Jun-2017

The impact of side reactions in sulphur recovery unit design

A full understanding of side reactions in the modified Claus process can help to improve sulphur recovery efficiency and optimise equipment design

SUNG-HO KIM, WON-SEOK JUNG, HEE-MUN LEE and GEUN-SOO CHANG
GS Engineering & Construction

Article Summary

The sulphur recovery unit (SRU) using the modified Claus process is a critical part of most hydrocarbon processing plants because an inability to recover sulphur will normally result in significant SO2 emissions and large fines or severely reduced operations as mandated by local governmental regulations.

A SRU has two sections: the thermal reactor section (same as the modified Claus reaction furnace section); and the catalytic converter section (same as the catalytic reactor section). A simple schematic is shown in Figure 1.

The main reaction in the reaction furnace has two steps. One is a fast, irreversible and highly energetic exothermic reaction:

H₂S + 3/2 O₂ ---> SO₂ + H₂O + 520 KJ (1)

The other is a slow endothermic reaction above 600°C:

2 H₂S + SO₂ ---> 3/x Sx + 2H₂O – 42 KJ (2)

In the thermal reactor section, around 60% of total feed sulphur compounds are converted to elemental sulphur resulting from the above reactions.

The main reaction in the catalytic converters (CR1, CR2, and CR3) is carried out as follows.

There is the same reaction mechanism (Reaction 2), but it is exothermic at lower temperatures (170-370°C) with catalyst:

2 H₂S + SO₂ ---> 3/x Sx + 2H₂O + 93 KJ (3)

In the catalytic converter section, 37% of total feed sulphur compounds are converted to elemental sulphur. The remaining unconverted 3% of sulphur compounds are fed to the incinerator or tail gas treating unit. Generally, three catalytic stages of Claus process can achieve 97% of total sulphur recovery.

In view of issues such as acid rain and air pollution, higher sulphur recovery, even 99.99+% (SO₂ 150 mg/Nm₃ at incinerator stack –World Bank guidelines), might be required.

To maximise efficiency in sulphur recovery, lots of process configurations, control schemes or operational philosophies, even optimised equipment designs, have been developed. However, lots of questions remain.

For further improved sulphur recovery efficiency, a full understanding of side reactions, especially kinetic reactions in the reaction furnace, can help to improve sulphur recovery efficiency and optimise equipment design. 

The effects of side reactions (kinetic) The main kinetic components participating in side reactions in the reaction furnace (thermal reactor) are hydrogen (H₂), carbon monoxide (CO), carbonyl sulphide (COS) and carbon disulphide (CS₂). These components affect both the combustion air requirement and adiabatic flame temperature in the reaction furnace. H2 is produced from thermal dissociation of H₂S (Reaction 4). CO is produced from the reaction between CO₂ and H₂S (Reaction 5) or directly from thermal dissociation of CO₂. It also provides a reduced O₂requirement for combustion. In the case of low flame temperature in the reaction furnace, production of H₂ and CO decreases because the thermal dissociation reactions of H₂S and CO₂ are both endothermic. Incorrect estimation of COS and CS₂ production in the reaction furnace can substantially affect sulphur recovery losses because only correctly estimated COS and CS₂ are hydrolysed in the first catalytic converter. COS and CS₂ reactions in the furnace also affect the O₂ requirement.

This article includes a literature study⁵ for the comparison of different furnace products resulting from equilibrium calculation by Gibbs free energy minimisation, from kinetic calculations by empirical correlation, and from real plant data. Then a case study is carried out based on five different projects in which GS Engineering & Construction has been involved.

In the case study, calculation of kinetic components COS and CS₂ is performed by a commercial simulator, ProMax ver.3.2, and compared with the data from several well known modified Claus process technology providers. Kinetic components H₂ and CO are also calculated and the results are analysed to evaluate the impact of temperature in the reaction furnace and waste heat boiler considering thermal dissociation and re-association reactions. Total required O₂ for combustion in the reaction furnace is additionally calculated with and without side reactions.

Empirical correlation

There are many well known empirical correlations to estimate furnace kinetic reactions in the sulphur recovery industry. Fischer2 published empirical correlations for H₂, CO, COS and CS₂. They are generated as monographs based on equilibrium constants (Kp) and ideal enthalpy of formation data. The data coincided well with experimental data. The only exception is CS₂ where the graphs are 200 times the calculated values to adapt them to measured results.

Luinstra D’Hane4 published an empirical correlation for CS2 as a mathematical model:

ln(CS₂)=A * t + B * ln(H₂S) + C

CS₂: CS₂ mol%(dry) at the inlet of the first catalytic converter

t: furnace residence time, s

A: -0.426

B: -0.25

C: 1.4

Sames and Paskall⁹ published empirical correlations for H₂, CO, COS and CS₂. These are called Western Research Corelations. The correlations are generated based on data from over 300 plants on 100 different sulphur trains including various front end configurations. Plant feed compositions also range from 8 mol% to 98 mol% H2S. The correlations are:

R(CO) = fraction of furnace inlet carbon that forms CO

= 0.002 * A0.0345 exp (4.53 * A)

R(H₂) = fraction of furnace inlet H₂S that cracks to H₂ and S

= 0.056 (±0.024)

R(COS) = fraction of furnace inlet carbon that forms COS

= 0.01 * tangent (100A) for 0<=A<=0.86

= 0.143 for A > 0.86

R(CS₂) = fraction of furnace inlet hydrocarbon that forms CS₂

= 2.6 * A0.971 exp (-0.965 * A)

Where A = mole fraction of H2S in the acid gas feed on a dry basis.

Equilibrium results vs empirical correlation vs plant data

Monnery₅ compared generated equilibrium data with the Western Research empirical correlations, plotted on graphs reproduced from Paskall6 which show the actual plant data based on samples taken from the waste heat boiler effluent. For the equilibrium calculation in the reaction furnace, he used the Gibbs free energy minimisation method and the result was verified against the commercial simulator Sulsim.

Figures 2 to 5 show the results of his study. Figure 2 shows that the equilibrium calculation predicts much more H₂ content than Western Research empirical correlations do for an acid gas feed above 360% H₂S concentration. This is because limited residence time in a real plant can quench the thermal dissociation reaction even though it is very fast at high temperatures. Mainly, the reverse reaction of thermal dissociation (re-association) in the waste heat boiler is considered because there is a huge difference between the equilibrium calculation and real plant data due to the thermal dissociation reaction, which is endothermic and favours the reverse reaction at low temperature (waste heat boiler section). 

Western Research empirical correlations seem to be quite similar to plant data.

The major H₂ forming/consuming reaction is:

H₂S <---> H₂ + 1/2 S₂ (4)

Figure 3 shows that the equilibrium calculation predicts much more CO content than the Western Research empirical correlations for all H₂S concentrations in the acid gas feed. The Western Research empirical correlations seem to be much closer to the plant data. Basically, CO₂ is the major source of CO formation (Reaction 5). The produced CO in the reaction furnace reacts with S to form COS rapidly (Reaction 6), both in the reaction furnace and waste heat boiler, so the real plant data has a lower content of CO compared to the equilibrium calculation.

The major CO forming/consuming reactions are:

H₂S + 2CO₂ <---> 2CO + H₂ + SO₂ (5)

CO + S <---> COS (6)

Figures 4 and 5 show that the equilibrium calculation predicts small amounts of COS and CS₂ in the reaction furnace nace but the Western Research empirical correlations estimate relatively high amounts. The Western Research correlations seem to be more similar to the plant data. This is because in the equilibrium condition with unlimited residence time, the produced COS and CS₂ (Reactions 7, 9) are fully hydrolysed back to H₂S in the reaction furnace (Reactions 8, 10). Due to the relatively slower reaction rate of the hydrolysis reaction, much more COS and CS₂could exist in the real plant data.

The major COS forming/consuming reactions are:

2CO₂ + 3/2S₂ ---> 2COS + SO₂ (7)

COS + H₂O ---> H₂S + CO₂ (8)

The major CS₂ forming/consuming reactions are:

CH₄ + 2H₂S ---> CS₂ + 4H₂ (9)

CS₂ + 2H₂O ---> CO₂ + 2H₂S (10

Besides the reactions of H₂, CO, COS and CS₂ shown above, many other reactions occur in the reaction furnace and waste heat boiler.

Because the empirical models are based on fitting data from the waste heat boiler effluent, H₂ and CO calculated from empirical correlations reflect both thermal dissociation and re-association reactions in the reaction furnace. This does not match the trend of real plant data exactly. After measuring the furnace effluent, the results of equilibrium calculations for H₂ and CO show a trend similar to real plant data. Fortunately, the results of the empirical correlations for CS₂ and COS are still similar to real plant data.^{1, 5, 11}

Case study

In this case study, the results of the equilibrium calculations, empirical correlations and sulphur recovery technology providers’ (licensors) data for kinetic components (COS and CS₂) in the reaction furnace were analysed to evaluate the amount of kinetic components by applying them to five different projects in which GS Engineering & Construction participated as the EPC contractor. Kinetic components (H₂ and CO) are also calculated and the results are analysed to evaluate the impact of temperature in the reaction furnace and waste heat boiler. Total required O₂deviations for combustion in the reaction furnace are additionally shown, with and without side reactions.

A commercial simulator, ProMax ver.3.2 was used to calculate the equilibrium calculation and empirical correlations. ProMax adopts the Gibbs free energy minimisation method for the equilibrium calculation and several empirical correlations can be selected as the reaction constraints.

The feed gas compositions of five projects, A-E, are shown in Table 1.

COS and CS₂ comparison

COS and CS₂ are kinetically limited components in the reaction furnace. Therefore, to estimate the products of COS and CS₂ correctly, various empirical correlations have been developed and applied in the reaction furnace reaction mechanism. To analyse the tendency of formation and consumption of COS and CS₂ components, five reference projects’ feed data are applied (see Table 1).

Two empirical correlations such as Fischer 1974 and NSERC 1993 are adopted. The correlations’ results were compared with the equilibrium calculation and licensors’ data. NSERC 1993 in ProMax can be considered similar to the Western Research correlations for COS and CS₂.

Table 2 shows the equilibrium results for COS and CS₂ in the reaction furnace of five reference projects and Table 3 shows the empirical correlations’ results and licensors’ data. Fisher 1974 and NSERC 1993 are used for empirical correlations.

For Project A, the contents of COS from the equilibrium calculations (see Table 2 and Figure 6) are higher than the results of NSERC 1993 and lower than Fischer 1974 respectively (see Table 3 and Figure 6). The content of CS₂ from the equilibrium calculation (see Table 2 and Figure 7) is negligible but the empirical correlations show high levels of CS₂ formation (see Table 3 and Figure 7).

Licensors’ data show similar results to Case 2 (see Table 3) regarding the amount of COS and CS₂ which are relatively higher contents than in other cases. From this result, it can be assumed that if the feed gas has a high content of CO₂, higher amounts of COS and CS₂ could be produced in the reaction furnace. The adiabatic temperature from the equilibrium calculation (see Table 2) is similar to Case 3 (Table 3). From this result, it can be assumed that lower production of COS and CS₂ could cause lower adiabatic temperature in the reaction furnace.

For Projects B, C and D, the contents of COS from the equilibrium calculations (see Table 2 and Figure 6) are higher than the results of Fischer 1974 and lower than NSERC 1993 respectively (see Table 3 and Figure 6). The contents of CS₂ from the equilibrium calculation (see Table 2 and Figure 7) are neglible but empirical correlations show a relatively high level of CS₂ formation (see Table 3 and Figure 7).

For Project E, the trend of COS formation is similar to Project A. Both projects have very lean acid gas compositions which are below 50% H₂S content. The content of COS from the equilibrium calculations (see Table 2 and Figure 6) are higher than the results of NSERC 1993 and lower than Fischer 1974 respectively (see Table 3 and Figure 6). The content of CS₂ from the equilibrium calculation (see Table 2 and Figure 7) is negligible but empirical correlations (only Fischer 1974) have a relatively high amount of CS₂ formation (see Table 3 and Figure 7).

The interesting thing is that the CS₂ contents from NSERC 1993 are also negligible. This is because the acid feed gas of Project E does not contain hydrocarbon. From this result, it can be assumed that if there is no hydrocarbon in the acid gas feed, NSERC 1993 estimates no CS₂formation in the reaction furnace due to the Western Research (same as NSERC 1993) correlation which calculates the CS₂ rate based on inlet hydrocarbon.

For Projects B and D, licensors’ data shows similar results to Case 3 (see Table 3) regarding the amount of COS and CS₂. The results show relatively higher content of COS and lower content of CS₂ than in other cases.

For Projects C and E, licensors’ data show results similar to Case 4 (see Table 3) regarding the amount of COS and CS₂ which both have relatively higher contents than in other cases.

The adiabatic temperatures from each case show different levels according to the amount of COS and CS₂. From this result, it can be assumed that less production of COS and CS₂ could cause lower adiabatic temperatures and more production of COS and CS₂ could cause higher adiabatic temperatures in the reaction furnace.

The result of this study seems to support a previous study.⁵ In the equilibrium circumstance, almost all generated CS₂ is hydrolysed back to H₂S so more CS₂ is found in the real plant data. However, the amounts of COS do not follow the previous study in that some of the empirical data show less COS than in the equilibrium calculation. Results from two projects (A and E) therefore show that more COS is found in the equilibrium calculation than in NSERC 1993. Results from three projects (B, C and D) show that more COS is found in the equilibrium calculation than in Fischer 1974.

As a result, according to this case study for COS and CS₂, it can be concluded that NSERC 1993 estimates more CS₂ content than the equilibrium calculation does in all acid gas feed compositions. However, Fischer 1974 estimates more COS formation than in the equilibrium calculation below approximately 75% H₂S content in the acid gas feed and NSERC 1993 estimates more COS formation above approximately 75%. Table 4 shows the results of the case study.

H2 and CO comparison

H₂ and CO are the other kinetically limited components in the reaction furnace and waste heat boiler. The main effects of H₂ and CO formation are on the combustion air requirement and adiabatic flame temperature.

As mentioned previously, the main H₂ formation reaction is thermal dissociation of H₂S (Reaction 4). The more H₂S that is involved in this dissociation reaction, the less air is required for combusting H₂S. In addition, the thermal dissociation reaction is endothermic so the adiabatic flame temperature is decreased. The produced H₂ is re-associated with S₂ to form H₂S in the waste heat boiler.^{1, 11}

CO is also produced by the thermal dissociation reaction but the main reaction is Reaction 5. The produced CO is consumed in the waste heat boiler by reacting with S₂ to produce COS (Reaction 6). In other words, CO entering the waste heat boiler is counterbalanced by the formation of COS and, because the reaction between CO and S₂ is very fast, the consumption of CO is attributed to the formation of COS.^{1, 11}

After measuring the furnace effluent, the equilibrium calculations result for H₂ and CO shows a trend more similar to real plant data than to empirical correlation at the furnace effluent.^{1, 5, 11}

Table 5 shows H₂ and CO formation and consumption in the reaction furnace and waste heat boiler of the five projects. The results are from ProMax ver.3.2. H₂ and CO formation in the reaction furnace were calculated by the equilibrium calculation because the result from the equilibrium calculation shows trends more similar to real plant data than to empirical correlation. ⁵ 

ProMax estimates re-association in the waste heat boiler by using quench temperature constraints³ which means the re-association reaction is stopped at a certain temperature and no more reactions occur in the waste heat boiler. From Table 5, it can be concluded that a higher flame temperature in the reaction furnace (Projects B, C and D) causes much more H₂ formation. This is because the thermal dissociation reaction is endothermic so a higher temperature favours the reaction. A relatively large portion of the produced H₂ is consumed and emerges in H₂S in the waste heat boiler. In other words, at low temperature (the waste heat boiler section), the reverse reaction (reassociation) is favoured. Generally, around 5-7% of inlet H₂S is converted to H₂ by the thermal dissociation and re-association reaction (Reaction 4).

A large portion of the produced CO is also consumed and converted to COS in the waste heat boiler (Reaction 6). However, if the furnace outlet temperature is below a certain level (1093°C), CO is not consumed and converted to COS (Projects A and E). 

In addition, if there is more CO₂ in the feed gas, more CO is produced in the reaction furnace (Projects A and E) because CO₂ is the main component of CO and COS formation reactions (Reactions 5 and 6). Table 5 was generated based on the calculation that H₂ and CO thermal dissociation reactions occur in the reaction furnace and H₂ and CO re-association reactions occur in the waste heat boiler.

If both reactions (H₂ and CO thermal dissociation and re-association) are assumed to occur in the reaction furnace but not in the waste heat boiler, what advantage can be obtained for process designers? Some licensors estimate that both reactions occur in the reaction furnace. Firstly, the reaction furnace temperature will be increased because less thermal dissociation occurs. This can result in more conservative refractory design for the reaction furnace. Secondly, the duty change of the waste heat boiler will be negligible because specific heat will be changed. Therefore waste heat boiler design should not be affected by increased reaction furnace temperature. 

Table 6 shows a temperature and duty comparison of the above two scenarios (Case 1: thermal dissociation and re-association in the reaction furnace; Case 2: thermal dissociation in the reaction furnace and re-association in the waste heat boiler).  

Requirement for combustion oxygen

Side reactions in the reaction furnace, especially kinetic component reactions, cause a decrease in combustion air. Thermal dissociation reactions of H₂S and COS/CS₂ formation would take away the combustion source of H₂S, so the O₂ requirement could be reduced. Table 7 shows the difference between combustion without side reactions and combustion with side reactions. Stoichiometric combustion can be calculated by Reaction 11:

C_nH_mO_hS_jN_k + (n+m/4–h/2+j) * O₂(g) ---> nCO₂(g) + m/2H₂O + k/2N₂(g) + jSO₂(g) (11)

Due to the effects of side reactions of kinetic components, the calculated O₂ with side reactions is relatively smaller than in combustion without side reactions. The more CO₂ included in the feed gas, the more deviation from stoichiometric data is expected.

Conclusion

Equilibrium calculations, empirical correlations and sulphur recovery technology providers’ (licensors’) data for kinetic components (H₂, CO, COS and CS₂) in the reaction furnace and waste heat boiler were analysed for five projects. In terms of COS formation, NSERC 1993 (Western Research correlations) produces more COS above approximately 75% H₂S composition in the acid gas feed, and Fisher 1974 produces more COS below approximately 75% H₂S composition in the acid gas feed. For tCS₂ formation, NSERC 1993 estimates more CS₂ for all H₂S feed compositions. Licensors generally estimate relatively high contents of COS but CS₂ has varying contents case by case, but not as much as COS.

The hydrolysis reaction of COS is relatively slow so some COS can be found at the waste heat boiler outlet. Therefore, estimating more COS seems to be more reliable from a SRU process design engineer’s point of view. Estimating more CS₂ also seems to be better for SRU process design engineers in terms of giving design margins to downstream equipment.

In terms of the formation and consumption of H₂, more H₂ is produced at higher adiabatic flame temperatures in the reaction furnace by causing thermal dissociation of about 15-19% of total H₂S inlet feed. The percentage participating in thermal dissociation reactions (including ing re-association reactions) at the waste heat boiler outlet is around 5-7% of inlet H₂S. For the formation and consumption of CO, certain amounts of produced CO in the reaction furnace are consumed as COS in the waste heat boiler.

From the process designer’s point of view, estimating less H₂ and CO in the reaction furnace, for instance 5-7% instead of 15-16% of inlet H₂S, delivers a higher furnace outlet temperature. Generally, the design temperature of refractory in a SRU reaction furnace is high enough (up to 1538°C).

In addition, COS, CS₂, H₂ and CO formation and consumption will substantially affect the requirement for combustion O₂ (5-19% depending on feed gas composition). Therefore, correct estimation of side reactions in the reaction furnace and waste heat boiler are critical in SRU design.

 

^References

1. ^{Dowling N I, Hyne J B, Brown D M, Kinetics of the reaction between hydrogen and sulphur under high-temperature Claus furnace conditions, Ind. Eng. Chem. Res., 29, 2327, 1990.}
2. ^{Fischer H, Burner/fire box design improves sulphur recovery, Hydrocarbon Process., Oct, 125, 1974.}
3. ^{Hegarty W P, Chen M S K, Goar B G, Comparison of Claus reaction furnace performance with air and COPETM O2-based operation, Air Products and Chemicals, Inc., presented at the AIChE Conference, 1990.} 
4. ^{Luinstra E A, d’Hasne P K, Catalyst added to Claus furnace reduces sulphur losses, Hydrocarbon Process., July, 53, 1989.}
5. ^{Monnery W D, Svrcek W Y, Behie L A, Modeling the modified Claus process reaction furnace and the implications on plant design and recovery, Canadian Journal of Chemical Engineering, 71: 711-724, 1993.}
6. ^{Paskall H G, Capability of the Modified Claus Process. A Final Report to the Department of Energy and Natural Resources of the Province of Alberta, Ch. IV, 1979.}
7. ^{Paskall H G, Reaction furnace chemistry and operational modes, Sulphur Recovery, published by Western Research, Calgary, Alberta, Canada, 1990 (originally published in Sulphur, Jun 1981).}
8. ^{Sames J A, Paskall H G, So you don’t have a COS/CS2 problem, eh?, Sulphur Recovery, published by Western Research, Calgary, Alberta, Canada, 1990 (originally published in Sulphur, Jun 1984).}
9. ^{Sames J A, Paskall H G, Simulation of reaction furnace kinetics for split-flow sulphur plants, Sulphur Recovery, published by Western Research, Calgary, Alberta, Canada, 1990 (originally published as a paper for the Sulphur ’85 conference, Nov 1985).}
10. ^{Sames J A, Dale P R, Wong B, Evaluation of reaction furnace variables in modified Claus Plants, Sulphur Recovery, published by Western Research, Calgary, Alberta, Canada, 1990 (originally published for the Laurence Reid Gas Conditioning Conference, Mar 1987).}
11. ^{Sames J A, Paskall H G, Brown D M, Chen M S K, Sulkowski D, Field measurements of hydrogen production in an oxygen enriched Claus furnace, Sulphur Recovery, published by Western Research, Calgary, Alberta, Canada, 1990 (also in preprints of Sulphur 1990 International Conference, Apr 1990).}