Computer simulation accurately determines volatile sulfur compounds

Oct. 21, 2002
We developed a computer simulation to determine the distribution of mercaptans in liquid petroleum products. We then applied this simulation to the NGL fractionation (NF) unit in the Bandar Imam Petrochemical Complex, Iran.

We developed a computer simulation to determine the distribution of mercaptans in liquid petroleum products. We then applied this simulation to the NGL fractionation (NF) unit in the Bandar Imam Petrochemical Complex, Iran.

The simulation used the Soave-Redlich-Kwong (SRK) equation of state (EOS) for all the vapor-liquid and thermodynamic calculations. Recently published experimental vapor-liquid equilibrium (VLE) data for mercaptans and light hydrocarbons helped determine optimum binary interaction coefficients for each system.

The data regression option in Aspen Plus determined these optimum values. We simulated the entire NF plant as one complete unit to ascertain the optimized binary interaction coefficients and distribution of each trace mercaptan.

NGLs, mercaptans

The removal of heavier components from natural gas streams leaving primary separators at field separation sites is a regular practice in natural-gas processing facilities. The level of removal depends on specific plant circumstances. Individual problems may differ but the solutions tend to be similar.

The first level of processing is typically the removal of heavier components to meet a specified hydrocarbon dew point for the effluent gas. This is usually accomplished with minimal processing and low operating costs. There is, however, little or no return on capital through product sales.

Click here to enlarge image

The operator can expend more effort, time, and capital towards liquids recovery processing if there is market for NGLs such as butane, propane, and even ethane; this creates a more profitable overall venture.

Removing the liquefiable components decreases the gas stream's volume. This "shrinkage" becomes part of the cost of liquids extraction.

Currently used commercial NGL extraction processes include:

Absorption using lean oil.

Adsorption, which usually consists of a short-cycle unit and hydrocarbon recovery units.

Condensation, which requires gas cooling that may include mechanical refrigeration, a turboexpander, or valve expansion.

NGLs include the hydrocarbon components in a produced gas stream that can be extracted and sold.

Common NGL products are ethane, propane, butanes, and natural gasoline (iC5+). Ethane recovery is justified in areas where a ready petrochemical market and a viable transportation network exist. Ethane is mainly a petrochemical feedstock.

Propane is a petrochemical feedstock, and also finds wide application as a domestic and industrial fuel. Propane is frequently sold as a mixture of propane and butane (LPG).

The market for butanes is primarily as a petrochemical feedstock, a fuel, and for gasoline blending. Isobutane is the most valuable NGL. Its primary use is as refinery feedstock for high-octane, gasoline-blending components.

Normal butane is a feedstock for the manufacture of ethylene, propylene, and butadiene. The largest use of butane is as a gasoline-blending component for octane.

Natural gasoline refers to the pentanes and heavier components; they are also commonly referred to as condensate or naphtha, and usually consist of straight and branched-chain paraffins.

Their most common use is as a refinery feedstock, although the petrochemical market for natural-gasoline-range hydrocarbons is growing. References 1 and 2 discuss natural gas processes and principles of operation.1 2

Bandar Imam

The Bandar Imam Petrochemical Complex (BIPC) in southern Iran required a facility to produce more feedstock. The natural gas feed to the planned facility, NGL 1200, contains a large amount of hydrogen sulfide and organic sulfur compounds such as mercaptans. These components made the feed unsuitable as a feedstock to the petrochemical complex.

The operator required an engineering study of the BIPC NGL fractionation (NF) plant to accurately determine the distribution and concentration of various mercaptans in the Bandar Imam NGL products; primarily due to their high sulfur concentrations and also due to the sweetening capacity limitation of the NF plant.

The design considered two alternatives to meet the National Iranian Oil Co. LPG product specification (NIOC LPG 010). The first alternative was to expand the NF plant's sweetening capabilities by removing all sulfur-containing compounds before separating the LPG from the other NGLs.

The second alternative was to lower the mercaptans in the NGL 1200 products before pipelining the NGL product to BIPC. This second alternative would also require a liquid polishing unit.

Equations of state

A key decision in any commercial process flow diagram simulation is to properly choose the proper EOS. Normally reliable and dependable EOSs such as the SRK3 and the Peng-Robinson (PR)4 can produce reliable results for systems of normal low-to-middle molecular weight hydrocarbons. We selected SRK for this study.

For hydrocarbon-mercaptan systems, one would expect that component-dependent interaction parameters (kij) are required for best results. The AspenTech databank included kij values for the requisite mercaptans, but information on the VLE data sources from which they were derived was unavailable.

Recent data5 are available for the components involved. We decided to use the new data and derive new values of kij for all components. These new optimized binary interaction coefficients were used in all of the process flow calculations for the BIPC NF plant.

SRK equation of state

The Equations box shows the SRK EOS, its mixing rules, and related expressions. The binary interaction coefficient, kij, used to calculate the "a" parameter, is an important tuning factor; cubic EOSs like the SRK are sensitive to this parameter.

Experimental binary VLE data are normally used to determine kij. The details and the impact of the kij parameters for cubic equations of state were presented and discussed in detail by Moshfeghian and Maddox.6

In this study, we used VLE data for mercaptans and hydrocarbons reported by Valtz, et al.,5 to determine the optimized binary interaction parameters. Valtz measured a complete set of VLE data for one, two, and three carbon mercaptan-propane systems and mercaptan-ethane systems.

With these VLE data and the data regression (DRS) of Aspen Plus,7 we optimized the required binary interaction parameters between volatile mercaptans and ethane, and volatile mercaptans and propane to minimize the errors between calculated and experimental phase compositions.

Click here to enlarge image

Table 1 presents a comparison between the experimental VLE data and those calculated by the SRK EOS using the optimized kij for the binary system C3H8-CS2. Agreement between calculated and experimental values is good.

Similar to the C3H8-CS2 system, we optimized the kij for other binary hydrocarbon-mercaptan systems. Table 2 summarizes the results, which demonstrate good agreement between calculated and experimental values.

Click here to enlarge image

As with any EOS, the accuracy of SRK calculations depends on the binary interaction coefficients. Any results based on incorrect values are wrong and should not be used.

Commercial simulation packages normally store binary interaction coefficients for many pairs of components in a library. Sometimes, however, the needed kij is missing or inaccurate.

One should ensure accurate kij values for process calculations. Checking calculation results using kij against reliable experimental VLE data is always advisable.

The NF unit

Chiyoda Corp. designed the BIPC NF unit. The NF unit consists of two identical parallel trains with a total capacity of 130,000 b/d of NGL feed. The plant includes a dehydration unit and sweetening-polishing unit for the ethane, propane, and butane products. This helps the operator meet the NIOC LPG 010 specification standard for each product.

Click here to enlarge image

Fig. 1 shows a simplified process flow diagram.

The NF unit has a depropanizer column; propane and the lighter components are distilled overhead and a C4+ stream is produced as the bottom product. The depropanizer overhead is fed to the de-ethanizer where propane (Stream 15) is the bottoms product. De-ethanizer overhead products include those components lighter than propane, which feed to the demethanizer column.

Demethanizer bottoms is an ethane product (Stream 16). Depropanizer bottoms feeds to the debutanizer where butane (Stream 20) is produced from the overhead.

Debutanizer bottoms include components heavier than butane, which feed to a depentanizer. The overhead product is pentane (Stream 23) and the bottoms product includes those components heavier than pentane (Stream 24).

Click here to enlarge image

NF feedstock, transported via pipeline, comes from several NGL units. For this study, we specified a feed containing large amounts of mercaptans (Table 3).

Click here to enlarge image

Table 4 presents the fractionating column specifications.

Simulations

We simulated the NF unit for three different cases.

Click here to enlarge image

Case 1, simulation test. In this case, we simulated the plant using the feed in Table 3 without any mercaptans to determine how calculated results compared with the calculations used for the original NF plant design.8 Table 5 shows a comparison between the results of this calculation and the original design. The comparison indicates good agreement between the two sets of calculations.

Click here to enlarge image

Case 2, simulation with Aspen databank kij values. In this case, we used the feed shown in Table 3 representing no mercaptan removal in the NGL 1200 unit. Table 3 represents the worst case scenario (highest level) of mercaptans. In this case, we used the original values kij in the Aspen databank for the interactions between hydrocarbons and mercaptans. Table 6 shows the distribution of mercaptans in the main product streams.

Click here to enlarge image

Case 3, simulation with optimized interaction parameters. Our calculations in this case were identical to Case 2 except that we used optimized kij values determined in this work (Table 2). Table 7 shows a summary of calculation results and the distribution of sulfur-containing compounds in LPG product streams.

Table 8 shows a summary of calculation results for Cases 2 and 3. This table also shows the distribution and concentration of volatile mercaptan-containing compounds in the product streams.

Click here to enlarge image

An analysis of Table 8 data indicates that the mole fraction of mercaptans in the propane product (Stream 15) using optimized kij is about six times larger than that calculated using the original kij in the Aspen databank. This difference is appreciable and cannot be ignored.

Table 8 verifies that the binary interaction coefficients play an important role in the accuracy of simulation results.

Click here to enlarge image

Table 9 shows the concentration of mercaptans in the feed stream to the sweetening unit for Cases 2 and 3. The highest concentration of mercaptans in this stream is 46 ppm, and the highest concentration of mercaptans in the feed to the C4 sweetening unit is 637 ppm.

Binary interaction coefficients

Simulation results for Cases 2 and 3 indicate that the binary interaction coefficients used in the SRK EOS play an important role in the VLE behavior of this system, and affect the distribution of the sulfur-containing compounds in the feed.

Using an improper or incorrect binary interaction coefficient can generate erroneous results. One must take care to use correct values for kij.

The results also indicate that the highest concentration of ethyl mercaptan (C2H5SH) and carbon disulfide (CS2) exists in the pentane product (Stream 23). The highest concentration of methyl mercaptan (CH3SH) is 74 ppm and is present in the propane product (Stream 15).

A close analysis of Table 8 data indicates that the kij's between mercaptans and light hydrocarbons from the Aspen databank are relatively accurate except for methyl mercaptan (CH3SH).

Based on the simulation results, which show a high content of mercaptans in the NGL products, and due to the limited sweetening capabilities of the NF unit, our recommendation is that the operator install a sweetening-polishing unit at the NGL 1200 plant.

Although the calculation results we present and discuss are specific to the liquid fractionation unit of the BIPC, there are some general conclusions that can be drawn from this study.

NGL product stream sulfur specifications are generally based on the mercaptan content of the liquid hydrocarbon streams. For example, the maximum allowable concentration of H2S in "sweet" natural gas is about 4 ppm, a small quantity. Allowable sulfur concentration levels of the other streams are on the same order.

In the BIPC fractionation train there were no concentration surprises. Mercaptans and other contaminants were distributed among the hydrocarbon products exactly as one would expect based on their volatilities and concentrations.

This should not lead to the assumption that there is no need to perform tray calculations on which to base process flow decisions. The operator should always make detailed process calculations to justify correct decisions when selecting process flow schemes.

References

  1. Maddox, R.N., and Lilly, L., "Gas Conditioning and Processing, Computer Applications for Production/Processing Facilities," John M. Campbell and Co., Norman, Okla., 1995.
  2. Maddox, R.N., and Morgan, D.J., "Gas Conditioning and Processing, Gas Treating and Sulfur Recovery Vol. 4," John M. Campbell and Co., Norman, Okla., 2000.
  3. Soave, G., Chem. Eng. Sci., 27, pp. 1197-1203, (1972).
  4. Peng, D.-Y., and Robinson, D.B., Ind. Eng. Chem. Fundam., 15, pp. 59-64, (1976).
  5. Valtz, A., Schwartzentruber, J., Richon, D., and Renon, H., "Vapor Liquid Equilibria for Volatile-Sulfur-Containing Systems. I. Propane Sulfur-Containing Systems and II. Ethane-Sulfur-Containing Systems," Research Report RR-135, Gas Processors Association, Tulsa, Okla., (1994).
  6. Moshfeghian, M., and Maddox, R.N., "Developing Binary Interaction Parameters for a Cubic Equation of State," Engineering J. of Qatar University, Vol. 4, pp. 7-14, (1991).
  7. Aspen User Manuals, Version 9.2, Aspen Technology, Inc., Cambridge, Ma., 1995.
  8. "The NGL Fractionation Operation Manual, Vol. 1," Krupp Lummus, Bandar Imam, (1991).

The authors

Click here to enlarge image

Abdulreda A. Alsaygh is assistant professor of chemical engineering at the University of Qatar, Doha. He holds a BS from the University of Qatar, an MS from Florida Institute of Technology, and a PhD from Oklahoma State University, all in chemical engineering.

Click here to enlarge image

Mahmood Moshfeghian is professor of chemical engineering at the University of Qatar. He previously served as professor of chemical engineering at the University of Shiraz, Iran. Moshfeghian holds a BS (1974), an MS (1975), and a PhD (1978) from Oklahoma State University, all in chemical engineering. He is a member of AIChE, the Iranian Association of Chemical Engineering, the Iranian Institute of Chemistry & Chemical Engineering, and the Iranian Petroleum Institute.

Click here to enlarge image

Mohammad R. Abbaszadeh is managing director of Energy Industries Engineering & Design (EIED), a subsidiary of Oil Industries Engineering Co., Tehran. Previously, he was advisor to the managing director of Oil Industries Engineering & Construction, technical director of National Petrochemical Co., managing director for Tabriz Petrochemical Co., and dean of the faculty of science at Shahid University, all in Iran. Abbaszadeh holds a PhD in physical organic chemistry from Essex University, UK.

Click here to enlarge image

Arland H. Johannes is professor of chemical engineering at Oklahoma State University, Stillwater. From 1977 to 1984, he was a professor of chemical engineering at Rensselaer Polytechnic Institute, Troy, NY. He holds a BS (1970) in chemistry and physics from Illinois State University, an MSE (1974) in civil engineering from West Virginia University, and a PhD (1977) in chemical engineering from the University of Kentucky. He is a member of AIChE.

Click here to enlarge image

Robert N. Maddox is the Leonard F. Sheerar chair emeritus professor of chemical engineering at Oklahoma State University. He served as head of the chemical engineering department from 1958 to 1978. He is a fellow of AIChE and a member of GPA.