Advanced Modeling of Incineration of Building Decontamination Residue


AWMA 98th Annual Meeting; Minneapolis, MN; June 21-24, 2005

Advanced Modeling of Incineration of Building Decontamination Residue

Martin Denison, Chris Montgomery, Wei Zhao, Mike Bockelie, Adel Sarofim
Reaction Engineering International

77 West 200 South, Suite 210
Salt Lake City, UT 84101 USA
deni son@reacti on-eng. com http://www.reaction-eng.com

Paul M. Lemieux
US EPA Office of Research and Development
109 T.W. Alexander Drive; E305-01
RTP,NC 27711 USA

ABSTRACT

In this paper we present recent development of the component models for a pilot scale rotary kiln
simulator for the incineration of building materials. A transient zonal model approach for use
with a computational fluid dynamics (CFD) model is presented. Comparisons are made between
the model and experimental data. The models predict complete destruction of the biological
agent that remains in the building material matrix when the incinerators and afterburners are
operated as per standard operating.

NOMENCLATURE

A

Arrhenius preexponential constant, 1/s

CP

Specific heat, J/kg/K

c

Z value form preexponential constant, 1/s

ER

Arrhenius activation energy to gas constant ratio, K

f

Mass fraction of 7th stream

h

Enthalpy, J/kg

k

Thermal Conductivity, W/m/K

m

Mass flow rate, kg/s

Oc

Convective heat transfer, W

Or

Radiative heat transfer, W

t

Time, s

V

Volume, m3

y

Viable spores

Y

Devolatilization yield

Z

Z value, K

T

Temperature, K

P

Density, kg/m3

l

-------
AWMA 98th Annual Meeting; Minneapolis, MN; June 21-24, 2005

NOTATION

CFD

CWA

Computational Fluid Dynamics
Chemical Warfare Agents
Deactivation Furnace System
Sarin

Mustard Agent

Reaction Engineering International
VX nerve gas

DFS

REI

INTRODUCTION

The EPA National Homeland Security Research Center (NHSRC) has initiated research to
develop better methods, technologies, and guidance relating to building decontamination in the
event of a terror attack. Incineration is one of the primary technologies being studied for the
disposal of contaminated building materials. As part of that effort, Reaction Engineering
International (REI), under contract to EPA, is developing computer simulation tools for
analyzing incinerators processing building material contaminated by biological weapon (BW)
agents or chemical weapon (CW) agents. These agents are collectively termed
chemical/biological (CB) agents. The model results are being compared against pilot-scale data
collected by EPA to characterize the behavior of BW agents, as would be found in a structure,
bound in various matrices and materials, typical of an office building/environment.

The models being developed will take advantage of a Department of Defense (DoD) Small
Business Innovative Research (SBIR) program recently completed1"4 by REI to develop a suite
of models for conducting detailed simulations of chemical demilitarization incinerator operation.
As part of that SBIR project, computational chemistry methods were used to develop detailed
chemical kinetic mechanisms to describe the decomposition of CW agents of mustard (HD), and
nerve agents GB and VX.

The purpose of the modeling effort is to provide facilities and permitting agencies to perform
"what if' type scenarios to assess the effect of operating conditions on destruction of CB agents.
In addition, by being able to simulate incinerator failure modes, risk assessors can analyze
situations such as the impact of power loss and system shutdown while CB agents are in the
incinerator.

Ultimately, the models are to be used to extrapolate heat transfer, fluid dynamics, and chemical
kinetic behavior to characterize the full-scale effect. Models for full-scale incinerators under
development include an oxygen starved medical waste incinerator and a commercial scale rotary
kiln incinerator. As part of this effort a model has been developed for the pilot-scale rotary kiln
simulator at the EPA NHSRC. The development of this laboratory kiln model is the focus of this
paper. The modeling approach for the incinerators is discussed. The kinetic models for BW
agents are developed and comparisons are made with the laboratory data. Finally, comparison
with laboratory scale kiln data is made.

-------
AWMA 98th Annual Meeting; Minneapolis, MN; June 21-24, 2005
ROTARY KILN SIMULATOR DESCRIPTION

Testing was performed at the EPA's Rotary Kiln Incinerator Simulator (RKIS) facility located in
Research Triangle Park, NC. The RKIS (shown in Figure 1) consists of a 73 kW (250,000
Btu/hr) natural gas-fired rotary kiln section and a 73 kW (250,000 Btu/hr) natural gas-fired
secondary combustion chamber (SCC). Following the SCC is a long duct that leads into a
dedicated flue gas cleaning system (FGCS) consisting of another afterburner, baghouse, and wet
scrubber. The RKIS is equipped with continuous emission monitors (CEMs) for oxygen (02),
carbon dioxide (CO2), carbon monoxide (CO), nitrogen oxides (NOx), and total hydrocarbons
(THCs). A series of Type-K thermocouples monitor the temperature throughout the system.
For the data reported here, the rotary kiln combustion air was flowing at a rate of 82.4 sm3/hr
(2909 scfh) and the burner natural gas fuel was flowing at a rate of 5.66 sm3/hr (200 scfh). The
static pressure in the rotary kiln section was maintained at -0.05 in. w.c. The typical gas
residence times are 2 seconds in the kiln, 3 seconds in the transition between the kiln and
secondary combustion chamber, and 7 to 8 seconds in the secondary combustion chamber. The
typical residence time for the solid matrix material is 10 minutes. This facility has been used for
testing a variety of wastes5.

Secondary Combustion Chamber

Figure 1. Rotary Kiln Incinerator Simulator.

MODELING APPROACH

In the kiln, the periodic loading of building materials results in an inherently time dependent
operation that must be adequately captured in the model. Modeling the operation with a true
transient computational fluid dynamics (CFD) simulation would require excessive computing
resources. To better represent the time dependent nature of the kiln in an efficient manner, we
use a combination of a transient "zonal" model and a steady state CFD model. The transient
zonal model captures the transient effect of building matrix combustion, on the furnace
temperature and overall gas composition. A steady state 3D CFD model is used to compute the
local mixing and destruction efficiency for a prescribed instant in time. The BW agent
destruction can be tracked within the matrix or traced along streamlines for spores released into

-------
AWMA 98th Annual Meeting; Minneapolis, MN; June 21-24, 2005

the gas. The release rate of spores is not calculated in the model but rather the log reduction
along gas phase streamlines of any spores that may be released. Chemical agent vaporization is
modeled and destruction is assumed to occur in the gas after release. A key aspect is that the
matrix combustion rate and wall thermal behavior determined with the transient zonal models are
used to define the boundary conditions, required by the CFD model. This paper focuses on the
zonal model and BW destruction in the matrix.

Transient Kiln Zonal Model

The transient zonal model consists of a gas phase zonal combustion submodel and a matrix
combustion model. The gas temperature and composition of each zone is assumed uniform.
Ordinary differential equations (ODE's) are integrated over time representing the material and
energy balances of the gas phase of each zone and the agent within each individual projectile.
The following ODE governs the material balance in each zone:

i 1 ( XTT • X r / 1 \

- fy-r - Woutfr + Z mtnfiM ( l)

dt pV

where f is the mass fraction of the ith stream, V is the zone volume, and p is the gas density. The
mass fraction of the various streams represents the elemental composition of the gas for use in
equilibrium calculations. The streams tracked are the fuel such as natural gas, dry ash-free matrix
fuel, air, and water. The mass flow rates, min, include the off-gas of water or dry ash-free (DAF)
fuel. Mass conservation for each zone is expressed as:

^ ~ • (2)

1? dt

The following gas enthalpy equation for each zone is also integrated:

dh 1 LhV±.ri,Mh+ji
-------
AWMA 98th Annual Meeting; Minneapolis, MN; June 21-24, 2005

rate is much smaller than the boiling evaporation rate and is neglected in the drying process. The
devolatilization process is modeled as a two-step reaction:

Cdaf -*-> Yt Vol + (1 - V,.) Char, / = 1,2

For the results presented in this paper, k2 is assumed to be 0 therefore producing a fixed volatile
yield. A global model based on the first order oxygen and total matrix surface area is employed
for modeling of char oxidization rate.

The model is advanced through time using a 4th order Runge-Kutta time stepping scheme with
adaptive time step size control7 to maintain prescribed accuracy tolerances. At each time step the
zones are updated with upstream zones first. The zone containing the outlet duct to the
afterburner is updated last. At each time step a chemical equilibrium calculation is performed to
obtain the zone furnace gas temperature and composition from the stream mass fractions and
enthalpy.

Output from the transient model includes as a function of time the temperature distribution
within each matrix piece, matrix volatile release and char production rate, char oxidation rate,
furnace gas temperature and gas composition. The vaporization rate and the burner, drying,
matrix combustion, and airflow rates are used as inputs (boundary conditions) to the 3D CFD
model of the kiln.

BW Agent Thermal Destruction Kinetics Development

Destruction of BW agent with temperature involves complex processes of microbiology. In order
to have a predictive tool to provide spore destruction trends within an incinerator a kinetic model
must be developed. In this paper the destruction is assumed to be first order

4- = -^ (5)

where y represents the number of viable spores and k is the kinetic rate constant which is
function of temperature. The log reduction of viable spores for given time of exposure at
temperature T for a first order rate is then given as

1 J 1 lOO k(T)t

log reduction = log10 — = (6)

{y J lnio

where yo is the initial amount of viable spores. The form of the rate constant as a function of
temperature is next needed. The concept of the Z value from microbiology8 is used as guidance
in providing this form. If D is the time required at a given temperature to achieve a log reduction
in organisms, then Z is defined as the number of degrees change in temperature needed to change
the D value by a factor of 10. For example, if Z is 10 degrees F, and if D is 5 minutes at 130F,
then D will be 50 minutes at 120F. It follows that Z = T2 '/) if

( v \

k(T2)D k(T,)\0D

log reduction = log10 — = = —. (7)

y J

lnIO lnIO

-------
AWMA 98th Annual Meeting; Minneapolis, MN; June 21-24, 2005

If the Z value is assumed constant the following form will satisfy Eq. 7.

In 10.

k = Ce

(8)

Here C is the preexponential constant. Alternatively, an Arrhenius form may be borrowed from
chemical kinetics.

k = Ae~™ (9)

where E is the activation energy, R is the gas constant and A is the preexponential constant. This
form yields a Z value that is a function of the two temperatures

E/R -

(10)

Parameters for both forms of the rate constant have been fit to data obtained by the EPA
NHSRC9. The fit for the building material types of ceiling tile and wallboard is shown in Figures
2 and 3, respectively. Table 1 lists the fit parameters. Since the destruction is assumed first order
all the fit curves are straight lines that pass through the zero time at zero log reduction. The
ceiling tile data exhibit more scatter.

a 300
¦ 400

~ 500

• 600

300

400

500

600

- -300

- -400

- -500

600

F Data
F Data
F Data
F Data
F Z value
F Z value
F Z value
F Z value
F Arrhenius
F Arrhenius
F Arrhenius
F Arrhenius

1000

2000
seconds

3000

4000

Figure 2. Log reduction versus time for B. subtilis on ceiling tile at various initial
temperatures.

-------
AWMA 98th Annual Meeting; Minneapolis, MN; June 21-24, 2005

9 7

8 -

+-»

26 -
u>
c

= 5-

^4 -

D
T3

¦- 3 -

u>
o

-*2 -

1 -
0 -

0 1000 2000 3000 4000 5000

Time (s)

Figure 3. Log reduction versus time for B. subtilis on wallboard at various initial
temperatures.

Table 1. Kinetic BW Destruction Parameters.

Z value

Arrhenius

C, Ms Z, K

A, 1/s E/R, K

ceiling tile
wallboard

1.534E-05 158.799
6.464E-05 280.803

30.01 3568.3
0.435 2354.7

RESULTS

The objective of the development of the pilot scale kiln simulator model is to provide a means of
calibrating the model for use in scaling up to evaluate larger systems that could be used in
building residue incineration following a terror attack. Data are much more accessible in the pilot
scale facility than would be in a commercial scale unit. Data were collected for processing nylon
6.6 carpet bundles in the unit by EPA and made available for testing the model. The bundles
were fed about every 10.5 minutes. The bundles were approximately 50% water.

The user interface of the software has a variety of matrix property inputs as well as operating
condition inputs. Matrix property inputs include matrix density, dimensions, thermal
conductivity, specific heat, and surface emissivity. The analysis of the matrix is also input which
includes ultimate analysis, ash content, water content, and higher heating value. Devolatilization
and char oxidation parameters are also input. The parameters adjusted to produce agreement with

~ 400 F Data
¦ 500 F Data
~ 600 F Data

400 F Z value

500 F Z value

600 F Z value

400 F Arrhenius

500 F Arrhenius

600 F Arrhenius

-------
AWMA 98th Annual Meeting; Minneapolis, MN; June 21-24, 2005

the temporal data set were the matrix surface emissivity and devolatilization rate parameters of
the matrix.

Figures 4 and 5 show the comparison between the model calculations and the measured data of
the kiln exit temperature and the SCC exit oxygen. Two successive carpet bundle dumps were
modeled. The time of the first dump corresponds to zero minutes. The drop in oxygen is the
result of matrix combustion. Both the model and the data show slight increases in temperature of
the second dump profile compared to the first. This is the result of wall refractory being heated.

2000

1500

Measured Data
— Model

5 10 15
Time, min.

Figure 4. Comparison between measured data and model calculations of the temporal
variation of the kiln exit temperature.

X
LU

o
o

(f)

Time, min

Measured Data
Model

Figure 5. Comparison between measured data and model calculations of the temporal
variation of the SCC exit oxygen.

-------
AWMA 98th Annual Meeting; Minneapolis, MN; June 21-24, 2005

The agreement between the model and the measured data is good following the adjustment of the
two parameters. The emissivity of about 0.5 produced agreement with the time of completion of
drying and the onset of matrix devolatilization. The devolatilization parameters thus obtained
where subsequently used in a calculation and compared with weight loss data from thermal
gravimetric analysis (TGA) for the nylon 6.610. The calculation captured the correct overall
weight loss over the temperature ramp period to within 20%. This agreement suggests that other
input parameters are a fairly good representation of the actual matrix. More data such as actual
temperatures within the matrix would provide a better calibration.

Since equilibrium is used in the gas combustion of the zonal model the CO is severely
underpredicted. However other major species such as C02 are well predicted by the approach.

Figure 6 shows the spatial minimum and maximum temperatures and water mass within the first
matrix as a function of time. Large gradients in temperature exist, in particular when water is
present. The outer surface temperature rises very rapidly after the water has vaporized from the
outer regions of the bundle.

In addition to integrating the governing ordinary and partial differential equation describing the
combustion and heat transfer processes, the destruction of the BW agent within the matrix is also
integrated. Figure 7 shows the log reduction of the BW agent as a function of time using both the
Z value and Arrhenius form for the rate constant (Eqs. 10 and 11). The kinetic parameters used
were those of B subtillis in ceiling tile. These parameters were used since at this writing, data for
BW in carpet were not available. Although both forms provide a reasonable fit to bench scale
data at low temperatures (see Figures 2 and 3), the difference between the Z value and Arrhenius
predictions become more severe at very high temperatures were there is no data. It is clear that
full destruction quickly ensues once a sufficient temperature has been reached. The Arrhenius

1.20

2500

2000

-------
AWMA 98th Annual Meeting; Minneapolis, MN; June 21-24, 2005

16.0

3
T3

O)
O

12.0

10.0

14.0

4.0

0.0

6.0

8.0

2.0

Z value
Arrhenius

0.0 1.0 2.0 3.0 4.0 5.0
Time, min

Figure 7. Log reduction of the biological weapons agent as calculated by the model.

expression yields the more conservative result. For the larger residence times associated with
spores trapped within the matrix both forms predict full destruction. However, when tracking the
destruction for spores that become entrained in the gas the resulting residence times would be
much smaller, although there would be no material heat transfer barriers for spores in flight. The
difference in the spatial minimum and maximum temperatures within the matrix shown in Figure
6 is indicative of such material heat transfer barriers.

CONCLUSION

This paper has demonstrated that zonal and CFD models of the laboratory scale kiln can be
constructed and provide useful information on the physical processes that affect furnace
performance in terms of microbiological destruction efficiency and operability. The pilot scale
rotary kiln simulator model has shown useful for calibration with easily attainable data. A
comparison has been made between the model and experimental kiln data obtained to provide
calibration, which will permit use with full-scale incinerator models. Only data concerning
operational characteristics were used in the comparison. No experimental data on biological
agent destruction within the kiln were available at this writing. The models predict complete BW
agent destruction for agent that remains within the matrix. The models may also be useful in
simulating incineration system upset conditions and failures that could lead to an agent release,
so that appropriate design and operational modifications can be made to mitigate such
occurrences.

-------
AWMA 98th Annual Meeting; Minneapolis, MN; June 21-24, 2005
REFERENCES

1. Denison, M.K., Montgomery, C.J., Sarofim, A.F., Bockelie, M.J., Magee, R., Gouldin, F.,
McGill, G., "Detailed Computational Modeling of Military Incinerators," 20th International
Conference On Incineration and Thermal Treatment Technologies, Philadelphia, PA, (May
2001).

2. Denison, M.K., Montgomery, C.J., Sarofim, A.F., Bockelie, M.J., Webster, A.G., and
Mellon, R.J., "Advanced Computational Modeling of Military Incinerators," 21st
International Conference On Incineration and Thermal Treatment Technologies, New
Orleans, LA, (May 2002).

3. Denison, M.K., Montgomery, C.J., Sarofim, A.F., Bockelie, M.J., and Webster, A.G.,
"Computational Modeling of a Chemical Demilitarization Deactivation Furnace System,"
22nd International Conference On Incineration and Thermal Treatment Technologies,
Orlando, FL, (May 2003).

4. Denison, M.K., Sadler, B.A., Montgomery, C.J., Sarofim, A.F., Bockelie, M.J., and Webster,
A.G., "Computational Modeling of a Chemical Liquid Incinerator Chamber," 23nd
International Conference On Incineration and Thermal Treatment Technologies, Phoenix,
AZ, (May 2004).

5. Lemieux, P.M., "Pilot-Scale Combustion of Building Decontamination Residue," A&WMA's
98th Annual Conference & Exhibition; Minneapolis, MN (June 21-24), 2005

6. Perry, R.H., Green, D.W. (Editors) "Perry's Chemical Engineers' Handbook", Section 12,
McGraw-Hill, New York, 1997.

7. Press, W.H., Flannery, B.P., Teukolsky, S.A., and Vetterling, W.T., Numerical Recipes in C,
Cambridge University Press, 1988.

8. Prescott, L.M., Klein, D.A., Harley, J.P., Microbiology, 5/e. McGraw Hill, 2002.

9. Lee, C.W., Wood, J.P., Betancourt, D., Linak, W.P., Lemieux, P.M., Novak, J., Griffin, N.,
"Study of Thermal Destruction of Surrogate Bio-contaminants Adsorbed on Building
Materials," A&WMA's 98th Annual Conference & Exhibition, Minneapolis, MN, (June
2005).

10. Lemieux, P.M., Personal Communication, (2004).

-------