energy
Conversion Systems. In.: Proceedings of the Second Stationary Source
Combustion Symposium, v.5, Addendum, EPA-600/7-77-073, July, 1977.
369
-------
9. Myerson, -A.L. The Reduction of Nitric Oxide in Simulated Combustion
Effluents by Hydrocarbon-Oxygen Mixtures. In: 15th Symposium (Inter-
national) on Combustion, The Combustion Institute, Pittsburgh, Pennsyl-
vania, 1975.
10. Muzio, L.J., Arano, J.K., and Teixeira, D.P- Gas Phase Decomposition
of Nitric Oxide in Combustion Products. In: 16th Symposium on Com-
bustion, The Combustion Institute, 1977. p. 199
11. Martin, G.B. NOX Considerations in Alternate Fuel Combustion. EPA-
600/1-76-149, U.S. Environmental Protection Agency. Washington, DC,
1976. p. 373
12. Heap, M.P., et al. Environmental Aspects of Low Btu Gas Combustion.
In: 16th Symposium on Combustion, The Combustion Institute, 1977-
p. 535
13. Folsom, B.A., Courtney, C.W., Heap, M.P., and Martin,' G.B. The Effect
of LEG Composition and Combustor Characteristics on Fuel NOX Formation.
In: 14th Annual International Gas Turbine Conference, A.S.M.E., Gas
Turbine Division, March 1979.
14. Kesselring, J.P., Krill, W.V., Kendall, R.M. Design Criteria for
Stationary Source Catalytic Combustors. EPA-600/7-77-073c, U.S.
Environmental Protection Agency, Washington, DC. July 1977. p 193
15. Matthews, R.D., and Sawyer, R.F. Fuel Nitrogen Conversion and Catalytic
Combustion. Presented at the Western States Section, The Combustion
Institute, Paper No. 77-40, LBL-6396, Fall Meeting, Palo Alto, CA,
October, 1977.
16. De Soete, G. La Formation Des Oxydes D'Azote Dans La Zone D'Oxydation
Des Flammes D'Hydrocarbures. No. 23 309, June 1975, Institut Francais
De Petrole, France.
17. Morley, C. The Formation and Destruction of Hydrogen Cyanide from
Atmospheric and Fuel Nitrogen in Rich Atmospheric-Pressure Flames.
Combustion and Flame, Vol. 27, 1976. pp. 189-204
18. Wendt, J.O.L., Corley, T.L., Morcomb, J.T. Interactions Between Sulfur
Oxides and Nitrogen Oxides in Combustion Processes. In: Proceedings
of the Second Stationary Source Combustion Symposium, Vol. IV., Funda-
mental Combustion Research. EPA-600/7-77-073d, July 1977. p 101
19. Tyson, T.J., Heap, M.P., Corley, T.L., and Kau, C.J. Fundamental Com-
bustion Research Applied to Pollutant Control. First Annual Report,
EPA Contract 68-02-2631. In preparation.
20. Sarofim, A.F., Pohl, J.H., Taylor, B.R. Mechanisms and Kinetics of
NOX Formation Recent Developments. Paper presented at 69th Annual
Meeting AIChE, Chicago. November, 1976.
370
-------
CO
Coal Input
Steam
Pressurized
Gasifier And
Cleanup System
Air
Inlet
Compressor
LEG
Adiabatic
Gas Generator
Steam Turbine
-e-
Generator
Condenser
Cooling Water
Turbine
I
Generator
Waste Heat
Boiler
Figure 1. Simplified Schematic of Basic LBG-Fired Combined Cycle
-------
35
CO
01
•H
U
g
•H
O
•H
14-1
4-1
w
30
25
22
Overall
Efficiency
TIT = 2000 F
Pressure Ratio = 12.
"""£» Intercooling
Heat Runs Steam Cycle
—O~~ Intercooling
Steam Injected Into Gasifier
^Q.— Intercooling
Heat Wasted
Steam Bottoming Cycle Efficiency
Gas Turbine
Cycle Efficiency
o
o.i
0.2
Heat Loss To Intercooling/Total Heat Release
Figure 2. Effect of LEG Intercooling on Combined Cycle Performance
372
-------
Coal
Steam
High Pressure
Gasifier and
Cleanup System
CO
^4
CO
Air
Inlet
Mid-Pressure
Gasifier and
Cleanup System
Adiabatic Gas
Turbine
Combustor
H.P.
Turbine
Coal
Steam
Adiabatic
Reheat
Combustor
L.P,
Turbine
Steam Turbine
-a-
Generator
Condenser
AAAAn
o
Feedwater
Pump
-3-
Generator-
Cooling Water
Figure 3. Simplified Schematic of LBG-Fired Combined Cycle With Reheat
T
Waste Heat
Recovery
Boiler
Exhaust
-------
45
40
cn
01
•H
O
§
•H
O
•H
14-1
U-l
W
35
30
25
Basic Cycle
Reheat Cycle
Overall
Efficiency
'*CL
Steam Cycle
Efficiency
Gas Turbine
Cycle Efficiency
I I
I
I
I
I
I
I
I
4 6 8 10 12 14 16 18 20 22 24 26 28 30
Pressure Ratio
Figure 4. LBG-Fired Combined Cycle: Basic Cycle Versus Reheat
374
-------
Exhaust
Stack
Monolith
T.C.
Steel
Shell
Water-Cooled
S.S. Sample Probe
Back Heat
T.C.
Preheat
T.C.
Graded Cell
Catalyst
Electrical
Heating
Element
1500 w.
100
Sintered
S.S. Disc
Castable
Refractory Cylinder
Preheated Reactants
Figure 5. Catalytic Reactor and Housing
375
-------
Co
~vj
cr>
Exhaus t
Stack
Gas Bubbler
Train for Wet
Sample
Zero and Span Gases
Teflon
Lined
Sample
Pump
High Purity Gas
Cylinders
9.9 9 9
8 Channel
Critical Orifice
Flow Metering
System
V Catalytic
I Reactor
In Situ
Calibration
System
Vent
Reactant
Preheater
500°C
Max.
Rotameters
N0/N0x
Chemiluminescent
Analyzer
CO
NDIR
co2
NDIR
°2
Paramagnetic
1 * * *
Bypass and
Wet Sample
Flowrate
Measurement
Figure 6. Flow System and Instrumentation
Vent
-------
oo
• NO - 800°C S.S. Teco
O NOX - Teco
AHCN - Specific Ion
• NH - Specific Ion
100 -
a
o
•H
O
u
PC
25
I
Full Conversion = 100%
200-
E
ft
cx
100 -'
200
Approximate
Full
Conversion
100
T.A.
% T.A.
200
Figure 7.
Processing in LBG-Fired Catalyst
-------
Full Conversion = 100%
100
CO
^1
co
25
X
CO
re
s
o
•H
ID
40
60
100
120 140 160
% T.A.
180
200
220
Figure 8. ZXN in NH Doped LBG-Fired Catalyst.
-------
O NO -
x
• NO
A HCN
• NH,
800 C S.S. Teco
Teco
Specific .Ion
Specific Ion
GO
100
B-S
O
•H
09
M
0)
50
I
Full Conversion = 100%
40
AAA A A
200
I,
ex
M
Q
100
Approximate
Full
Conversion
70
100
200
40
AAA A A-
% T.A.
100
200
% T.A.
Figure 9. NO Processing in LBG-Fired Catalyst
-------
00
CO
o
o
•u
o
•H
05
O
u
® NO -
x
O NO -
A
800 C S.S. Teco
Teco
Specific Ion
Specific Ion
100 -
Full Conversion = 100%
600
500
400 ~
FM
FM
300
i-i
Q
200 -
100 -
100
200
Approximate
Full
Conversion
100
200
% T.A.
% T.A.
Figure 10. NH Processing In CH,-Fired Catalyst
-------
100
co
CO
o
H
C
O
•H
CO
e
o
o
CO
33
3
50
40
60
80
100
120
140
160
180
200
% T.A.
Figure 11. EXN in CH,-Fired Catalyst
-------
Fuel-Rich
Stoichiometric
Fuel-Lean
Heat
Addition
OJ
co
in
Residence
Time At
High
Temperature
Heat
Removal
Q.
Q.
Air
Burnout
Combustibles
t Near
Stoichio-
metric
Heat Addition
And Dilution
To Final
Temperature
Lean
Exhaust
Products
(a)
Figure 12. Low NOX Combustor Concepts
-------
f
1
CO
00
CO
Air
Fuel
2
*3
Second
Stage
Burnout
Figure 12b. Fuel-rich primary
-------
THE ADVANCED LOW - EMISSIONS
CATALYTIC -COMBUSTOR PROGRAM
PHASE I - DESCRIPTION AND STATUS
By:
Andrew J. Szaniszlo
NASA - Lewis Research Center
385
-------
THE ADVANCED LOW-EMISSIONS CATALYTIC-COMBUSTOR PROGRAM
PHASE I - DESCRIPTION AND STATUS
by Andrew J. Szaniszlo
NASA - Lewis Research Center
ABSTRACT
The Advanced Low-Emissions Catalytic-Combustor Program is an
ongoing three-phase contract effort with the primary objective of
evolving the technology required for incorporating catalytic
combustors into advanced aircraft gas-turbine engines. Phase I
is currently in progress. At the present time, analytical evaluation
is being conducted on advanced catalytic-combustor concepts - including
variable geometry - with their known inherent potential advantages of
low-level pollutant emissions, improved operational stability, reduced
specific fuel consumption, and reduced pattern factor for longer tur-
bine life. Phase II and III consist of experimental development.
387
-------
CATALYST-COATED FLAMEHOLDER
FOR LEAN STABILITY IMPROVEMENT
By:
E. J. Szetela
J. B. McVey
United Technologies Research Center
East Hartford, Connecticut 06108
389
-------
ABSTRACT
A program is being carried out at United Technologies Research Center
to improve lean stability in premixed, prevaporized combustor concepts de-
signed for reducing gas turbine pollutant emissions. Among the techniques
being investigated is the use of a catalyst-coated flameholder similar in
principal to a perforated plate. The flameholder consists of a multitude
of coated cylindrical passages which terminate at the combustor inlet. A
mixture of fuel and air is flowed through the passages and at the discharge
end in the combustor, recirculation regions .are produced between the ports.
The recirculation regions extend the residence time of a portion of the mix-
ture and form a pilot for the flame. The catalyst-coated passages were de-
signed to produce a heated boundary layer that enters the recirculation
region and extends the piloting capability of the flameholder to leaner
mixtures.
391
-------
CONTENTS
Introduction 395
Program Goals 395
Approach 396
Planned Experiment 398
References 399
392
-------
LIST OF FIGURES
Figure 1 Flame Stabilization Process 400
Figure 2 Catalyzed Tube Performance 401
Figure 3 Catalyzed Flameholder 402
393
-------
LIST OF SYMBOLS
A - Area, ft2
C - Concentration, lb/ft3
d - Diameter, ft
D - Diffusion Coefficient, ft2/hr
G - Mass Velocity, rb/hr-ft2
M - Mass Flow Rate, lb/hr
q. - Heat Transfer Rate, BTU/hr
M - Absolute Viscosity, lb/hr-ft
P - Density, lb/ft3
Subscripts
D - Diffusion
F - Fuel
r - Recirculation
S - Surface
394
-------
INTRODUCTION
The operating characteristics of the combustors in current aircraft gas
turbines result in the emission of objectionable quantities of air pollu-
tants over a large part of the engine operating cycle. Considerable effort
has been devoted to reduction of pollutant emissions by modifying conven-
tional concepts (Ref. 1-3). Current emission regulations are concerned with
emissions generated during the takeoff-landing phase of aircraft operation.
There is also a concern that pollutants emitted during stratospheric cruise
may prove to be a serious environmental problem. Several NASA studies
(Ref. h) are being carried out to develop approaches to the solution of the
stratospheric cruise emissions problem. One such investigation is the Lean
Stability Augmentation Program that is being conducted at United Technologies
Research Center with sponsorship by NASA Lewis Research Center under Contract
NAS3-20801*.
The concept that is being explored in this program is the combustion of
lean, uniform gaseous fuel and air mixtures in a low temperature zone with
low residence time. The key element in this concept is bluff-body flame
stabilizer that is composed of a large number of circular passages and is
commonly called a perforated plate flameholder. Although it can produce
combustion products with low emissions, its natural operating range does
not satisfy the requirements of a gas turbine engine because the engine
equivalence ratio covers a range beyond the lean extinction limit of the
flameholder.
A large number of methods and variations for augmenting the lean sta-
bility of a perforated plate flameholder were considered in this program.
After study and evaluation of predicted performance, initial cost, and
maintenance cost, the list was reduced to three main methods. Among these
is the catalyst-coated flameholder. The other two involved addition of fuel
downstream of the flameholder and geometric variations of the flameholder
design.
PROGRAM GOALS
The goals of the program are defined by the emissions index, El (grams
per kilogram of fuel), which is specified as follows:
395
-------
EIjjQ 1.0 at design
A
3.0 over specified operating range
EICO 10 at design
EIUHC 1 at design
These values are consistent with the specified combustion efficiency which is
99% over a range of equivalence ratio of 0.3 to 0.6. The operating condi-
tions for the flameholder are specified in terms of inlet conditions as
shown below:
Pressure - 10 atm
Inlet Temperature - 600 to 800 K
Equivalence Ratio - 0.25 to 0.6
Reference (approach) Velocity - 25 m/sec
The program goals are accompanied by a number of constraints on the
methods and variations that are employed in the program in order to avoid
duplication of the work being done elsewhere. For instance, the use of a
catalytic combustor as a substitute for the flameholder is not permitted.
The use of a number of stages with catalytic combustors in some of the
stages is also not permitted. These constraints led to the use of catalytic
effects in a manner that augments the lean stability of a flameholder by
a reduction of its natural lean extinction limit.
APPROACH
The merit of using a catalytic surface to extend the lean extinction
limit of a flameholder was examined in relation to a model of the bluff body
flame stabilization process. The model (Ref. 5) is based on a heat balance
between heat generated at the flame front and the heat transferred to the
incoming mixture as shown in Fig. 1. It can be seen that sudden expansion
of the incoming gases forms a recirculation zone which consists primarily
of combustion products. Heat is transferred from this region by direct
mixing of recirculated products (M ) with the incoming mixture and by con-
vection to the wall of the bluff body. Burning is possible only when the
flow of heat out of the recirculation zone is limited to an acceptable
level. A reduction in the equivalence ratio results in a reduction in the
temperature of the recirculation zone. When the temperature falls below
a minimum level, the flame is extinguished.
An increase in the temperature of the incoming mixture near the wall
results in an increase in the temperature of the recirculation region. A
396
-------
higher temperature should produce a reduction in the lean extinction limit of
the burner- One method of heating the layer at the wall is the initiation of
local combustion reactions by catalyzing the surface. It has been estimated
from the data contained in Ref. 6 that approximately 5$ of the incoming mix-
ture enters the recirculation zone. Therefore, it is felt that the catalyzed
wall should be designed to produce a reaction of close to 5$ of the fuel in
the gas mixture.
DESIGN OF CATALYZED WALL
The reaction of fuel at the wall of the gas passage was estimated by
assuming that diffusion of fuel to the wall is the controlling mechanism.
The mass transfer rate can be expressed by an analogy to heat transfer in
a manner described in Ref. 7 as shown below:
WF/AS = hD (Coo - Cy)
The mass transfer coefficient, h-, is described in the Sherwood No., N ,
•L* on
as follows:
For fully developed turbulent flow, Sherwood Wo. can be related to the
Reynolds No. and Schmidt No. as shown below:
»SH . 0.033 (»Ee)0'83 (nsc)°-hk
where
w Gd
N_ = —.-.-
Re /*
and Nqp = JL
SC
For laminar flow in a tube, Sherwood No. is a constant with an approximate
value of k.
These equations were used after making two important simplifying as-
sumptions as follows :
1. Jet A fuel can be described as a pure hydrocarbon with a
molecular weight of 150.
2. Reaction of fuel at the wall is instantaneous.
For the case of the flameholder, the calculated value of reacted fuel flow
397
-------
is shown in Fig. 2. From these results, it can be seen that h% of the fuel
will be reacted after the mixture flows through a 2 in. long circular pas-
sage. This result is very conservative because fully developed flow only
occurs near the end of the passage. When corrected for entrance effects,
the calculated value of reacted fuel flow can be expected to reach close
to *>%.
To verify this analysis, the calculation was repeated using a monolithic
catalytic combustor based on a substrate having a cell density of 35 cells/
cm^ where the flow was assumed to be laminar. The results indicated that
approximately 50% of the flow will be reacted in a 2 in. long monolith. The
data of Ref. 8, which were obtained with propane-air mixtures, showed that
approximately hO% of the fuel was reacted in the first two inches of a
similar combustor operating at an equivalence ratio of 0.3 and a reference
velocity of 20 m/sec.
PLANNED EXPERIMENT
An experiment will be run to measure the effect of a catalyzed flame-
holder on the lean extinction limit. The flameholder consists of a group
of 22 discharge ports in a circular plate. The open area of the ports is
25% of the total area. Each port is fed from an individual tube as shown
in Fig. 3. The tube has an internal diameter of O.Uj in. and a catalyzed
length of 3-5 in. The tube material is AISI 321. The internal surface of
each tube and the downstream face of the plate have been catalyzed by Oxy-
Catalyst of West Chester, Pennsylvania. To apply the catalyst, the surface
was first grit-blasted for adherence of a coating of alumina. The alumina
r~)
was next applied at a loading of 5-10 mg/in. . A final application of
Pt-Rh catalyst was added to the alumina surface at a loading of 2-3$ of the
alumina loading.
The external surface of the tubes will be cooled by a crossflow of
cooling air- A second cooling stream will be used to control the tempera-
ture of the circular plate. Thermocouples will be used to monitor the wall
temperatures during the test.
The test of the flameholder will consist of the determination of the
lean extinction limit at the specified inlet temperature, pressure, and
velocity. Following this, the downstream face of the flameholder will be
scraped to remove the catalyst and the extinction limit will be again
determined. Finally, the internal surface of the forward two inches of the
tubes will be scraped to remove the catalyst and thereby to determine the
sensitivity of extinction limit behavior to catalyzed tube length.
398
-------
REFERENCES
1. Roberts, R. , A. Peduzzi, and G. E. Vitti: Experimental Clean Combustor
Program. Alternate Fuels Addendum. PWA-5370. NASA CR-13^970. July,
1976.
2. Gleason, C. C. and D. ¥. Bahr: Experimental Clean Combustor Program.
Alternate Fuels Addendum. GE R76 AEG 268. NASA CR-13^972. January,
1976.
3. Jones, R. E. : Emissions Reduction Technology Program. NASA Conference
Publication 2021. May, 1977.
k. Diehl, L. A., G. M. Reck, C. J. Marek, A. J. Szaniszlo: Stratospheric
Cruise Emission Reduction Program. NASA Conference Publication 2021.
May, 1977.
5. Cheng, S. I. and A. A. Kovitz: Theory of Flame Stabilization by a Bluff
Body. Seventh International Symposium on Combustion. Butterworth's
Scientific Publications, pp. 681-691. 1959.
6. Ozawa, R. I.: Survey of Basic Data on Flame Stabilization and Propa-
gation for High Speed Combustion Systems. The Marquardt Co. Technical
Report AFAPL-TR-70-81. January, 1971.
7. Rohsenow, W. M. and H. Choi: Heat, Mass and Momentum Transfer. Pren-
tice-Hall, Inc. 1961.
8. Anderson, D. N.: Preliminary Results from Screening Tests of Commercial
Catalysts with Potential Use in Gas Turbine Combustors. NASA TMX-
731*12. May, 1976.
399
-------
IGNITION POINT
o
o
FLAME FRONT
COMBUSTION PRODUCTS
RECIRCULATION
ZONE BOUNDARY
Figure 1. Flame stabilization process.
-------
LL
O
2
O
O
oc
ELEMENT LENGTH-CM
Figure 2. Catalyzed tube performance - fraction of
fuel contacting surface.
Figure 2. Catalyzed tube performance - fraction of fuel contacting surface.
401
-------
I FLOW DIRECTION
SECONDARY
COOLANT FLOW
COOLANT FLOW
CATALYZED
SURFACES
THERMOCOUPLES
FIGURE 3. CATALYZED FLAME HOLDER
402
78-08-84-3
-------
EFFECT OF INLET TEMPERATURE ON THE
PERFORMANCE OF A CATALYTIC REACTOR
by
David N. Anderson
NASA-Lewis Research Center
Cleveland, Ohio 44135
403
-------
EFFECT OF INLET TEMPERATURE ON THE PERFORMANCE
OF A CATALYTIC REACTOR
by David N. Anderson
National Aeronautics and Space Administration
Lewis Research Center
Cleveland, Ohio 44135,
ABSTRACT
A 12-cm diameter by 15-cm long catalytic reactor was tested with No. 2
diesel fuel in a combustion test rig at inlet temperatures of 700, 800, 900,
and 1000 K. Other test conditions included pressures of 3 and 6^10 Pa, ref-
erence velocities of 10, 15, and 20 m/s, and adiabatic combustion temperatures
in the range 1100 to 1400 K. The combustion efficiency was calculated from
measurements of carbon monoxide and unburned hydrocarbon emissions. Nitrogen
oxide emissions and reactor pressure drop were also measured. At a reference
velocity of 10 m/s, the CO and unburned hydrocarbons emissions, and, there-
fore, the combustion efficiency, were independent of inlet temperature. At an
inlet temperature of 1000 K, they were independent of reference velocity. Ni-
trogen oxides emissions resulted from conversion of the small amount (135 ppm)
of fuel-bound nitrogen in the fuel. Up to 90 percent conversion was observed
with no apparent effect of any of the test variables. For typical gas turbine
operating conditions, all three pollutants were below levels which would per-
mit the most stringent proposed automotive emissions standards to be met. The
pressure drop increased linearly with reference velocity and decreased
slightly as the inlet temperature was raised. Pressure drop increased lin-
early with velocity to a maximum value of 1.5 percent at a reference velocity
of 20 m/s.
INTRODUCTION
Probably the most important aspect of catalytic combustion is the produc-
405
-------
tion of extremely low concentrations of thermal NO . Because of this feature,
X
evaluations are being made of catalytic combustion for aircraft (refs. 1 to 3),
stationary (refs. 4 to 6), and automotive applications.
Catalytic combustion is being studied at the NASA-Lewis Research Center for
the Gas Turbine Highway Vehicle Systems Project which is supported by the De-
partment of Energy. The goal of this project is to demonstrate the technology
for an automotive gas turbine engine which would improve the thermal efficiency
over current spark-ignition piston engines. This improved gas turbine is pre-
sently being defined by each of the four major American automobile manufactur-
ers under development contracts to the DOE. The engine is expected to operate
with a turbine inlet temperature of about 1300 K and with combustor inlet tem-
peratures which will decrease from 1200 K at idle to 1000 K at full speed
(ref. 7).
Combustor work in support of this Project involves studies of fuel-air
premixing/prevaporizing systems (refs. 8 and 9), and the evaluation of cata-
lytic reactors (refs. 10 and 11). Monolithic catalytic reactors have been
tested at steady-state conditions which simulated those of the improved engine
combustor except that the maximum test section inlet temperature was only
800 K. Several studies (refs. 1, 10, and 12) have shown that an increase in
inlet temperature results in an improvement in catalyst performance; however,
none of those studies reported results for temperatures higher than 800 K.
The present study was made with test section inlet temperatures as high as
1000 K. While this temperature matches the improved gas turbine combustor in-
let at full speed, it is 200 K lower than the idle inlet temperature. To pro-
vide a basis for extrapolation of results to the idle condition, reactor per-
formance was determined at inlet temperatures of 700, 800, and 900 K as well.
Other test conditions included pressures of 3 and 6x10 Pa, reference veloci-
ties (catalytic reactor inlet velocities) of 10, 15, and 20 m/s, and a range
of adiabatic combustion temperatures of 1100 to 1400 K. The catalytic reactor
was 12 cm in diameter and 15 cm long. No. 2 diesel fuel was used for all
tests.
EXPERIMENTAL DETAILS
The test rig is described in figure l(a). All ducting was made from
15.2-cm (6-in. nominal) diameter stainless steel pipe with 12 cm ID by 15.2 cm
406
-------
OD Carborundum T30R Fiberfrax tube insulation inserted inside the pipe to give
a flow diameter of 12 cm.
Test section inlet air was indirectly preheated to temperatures in the
range of 700 to 1000 K. The inlet temperature was measured before fuel injec-
tion at a plane 40 cm upstream of the catalytic reactor. Twelve Chromel-
Alumel thermocouples were positioned in this plane at the centers of equal duct
cross-sectional areas. In previous studies at Lewis (refs. 7 and 11), the in-
let fuel-air mixture temperatures were measured nearer the reactor inlet plane
to avoid cooling errors associated with the poorly insulated pipes used in
those tests. The difference between the inlet air temperature and the inlet
fuel-air mixture temperature depends on the fuel-air ratio, but should be no
more than 15 to 20 K for the conditions of these tests.
A multiple conical tube fuel injector was located 15 cm downstream of the
inlet thermocouple plane and 25 cm upstream of the catalytic reactor. This
fuel injector was of the same type developed by Tacina (refs. 8 and 9); it is
pictured in figure l(b). Two separate sets of 21 equal-length fuel tubes in-
troduced fuel at the small-diameter (high-air-velocity), upstream end of the
conical airflow tubes. The large-diameter (1.6 mm) fuel tubes were for pro-
pane which was not used in this study, and the small-diameter (0.5 mm) were for
the No. 2 diesel fuel used in this study. Tests with this injector type have
shown that, 24 cm downstream of the fuel injector inlet, the fuel-air ratio and
the velocity varied by less than ±10 percent over the duct cross-section
(ref. 8). With an inlet air temperature of 700 K or greater, fuel vaporization
approached 100 percent at a distance of 17.8 cm (ref. 9). Therefore, at 25 cm
the fuel should be fully vaporized and mixed. The pressure drop reported in
reference 13 for this type of fuel injector was about 0.25 percent at a refer-
ence velocity of 10 m/s, increasing to 1 percent at 20 m/s.
The reactor inlet pressure was measured at a tap 13 cm upstream of the cat-
alytic reactor. At the same location a single Chromel-Alumel thermocouple was
used to detect burning upstream of the reactor. None was observed.
The catalytic reactor was the same J4 reactor used in two earlier studies
(refs. 11 and 14). It consisted of two 12-cm diameter and 7.5-cm long metal-
foil monolithic elements placed end to end and separated by a 0.31-cm diameter
Pt vs Pt-13 percent Rh thermocouple which measured the temperature at the duct
centerline. The elements are described in Table I; they were identical except
407
-------
that the first element used a Pt catalyst and the second Pd. Previous to the
start of testing this reactor had been operated for about 30 hours at adia-
batic combustion temperatures as high as 1500 K.
An array of 12 Pt vs Pt-13 percent Rh thermocouples were used to measure
the temperature at the reactor exit plane (see fig. l(a)). The reactor pres-
sure drop was measured with a differential pressure transducer which indicated
the difference between the inlet static pressure and the static pressure at a
tap 7 cm downstream of the reactor.
An array of 11 Pt vs Pt-13 percent Rh thermocouples measured the average
temperature at a plane 22 cm downstream of the reactor. At the same plane a
sample of the exhaust gas was obtained at the centerline of the duct with a
single-point probe. The probe was water-cooled to quench the sample near the
entrance port. The quenched gases flowed from the probe through an 18-m length
of 0.5-cm diameter electrically heated stainless-steel tubing to the gas ana-
lyzers. This sample line was maintained at 410 to 450 K to prevent the con-
densation of any unburned hydrocarbons in the sample. Concentrations of CO
and CO were determined with Beckman Model 315B nondispersive infrared analy-
zers, unburned hydrocabons with a Beckman Model 402 flame ionization detector,
and nitrogen oxides (NO and N0~) with a Thermo Electron Model 10A chemilumi-
nescent analyzer. Before analyzing for CO, C09, or NO , water vapor was re-
^ X
moved with a Hankinson Series E refrigeration-type dryer. Corrections were
made to the measured concentrations to obtain the actual, wet-basis concentra-
tions of these three constituents.
After discharging from the downstream instrumentation section, the exhaust
gas passed through a water spray to cool the combustion gases, then through a
back-pressure valve for control of rig pressure.
MEASUREMENTS AND COMPUTATIONS
For each setting of inlet airflow, temperature, and pressure, data were ob-
tained at several different fuel flowrates, first with fuel flow increasing,
then decreasing. This procedure permitted a check to be made of the repeata-
bility of the data, and to determine if hysteresis effects occurred. The re-
peatability was excellent and no hysteresis was evident.
The reference velocity was computed from the measured mass flowrate, the
average temperature measured at the test section inlet plane, the duct cross-
408
-------
sectional area, and the pressure measured at the test section inlet. Thus,
the reference velocity is the same as the reactor inlet velocity.
The emissions were measured as concentrations in ppm by volume and conver
ted to emission indexes using the expression
(E. I.)
x
where
(E.I.) emission index of specie; x, g /kg
X X
C concentration of specie x, ppm V
f fuel-air weight ratio, (kg/s)fuel/(kg/s)alr
M molecular weight of specie x, g /mole
X 'XX
Mp molecular weight of combustion products, 8products/moleproducts
The combustion efficiency was computed from the CO and HC emissions mea-
surements. The difference between the measured and equilibrium values of
these two emittants represents available chemical energy which has not been
released in the combustion process. Thus, the combustion efficiency in per-
cent is
EFF = 100 - 0.1 (E.I.
"~(HV)
co
-0.1
where
EFF combustion efficiency, percent
(HV) heating value of x, J/kg
X
The fuel-air ratio was determined both from the metered fuel and air flow-
rates and by making a carbon balance from the measured concentrations of CO,
CO-, and unburned hydrocarbons. The carbon-balance fuel-air ratio has the ad-
vantage that it is the local fuel-air ratio at which the emissions data are ob-
tained. The equilibrium concentrations of each important specie and the adia-
409
-------
batic combustion temperature were obtained using the carbon-balance fuel-air
ratio with the computer program of reference 15.
RESULTS AND DISCUSSION
Acceptable carbon-balance fuel-air ratios were between 90 and 112 percent
of the fuel-air ratio determined from fuel and airflow measurements. Several
data points (8 percent of all data taken) had values below this range and were
rejected; none were above. Virtually all of those data rejected were obtained
at extremely low fuel flow rates which were difficult to measure accurately.
Eighty two percent of the data retained had a carbon-balance fuel-air ratio
which was between 94 and 106 percent of the measured fuel-air ratio. Thus, the
carbon-balance fuel-air ratio was generally a valid representation of the
fuel-air ratio obtained from measured flows.
Combustion Efficiency
The combustion efficiency is shown in figure 2 as a function of the adia-
batic combustion temperature. Figure 2(a) gives the results at a reference
velocity of 10 m/s and for inlet temperatures of 700 to 1000 K with pressures
of 3 and 6x10 Pa. For each set of conditions the combustion efficiency in-
creased with adiabatic combustion temperature. At a pressure of 3x10 Pa the
efficiency was above 99 percent for all inlet temperatures when the adiabatic
combustion temperature was higher than 1200 K. At combustion temperatures be-
low 1200 K, the effect of raising the inlet temperature was to improve the
combustion efficiency.
The combustion efficiency decreased with an increase in pressure at inlet
temperatures of 800 and 900 K, but it increased with pressure at 1000 K. The
two different effects of pressure resulted from the relative roles of surface
and gas-phase reactions. The surface reation rate is limited by that of mass
diffusion, which is inversely proportional to pressure. Surface reactions
cannot provide complete combustion, and gas-phase reactions, which increase
with pressure, are necessary to achieve high combustion efficiency (refs. 4,
12, and 16). Thus, when the pressure is increased, combustion efficiency may
either increase or decrease depending on whether gas-phase reactions become im-
portant near the reactor inlet or not until much later.
At reference velocities of 15 and 20 m/s, data were taken only with a pres-
sure of 3x10 Pa (see figs. 2(b) and (c)). The effect of inlet temperature on
410
-------
combustion efficiency was much greater than at 10 m/s and 3*10 Pa. As the
inlet temperature was increased, the combustion efficiency also increased.
Reference 11 reported combustion efficiencies at an inlet temperature of
800 K and a pressure of 3x10 Pa for the same reactor tested in this study.
Higher combustion temperatures were required in the experiments of refer-
ence 11 to achieve the same efficiencies as this study. The primary reason
that better performance was obtained in the present study was the elimination
of most of the test section heat loss through the use of better insulation.
Heat loss was calculated to be less than 2 percent, and measured exit thermo-
couples agreed with the adiabatic combustion temperatures within a few degrees
for the present experiments. Exhaust-gas sampling techniques were different
for the two experiments, as well. The single-point probe of this study was
located 8 cm farther downstream than the multi-point probe of reference 11.
Both of these differences would explain why higher combustion efficiencies
were obtained for the present study.
Emissions
Emissions goals for the Gas Turbine Highway Vehicle Systems Project have
been established (ref. 17) as 13.6 g CO/kg fuel, 1.64 g HC/kg fuel, and 1.60 g
NO /kg fuel. These goals are used as reference values to help evaluate the
X
measured emissions.
Carbon Monoxide
The carbon monoxide emission index is plotted as a function of the adia-
batic combustion temperature in figures 3(a) to (c). Figure 3(a) gives the
emissions at a reference velocity of 10 m/s for pressures of 3 and 6x10^ Pa.
Emissions decreased as the adiabatic combustion temperature increased. Values
below the reference level were achieved at a pressure of 3x10 Pa when the
adiabatic combustion temperature was higher than 1220 K for all four inlet
temperatures.
As with the combustion efficiency results, the nature of the pressure ef-
fect depended on the inlet temperature. Emissions at an inlet temperature of
1000 K decreased when the pressure was doubled from 3 to 6x10 Pa. In con-
trast, at an inlet temperature of 800 or 900 K, doubling the pressure produced
an increase in the CO emission index for adiabatic combustion temperatures be-
411
-------
low 1250 K, but a decrease in emissions for higher combustion temperatures.
With a reference velocity of 15 m/s (fig. 3(b)) the carbon monoxide emis-
sions decreased with increases in either inlet temperature or adiabatic com-
bustion temperature. The same trends are shown in figure 3(c) for a refer-
ence velocity of 20 m/s.
If figures 3 (a) to (c) are compared at an inlet temperature of 1000 K it
can be seen that there is little difference in the emissions produced at 10,
15, or 20 m/s reference velocities. Overall reaction rates are high enough at
this inlet temperature that an increase in residence time (decrease in refer-
ence velocity) provided little benefit. Similarly, it was seen in figure 3(a)
that at a reference velocity of 10 m/s there was enough residence time that an
increase in inlet temperature (i.e., initial reaction rate) had little effect
on the CO emissions produced.
Unburned Hydrocarbons
The emission index of unburned hydrocarbons is plotted as a function of
the adiabatic combustion temperature in figure 4. Figure 4 (a) gives the re-
sults at a reference velocity of 10 m/s and shows no significant effect of
either inlet temperature or pressure. The temperature required to meet the
hydrocarbon emission index goal was 40 K lower than that required to meet the
CO emission index goal.
An effect of inlet temperature on hydrocarbon emissions was clearly seen
when the reference velocity was increased to 15 m/s (fig. 4(b)). The combus-
tion temperatures required to meet the hydrocarbon emissions goal were 30 to
80 K less than those required to meet the CO goal at this velocity.
Similarly, at 20 m/s (fig. 4(c)) emissions were higher with the lower in-
let temperatures. The hydrocarbon emissions goal was achieved at combustion
temperatures which were 10 to 95 K lower than those required to meet the CO
goal.
As with the CO emissions, the reference velocity had little effect on the
hydrocarbons emissions at an inlet temperature of 1000 K, while inlet temper-
ature had little effect at a reference velocity of 10 m/s.
Nitrogen Oxides
The third pollutant of interest in automotive combustion studies is nitro-
gen oxides. Because of the low temperatures at which catalytic combustion
412
-------
takes place, thermal NO emissions are on the order of 1 ppm or less. How-
X
ever, conversion of nitrogen in the fuel can produce significantly higher
levels than this.
The fuel used in this study contained 135 ppm of N. Eightly percent of
the NO emissions data ranged in value from 0.2 to 0.4 g N0»/kg fuel which
X "
represents conversion of from 45 to 90 percent of the fuel N. It should be
noted, however, that there is a great deal of uncertainty associated with the
measurement of such low levels of NO . No effect of inlet temperature, ref-
1 X
erence velocity, pressure, or adiabatic combustion temperature was apparent
within the scatter of the data. The emissions were well below the reference
emission index of 1.60 g NO /kg fuel for all test conditions.
Minimum Operating Temperature
The minimum operating temperature is the adiabatic combustion temperature
above which the reactor must be operated to insure that all three pollutants
meet the project goals. In this study, NO emissions were below the reference
value at all test conditions, and the hydrocarbons goal was met at lower com-
bustion temperatures than the CO goal. Thus, the minimum operating tempera-
ture was identical with the minimum adiabatic combustion temperature required
to meet the CO goal.
Figure 5 gives the minimum operating temperature as a function of the in-
let temperature. As noted in the discussion of the CO emissions, at a refer-
ence velocity of 10 m/s and a pressure of 3x10 Pa, excess residence time was
available so that an increase in inlet temperature (and, thus, initial reac-
tion rate) had no effect on the minimum operating temperature. A shorter re-
actor would probably perform equally well at these conditions.
When the pressure was increased to 6><10 Pa, the adiabatic combustion
temperature required to meet the emissions goals increased at 800 and 900 K
inlet temperatures but decreased at 1000 K. This result was due to changes
in the relative roles of surface and gas-phase reactions as discussed under
CO emissions.
At reference velocities of 15 and 20 m/s an increase in inlet temperature
produced a decrease in the minimum operating temperature. For inlet tempera-
tures above 1000 K the extrapolated results in figure 5 show that the minimum
operating temperature approaches 1220 for all three velocities independent of
413
-------
inlet temperature. Thus, at an inlet temperature of 1220 K, the minimum op-
erating temperature and the inlet temperature will be equal.
Reactor Pressure Drop
In addition to operating with very low emissions, an automotive gas tur-
bine combustor must have minimal pressure losses to permit high cycle effi-
ciencies. In general, the pressure drop for a catalytic reactor includes both
friction losses and entry/exit losses. For the reactor and the test condi-
tions of this study, however, friction losses dominated. The Reynolds number
was less than 1000 for all flow conditions; thus, the flow through the reactor
passages was laminar. For laminar flow, the friction drop over a small ele-
ment of length, d£, is
where
k a constant
V local velocity
T local temperature
p pressure
The local velocity is equal to the reference velocity V f, times the ratio of
the local to inlet temperature, T/T. . Then the pressure drop over the reac-
tor length, £, is
P Tin P
The pressure drop as a percent of upstream pressure is presented in fig-
ure 6 as a function of V /p. For an inlet temperature of 700 K the data
show a direct proportionality between Ap/p and V f/p. This result sug-
gests that the temperature history, T = f (£), does not change appreciably with
reference velocity. At higher inlet temperatures, some decrease in pressure
drop was observed as would be expected for laminar flow friction loss. For
example, at a reference velocity of 10 m/s and a pressure of 3*105 Pa
414
-------
(V /p = 3.3x10 5 Yl» the Pressure dr°P decreased from 0.8; to 0.66 percent
when the inlet temperature was raised from 700 to 1000 K.
The pressure drop data in figure 6 at an inlet temperature of 800 K and a
pressure of 3*10 Pa were about half the values reported in references 11
and 14 for the same reactor. In those previous studies, however, the thermo-
couple instrumentation within the reactor housing was more dense than that
used for the present study. In addition, the earlier data included the pres-
sure drop across the downstream instrumentation section. For the very low
pressure drops at which this reactor operates, these additional instrumenta-
tion pressure drops were significant.
CONCLUSIONS
This study of the effect of inlet temperature on the performance of a cat-
alytic reactor has shown that if sufficient residence time was provided, as
occurred at a reference velocity of 10 m/s, inlet temperatures in the range
700 to 1000 K had little effect on combustion efficiency or emissions. At
700 K inlet temperature, significantly better performance was achieved at
10 m/s than at higher reference velocities. Thus, for applications with low
inlet temperatures, such as the aircraft or stationary gas turbines, long res-
idence times (low reference velocities) must be provided to achieve high com-
bustion efficiencies.
In contrast, the reference velocity had little effect on the combustion
efficiency or emissions when the inlet temperature was 1000 K. This result
suggests that for combustor applications with high inlet temperature, such as
the regenerative automotive gas turbine, the upper limit on reference velocity
may be determined by permissible pressure drop rather than combustion effi-
ciency .
The pressure drop in this study resulted primarily from friction loss
through the reactor passages. Because the flow was laminar, the percent pres-
sure drop decreased slightly with increasing inlet temperature at a constant
adiabatic combustion temperature.
The improved catalyst performance which results from operation at high in-
inlet temperatures makes the automotive gas turbine an especially attractive
application for catalytic combusion. In this study, at 1000 K inlet tempera-
ture, 1300 K exit temperature, and 20 m/s reference velocity, emissions of CO,
415
-------
unburned hydrocarbons, and NO were below levels which would permit the -
A
stringent automotive standards to be achieved, and pressure drop was only
1.5 percent.
REFERENCES
1. Blazowski, W. S., and D. E. Walsh. Catalytic Combustion: An Important
Consideration for Future Applications. Combust. Sci. Technol., 10(5/6)
233-244, 1975.
2. Rosfjord, T. J. Catalytic Combustors for Gas Turbine Engines. AIAA Paper
76-46, 1976.
3. Siminski, V. J., and H. Shaw. Development of a Catalytic Combustor for
Aircraft Gas Turbine Engines. EXXON/G.R.U. 1 BFA.76, Exxon Research and
Engineering Co., Linden, N. J., 1976. (AFAPL-TR-76-80, AD-A040135.)
4. DeCorso, S. M., S. Mumford, R. Carrubba, and R. Heck. Catalysis for Gas
Turbine Combustors - Experimental Test Results. ASME Paper 76-GT-4, 1976.
5. Kesselring, J. P., R. A. Brown, R. J. Schreiber, and C. B. Mover. Cata-
lytic Oxidation of Fuels for NO Control from Area Sources. EPA-600/2-76-
-X.
037, Environmental Protection Agency, 1976.
6. Kesselring, J. P., W. V. Krill, and R. M. Kendall. Design Criteria for
Stationary Source Catalytic Combustors. Paper presented at the 1977 Fall
Meeting, Western States Section, Combustion Institute, WSS/CI Paper 77-32.
7. Anderson, D. N., R. R. Tacina, and T. S. Mroz. Catalytic Combustion for
the Automotive Gas Turbine Engine. NASA TM X-73589, 1977.
8. Tacina, R. R. Experimental Evaluation of Premixing-Pre-vaporizing Fuel In-
jection Concepts for a Gas Turbine Catalytic Combustor. NASA TM-73755,
1977.
9. Tacina, R. R. Degree of Vaporization Using an Air-Blast Type Injector for
a Premixed-Prevaporized Combustor. NASA TM-78836, 1978.
10. Anderson, D. N., R. R. Tacina, and T. S. Mroz. Performance of a Catalytic
Reactor at Simulated Gas Turbine Combustor Operating Conditions. NASA TM
X-71747, 1975.
11. Anderson, D. N. Performance and Emissions of, a Catalytic Reactor with
Propane, Diesel, and Jet A Fuels. NASA TM-73786, 1977.
416
-------
12. Pfefferle, W. C., R. V. Carrubba, R. M. Heck, and G. W. Roberts. Cata-
thermal^ Combustion: A New Process for Low-Emissions Fuel Conversion.
ASME Paper 75-WA/Fu-l, 1975.
13. Tacina, R. R. Experimental Evaluation of Fuel Preparation Systems for an
Automotive Gas Turbine Catalytic Combustor. NASA TM-78856, 1977.
14. Anderson, D. N. Effect of Catalytic Reactor Length and Cell Density on
Performance. Paper presented at the Second Workshop on Catalytic Combus-
tion, Raleigh, N. C., June 21-22, 1977- Sponsored by Environmental Pro-
tection Agency.
15. Gordon, S., and B. J. McBride. Computer Program for Calculation of Com-
plex Chemical Equilibrium Compositions, Rocket Performance, Incident and
Reflected Shocks, and Chapman-Jouguet Detonations. NASA SP-273, 1971.
16. Wampler, F. B., D. W. Clark, and F. A. Gaines. Catalytic Combustion of
C-jHg on Pt Coated Monolith. Combust. Sci. Technol. , 14,25-31, 1976.
17. Aircraft Engines Emissions, NASA CP 2021, 1977.
417
-------
TABLE I. - DESCRIPTION OF CATALYST ELEMENTS
Element number
Element designation
Position in reactor
Manufacturer
Catalyst
•D
Loading, kg/m
Substrate
Cell density, cells/cm^
Element diameter, cm
Element length, cm
1
JM1
Upstream
Johnson Matthey, Ltd.
Pt
5.3
Metal foil, corrugated
and wound into a
cylinder
62
12
7.6
2
JM2
Downstream
Johnson Matthey, Ltd.
Pd
5.3
Metal foil, corrugated
and wound into a
cylinder
62
12
7.6
418
-------
r INLET PLANE:
\12THERMOCOUPLES.
\1 PRESSURE TAP
\
rPRESSURE TAP AND
\ THERMOCOUPLE
\
i-REACTOR
/ THERMO-
r-MULTIPLE-
\ CONICAL-TUBE
\FUEL INJECTOR
\
r INSULATION
r SAMPLING STATION:
' 1 SAMPLE PROBE.
11 THERMOCOUPLES
*-EXIT PLANE:
12 THERMOCOUPLES;
(a) TEST SECTION (DIMENSIONS IN cm).
Figure 1. - Experimental apparatus.
-------
NJ
o
C-78-1793
(b) MULTIPLE CONICAL TUBE FUEL INJECTOR.
Figure 1. - Concluded.
-------
100 r—
(a) REFERENCE VELOCITY. 10 m/s.
PRESSURE.
3 6
O
D •
O '*
A A
INLET
TEMP.,
K
700
800
900
1000
I
I
1300
1400
1300 1400 1100 1200
ADIABATIC COMBUSTION TEMPERATURE. K
(b) REFERENCE VELOCITY, 15 m/S; PRESSURE, 3x10* Pa. (c) REFERENCE VELOCITY, 20 m/s: PRESSURE, SxlO5 Pa.
Figure 2. - Combustion efficiency.
421
-------
600
100
50
10
5
.4
PRESSURE, INLET
pa TEMP .
700
800
900
1000
3
O
D
O
A
REFERENCE EMISSION
INDEX, 13.6g/kg
(a) REFERENCE VELOCITY, lOm/s.
8
1100
1200
1300
1300 1400 1100 1200
ANABATIC COMBUSTION TEMPERATURE, K
(b) REFERENCE VELOCITY, 15 m/s: PRESSURE, SxlO5 Pa. Id REFERENCE VELOCITY, 20 m/S; PRESSURE, 3X105 Pa.
Figure 3. - Carbon monoxide emissions.
1400
422
-------
10 i—
I
5.
PRESSURE, INLET
105 Pa TEMP.,
3 6
O
D •
O *
A A
700
800
900
1000
REFERENCE EMISSION
INDEX, 1.64g/kg
(a) REFERENCE VELOCITY, 10 m/s.
1200
1300
1300 1400 1100 1200
ADIABATIC COMBUSTION TEMPERATURE, K
(b) REFERENCE VELOCITY, 15 m/S; PRESSURE, SxlO5 Pa. (c) REFERENCE VELOCITY, 20 m/s; pressure, SxlO5 Pa.
Figure 4. - Unburned hydrocarbons emissions.
1400
423
-------
-p-
Is3
o* 1*»<
< UJ"
o ^ f^^
«£g 1300
MINIMUM REQU
COMBUSTION T
i— • >-*
o S
~"
PRESSURE, REFERENCE
105 Pa VELOCITY,
3 6 m/s
V -o- -^ 10
~ \ a 15
^\^ Nx^. ° 20
* fc"^~I*-~"^_v
~~ %>»
1 1 1 1 1 1 1 1 1 1
DO 800 900 1000 1100 1200
INLET TEMPERATURE, K
Figure 5. - Effect of inlet temperature on minimum
operating temperature required for low emissions.
-------
=
PRESSURE. INLET
Ifl'Pa TEMP.,
K
0 1234567
REFERENCE VELOCITY/PRESSURE. 10"5(m/s)/Pa
Figure 6. - Reactor pressure drop at an adiabatic com-
bustion temperature of 1300 K.
425
-------
PERFORMANCE OF MULTIPLE-VENTURI
FUEL PREPARATION SYSTEM
by
Robert R. Tacina
NASA-Lewis Research Center
Cleveland, Ohio 44135
427
-------
PERFORMANCE OF MULTIPLE-VENTURI
FUEL PREPARATION SYSTEM
by
Robert R. Tacina
NASA Lewis Research Center
Cleveland, Ohio 44135
ABSTRACT
An airblast-type fuel-injection system was evaluated for use
with a catalytic reactor in a gas turbine application. Multiple,
venturi-type tubes were used to provide high-velocity air for atom-
ization and would also straighten the inlet airflow. Spatial fuel-
air distributions, degree of vaporization, and pressure drop were meas
ured 16.5 cm downstream of the fuel injection plane. Tests were per-
formed in a 12 cm tubular duct. Test conditions were: a pressure
of 0.3 MPa, inlet air temperatures from 400K to 800K, air velocities
of 10 and 20 m/s, and fuel-air ratios of 0.010 and 0.020. The fuel
was diesel #2. Spatial fuel-air distributions were within ± 20 per-
cent of the mean at inlet air temperatures above 450K. At an inlet
air temperature of 400K, the fuel-air distribution was within ^ 30
percent of the mean. No distortion in the fuel-air distribution was
measured when a 50 percent blockage plate was placed 9.2 cm upstream
of the fuel injection plane to distort the inlet air velocity pro-
fiel. Vaporization of the fuel was 50 percent complete at an inlet
air temperature of 400K and the percentage increased linearly with
temperature to complete vaporization at 600K.
429
-------
CORRELATIONS OF CATALYTIC COMBUSTION
PERFORMANCE PARAMETERS
Daniel L. Bulzan
NASA-Lewis Research Center
Cleveland, Ohio
431
-------
ABSTRACT
Performance parameters were correlated for a catalytic combustor.
Correlations for combustion efficiency, percentage pressure drop and the
minimum required adiabatic reaction temperature necessary to meet emis-
sions goals of 13.6g CO/kg fuel and 1.64g HC/kg fuel are presented.
Combustion efficiency was found to be a function of the cell density,
cell circumference, reactor length, reference velocity and adiabatic
reaction temperature. The percentage pressure drop at an adiabatic
reaction temperature of 1450K was found to be proportional to the refer-
ence velocity to the 1.5 power. The percentage pressure drop was also
found to be proportional to the reactor length and inversely proportional
to the pressure, cell hydraulic diameter and fractional open area. The
minimum required adiabatic reaction temperature needed to meet the emis-
sions goals was found to increase with reference velocity and decrease
with cell circumference, cell density and reactor length. A catalyst
factor was introduced into the correlations to account for differences
between catalysts. Combustion efficiency, the percentage pressure drop,
and the minimum required adiabatic reaction temperature were found to be
a function of the catalyst factor. The catalyst factor ranged from .12
to 1,52. The data was from a 12 cm. diameter test rig with noble metal
reactors using propane fuel at an inlet temperature of 800K, a pressure
of 3 x 10 Pa, and reference velocities from 10 to 20 m/s.
433
-------
CATALYST DESIGN STUDIES IN
LOW BTU GAS COMBUSTION
by
R. Carrubba
I. T. Osgerby
Engelhard Industries Division
Edison, New Jersey
This paper was not received in time for
publication. Contact the authors for
details.
435
-------
CATALYZED COMBUSTION OF H2/AIR MIXTURES
IN A FLAT PLATE BOUNDARY LAYER
By:
R.W. Schefer
F.S. Robben
R.K. Cheng
Lawrence Berkeley Laboratory
University of California
Berkeley, California 94720
*This work was supported by DOE, Division of Conservation
Research and Technology
437
-------
ABSTRACT
A study has been made of the combustion characteristics of lean hydrogen-
air mixtures flowing over a heated catalytic platinum plate. The objectives
of the investigation are to develop a better understanding of the interaction
between fluid mechanics, gas phase combustion, and the surface catalytic re-
actions in high temperature surface catalyzed combustion.
The experimental system consists of a thin quartz plate with vacuum de-
posited platinum heating strips mounted over an open atmospheric pressure jet
of premixed hydrogen and air. Boundary layer density profiles were measured
using differential interferometry for flow visualization studies and Rayleigh
scattering for point density measurements. The presence of heat release due
to surface reaction was determined from changes in heating strip power inputs
with fuel addition.
Results are presented for a range of equivalence ratios from 0.05 to 0.30
and plate surface temperatures from 470K to 1300K. Significant surface heat
release was found for all mixtures at plate temperatures as low as 470K. Only
at increased equivalence ratios and plate temperatures was heat release due
to gas phase reaction present. This was characterized by local increases in
temperature across the boundary layer and an increase in thermal boundary lay-
er thickness.
The effect of various surface boundary conditions was studied using a
computer program to solve the boundary layer equations for flow over a heated
plate with H2/air combustion. Surface conditions investigated include: 1)
a noncatalytic surface; 2) finite rate surface oxidation of H2; and 3)
radical recombination at the plate surface in addition to surface oxidation
of H2. Surface reaction is found to have a strong quenching effect on the
initiation of gas phase combustion in the boundary layer due to the removal
of H_ near the plate surface and quenching of radical species generated in
the gas phase.
439
-------
SECTION 1
INTRODUCTION
Interest in the use of catalytic surfaces to promote combustion reactions
has increased greatly over the past several years. A number of studies have
been undertaken which indicate that catalytically supported combustion makes
efficient burning of a variety of fuels possible under fuel lean conditions
with a substantial reduction in NO levels (1,2). While initial emphasis was
X
based on the use of catalytic combustors for aircraft gas turbine application
its potential application to stationary power sources has also been studied
with favorable results (3).
Much of the work to date has involved parametric investigations of pro-
totype catalytic combustor configurations. The results of such studies are
both necessary to the long term development of catalytic combustors and have
been very promising. However, such studies are somewhat limited in terms of
deriving a more fundamental understanding of catalytically supported combus-
tion. Two such areas where a greater knowledge would be desirable are the
role of internal heat and mass transfer in the catalytic combustion process,
and the role of homogeneous as opposed to catalytic surface reactions. An
understanding of these and related processes is necessary for the optimum
evaluation and application of the catalytic concept to practical combustion
system design.
The approach of the present investigation has been to examine catalytical-
ly supported combustion in a well characterized system in which most of the
important physical and chemical processes found in larger scale catalytic
combustors are present. The geometry chosen consists of combustion in the
boundary layer of heated catalytic flat plate. This system provides a suit-
able geometry for both experimental and numerical modeling studies. The oper-
ation of such a system under a suitably selected range of conditions greatly
441
-------
facilitates determination of the roles of the various processes involved.
The experimental study has concentrated on the use of two optical diagnos-
tic techniques, Rayleigh scattering and differential interferometry, for boun-
dary layer density measurements. Optical diagnostic techniques were used be-
cause they are non-perturbing to the flow and provide good spatial resolution.
The advantage of a flow visualization method such as interferometry is that a
relatively rapid study of boundary layer behavior under combustion conditions
is possible. Rayleigh scattering was then used for a more detailed investi-
gation of the boundary layer structure under those conditions of greatest
interest.
In Section 2 of this paper the experimental facility and diagnostics are
described, and results are presented for the flow of chemically reacting H-/
air mixtures over a heated platinum catalytic surface. The numerical model
and procedure are described in Section 3. Results are presented for the com-
bustion of H_/air at an equivalence ratio of 0.1 over a heated noncatalytic
plate with a surface temperature of 1100 K. The model is then extended to
include finite rate surface reaction. Two models are formulated to account
for catalytic reactions at the plate surface. These are based on: 1) a
boundary condition which includes finite rate surface oxidation of H« (case 1);
and 2) a condition which includes radical recombination at the surface in
addition to surface oxidation of H- (case 2). Comparisons are made with the
experimental results.
442
-------
SECTION 2
EXPERIMENTAL INVESTIGATION
APPARATUS
Flow System and Plate Design
The experimental configuration consists of a thin quartz plate with
vacuum deposited platinum heating strips suspended vertically over an open,
atmospheric pressure jet of premixed hydrogen and air. The jet is produced
by a stagnation chamber with a converging nozzle at the exit. The stagnation
chamber is 20 mm in diameter and is fitted with a 5 cm diameter nozzle at the
exit. It has three internal screens and produces a uniform, low turbulence
flow of premixed combustible mixture which is directed vertically upwards.
Lower flow rates of gases are metered by calibrated rotometers, and larger air
• -• • -»
flow rates are metered using a standard orifice gauge and water manometer.
The house air supply is passed through a dryer and filter combination which
eliminates particulate matter. The flat plate and stagnation chamber are
mounted on a three-dimensional traverse mechanism (adapted from a milling
machine) with a 0.001 cm positioning sensitivity.
Details of the flat plate and holder design are shown in Fig. 1. The
plate is made from a 1.5 mm thick, 75 mm squate quartz sheet with a sharp
leading edge angle of 2°. Five platinum heating strips are vacuum deposited
on the plate surface. The central region of the plate is coated to a thick-
ness of 0.3 microns, while the outer edges are coated to approximately three
times this thickness. This keeps the outer edges of the plate cool so that
gold foil can be used to make electrical contact with the stainless steel
fingers. The strips are oriented perpendicular to the flow and are of vary-
ing widths to improve temperature control near the plate leading edge. Elec-
trical power is individually controlled to each of these strips and as a result
a reasonably uniform plate surface temperature can be maintained.
443
-------
To reduce the interference of surface scattering of the laser beam with
the Rayleigh scattering in the boundary layer, important when measurements are
being made close to the plate, it has been found advantageous to have a small
bend, about 2°, in the center of the plate. The Rayleigh scattering measure-
ments are made in the shadow of this bend.
Surface temperatures greater than 1000°K are measured using a disappear-
ing filament type optical pyrometer with an emissivity estimate of 0.65 for
the platinum surface. For temperatures below 1000°K, the surface temperature
is determined from the measured resistivity of the platinum strips.
Optical Diagnostics
Rayleigh Scattering System
Rayleigh scattering refers to the light scattered from gas molecules with-
out any shift in wavelength (4). The intensity is proportional to the gas
density and gives approximately the same information as an interferometric
measurement of gas density. The principal advantage over interferometric
techniques (5) is that it measures the density at a single point rather than
integrating along the optical path. The principal disadvantage, compared to
an interferometric photograph, is that the density is measured at only one
point at a time.
The optical arrangement for the Rayleigh scattering system is shown in
Fig. 2. A Spectra Physics argon-ion laser operating at 488 nm with 1 watt
power is used in the system. The Rayleigh scattering is measured at right
angles to the focused laser beam by a monochrometer and photomultiplier.
For measurements close to the plate surface scattering appears as a back-
ground signal and thus limits how close measurements can be made. In order
to minimize the surface scattering, spatial filters are used -to "clean up"
the laser beam. A waist diameter of 50 microns was chosen to give good spatial
resolution in the boundary layer.
An f/1.2, 55 mm focal length camera lens is used to collect the scattered
light. An image of the monochrometer slit, about 100 microns wide and 1 mm
long, is then focused on the waist region of the laser beam. The optics and
monochrometer (a 0.3 m McPherson) are arranged such that all the rays from the
lens fall onto the photomultiplier, an RCA type 1P28. The quantum efficiency
444
-------
of this type of photomultiplier is about 15% at 488 nm. The output current
of the photomultiplier is measured using a Keithley electrometer.
As noted previously the slight bend (about 2°) down the middle of the
plate effectively shadows the second half of the surface of the plate from
the laser beam. This results in a substantial reduction in plate scattered
light and makes Rayleigh measurements possible as close as 0.2 mm from the sur-
face.
Differential Interferometer
. »
To study changes in the flow pattern as a function of variations in
operating parameters, differential interferometry was used to visualize the
flow field. This technique is based on the interference of two beams passing
through adjacent points in the test space. Creeden (6) has shown that in this
situation, the fringe displacement is proportional to the deflection, 0, rather
than to the index of refraction as in the case of a conventional interferometer.
0 is defined as:
.- r dn ,
0 = / -j— dz
dy
where n is the index of refraction, z and y are the coordinates along and nor-
mal to the optical path respectively. Since the index of refraction is linear-
ly proportional to the density, the fringe displacement is proportional to the
density gradient in the test space provided that all other index of refraction
fields along the optical path are negligible.
An example of an interferogram is shown in Fig. 3 for a plate surface
temperature of 1200 K and an equivalence ratio of 0.3. The distribution of
the gradient of the refractive index in the boundary layer is shown by the
displacement of the fringes from the undisturbed position. Details of the
optical arrangement and analysis procedure can be found in reference (7).
To obtain density or temperature profiles across the boundary layer, the
fringe displacement is measured directly from the interferograms. Numerical
integration of the area under the fringe displacement curves then gives the
difference between the density of the gases in the free stream (p^), and the
local gas density. The corresponding temperature profiles are obtained using
the ideal gas law.
445
-------
RESULTS
Differential Interferometer Study
Typical temperature profiles obtained from an analysis of the interfero-
grams are shown in Fig. 4 for a plate temperature of 1270 K and various equiv-
alence ratios. The profiles for equivalence ratios of 0.0 and 0.1 are typical
of those with no heat release due to gas phase reaction. Under these conditions
the profiles are self-similar. For.mixtures with higher equivalence ratios
the temperature profiles are significantly different. In this case the heat
release rates due to gas phase combustion are considerably greater than at
lower equivalence ratios. As a result the local gas temperature increases
and the thermal boundary layer becomes thicker.
It should be noted that the temperature profiles obtained from the inter-
ferograms are only approximations to the actual profiles measured with Rayleigh
scattering and predicted by the numerical calculations. This can be attributed
to the various assumptions regarding the refractive surfaces which were neces-
sary for the analysis (7). As shown in Fig. 4 the interferograms do, however,
provide the information needed to distinguish between various stages of gas
phase reaction. Further analysis of the data over a range of conditions shows
that the existence of gas phase combustion in the boundary layer depends on
mixture composition, surface temperature, and distance from the: leading edge.
Significant surface reaction was also found to occur under certain con-
ditions. This was apparent as an increase in plate surface temperature due
to surface energy release as fuel is added to the flow. By decreasing the
power input to the individual heating strips when surface reaction was present
the plate temperature could be maintained at the desired value.
The conditions under which the various phenomena occur are summarized in
Fig. 5. These results are based on an analysis of data taken at a distance of
50 mm downstream from the plate leading edge. At distances greater than this
fluctuations in the jet mixing layer become a significant source of disturbance
to the boundary layer. Surface reaction was found to occur at temperatures
as low as 470 K, the minimum surface temperature at which measurements were
taken. At increased temperatures and equivalence ratios a region exists over
which both surface reaction and boundary layer combustion occur simultaneously.
At the highest temperatures and equivalence ratios the primary heat release, or
446
-------
reaction, zone moves out of the boundary layer and takes on a flame-like struc-
ture characterized by a steep temperature gradient and an approximately constant
temperature region extending to the plate surface.
Rayleigh Scattering Study
Based on the results of the differential interferometer study a set of
conditions was chosen for Rayleigh scattering measurements in which various
stages of gas phase reaction were present. The set of conditions at whicl*
Rayleigh scattering measurements were taken is indicated in Fig. 5. These
measurements, in addition to providing a better understanding of boundary
layer behavior under combustion conditions, also provide a set of data upon
which the development of the numerical model can be based.
Density profiles at a distance of 30 mm downstream from the plate leading
edge are shown in Fig. 6 for a surface temperature of 1170 K and various
equivalence ratios. Also shown as a solid line is the corresponding density
profile obtained from the numerical calculations for the case of no combustion.
At equivalence ratios less than 0.15 good agreement exists between the experi-
mental and predicted profiles, indicating that no significant heat release due
to gas phase reaction occurs under these conditions. At increased equivalence
ratios the thermal boundary layer becomes considerably thicker due to gas phase
heat release and a region of approximately constant density exists across much
of the boundary layer.
The effect of plate surface temperature on thermal boundary layer thick-
ness is shown in Fig. 7 for an equivalence ratio of 0.20. The thermal boundary
layer thickness, ST, is defined here as the point at which p/p^ =0.5. The
corresponding results from numerical calculations for the case of no combustion
are indicated by solid lines. At a plate temperature of 1070 K the agreement
between experimental and predicted profiles is excellent. From this it is
concluded that no gas phase combustion is present under these conditions. At
temperatures of 1170 K and 1270 K a noticeable increase in & occurs between
15 mm and 20 mm from the leading 'edge. Farther downstream the thermal boundary
layer growth rates again become comparable for the two plate temperatures.
447
-------
SECTION 3
NUMBERICAL INVESTIGATION
NUMERICAL MODEL
Governing Equations
The governing differential equations for laminar boundary layer flow of a
chemically reacting mixture of gases over a heated flat plate are the conserva-
tion of mass, the conservation of momentum, the conservation of energy, and
the species conservation equations. Introducing the similarity variables
£00 = (t>V)J3JS.
Uoo
n(x,y) = °° fj pdy
1/25
these conservation equations can be written as follows:
Continuity equation
Momentum equation
"e
Energy equation
448
-------
Species equation
>. 32a.
In the above equations the following dimensionless quantities have been
introduced
' - (py)^ - 9x ^ ' ' ^ Uoo
- £CP
c = : b = : e = 1—
Pr i Pr ,0. v
r,- 9b
PU —, 1 9c . , 1 i
£ = ^- ; c = ; b! =
ne represents the value of n at the outer edge of the boundary layer.
The chemical kinetic source term S in the species equation is defined
ai
as
JJ
s = - y
0\ -i-1
where R and R are the forward and reverse rates of reaction j
3 -J
respectively, and a. . is the stoiciometric coefficient for the i species
in the j reaction as defined in the general reaction expression
JJ i _ NR
J a. . a. + a. M = 7 a'.', a. + a. M j = 1,2, . . .,JJ.
= ^ x 3 1J i J
An additional relation is provided by the ideal gas equation of state
P =
449
-------
The mixture viscosity and thermal conductivity are calculated from single
component transport properties using the semi-empirical expressions of Wilke (8),
Since the principle constituent in the present system is NZ, trace diffusion
coefficients for the ith species in N, were used to simplify the calculations.
Single component properties and the binary diffusion coefficients, D^ , were
calculated from the Chapmen-Enskog theory with molecular parameters evaluated
using a Lennard Jones potential model (8). Thermodynamic data for specific
heat and enthalpy were calculated from JANAF data using a formlism based on
the NASA Complex Chemical Equilibrium Code (9).
Boundary Conditions
Boundary conditions for the above equations are required at the plate
surface and at the outer edge of the boundary layer. In the present case the
veloicty is zero at the plate surface (no slip condition) and the surface temp-
erature is specified. These conditions can be written
f (5,0) «= 0; h(5,0) = hw(x) (5)
The general form of the species boundary condition is given by
(6)
pr
where R is the surface reaction rate per unit area. In the case of a non-
s r
catalytic wall the surface reaction rate is equal to zero. At the outer edge
of the boundary layer
f'U,ne) = 1 ; hU,ne) «= h
(7)
oico
where the free stream conditions are assumed to be constant along the plate.
Surface Reaction Model
While the general role of a catalyst in promoting chemical reaction is
well known, there is considerable uncertainty in many systems concerning the
kinetic mechanisms involved. This is particularly true at the high temperatures
and pressures encountered in combustion systems. Experimental results presented
in Section 2 and in a previous report (7) indicate the existence of significant
surface reaction for H_/air mixtures flowing over a platinum surface at plate
450
-------
temperatures as low as'470 C. This was modeled as a one step surface reaction
in which H. is oxidized to form H20 as a product (7)
H2 + 1/2 02 SURFACE> H20
The resulting surface energy release rates were found to correlate well with
an Arrhenius type rate expression given by
R = 1.4 (IL. ) exp(-3850/RT)m°*eS (8)
S n2 S cm sec
where (i^ ) is the concentration of H2 at the surface.
An additional effect which must be considered is that of the plate surface
on radical concentrations. It is likely that radical recombination will be
important at the plate surface for the relatively low surface temperatures
being considered. To include this effect in the calculations radical recombination
.at the surface was modeled by the set of reactions shown in Table I where the
individual reaction rate expressions are of the form
Y(n.)sC 2
Rg = (n^g kg = i^Ji (mole/cm sec) (9)
Y is the reaction probability per collision, (n.) is the concentration of
1. O
species i at the plate surface and C. is the molecular velocity. The quantity
th
niC1/4 thus represents the collision frequency of the i specie with the plate
surface. Substitution of the surface reaction rate expressions (8) and (9)
into Eq. (8) provides the necessary species boundary conditions.
TABLE I. Radical Recombination Surface Reactions
Reaction Rate Constant, kg
0 + S * 1/2 02 + S Y0C0/4
H + S * 1/2 H2 + S >HCH/4
OH + 8 + 1/2 H2 + 1/2 02 + S fOHCOH/4
H02 + S * 1/2 H + 02 + S YHQ CHO /4
2 2
451
-------
To assess the relative importance of surface oxidation of HZ and radical
recombination, the calculations were carried out for two models of catalytic
wall reaction. In case 1 only the surface oxidation of H2 is considered. In
case 2 surface radical recombination is considered in addition to surface oxi-
dation of H2. To determine the maximum possible influence of radical recombin-
ation a reaction probability of unity (y-D was used in the latter case. The
plate surface was considered noncatalytic (i.e. Rg = 0) for all species not
undergoing surface reaction.
Gas Phase Kinetics
The forward and reverse reaction rates of the i reaction can be written
a. NS a.' .
• R = k. (pa) J IT (pa.) 13 , j = 1.2...J
m =
- MO ' '
a NS a. .
R = k (pa ) 3 TT (pa.) J , j = 1.2...J
-J -J m i=1 i
where k. is an Arrhenius rate coefficient of the form
k.. * 10BJTN:) exp (-E./RT).
Reverse rate coefficients were assumed to be related to the forward rate con
stant through the equilibrium constant K .
k.
1
c
The reaction mechanism chosen for H^/air combustion consists of 13 reactions
8 chemical species. These are shown in Table II.
Numerical Procedure
The above system of equations was solved using an implicit finite differ-
ence procedure. The resulting difference equations and the boundary conditions
form a system of equations of the tridiagonal type. The approach of the present
investigation was to use a standard tridiagonal matrix solution algorithm to
solve the hydrodynamic equations for conservation of mass and momentum (16).
452
-------
The coupled energy and species conservation equations were then solved using
a Newton Raphson correctional scheme developed by Pratt (17) .
For the results presented a grid consisting of 50 points across the
boundary layer between r\ = 0 and n =9.0 was found to be sufficient. To
improve accuracy near the wall where the steepest gradients occur a nonuniform
grid was used with An given by Eq. (10) with k = 1.07.
_ 1} ' N = -l ' (10)
Step size in the streamwise direction, A£, was adjusted continuously to meet
the desired accuracy of 5%. Minimum step size corresponded to a time step on
the order of 10 sec.
RESULTS
Noncatalytic Plate Surface
The results of the numerical calculations for the case of H2/air flow
over a noncatalytic heated plate with combustion at an equivalence ratio of
0.1 are shown in Figs. 8-10. In the case presented the surface temperature
was assumed constant at 1100 K and the free stream velocity and temperature
were 3.17m/s and 293 K respectively.
Temperature profiles are plotted against the nondimensional boundary layer
coordinate r| in Fig. 8 for several distances downstream of the plate leading
edge. The temperature profiles remain self similar until a distance of approx-
imately 0.6 mm along the plate. Upstream of this location little heat release
due to chemical reaction has occurred. At x = 1.89 mm combustion is apparent
near the surface as can be seen by the peak in the temperature profile. A
short distance downstream (x = 2.00 mm) the heat release has become significant
with the temperature reaching a maximum of 1470 K. Combustion has also spread
farther out into the boundary layer. At x = 2.80 mm the heat release rate has
decreased in regions near the plate surface as H9 is used up and heat transfer
to the plate has resulted in a decrease in gas temperature. Combustion of in-
coming reactants continues to occur farther out in the boundary layer but at
a decreasing rate due to the lower temperatures.
Typical species concentration profiles for several distances downstream
from the plate leading edge are shown in Figs. 9 and 10. Profiles at x = 1.35 mm
453
-------
correspond to the region before any heat release due to chemical reaction has
occurred. 0, H, OH, and HCL radical concentrations are rapidly increasing in
the high temperature region near the plate surface and the H20 concentration,
while comparable to the radical concentrations, is still quite low. HZ has
undergone little reaction at this point.
At x - 1.89 mm a small amount of chemical heat release has occurred in
the boundary layer (the corresponding temperature profile is shown in Fig. 8).
This release corresponds to the reaction of HZ and 02 near the plate surface,
and the formation of relatively large amounts of H20. Radical concentrations
of H, 0, and OH have attained their peak values near the plate and are begin-
ning to decrease as recombination reactions become important.
Catalytic Plate Surface
The effect of including surface oxidation of HZ on boundary layer com-
bustion (case 1) is shown in Fig. 11 and 12. As can be seen from Fig. 11
surface oxidation of H~ effectively inhibits the onset of heat release due
to gas phase combustion. In this case no significant heat release is found
to occur prior to a distance of 4.0 mm from the leading edge. Farther down-
stream gas phase combustion once again results in local increases in temperature
across the boundary layer. The heat release rate is, however, considerably
lower than that obtained with the noncatalytic plate. Temperature increases
are limited to a maximum of approximately 100 K and only a relatively small
increase in thermal boundary layer thickness occurs (compare Fig. 8 for a
noncatalytic plate).
The reason for the reduced gas phase heat release rates can be seen in
Fig. 12 where H2 and OH concentration profiles are shown for case 1. Surface
oxidation directly results in a reduction in H~ concentration near the plate
surface where maximum heat release was found to occur in the case of the non-
catalytic plate. Since less H~ is available for reaction in this region the
overall gas phase heat release rate is significantly reduced. In addition,
the onset of gas phase combustion is delayed until farther downstream due to a
slower increase in radical concentrations near the leading edge.
The effect of adding surface radical recombination in case 2 is shown in
Fig. 13 where the H_ and OH concentration profiles are shown for several dis-
tances downstream of the plate leading edge. With both the noncatalytic plate
454
-------
and case 1 maximum radical concentrations were found to occur in the high
temperature region near the plate surface. As can be seen from Fig. 13 radical
recombination results in a rapid decrease in OH concentration in this region
and a slower initial radical buildup.
The predicted thermal boundary layer thickness, <$„,, is shown as a function
of distance from the plate leading edge in Fig. 14 for the surface boundary
conditions considered above. Also shown for comparison are results for a
heated plate with no combustion. In the case of a noncatalytic plate, the
high gas phase heat release rate results in a rapid increase in 6 at approx-
imately 2 mm downstream from the leading edge. With a catalytic surface the
increase in 6 is only about 30 percent greater than for the case of no com-
bustion at a distance of 20 mm downstream, with the difference between cases
1 and 2 being relatively small. In case 1 gas phase heat release is first
apparent at a distance of 4 mm downstream from the leading edge. In case 2
radical recombination at the surface results in boundary layer combustion
being delayed until approximately 10 mm downstream.
COMPARISON WITH EXPERIMENT
The above study of surface boundary condition effects indicates that both
models for surface reaction predict a relatively small increase in 6 over that
obtained with no combustion. This behavior agrees well with the experimental
results shown in Fig. 7. Since it is likely that both radical recombination
and surface oxidation of H_ are important in the present system, this was
chosen as the most realistic boundary condition to apply in the case of a cat-
alytic surface and the calculations were repeated for a set of conditions at
which Rayleigh data were taken. An equivalence ratio of 0.2, a plate tempera-
ture of 1170 K, and a free stream velocity of 1.5 m/s was chosen.
The resulting thermal boundary layer profile is shown with the experimental
results for comparison in Fig. 15. It can be seen that the predicted increase
in thickness is much more abrupt and occurs farther upstream (approximately
2 mm from the leading edge) than is found experimentally. Since good agreement
was obtained between predicted and experimental results when no combustion was
present, it would appear that the discrepancy in boundary layer thickness is due
either to the kinetic mechanism used or to the model employed for surface re-
action. The rate expression used for surface oxidation of H_ was based on sur-
face heat release rates measured in the present system under similar operating
455
-------
conditions and thus is expected to provide a good model for HZ oxidation. Since
a reaction probability of unity was used in the surface radical recombination
rates the calculations should reflect the maximum possible influence of radical
recombination on the combustion process. Therefore it appears that the most
likely cause of the observed discrepancy is the gas phase reaction mechanism
employed.
It would be difficult at this time to conclude which reaction rates are
the primary cause of the discrepancy since the occurrence of combustion is a
function of both the initial radical buildup rate and energy release rates in
the primary combustion zone. A preliminary examination of reaction rates dur-
ing the combustion process shows that initial radical buildup is controlled by
the bimolecular branching reactions (1) , (2), and (3), and reaction (13) in-
volving HO. (Table II), whereas energy release is due primarily to reactions
(2) and (8). Rates for the bimolecular branching reactions (1) through (3)
are relatively well known, for the most part to within 30%. Expected uncer-
tainty in rates for the three body recombination reactions is somewhat greater,
but these reactions were found to be relatively unimportant to the combustion
process (except for reaction (8) involving EO ). The greatest uncertainty in
the present kinetic mechanism appears to exist in the H0? reactions, the two
most important being reactions (8f) and (13r). A more detailed study of the
reaction mechanism is currently being made to determine if the above uncer-
tainties in reaction rate constants could account for the observed discrepancy.
456
-------
SECTION 4
SUMMARY
A study of boundary layer combustion for lean H_/air mixtures flowing over
a heated platinum plate indicated the presence of significant surface heat re-
lease for all mixture ratios at plate temperatures as low as 470 K. At the
maximum equivalence ratio studied ( = 0.3) stable boundary layer combustion
was observed for plate surface temperatures between 1070 K and 1170 K. At
temperatures greater than 1170 K combustion was apparent very near the plate
leading edge and the primary reaction zone took on a flame like structure
characterized by a steep temperature gradient which extended well out into the
free stream. For lower equivalence ratios the plate surface temperature needed
to maintain gas phase combustion increased significantly.
Numerical calculations for a lean ($ = 0.1) H»/air mixture flowing over
a heated noncatalytic plate with a wall temperature of 1100 K indicated the
existence of several stages of boundary layer combustion. These included an
upstream region in which radical concentrations increased with little assoc-
iated heat release and a downstream region in which heat release due to the
reaction of H? became appreciable. In the upstream region temperature pro-
files were approximately similar and boundary layer thickness corresponded to
that for no combustion. Heat release due to gas phase combustion resulted in
a significant increase in thermal boundary thickness and gas temperatures
greater than the wall temperature were achieved.
The effect of surface reaction on the combustion process was investigated
using boundary conditions for a catalytic surface based on 1) a condition which
includes finite rate surface oxidation of H_ and 2) a condition which includes
radical recombination at the plate surface in addition to surface oxidation of
H2< Surface reaction was found to have a strong quenching effect on the initia-
tion of gas phase combustion in the boundary layer. This was due to the deple-
tion of H_ near the plate surface and quenching of radical species generated
457
-------
in the gas phase.
A comparison between experimental results and numerical calculations
utilizing condition 2) above showed poor agreement. The predicted increase
in thermal boundary layer thickness was more abrupt and occurred farther
upstream than was found experimentally. This discrepancy was attributed to
uncertainties in the gas phase kinetic mechanism since condition 2) should
result in the maximum possible surface quenching of gas phase combustion.
458
-------
REFERENCES
1. Blazowski, W.S. and Walsh, D.E., Combust. Sci. Techno1. 10,
p. 233 (1975).
2. Pfefferle, W.C., Carrubba, R.V., Heck, R.M. and Roberts, G.W., "Catathermal
Combustion: A New Process for Low Emissions Fuel Conversion," ASME Paper
75-WA/FU-l (Dec. 1975).
3. Kesselring, J.P., Brown, R.A., Schreiber, R.J., and Moyer, C.B., "Catalytic
Oxidation of Fuels for NO Control from Area Sources," Environmental Pro-
tection Agency Report EPA-600/2-76-037 (February, 1976).
4. Jenkins, F.A. and White, H.E., Fundamentals of Optics, McGraw-Hill, New
York, p. 462 (1957).
5. Weinberg, F.J., Optics of Flames, Butterworths, London (1963).
6. Creeden, J.E., Fistrom, R.M., Grunfelder, C. and Weinberg, F.J., J. Phys.
D, 5, 1063 (1972).
7. Schefer, R., Cheng, R., Robben, F., and Brown, N., "Catalyzed Combustion
of H?/Air Mixtures on a Heated Platinum Plate," presented at the Western
States Section/The Combustion Institute, Boulder, CO (1978).
8. Bird, R.B., Stewart, W.E. and Lightfoot, E.N. Transport Phenomena, John
Wiley and Sons, New York (1960).
9. Gordon, S. and McBride, B., "Computer Program for Calculation of Complex
Chemical; Equilibrium Composition," NASA SP-273 (1971).
10. Baulch, D.L., Drysdale, D.D., and Home, D.G., Fourteenth Symposium (Inter-
national) on Combustion, The Combustion Institute, Pittsburgh, 107 (1973).
11. Wilson, W.E., Jr., J_. Chem. Ref. Data 1, 535 (1972).
12. Baulch, D.L., Drysdale, D.D., Home, D.G., and Lloyd, A.C., Evaluated
Kinetic Data for High Temperature Reactions, Butterworths, London, Vol. 1
(1972).
13. Myerson, A.L. and Watt, W.D., J. Chem. Phys. 49, 425 (1968).
459
-------
14. Johnston, H.S., "Gas Phase Reaction Kinetics of Neutral Oxygen Species,"
NSRDS-NBS 20 (1968).
15. Lloyd, A.C., Int. ,J. Chem. Kinetics 6, 169 (1974).
16. Blottner, F.G., Johnson, M. and Ellis, M., "Chemically Reacting Viscous
Flow Program for Multi-Component Gas Mixtures," Sandia Laboratories
Report SC-RR-70-754 (1971).
17. Pratt, D.T., and Wormeck, J.J., "CREK - a computer Program for Calculation
of Combustion Reaction Equilibrium and Kinetics in Laminar or Turbulent
Flow," Washington State University Report WSU-ME-TEL-76-1 (1976).
460
-------
TABLE I
H- - Air Reaction Mechanism
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
Reaction
0 + H2 £ OH + H
H2 + OH J H20 + H
H + 02 j 0 + OH
OH + OH j H20 + 0
H2+M^H + H + M
H + OH + M £ H20 + M
0 + 0 + M Z 02 + M
H + 02 + M + H02 + M
H02 + H j OH + OH
HO- + OH •*• H90 + 0,
£ ~^~ £ £
H02 + 0 J OH + 02
H02 + H £H20 + 0
H02 + H + H2 + 02
k
10? ' 255T1 ' °exp (-4479 . 0/T)
1010-35exp(-2617.0/T)
1011-342exp(-8450.0/T)
9 Ron
10 exp(-550.0/T)
. 9.348 m0.5 , ,,„,«„ n,^
10 T exp (-48414. 0/T)
1016.342 T-2.0
1011.410 T-l-Oexp(_171.5/T)
109>35°exp(500.0/T)
1011-4°°exp(-950.0/T)
1010-6"exp(-503.3/T)
1010'6"exp(-503.3/T)
1010'6"exp(-503.3/T)
1010-398exp(-350.0/T)
Ref.
10
11
12
12
13
12
14
12
12
15
15
15
15
"'Units - k : (m kg mole" )n~ sec" where n is reaction order
T : °K
461
-------
Fl ow
XBL 771 -127
Figure 1. Design of the quartz flat plate and the holder.
462
-------
OJ
Catalytic flat
plate
Laser
Spectrometer
Photomultiplier
I
Laser beam trap
Rayleigh detector
Combustion in the boundary layer over a heated catalytic flat plate
Details of flat plate
VRL 764-2757A
Figure 2. Schematic of the experimental apparatus for measurement of
velocity and density boundary layer profiles over a flat plate.
-------
Figure 3. Deflection mapping photograph of the flat plate boundary layer.
H /air combustion. 4> = 0.3, T = 1200 K, U^ = 1.5 m/s,
T = 293 K.
00
464
-------
1200
1000
T(*K)
800
600
400
200
Y(mm)
6
XBL 784-7919
Figure 4. Temperature distribution as a function of distance above the plate.
Temperatures are based on differential interferometer measurements.
H_/air combustion. T = 1270 K, Uro = 1.5 m/s, T^ = 293 K.
£* S
465
-------
.30
.20
UJ
o
z
UJ
A
SURFACE
COMBUSTION
.10
BOUNDARY
LAYER
SURFACE
COMBUSTION
T
o
(U
_L
_L
770
970
1170 1370
T(°K)
1570 1770
XBL 784-7918A
Figure 5.
Combustion regions for H2/air mixtures over a heated platinum plate
U = 1.5 m/s, T = 293 K. A indicates a point at which Rayleigh
data was taken.
466
-------
0
4
Y (mm)
8
XBL 788-10512
Figure 6. Normalized density profiles over a heated platinum plate at a dis-
tance of 30 mm from the leading edge. H_/air combustion T = 1170 K,
U^ = 1.5 m/s, TOT = 293 K, X, = 0.15; 0, = 0.20; A, S = 0.25.
Solid line indicates numerical results for no combustion.
467
-------
D
tl
I070K
10 15 20 25
X (mm)
30
35
40
XBL 788-10509
Figure 7. Effect of plate surface temperature on thermal boundary layer thick-
ness (p/p^ = 0.5) for flow over a heated platinum plate. H2/air
combustion. <{> = 0.20, U^ =1.5 m/s. Solid lines indicate numerical
results for no combustion.
468
-------
1400 -
200
.6 1.2 1.8 2.4 3.0 3.6 4.2 4.8
XBL 788-10507
Figure 8. Boundary layer temperature distribution for a noncatalytic plate.
H0/air combustion.
-------
1,2 1,8 2.4 3,0 3,6 4,2
XBL779-2007A
Figure 9.
Concentration profiles as a function of nondimensional distance above
plate for a noncatalytic plate, x = 1.35 mm downstream from leading
edge. H2/air combustion. Q - 0.10, Tg =1100 K, U^ - 3.17 m/s.
470
-------
10"'
10
-3
I 10
o
10
I I I '_
0 0.6 1,2 1,8 2.4 3.0 3.6 4,2 4.8
XBL779-2006A
Figure 10. Concentration profiles as a function of nondimensional distance above
plate for a noncatalytic plate, x = 1.89 mm from leading edge. I^/
air combustion.
-------
1100
1000
900
800
~ 700
500
400
300
200
,X=4.65
-X =22.57
XBL 788-10504
Figure 11. Boundary layer temperature profiles for a catalytic plate. Case 1.
Surface oxidation of H . H /air combustion, (j) = 0.1, T = 1100 K,
U^ = 3.17 m/s, Tro = 293 K. b
472
-------
o>
"o
E
CJ>
XBL 788-10513
Figure 12. H» and OH concentration profiles as a function of nondimensional
distance above plate surface for a catalytic plate. Case 1. Sur-
face oxidation of H-. H_/air combustion. 4> = 0.1, T = 1100 K,
UK, = 3.17 m/s, T = 293K7 , a ; , a .
rt/} OH.
473
-------
o>
o>
o
E
o>
JC
10
"
F
j
- 8.9mm
4-
A
i \
i \
\
\
\
\
\
\
\
\
\
\
\
, \
\\
\\
\\
\\
\\
u
n
\\
\
\
t l i
H j _
OH (X 10* )
-
—
I j i
0
f -
Figure 13. H and OH concentration profiles as a function of nondimensional
distance above plate surface of a catalytic plate. Case 2 Sur-
face oxidation of H with radical recombination. H0/air combus-
tion.
-------
NONCATALYTIC
WALL
0
0
20
XBL 788-10506
Figure 14. The effect of surface boundary condition on thermal boundary layer
thickness. H_/air combustion. (f> = 0.1, T = 1100 K, U = 3.17 m/s,
T = 293 K. z b °°
475
-------
6
3
2
0
0
COMBUSTION
NO COMBUSTION
Figure 15.
10 15 20 25 30 35 40
X (mm)
XBL 788-10505
Comparison of experimental and predicted thermal boundary layer
thickness profiles for a heated platinum plate. H_/air combustion.
- 0.20, T = 1170 K, U = 1.5 m/s. 2
476
-------
APPLICATION OF RICH CATALYTIC
COMBUSTION TO AIRCRAFT ENGINES
by:
G. E. Voecks
D. J. Cerini
Jet Propulsion Laboratory
California Institute of Technology
4800 Oak Grove Drive
Pasadena, CA 91103
477
-------
ABSTRACT
This program was aimed at evaluating the hydrogen enriched operation of
an aircraft piston engine as a means of safely operating the engine at a lean
air/fuel ratio during cruise mode. An earlier demonstration of this approach
to automobile operation indicated a gain in economy and it was expected that
most of the same technology and procedures could be transferred to the aircraft.
The hydrogen generator utilizes a catalytic rich combustion process to produce
a gaseous mixture of hydrogen, carbon monoxide and nitrogen from liquid fuels
and air. The hydrogen generator was expected to involve little modifications
from the preceding system. However, constraints peculiar to the aircraft
application required a series of modifications in the generator (or combustor)
design. To meet the aircraft system requirements, a generator scale-up, a
change in catalyst support, an inlet design change, a different start-up
approach, and a modified generator configuration were implemented. Optimi-
zation of each modification was not pursued because of the program constraints
concerning the generator development. Use of a monolith catalyst was shown
to be superior to the pelleted catalyst used in the previous automotive program
in performance and in solving the problem of generating "fines" resulting from
vibration. Both electrical and hot gas start-up modes were implemented. A
successful operation of the rich combustor with liquid fuel over a range of
fuel feeds from 4.2 Ib/hr. to 25 Ib/hr. at an air-fuel ratio of 5.4 was
achieved. The resulting catalytic reactor was operated on the aircraft in the
final phase of the program. This application constitutes but one segment of
staged combustion in which catalysts will ultimately contribute greatly.
479
-------
SECTION 1
APPLICATION OF RICH CATALYTIC COMBUSTION TO AIRCRAFT ENGINES
INTRODUCTION
The concept of operating a piston engine at equivalence ratios leaner
than the lean burn limit of the fuel in order to reduce the NO levels in the
(1) X
emission has recently been demonstrated . This was achieved by generating
hydrogen on-board a vehicle and subsequently mixing the hydrogen with the
gasoline/air mixture. The method chosen for generating hydrogen which was
deemed most compatible with a vehicle was catalytic partial oxidation, i.e.,
rich combustion; (-CH^f ^02(air) - > CO + H2>
During the past year, a joint NASA/Lycoming/Beech/JPL effort was under-
taken to evaluate the application of this concept to piston driven aircraft.
Since the current engines are operated constantly at rich equivalence ratios,
it was anticipated that the engine could be operated leaner than stoichio-
metric to extend the range of operation while simultaneously not produce
excessive NO emissions. During this program an aircraft was successfully
< X
operated at lean conditions with liquid fuel and liquid fuel/gaseous hydrogen
mixtures . The catalytic hydrogen generator constructed for this purpose
provided several advances in the area of rich catalytic combustion. The unit
constructed will be discussed in the following text.
Although this catalytic partial oxidation or rich combustion unit was
installed as a hydrogen generating device on transportation vehicles, the
concept is the same as two-stage combustion in stationary applications. By
catalytically generating hydrogen (and carbon monoxide) from a liquid fuel and
air mixture, the ensuing mixture with more air and fuel can be completely
combusted '(as in the combustion chamber of the piston engine) for driving
turbines or heating boilers at temperatures below the NO formation linit.
A
Stationary installations, however, would appear to place somewhat less string-
ent requirements on the first stage combustor design than were encountered
during the aircraft project. •
481
-------
SECTION 2
DESCRIPTION OF THE CATALYTIC PARTIAL OXIDATION AIRCRAFT GENERATOR
The decision to use catalytic partial oxidation to generate hydrogen on-
board the aircraft was based on the .previous assessment made during the NASA
sponsored Low Pollution Engine Program for automobile implementation. Earlier
efforts to partially oxidize the fuel thermally (i.e., no catalyst), whereby
controls and operation would be even simpler, were unsuccessful. This was
due to the formation of soot. Catalytically, however, rich combustion can be
maintained near the equilibrium soot limit where the hydrogen yield is maxi-
mized (Figure 1) without soot production.
In addition to the non-soot/high hydrogen yield requirement, the aircraft
application demanded several other specific design considerations for success-
ful operation. Because of the limited space within the engine housing only a
3
volume of 2.5 ft. was available for the entire reactor. This included the
fuel vaporizer, air-fuel mixer, pre-heat, control valves and mounting fixtures.
With respect to the reactor itself, the high conversion efficiency had
to be maintained over a range of 0.5 Ib/hr. to 3.0 Ib/hr. of hydrogen. Based
on the fuels used [Indolene (Federal Register specified gasoline) and aviation
fuel], 8.3 Ib/hr. reacting with 45.2 Ib/hr. air (air/fuel ratio = 5.4 to 5.5)
would generate one Ib/hr. of hydrogen. The resulting space velocity variation
for the generator operation in the final analysis was 2,700 to 16,200 hr~ .
This constituted a two-fold increase in both maximum flow and turn down ratio
over which the generator had previously been constructed for the automobile
program.
Two other constraints 'peculiar to this application were imposed on the rich
combustor: two modes of start-up were requested, and the catalyst and reactor
would have to withstand the high vibration intensities and frequencies which
the engine created during operation.
Start-up had previously been accomplished by pre-heating the catalyst bed
to the light-off temperature with a flow of hot air. In the aircraft
482
-------
application the exhaust gas. from the engine (which started in the normal rich
mode) was routed through the reactor to heat the catalyst. The second method
for starting, which was to be a backup approach, was to pass an electric
current through wires coiled within the front section of the catalyst bed.
With no air flow through the generator, the batteries on-board the aircraft
were more than adequate for bringing the inlet section of the catalyst to light-
off temperature, typically 1000°F.
Assessing the vibration problem was difficult initially because the
combined testing of the engine and generator was completed with the engine in
a test dynomoeter and the gearator mounted separately. The severity of. the
problem was realized during testing in which the generator was mounted on the
engine.
Catalyst Description
Initial tests were conducted with the commercial pellet catalyst type
employed during the automotive hydrogen enrichment program; a nickel reforming
catalyst. This catalyst was 25 wt% nickel oxide coprecipitated with alumina
2
and had an initial surface area of 60 m /g. A nickel catalyst had been chosen
for the rich combustion because it maintained good activity with various fuels
and because a non-noble metal was preferred. Although monolithic catalysts
had been tested under rich and lean conditions previously at JPL, none had
actually been used in a generator prior to this program. Two distinct
advantages were shown to favor the monolith catalyst compared to the pellet
catalyst. During the aircraft engine dynomometer tests with the generator
installed on the engine, the pellet catalyst produced fines which were carried
into the engine resulting in mechanical damage. By proper packing, the 300
2
cell/in Corning cordierite monoliths (5.63 in diameter) eliminated this
problem. Nickel oxide was also used as the catalyst on the monoliths to provide
a closer comparison to the pellets. The nickel oxide, applied on the washcoat,
2
was nominally 12 wt% and had a surface area of approximately 22 m /g. The
2
hydrogen adsorption measurement of the nickel on a fresh catalyst was 3.8 m /g.
Catalytica Associates achieved the high nickel loading by multiple applications
of nickel salts. In Figure 2 the variation of hydrogen yield is shown for both
pellet and monolith catalysts as a function of air to fuel ratio. In Figure 3
the catalyst temperatures during start-up and operation are compared between
the monolith and pellet catalysts. The monolith catalyst demonstrated this
superior performance throughout testing.
483
-------
The generator size constraints required a different length to diameter
ratio than had previously been used for automotive use in order to provide
enough catalyst for operating over the entire range of air and fuel through-
puts. Scale-up from a 3.66 in. diameter to 5.63 in. diameter and from 6 in
catalyst bed length to 8 in. was found to be appropriate.
Use of the electric preheat with the monolith catalysts was accomplished
by sectioning the first monolith into segments. The coils of heater wire
were placed between the segments in the upper half of the catalyst bed. The
reactor with an expanded view of the monolith before installation is shown
in Figure 4.
The mounting site constraints also inhibited the length of the mixing and
preheat sections, and required a redesigning of the inlet system to maintain
uniform flow through the monoliths. Figure 5 shows a cutaway illustration of
the rich combustor. Air is preheated by the heat from the reaction when
passing through the spiral vanes surrounding the inside wall of the reactor.
The preheated air then passes through the atomized fuel inlet, proceeds through
an air/fuel mixer and is expanded into an outer manifold prior to entering
the inlet of the reactor. The air/fuel mixture is then introduced through a
ring of holes above a diffuser at the top of the monolith catalyst.
Summary
Use of a rich combustion reactor to generate hydrogen for two-stage
combustion was successfully implemented with a piston engine on an aircraft.
The peculiar problems involved in this application were (1) use of a catalyst
which with proper packing would maintain adequate heat transfer, match the
expansion of the reactor and maintain shock and attrition resistance, (2)
scale-up a reactor disproportionately to meet size constraints, and (3) design
inlet conditions for a wide range of air-fuel flows and maintain uniform
distribution through the monolith catalyst. Application of staged combustion
to turbines and boilers with the use of a catalyst used in the rich combustion
stage would offer the same advantages realized in the aircraft and automobile
applications, i.e., high hydrogen yield, no soot. Second stage combustion
could be achieved either thermally (as in the piston engines) or catalytically
to meet whatever heat realease, temperature, emission, or geometrical require-
ments which may be imposed by the application. The scale-up achieved in the
work reported here is but one step toward realizing larger industrial
utilization of the same principle.
484
-------
References
1. Houseman, J., and Cerini, D. J., "On Board Hydrogen Generator for
a Partial Hydrogen Injection Internal Combustion Engine," SAE -
740600, 1974.
2. Chirivella, J. E., Duke, L. A., and Menard, W. A., "High Fuel
Economy in an Aircraft Piston Engine When Operating Ultralean,""
SAE-770488, 1977-
485
-------
10
z
LU
u
a:
LU
a.
z
o
= 20h
O
Q.
5
O
u
i—
u
D
Q
O
10 h
0
FIGURE 1.
5 10
AIR/FUEL MASS RATIO
Theoretical Equilibrium Product Composition for Indolene
(CH )/Air Adiabatic Combustion at 80°P, 44 psia.
15
486
-------
1.0
0. 9
a.
o
CD
o
Q
o
o
Of
0.8
0.7
0.6
EQUILIBRIUM REACTION AIR + CH 1.92
CATALYST TYPE
•-0- MONOLITH
PELLET
2.2
2.4
2.6
2.8
AIR TO CARBON, mole ratio
i i i i
FIGURE 2.
5.2 5.3 5.4 5.5 5.6
AIR TO FUEL, mass ratio
Molar Ratio of Hydrogen Produced to Fuel Carbon Input as
a Function of Air to Carbon Molar Ratio for the Mono-
lith and Pellet Catalysts.
487
-------
2000
5 1000
UJ
Q_
00
00
0.5
2.0
CATALYST TYPE
MONOLITH
PELLET
X = DI STANCE FROM
TOP OF CATALYST, in.
0
I I
0
100
200
300
TIME, sec
FIGURE 3.
Bed Temperature Profile of the Monolith and Pellet
Catalyst During Light-Off and Steady State Operation.
-------
r
CO
Figure 4. Aircraft generator with expanded catalyst bed.
-------
CLAMP
BYPASS AIR
AIR/FUEL
MANIFOLD
VANES
AIR IN
ELfCTRICAL
HEAT ING ELEMENT
BELLOWS
FUEL ATOMIZER
MONOLITH
CATALYSTS
AIR/FUEL
MIXER
PRODUCT
FIGURE 5. Cutaway of Aircraft Generator
490
-------
CATALYTIC COMBUSTION IN ACTUAL ENGINES
A summary of Engine and Rig tests
By:
B.E. Enga
Johnson Matthey Research Centre
Reading, Berkshire, England.
491
-------
ABSTRACT
This paper outlines work carried out at JMRC over a period of years into
the application of catalytic combustion to gas turbine engines.
The technique adopted consists of examination of the requrements of an
actual gas turbine followed by rig and computer experimental work to solve
each problem.
A brief outline of the problems of the flow and thermal transients together
with start up considerations is given.
An account of the preliminary series of tests carried out on a 65
BHP gas turbine is included together with recorded emissions from this engine.
The system designed to run this engine and overcome the problems exper-
ienced is described and attention drawn to the fact that a system is required
to ensure efficient use of the concept. It is not sufficient to simply place
a catalyst in a combustor.
493
-------
Introduction
At Johnson Matthey Research Centre we have been studying catalytic
combustion phenomena for some years. Based upon many series of tests and
rig evaluation very accurate computer simulation models of JM catalyst
systems were developed for catalytic combustion and thus the basic param-
eters of the catalyst-support are rapidly computable for any given flow
and fuelling system.
In parallel with this work the required catalyst system to power
various applications were determine from the engineering point of view.
A TARGET GAS TURBINE:
1. Theoretical considerations
The simple gas turbine cycle comprises the compression of the working
fluid, energy addition at constant pressure and expansion of the working
fluid. The loss incurred in the system must include compression and
expansion Inefficiencies, pressure losses through system, and above
atmospheric pressure - work lost in the exhaust annulus. Thus the overall
thermal efficiency of the system may be stated as:
Eff.
EXPANSION Work - Compressor Work
Energy Supplied
-(1)
This can now be resolved into its component form.
fo-dffir
U-*
-------
Where
Effa = Overall thermal efficiency
(")« = Combustion efficiency
O = Turbine isentropic efficiency
O = Compressor isentropic efficiency
f = Fuel/air ratio
Cpe = Mean specific heat at constant pressure during expansion
Cp = Mean specific heat at constant pressure during compression
Cpf = Mean specific heat at constant pressure during combustion
^t = Mean ratio of specific heats during expansion
&c = Mean ratio of specific heats during compression
PI = Delivery pressure
A PC = Pressure loss through combustion chamber as % inlet P
A Pt = Pressure loss in power turbine annulus as % inlet P
TI = Inlet temperature
TO = Compressor delivery temperature
T- = Gas generator turbine inlet temperature
T^ = Gas generator turbine exit temperature
T,- = Power turbine exhaust annulus temperature
From equation 2 it can easily be seen that for a standard engine working
at any given settings the only factors that the combustion process affects
are
-------
This thus served to give the area of AP that we were to work in.
B CATALYST SUPPORT
Having determined the importance of pressure drop many forms of support
material were then tested on our 10 Ibs/sec flow rig.
From this work it was concluded that for a given cell density and cell shape
the overriding parameter is the open area. As this is easily variable with
metal support system the concept was then scaled up.
1. Tjarget Engine flow
There are two systems of catalyst that can be applied to our target
engine, an annular combustor and multi canular combustors. Scale down
test runs of the two systems gave the following results:
2
Annular Combustor 0.617m x 0.355m long target size
I
Reynolds No
AP psi
A P % inlet
Open Area 0.8
3,430
5
3.06
Open Area 0.9
3,049
4.3
2.63
Open Area 0.96
2,859
3.97
2.43
8 canular combustors, 0.4257 m x 0.355 m long target size
Reynolds No
A P Psi
A p #
Open Area 0.8
4,971
9.69
5.95
Open Area 0.9
4,419
6.98
4.68
Open Area 0.96
4,144
6.39
3.92
The 0.9 open area support corresponds to the JM 400 cell metal support and
the 0.96 open area corresponds to the JM 200 cell metal support.
Pressure drop evaluation was continued at approach reference velocity
of up to 90 metres/sec and mass flows of up to 100 Kgs/m^/sec and from this
it became apparent that no simple relationship with exterior parameters was
present. Test results for the 400 cells and 200 cell materials are give in
Figure 1-4.
497
-------
Compressor isentropic efficiency
Turbine isentropic efficiency
Combustion efficiency
Standard APc
Intake Pressure
Intake Temperature
Final exhaust pressure
Gas generator exhaust annulus diameter
0.82
0.85
1.00
4.0%
1.0137 bar
288° K
1.0344 bar
622.3 mm
4.0$ A PC
T4°K
LP Compressor speed rpm
Exhaust man flow rate Kg/S
P2 Bar
T, °K
3
T2 °K
Fuel/air ratio
T IT
Expansion work KV/Kg
Compression work KV/Kg
Energy supplied KV/Kg
LOAD
Emergency
923
6550
107.72
10.344
1192
605
0.0163
745.9
512.1
328.3
677.3
Desicm
886
6340
104.54
9.862
1157
595
0.01536
730
488.4
319.0
645.6
Base
823
5950
96.36
8.793
1080
574
0.0136
690
435.1
298.1
570.2
Thus the following can be related
LOAD
APc
4.0$
6.0$
10.0$
15.0$
Effect on performance
Overall thermal efficiency $
Power loss $
)verall thermal efficiency $
Power loss $
Overall thermal efficiency $
Power loss $
Overall thermal efficiency
Emergency
27.14
2.013
26.58
6.935
25.48
11.42
24.04
Design
26.23
2.007
25.7
6.434
24.55
12.1
23.06
Base
24.02
2.48
23.44
7.44
22.25
14.12
20.06
498
-------
The following guide lines were established.
1. AP in the turbulent zone in proportional to the approach velocity squared.
2. AP$ of upstream pressure appeared to decrease proportionally with the
reciprocal of upstream pressure squared.
3. The pressure drop of a combination of supports is not the sum of theAP
across each support.
4. The friction factor for the support in the turbulent zone becomes
constant and is a direct function of washcoat type. This friction
factor differs radically from established published friction factors
for example a 200 cell type II profile metal support 7.62 cm long,
at a Reynolds number 1860Q has a constant friction factor of 0.0312
for type 6 washcoat.
c START UP TRANSIENTS
The consideration of start up problems led to a series of experimental
runs and computer simulations. A typical computer simulation is shown in
Figure 5.
Under the mass flow conditions that are normal with gas turbines and
the relatively low flows on start up the phenomenon of kinetic limited support
surface temperature effects were studied.
This can be stated roughly as the effect that under inlet flow
conditions such that the catalysts is obtaining 50% conversion of the fuel
delivered during a transient condition then it can be shown that the surface
temperature approimates to twice the gas temperature.
This ball park rule is simply modelled and the controlling parameters
are found to. be approach velocity, inlet pressure and cell density.
Experimental runs under fuel-on conditions gave the following
results:
499
-------
Inlet 200° C 1 Bar, adiabatic fuelling to give 850 C
Time
Sec.
0
1.28
6.40
8.51
21.76
28.16
43.5
0
2.56
3.84
7.68
8.96
10.80
11.54
12.74
Catalyst Exit
T°C
200.0
227.0
540.0
675.3
849.4
851.2
851.1
200.3
295.9
520.4
635.0
749.4
847.2
852.5
852.6
Catalyst Surface
T°C
200.0
329.8
819.8
1100.4
851.3
851.2
851.3
200.1
421.4
852.5
1121.4
1290.4
913.4
854.4
852.6
This test was repeated nine times for each support. A range of ceramic
supports was also tested.
It was observed that the activity of the catalyst as measured by the light
off temperature on a separate rig decreased with each start up test, this
decrease was vastly more than the decrease observed after many hours of
constant running after a start up.
In addition it was observed that the ceramic supports exhibited a wide
range of thermal shock failures to the point where no ceramic support withstood
20 start ups. Based upon this it was decided not to reconsider ceramic systems.
500
-------
Cold Starting:
For a practical engine, the ability to start up under cold conditions
rapidly and reliably is essential. With the problem that catalyst light off
is in the region 200°C and above it is not possible on cranking to obtain
compressor air at this temperature.
A variety of systems were considered.
1. Start up fuel
The addition of small amounts of H2 to the delivered air enables the
catalyst to start up i.e. light off instantly. This however, requires a
large amount of additional hardware and also an additional expensive fuel.
2. Start up combustor
A small combustor designed to flame fire up the engine until the
engine is running and the catalyst system can take over. This however,
requires hardware and presents redundancy problems especially in terms of
fluid, flow control.
The system adopted consists of a pre-burner burning a small amount of
fuel up stream of the catalyst. The main flow air is then dumped into the
burner flow to quench the flame and pre-heat the feed air to the catalyst.
The main power fuel is then added and the heated pre-mixed fuel is then
passed through the catalyst as in Fig. 6.
E ENGINE TESTS
The Rover 65 BHP automotive and air-craft auxiliary generator engine
was selected as suitable for the engine tests of this system.
A typical data plot and emissions output from this engine was taken and
is shown in Table I.
The combustor shown in Fig. 7 was then constructed and the engine test
undertaken.
The standard test equipment on the Rover gas turbine was employed. The
load and speed were measured by a Heenan and Frou.de (r type dynamometer and
tachometer. Pressures were taken to record impeller tip and compressor
delivery pressure, water manometers and mercury manometer were used to take
air flow and pressure drop.
501
-------
At full power the mark one system exhibited problems with fuel preparation
but even so the NOx emissions were 0.44 ppm despite rather high CO figures.
A series of test result emission indices were then taken:
STANDARD ENGINE;
Equivalence Ratio
0.2319
Q_2658
0.327
0.380
0.225
CATALYST SYSTEM: 400 cell
0.35168
0.335
0.349
0.377
0.384
CATALYST SYSTEM: 200 cell
0.353
0.337
0.40
CO
gm/Kg Fuel
107.39
9.82
5.17
4.91
119.46
0.099
0.099
0.096
0.075
0.039
7.43
7.45
0.04
HC
gm/ICs Fuel
38.88
17.68
11.99
9.21
41.35
1.86
1.87
1.78
1.46
0.91
2.22
2.24
1.05
NOx
gm/Kg Fuel
2.68
2.58
3.76
4.18
2.26
0.032
0.033
0.031
0.029
0.029
0.030
0.033
0.056
1. The size of catalyst used was 101.6 mm diameter by 76.2 mm long,
2. The reference velocity to the catalyst 60 metres/sec.
3. Pressure drops 4.2$.
4. By-pass to main flow split for working fluid
68$ main 32$ by-pass
5. Pre burner delivery temperature 420°C.
502
-------
Fuel Variations;
Many variations of fuel specification have been run; the standard fuel
used is diesel oil but test runs have been carried out using wide cut
gasoline, "washed" number 6, Avcat and Avtur kerosines, high smoke point
kerosine, natural gas, propane and low BTU mixtures, including H2/CH^,
CO/H2, CO/GH- amd various nitrogen diluted mixtures.
No unanticipated problems occurred with these fuel changes and apart
from the problems of oinjecting the Number 6 oil and prevention of auto ignition
with the wide cut gasoline, the catalysts performed well and the only adjust-
ments required were to the preheater burner output.
Low BTU Fuels;
The throughput on diesel being in the range of 44 litres/hr, thus creates
difficulties due to the extreme bulk of low BTU fuel required to produce
the same output.
It was found very difficult to thermally combust some of the very low BTU
fuels with high nitrogen dilutant, however catalytic combustion of these fuels
readily took place and it would seem that for these fuels catalytic combustion
presents the best solution.
Combustor Exit Temperatures;
Analysis of required exit temperatures is complicated by variation of
by-pass air, however the range of metal and metalloid supports fabricated
for the test programme fell into two classes:
a) Type I, a low cost support capable of continuous running at
1200 K; the majority of test work h as been carried out using this system
which is adequate for the majority of industrial engines.
b) Type II, an expensive high technology support which has a continuous
running temperature of 1700 K. This support and material was developed to
provide the flexibility to power advanced technology engines.
1500 K exit temperature tests;
Two series of tests were run on the blowing rig: The first series used
the ultra high temperature support by itself with a stabilisec high temperature
platinum catalyst. The performance of the unit was as per the normal supported
catalyst but with the ability to run up to high exit temperatures without
503
-------
damage. This support was also found to be very tolerant to premix fuel
quality and suffered no damage when fed a fine spray instead of vaporized
fuel.
The second series of tests involved running a standard catalyst unit to
provide an exit gas of 1100°K continuous and adding further fuel before
passing through an extremely small high temperature support. In this mode
the high temperature support was essentially acting as a 'bluff body1 and
the text book phenomenon of hot body ignition could be observed. However,
this hot body ignition led to an interesting series of experiments.
504
-------
TABLE I. BASE LINE TEST ON ROVER GAS TURBINE
Base Line Test on Rover Gas Turbine
Test No.
RPM
Brake load
Impeller tip pressure - psig.
Compressor delivery pressure - psig
ii it ii _ aDB
Pressure drop in combustion chamber - cm Hg
Exhaust pressure - cm paraffin
Air venture depression - mm paraffin
Air intake temperature A
Compressor delivery temperature (Measured) R
Jet Pipe temperature
Time for 2000 car fuel o
Compressor delivery temperature (Calc.) K
Air flow rate Kg/s
Fuel flowrate Kg/s
Air/fuel ratio
Brake power kV
Specific fuel consumption Kg/kW h
Brake thermal efficiency
Combustion efficiency
Compression ratio
Turbine inlet temperature
Turbine inlet pressure (p.s.i.abs)
Expansion ratio P/./^c
Exhaust Gas Analysis
EC ppm
No ppm
NO ppm
COppm
CO, %
"z*
Barometric pressure 775.5'H'ii Hg
1234
N
P
P
P
p^
• ^
he5
"c
h5
T
T2
T":
t^
T2
M
F
f
* V
SFC
^_
« OrtinTi
** womu
A
T.
P4
E*
3000
10
3.4
26
41.0
15.5
5.3
24.6
293
433
628
230
433
.667
.00735
90.7/1
5.06
5.23
1.56
74.3
' 2.733
762
33.4
2.22
350
13
22
1450
2.5
18
3000
40
3.6
26.6
41.6
15.7
3.2
22.4
293
-
726
201
434
.637
.00841
75.7/1
20.23
1.497
5.45
86.2
2.773
882 '
33.9
2.25
180,
22
24
150
3
17.5
3000
70
3.7
27.9
42.9
14.4
2.0
22.0
293
808
167
434.1
.6311
JD101
62.5/1
35.4
1.027
7.7
89.3
2.86
985
35.8
23.8
150
33
42
97
3.7
16.7
3000
95.2
3.8
28.1
43.1
14.0
-2.3
20.5
293
4930
867
145
435.1
.609
.0117
52/1
48.15
0.8750
9.04
86.3
2.873
1065
36.2
2.42
140
44
58
112
4.5
15.7
5
3000
10
3.5
26
41.0
15.7
4.3
24.5
293
433
651
230
433
.666
.00735
90.6/1
5.06
5.23
1.56
74.2
2.733
762
33.3
2.23
360
17
18
1560
2.4
18,2
505
-------
10 =
1
8
7
5
4
1
1
AP ,
PA
1 '
8
7
a
'
t
1359
1.703
2393
. 2738
4OO CELL
-" . Ai~\r\ t-c\ i
-
^^lillBlB I jig
i:
Ii I 1 1
« i 4
l_ Y»MOn^\^T
BAR ABS.
H tf
NOT WASHCOA
NOT WASHCOA
= |1
^ ;:E;;;;;;::;;;: ; f ; -|-
tP*- i 1 tn »/ /
ra-wgr-l1 |^^^
illjjipEE
t / t P- i
, Jt1 ^
,. L r.v i
T i I
R' r '
' ' f '" "T""1
\
T
T
y
— J
/
!_L/ /.U^ tin !->•
ED 7.62 cm. L(
ED 254 cm. L(
— 1 — — — - - -- J — 1 Li f j ; i , i I
_ ._ /PTi Ifr ^
ri J/ ' ' 1
, n Lf i i[
J'? 2- ' 1
^ ::::::::::::::::::
T
» • » e t i a >
10
}NG
DNG /
p? tit t £ it? : Jff : * fff ip
rftS^TTJ PT -l£ J^Hf -1
i/ 1 ' i ! ' i i ' i ' : I ' '
/I i 1 1 [ ' ! ! i . ,'
j|l. i • j • , ' ' 'l • ;.
/ ! jll I ! >|i ,
1 j |!| !| ] \\ |'|
TTT ' l*"jL
g|| EE=;;;;!; illl
... (. at. [J
.Li t+3
' I'l •!]
T IE iii
jty in
* t • 7 • • 1
IO
VELOCITY 'm/s
Figure I, AP/PA vs. velocity for 400 cell washcoated support.
506
-------
4OO CELL
WASHCOATED 7.62 c^ LONG
4£ VERSUS PA
PA
FOR CONSTANT VALUES OF G*
3 4 I • 7 « • I
IOO
PRESSURE P (ABS.)'
7 I • I
IOOO
p.s. i '•',.
Figure 2. AP/PA vs. PA for 400 cell washcoated support.
507
-------
4OO CELL NOT WASHCOATED 7.62 cm LONG
Figure 3. AP_ vs. G for 400 cell washcoated support,
PA
508
-------
20O CELL WASHCOATED 7.62 cm. LONG
IOOO
•*
Figure 4. AP vs G for 200 cell support.
P7T
509
-------
CRTKIN - MULTIPLE RUN OUERLRY
TEMPERRTURE
JQOS
Solid
08/07x78
15:09(307
100 -]
en
O
Figure 5. Relative distance through catalyst.
-------
OUEA)CH HOLES
HOLES
A/ozz<.£.
ST«TIO*J
Figure 6. Schematic diagram of comubstion chamber.
-------
en
—i
ro
Figure 7. Experimental engine test setup.
-------
HIGH COMBUSTION EFFICIENCY AND LOW POLLUTANT EMISSION
BY CATALYTIC AND OTHER MEANS
By:
K. C. Salooja
Esso Research Centre
Abingdon, Oxfordshire, U.K.
513
-------
ABSTRACT
Much effort is being expended these days to maximize combustion
efficiency while ensuring low levels of pollutant emission. We have now
achieved these objectives using a staged combustion system where particular
attention was paid to the choice of burner, the design of the combustion
chamber, and the use of non-volatile, non-dischargeable, catalysts in
different combustion stages. Combustion of fuel oil was carried out under
almost stoichiometric conditions with negligible smoke and virtually no
CO, or unburned hydrocarbons, and with very low emission of NO and SO,.
515
-------
HIGH COMBUSTION EFFICIENCY AND LOW POLLUTANT EMISSION
BY CATALYTIC AND OTHER MEANS
INTRODUCTION
The growing concern about both energy conservation and atmospheric
pollution requires that combustion systems should be opreated with maximum
efficiency and with pollutant emissions controlled to within environ-
*
mentally acceptable limits. The two goals, however, are not readily
compatible since any reduction in excess air to promote combustion
efficiency increases smoke, carbon monoxide, unburned hydrocarbons, and
often also oxides of nitrogen.
We have now developed a system where both objectives can be achieved.
The system, basically, is a staged combustion set-up where particular
attention has been given to the choice of the burner, the design of the
combustion chamber and the use of'non-volatile, non-dischargeable,
catalysts within the combustion chamber.
METHOD
Preliminary studies were carried out in a combustion rig of con-
ventional design using a Hamworthy Mark 6A medium pressure burner of 1 M
BTU/hr rating firing into a refractory lined water-cooled combustion
chamber, 24 in. I.D, and 48 in. long. This system produced such large
amounts of carbon even under stoichiometric conditions that for staged
combustion a ,high intensity Hygrotherm R 10 burner was used. A new
combustion chamber, which is described later, had to be designed to
further greatly minimize production of undesirable products.
517
-------
Temperatures of fuel and air to the burner, of cooling water inlet
4
and outlet, and of the flue gas were monitored. Pressures of the fuel and
air were also determined. The fuel firing rate was checked by periodically
monitoring the weight of the feed drum.
Flue gas was analyzed continuously for Op, CO, COp, H-, unburned
hydrocarbons (HC), NO and smoke. For analysis of 02, CO, CO^ and H2 the
gas drawn from the stack was filtered and passed through a refrigerator
and then through a column of dry silica gel to remove particulate matter
and moisture. CO and C0_ were analyzed by non-dispersive infra-red
analyzers (Maihak 'Unor 21), Op by a paramagnetic analyzer (Servomex
'OA 137'), and H_ by a thermal conductivity analyzer (Hartmann and
Braun 'Caldos 2').
HC were anlayzed by a flame ionization detector (manufactured by
Curzon Technical Supplies), and smoke by a von Brand analyzer.
NO were analyzed by a chemiluminescent analyzer (Thermo Electron
Model 10A). Flue gas sample was drawn through a quartz tube, which was
frequently replaced since the deposits that gradually build up affected
the readings.
A sensitive technique was developed for SO, analysis. This has been
described elsewhere (Reference 1 ) .
All the gas analyzers were regularly calibrated with standard gas
samples.
Tests were carried out using fuel oil 2 and^LS. Both these were
I
doped with pyridine to increase their nitrogen content as high as 0.76$ to
verify how well the system could curb NO even under such exceptional
condition.
518
-------
RESULTS AND DISCUSSION
Pollutant control becomes increasingly difficult when, in the interest
of energy conservation, combustion is carried out closer to the stoichio-
metric condition. Of the various undesirable products, CO, HC and H2 are
not so difficult to control as solid carbon and NO (and also SO,, when
sulphur is present in the fuel). Our efforts at the outset were therefore
largely directed towards the control of carbon and NO . Amongst the
various NO control techniques hitherto developed staged combustion is
considered highly satisfactory since it alone, can control both thermal and
fuel NO^ (Reference 2). However, even with this technique attempts to
reduce NO substantially present serious problems with the control of
carbon. This is because the effectiveness of NO control mainly depends
on the degree of fuel-richness of the first stage, but the richer the
first stage, the greater is the formation of carbon and the other products
of incomplete combustion. Whilst the gaseous combustible products can be
effectively burned in the second stage, carbon is notoriously difficult to
burn within a practical system. Apart from the combustion difficulties,
excessive carbon formation can seriously interfere with additional control
measures. The success of the system, therefore, largely depends on how
well carbon formation is curbed in the first stage.
The results in Figure 1 typify the severity of carbon formation with
a conventional burner even under normal single stage condition. We,
therefore, tested a variety of burners, and found that the Hygrotherm
burner - which recycles a considerable proportion of the flame products -
produces markedly less carbon and also, unlike several other burners,
produce less NO^ (Figure 1).
Although the 24 in. I.D. combustion chamber used in the above experi-
ment was adequate for the conventional burner, it was far too wide for the
Hygrotherm flame, barely 4 to 7 in. in diameter. One could not therefore
attempt a further reduction in carbon and NO by further promoting mixing
and recriculation of products in the main body of the flame and/or provide
519
-------
a convenient means of treating the combustion medium with catalysts.
Studies were, therefore, carried out with narrow combustion chambers.
These proved far more amenable to the control of all the undesirable
products. The influence of the chamber length was next investigated.
Attempts were then made to increase recirculation of combustion gases by
inserting baffles in the chamber. Next design studies on the second stage
burner (air injector) were carried out. Finally the application of
catalysts was investigated. The various developments, together, helped to
produce a system which could operate with maximum combustion efficiency
(without any excess air) and with exceptionally low pollutants.
INFLUENCE OF COMBUSTION CHAMBER DIAMETER
Comparative performance of Hygrotherm burner in 24 in. I.D. chamber
and an 8 in. I.D. chamber, both of the same length, is shown in Figure 2.
In the narrower chamber carbon formation was far less, although more NO
was formed. These effects are, obviously, due to improved mixing and
interaction of the flame products, and higher wall temperatures.
Attempts to use a still narrower chamber (5 in. I.D.) showed that now
impracticably long length was necessary to effect complete combustion. So
most of the further development was carried out with the 8 in. I.D.
chamber.
INFLUENCE OF CHAMBER LENGTH
Carbon forming tendency was reduced as the chamber length was increased
from 50 in. to 75 in., and then to 90 in.; although NO formation was
slightly increased. The magnitude of both these effects, however, was far
less than with changing the chamber diameter.
520
-------
fuel oil, the combined effect of reduced chamber diameter
(8 in. I.D.) and increased length (90 in.) was such that virtually smoke-
free combustion could be carried out with barely 56% of the stoichiometric
air. The system, therefore, is capable of generating reducing atmospheres
of a quality such as hitherto been achieved mainly with gaseous fuels.
INFLUENCE OP BAFFLES
Baffles were introduced into the combustion chamber to
• explore if carbon formation can be further reduced and the more
readily combustible CO and H_ increased,
• find if their use can help reduce chamber length,
• check what effects can arise on the insertion of perforated
refractory blocks as catalyst support.
Three refractory baffles, each with a 2.5 in. hole in the centre,
were spaced in the combustion chamber at 16 in., 30 in. and 48 in. from
the burner. Their presence, as shown in Figure 3, considerably reduced
carbon formation, increased CO and H_ and reduced COp. NO was not much
affected; its concentration, with air 50-60$ of the stoichiometric, was
less than 3 ppm. The baffles entailed only slight - ca 1 in. w.g. -
pressure drop.
It is thus clear that the insertion of baffles, and hence also of
catalyst supports of similar design, is mainly beneficial.
521
-------
SECOND STAGE BURNER
The principal requirement of the second stage burner - air injector -
is that it should cause the air and the hot products from the first stage
to mix well quickly in a short space. Several air injectors were designed,
but their performance showed little difference. Par greater effect,
mainly on carbon formation, was that of its distance from the first
burner. The closer it was to the upstream burner, the less the carbon
formed. The effect on NO , CO, HC and Hg was much less.
VARIATIONS IN PRIMARY TO SECONDARY AIR RATIO
Whereas according to previous studies NOx reduction increased as the
proportion of the primary air is reduced, (Reference 2), our studies show
that the reduction reaches a maximum when the primary air is reduced to
ca 60% of the stoichiometric. A further reduction in primary air to
ca 50/6 of the stoichiometric increased rather than reduced NO. This
reversal in effect is most probably caused by less complete degradation of
the organic nitrogen species in the first stage, and their subsequent
oxidation in the second stage.
INFLUENCE OF FUEL-NITROGEN CONTENT
The influence of increasing the nitrogen content of fuel,' by doping
it with pyridine, was studied with the medium pressure burner firing into
the 24 in. I.D. combustion chamber, and also with the Hygrotherm burner
firing into the 8 in. I.D. chamber. The results showed that Whilst NO
increases with increasing nitrogen content, the increase is not directly
proportional to the nitrogen content. The nature of the combustion system
522
-------
and the air to fuel ratio also affect it. The latter factor has a par-
ticularly pronounced effect under sub-stoichiometric conditions. Here the
more fuel rich the combustion, the smaller is the conversion factor. This
is understandable since the more fuel rich the combustion the more intense
are the reducing conditions, and hence the less the NO that can form or
survive reduction.
INFLUENCE OF FUEL FIRING RATE
It has been previously established that a decrease in fuel firing
rate reduces NO (Reference 3, 4). To verify the extent of the effect
under our experimental conditions, the influence of changing the firing
rate from 2 to 3 gph was examined. Under excess air firing conditions
overall, and with the first stage operating with ca 80% of the stoichio-
metric air, NO indeed decreased as the firing rate was reduced* However,
under stoichiometric firing conditions overall, and the first stage
operating with considerably less than 80% of the stoichiometric air,
increased firing rate produced less, rather than more, NO. There was
nothing else unusual about the combustion process: higher firing rate
produced more CO, HC and H-, and the variation in the proportion of COp to
CO followed the expected pattern. Obviously, the higher firing rate
produces more severe reducing conditions and this gives rise to less
NO.
INFLUENCE OF CATALYSTS
We have previously shown how NO formation can be markedly reduced by
placing catalysts in the flame (Reference 5)« In a staged combustion
system, the two flames, as well as the interstage zone of a highly reducing
523
-------
nature, should all be readily amenable to HO destruction by catalysts.
This possibility was explored with following results.
Our previous study, with several catalysts, has shown that Fe/Cr is a
highly effective catalyst. This was again found to be the case in the
present study and the results presented here are mainly with this catalyst.
The catalyst was supported on alumina/silica refractory capable of with-
standing temperatures up to 1650°C. For its use in the interstage zone,
the refractory block was 3 in. thick and had numerous 1/2 in. diameter
channels across it. Catalysts were applied to the block by coating it
with an aqueous solution of the nitrates of the metals and then firing it
till the nitrates decomposed.
Catalysts in the Interstage Zone
The catalyst block was placed in turn at 40 in., 52 in. and 66 in.
from the first stage burner at positions where the temperature with fuel
oil^5, firing at 2.5 gph rate, and air 15% of the stoichiometric, was
respectively 1260°, 1180° and 1050°. The catalyst reduced NO, the effect
being greater the more fuel rich the combustion (curves 1 and 2, Figure
4). With the catalyst 52 in. from the burner and air 15% of the stoichio-
metric NO was reduced from 300 pm to 130 ppm (57$ reduction). The NO
reduction was less (46$) with the catalyst 40 in. from the burner, but
more (64$) when it was 66 in. from the burner. Whilst NO was thus markedly
reduced by the catalyst, carbon formation was somewhat increased. However,
no difficultly was experienced in curbing this in the second stage.
The second stage opeation (with! the air injector 3 in. downstream of
the catalyst) maintained the marked reduction effected by the interstage
catalyst. With the primary air 73# of the stoichiometric, and the overall
air input stoichiometric, NO emission was 140 ppm (curve 4, Figure 4).
Without the interstage catalyst, staged combustion under these conditions
produced 380 ppm NO (curve 3, Figure 4).
524
-------
Catalysts in the Second Stage Flame
An Fe/Cr treated refractory block, similar to the interstage catalyst,
but only half its thickness (1.5 in.)» was placed ca 3 in. downstream of
the air injector. This brought about a further reduction in NO (curve 5,
Figure 4). With the primary air 73# of the stoichiometric and the overall
air input stoichiometric, the second stage catalyst reduced NO from 140
ppm to 120 ppm.
The effect of using the second stage catalyst in the form of an
elonaged ellipsoid, ca 6 in. and 2 in. in axes, supported ca 2 in. from
»
the burner along its central axis was also studied. This had virtually
the same effect as the perforated block catalyst.
Catalyst in the First Stage Flame
The use of the catalyst in the first stage, by mere treatment of the
burner throat wherefrom the flame emerges into the combustion chamber,
brought about a further substantial reduction in NO. The NO concentration
in the flue gas, with the system operating with stoichiometric air overall,
was now 95 ppm.
The first stage catalyst did cause a slight increase in smoke, but
its emission in the stack gas was still fairly low - below 3 Bacharach
number. CO, HC and H_ were hardly detectable.
SO- CONTROL
SO, formation from fuels containing sulphur is, of course, highly
undesirable. The factors which are responsible for NO formation also
525
-------
promote SO, formation. Hence the techniques developed for NO^ control
should also help reduce SO,. In our studies with fuel oil^t5 containing
2.5$ sulphur, SO, emission from the conventional system, operating with
2.1$ excess oxygen was 44 ppm. In the staged combustion system, at the
same firing rate and with the same excess air, SO, emission was 26 ppm.
The use of catalysts reduced this to 18 ppm. Since the staged set-up,
unlike the conventional system, could be operated smoke-free under almost
stoichiometric condition - and SO, formation is strongly curbed by curtail-
ment of excess air - SO- was reduced to only 5 ppm.
CONCLUSIONS
This study demonstrates a combustion system which can be operated
with maximum efficiency (with no excess air) and with very low emission of
pollutants. Fuels which, under conventional combustion conditions,
generate ca 950 ppm of NO, now give rise to only 95 ppm. Moreover,
despite the fact that the system is operating without any excess air, CO,
HC and H. concentrations are negligible and that of smoke is below the
visible level. SO, formation, when sulphur is present in the fuel, has
also been greatly reduced. Whilst conventional combustion conditions
generate 44 ppm of SO,, the new system produces barely 5 ppm.
This study has included examination of fuels with very high nitrogen
content (0.76$). Normal fuels, with nitrogen concentrations ranging from
0.1 to 0.4$ produce less NO and with these the present system could give
rise to very low NO emission. One NO control technique which has been
studied in the past but not incorporated in this investigation - interstage
cooling (Reference 2) - could, if needed, help to reduce NO even further.
The staged combustion system adopted has been developed to generate
very low concentrations of carbon. This capability is reflected by the
fact that fuel oil*#*2 can be fired virtually smoke-free with barely 56$
526
-------
of the stoichiometric air. In conventional systems, in contrast,
smoke-free operation would require air considerably in excess of the
stoichiometric.
The exceptionally low emissions of NO and SO, that have now been
achieved arise primarily through the use of catalyst in the different
zones of staged combustion. The effective catalyst (Fe/Cr) as well as the
catalyst support (AlpO,/Si02 refractory) are inexpensive materials which
can well withstand the furnace temperatures involved in staged combustion,
and they do not give rise to secondary pollution problem through loss from
the system.
REFERENCES
1. 'Salooja, K.C., "Some serious errors in SO, determination in flue
gases", Energy World, March 1974, 10.
2. Siegmund, C.W. and Turner, D.W., "Influence of combustion modification
and fuel nitrogen content on nitrogen oxides emissions from fuel oil
combustion , Combustion, August 1972, p.21; "NO emissions from
industrial boilers: potential control methods", Combustion, October
1973.
3. Bartok, W., "Status of EPA-sponsored studies of stationary NO
control at Esso Research & Engineering Co.", ASMS Meeting, November
28-December 2, 1971, Washington D.C.
4* Blakeslee, C.E. and Burbach, H.E., "Controlling NO emissions from
steam generators", J. Air Poll. Control Asaoc. 23: 37 (1973).
5. Salooja, K.C., "Combustion control by novel catalyst means", Nature,
240: 350 (1972). '
527
-------
250
(FUEL OIL #5 (N, 0.15%); FUEL FIRING RATE 2.0 gph)
/
MEDIUM PRESSURE ATOMISING BURNER
HIGH RECIRCULATION RATE BURNER
200
150
10
o
111
IT
*
O
- 100
£
50
_L
J_
0.7 0.8 0.9 1.0 I.I
ACTUAL/STOICHIOMETRIC AIR
1.2
1.3
o
CD
u
Figure 1. Smoke and NO emission performance of a high
recirculation rate burner and a medium pressure
atomising burner in a conventional, unstaged
system.
528
-------
(FUEL OIL #5 IN, 0-15%) : FIRING RATE 2-Ogph)
— COMBUSTION CHAMBER 8" I.D.
• -COMBUSTION CHAMBER 24" I.D.
\
250
5?
fO
Q
UJ
a
-------
15
14
13
12
CM
X
10
CJ
8 9
8
(FUEL OIL # 2 (N, 0-01%): FIRING RATE 2-5gph)-
I - WITHOUT BAFFLES
2- WITH BAFFLES
6
X
o
o
d
z
ui
*
O
0.5 0.6 0.7
ACTUAL/STOICHIOMETR 1C AIR
Figure 3. Influence of baffles in combustion chamber on
smoke, CO, C02 and H2-
530
-------
900
800
(O
0600
o
UJ
5500
LU
-------
APPENDIX A
Summaries
First and Second Workshops on Catalytic Combustion
A-l
-------
SUMMARY
SECOND WORKSHOP ON CATALYTIC COMBUSTION
The Plantation Inn
Raleigh, North Carolina
June 21-22, 1977
Sponsored by the U.S. Environmental Protection Agency
Summary prepared by J. P. Kesselring, Acurex Corporation/Aerotherm Division
The Second Workshop on Catalytic Combustion was held in Raleigh, North
Carolina on June 21-22, 1977. Forty-three people, representing various
governmental, industrial, and academic organizations, attended the workshop;
a list of attendees is given at the end of this summary. The purpose of the
workshop, sponsored by the U.S. Environmental Protection Agency and organized
by the Aerotherm Division of Acurex Corporation, was to provide an overall
summary of the current state of the art of catalytic combustion. This was
accomplished by bringing catalytic combustion research people together for
the exchange of results obtained since the first workshop was held in May
1976.
The meeting began on June 21 with introductory remarks by Mr. G. B.
Martin of the Environmental Protection Agency. Mr. Martin emphasized the
potential role of catalytic combustion in the control of NOX emissions. For
conventional combustion systems, significant amounts of thermal NOX are
produced at temperatures above 2800 °F, and large fuel NOX emissions are
found in lean systems. Analytical and very limited experimental results are
beginning to show that catalytic combustion systems may be able to minimize
NOX emissions in systems that operate above 2800 °F and use fuels which
contain large amounts of bound nitrogen. The importance of system design on
the combustor approach used must also be considered. As an example, firetube
or watertube boilers operate under different conditions than gas turbines,
and knowledge of the operating regime for the system is required to design
the combustor. Mr. Martin then introduced Mr. Dale Johnson of the Institute
of Gas Technology.
A-3
-------
Institute of Gas Technology Programs
Mr. Johnson of I6T described work carried out at the Institute of Gas
Technology on ventless appliances and catalytic igniters. During 1972 to
1976, under the sponsorship of Southern California Gas Company of Los
Angeles, IGT invented and developed several conceptual burners and model
ventless .appliances for catalytic combustion of hydrogen and steam-reformed
natural gas. During 1975, the U.S. Environmental Protection Agency and
Southern California Gas Company cofunded a program at IGT to develop a
catalytic range-top burner. These model catalytic appliances operate at low
combustion temperatures and produce very low emissions of nitrogen oxides
(NOX < 8.0 x 10~4 lb/106 Btu). Because of the low emission levels, outside
venting of the products of combustion is not required.
Since 1974, IGT has been working on the development of an
instantaneous ignition system for gas appliances using hydrogen fuel ignited
in air with a platinum catalyst. The hydrogen is stored in the form 'of a
metal hydride, with small quantities of hydrogen released by valving when
ignition is required., This catalytic ignition system could become an
alternate to standing pilot or electric systems.
Westinghouse Program
Mr. S. M. DeCorso of Westinghouse described efforts conducted by
Westinghouse to assess the applicability of catalytic combustors for
stationary gas turbines. In order to achieve a significant reduction in NOX
emissions, new combustor technology is needed. Westinghouse has been working
in four areas of gas turbine combustion to develop this technology: (1)
conventional combustors, (2) premixed, (3) catalytic, and (4) new wall
cooling techniques.
The catalytic combustor approach has been identified as one of the
major ways to solve the emissions problem. Experimental test results with
No. 2 distillate oil and low Btu gas show much lower emissions than the
conventional combustor, with NOX emissions <1 ppm. Pattern factors measured
1 foot downstream of the catalytic combustor exit were excellent, allowing
the air requirement normally reserved for the pattern factor to be
distributed elsewhere.
Mr. DeCorso presented catalytic combustor development situations for
stationary gas turbines based on a program approach emphasizing integration
of system design thoughts throughout the program. Application of these
combustors would be in both retrofit and new systems, leading to the
combustion of coal gas, waste gas, coal liquids, and bound nitrogen fuels
with minimal NOX emissions.
Based on work done at Westinghouse, the catalytic combustor has been
shown to have the characteristics required to help solve problems of fuel
conservation, emission requirements, and use of new fuels. Development of
the catalytic combustor in a total system design is now needed while research
efforts on the catalytic components continue.
A-4
-------
U.S. Air Force Programs
The U.S. Air Force is sponsoring programs in catalytic combustion
through the Aero Propulsion Laboratory and the Air Force Office of Scientific
Research. The Aero Propulsion Laboratory programs deal with the application
of catalytic combustors to aircraft gas turbine engines and afterburners.
The Air Force Office of Scientific Research programs are concerned with
catalytic combustion modeling and the initiation of combustion on catalytic
surfaces.
Aero Propulsion Laboratory Program
Captain Thomas J. Rosfjord of the AFAPL described the AFAPL in-house
program for catalytic combustion application to aircraft systems. The
program addresses the use of catalysts in both mainburner and afterburner
applications. Mainburner rig testing will cover the following range of
conditions:
o 2 < p < 18 atm
o 400 < T1nlet < 800 K
o 20 < Vref < 40 m/s
o 0.1 < $ < 0.4
Afterburner tests will be focused on the ability of a catalytic combustor to
convert fuel energy in an efficient and stable manner.
A honeycomb flameholder consisting of silicon nitride monolith
segments coated with Pt/Pd catalyst has been fabricated and will be tested in
July. Inlet temperature for this system will be 1300 °F, with the surface
temperature between 2500 and 3000 °F.
Dr. Henry Shaw of Exxon Research and Engineering discussed work to
develop a hybrid catalytic combustor for aircraft turbine applications. This
material is discussed in detail in Air Force Aero Propulsion Laboratory
Report AFAPL-TR-76-80. The hybrid catalytic combustor minimizes pollution
problems associated with unburned hydrocarbons and carbon monoxide in the
idle mode, and NOX and smoke production in the power mode of aircraft gas
turbine operation. This combustor consists of a fuel-rich thermal
precombustor, secondary air quenching zone, and monolithic catalyst stage
which rapidly oxidizes CO and UHC produced in the precombustor.
Noble metal catalysts on various monolithic support materials and
geometries were found to be the most active materials for CO and UHC -
oxidation in the temperature range of 700 to 1200 K. the hybrid catalytic
combustor combustion efficiency for JP-4 fuel containing 535 ppm sulfur was
found to be 99.8 percent under realistic conditions. Combustor pressure drop
was less than 6 percent. For a Johnson Matthey metal-supported Pt catalyst,
average emission indices of CO, UHC, and NOX were 0.95, 0.43, and 1.8 g/kg of
fuel, respectively. This catalyst was effective in reducing CO,by 86 percent
A-5
-------
and UHC by 94 percent, while increasing NOX by 68 percent relative to
catalyst inlet values. It was estimated that the hybrid catalytic combustor
can meet the 1979 new aircraft emission standards, but must be modified
slightly to reduce UHC emissions to meet the 1981 new aircraft emission
standards.
Gerald Roffe of General Applied Science Laboratories, Inc., then
described work conducted at GASL to develop a fuel preparation system for the
catalytic combustor, which would provide uniform velocity and fuel
distribution profiles, completely vaporize the fuel, and have a reasonable
pressure drop. Operating conditions for the system include flow velocities
of 50 to 125 ft/sec, pressures of 100 to 200 psia, temperatures of 700 to 1000
°F, and fuel air ratios of 0.018 to 0.028. The approach taken in developing
the fuel preparation system was to design for a limited residence time to
prevent autoignition, to provide for adequate exit blockage to prevent
flashback, to produce small droplets to get good evaporation, to use a high
flow "velocity to preclude flame stabilization prior to entering the
combustor, and to obtain the best possible initial dispersion of the fuel to
enhance mixing. Three candidate designs were tested, consisting of pressure
atomization, air blast atomization, and air assist atomization systems.
Following testing of these systems and some modifications, an air assist
atomization system with an upstream swirl generator was selected as the final
design.
Captain Rosfjord then summarized some areas in need of investigation.
One obvious area is the physical, chemical, and thermal endurance of a
catalytic combustion system. The investigation of an all-catalytic system,
rather than a hybrid system, for low power application also needs to be
addressed.
Air Force Office of Scientific Research Programs
Dr. B. T. Wolfson of AFOSR then briefly discussed the AFOSR interests
in catalytic combustion for air-breathing propulsion systems. Two programs
are currently being funded by the AFOSR in catalytic combustion. Dr. F. B.
Bracco of Princeton University is performing a theoretical-experimental study
of the initiation of combustion on catalytic surfaces, and Dr. C. M. Ablow of
Stanford Research Institute is developing a theoretical model for catalytic
combustion. In addition to these efforts, advanced diagnostic techniques and
other fundamental studies are of interest to the AFOSR.
Joint National Science Foundation/Environmental Protection Agency/Exxon
Program ~~
Dr. Anthony E. Cerkanowicz of Exxon Research and Engineering described
this jointly-funded effort to develop a practical model of catalytic
combustor operating characteristics for analysis of advanced power systems
and data analysis. The major physical assumptions of the steady-state model
are:
A-6
-------
o Uniform gas-phase properties at a cross section
o Catalyst/substrate temperature and fuel concentration are uniform
at a cross section
. o Conversion of reactants to products can occur at the catalyst
surface and in the gas phase
o Axial variations in velocity are allowed
o Axial heat conduction in the substrate is included, but radiation
and gas-phase conduction are neglected
The model consists of a variable-order finite-difference method to
solve the two-point boundary value problem. Comparisons of catalytic
combustor data and model predictions show excellent agreement. When the
model is used to examine conversion as a function of gas inlet temperature
for typical lean operating conditions with a noble metal catalyst, a sharp
catalytic lightoff temperature is predicted, as is a lightoff/extinction.
hysteresis. At higher temperatures the onset of gas-phase combustion occurs,
resulting in complete conversion.
Further work on expanding the model to include internal heat removal,
CO kinetics, multiple fuel species, and NOX kinetics is planned.
U.S. Environmental Protection Agency Program
Mr. G. Blair Martin of the EPA briefly described the goals of the
catalytic combustor application to stationary combustion systems. In
stationary systems, it is desirable to operate at high system temperatures,
at low overall excess air, and at minimum flue gas temperature. It is also
desirable to be able to burn fuels containing bound nitrogen without
producing excessive NOX emissions. These are among the areas being addressed
in the EPA program. 1
Acurex Corporatlon/Aerotherm Division Program
*
Under EPA sponsorship, the Aerotherm Division of Acurex Corporation is
conducting a study to determine design criteria for catalytic combustors with
application to stationary sources. Dr. John P. Kesselring of Aerotherm
described the program effort aimed at identifying concepts for new combustion
systems capable of extremely low emissions performance at the highest system
efficiency achievable.
The development of catalytic combustors for systems such as industrial
and commercial boilers depends on establishing design criteria for the
combustion system. These design criteria include information on the
catalyst/washcoat/substrate, the catalyst life, the preheat requirement for
combustion air, and the maximum flow velocity and bed temperature that can be
achieved in the system. Additional design criteria include the amount of
A-7
-------
heat extracted from the combustion system for a given extraction mode and
catalytic combustor geometry.
The program began with a review of available catalyst materials
(precious metals and metal oxides, washcoats, and substrate materials). This
review was interfaced with theoretical calculations for heat and mass
transfer in a catalyst bed to determine the materials required for the
catalyst systems. Catalyst systems were then prepared in-house and obtained
from outside suppliers. Prior to testing, all catalyst systems were
characterized by total and active (precious metal) surface area measurements.
The catalysts were then instrumented with in-depth thermocouples to provide
bed temperature distribution measurements. One-atmosphere screening tests
were then conducted with up to four fuels (natural gas, propane, indolene,
and methanol) for a test time sufficiently long to determine the catalyst
performance after the initial rapid degradation was completed and a constant
performance level achieved.
A total of thirty-four screening tests were run on a variety of
monolith materials, washcoats, and catalysts. Based upon the results of the
screening tests, it was found that:
o Alumina was the best high-temperature (2700 °F) monolith
o a-alumina or presintered y-alumina washcoat enhanced long term
combustion performance at low preheat, compared to y-alumina
washcoats
o H?S treatment of Pt helped stabilize the catalyst at high
temperature
o Increasing Pt loading gave significantly greater activity
o Large cell monoliths passed many UHC, but operated at extremely
high mass throughputs
o Small cell monoliths provided excellent cleanup of pollutants
Based upon the screening test results, the graded cell catalyst system
was developed and tested. This system used a large-cell monolith segment at
the front end, a small-cell segment at the back end, and one or more
intermediate-sized segments in the middle. Graded cell catalysts have been
operated at maximum bed temperatures of 2735 °F, space velocities of 413,000
(ft3/hr)gas/ft3catalyst, and for tests on individual catalysts of up to 78
hours duration. Limited test data indicates less conversion of NH3 dopant (a
NOX precursor) to NOX at low pressure and high velocity under lean conditions
than is found in conventional systems.
The first catalytic combustion system to be tested incorporating
direct bed cooling used catalyst-coated cylinders surrounded by water tubes.
Surface energy removal was accomplished by radiation from the catalyst to the
water tube, allowing the catalyst surface to operate at less than 2000 °F
temperature, even under stoichiometric conditions. The water tubes remove
approximately 25 percent of the available energy in the fuel. For boiler
A-8
-------
application, the radiative catalytic/water tube system would be followed by
a graded cell catalyst to complete combustion and achieve the goal of
extremely low emissions and high system efficiency. Further testing of
graded cell catalysts and combustion systems is ongoing, with prototype
system design to be completed by the end of the year.
Johnson Mattney Research Center Program
Dr. Gary Acres of the Johnson Matthey Research Center described
platinum metal crystallite growth at elevated temperature. Test results have
shown that catalytic activity is not proportional to crystallite size or
metal area, but that the lightoff temperature is a function of the metal
crystallite size. Lightoff is a kinetically controlled phenomenon, whereas
high conversion of reactants is diffusion controlled. Increasing metal
loading, or using promoters, helps increase the kinetic region activity.
The parameters which affect metal sintering are:
o Temperature, with high-temperature catalytic combustion
applications always increasing sintering
o Catalyst environment, with inert atmospheres showing slow
sintering and oxygen-combustion gas atmospheres exhibiting rapid
sintering
The catalyst may move about on the support by vapor transport in the
gas phase, by a surface migration of metal atoms from one crystallite to
another, by the combining of crystallites, or by metal-ceramic support
interactions. Dr. Acres presented a movie of platinum on silica and
platinum/rhodium sintering, giving visual confirmation of the migration and
sintering of platinum.
United Technologies Research Center Program
The UTRC catalytic combustion program was described by Dr. Pierre
Martenay. The program seeks to determine the feasibility of utilizing
catalyzed surface reactors in the combustion of muHi component fuels. , The
program is currently involved in simple burner experiments using propane fuel
and in an analytical study combining heat and mass transfer and homogeneous
reactions in the analysis.
The,experimental program uses a catalytic burner apparatus to
determine ignition temperature, steady-state operating conditions, and
species concentrations by varying mixture ratio, temperature, bed length,
flow velocity, bed material, and diluent. The combustible mixture is usually
oxygen-propane with argon diluent, but helium and nitrogen diluent have also
been used. Isothermal experiments are then run to determine ignition
temperature.
The modeling program takes into account both heterogeneous and
homogeneous reactions, and seeks to fit experimental data over a range of
A-9
-------
temperatures with single heterogeneous and homogeneous rate constants.
Matching of temperature rise and concentration data has been done for propane
combustion with argon diluent.
NASA-Lewis Research Center Programs
Dr. David N. Anderson began the meeting on June 22nd with a brief
review of NASA's interests in catalytic combustion. The Air Breathing
Engines Division of the Emission Technology Branch is pursuing the
application of catalytic combustors to aircraft gas turbines, while the Power
Generation and"Storage Division of the Combustion Power Section is interested
in catalytic combustors for automotive gas turbines.
Dr. Anderson then described the catalytic combustion automotive gas
turbine program in more detail. The proposed gas turbine operating
conditions are:
o Pressure 1.5 to 4.5 atm
o Exit temperature 1310 °K
o Primary zone temperature 1350 to 1425 °K
o Reference velocity 11.4 to 12.9 m/sec
o Airflow 0.1 to 0.5 kg/sec
o Inlet temperature 1210 to 970 °K ,
The NASA test rig can duplicate all these conditions except the inlet
temperature, which is limited to 800 °K. The goals are to limit emissions
from the combustor to half of those required by the most stringent standards,
and to keep the total combustor system (fuel preparation plus combustion
chamber) pressure drop under 3 percent.
For the fuel preparation system, the program goals are:
r\.
o Spatial fuel distribution within 10 percent of mean
o 90 percent of fuel vaporized at 800 °K
o Velocity distribution within 10 percent of mean
o No auto ignition
o Less than 1 percent pressure drop
Four different fuel injectors were tested; (1) air assist sonicore
injector, (2) splash-groove injector, (3) multiple-jet injector, and (4)
multiple conical tube injector. Air swirlers were used with the sonicore and
A-10
-------
splash-groove injectors to improve spatial fuel-air distributions. The
multiple conical tube fuel injector was generally able to meet the program
goals if sufficient mixing length was allowed.
In the catalyst evaluation program, the objectives are:
o Emissions
- 1.6 g N02/kg fuel
- 13.6 g CO/kg fuel
- 1.64 g HC/kg fuel
o Pressure drop
~ Less than 2 percent
i
The combustion test rig used has a long mixing section, and the reactor can
hold from one to six individual catalyst elements in series. Each element is
located between thermocouple arrays. All tests were performed at an inlet
temperature of 800°K, pressure,of 3 x 1CP Pa, and a range of velocities from
10 to 25 m/sec. Adiabatic reaction temperatures from 1100 to 1600°K were
obtained by varying fuel-air ratio.
Catalyst elements were obtained from Engelhard Industries, W. R.
Grace and Co., Johnson Matthey Corporation, and Oxy-Catalyst, Inc. All
elements were 12 cm in diameter. The Johnson Matthey elements used a metal
substrate; all other elements used a ceramic substrate. Emissions
measurements were then made in combustion tests to determine the minimum exit
temperature at which the reactors should be operated to obtain the steady-
state emission objectives. Effects of reactor length, cell density, and gas
phase reactions were considered.
The feasibility of using a catalytic reactor in an automotive gas
turbine engine and meeting both emissions and pressure drop goals was
demonstrated in this program. Potential problems for such a system include
the loss of catalytic activity with time and transient operation
characteristics.
In addition to the automotive work, NASA is pursuing aircraft gas
turbine engine emission reduction programs using catalytic combustors. Dr.
Edward J. Mularz described this work. The purpose of the experimental
program being conducted by General Electric is to evolve jet aircraft engine
combustion technology and reduce low-power CO and HC emissions to extremely
low levels. Combustor concepts are tested in a 60-degree sector configuration
sized to engines of large commercial aircraft. Three design concepts will be
screened over a limited range of operating conditions. The three designs
include a hot wall combustor, a recuperative combustor, and a catalytic
combustor.
A new program, the advanced low emissions catalytic combustor program,
will be starting shortly. The purpose of this program is to evaluate the
A-ll
-------
feasibility of employing catalytic combustion technology in the aircraft gas
turbine engine field to achieve the control of NOX emissions for subsonic,
stratospheric cruise operation, while retaining or improving system
performance. This program is jointly funded by NASA and the Air Force, and
will involve three phases: Phase I is a design study, Phase II involves
further design with combustion testing, and Phase III will involve full
annular or sector combustion testing. This program will begin in October
1977.
Engelhard Industries Program
In addition to in-house research, Engelhard Industries has active
externally supported programs in catalytic combustion. Engelhard is a
subcontractor to General Electric on the NASA-sponsored low emissions
aircraft combustor program, and has also completed a program on catalyst
selection and life test durability in No. 2 diesel oil combustion for NASA
under ERDA funding.
Dr. Robert V. Carrubba described the work done for 6.E. on catalyst
design for low power operation of a low emissions aircraft combustor. As was
the case with the Exxon program, Engelhard decided to pursue a hybrid
combustor design using a preburner followed by a catalytic cleanup of the
vitiated air stream. Performance goals for the preburner/catalyst system are
1 ppm HC, 10 ppm CO, and 4 ppm NOX, with a 5-percent overall pressure drop in
the combustor and 3 percent in the catalyst.
Screening tests were carried out to select a perferred catalyst
configuration. Test conditions were 2-atm pressure, temperatures of 1000 and
1400 °F, velocities of 80, 100, 120, and 140 ft/sec, and an addition of 200
and 400 ppm by volume of both CO and HC.
From the screening tests it was found that both CO and unburned
hydrocarbon conversion were higher than expected, due to thermal reactions.
Thus the cleanup mode, with air bypass, was found to be a workable design for
low emissions idle mode gas turbine combustors in the laboratory. Further
testing of these catalyst systems at General Electric are forthcoming.
Dr. Ronald M. Heck of Engelhard presented the work done on catalyst
life testing for NASA-Lewis Research Center. This work was performed in
support of the ERDA highway vehicle gas turbine engine program, and the
objective was to investigate the durability of two selected proprietary
Engelhard catalysts. The test sequence for the two catalysts involves a 24-
hour break-in period, a CO activity test, a test with propane to determine
the performance range, another CO activity test, a 1000-hour life test with
No. 2 diesel fuel (with CO activity tests every 250 hours), and a final
propane test to determine performance range changes.
A summary of the test results showed emissions to be the same for both
catalysts during the 1000-hour test with No. 2 diesel fuel. One catalyst
required a higher inlet temperature to maintain low emissions after 600 hours
of testing. The propane parametric studies showed this catalyst had
deactivated completely for high efficiency combustion, and the CO activity
A-12
-------
test showed significant deactivation of both catalysts between 24 and 250
hours of aging.
Emissions with No. 2 diesel fuel after the 1000-hour test were 4 ppm
HC, 50 ppm CO, and 4 ppm NOX. These emissions are well below the 1977 and
1978 automotive standards. Further work to perform life testing at 5-
atmosphere pressure is now underway.
\
Stanford Research Institute Program
Dr. C. M. Ablow of SRI described work now being undertaken at SRI
under AFOSR funding. SRI has begun an analytical study to determine the
contribution of catalytic wall reactions to combustion initiation. The
temperature distribution on the duct wall is found, taking into account wall
heat conduction, convective heat transfer, and heat generation and fuel
consumption. For Lewis numbers greater than one, the temperature increases
with distance down the duct, while for smaller Lewis numbers the temperature
passes through a maximum whose value depends on flow speed. This program is
just being initiated at this time, but the results should be useful in
determining how a catalyst should be distributed for startup and shutdown in
a practical system.
Energy Systems Corporation Program
R.ichard E. LeCompte of Energy Systems Corporation presented work done
at ESC on thermal protection equipment using catalytic combustors. Existing
products include belt-mounted Arctic ambulatory heater systems, SCUBA diver
heaters, hypothermia prevention and treatment systems, casualty evacuation
bag heaters, and downed airman power sources.
The downed airman power source system supplies warmth to airmen in
life rafts by circulating heated water through turbulated undergarments or
blankets. These systems use either propane or propylene fuel combusted on
1-percent Pt on alumina pellets manufactured by Matthey Bishop. Catalyst bed
temperatures are between 1200 and 1400 °F. Heat extraction pins or fins
conduct heat to a hot plate and finally to the fluid heat exchanger.
ESC is also currently developing a catalytic/thermoelectric SCUBA
diver heater for the U.S. Navy. This system will be capable of delivering
500 thermal watts to a diver in 35 °F seawater.
British Gas Corporation Paper
Dr. Robert M. Kendall of Acurex Corporation discussed a paper sent to
the workshop by A. Melvin of the British Gas Corporation. This paper
summarizes research conducted on the modeling of diffusive catalytic
combustion.
Most discussion centered around the conclusions reached in the paper,
which states that diffusive catalytic combustion has disadvantages that
A-13
-------
preclude its extensive use, and that premixed combustion is really
impractical in domestic, commercial, and process applications. Dr. Kendall
pointed out that it is a delicate control problem to get air to the surface
of a diffusive combustion system, but with a good catalyst it should be
possible to produce nearly all C02 at the surface if the fuel flowrate is
controlled. The importance of boundary layer events in the combustion
reaction must also be emphasized. Finally, reference was made to the
recently developed Bratko furnace, which operates successfully as a premixed
diffusive combustor and has emissions of CO and HC < 30 ppm, and NOX < 15
ppm.
Dr. Kendall then acted as moderator for brief presentations by
substrate and catalyst manufacturers.
Substrate Manufacturer Presentations
The monolith manufacturers were represented by Corning Glass Works and
3M's American Lava Corporation.
Corning Glass Works Presentation
Willard A. Boyer of Corning described high-temperature substrates
being developed at Corning for catalytic combustion. Materials, such as
mullite-alumina-titanate and zirconia spinel are being produced to
incorporate high use temperature with good thermal shock properties. In
addition to materials research, Mr. Boyer described the new extruded ceramic
honeycomb cell shape called "flexible rectangles," which allows cell walls to
bend in thermal shock and thereby minimize fracture.
American Lava Corporation Presentation
P- A. Coates of American Lava described Thermacomb brand corrugated
ceramics and their applications to catalytic combustion applications.
Corrugated ceramics are made using conventional paper-making equipment.
Catalyst Manufacturer Presentations
The catalyst manufacturers were represented by Oxy-Catalyst, Inc.,
Engelhard Industries, Universal Oil Products, Matthey Bishop, Inc., and W. R.
Grace and Company.
Oxy-Catalyst, Inc. Presentation
Dr. William B. Retallick of Oxy-Catalyst discussed a current concern;
how much washcoat depth can actually be used during the combustion process?
A mathematical model was developed to calculate how far the reactants can
A-14
-------
diffuse into the catalyst during the contact time. Since kinetic constants
are unknown, estimations are necessary. Based on this analysis, a
significant penetration of the reactants into the washcoat is predicted.
Engelhard Industries Presentation
Dr. Larry Campbell of Engelhard gave an overview of the Engelhard
program in catalytic combustion. Engelhard began with automotive monolithic
catalysts, and proceeded through the development of catalytically supported
thermal combustors to life testing. Special requirements for the materials
used include:
o Monolith
— Thermal shock properties
— Use temperature
~ Melting temperature
~ Porosity
-- Void fraction (pressure drop)
— Geometric surface area
o Washcoat
~ Surface area stability (>20 m2/g after 1200 °C exposure)
-- Adherence
-- Reactivity
o Metals
« Activity
— Stability
— Specific fuels
Engelhard has developed high-temperature combustion catalysts,
demonstrated those catalysts for over 1000 hours in life tests, and will now
be pursuing commercial development of those catalysts in systems.
Universal Oil Products Presentation
Dr. George R. Lester of UOP sees a significant market potential for
this area. Clean fuel catalytic combustion has been fairly convincingly
demonstrated, and the extension to fuels with bound nitrogen is now of major
A-15
-------
interest. UOP will be focusing on optimizing catalysts to produce a minimum
conversion of bound nitrogen to NOX.
Matthey Bishop Presentation
Dr. A. S. D'Allessandro of Matthey Bishop described MBI's interest in
developing washcoats with the maximum stable surface area while maintaining
adherence to precious metals. MBI is also developing high open area,
high surface area, high-temperature metal sheets for monolith application. „
Catalysts have been tested to space velocities of 106/hr.
W. R. Grace and Company Presentation
Mr. Philip A. Smith of W. R. Grace described their Poramic monoliths.
Ceramic components are fluxed with polyethylene and passed through rollers to
form ribs on sheets of material. The sheets are then rolled up and heat-
sealed together, and the structure is fired. During firing the polyethylene
is removed and the ribs are fixed to the ceramic structure. Annular
substrates up to 2.5 feet in diameter can be made in this manner.
Meeting Summary
Blair Martin concluded the workshop with some summary comments.
Good technical progress has been made in the last year in catalytic
combustion. The technology is developing rapidly, more data is available,
and cooperation has been solidified between the R&D community, the
catalyst/substrate manufacturers and, hopefully, the users of this
technology.
Clean fuels are getting in shorter supply every day. Use of lower-
grade fuels necessitates coping with the bound nitrogen problem. Low Btu gas
can give up to 4,000 ppm m^, dependent on the ^S cleanup process, and shale
oils can have significant fuel nitrogen volumes. Catalytic reduction of NO
with ammonia is possible at low temperatures, and may be possible at high
temperatures as well. Thus, it may be possible to control fuel nitrogen,
even in lean combustion, without staging. Tailoring of the catalyst to get
the minimum conversion of fuel nitrogen to NOX is definitely of interest.
Combustion of heavier fuels (No. 4, No. 5, No. 6, and shale-derived
oils; with catalysts is also of great importance. Even if fuel NOX can be
controlled by proper catalyst system design, some fuel upgrading may be
required for particulate formation control. The degree of fuel treatment
will strongly influence fuel cost and the minimum treatment possible should
be addressed.
Thanks were extended to the authors and presenters, the catalyst and
monolith manufacturers, and the attendees for their efforts and interest.
A-16
-------
SECOND WORKSHOP ON CATALYTIC COMBUSTION
Attendees
G. Blair Martin
W. Steven Lanier
Dr. David N. Anderson
Dr. Robert M. Kendall
Dr. John P. Kesselring
William H. Nurick
Dr. Stanley A. Mosier
Captain Thomas J. Rosfjord
Dr. Henry Shaw
Dr. Anthony Cerkanowicz
Dr. William B. Retallick
Dr. George R. Lester
Dr. C. R. Krishna
Dr. Larry Campbell
Dr. Robert V. Carrubba
Dr. Ronald M; Heck
Dr. Karl Bastress
Philip A. Smith
IERL/U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
IERL/U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
NASA-Lewis Research Center, Cleveland,
Ohio
Acurex Corporation, Mountain View, California
Acurex Corporation/Aerotherm, Mountain
View, California
Acurex Corporation/Aerotherm, Mountain
View, California
Pratt & Whitney Aircraft, West Palm Beach,
Florida
Aero Propulsion Laboratory, Wright-Patterson
Air Force Base, Ohio
Exxon Research & Engineering Company,
Linden, New Jersey
Exxon Research & Engineering Company,
Linden, New Jersey
Oxy-Catalyst, Inc., West Chester, Pennsylvania
Universal Oil Products, Des Plaines,
Illinois
Brookhaven National Laboratory, Upton,
New York
Engelhard Industries, Edison, New Jersey
Engelhard Industries, Edison, New Jersey
Engelhard Industries, Edison, New Jersey
Energy Research and Development Administration,
Washington, D.C.
W. R. Grace & Co., Baltimore, Maryland
A-17
-------
Willard A. Boyer
Rodney I. Frost
Phil Coates
S. Mario DeCorso
Dr. Pierre J. Marteney
William Cain
Gerald Roffe
Dr. C. M. Ablow .
Dr. Thomas Tyson
R. V. Dumke
Dr. Edward J. Mularz
Dr. Bernard T. Wolfson
Dr. William C. Pfefferle
Dr. Gerald Voecks
Dr. Frank A. Robben
Dr. G. J. K. Acres
Dr. A. S. D'Alessandro
Walter Jasionowski
Dale Johnson
Corning Glass Works, Corning, New York
Corning Glass Works, Corning, New York
3M Company, St. Paul, Minnesota
Westinghouse Electric Company, Lester,
Pennsylvania
United Technologies Research Center
East Hartford, Connecticut
U.S. Environmental Protection Agency,
Cincinnati, .Ohio
General Applied Science Laboratories,
Westbury, N.Y.
Stanford Research Institute, Menlo Park,
California
Energy & Environmental Research, Inc.,
Irvine, California
Engelhard Industries, Edison, New Jersey
NASA-Lewis Research Center, Cleveland,
Ohio
Air Force Office of Scientific Research
Washington, D.C.
Private Consultant
Middletown, New Jersey
Jet Propulsion Laboratory
California Institute of Technology,
Pasadena, California
Lawrence Berkeley Laboratory, Berkeley,
California
Johnson Matthey Research Centre, Reading,
United Kingdom
Matthey Bishop, Inc., Malvern, Pennsylvania
Institute of Gas Technology, Chicago,
Illinois
Institute of Gas Technology, Chicago,
Illinois
A-18
-------
C. C. Gleason
William Moore
Eugene Szetela
Dr. Frediano Bracco
Peter Matuschak
Richard E. LeCompte
General Electric Company, Cincinnati,
Ohio
Energy Research and Development Administration,
Washington, D.C.
United Technologies Research Center,
East Hartford, Connecticut
Princeton University, Princeton, New
Jersey
Garrett Corp., Phoenix, Arizona
Energy Systems Corp., Nashua, New Hampshire
A-19
-------
SUMMARY
WORKSHOP ON CATALYTIC COMBUSTION
The Plantation Inn
Raleigh, North Carolina
May 25-26, 1976
Sponsored by the U.S. Environmental Protection Agency
Sumnary prepared by J. P. Kesselring, Aerotherm/Acurex Corporation
The first Workshop on Catalytic Combustion was held 1n Raleigh, North Carolina on May 25-26,
1976. Forty people, representing various government, Industrial, and academic organizations, attend-
ed the Workshop; a list of attendees is given at the end of this summary. The purpose of the Work-
shop, sponsored by the U.S. Environmental Protection Agency and organized by the Aerotherm Division
of Acurex Corporation, was to provide an overall summary of the current state-of-the-art of catalytic
combustion. This was accomplished by bringing catalytic combustion research people together for the
exchange of results and ideas.
The meeting began on May 25 with Introductory remarks by Mr. G. B. Martin of the Environmental
Protection Agency. Mr. Martin described the two main areas of research in catalytic combustion at
this time: the gas turbine research, where high excess air levels are used and low NO emissions are
found, and the stationary source research, where the goal is a low NO emission level but with a dif-
ferent system requirement from the gas turbine. For stationary combustion sources, obtaining high
system efficiency is the primary problem, and for these systems it is necessary to minimize excess air
and exhaust gas temperature to maximize system efficiency. The need to control the conversion of fuel
nitrogen in stationary sources was also described. Mr. Martin then introduced Dr. R. V. Carrubba of
the Engelhard Industries Division of Engelhard Minerals and Chemicals Corporation. Engelhard Industries
Is the acknowledged pioneer in the field of high temperature, high heat release catalytic combustion.
Engelhard Industries Program
Dr. Carrubba discussed the necessity of having both catalytic and thermal combustion processes
operating in a high heat release combustor; the purely catalytic combustion processes are mass-
transfer limited, and cannot achieve the required high combustion efficiency on their own. The
performance range of the catalytically supported thermal combustor Is defined by the two processes,
with the catalyst determining the minimum Inlet temperature and maximum operating temperature for a
given fuel, and the thermal reactions determining the minimum flame temperature as defined by the
onset of thermal combustion reactions. For propane, a typical operating range with a proprietary
Engelhard catalyst is:
A-20
-------
• Inlet temperature 300-550°C
• Maximum operating temperature 1500°C (material and catalyst failure becomes a problem
above this value)
• Minimum operating temperature 1000°C (below this value, thermal reactions become less
significant
• Minimum Inlet velocity 3 m/sec (have flashback below this value)
• Maximum Inlet velocity 25 m/sec
• Typical emissions
- UHC : 2 ppmv
- CO : 30 ppmv
- NOX : 1 ppmv
Dr. Carrubba described the Engelhard laboratory test facility, used for life tests and studies
on small (1-inch diameter by 6-1nches 1n length) monolith catalyst systems, and showed photographs of
larger monolith catalyst systems tested at NASA-Lewis Research Center and the A1r Force Aero Propulsion
Laboratory. A comparison of data obtained on small catalysts at Engelhard and on larger systems at
NASA-Lewis show Identical results under fuel-lean conditions, Indicating scale-up of catalyst perfor-
mance Is achievable simply In terms of Increasing frontal area.
Data was presented for a variety of fuels tested with a single catalyst. The data demonstrated
that acceptable performance can be obtained from a single catalyst on a wide range of fuels, and that
the typical Inlet temperature for low emissions 1s below the catalytic ignition or "lightoff" tempera-
ture. Lean tests with fuels doped with pyridine and ammonia Indicated about 80 percent conversion of
fuel nitrogen to NO , which Is typical of premixed noncatalytlc systems.
Data describing the pressure drop as a function of reference velocity for 4-atm tests with
propane and a typical monolith catalyst showed the pressure drop to be given by the equation
(m/sec)
for several inlet temperatures and reference velocities between 10 and 25 m/sec. Fuel-Injection and
mixing pressure drops should be added to the catalyst bed pressure drop to estimate the total com-
bustion pressure loss.
A-21
-------
In hardware development, Engelhard has developed complete working catalytic combustors capable
of up to 106 Btu/hr heat release, gaseous or liquid fuel feed, ambient start-up, good turndown, and
low emissions for a variety of relatively clean fuels.
In conclusion, a low emissions capability has been demonstrated for a broad range of gaseous
and light distillate fuels. Gaseous fuels of low heating values (60 Btu/ft3 and up) can be com-
busted efficiently, and the scale-up to larger size combustors has been demonstrated in field tests.
Unpublished results indicate it is feasible to design and fabricate commercially useful gas turbine
and atmospheric burners. Detailed work in terms of poisons and potential use with heavier fuels
still needs to be done.
Catalyst Manufacturer's Presentations
Dr. Robert M. Kendall, Corporate Chief Scientist of the Acurex Corporation, acted as moderator
for presentations from five catalyst manufacturers and one refractory support manufacturer. The
presentations centered on the current state-of-the-art of manufacturing combustion catalysts.
Universal Oil Products. Dr. George R. Lester presented considerations for the selection of
combustion catalysts. This selection is based upon many parameters of the specific application,
such as fuel type, required fuel/air ratio for acceptable NO emissions, minimum required Ignition
temperatures, range of operating linear velocities, acceptable pressure drop, required life, pre-
sence of poisons, and cost restraints. By specifying these parameters, the catalyst manufacturer
is allowed to use his experience and skill to develop an optimal catalyst for the application.
Dr. Lester also presented data obtained on an experimental UOP catalyst used to combust
propane. At an equivalence ratio of 0.321, Inlet temperature of 700°F, and inlet velocity of
75 ft/sec, an outlet temperature of 1900°F was achieved with greater than 99.9 percent conversion
of hydrocarbons.
Oxy-Catalyst Division, Research-Cottrell. Dr. William B. RetalUck presented his design for
an isothermal microreactor system used 1n the screening of oxidation catalysts. Since many catalysts
must be screened, the test method must be fast and simple to perform. The tests screen the catalyst
for fresh activity and activity after heat aging has occurred. The microeactor system used consists
of a quartz tube in which the catalyst 1s placed. A gas flow of 1 percent CO, 1 percent 02 in
helium is passed over the catalyst, and a sample of the exhaust is sent to a gas chromatograph for
analysis. For heat aging tests, a single stainless steel block could hold several microreactors
to maximize the screening rate.
A-22
-------
W. R. Grace and Co. Dr. James Maselli presented thoughts on the current catalyst systems.
Automotive catalysts will soon be surpassed technologically by greatly Improved systems currently
under development. The materials problem 1s critical, since the degradation of the support enhances
metal sintering. The stability of the monolith material at high temperatures is also a prime con-
sideration. Grace prefers to take a fundamental solid-state chemistry approach to developmental sys-
tems, rather than an empirical approach. The new catalysts being developed under this approach will
exhibit major Improvements 1n lightoff temperature and resistance to poisons.
Engelhard Minerals and Chemicals Corporation. Dr. George W. Roberts presented some of the
thoughts that directed the development of the Catathermal combustion systems. By comparing the
catalyst system pressure drop and reaction rate for both particulate and honeycomb supports, the
monolithic honeycomb requires significantly less catalyst volume than the particulate 1n the mass-
transfer controlled regime, and is therefore the preferred support. In order to prolong support
life, operation should be in the mass-transfer controlled regime, but unstabilized AO, washcoat
total surface area will change from approximately 240 mVg of washcoat to 10 mz/g after operation
at 1200°C. The solution to this problem 1s to either stabilize the AO, washcoat or look at other
washcoats 1n order to maintain a high total surface area.
Typical operating parameters for the Catathermal combustors with propane and #2 dlesel oil
fuels were reported as adiabatlc flame temperature between 1800 and 2800°F, pressure drop between
1 and 6 percent, heat release rates from 10 x 106 to 20 x 106 Btu/hr-atm-ft3, reference velocities
between 25 and 100 ft/sec, and a catalyst life 1n excess of 1000 hours. These systems are still
under development.
E. I. DuPont deNemours and Co. Dr. Donald M. Sowards presented considerations from the point
of view of the monolith manufacturer. The monolith manufacturer must do coating 1n order to learn
to make an Improved monolith. DuPont has used two different washcoats (y-alumina, and ex-alumina
with some other phases) In their work. The
-------
to a monolith with 400 cells/In2, but the pressure drop 1s now Increased and the open area goes down.
It appears that a newly-developed platinum catalyst on a metal-based monolith 1s capable of achieving
90 percent open area and 400 cells/in2. Preliminary data with kerosene fuel at 60 psig pressure and
250°C inlet temperature are: space velocity = 1.17 x 106/hr @ STP, 2 percent pressure drop, 100
percent conversion of hydrocarbons at an equivalence ratio of 0.38. The pressure drop is only half
that of 200 cell/in-" ceramic monoliths for velocities from 2 to 50 ft/sec. The system should have
long lifetime at 1100°C, with 1375°C being the maximum use temperature of the iron-chrome metal mono-
lith. The washcoat is alumina, based on the automotive catalyst technology.
Institute of Gas Technology Programs
Dr. Jon B. Pangborn of IGT described work currently In progress at the Institute of Gas Tech-
nology. Starting 1n 1972, IGT has been developing conceptual burners and model ventless appliances
for catalytic combustion of hydrogen and reformed natural gas under the sponsorshop of the Southern
California Gas Company of Los Angeles. During 1975, the U.S. Environmental Protection Agency and
Southern California Gas co-funded a program at IGT to develop a catalytic range-top burner.
The burners designed for catalytic combustion must be self-starting in order to be practical;
flame combustion, standing pilot, or electrical ignition systems were not allowed. These burners
were developed for hydrogen-rich gas fuel, and the catalyst configurations were to completely com-
bust the fuel 1n the system configuration. The system was also to be flashback-free. Based on
these requirements, IGT has developed burners and two types of hot-water heaters based on catalytic
combustion. The instantaneous catalytic hot-water heater, which uses a noble metal catalyst on a
specially-treated aluminum surface, is capable of 99+ percent combustion efficiency, 75 - 80 percent
thermal efficiency, and a heat-up time (50°F to 150"F) from 30 to 45 seconds for 6 gallons/hour
capacity.
In summary, the catalytic combustion systems developed at IGT have several advantages over
current systems. These include: minimal pollutants (very low NO and CO emissions), less expensive
because no venting is required, provide for humidification of the home, high system efficiency, low
operating temperature and lower attendant hazard, and self-ignition. The biggest problem to date
has been cracking and peeling of th¥ catalyst from the surface; the need to have either a methane re-
former or a hydrogen fuel supply 1s also evident.
NASA Lewis Research Center Program
Dr. David N. Anderson of NASA Lewis Research Center introduced the NASA Lewis program and de-
scribed the three major research areas: fuel-air preparation system studies, catalyst evaluation
program, and catalyst life test program. This work is directed toward finding a catalyst combustor
for automative gas turbine applications, and is sponsored by ERDA.
A-24
-------
Fuel-Air Preparation System Studies. Robert R. Tacina of NASA Lewis presented Information on
the fuel preparation and introduction system accomplished to date. Premixed, prevaporlzed fuel-air
systems are required to avoid local rich zones and subsequent NO emissions. The goals of the
current program are threefold:
1. 90 percent vaporization of the fuel at 1000°F inlet temperature,
2. A spatial fuel distribution ±]Q% of the mean value across the duct cross-section,
3. No autoignition at 1400°F.
The spatial fuel distribution and degree of vaporization were measured using Jet A fuel. Fuel
injectors tested include two types of air blast injectors, a splash-groove injector, and a multiple-
jet cross-stream injector. Air swirlers with vane angles of 15° and 30° were used to improve the
spatial fuel distribution.
As a result of these tests, it was found that the multiple-jet cross-stream injector and the
splash-groove Injector with a 30° air swirler had a uniform fuel distribution and a high degree of
vaporization with little total pressure drop at the conditions tested. At 800°K inlet air tempera-
ture, fuel oxidation reactions were noted. Since the automotive gas turbine will probably encounter
higher inlet air temperatures, further work is required for the application of catalytic combustors
to automotive gas turbine engines.
Catalyst Evaluation Program. Dr. Anderson described the two elements of the catalyst evalua-
tion program; the furnace screening tests of monoliths and pellets of small size by passing 500
ppm of propane in air over the catalyst, and the combustion tests of monoliths 12 cm in diameter
with 800°K Inlet temperature and premixed propane in air at equivalence ratios between 0.1 and 0.3.
Results from the furnace screening tests are plotted as oxidized fuel fraction versus catalyst
temperature, and serve to identify the most suitable catalysts for further testing. The furnace
tests have indicated that the most effective catalysts for gas turbine combustor applications will
probably be noble metals on monoliths. Based only on catalytic activity, the furnace tests have
shown that the optimum cell density should be between 30 and 45 cells/cm2. However, most of the
catalysts used in the study were developed for use as automobile exhaust emission control reactors,
and not for the current application.
The combustion tests were run on catalyst beds consisting of two to four elements of 12-
centimeter diameter by 2.5 centimeter long monoliths. Thermocouples at the back end of each monolith
r
element measure the gas phase temperature. Two cell configurations have been used in the combustion
tests to date: small square cells in the Grace Poramie 290 cordlerlte, with a cell density of 45
A-25
-------
cells/cm2, and round cells in General Refractories1 Versagrid mullite, with a cell density of 10
cells/cm2. Catalysts used include platinum, palladium, mixed platinum/palladium, cobalt oxide, and
cerla.
Results of the combustion tests are given by combustion efficiency as a function of adiabatlc
reaction temperature. Results indicate that the temperature rise across catalyst elements decreases
as the reference velocity is increased, and that most of the temperature rise (and thus most of the
reaction) occurs across the first two elements. A comparison of the Grace Davex 512B catalyst (with
small square cells) and the Oxy-Catalyst 4 catalyst (with larger round cells) shows that the Oxy-Cat
system required higher temperatures to achieve the same combustion efficiency as the Grace system.
A detailed study of cell shape and cell size for optimum performance has not been done, however.
As a result of'this program, several potential problem and developmental areas have been
determined. The narrow operating range in temperature between the minimum temperature for good
efficiency and the maximum temperature for long catalyst life must be addressed. The transient
response has not yet been determined, and the loss in activity with test time will require more
durable catalysts to be developed. In addition, the possibility of improved combustor performance
from variable catalyst system geometry must be explored.
Life Test Studies. Dr. Robert V. Carrubba of Engelhard Industries returned to describe the
life testing Engelhard has carried out for NASA Lewis. The objectives of this program were to
select combustion catalysts for testing, to construct a catalyst test rig, to demonstrate low
emissions combustion using #2 oil for up to 1000 hours, and to compare catalysts using propane in
/
parametric tests. The test rig is capable of testing catalysts one inch in diameter by 6 inches in
length with a maximum inlet temperature of 1000°F. The life test conditions used are 1 atm pres-
sure, 680°F inlet temperature, 45 ft/sec reference velocity, and air/fuel ratio of 38 with #2 oil.
At these conditions, one catalyst has demonstrated at least 1000 hours life with emissions which
exceed Federal requirements for automotive engines based on a single, static Hg operating point.
The catalyst showed an Initial activity decline to 250 hours, with subsequent stable operation to
1000 hours.
Parametric test results with propane show un operating range decline after 1000 hours com-
pared to #2 oil after 1000 hours on the same catalyst. Further experimentation will be done to
compare the fuels capability of the catalyst.
Jet Propulsion Laboratory Program
Dr. John Houseman of JPL described the JPL partial oxidation synthetic gas generator program.
The objectives of this program are to design and build a partial oxidation syn-gas generator, to
A-26
-------
package the reactor and controls 1n separate consoles, and to prepare an operating manual. The syn-
gas generator will be used In EPA studies to generate a simulated low Btu coal gasification product.
It has also been used to produce a hydrogen-rich gas stream for combustion 1n an automotive engine.
The gas generator consists of an igniter and start-up burner to provide a preheat source for
the catalyst bed, a start-up combustion zone, and a steady state reaction zone. The steady state
reaction zone consists of nickel on alumina pellets, which act as the catalyst. The generator is
normally run in the fuel-rich mode, but could run lean as well. At an equivalence ratio of 2.8, the
product gas from the generator is approximately 22 percent H2 and 25 percent CO by volume, when indo-
lene is used as the fuel. Benzene and methane have also been used as fuels.
U.S. Environmental Protection Agency Program
Blair Martin began the workshop on Wednesday, May 26, by describing the need to improve system
efficiency in stationary combustion systems. Conventional flame combustion is very efficient in terms
of combustion efficiency, but improvements 1n system efficiency occur when excess air is decreased and
stack temperature is reduced. These requirements are not compatible with the capabilities of current
one-stage lean catalytic combustors being developed for gas turbines. If stationary heat and steam
generators require the use of catalytic combustion to achieve very low emissions of NOX with high com-
bustion efficiency, alternate approaches to design must be explored. These include staged combustion,
flue gas recirculation, and bed heat removal. The EPA program is directed to establishing the tech-
nical feasibility of these concepts as applied to stationary sources.
The potential of using two-stage catalytic combustion to suppress fuel NO production was also
discussed. Fuel NO Increases with an Increase in primary stoichiometric ratio, and it appears that
90 percent control of fuel NO is possible with staged combustion.
Aerotherm/Acurex Corporation Program. Under EPA sponsorship, the Aerotherm Division of Acurex
Corporation is conducting a study to determine design criteria for catalytic combustors with applica-
tion to stationary sources. Dr. John P. Kesselring of Aerotherm began the presentation by discussing
the program plan to screen experimental small scale catalysts and combustion system concepts, to scale
the promising concepts up, both experimentally and theoretically, and to do preliminary prototype de-
sign of systems based on the most promising concepts. The operation of a high-temperature catalytic
combustor is based on an Interaction of fluid mechanics, radiative, convectlve, and conductive heat
transfer, and heterogeneous and homogeneous chemical reactions, and a proper understanding of these
phenomena is basic to the development of a successful system.
A-27
-------
Dr. Robert M. Kendall of the Acurex Corporation then discussed basic performance and modeling
considerations. A simplified treatment of the surface, 1n which the wall 1s assumed adlabatlc, the
enthalpy assumed constant, and an Arrhenlus rate law assumed for the consumption of the lean reactant
gives certain Information about the blowout point. Blowout, defined as that condition where both
homogeneous and heterogeneous reactions cease, can be avoided by having large diameter cells, high
thermal conductivity of the monolith, and high surface activity. The large diameter cells, however,
will have relatively poor conversion characteristics, and It 1s known that small cells help support
homogeneous reactions and complete conversion. Therefore, for a system 1t appears desirable to have
large cells at the front of the catalyst bed to prevent blowout, and small cells at the back of the
bed to prevent breakthrough (passing of unburned hydrocarbons).
The screening program performed consists of catalyst preparation and characterization, catalyst
instrumentation with in-depth bed thermocouples, combustion screening tests, and post-test characteri-
zation and data analysis. The objectives of the screening program are to develop a catalyst which 1s
fully characterized In terms of catalyst properties which 1s capable of a volumetric heat release rate
of 2.5 x 106 Btu/hr-atm-ft3, and can run under rich, lean, or stolchlometric conditions for a minimum
of 50 hours.
Dr. James A. Cusumano of Catalytica Associates described the preparation and characterization
procedures for combustion catalysts. Catalytica Associates is working on this program under subcon-
tract from Aerotherm.
The catalyst preparation objectives are to produce a catalyst with good light-off and heat re-
lease characteristics that is durable at high temperatures. The characterization procedure helps
monitor progress in this direction. Use of total surface area and dispersion measurements gives quan-
tification and direction to improvements in catalyst manufacture. The catalyst system can deactivate
at high temperatures because of sintering of the active component (e.g., Pt), sintering of the wash-
coat and subsequent loss in total surface area, and mechanical degradation. Methods used for cata-
lyst stabilization Include sulfide pretreatment, washcoat presinterlng, and metal-support interactions.
In the combustion screening tests, monoliths 3.66 Inches In diameter and 3 Inches 1n length
have been tested 1n a variety of materials, cell sizes, and platinum loadings. The operating temper-
ature selected is 2000°F, and methane has been the primary test fuel, with some propane tests being
run. Eventually, low Btu gas, hydrogen, #2 diesel, and methanol will also be tested.
Combustion screening tests to date have been run primarily in the existing catalytic combustion
test facility at the Jet Propulsion Laboratory, with some tests being run 1n the recently-completed
test facility at Aerotherm. Dr. Gerald E. Voecks of JPL described the combustion testing at JPL.
A-28
-------
The objective of the combustion test program is to evaluate the catalyst in its ability to
conduct a variety of fuels over a wide range of conditions. The approach has been to determine
catalyst activity under rich, stoichiometric, and lean conditions, to determine blowout or break-
through performance for rich and lean conditions, and to relate catalyst activity to catalyst prop-
erties. The JPL reactor 1s made of quartz, allowing the monolith bed to be observed during a test
when the insulating material is removed.
Results of combustion tests to date have shown considerable degradation with time for all
catalyst systems. This degradation levels out to a nearly constant activity after several hours,
however, and it is this activity which must be determined for system design considerations. Tests
run with a 5 percent platinum on alumina first catalyst element with large cells, followed by 0.7
percent platinum on alumina second and third elements with small cells showed excellent performance
after several hours of operation in terms of activity and emissions. The testing of this system
thus verifies the basic considerations described earlier.
ERDA Considerations
Mr. William E. Moore of the Fossil Energy Administration of ERDA and Mr. Robert Ebeling of
Gilbert Associates described some of the needs of the catalytic combustion program from their point
of view.
ERDA's advanced turbine development program couples a high turbine inlet termperature require-
ment (2600°F and up) with emission problems imposed by coal-based fuels. When burning coal-based
fuels the emissions are of primary concern due to the large nitrogen and aromatic content of the
fuels. The need for advanced combustors to avoid the conversion of fuel nitrogen to NO is apparent.
At some point, the cost tradeoff between fuel enhancement and turbine combustor development
will dictate what fuel and combustor combination will be most economical to use. The determination
of this point Is critical.
United Technologies Research Center Program ,
Dr. Arthur S. Kesten of UTRC described their catalytic combustion program. The UTRC program
involves experimental and theoretical studies of the use of catalyzed surfaces to promote hydrocarbon
combustion. The ultimate objective of the program is to assess the feasibility of using catalyzed
surface reactors to promote the efficient combustion of multicomponent liquid fuels 1n practical
combustion systems where the fuel is only partially prevaporized before entering the reaction cham-
ber. Initial objectives of the program are to extract Information on catalytic reaction mechanisms
and rates for pure gaseous fuels and monolithic catalyst structures.
A-29
-------
The experimental program is studying the effects on reactor behavior of fuel, catalyst type,
monolith structure, temperature, stoichiometry, mass flowrate, bed length, and diluent concentration.
Catalyst beds one-inch in diameter are tested in the reactor. Propane fuel has been used to Investi-
gate the variation in ignition temperature with inlet velocity and equivalence ratio, and the combus-
tion efficiency change with bed temperature. A mathematical model to aid in evaluating the data was
also described. This model is a steady-state analysis which assumes axial variations in bed temper-
ature, bulk-gas temperature, and reactant concentration.
U.S. Air Force Programs
The U.S. Air Force is sponsoring two programs dealing with catalytic combustion. The first
program, sponsored by the Aero Propulsion Laboratory, deals with the application of catalytic combus-
tion to aircraft gas turbine engines. The second program, sponsored by the Air Force Office of
Scientific Research, deals with the initiation of combustion on catalytic surfaces.
Aero Propulsion Laboratory Program. Captain Thomas J. Rosfjord of the AFAPL described current
activity in developing the catalytic combustion concept for application to aircraft gas turbine en-
gines. At this time, the program is addressing the use of catalysts in after burners. Two specific
applications are of interest: igniters and flameholders. A more simple system with increased per-
formance can result with successful application of the concept. Previous work done at the AFAPL dealt
with the use of catalysts in the main burner of the gas turbine engine.
The aircraft application imposes a great demand for throttling the power level and presents
a large variation in combustor inlet temperature as well. The current studies are divided between
low power investigations, subcontracted to Exxon initially and to be followed by extensive In-house
testing, and high power investigations, being conducted in-house.
Current focus of applications of catalytic combustion is on Ignitors or f1ameholders. The
catalytic ignitor can provide a simple, reliable ignition source, with the prime concern being the
location of the ignitor. Use as a f landholder can promote an Increase 1n the axial rate of energy
release. Flight tests at aircraft approach conditions have shown that the catalytic Ignitor alone
is unsatisfactory, but with the pilot burner an expansion in relight envelope can be achieved.
The work to develop a catalytic combustor to operate in the engine Idle-mode was described by
Mr. V. J. Siminski of Exxon. Since modifications to conventional aircraft turbine engine combustors
have not been able to meet 1983 EPA standards, and there is low combustion efficiency in the Idle-
mode, the use of catalytic combustors is attractive. The catalytic combustor can operate efficient-
ly far off stoichiometric, and generate low NOX at the same time. The objectives of the program are:
1) to conduct a thorough literature review, defining the best catalysts and substrates, 2) to design
A-30
-------
and fabricate a combustion system, modifying the existing catalytic combustor at Exxon to Include
the Idle operation mode, 3) to evaluate a minimum of 10 catalyst/substrate systems and select the
best two candidates, 4) to complete a detailed design of each recommended system and fabricate two
copies for delivery.
Important considerations for the catalyst system include activity, lifetime, pressure drop,
cost, and poisoning. Catalysts similar to automotive catalysts have been tested, and the following
conclusions have been made:
• The hybrid catalytic combustion technique can be used to meet the 1983 aircraft emission
standards. In this system the catalyst is operating 1n the clean-up mode, with low inlet
hydrocarbons and CO.
t An overall combustion efficiency of 99.8 percent has been achieved
• Operating conditions are selected to avoid catalyst degradation (<1200K)
• The optimum support surface/volume ratio required is 1500 mz/m3
• The hybrid catalytic combustion technique provides stable and efficient combustion over a
wide range of fuel-air mixtures.
Air Force Office of Scientific Research Program. Professor F. B. Bracco of Princeton University
presented a discussion of work now beginning under AFOSR funding. This program will consist of a theo-
retical-experimental study of the Initiation of combustion on catalytic surfaces. Initially, an Iso-
thermal boundary layer will be considered with gas phase and surface reactions of trace species only.
Then, the fully reactive two-dimensional flow system will be studied. The experimental program will
be carried out with a steady combustor and will Include measurement of stable and unstable species by
a continuous argon ion laser system.
Exxon-National Science Foundation Program
Dr. Anthony E. Cerkanowlcz of Exxon presented the work done under joint Exxon-NSF sponsorship
on catalytic combustion modeling. The objective of this program 1s to develop a usable model of
catalytic combustion reactor operation which provides for optimization, design, scaling, and data
analysis. The major assumptions 1n the model are one step global kinetics, first order kinetics,
and negligible gas phase reactions and radiant heat transport. Based on the analysis, 1t has bean
observed that multiple steady state solutions exist, and that the gas residence time 1n the reactor
\
Is a critical factor for design. Future work will correlate experimental data, Include gas phase
kinetics, Include heat transfer effects, and consider transient operation.
A-31
-------
Hestinghouse-Engelhard Joint Program
Mr. S. M. DeCorso of Westlnghouse described a program conducted jointly by Westlnghouse and
Engelhard Industries to assess the applicability of catalytic combustors for gas turbines. Test
fuels used were #2 distillate oil and low Btu synthetic coal gas. The tests were carried out over a
range of pressure, temperature, and mass flow conditions.
Currently, combustors are based on the diffusion flame, are not scaleable, and are emissions
and performance limited. The catalytic combustor can avoid NO formation, and also avoid the relia-
bility problems of high temperature operation in conventional systems.
Experimental test results were obtained for #2 distillate oil and low Btu gas (126 Btu/ft3
lower heating value). Pressure, temperature, and mass flowrate were varied during the tests. The
joint program began in 1973 and is still going on. A total of 34 tests have been run, with combustor
pressure drop varying from 2 to 12 percent as reference velocity increases. The catalyst bed temp-
erature profile at the bed exit was very uniform for low Btu gas, but not as uniform for #2 oil.
Exceptionally low emissions (2-3 ppm NO , 20-30 ppm CO) were achieved for either fuel, and unburned
A
hydrocarbons were < 1 ppm. The catalytic combustor has shown excellent potential for gas turbines,
but much additional systems work is needed before it can be reduced to practice.
University of California Programs
Dr. Frank Robben of the Lawrence Berkeley Laboratory of UC described an ERDA-sponsored pro-
gram now underway at LBL. The program consists of an experimental and computational study of com-
bustion in the boundary layer of a catalytic flat plate in a premixed fuel-air stream. The velocity,
gas density, and degree of reaction in the boundary layer will be measured using laser light scatter-
ing techniques. The principal goals of the program are: 1) a fuller understanding of the relation
between catalyzed heterogeneous reactions at the wall and homogeneous reactions In the boundary
layer, 2) the development of a suitable computational model to enable these processes to be predicted
in more complex geometry, and 3) to test and develop suitable catalytic surfaces for combustion.
Dr. Robert F. Sawyer of the Mechanical Engineering Department at UC is beginning a study of
catalytic combustion and its role in solving the fuel-nitrogen problem. The desired goal is the con-
version of fuel nitrogen to N2 during the combustion process, but this appears difficult. Since the
use of alternate or synthetic fuels will be likely to include those with relatively high organically
bound nitrogen content, work on the fate of these compounds In catalytic combustion processes Is
needed.
A-32
-------
There 1s no indication that the low operating temperatures of catalytic combustion which
effectively suppress thermal NO formation will similarly favor the conversion of fuel nitrogen to
N2- On the contrary, lean conditions should favor the formation of NOX and fuel rich conditions may
favor the formation of a number of nitrogen compounds.
Meeting Summary
The Workshop presentations give an indication of the current state-of-the-art of catalytic
combustion. Research in catalytic combustion at the present time is 1n the areas of high tempera-
ture catalysts, gas turbines, stationary combustion systems, and home appliances.
For almost all applications of catalytic combustion of current interest, higher temperature
catalysts are required. In this regard, current automotive catalyst technology needs to be upgraded
to obtain catalyst materials that are capable of high temperature (i.e., 2600°F) operation with good
activity, resistance to poisoning, long lifetime, and good lightoff characteristics. Scale-up of
performance for catalysts is achievable by scaling the frontal area of the catalyst bed. In the bed
itself, both heterogeneous and homogeneous reactions are important, and the monolith bed configura-
tion is the preferred substrate.. Tailoring of the geometric cell configuration in the bed, by using
large cells at the bed inlet to prevent blowout and small cells at the bed exit to achieve cleanup,
is postulated to improve catalyst performance for a given catalyst activity. Further work on tailor-
\
ing of the catalyst in the bed, and the acceptance of some unvaporized fuel in the bed, is needed.
An initiation of work aimed at developing a combustor which does not convert fuel nitrogen to NO is
A
also needed; rich catalytic combustion may be applicable here.
The use of catalytic combustors for gas turbines has been shown applicable by a variety of
test programs. Further development work is needed, however, to answer questions regarding system Im-
pacts of catalytic combustors. While the catalyst performance appears scaleable in size, the perfor-
mance of the fuel preparation and introduction system is not, and much developmental work is needed
there. Tests of a hybrid gas £urbine system, with the catalytic combustor acting as a low-temperature
cleanup stage, have shown that low emissions can be obtained with good efficiency. Further testing 1s
needed to determine if a catalytic combustor can operate effectively over the entire range of condi-
tions that an aircraft gas turbine encounters, or if a hybrid system is needed.
In the field of stationary combustion systems, testing is needed to show the potential of
catalytic combustion for obtaining high system efficiency. Experimental work involving various heat
removal mechanisms needs to be done to determine system Impacts, followed by further catalyst develop-
ment work. In particular, the effects of coal-derived fuels on various catalysts must be determined.
A-33
-------
For home appHcances, the use of lower catalyst lightoff temperatures has been possible be-
cause of the fuels used (hydrogen and reformed natural gas). The supply of these fuels 1s somewhat
problematic, however, and this work should be extended to other fuels as well.
Thanks were extended to all speakers and participants for making this workshop a success.
A-34
-------
WORKSHOP ON CATALYTIC COMBUSTION
Attendees
Name
G. Blair Martin
W. Steven Lanier
Joshua S. Bowen
Dr. David N. Anderson
Robert R. Tacina
Dr. Gerald E. Voecks
Dr. John Houseman
Dr. Stanley A. Hosier
Dr. John P. Kesselring
Dr. Robert M. Kendall
Dr. Kimble J. Clark
Dr. James A. Cusumano
Captain Thomas J. Rosfjord
Dr. William B. Retail1ck
Vincent Simlnski
Dr. Henry Shaw
Dr. Anthony Cerkanowicz
Daniel Kahn
Dr. Jon Pangborn
Organization
IERL/U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
IERL/U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
IERL/U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
NASA Lewis Research Center
Cleveland, Ohio
NASA Lewis Research Center
Cleveland, Ohio
Jet Propulsion Laboratory
Pasadena, California
Jet Propulsion Laboratory
Pasadena, California
Pratt & Whitney Aircraft
West Palm Beach, Florida
Aerotherm/Acurex Corporation
Mountain View, California
Acurex Corporation
Mountain View, California
Aerotherm/Acurex Corporation
Mountain View, California
Catalytica Associates, Inc.
Palo Alto, California
Aero Propulsion Laboratory
Wright-Patterson Air Force Base, Ohio
Oxy-Catalyst, Inc.
West Chester, Pennsylvania
Exxon
Linden, New Jersey
Exxon
Linden, New Jersey
Exxon
Linden, New Jersey
Rocketdyne Div., Rockwell Int'l Corp.
Canoga Park, California
Institute of Gas Technology
Chicago, Illinois
A-35
-------
Name
Organization
Dr. Arthur Kesten
Robert Ebeling
Dr. Robert V. Carrubba
Dr. G. W. Roberts
Dr. James Maselli
Dr. Fred Bracco
Dr. Bernard T. Wolfson
Rudolph Bratko
Dr. C. C. Lee
Dr. George McGuIre
Dr. Robert Sawyer
Dr. Thomas Tyson
S. Mario DeCorso
Dr. George R. Lester
William Williams
Paul Jensen
Dr. David Clark
Dr. Frank Robben
Dr. Donald M. Sowards
William E. Moore
Dr. G. J. K. Acres
United Technologies Research Lab.
East Hartford, Connecticut
Gilbert Associates
Reading, Pennsylvania
Engelhard Industries
Edison, New Jersey
Engelhard Industries
Edison, New Jersey
W. R. Grace Chemical Co.
Baltimore, Maryland
Princeton University
Princeton, New Jersey
Air Force Office of Scientific Research
Washington, D. C.
Bratko Corporation
Cleveland, Ohio
IERL/U.S. Environmental Protection Agency
Cincinnati, Ohio
Matthey Bishop, Inc.
Malvern, Pennsylvania
University of California
Berkeley, California
Ultrasystems
Irvine, California
Westinghouse Electric Company
Lester, Pennsylvania
Universal Oil Products
Des Plalnes, Illinois
ERDA
Washington, D.C.
ERDA
Idaho Falls, Idaho
Detroit Diesel Allison
Indianapolis, Indiana
\
Lawrence Berkeley Laboratory
Berkeley, California
E. I. DuPont deNemours & Co.
Wilmington, Delaware
ERDA, Fossil Energy Administration
Washington, D.C.
Johnson Matthey Research Centre
Reading, England
A-36
-------
APPENDIX B
LIST OF ATTENDEES
B-l
-------
THIRD WORKSHOP ON CATALYTIC COMBUSTION
ATTENDEES
Dr. Robert V. Carrubba
G. Blair Martin
Dr. John T. Pogson
Dr. John P. Kesselring
Dr. Robert M. Kendall
Mil lard A. Boyer
Dr. Frank A. Robben
Dr. Robert W. Schefer
Wayne V. Krill
Edward K. Chu
Dr. George R. Lester
Engelhard Industries, Menlo Park
Edison, New Jersey 08817
MD-65,. IERL/U.S. Environmental Pro-
tection Agency, Research Triangle Park,
N.C. 27711
Acurex Corp. / E&E Division
485 Clyde Ave.
Mt. View, CA 94042
Acurex Corp./E&E Division
485 Clyde Ave.
Mt. View, CA 94042
Acurex Corp./E&E Division
485 Clyde Ave.
Mt. View, CA 94042
Industrial Products Dept.
Ceramic Products Div.
Corning Glass Works
Corning, N.Y. 14830
Lawrence Berkeley Laboratory, 70-158
Berkeley, CA 94720
Lawrence Berkeley Laboratory, 70-158
Berkeley, CA 94720
Acurex Corp./E&E Division
485 Clyde Ave.
Mt. View, CA 94042
Acurex Corp./E&E Division
485 Clyde Ave.
Mt. View, CA 94042
Director - Materials Science
Universal Oil Products Company
Corporate Research Center
Ten UOP Plaza
Des Plaines, Illinois 60016
B-3
-------
Page 2
Dr. Gerald E. Voecks
Dr. Edward J. Mularz
Y. Ishihara
Dr. David N. Anderson
Dr. Blair A. Folsom
Dr. William B. Retallick
Dr. Stanley A. Hosier
Dr. Pierre J. Marteney
Robert R. Tacina
Daniel L. Bulzan
Jet Propulsion Laboratory
California Institute of Technology
4800 Oak Grove Drive
Pasedena, California 91103
Propulsion Lab, USARTL (AVRADCOM)
Mail Stop 60-4
NASA-Lewis Research Center
21000 Brookpark Rd.
Cleveland, Ohio 44135
Mgr., Environmental Chemistry Dept.
Energy & Environmental Lab
Central Research Institute of Electric
Power Industry
11-1, Iwato Kita 2 Chome; Komae-Shi,
Tokyo 182, Japan
Mail Stop 500-202
NASA-Lewis Research Center
21000 Brookpark Rd.
Cleveland, Ohio 44135
Energy & Environmental Research Corp.
8001 Irvine Boulevard
Santa Ana, CA 92705
Vice President, R&D
Oxy-Catalyst
E. Biddle Street
West Chester, Pennsylvania
19380
Mail Stop E-75
P. 0. Box 2691
Pratt & Whitney Aircraft
West Palm Beach, Florida 33402
United Technologies Research Center
400 Main Street
East Hartford, Connecticut 06108
Mail Stop 500-202
NASA-Lewis Research Center
21000 Brookpark Road
Cleveland, OH 44135
Mail Stop 500-202
NASA-Lewis Research Center
21000 Brookpark Road
Cleveland, Ohio 44135
B-4
-------
Page 3
Dr. William C. Pfefferle
Leonard C. Angello
S. Mario DeCorso
Albert C. Do!bee
Dr. Anthony Cerkanowicz
Dr. Victor J. O'Brien
Dr. Clarence M. Ablow
Dr. K. C. Salooja
Frank Zimar
Dr. Irwin M. Lachman
Dr. Edmond R. Tucci
51 Woodland Drive
Middletown, New Jersey 07748
AFAPL/SFF
Wright-Patterson Air Force Base
Ohio 45433
Mail Stop A-702
Westinghouse Electric Co.
Gas Turbine Division
Lester, PA 19113
Electric Power Research Institute
3412 Hi 11 view Ave.
P. 0. Box 10412
Palo Alto, CA 94303
Exxon Research & Engineering Co.
Government Research Labs
P. 0. Box 8
Linden, New Jersey 07036
Houdry Division, P&C
Air Products and Chemicals, Inc.
Chemicals Group
P. 0. Box 538
Allentown, PA 18105
Stanford Research Institute
333 Ravenswood Avenue
Menlo Park, Calif. 94025
Esso Research Centre
Abingdon, Oxfordshire
United Kingdom
Mechanical Product Development
Corning Glass Works
Sullivan Science Park
Painted Post, New York 14870
Specialty Ceramic Dept.
Corning Glass Works
Sullivan Science Park
Painted Post, New York 14870
Matthey Bishop, Inc.
Malvern, PA 19355
B-5
-------
Page 4
Dr. A. S. D'Allessandro
Michael Bak
B. E. Enga
Dr. Geoffrey J. Sturgess
Dr. Frediano A. Bracco
Dr. Nancy D. Fitzroy
Walter Jasionowski
Andrew J. Szaniszlo
Dr. Donald M. Sowards
W. Steven Lanier
Will Dodds
Matthey Bishop, Inc.
Malvern, PA 19355
Williams Research Corporation
48080 Maple Road
Walled Lake, Michigan 48080
Johnson Matthey Research Centre
Blount's Court
Sonning Common
Reading RG4 9NH
United Kingsom
Pratt & Whitney Aircraft
400 Main Street
East Hartford, Conn. 06108
Dept. of Mechanical and Aerospace
Engineering
Princeton University
Princeton, New Jersey 08540
Bldg. 500, Room 224
GTD.:;
General Electric Company
Schenectady, New York 12345
Institute of Gas Technology
3424 South State Street
Chicago, Illinois 60616
Mail Stop 60-4
NASA-Lewis Research Center
21000 Brookpark Rd.
Cleveland, Ohio 44135
Industrial Chemicals Dept.
Chestnut Run Laboratory
E.I. DuPont de Nemours & Co.
Wilmington,Del aware 19898
MD-65
Industrial Environmental Research
Lab.
Environmental Protection Agency
Research Triangle Park, N.C. 27711
General Electric Company
Cincinnati, Ohio 45215
B-6
-------
Page 5
Warren Bunker
John C. Bonacci
Dr. T. J. Rosfjord
Dr. Eugene Szetela
Dr. Joshua A. Bowen
Warren P. Richard
Dr. E. A. DeZubay
Dr. R. D. Matthews
Department of Energy
Washington, D.C. 20545
Engelhard Industries
Menlo Park
Edison, N.J. 08817
Mail Stop 18
United Technologies Research Ctr
400 Main Street
East Hartford, Conn. 06108
United Technologies Research Ctr
400 Main Street
East Hartford, Conn. 06108
MD-65
Industrial Environmental Research
Lab.
Environmental Protection Agency
Research Triangle Park, N.C. 27711
East Shore Drive
Lake Toxaway, N.C. 28747
Westinghouse Research & Development
Center
Beulah Road
Pittsburgh, Pennsylvania 15235
Dept. of Mechanical & Industrial
Engineering
University of Illinois at Urbana-
Champaign
144 Mechanical Engineering Building
Urbana, 111. 61801
B-7
-------
Page 6
Howard Lowenstein
Daniel Carl
Peter Walsh
Richard Sederquist
Ned R. Baker
Industrial Products Dept.
Ceramic Products Division
Corning Glass Works
Corning, N.Y. 14830
Westinghouse Electric Co.
Gas Turbine Division
Lester, PA 19113
Dept. of Mechanical & Aerospace
Engineering
Princeton University
Princeton, New Jersey 08540
Power Systems Division,
United Technologies
P.O. Box 109
South Windsor, Conn. 06074
Institute of Gas Technology
3424 South State Street
Chicago, Illinois 60616
B-8
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/7-79-038
2.
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
Proceedings: Third Workshop on Catalytic
Combustion (Asheville, NC, October 1978)
5. REPORT DATE
February 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
John P. Kesselring, Compiler
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Acurex Corporation/Aerotherm Division
485 Clyde Avenue
Mountain View, California 94042
10. PROGRAM ELEMENT NO.
EHE624A
11. CONTRACT/GRANT NO.
68-02-2611, Task 30
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Proceedings: 5-11/78
14. SPONSORING AGENCY CODE
EPA/600/13
15. SUPPLEMENTARY NOTES
2235.
project officer is G. Blair Martin, MD-65, 919/541-
i6. ABSTRACT Tne proceedings document the major presentations at the Third Workshop
on Catalytic Combustion, in Asheville, North Carolina, October 3-4, 1978. Spon-
sored by the Combustion Research Branch of EPA's Industrial Environmental Re-
search Laboratory—Research Triangle Park, the workshop served as a forum for
the presentation of results of recent research in the areas of catalyst and catalytic
combustion system development. The workshop provided industrial, university,
and government representatives with the current state-of-the-art in the application
of catalyst systems for pollution control and performance improvement. Applications
include firetube and watertube boilers and gas turbines for utility, industrial, auto-
motive, and aircraft systems.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
Pollution
Combustion
Catalysis
Catalysts
Boilers
Gas Turbines
Automotive Engin-
eering
Aircraft Industry
Utilities
Industries
Pollution Control
Stationary Sources
Catalytic Combustion
13B
21B
07D
13A
13G
13F
01C
05C
8. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (ThisReport)'
Unclassified
21. NO. OF PAGES
590
20. SECURITY CLASS (Thispage)
classified
Unc
22. PRICE
EPA Form 2220-1 (9-73) B_9
*U.S. GOVERNMENT PRINTING OFFICE: 1979-61+0-013- 418 9REGION NO. 4
------- |