CFD STUDY OF FUEL GAS ASSISTED BTX DESTRUCTION IN A SPLIT

FLOW REACTION FURNACE

Dave Sikorski Anthony Corriveau, Ph.D. Nick Roussakis, P.Eng

HEC Technologies Stream Function Inc. HEC Canada Inc.

Abstract

At 6.5 mole % H2S in the acid gas feed, it is impossible to sustain combustion or achieve
reasonable BTX destruction without resorting to additional measures. Using a
combination of fuel gas assist, acid gas bypass and good furnace design, it is possible to
achieve excellent flame stability, complete BTX destruction, and good sulphur recovery
in a traditional modified Claus SRU with only 2 catalytic stages. This paper presents a
CFD study of a lean acid gas reaction furnace that has been operating trouble free since
start up six years ago.

Introduction

The Petro-Canada Wilson Creek gas plant is somewhat unique in that it uses the
traditional modified Claus process to recover sulphur from an acid gas feed stream of
approximately 6.5% H2S. Consisting of a reaction furnace and 2 catalytic stages, the
plant is able to achieve over 93% sulphur recovery. In addition, the acid gas feed often
has more than 600ppmv BTX (benzene, toluene and xylene). Despite this, the high BTX
destruction efficiency of the reaction furnace has resulted in catalyst bed operation free of
coke deposition (1). After some initial commissioning hurdles, the Wilson Creek SRU
has been running reliably for the past six years.

HEC Technologies has presented papers (2, 3, 4) showing CFD (computational fluid
dynamics) to be a useful tool for reaction furnace design and troubleshooting. However,
CFD can also be very useful as a means for analysing the design and behaviour of
unusual systems that perform well. The Wilson Creek reaction furnace and burner were
designed without the aid of CFD. The purpose of this study was to analyse this system
using CFD and determine what lessons could be learned regarding the design and
operation thereof.

Acid Gas Combustion- Flame Stability

The current acid gas feed to the Wilson Creek SRU consists of 6.5% H2S, 1%
hydrocarbons (including 600ppmv BTX), 2.5% H2O, up to 10,000ppm of methanol and
about 90% CO2. Based on typical modified Claus reaction furnace air requirements-
approximately one third of stoichiometric air for H2S and almost full stoichiometric air
for hydrocarbons- the Wilson Creek acid gas would yield an adiabatic reaction furnace
temperature of only 545C (based on acid gas T=50C and air T=90C). One of the means
utilized for increasing the reaction furnace temperature involves preheating the air and
acid gas. Based on current operation, the air and acid gas are preheated to 198 and 216C
respectively. Based on preheat alone, the calculated adiabatic reaction furnace
temperature goes up to 665C.
Under typical reaction furnace air/acid gas equivalence ratios, severe flame instability
will occur as the acid gas flame temperature drops to about 850C. This value is
supported by examining the flammability limit characteristics of pure H2S in air. At the
lower limit of flammability (LFL), 4.3 volume % H2S, the theoretical adiabatic flame
temperature is 773C. The LFL is characterized by excess air combustion with the
unburned mixture equivalence ratio corresponding to 310% of stoichiometric air.

For the upper flammability limit (UFL), 45 volume % H2S, the theoretical adiabatic
flame temperature is 960C. The UFL is characterized by substoichiometric combustion
(oxygen deficient) with the unburned mixture equivalence ratio corresponding to only
17% of stoichiometric air. In a typical reaction furnace, optimum sulphur conversion
occurs for air/acid gas equivalence ratios corresponding to approximately 33% of
stoichiometric air. Since this lies in between the LFL (310%) and UFL (17%)
equivalence ratios, it is not surprising that the typical lower temperature stability limit of
850C for an acid gas flame lies in between these values. There are additional factors that
affect this limit; however, systems are generally operated well above this point because
sulphur recovery and contaminant (BTX) destruction can be significantly compromised
when a furnace is operated at very low temperatures.

As stated above, based on the current H2S content of 6.5 volume %, the use of preheat
alone will only achieve a calculated temperature of 665C. This is far below the limit for
flame stability or good BTX destruction. These aspects of system performance were
assured by incorporating fuel gas assist and bypass operation. In order to ensure good
BTX destruction, the system was designed with fuel gas assist to yield a fully mixed
adiabatic temperature of 1100C and a total system residence time of 3 seconds. Good
BTX destruction typically requires a temperature of 1050C and a residence time of 1
second.

Fuel Gas Assisted Combustion

The effectiveness of fuel as addition to an acid gas flame depends primarily on two
factors: the flame temperature and the fraction of the heat release attributable to the
hydrocarbons. As the portion of the heat release from the fuel gas increases, the greater
the system temperature required to ensure proper combustion. Since fuel gas usually
consists overwhelmingly of methane, it is to be expected that the pure component
combustion characteristics of this component will give the best indication of the
hydrocarbon combustion behaviour compared to H2S. At 1268 and 1491C respectively,
the LFL and UFL temperatures of methane in air are about 500 degrees C higher than the
corresponding values of H2S. This indicates that the oxidation kinetics of methane are
slower than those of H2S and thus the need for higher temperatures to sustain combustion
at limiting conditions.

Based on the original acid gas composition (8 to 11 volume%) and the design process
conditions, the resulting H2S contribution to the overall heat of combustion was to be less
than 25% of the overall heat release. In order to ensure adequate flame temperatures for
good fuel gas combustion, the original specification stipulated that the reaction furnace
burner must be able to operate with up to at least 50% acid gas bypass away from the
burner and directly to zone 2. This would result in a flame temperature of about 1500C
as compared to the fully mixed system temperature of 1100C.

CFD ANALYSIS

Figures 1a to c present the CFD hydraulic domain for the burner (HEC Technologies) and
the reaction furnace system. The burner air, acid gas and fuel gas injection point is
located at the far left end of the system shown in figure 1a. Integral to the burner is the
cylindrical combustor with an inside diameter that is slightly smaller than half the
chamber diameter. The single bypass gas injector is located about one chamber diameter
downstream of the combustor discharge. The bypass flow and the hot combustor effluent
are channelled through the choke ring followed by the checker wall before entering the
waste heat boiler. The adiabatic (zero heat loss) assumption was made for all cases. The
computational domain contained 5x106 volume elements. The simulations were carried
out using ANSYS CFX version 11.0 CFD software.

The Wilson Creek SRU was designed for 4 LTPD of sulphur production with 8 to 11%
H2S acid gas. At the time of this writing the plant was operating at about 1 LTPD of
production. Since turndown operation generally represents a greater challenge regarding
flame stability and mixing, it was decided to base the CFD simulations on the current
25% production case with 6.5% H2S acid gas. Ismailian et al (1) reported good sulphur
recovery and BTX destruction based on SRU testing conducted with 70% of the acid gas
feed going to bypass and only 30% to the burner. Petro-Canada personnel have also
reported that the system has been operated at 100% bypass without compromising
sulphur recovery. For these reasons, it was decided to simulate a 70% bypass case and,
as an extreme, to simulate the system behaviour with 100% of the acid gas going directly
to zone 2. Both of these cases represent oxidizing (excess air) conditions in the main
flame. As a contrast to those cases, the third case was based on the extreme of 100% acid
gas to the flame, hence zero flow to the bypass.

Results and Discussion

Figures 2a to c show the simulated temperature distributions for 70%, 100% and 0%
bypass flow respectively. Temperature is a good indicator of flame mixing/reaction
within the combustion zone and for the mixing/reaction of the hot combustor effluent
with the bypass stream (zone 2).

Figures 2a and 2b predict very high temperatures within the combustor. This was
expected because in one case most and in the other case all of the acid gas is bypassed
directly to zone 2. Because of the relatively high velocity of the hot combustor effluent,
the fuel gas combustion remains reasonably well segregated from the cross-flowing
bypass gas. In Figure 2b, some of the bypass backwash can be seen penetrating a short
way into the combustor. The high combustor gas temperatures for the 70 and 100%
bypass cases emphasize the need to ensure that refractory thermal design reflects the true
variability of operating conditions in a reaction furnace system.

For the 0% bypass mode, Figure 2c shows a relatively low temperature within the
combustor. This mode of operation has not been tested at the plant and represents a
hypothetical scenario. As discussed above, the overwhelming majority of the combustion
heat release is contributed by methane which can have questionable combustion stability
and compromise furnace conversion at a fully mixed theoretical temperature of 1100C.

The adiabatic exit temperature variation is a good indicator of overall system mixing
performance. The 0% bypass case shows the best overall mixing with an exit delta T of
1C compared to 186 and 104C for the 70% and 100% bypass cases respectively.

At Wilson Creek, the relative fuel gas and air addition to the process is based on the
target reaction furnace temperature required for good BTX destruction. T1 pyrometer,
between the combustor discharge and the bypass inlet, is used as the basis for increasing
or decreasing fuel gas and air flow. It should be noted that figures 2a and b indicate that
the furnace zone monitored by T1 contains a highly non-homogeneous gas composition.
Poorly mixed gas in the T1 zone may be generating inconsistent temperature readings. In
general, it would be expected that a more homogeneously mixed location, such as T3,
may provide a better control point for determining fuel gas addition. Ismalian et al (1)
reported that there was a distinct drop in temperature between T1 and T3. This was
attributed to non-homogeneous composition at T1 and significant system heat loss.

A very unusual aspect of this application is the fact that it performs well even when the
reaction furnace is operated with oxidizing (excess air) conditions in zone 1 (flame).
This was the case during the 70% bypass operating condition during the plant testing
reported by Ismailian et al (1). As already mentioned, the system has been operated with
100% of acid gas to bypass with good performance results. Typically, bypass systems are
limited to bypass fractions that yield reducing conditions in zone 1. This strategy is
intended to ensure complete oxygen consumption and thus prevent breakthrough to the
first catalyst, as would occur if the zone 1 effluent does not mix well with the bypass gas
in zone 2. In addition, there is significant concern that zone 2 mixing and residence time
may not suffice to ensure the complete reduction of oxidizing species such as SO3 and
NO that could form in zone 1.

Figures 3a to c are oxygen concentration plots for 70%, 100% and 0% bypass cases
respectively. Due to zone 1 oxidizing conditions in figures 3a and b, unreacted oxygen is
present in the combustor effluent. However, the oxygen profiles predict that the oxygen
is completely consumed at the choke ring for the 70% bypass case and just downstream
of the choke ring for the 100% bypass case. This would leave about 60% of the total
system residence time for the reduction of unwanted oxidizing species such as SO3 and
NO.

The oxygen profile of Figure 3c forecasts that all oxygen would be consumed well within
the combustor. As stated previously, this case is somewhat hypothetical because in reality
it could yield questionable fuel gas combustion stability and extreme levels of CS2 in the
reaction furnace.

In the plant testing reported by Ismailian et al (1), the temperature drop from T1 to T3
was about 180C at a sulphur production level of 2.7 LTPD. At the current production rate
of 1 LTPD the pyrometer temperature readings are 1065, 872 and 850C respectively.
This represents a T1 to T3 temperature decline of 215C. The increased temperature loss
is expected from the greater non-adiabatic behaviour characteristic of decreasing system
mass flow. Despite the fact that the average zone 2 temperature is well below the typical
1050C target, the furnace yields excellent BTX destruction and no indication of catalyst
bed coking. This may be attributable to the fact that the system was designed with a much
larger than normal residence time of 3 seconds at full load operation. Based on the
current 25% load case the residence time will be 12 seconds – 4 times the original.

At the present time there is a lack of chemical kinetic expressions that can be used in
conjunction with CFD models to predict the destruction behaviour of BTX compounds in
reaction furnaces. The development of such expressions will improve equipment design
capability. In the case of the Petro-Canada Wilson Creek reaction furnace such data can
be used to optimized reaction furnace size and operating temperature.

Conclusions and Recommendations

Using bypass operation and fuel gas assist, good BTX destruction and sulphur conversion
are possible in extremely lean acid gas applications. Such systems need to be carefully
designed to ensure flame stability, refractory integrity, good post flame mixing and
sufficient residence time for contaminant destruction.

Extreme acid gas bypass ratios - oxidizing conditions in zone 1 – can be tolerated if zone
2 mixing and residence time ensure complete consumption of residual oxygen and any
other oxidizing species that can compromise the catalyst beds.

For similar low H2S acid gas applications, consideration should be given to a system
design based on 100% acid gas bypass; only fuel gas is burned in the flame.

For the Wilson Creek reaction furnace the T3 pyrometer may provide more consistent
temperature readings than the T1 location.

The development of kinetic expressions for the destruction of BTX compounds will
improve the ability to optimize reaction furnace size and temperature.

Acknowledgements
The authors would like to express appreciation to Sam Tse, Fred Ismailian, Keith Ng and
Gavin Proudfoot, all of Petro-Canada, for graciously providing Wilson Creek plant
information and technical data to the authors of this paper.
References

1) Ismailian, F., Proudfoot, G., Ng, K., (2003) “Pushing the Claus Envelope 10% H2S
Acid Gas with Significant Aromatics” Laurance Reid Gas Conditioning
Conference, Norman Oklahoma, Feb. 23-26, 2003

2) Sikorski, D., Roussakis, N., and Corriveau, A., (2002) "CFD Un‘Vail’ed: The
Application of CFD for the Analysis of Reaction Furnaces" Brimstone Sulfur
Recovery Symposium, Vail Colorado, Sept. 9-13, 2002

3) Sikorski, D., Roussakis, N., and Corriveau, A., (2003) “CFD Analysis of Bypass
Mixing in Reaction Furnaces” Laurance Reid Gas Conditioning Conference,
Norman Oklahoma, Feb. 23-26, 2003

4) Sikorski, D., Roussakis, N., and Corriveau, A., (2005) “Flame Temperature Issues in
SRU Equipment” Brimstone Sulfur Recovery Symposium, Vail Colorado, Sept.
2005

Contact information:
HEC Technologies
Calgary, Alberta, Canada
1-403-247-0578
www.hecombustion.com
1.6m
`