Simulation And Sustainability Assessment Of H 2 S Utilization From Acid Gas On Haldor Topse Wet Gas Sulfuric Acid And Claus Processes

ABST�CT: Claus process is a widely adopted process to reduce emissions from re�neries by converting H 2 S into elemental sulfur. On the other hand, Haldor Topsoe’s Wet Gas Sulfuric Acid (WSA) is an alternative to convert H 2 S directly into sulfuric acid. �e purpose of this project was to simulate both of these state-of-the-art technologies and evaluate their suitability for various acid gas capacity and H 2 S concentrations. �ree sustainability pillars of people (safety), planet (environment), and pro�t were used as the comparison metrics. �e developed simulation (1 st principle) models were used to generate lots of data as the basis for subsequent development of regression models. �e la�er models were used in the comparisons for they are much faster in calculations than the 1 st principle models. �e results showed that the WSA process was safer (lower Fire and Explosion Damage Index), more environmentally friendly (lower Global Warming Potential), and more pro�table (higher annual pro�t) in most of the evaluated operating conditions.


INTRODUCTION
H 2 S is a typical contaminant found in crude natural gas, extracted from the oil and gas wells and re neries. Its amount in natural gas varies signi cantly from wells to wells. It is a colourless, ammable and poisonous gas. It has a molecular mass of 38 g/mol with boiling point at -60. 3 ºC. e mixture of H 2 S and air is explosive. It is a very corrosive gas which can cause corrosion in pipeline and other equipment. In fact, it also poses hazards to the environment and potential risks to human health [1]. At the present time, more natural gases contain H 2 S is being produced in Malaysia as the consumption of natural gas is gradually increased. Total natural gas consumption is expected to grow at 18% annually between 2007 and 2035 [2]. Table 1 Range of acid gas composition (mole %) in typical re neries [3] Recently, concerns over H 2 S emissions from acid gas continue to draw ever increasing consideration towards the development of alternative sulfur recovery technology. e conversion of H 2 S to a commercial product is an effort to reduce its emissions. Typical acid gas composition coming from re neries is shown in Table 1.
ere are several technologies for sulfur recovery such as Claus and Wet Gas Sulfuric Acid (WSA) processes. e Claus process has been known for decades and has been used in industries to convert H 2 S into elemental sulfur. e majority usage of this elemental sulfur is sulfuric acid plants.
us, an alternative to directly convert H 2 S into sulfuric acid is found in a technology called Wet gas Sulfuric Acid (WSA), developed by Haldor Topsoe. e WSA process has been found to be an efficient process for the production of concentrated sulfuric acid from acid gas [4].
From a sustainability point of view, it is very important to reduce this H 2 S emission while at the same time producing a more useful product in a safer and more eco-friendly way. In this regard, a concept of 3P's (People, Planet, Pro t) has been used as three important pillars of sustainability [5,6]. e rst pillar ("People") means that it is a safer process, while "Planet" symbolizes a more environmentally friendly process, and lastly, "Pro t" means that the process itself has to make a pro t in order to sustain.
Despite the presence of these two state-of-the-art technologies, acid gas comes with various capacities and H 2 S concentrations.
is research work aims to technically  [7,8]. is safety assessment considers the chemicals used and the operating conditions of the process. For the "Planet" pillar, Global Warming Potential (GWP) is selected as the index for the environmental impact. Lastly and obviously, the annual pro t of the process is taken as the "Pro t" pillar.
In this work, both Claus and WSA processes modelled and simulated using a process simulation so ware called Symmetry-iCON. Based on the mass balance obtained from the simulations, the indices of FEDI, TDI, GWP, and annual pro t are calculated accordingly. Due to the required granularities within the applicable range of capacities and concentrations, running smaller steps of variations is required. In this case, a surrogate approach of merging machine learning and rst principles [9] is adopted. us, required simulation runs are obtained via a Design Of Experiment (DOE) using Central Composite Design (CCD) [10]. e data is then used to develop the surrogate models, which are then used to create a surface map for each process in each index. Finally, surface maps of both processes are plo ed together for each index for comparison. Figure 1 shows the overall methodology for this work where both Claus and WSA processes were modelled to calculate FEDI, TDI, GWP, and annual pro t. CCD based simulation runs were then performed for the development of surrogate models. en surface maps of the models were plo ed for each index for comparisons. e components taken in the simulation were hydrogen sul de (H 2 S), carbon dioxide (CO 2 ), carbon monoxide (CO), water (H 2 O), methane (CH 4 ), sulfur dioxide (SO 2 ), sulfur trioxide (SO 3 ), nitrogen (N 2 ), oxygen (O 2 ), hydrogen (H 2 ), sulfuric acid (H 2 SO 4 ), and elemental sulfur (S 2 and S 8 ). Typical unit operations used in this simulation were furnaces, conversion reactors, heat exchangers, and separator vessels. In the WSA process, its condenser had three basic functions, namely a place to react (converting SO 3 to H 2 SO 4 ), to reduce temperature and to condense (condensing SO 3 gas and H 2 SO 4 gas to H 2 SO 4 liquid), and to separate the gas from the condensed phase (separating liquid H 2 SO 4 and clean gas) (11). In Symmetry-iCON, there is no WSA condenser unit per se available to do these three functions. Hence, this three equipment (conversion reactor, heat exchanger, and vessel) were used to simulate one WSA condenser.

Fig. 1. Research Methodology
In this project, H 2 S concentration and feed gas capacity were varied as shown in Table 2 by using the CCD  method. is method was used to minimize necessary simulation runs while still meeting enough variations to develop accurate regression or surrogate models for both WSA and Claus processes. ese regression models were used further instead of 1 st principles modelling from Symmetry-iCON due to their much faster calculations. Table 2 Variation of feed capacity and H 2 S concentration In each simulation, the operating conditions results obtained were then used to calculate the above-mentioned indices representing the three pillars of sustainability (3Ps). More details on how to calculate FEDI, TDI, GWP, and annual pro t are taken in one of our previous works [12].
e regression models were developed as a function of two variables, namely feed capacity (x 1 ) and H 2 S concentration (x 2 ). A general form of a second order regression model was used in this work. It is shown as follows. Y = ax 1 +bx 2 + cx 1 2 +dx 2 2 + ex 1 x 2 + f Hence, there were four models developed for each Claus and WSA processes. e dependent variables (y) involve TDI, FEDI, GWP and annual pro t as described above. Further details on the methodology and the subsequent results are covered in the literature [13].

WSA Process Simulations Description
In the WSA process, the feed air and acid gas are fed into a combustor, where there are various reactions occurred during the combustion process. e main reaction is the H 2 S is converted to SO 2 and CH 4 is bunt into CO 2 since it was one of the main feed gas compositions. e gas stream leaves the combustor at 842ºC and is cooled at 430ºC before entering the reactor. e stream enters three consecutive conversion reactors with different required temperatures of inlet streams. e reaction that occurred in the reactor is to convert SO 2 to SO 3 , which is a highly exothermic reaction, in a step-wise manner. Reactions occurring in the combustor are as follows: e reactions are conducted under an adiabatic condition. To achieve a high conversion the reaction is then cooled in heat exchangers before entering the next conversion reactor.
is cooling process produces steam which can be sold and taken as part of the annual pro t calculation. A er the third conversion reactor, the process gas is cooled from 443 ºC to the acid dew point temperature of 260 ºC in the condenser, producing the sulfuric acid. Reactions happening in the condenser are: In this manner, this process achieves current requirement of acid mist emissions of about 20 ppmv without depending on air dilution. e conversions of the three reactors are shown in Table 3. Table 3 Conversion in WSA reactors [4] e vapor sulfuric acid stream was then cooled in a heat exchanger by cooling water at 20 ºC before entering a 3-phase separator. At this point, the condensation of sulfuric took place where the hot sulfuric acid was condensed into liquid phase as the expected product. e clean gas is released at the top of separator. e developed owsheet of this WSA process is shown in Figure 2.

Claus Process Simulations Description
e Claus process is one of the most common processes for sulfur recovery from acid gas generated in oil and gas re ning. Elemental sulfur (S) is the nal product, produced from the reaction between H 2 S and SO 2 . e SO 2 itself is produced from the combustion of H 2 S and O 2 in the furnace [14]. In this simulation, the considered elemental sulfur is S 2 and S 8 . e reactions that occurred in the combustor are as follows:

Research Article
Applied Materials and Technology

Applied Materials and Technology
e outlet from this reactor was then cooled to 200 ºC and separated from the gas stream as the product stream. e upper gas stream was sent to the second reactor a er heated up to 240 ºC. In this reactor some of the produced SO 2 reacts with H 2 S to produce sulfur S 8 . is product stream is cooled down and enters a second separator at 200 ºC producing another S 8 according to the following reaction: e outlet stream contained some of S 8 , S 2 and H 2 S is then sent to a heater to heat up before entering the last reactor at 215 ºC. is reactor produces elemental sulfur S 8 as the nal product according to the following reaction: ree stages of reactors are used in series to increase the yield of recovery process. In addition, the steam is generated at 1 barg as Low-Pressure Steam (LPS) during the cooling down/condensed process. Simulation ow sheet of the Claus process is shown in Figure 3.

Evaluation of Operating Conditions
e steam production is used to calculate the pro tability of both processes. WSA process generates High Pressure Steam (HPS) at 40 barg (270 ºC) while Claus produces Low Pressure Steam (LPS) at 1 barg. HPS and LPS are worth 29.97 USD/ton and 29.29 USD/ton, respectively (15). Based on the simulation runs, regression/ surrogate models of the pro t functions (in $/hr) for both WSA and Claus processes are developed as function of feed capacity (x 1 ) and H 2 S concentration (x 2 ), shown as follows: Pro t WSA = -3.5x 1 + 10.5x 2 + 0.016x 1 2 -0.11x 2 2 + 0.045x 1 x 2 -15.58 Pro t Claus = 1.88x 1 -1.75x 2 -0.00252x 1 2 -0.0167x 2 2 + 0.0025x 1 x 2 +38.32 e accuracies of these models are shown by their multiple R values (square root of R 2 ). ese multiple R values show how strong the relationship between the data and the model where 1 or 100% being the strongest. us, equation 16 and 17 have the values of 97% and 85%, respectively. ese values are close to 100%, showing their reasonable accuracies. Figure 4 show the plots of these surrogate models for both processes in terms of pro tability (in $/hr). It can be seen that for most of the H 2 S concentrations and capacities, the WSA process (dark colour) is more pro table than the Claus process (light colour). e reason why the WSA process is more pro table in most of the operating conditions is due to the exothermic reactions occurred in WSA process that allows for more high -pressure steam to be generated. However, there is a narrow region where the Claus process is more pro table, which is at the lower concentration range of H 2 S. At this low concentration region, not enough high-pressure steam is generated to outweigh the necessary capitals required to build such facilities.

Planet (Environment) Regression Analysis
Based on result of GWP calculation, regression models of WSA and Claus processes are developed also as a function of feed capacity (x 1 ) and H 2 S concentration (x 2 ). ey are shown as follows.
ese models are plo ed as shown in Figure 5. From these plots, WSA is seen to be more environmentally friendly through its lower GWP, if not the same (the mid part of the region). For equation 20 and 21, their multiple R values are 80% and 71%, respectively. For this FEDI criteria, both equations have the lowest relationship with the data. Nonetheless, they are reasonably good (> 70%). ese values again suggest that the 2 nd order equation may have to be revisit in the future work. Figure 6 illustrates the plots of the regression models of the FEDI. Due to more efficient conversions of WSA process, the amount of ammable materials in the system is much lower. In can be seen from the plots as well that the higher the quantity of H 2 S (its concentration and feed capacity), the higher the FEDI index.

People (Safety) Regression
Toxicity hazard models via TDI are shown as follows and the corresponding plots are shown in Figure 9.  As shown in Figure 7, the WSA process has a higher TDI than the Claus process in the most part of the operating ranges. is is because the production of sulfuric acid that occurs in the WSA process, while it is not considered in the Claus process. In this regard, future work on comparing the same input and output of the processes are necessary.
Nonetheless, the current work has shown that in three sustainability indices, the WSA process seems to be more superior than the Claus process. In the most part of the operating conditions, the WSA process is more pro table, lower GWP, lower FEDI, and higher TDI. Future work should consider the Claus process to be integrated with a sulfuric acid plant, which can then be compared with the WSA process to get an apple-to-apple comparison.

CONCLUSIONS
In this work, Haldor Topsoe's Wet Gas Sulfuric Acid (WSA) and Claus processes were simulated using Symmetry -iCON as a 1 st principles modelling tool. e resulted mass and energy balance were then used to assess the suitability of these processes in various feed gas capacity and H 2 S concentrations.
ree pillars of sustainability of people (safety), planet (environment), and pro t were used as the comparison metrics. To account for more detailed comparisons, a huge number of simulation runs were needed. To do this much faster, regression models were developed from selected simulation runs via Central Composite Design (CCD) of Design of Experiment. e WSA process has been shown to be the more sustainable process for H 2 S conversion at most of the concentrations and capacity ranges. To be more speci c, the WSA process is more pro table, producing lower GWP index, lower FEDI, and higher TDI.