United States
                    Environmental Protection
                    Agency
Industrial Environmental Research
Laboratory
Research Triangle Park NC 27711
                    Research and Development
EPA-600/S7-83-048 Dec. 1983
f/EPA         Project  Summary
                    Shell   NOX/S02  Flue  Gas
                    Treatment  Process:  Pilot  Plant
                    Evaluation
                    J.B. Pohlenz and A.O. Braun
                      The Shell Flue Gas Treatment process
                    was evaluated in a pilot-scale test for
                    reducing simultaneously the emissions
                    of sulfur dioxide  (SO2) and nitrogen
                    oxides (NO,) from flue gas produced in
                    a coal-fired utility boiler.  Flue gas
                    leaving the economizer at about 400°C
                    with a full load of fly ash is contacted
                    with a copper-on-alumina acceptor/cat-
                    alyst in a parallel passage reactor. SOz
                    is removed by reaction with the acceptor,
                    and the selective reduction of NO, with
                    NH3 is catalyzed.  Regeneration of the
                    acceptor at 400°C with  hydrogen
                    releases the sulfur as concentrated SOz
                    which can be processed into a marketa-
                    ble  byproduct.  Fly ash was found to
                    increase the diffusional resistance to
                    mass transfer of SO2 and SOs to copper
                    oxide, requiring an increase of 25% of
                    acceptor/catalyst  over that  for  clean
                    service to achieve 90% SO, reduction.
                    NOx reduction with NH3 was found to
                    be non-selective at the beginning of
                    acceptance due to the temperature
                    increase accompanying copper oxidation,
                    requiring a further increase in acceptor/
                    catalyst of  18% to obtain  90% NOx
                    reduction simultaneously. Estimates of
                    capital and operating costs increased by
                    12 and 21%, respectively, due to the
                    effect of fly ash with further increases
                    of 7 and 5% to compensate for the
                    combustion of NH3 at initiation of
                    acceptance.  The objective performance
                    of 90% simultaneous NO,/SO« reduction
                    was achieved; however, the demonstra-
                    tion was not continued for the initially
                    planned 90 days.
                      This Project Summary was developed
                    by EPA's Industrial Environmental
                    Research Laboratory. Research Triangle
                    Park, NC, to announce key findings of
the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).

Introduction
 In  1973, UOP constructed a 0.5 MW
pilot unit at  the coal-fired Big Bend
Station of Tampa  Electric Company
(TECO), based on the Shell Flue Gas
Treatment (SFGT) process for removing
sulfur oxides (SO*). This process was in
full  scale operation (40 MW) on a boiler
fired with fuel oil in the Showa Yokkaichi
Sekiyu (SYS) refinery near Yokkaichi,
Japan, but its performance in coal-fired
service had not been determined. The
TECO  pilot plant was  operated as a
desulfurization unit by UOP during 1974-
1976, during  which it was established
that the SOX accepting material (copper
oxide supported on alumina) is stable in
this service and that the reactor design
(parallel passage) can function effectively
on flue gas containing a full load of coal
fly ash.
  Between 1973 and 1975, laboratory
experiments and field tests (SYS) proved
the activity of copper as the oxide or the
sulfate to catalyze the selective reduction
of nitrous and nitric oxides with ammonia
to nitrogen and water. The feature of NOx
reduction,  simultaneous with SO,reduc-
tion, was successfully added to the SYS
operation in 1975.
  In September 1976, the EPA  invited
proposals (RFP No. DU-76-A122) for the
operation of pilot plants to treat flue gas
from coal-fired utility boilers  for the
reduction of NO»and/or the simultaneous
reduction of NO, and SO,. TKe work was
to be performed on a scale of about 0.5
MW, and  the scope was intended to

-------
provide information to permit a technical
and economic evaluation of the treatment
process that was proposed.
  In  the  response to EPA's RFP, UOP
proposed to demonstrate the SFGT
technology at the TECO pilot plant.
Coupled  with the combined  NO»/SOx
reduction in the commercial operation on
oil in Japan and the desulfurization pilot
plant work on coal at TECO, the simulta-
neous  NOx/SOx operation at TECO on a
pilot scale would provide the last develop-
mental stage  before performance in a
full-scale application to a coal-fired boiler
could be represented and guaranteed.
  Consequently, EPA awarded Contract
68-02-2676 to UOP  early in 1978. The
technical objective of the contract was
simultaneous reduction of NOX and SOx
by 90% each in a pilot unit when treating
flue gas from a coal-fired utility boiler at a
rate equivalent to 0.5 MW. The demonstra-
tion  period of continuous operation was
to be a  minimum of 90days and was to in-
clude 75 days in compliance with the 90%
reduction requirement of each pollutant.

Process Description

  The  underlying  chemistry of the
process,  described extensively  in the
technical  literature (1, 2,  3,  4), is
summarized here.

Technical Basis
  Copper, as  copper oxide or copper
sulfate, catalyzes the selective reduction
of NO  and NO2  (NOx) in flue gas to
nitrogen and water with ammonia (NH3)
at temperatures of 350-450°C. Conver-
sions and efficiency (ammonia utilization)
are high, resulting in low concentrations
of both NO and NH3 in the treated gas.
  If flue gas containing SOX, NOx, and the
reductant  ammonia  is processed over
copper  at 400°C, the copper is converted
first to the oxide, then to the sulfate, and
NOx reduction begins. As the conversion
to copper sulfate continues, the NOx
concentration of the treated gas decreases
to a minimum value and the concentration
of SO2  increases, approaching that of the
untreated gas.
  Copper sulfate can be reduced with a
variety of fuels (e.g., Hz, CO, CH4) at
400°C, yielding a concentrated stream of
SO2  and water and-changing the copper
to elemental form.
  Thus, the copper system provides the
technical base for flue gas treatment
capable of NOX reduction, SOX reduction,
and the simultaneous reduction of both
(Figure 1). It offers the potential of a dry
process,  without waste products, and
with modest energy  requirements. The
technical base  has been  applied (i.e.,
reduced to practice) in several forms: first
by type (NOx-only, SOx-only, and com-
bined NOx/SOx), and then by function (re-
search/development,  pi lot/demonstra-
tion, and commercial- or full-scale).

Reduction to Commercial
Practice
  The full-scale  application  of this
technology to a NOx-only process is as a
fixed-bed reactor with provisions  for
ammonia introduction and mixing. The
reactor is of special design-to accommodate
fly ash  (Figure 2).  The catalyst  is
contained in woven-wire  baskets sus-
pended in the gas stream so that the gas
flows  in the open channels between
baskets (parallel passages). The reactants
enter and the products leave the catalyst
bed  by radial diffusion. The design is
'modularized in  units  of  cells, 0.5  m
square by 1 m long, which can be stacked
one atop another for the required space-
time.
  For SOx removal, the operation is cyclic
and requires that the reactor  be  isolated
from the flue gas circuit for regeneration,
producing  a concentrate of S02 and
restoring the copper to elemental form.
Flue gas is processed continuously with
multiple reactors, at least one of which is
always in regeneration. A unit designed
for SO, removal can operate in NOx -only
reduction mode by adding ammonia and
eliminating the regeneration step. A unit
designed for SOX removal also can
operate  to provide simultaneous NOx
reduction by introducing ammonia to the
untreated flue gas during the acceptance
step.
  The performance characteristics of the
design are such that the  percent SOa
removal is high at the  initiation  of
acceptance and  "slipped"  S02  (SO2  in
treated gas) increases as  the sulfur
loading of the acceptor material increases
with acceptance time.  Control of time-
averaged percent de-SOx is obtained by
adjusting the acceptance time.  Slipped
NOx with time  is the reverse (mirror
image) of that for S02; NOX slip is high at
the beginning of acceptance and decreases
with acceptance time (see Figure 3).
  There are two ways to increase the de-
NOx performance while  holding  the SOX
removal constant at 90%. In the first, the
initial NOX slip is decreased using some
combination of: (1) ammonia  injection
time and rate, (2) incomplete regeneration
to leave  a heel of su(fated copper which
provides a catalytic function immediately
when  flue gas  is  introduced,  and  (3)
partial oxidation with air and temperature
reduction at the reactor inlet before
initiation of acceptance. In the second
method, the acceptance period is increased
with enough reduced space velocity to
offset the initial NOX slip to the desired
average level.
  This reduction-to-practice is called the
SFGT process. Commercial applications
have been  in operation since 1973
treating flue gas from various fuels not
including coal.

Equipment and Operation
  A  simplified process flow of the test
unit used in this program (UOP Pilot Plant
247) is  shown in Figure 4. Flue gas is
taken from boiler No.  2 at Tampa Electric
Company's Big Bend Station at the outlet
of the economizer through approximately
50 m of insulated 14-in. (35.56 cm) pipe;
                1 1/2 Oz
                ^-	
 (a+1)H2—+y.
          Jf    aCuSQt
       aSOz    d ~ <*> CuO
    (a + 1) H20
   Regen. Off-Gas

Figure  1.    Basic chemistry.
      Flue Gas
       Outlet
                         Flue Gas
                         Passage

                         Acceptor
                         Material

                         Gauze Layer
Figure 2.
         Flue Gas
           Inlet
Principle of parallel passage
reactor.

-------
   450
   400
           i     I     i     r    r
            At Reactor Inlet:
I
I
c

-------
pressure testing of the reactor, cell-stack
bypassing, ammonia and flue gas mixing,
and distribution. Calibration of instruments
— flow meters, and temperature  and
pressure indicators — was frequent and
conformed to UOP standard pilot plant
practice.
  Each  start-up after a turnaround was
carried out in a "clean" plant, one from
which fly ash had been removed mechani-
cally as part of the  turnaround. Initial
operation used air with pollutants added.
The "air" base performance for that run
was determined in tests of 99-minute
acceptance time  (maximum  for  the
sequencer) during which the reactor inlet
and outlet concentrations of S02 and NO*
were monitored continuously yielding
"slip" curves for  these  species.  (Slip
curves are plots of pollutant concentrations
in the  treated gas  versus  time;  i.e.,
pollutants which  slipped through the
monitor.)

Process Performance Testing

  NO and NO2 were analyzed continuously
with on-line analyzers in the reactor inlet
and outlet. SOz was also monitored with
on-line  analyzers in the reactor inlet and
outlet, and in the  regeneration off-gas.
NH3/NOx  ratio was  calculated from a
metered rate of NH3 addition, a metered
rate of  flue gas, and  NOX  analysis  in
untreated flue gas. Continuous SO3 and
NH3 monitoring was attempted but never
achieved; instead wet absorption methods
were used late in the program which,  by
their nature, represent time-averages.

Results

  Eleven runs were  carried out in the
demonstration/pilot unit (UOP Pilot Plant
247). Runs 1 -6 were completed in 1974-
1976 for reduction of  SOx-only on the
same load of acceptor; the results have
been reported (5). Runs 7-10, under EPA
Contract 68-02-2676, included cyclic
tests for simultaneous NO«/SOX reduc-
tion and non-cyclic tests for NOx-only re-
duction on  sulfated  copper acceptors,
commercial preparations (SK-501 and
SC-501) furnished by Shell, and a proto-
type UOP acceptor (SOx-1 -2). Run 11 de-
monstrated the selective catalytic reduc-
tion of NO, over a catalyst in commercial
use in both Japan and the U.S. (at Flet-
cher Refining Company). The catalyst (JP-
501) is  promoted vanadium on a support
resistant to sulfation. Although the test
work on Run 11 was not funded by the
EPA contract, it is included in the report to
indicate the performance for this catalyst
in coal-fired service.
  An operational summary covering
Runs 7-11  is shown in Table 1. There
were approximately 300 days of cyclic
and non-cyclic operation. For simultaneous
NOx/SOx operation, there were over
2000 cycles of 1-2 hours  cycle length.
During  the  NOx-only operation, tests
were represented by operating data
recorded approximately every 1  or 2
hours.

  Shown in  Table 1 is the performance
average for  tests on those days which
qualify as "demonstration" days in that
the unit was in test for 18 hours of each
24  hour  period  of continuous plant
operation; i.e., the ratio of tests to cycles
was equal to or greater than 0.75. There
were three such periods, one in Run 9
and two in Run 10. In each, the unit was
operated for  90% de-SOx and the de-NOx
determined  at  NH3/NO«  =  1.2-1.25.
Under these  constraints,  the  de-NOx
increased from 70 to 80 to 90%; plant
operation   was  stable   in  each
demonstration period.
  The  step-wise  improvement  in NO»
performance  was   attained   by
lengthening  the acceptance  period to
increase the cumulative NOX reduction.
This was possible by adjusting operating
conditions and capacity of acceptor  for
sulfur in order to hold cumulative de-SOx
at 90%  as  the  acceptance time was
increased from 63 to 86 to 99 minutes
(see Table 1).
  Although   the   project  performance
objective of  90% reduction of both SO*
and NOx was achieved, the demonstration
was   not  continued for  the  initially
planned 90 days.

/VOx/SOx  Operation

  During the initial phases of Run 10,
base  or reference  performance  was
established with air plus added pollutants
and  a 1-week performance evaluation
was  completed preceding  a scheduled
outage of the TECO boiler. Tests with flue
gas began with Cycle 43 and continued
for 7 days. Acceptance time was set at 85
minutes  to give  an average de-SOx of
90%, and with an NH3/NOX ratio of 1.20,
the average de-NOx was 79%.
  Following  the  boiler outage, the unit
was brought up on air only to operating
temperature, the flue gas circuit was
established through the bypass, and flue
gas was processed  through the reactor
without regeneration  to  convert
completely   the  oxide  to sulfate  in
preparation for non-cyclic NOX reduction.
The first 99 minutes of acceptance were
monitored, yielding the following results:
  Reynolds No. (NRe)*
  Inlet SOz/NOx, ppmv
  Reactor Inlet Temp., °C
  NH3/NO
  Cumulative Reduction, %
      SOz
      NOx
2000
1940/407
390
1.21

91.4
90.3
"In this work, Reynolds No. is directly proportional to
 the gas mass flow rate through the reactor.
  The  de-SOx performance  was  as
expected; i.e., the sulfur loading capacity
of the  bed at  90% cumulative removal
was 90% of the 2000 ppmv S02 entering
the reactor at 2000 Reynolds No. for 99
minutes. The NOx reduction showed that
99 minutes of acceptance is long enough
to average  down the  high  initial slip
which occurs during the first 10 minutes
into acceptance. But two other conditions
existed during this test which affect both
SO2 and NOxSlip curves; i.e., acceptance
was started with copper as the oxide and
the oxidation temperature wave traversed
the reactor so that acceptor temperature
at the  reactor inlet was about 400°C.
Laboratory  experiments with copper
oxide and  a fixed  NH3/NO« ratio  of  1.0
show that the characteristic maximum in
NOX reduction with temperature occurs at
about 350°C. At 400°C  the net relative
rate of oxidation of NH3 to reduction of
NOX has increased such that net reduction
of NOx has declined; i.e., at 400°C copper
oxide no longer selectively catalyzes the
desired reaction.  Similar laboratory
experiments with  copper in the sulfate
state show greater selectivity toward
promoting the reduction over the oxidation
reactions  up to 450°C, beyond  which
oxidation takes over.
  The reaction rates of selective catalytic
reduction of NOx or acceptance of SOX are
not very temperature sensitive at 400°C,
but NH3 is oxidized readily at 450°C in the
presence of copper oxide and competes
successfully with  copper for oxygen at
500°C+, both conditions which exist at
the entrance of the catalyst bed when flue
gas first contacts regenerated acceptor.
The time-averaged NH3  emission value
for  the first  15 minutes  (i.e., during the
oxidation period) was measured at 2 ppm;
that from 50 to 77 minutes averaged 17
ppm. Also explained is the reduced NOX
slip at initiation of acceptance when
processing  air plus pollutants,  even
though (in  steady-state operation for NOX
reduction only) performance is essentially
the same  for flue gas and  air plus
                                   4

-------
Table 1. SFGT Operation Summary, EPA Contract 68-02-2676
Run Number 7 8
Dates of Operation 6/2/79-12/17/79 2/8/80-4/19/80
Acceptor Catalyst SK-501 Blend SC-501
Cells Loaded 6 (Fresh)

KG Acceptor Loaded 40 1
GMo/es Copper Loaded 327
Total Operating Days (Cycles) 99/10747)
Days of Cyclic Air • Pollutants go
Days of Cyclic Flue Gas 9
Days of Steady State A • P 0
Days of Steady State FG 0
Total Days Plant Outage 97
Total Days TECO Outage 2
SO, NO, Demonstration
Dates of Operation
Operating Days (Cycles)
Compliance Days
Avg. Performance
DeSO,%
DeNOf/o
Acceptance Time Min
NHi/NO,
/VR.
Sulfur R,<%
R, Inlet °C
NO, Only Demonstration
Dates of Operation
Operating Days
Compliance Days
Avg. Performance
DeNO,%
NHs/NO,
/Vn.
R,Avg°C
Key Descriptor Mechanical
Shakedown

6 (Fresh)
(New Cells)
357
300
54(5341
34
14
4
2
13
20





















Process
Shakedown

9
5/9/80-8/24/80
SC-501
6 (Screened)
1 (Fresh)
471
389
89(511)
11
34
8
36
19
11.
/\
6/13/80-7/7/80
22(291)
17

904
69.0
62.8
1.24
2000
100.8
412

7/8/80-8/18/80
35
31

89.5
1.26
1998
416.0
Demo With
Shell Acceptor

10 11
9/ 1 5/80- 1 1 /3/80 11/1 2/80- 1 1 / 24/80
SO.-? -2 JP-501
6 (Fresh)

270
472
48(198)
7
26
7
7
3
4
NO PREOX PREOX
9/30/80-10/7/80 10/28/80-11/3/80
7(561 5(34)
6 4

90.0 90.5
79.0 90.3
85.9 99.0
1.20 1.23
2033 1498
95.9 96.4
388 386

10/14/80-10/20/80
7
Variable Study





Demo With
U OP Acceptor

2 (Fresh)

192
—
17
0
0
?
15
0
0





















NO, Testing
W Fletcher
Catalyst
pollutants. Copper oxidation is 10 to 20
times faster for air than for flue gas since
the oxidation rate is limited by the rate at
which oxygen is supplied to the oxidation
front.
  To  circumvent  the  NHa combustion
problem and  high initial  slip of NO,, a
preferred procedure would be to leave an
unregenerated "heel"  of  copper sulfate
in the bottom cell for  NO, redu'ction at
initiation of acceptance, but this would
be  at the  expense of  sulfur loading
capacity and acceptance time and would
not avoid the temperature wave resulting
from copper oxide formation.
  Ammonia emissions,  time-averaged
over the acceptance period, were 10-20
ppmv with an NH3/NO, ratio of 1.2. This
average was the same for the conventional
cycles with 90/80% SOx/NO, reduction
and the modified cycles  with separate
oxidation/cooling steps  in which  the
reduction was 90/90%. The NH3 required
for the increase in de-NO, with modified
cycles using the same  ratio was burned
during oxidation  in the conventional
cycles.
  Measured SOa content of the treated
flue gas, time-averaged over the accep-
tance period, was less than  3  ppmv,
indicating SOa reduction of 70-90%. This
result, consistent with the proposed
mechanism  of the S02/CuO  reaction,
was not unexpected. Copper oxide reacts
directly with SOa at acceptance conditions
and will remove SOa and S02 from flue
gas,  the latter  probably as SOa after
surface  oxidation of  adsorbed S02.
Laboratory experiments indicate that the
SOa  in the treated gas  increases with
temperature and sulfur loading and is not
dependent  on the SOa  content of the
untreated gas. At a given temperature,
the SO3 increases to a maximum as the
sulfur/copper ratio approaches unity and
then declines rapidly with higher sulfur
loadings, indicating that CuS04  does not
catalyze S02 to S03 oxidation.

NOi-Only Operation


  The performance of the fully  sulfated
acceptor  was determined  in the early
stages of Run 10 with air plus added SO,
and NO, and with the plant free of fly ash.
Later in the run after processing flue gas
for about 3  weeks in simultaneous
N0,/S0x operation,  the  acceptor was
completely sulfated  in flue gas, and de-
NO,  performance was measured again.
Recognizing  that (1) control  of the
adiabatic heaters  was  not uniform,  (2)
NHa/NO, ratios were not precise, and (3)
the oxygen content of the untreated gas
was  at times low due to fuel-rich
operation of the in-line heating units, the
NOx reduction was judged to be equivalent
to that observed  early in the  run,
indicating that the presence of fly ash has
little  effect on catalytic reduction activity.
  The  performance of the  sulfated
acceptor may be summarized as follows:

Temperature, °C
NOx Reduction,
% (6 stages)
NOx Reduction,
%/Stage
NHa Slip, ppmv
NHa/NOx
1.0
415
85
27
10
Mole Ratio
1.2
415
95
39
30

-------
  Laboratory studies of the copper-SOa
system showed that copper sulfate does
not catalyze the  S02 oxidation. It was,
therefore, expected that SOa in and out of
the reactor should be substantially the
same over sulfated acceptor. Given that
both  accuracy and precision of the
controlled condensation method for SOs
analysis  are  low,  this conclusion was
essentially confirmed with five tests on
five days showing consistently lower SOs
content in the treated gas than  in the
feed.
  In preparation for  Run 11, two cells
were loaded  with  JP-501 catalyst and
positioned in  the reactor atop three and
below two empty cells in a stack of seven.
With this arrangement, seals and gasket-
ing procedures were unchanged from the
previous run. De-NOx performance at
400°C with a gas rate of 28 standard
(16°C) cubic meters  per minute (scmm)
(2000 IMRe)  was 45, 65, and 85% at
NH3/NOx ratios  of 0.5,  0.75,  and 1.0,
respectively. The  corresponding  NH3
content inthe treated gas was 15, 25,and
50 ppmv, consistent  with little NH3 loss
due to oxidation. Further, the percent NOX
reduction was insensitive to throughput,
temperature,  and fly ash. At a ratio of
0.75, 65% NOX reduction was measured
at flue gas rates  of 21,28, and 35 scmm
(1500-2500 NRe) and decreased to 63%
as the temperature  was reduced from
400  to  325°C. Tests  with  air plus
pollutants at the beginning of the run in a
clean plant showed the same 85% NOX
reduction (61% per stage) as with flue gas
at a NH3/NOx ratio of 1.0.

Commercial Design
   UOP's proposal  to the  EPA in 1976
contained process specifications for the
design of a SFGT unit to reduce by 90%
the SOx and  NOX in the flue gas from a
500 MW, coal-fired  utility boiler when
operating at rated load. This was the basis
for an estimate of capital requirements
and first year annual operating charges.
   In  this design, flue gas  from the
economizer is fed directly to a battery of
eight reactors, six  of which process flue
gas while two are in regeneration. Fans in
series with the reactor system supply the
pressure drop for  the  entire  treater
circuit.  The fans/reactor complex is in
parallel with flue gas ducting, between
the economizer  and preheater (see
Figure  5).  With this arrangement, the
flue gas flow bypassing the treatment
system is boiler output less treater feed.
The fans are selected in size and number
so that there could  be some recycle of
treated gas at any  boiler load. The boiler
and the SFGT sections are isolated from
each other, and turndown of the treatment
unit  is continuous from zero to 100% of
rated load. The location of the treatment
system  upstream of the air preheater
permits recovery in the combustion air of
the sensible heat in thef lue gas above the
acid dewpoint temperature. This includes
reaction heat absorbed by the treated gas
and  incremental sensible heat available
due to reduction of S03 in the treater.
  The regeneration  off-gas, containing
SOz, water, slipped hydrogen, and small
amounts of CO2, CO, and ChU, is cooled,
water is condensed, and the concentrated
SO2  compressed into a gas holder from
which it is  released  on flow control to a
modified Claus.  The first stage  of the
               Claus prepares Claus feed (2 parts H2S
               and 1 part S02) from the S02 in the gas
               holder product in a thermal and catalytic
               reactor. Part of the reductant is the
               slipped H2, CO, and Cmfrom regeneration.
               The  balance  is  reducing gas supplied
               from the hydrogen manufacturing unit.
               Claus tail gas is recycled to the SFGT unit
               through the furnace of  the boiler (see
               Figure 6).
                 The feed  and fuel  to the hydrogen
               manufacturing  unit may be any low-
               sulfur light-hydrocarbon stream from
               methane through light  naphtha.  Since
               the feed/fuel requirements, expressed as
               heat of combustion per unit of hydrogen
               produced, is  constant,  the selection is
               economic; i.e., natural gas would be used
Pulverized
  Coal
                                           Stack
                     F.D. Fan
Figure 5.   Flow diagram for flue gas from steam generator, showing tie-in for flue gas
           treatment.
Producing Gas
Supply Section
Reactor
Section
Flow Smoothing
Section
	 1 	 1 	
Work-Up
Section
	 1
      Steam
       Light
    Hydrocarbon
    Reforming
Reactor
Section
 Gasholder
Compressor
Modified
 Claus,
                    NH3
                          Flue
                          Gas
Figure 6.   Block diagram of SFGT system.

-------
 in the current market. The hydrogen plant
 has one stage of shift  conversion and
 pressurized swing adsorption  for CO2
 removal.
  The process specificationsforthe SFGT
 system were developed with reaction
 kinetics applicable to flue gas containing
 low concentrations of fly ash;  e.g., flue
 gas from refinery fuel gas and fuel oil.
 Qualitatively it was known that fly ash in
 the interstitial space and pore structure of
 the acceptor would increase diffusional
 resistance and reduce acceptance time
 and sulfur loading. To  compensate for
 this effect, the amount  of acceptor was
 increased by 20%; i.e., the number of
 stages was increased from five to six.
 From NOx-only and combined  NOx/SOx
 tests in laboratory reactors (isothermal), it
 was estimated that 90% NO* reduction
 could be achieved with a  NHs/NOx ratio of
 1.0. The design specifications prepared
 for the proposal are summarized under
 Case 1 of Table 2.
  The design specifications of  the pro-
 posal were modified in two steps. In the
 first,  the reaction  kinetics for SOx
 reduction were adjusted from "Clean" to
 "Dirty" service using the data generated
 in the demonstration tests of Run 9. The
 results are shown in Case 2 of Table 2.
 To retain the same cycle (acceptance time
 and regeneration time), treated  gas is
 recycled (10%) and parallel passage
 Reynolds  No.  reduced to  1860. The
 reduction of S02 and  NOx across the
 reactor is 89 and 69%; and across the
 system, 90 and 71%.
   In the second modification,  the  bed
 length is increased by a  stage containing
 the JP-501 NOx reduction catalyst tested
 in  Run  11.  This stage, placed  at the
 reactor inlet, reduces the NOX concentra-
 tion at the inlet of the acceptor stages to
 35% of that in the reactor feed, giving a
 reduction of 89% across the reactor [65 +
 (35 x 0.69)] and 90% across the system.
   From  these specifications, given in
 Table  2, Case  3, economics were deve-
 loped. Capital and operating costs
 increased by 12 and 21%, respectively,
 due to the effect of fly  ash, and further
 increases of 7 and 5% in  capital  and
 operating costs  compensate for the
 combustion of Nhb at  the initiation of
 acceptance.

Conclusions

  Within the constraints under which this
project was carried out, a number of
conclusions  can be  drawn which are
consistent with and confirmed by the test
 lata obtained.
  These constraints include:
  • The flue gas processed was derived
     from the combustion  of blends of
     coal from Western Kentucky, Poland,
     and Australia in a Riley-Stokerturbo
     (wet bottom) boiler of approximately
     400 MW capacity; the sulfur content
     of the fuel was about 3%.
  • The slipstream of flue gas was taken
     between the economizer outlet and
     air preheater inlet,  transported at
     high velocity to the demonstration
     plant to avoid saltation, and heated
     by in-line combustion of propane to
     replace  heat losses. Approximately
     1600 NmVh (ca. 0.5 MW equivalent)
     was processed for SOX and/or NOX
     reduction.
  • The on-line analyzers monitoring
     NOx/SOa into and out of the reactor
     provided accurate, reproducible,
     continuous measurements, both for
     simultaneous NOx/SO2 and  NOx-
     only operation. The NH3 and SOa
     analyses  were based on  time-
     averaged values from wet chemical
     methods.
  • The usual range of S0» in flue gas
     was 2000-2500 ppmv; that of NOx,
     250-400 ppmv.
  • Three acceptor/catalysts with sup-
     ported copper were used in the five
     runs which constituted the opera-
    tional  phases of the program. The
    first two, SK-501 and SC-501,
    represented different commercial
    lots. The third,  SO«-1-2, was a
    prototype not in use commercially,
    with increased capacity for sulfur.
    JP-501, a commercial selective NOX
    reduction catalyst, was given a brief
    process evaluation in the last run.
  • For desulfurization, process time
    was composed of a series of cycles,
    each of which consisted of accept-
    ance,  purge, regeneration,  and
    purge.  Such a cycle is referred to as
    a standard or conventional cycle. For
    the final stages of the simultaneous
    NOx/SOx program,  the standard
    cycle was modified to  remove  the
    heat released from copper oxidation
    before initiating acceptance. This is
    defined as a modified cycle. Accept-
    ance was upflow; regeneration,
    cooling, and purging were downf low.
  • The parallel-passage reactor was used
    exclusively throughout the program.

Conclusions are:
  1. With the copper acceptor/catalyst,
    pollutant reduction exceeding 90%
    can be  obtained for either SOxJSOa +
    S02) or NOx (NO + N02), the former as
    a  non-steady state  operation with
    standard  cycles, the latter  as a
 Table 2.   Design Specifications, SFGT System
 Nominal MW of Boiler

 Flow Gas from Economizer

 Rate. Nm3/h x W6
 Temperature, °C
 SO* vol-%
 No» vol-%

 SFGT Reactors

Feed Rate, Nm3/\\ x 1CT6
 Temperature °C
SO*, vol-%
NO* vol-%
Recycle, %

 Cell Reynolds No.
Bed Length, m
Reactor Diameter, m
Reactors, A ccepting/Regenerating
Acceptor/Catalyst, kg/reactor
Acceptance Time, minutes
Acceptor Loading, mot S/mol Cu
 % cte-SO,
% de-NOt

SFGT System

System AP, cm HZO
% de-SOt
                       500
                      1.657
                       400
                     0.2580
                     0.0634
Case 1
1.657
400
0.2580
0.0634
0
Case 2
1.823
400
0.2368
0.0593
10
Case 3
1.823
400
0.2368
0.0582
10
2,OOO
6
7.4
6/2
54.000
66.0
0.74
90.0
90.0
1,860
6
8.0
6/2
68,000
66.0
0.56
89.0
69.0
1,860
7
8.0
6/2
80,000
66.0
0.56
89.0
89.0
              44
             90.0
             90.0
 30
90.0
71.0
 32
90.0
90.0

-------
       continuous operation with sulfated
       acceptor as a  selective  reduction
       catalyst.
     2. For simultaneous reduction of SOx/
       NOX with cyclic operation using
       standard cycles, and  with  cycle
       time  adjusted for 90% de-SOx,
       time-averaged de-NOx was limited
       to 80% at a NH3/NOx ratio of 1.2.
     3. For simultaneous reduction of
       SOx/NOx with modified cycles,
       timed for 90%de-SOx, time-averaged
       de-NOx increased to 90% at a
       NHs/NOx ratio of 1.2.
     4. Including oxidation of copper, with
       the accompanying temperature
       front,  into the  acceptance  mode
       may result in a net production of
       NOx from NH3  at the beginning of
       oxidation.
     5. Time-averaged over  the entire
       acceptance period, the  slipped  NH3
       at NH3/NOx =1.2 was 10-20 ppmv.
     6. Copper oxide  reacts readily with
        S03;  SOa oxidation to S03 is  not
       catalyzed by copper sulfate.
     7.  For a given concentration of copper
       oxide  in the  acceptor,  the rate of
       SO, reduction  is decreased by fly
        ash in the catalyst envelope, by an
        increase  in flue gas  throughput,
        and by a decrease in temperature.
        Both mass transfer and  the effective
        reaction rate represent significant
        resistances.
     8.  NOx  reduction is insensitive to
        fly ash, throughput, and tempera-
        ture. Mass transfer controls.
     9.  The promoted vanadium catalyst is
        an active de-NOx catalyst usable
        over a wide range  of temperatures,
        but may be unsuitable in applica-
        tions  with high concentrations of
        SO2 due to SO3 production.
         10. The  parallel passage design is an
            effective contacting device  for
            catalytic systems processing flue
            gas  from coal-fired boilers in  the
            presence of full gas loadings of fly
            ash.
       References

          1. AVCO Corporation, A Survey of
            Metal Oxides as Sorbents for
            Oxides of Sulfur. EPA Report APTD
            1178 (NTIS PB 185190), February
            1969.
          2. McCrea, S.H.,A.J.Formey,andJ.G.
            Meyers, Paper presented at 63rd
            Annual Meeting  of Air Pollution
            Control Association, St. Louis, MO,
            June 1970.
          3. Dauzenberg, F.M., J.E.  Naber, and
            A.J.J. van Ginneken, The Shell Flue
   Gas Desulfurization Process, Chem.
   Eng. Prog., 67, 1971.
4. Pohlenz J.B., NOX and SO, Emissions
   Control with  the  Shell Flue Gas
   Treating Process. 72nd Air Pollution
   Control Association Meeting, Cincin-
   nati, OH, June 1979.
5. Arneson, A.D., F.M. Nooy,  and J.B.
   Pohlenz, The Shell FGD Process —
   Pilot  Plant  Experience at Tampa
   Electric. In: Proceedings: Symposium
   on Flue  Gas  Desulfurization  —
   Hollywood, FL, November 1977
   (Vol. II). EPA-600/7-78-058b (NTIS
   PB 282091),  pp 752-793, March
   1978.
6. Imanari, M.,Y. Watanabe, S. Matsuda,
   and F. Nakajima, 7th Int. Congress
   on Catalysis, Tokyo, Japan, 1980.
7. Levenspiel, Octave, Chemical React-
   tion Engineering, John Wiley and
   Sons, New York, NY, 1972.
          J. B. Pohlenz and A. 0. Braun are with UOP, Inc., Des Plaines, IL 60016.
          J. David Mobley is the EPA Project Officer (see below).
          The complete report, entitled "Shell /V0,/S02 Flue Gas Treatment Process: Pilot
            Plant Evaluation," (Order No. PB 84-102 367; Cost: $49.00, subject to change)
            will be available only from:
                  National Technical Information Service
                  5285 Port Royal Road
                  Springfield, VA 22161
                  Telephone: 703-487-4650
          The EPA Project Officer can be contacted at:
                  Industrial Environmental Research Laboratory
                  U.S. Environmental Protection Agency
                  Research Triangle Park, NC 27711
                                                     US. OFFICIAL MAIL.
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
                                                                                 it U.S. GOVERNMENT PRINTING OFFICE 1984-759-102/816

-------