Nov-2019

Simulation of Claus unit performance

Simulating a sulphur recovery unit using a fundamental approach enables the effects of process changes on a host of performance metrics to be reliably assessed.

SIMON WEILAND and RALPH WEILAND
Optimized Gas Treating, Inc.

Article Summary

Whether at the level of an individual piece of equipment or an entire plant, optimisation is usually done using simulation tools. The first step, however, is to verify that the tools chosen are able to reproduce the measured performance of the unit under its present operating conditions. This article benchmarks the SulphurPro SRU simulator against the measured performance of a sulphur recovery unit (SRU) processing an ammonia-rich acid gas in which a significant concentration of COS is formed and recycled back to the reaction furnace. Combustion air is oxygen enriched and, at just over 6 mol%, the ammonia content of the combined acid gas stream, amine acid gas (AAG) plus sour water acid gas (SWAG), is greater than the CO2 content. This provides a fairly stringent test of a model that relies on fundamental chemical reaction kinetics and heat transfer rate calculations to predict performance, rather than one that uses curve fits to forecast performance by looking in retrospect at how similar plants have performed in the past.

The discussion begins with the case study of a refinery SRU producing about 125 t/d of sulphur from 3.4 MMscfd of AAG with 91% H2S plus 6.8% CO2, balance nitrogen on a dry basis comingled with a sour water acid gas flow of 0.6 MMscfd that is 45 mol% H2S and 55 mol% ammonia. This forms the base case. Simulation includes SRU feed gas preparation, the main body of the SRU itself, and the treatment of the effluent gas from the last sulphur condenser through the final burner, the hydrogenation reactor, and the quench tower. In other words, the SRU from introduction of AAG, SWAG, air, and enriching oxygen through to the point of entry of the tail gas into the tail gas treating unit (TGTU). The TGTU was not part of the study so it was not simulated; however, it could have been completely integrated into the SRU model in a single flowsheet and simulated on a mass transfer rate basis with recycle of the recovered H2S back to the start of the SRU.

Base case
Figure 1 shows the flowsheet for the base case. Pure oxygen is used to enrich the oxygen content of the intake air to 28.5%. The total combined (enriched air) flow is adjusted using the solver block marked ADA (air demand analyser) to ensure the total air flow rate from the flow multiplier (marked MULT) in the figure results in a H2S to SO2 molar ratio of 1.9 in the gas leaving the final sulphur condenser.

The purpose of this article is to use simulation to quantify the effect of oxygen enrichment on the sulphur processing capacity of the plant, and in the same processing show unit how sensitive such parameters as SRU throughput, ammonia destruction, and WHB performance parameters are to the level of oxygen enrichment. To this end, an additional computational (solver) loop was added to the flowsheet in order to calculate the plant feed sulphur flow necessary to have the same total molar flow rate of gas from the first condenser into the converter (Stream 26 in Figure 1). The flow rate of this stream is taken to represent the SRU’s gas handling capacity, as discussed more fully below. For the base case study, however, this solver block was left disabled, and the gas flow calculated from the first condenser (but with the COPE recycle removed) formed the capacity basis for other levels of oxygen enrichment.

To ensure simulator credibility, the first task was to compare simulated versus measured performance indicators. Table 1 shows the relative percentage deviation of simulated results from measured data for some 25 of the parameters for which measurements were made. Every parameter is predicted within a few percent of measured levels.

The SulphurPro simulator can also provide a little more detailed picture of the operation of the catalytic converters by breaking the reactors into a number of discrete incremental depths of catalyst bed in order to assess temperature and conversion profiles across the beds. This can be done in both design and rating modes. SulphurPro allows the catalyst beds to have varying levels of activity relative to the fresh catalyst.

Figure 2 shows temperature profiles across the catalyst beds in the three converters. Points are the average of two thermocouple readings at opposite ends of a bed diameter. Converters-2 and -3 were simulated as having fresh catalyst beds. When Converter-1 was simulated this way, it was obvious that the catalyst was to some degree aged or deactivated. When simulated as just under 60% deactivated, the solid black line was obtained, so in an overall sense the bed does indeed appear to have a little over half of its original activity remaining.

It is interesting to note that, as measured, the temperature profile in the first converter actually has two rather flat sections joined by a rapid transition from one to the other. This is a classic profile of what one would expect for a catalyst being poisoned by a contaminant in the feed. However, the contaminant, whatever it is, seems to be readily adsorbed onto the catalyst and deactivation will move through the converter’s catalyst bed like a wave. Possible contaminants (poisons) might include BTEX components or even the soot formed from their decomposition.
 SulphurPro gives a faithful rendition of what was actually observed in the operating unit; therefore, one can only conclude that at least in this instance it is a reliable model. It is important to emphasise that model calculations are pure predictions done completely without tuning of any kind. The basis for the predictions is fundamental laboratory measurements of chemical reaction kinetics combined with well-founded models for heat transfer. There is no direct reliance on tuning any parameters to plant performance data; thus, SulphurPro is solidly placed to predict the performance of individual units and the SRU as a whole, both  robustly and with good accuracy.

Effect of air enrichment on performance
The base case already discussed used 28.5% oxygen in the enriched air. The objective now is to determine exactly how enrichment affects several key performance parameters in the as-built plant. Including the base case, six enrichment levels were considered: air containing 21%, 28.5%, 40%, 60%, 80%, and 100% O2. It should be understood that this is not a design study; rather, the SRU under all the conditions being studied must operate within the constraints of the existing equipment and flowsheet configuration without modifications.

There are several solver blocks that could be used in the flowsheet solution process for this study. A solver block in SulphurPro is intended to determine the value of a parameter within an operating block needed to achieve a specified value of a specified parameter in a certain stream:
• The function of the solver labelled ADA is to calculate the total enriched air flow to the reaction furnace needed to maintain the H2S:SO2 ratio of about two in the vapour from the final condenser (Stream 20).

• Another solver (acid gas flow control) could be used to calculate the total flow rate of sulphur-bearing gas into the SRU (Stream 46) corresponding to a given total flow rate of gas from the first condenser and into the converter/condenser bank (Stream 26). If no COPE1 recycle is needed (corresponding to zero to low enrichment), the gas flow from the first condenser (Stream 43) is often taken as a fair measure of the SRU’s gas handling capacity. However, if higher levels of enrichment are contemplated, COPE recycle is necessary to keep the reaction furnace outlet temperature below the thermal limits of the furnace refractory. If one is interested in high levels of oxygen enrichment, the gas flow in Stream 26 is the practical measure of capacity. Thus, acid gas flow control would ensure the SRU is fed with the right amount of gas to keep it at full capacity (Stream 26).

• With oxygen enriched air, furnace temperatures can become high enough to exceed the thermal limits of the furnace’s refractory and of the material in the waste heat boiler, especially with regard to corrosion. In the original case, 28.5% enriched air resulted in a simulated furnace effluent temperature of about 2400°F (1316°C).  For each subsequent case in this study, the solver block COPE recycle could be set up to calculate the recycle gas flow by solving for the split of first-condenser gas between the COPE recycle (Stream 8) and the feed to the first reheater (Stream 26) necessary to keep the reaction furnace effluent at 2400°F.