Category: Gas Processing

The best way to prevent hydrate formation (and corrosion) is to keep pipelines, tubing and equipment dry of liquid water. There are occasions, right or wrong, when the decision is made to operate a line or process containing liquid water. If this decision is made, and the process temperature is below the hydrate point, inhibition of this water is necessary. This is of particular interest in gas gathering systems [1] and subsea operations [2] during normal production as well as during shut down.

Many materials may be added to water to depress both the hydrate and freezing temperatures. For many practical reasons, a thermodynamic hydrate inhibitor (THI) such as an alcohol or one of the glycols is injected, usually methanol, diethylene glycol (DEG) or monoethylene glycol (MEG). All may be recovered and recirculated, but the economics of methanol recovery may not be favorable in many cases. Hydrate prevention with methanol and or glycols can be quite expensive because of the high effective dosage required (10% to 60% of the water phase). Large concentrations of solvents aggravate potential scale problems by lowering the solubility of scaling salts in water and precipitating most known scale inhibitors. The total injection rate of inhibitor required is the amount/concentration of inhibitor in the liquid water phase for the desired hydrate temperature suppression, plus the amount of inhibitor that will distribute in the vapor and liquid hydrocarbon phases. Any inhibitor in the vapor phase or liquid hydrocarbon phase has little effect on hydrate formation conditions.  Due to the accuracy limitations of the hydrate depression calculations and flow distribution in the process, it is recommended that the hydrate formation temperature with inhibition be chosen with a design factor below the coldest expected operating temperature of the system to ensure adequate inhibitor injection rates.

Solubility loss of MEG in the gas phase is negligible and loss to the liquid hydrocarbon phase is very low, 3.5 L/10⁶Sm³ (0.23 lbm/MMscf) [3]. Methanol losses are more significant, particularly vapor phase losses.  Based on Figure 6.20 in reference [3], depending on operating conditions, the solubility loss of methanol into the gas phase can be very high, typically 16 mg/Sm³ (1 lbm/MMscf) for every percent methanol in water phase. Losses to the liquid hydrocarbon are higher than for MEG but usually less than 1-2 % of hydrocarbon volume. At typical pipeline inhibition conditions, a solubility of about 0.4 kg/m³ (0.15 lbm/bbl) is generally adequate for planning purposes [3]. Depending on solubility losses, chemical makeup requirements for methanol can be very large and expensive for both once-through systems and methanol recovery units. In addition, the downstream processes like petrochemical and LNG plants cannot tolerate methanol in the feed gas.

Determination of the amount and concentration of inhibitors and their distribution in different phases is very important for practical purposes and industrial applications. Therefore, to determine the required amount and concentration of these inhibitors, several thermodynamic models for hand and rigorous calculations have been developed and incorporated into computer software [4].

As previously stated, a significant amount of methanol would be lost to the hydrocarbon phases, which may cause problems for refineries, petrochemical, and gas plants downstream. In gas plants where there is propane recovery the methanol will follow the propane product and be a potential cause for propane to go off specification. Methanol has also been known to cause premature failure in molecular sieves. In refineries the methanol must be washed out of the crude/condensate, where it presents a problem in wastewater treatment. In petrochemical plants methanol is also considered poison for catalysts.

In offshore production, gas lift and gas injection for pressure maintenance are becoming common practice. The associated gas produced with crude oil and water is separated, compressed and normally dehydrated with triethylene glycol (TEG) before injection or export. Since the produced oil/water/gas lines pose hydrate formation during normal production or shut downs, methanol is commonly injected to prevent hydrate formation and plugging of flow lines and gas lift/reinjection lines. Occasionally, TEG dehydration units shut down or may produce off spec dry gas which requires methanol injection. Consequently, some of the injected methanol ends up back in the produced oil/water/gas stream.

In this TOTM we will consider the presence of methanol in the produced oil/water/gas stream and determine the quantitative traces of methanol ending up in the TEG dehydrated gas. To achieve this, we simulated by computer an offshore production facility consisting of oil/water/gas multistage-separation, compression and TEG dehydration processes and determined the methanol concentration in the dried gas. We also studied the effect of wet gas temperature, the number of theoretical trays in the TEG contactor, the water content spec of dry gas, and lean TEG circulation rate on the dried gas methanol content. For this purpose methanol content in the production stream was assumed to vary from zero to 350 PPM (V).

Case Study:

A simplified process flow diagram (PFD) for the offshore production facilities considered in this study is shown in Figure 1. The production stream (oil, water, gas, and methanol) was passed through the high pressure separator where free water and gas were separated and the oil was passed through the intermediate and low pressure separators for subsequent gas separation from oil. The separators’ off gas streams were recompressed and cooled to 48 bara and 35 °C (696 psia and 95 °F) before entering the TEG contactor for dehydration. The dried gas was compressed further (not shown in the PFD) to 232 bara (3365 psia) for reinjection or export purposes. To meet a water content spec of 32 mg/Sm³ (2 lbm/MMscf) or lower, a lean TEG concentration of 99.95 weight percent was used in all of the simulation runs.

To study the impact of methanol (MeOH) concentration and determine its traces in the TEG dehydrated gas, the MeOH content of the production stream feed to the high pressure separator was assumed to vary from 0 to 350 PPM (V). This variation of MeOH content was chosen due to the uncertainty of its concentration in the production stream. The wet compressed gas temperature is an important parameter in the operation of a TEG unit and affects the water content of dried gas and the required lean TEG solution circulation rate and/or the number of required theoretical trays. Depending on the design and/or operational problem like scaling on the cooling side of the gas cooler, the wet gas temperature may be higher than 35 °C (95 °F). Therefore, the wet gas temperature was assumed to vary from 35 to 50 °C with 5 °C increment (95 to 122 °F and 9 °F increment). Depending on the requirement, 2 or 3 trays theoretical was used in the contactor unit. For each case the lean TEG solution rate was varied to meet the desired water content specification for each case.

Figure 1. Simple process flow diagram used in this case study

Results and Discussion:

The ProMax simulation software [6] was used to perform computer simulations for different cases of interest and determined the concentration/traces of MeOH in the TEG dehydrated gas. Twelve cases were studied in which the number of theoretical trays, wet feed gas temperature and pressure, dried gas water content, lean TEG solution weight percent, circulation rate, pressure and temperature were specified. For each case the MeOH content in the feed to the high pressure (HP) separator was assumed to vary from 0 to 350 part per million by volume, PPM (V) and the corresponding MeOH concentration in dried gas was determined. Table 1 presents the computer simulation results for one the 12 case studies. The absorption % presented in the last two columns of Table 1 is defined as:

Overall % = 100 (MeOH PPM in feed to the HP Separator – MeOH PPM in dry gas) /( MeOH PPM in feed to HP Separator)

TEG Contactor % = 100 (MeOH PPM in wet gas – MeOH PPM in dry gas) /( MeOH PPM in wet gas).

Notice the calculated MeOH absorption percents are relatively constant and independent of MeOH PPM in the feed to HP separator or wet gas. As shown in the last row of Table 1, the process overall and TEG contactor MeOH absorption percents are 61.1, and 30.6, respectively.

Table 1. Typical computer simulation results

Lean TEG Wt%=99.95			Lean TEG Solution Std Liquid Vol Rate = 2.91 m3/h (16.9 L TEG/kg Water)
No. of Theoretical Tray=3			Lean TEG Solution Temperature = 38 °C
Wet Gas To TEG      T, °C	MeOH, PPM (V)				Dry Gas			MeOH Absorption
	In the Feed of Inlet HP Separator	In Wet Gas To TEG		In TEG Dry Gas	Hydrate Formation T, °C	Water Content lbm/MMSCF	Water Content mg/Sm3	TEG Contactor %	Overall %
35	0	0		0.0	-20.0	1.00	16.0	NA	NA
	25.6	14.4		10.0	-20.0	1.00	16.0	30.6	60.9
	70.0	39.4		27.4	-20.0	1.00	15.9	30.5	60.8
	104.9	59.0		41.0	-20.0	0.99	15.9	30.5	60.9
	139.8	78.5		54.5	-20.0	0.99	15.9	30.6	61.0
	174.6	97.9		68.0	-20.1	0.99	15.8	30.5	61.1
	209.4	117.0		81.4	-20.1	0.99	15.8	30.4	61.1
	244.2	137.0		94.8	-20.1	0.98	15.8	30.8	61.2
	278.9	156.0		108.0	-20.1	0.98	15.7	30.8	61.3
	313.6	175.0		121.0	-20.1	0.98	15.7	30.9	61.4
	348.3	194.0		134.0	-20.2	0.98	15.7	30.9	61.5
Average								30.6	61.1

The MeOH concentration profile in the dried gas as a function of MeOH concentration in the feed to HP separator and MeOH concentration in the wet gas for 4 of these cases are presented in Figures 2 and 3, respectively.  Notice the number of theoretical trays (N=3) and dried gas water content spec (16 mg/Sm³) were kept constant in these two figures. These two figures indicate that as the wet gas temperature increases, the lean TEG solution increases; therefore, more MeOH is picked up (absorbed) by TEG solution. This is explained by the fact that as the wet gas temperature increases, its capacity to hold water vapor increases. Since the dry gas water content spec had to remain constant, more lean TEG solution is required to remove the water. Because of higher lean TEG circulation rate, more MeOH is also absorbed in the contactor.

Table 2 presents the summary results for all 12 cases investigated. As shown in this table, for dry gas water content spec of 16 mg/Sm³ (1 lbm/MMscf), in comparison to three theoretical trays, the case of two theoretical trays requires a higher circulation rate; therefore, the MeOH absorption factor increases drastically (lower traces of MeOH in dry gas). The MeOH absorption is a function of the TEG circulation rate through the contactor.  The greater the circulation rate, the greater the absorption.  This high MeOH absorption % is favorable for the downstream process units; however, for the cases of N=2 and wet gas temperatures of 45 and 50 °C, the required liter of TEG per kg of water removed is much higher than the recommended value. According to Chapter 18 of reference [5], the recommended range is from 16 to 50 liter of TEG per kg of water removed (2 to 6 gal TEG/lb water removed). For dry gas water content spec of 32 mg/Sm³(32 lbm/MMscf) and two theoretical trays, the required lean TEG circulation rate drops down within the recommended range for all four wet gas temperatures.

Figure 2. Variation of MeOH content of the TEG dehydrated gas as a function of MeOH content in the inlet separatoor feed and wet gas temperature

Figure 3. Variation of MeOH content of the TEG dehydrated gas as a function of the wet gas MeOH content temperature

Table 2. Summary of all simulation results

The results in Table 2 are also represented in Figures 4 and 5. These figures can be used to estimate quickly the MeOH content of TEG dehydrated gas for a specified number of theoretical trays, and the wet gas temperature and MeOH content.

In summary, a case study based on an offshore production facility was undertaken to investigate the impact of design and operational parameters on the trace of MeOH in the TEG dehydrated gas. The results can be summarized as follows:

1. The MeOH concentration in TEG dehydrated gas is proportional to the MeOH concentration in the feed to HP separator or in the wet gas to the TEG contactor.
2. The MeOH concentration in TEG dehydrated gas decreases as the wet gas temperature increases.
3. The MeOH concentration in TEG dehydrated gas decreases as the theoretical number of trays decreases, (or as the TEG circulation rate increases).

Notice, the above results were drawn based on twelve simulation runs for a single case study. They may be used for general guidelines. We believe each specific case should be analyzed separately and thoroughly.

To learn more about similar cases and how to minimize operational problems, we suggest attending the John M. Campbell courses; G4 (Gas Conditioning and Processing) and G5 (Gas Conditioning and Processing-Special).

John M. Campbell Consulting (JMCC) offers consulting expertise on this subject and many others. For more information about the services JMCC provides, visit our website at www.jmcampbellconsulting.com, or email your consulting needs to consulting@jmcampbell.com.

By: Dr. Mahmood Moshfeghian

Reference:

1. Bullin, K.A., Bullin, J.A., “Optimizing methanol usage for hydrate inhibition in a gas gathering system,” Presented at the 83^rd Annual GPA Convention – March 15, 2004.
2. Szymczak, S., Sanders, K., Pakulski, M., Higgins, T.; “Chemical Compromise: A Thermodynamic and Low-Dose Hydrate-Inhibitor Solution for Hydrate Control in the Gulf of Mexico,” SPE Projects, Facilities & Construction, (Dec 2006).
3. Campbell, J. M., “Gas Conditioning and Processing”, Vol. 1, The Basic Principles, 8^th Ed., Second Printing, J. M. Campbell and Company, Norman, Oklahoma, (2002).
4. Asadi Zeydabadi, B., Haghshenas, M., Roshani, S., and Moshfeghian, M., “Prevent system hydrate formation during sudden depressurization,” Hydrocarbon Processing, J., pp 83-91, April 2006.
5. Campbell, J. M., “Gas Conditioning and Processing”, Vol. 2, The Equipment Module, 8^th Ed., Second Printing, J. M. Campbell and Company, Norman, Oklahoma, (2002).
6. ProMax 3.1, Bryan Research and Engineering, Inc, Bryan, Texas, 2009.

Figure 4. Effect of wet gas temperature and number of theoretical trays absorption %  on the TEG contactor MeOH absorption %

Figure 5. Effect of wet gas temperature and dry gas water content spec on the TEG contactor MeOH absorption %

Many materials may be added to water to depress the hydrate temperature. For many practical reasons, a thermodynamic hydrate inhibitor (THI) such as an alcohol or one of the glycols is injected, usually methanol, diethylene glycol (DEG) or monoethylene glycol (MEG). All may be recovered and recirculated, but the economics of methanol recovery may not be favorable in many cases. Hydrate prevention with methanol and or glycols can be quite expensive because of the high effective dosage required (10 to 60% of the water phase). Large concentrations of solvents can aggravate potential scale problems by lowering the solubility of scaling salts in water and precipitating most known scale inhibitors. The high rates of methanol create a logistical problem as well as a health, safety, and environmental (HS&E) concern because of the handling issues associated with methanol. The total injection rate of inhibitor required is the amount/concentration of inhibitor in the liquid water phase for the desired hydrate temperature suppression, plus the amount of inhibitor that will distribute in the vapor and liquid hydrocarbon phases. Any inhibitor in the vapor phase or liquid hydrocarbon phase has little effect on hydrate formation conditions.  Due to the accuracy limitations of the hydrate depression calculations and flow distribution in the process, it is recommended that the hydrate formation temperature with inhibition be chosen with a design factor below the coldest expected operating temperature of the system to ensure adequate inhibitor injection rates.

Low dosage hydrate inhibitors (LDHIs) are relatively new and only recently reaching the “proven technology” stage in oil and gas processing.  Although LDHIs move the hydrate formation line to the left, it is only temporary. In typical systems they will “delay” the formation of hydrates for about 12 hours. The LDHIs are two classes of chemicals: Kinetic inhibitors (KHIs) and Anti-Agglomerants (AAs). A KHI can prevent hydrate formation but contrary to methanol cannot dissolve an already formed hydrate. Current KHIs have a difficult time overcoming a subcooling temperature (ΔT) threshold of about 15 °C (27 °F). AAs allow hydrates to form and maintain a stable dispersion of hydrate crystals in the hydrocarbon liquid. AAs form stable water in oil micro-emulsion. AAs adsorb onto the hydrate crystal lattice and disrupt further crystal growth but must have a liquid hydrocarbon phase present and the maximum water to oil ratio is about 40-50%.

Laboratory studies and field experiences indicate hydrate-inhibition synergy is gained through the combination of a THI and LDHI [1]. This is termed a hybrid hydrate inhibition (HHI). In the June 2010 tip of the month (TOTM) we demonstrated the synergy effect of mixed THIs like NaCl and MEG solution and presented a shortcut method to estimate the synergy effect of brine and MEG solution. In this TOTM, we will discuss the results of a successful application of combined methanol and a KHI solution for a well producing natural gas, condensate and water in the Gulf of Mexico (GOM). The following sections are based on the paper presented by Szymczak et al. [1].

As mentioned earlier, THIs are used in concentration ranging from 10 to 60 weight percent in water and LDHIs are used in concentration normally less than 5 weight percent. Proper combination of THI and LDHI will result in lower injection rates of the combined inhibitor mixture while controlling hydrate formation. In addition, the combined inhibitor mixture provides the ability to dissociate any hydrates that may form. Table 2 extracted from reference [1] presents the cost comparison between LDHI and methanol for various related activities. As can be seen in this table the cost of HHI for most activities is low and medium for unit cost and volume usage.

Table 1- Cost comparison of LDHI, Methanol and HHI for an offshore application [1]

Cost Factor	LDHI	Methanol	HHI
Unit Cost	Very High	Low	Medium
Transportation	Low	High	Low
Pump	High	High	Low
Storage	Low	High	Low
Crane Lifts	Low	High	Low
Corrosion	Low	High	Low
Volume	Low	High	Medium

Field Study:

To demonstrate the synergy effect of THI plus LDHI (HHI) and to illustrate the advantage of using HHI, we will discuss the results of a field study in the GOM reported by Szymczak et al. [1]. The well production flows 5.6 km (3½ miles) through 114 mm (4½-in) flowline to a production platform where natural gas, condensate and water are separated. There was a seven-line umbilical bundle that included a 9.5 mm (3/8-in) outside diameter line for methanol and/or LDHI injection. The hydrate-inhibitor injection point was at the tree. The recent gas composition is presented in Table 2 while detailed system information is shown in Table 3.

Table 2- Field Gas Composition [1]

Component	Mole %
Nitrogen	0.2045
Carbon Dioxide	0.5893
Methane	95.7432
Ethane	0.4462
Propane	0.3431
i-Butane	0.1508
n-Butane	0.1823
i-Pentane	0.1262
n-Pentane	0.1088
Hexane	0.1663
C7+	1.9392

To inhibit hydrate formation, a sufficient rate of methanol was injected to assure hydrate-free operation. Knowing the rate of water production, methanol was injected at approximately 0.019 m³/h (5 gal/hr). The injection rates were monitored and adjusted by comparing the chemical feed-line pressure at the wellhead and the flowline pressure measured at the platform.

Monitoring pressure drop between the inlet and outlet of pipelines is an industry-wide standard method of flow assessment. Fluctuating pressure drop values provide the operator with instant information concerning flow irregularities or obstructions. Only formed and dislodged hydrates manifest as rapid pressure fluctuations, whereas flow regime change or wax and scale build up result in gradual pressure changes. The GOM facilities operating experience showed that with only methanol in the system, the pressure difference between the wellhead and the flowline at the platform changed rapidly. The differential pressure changed as much as 345 kPa (50 psi) daily and was always between 1034 and 1724 kPa (150 and 250 psi) [1].

Table 3- Flowline Data [1]

Terrain	Flat
Gas Flow Rate	0.5663 x106 std m3/d (20 MMSCF/D)
Line Length	5.6 km (3.5 miles)
Line Diameter	114 mm (4.5 in)
Water Flow Rate	0.023 m3/d (6 gal/day)
Condensate	Traces
High Pressure	35, 853 kPa (5,200 psi)
Low Pressure	7,584 kPa (1,100 psi)
Average Pressure	27,579 kPa (4,000 psi)
Flow Speed	3.66 to 6.096 m/s (12 to 20 ft/sec)
Practical Methanol Rate	0.019 m3/h (5 gal/hr)
Sea Temperature	5 °C (41 °F)
Outlet Temperature	12.8 °C (55 °F)

Table 4 presents a summary of Szymczak et al. [1] calculation results for the worst case-scenario methanol injection rate. The relatively large dosage of methanol required was the result of a combination of temperature and gas volume conditions in the pipeline resulting in most of the injected methanol going into the vapor phase of the system at equilibrium conditions. For the detail of calculations, refer to Chapter 6, Volume 1, Gas Conditioning and Processing [2]. For methanol concentration below 25 weight percent, the Hammerschmidt [3] equation may be used. The practical 0.019 m³/h (5 gal/hr) rate of methanol applied resulted in borderline operating conditions between obstructed flow and line plugging. Szymczak et al. stated that the short fluid residence time in the flowline prevented the formation of a complete hydrate plug. Note that the high values of subcooling temperature eliminated KHI as the sole hydrate-prevention method. Known KHIs become ineffective inhibitors at approximately ΔT>15 °C (ΔT>27 °F) [1].

HHI Results

Szymczak et al. [1] reported that the inhibitor usage was reduced dramatically from 0.019 m³/h (5 gal/hr) of methanol to 0.0028 m³/h (0.75 gal/hr) of HHI and the pressure drop showed a lowering trend. They optimized the HHI dosage at approximately 0.0025 m³/h (0.67 gal/hr), a HHI rate sufficient to protect the flowline from producing hydrates in any case of rate or pressure/temperature fluctuation. This HHI rate represented an 80% reduction compared to the methanol injection rate. As a result of the injection rate reduction, the costs of transportation, pump maintenance, storage on the platform, corrosion inhibition of the flowline, labor and safety costs related to crane lifts, and pressure drop were reduced. For further detail on this field study, refer to Szymczak et al. paper [1].

Table 4- The worst case-scenario theoretical methanol injection rate requirement

Flowline Pressure Option	35, 853 kPa (5,200 psi)	27,579 kPa (4000 psi)
Hydrate depression (Subcooling)	23 °C (41.4 °F)	20 °C (36 °F)
Weight % methanol in water phase	23	20
Injection rate	0.045 m3/h (12 gal/hr)	0.035 m3/h (9.2 gal/hr)

In summary, HHI provides both thermodynamic and LDHI inhibition. From a cost standpoint, the HHI is cost-efficient compared to THIs. Additionally, the HHI can reduce corrosion and may eliminate the need for corrosion inhibitor.  From an offshore operational standpoint, the HHI significantly reduces logistical costs related to shipping, storage, handling, and chemical pumping. In addition to cost reduction, the problems related to health, safety, and environment (HS&E) would reduce too.

To learn more about similar cases and how to minimize operational problems, we suggest attending our G4 (Gas Conditioning and Processing) and G5 (Gas Conditioning and Processing – Special).

John M. Campbell Consulting (JMCC) offers consulting expertise on this subject and many others. For more information about the services JMCC provides, visit our website at www.jmcampbellconsulting.com, or email your consulting needs to consulting@jmcampbell.com.

By: Dr. Mahmood Moshfeghian

Reference:

1. Szymczak, S., Sanders, K., Pakulski, M., Higgins, T.; “Chemical Compromise: A Thermodynamic and Low-Dose Hydrate-Inhibitor Solution for Hydrate Control in the Gulf of Mexico,” SPE Projects, Facilities & Construction, (Dec 2006).
2. Campbell, J. M., “Gas Conditioning and Processing”, Vol. 1, The Basic Principles, 8^th Ed., Second Printing, J. M. Campbell and Company, Norman, Oklahoma, (2002).
3. Hammerschmidt, E. G. “Formation of Gas Hydrate in Natural Gas Transmission Lines”, Ind. Eng. Chem., 26, 851-855, (1934).

The best way to prevent hydrate formation (and corrosion) is to keep the pipelines, tubing and equipment dry of liquid water. There are occasions, right or wrong, when the decision is made to operate a line or process containing liquid water. If this decision is made, and the process temperature is below the hydrate point, inhibition of this water is necessary.

Many materials may be added to water to depress both the hydrate and freezing temperatures. For many practical reasons, a thermodynamic hydrate inhibitor (THI) such as an alcohol or one of the glycols is injected, usually methanol, diethylene glycol (DEG) or monoethylene glycol (MEG). All may be recovered and recirculated, but the economics of methanol recovery may not be favorable in many cases. Hydrate prevention with methanol and or glycols can be quite expensive because of the high effective dosage required (10% to 60% of the water phase). Large concentrations of solvents aggravate potential scale problems by lowering the solubility of scaling salts in water and precipitating most known scale inhibitors. The total injection rate of inhibitor required is the amount/concentration of inhibitor in the liquid water phase for the desired hydrate temperature suppression, plus the amount of inhibitor that will distribute in the vapor and liquid hydrocarbon phases. Any inhibitor in the vapor phase or liquid hydrocarbon phase has little effect on hydrate formation conditions.  Due to the accuracy limitations of the hydrate depression calculations and flow distribution in the process, it is recommended that the hydrate formation temperature with inhibition be chosen with a design factor below the coldest expected operating temperature of the system to ensure adequate inhibitor injection rates.

Determination of the amount and concentration of inhibitors and their distribution in different phases are very important for practical purposes and industrial applications. Therefore, to determine the required amount and concentration of these inhibitors, several thermodynamic models for hand and rigorous calculations have been developed and incorporated into computer software.

Low dosage inhibitors are relatively new and only recently reaching the “proven technology” stage in oil and gas processing.  Although these systems move the hydrate formation line to the left, it is only temporary.  In typical systems they will “delay” the formation of hydrates for about 12 hours. Low Dosage Hydrate Inhibitors (LDHIs) are two class of chemicals: Kinetic inhibitors (KHIs) and Anti-Agglomerants (AAs). A KHI can prevent hydrate formation but cannot dissolve an already formed hydrate. Current KHIs have a difficult time overcoming a subcooling temperature (ΔT) threshold of 15 °C (27 °F). AAs allow hydrates to form and maintain a stable dispersion of hydrate crystals in the hydrocarbon liquid. AAs form stable water in oil micro-emulsion. AAs adsorb onto the hydrate crystal lattice and disrupt further crystal growth but must have a liquid hydrocarbon phase and the maximum water oil ratio is about 40-50%.

Laboratory studies and field experiences indicate hydrate-inhibition synergy is gained through the combination of two or more THIs [1] or THI and LDHI [2]. This is termed a hybrid hydrate inhibition (HHI).

In this TOTM we will demonstrate the synergy effect of mixed THIs like NaCl and MEG solution. In the next TOTM, we will discuss the results of a successful application of combined methanol and a KHI solution for a well producing natural gas, condensate and water in the Gulf of Mexico (GOM).

Combined THIs (MEG + NaCl or MEG + KCl)
The produced water from natural gas reservoirs contains an electrolyte solution such as NaCl, KCl, and CaCl2. In order to estimate the hydrate formation temperature in the presence of mixed thermodynamic inhibitors, we propose to add up the depression temperature due to each individual inhibitor. The steps are summarized below:

1. Using a conventional method described in reference [3], estimate the hydrate formation temperature in the presence of pure water, To.
2. Using a method similar to Javanmardi et al. [1], estimate the hydrate depression temperature due to the presence of salt solution, salt ΔT.
3. Using a method similar to Hammerschmidt [4], estimate the hydrate depression temperature due to the presence of MEG solution, MEG ΔT.
4. Add up Salt ΔT and MEG ΔT, Total ΔT.
5. The hydrate formation temperature is calculated by subtracting total ΔT from To.

As an example, Table 1 presents the detail of calculation and the contribution of each inhibitor to the hydrate formation temperature for methane gas at different pressures and mixed inhibitor concentration. Comparison of the estimated hydrate formation temperature (last column of Table 1) with the experimental data (the fifth column) measured by Masoudi et al. [5] indicates a relatively good agreement. Figures 1 and 2 also present the contribution of each inhibitor to the hydrate formation temperature as described above for mixed solution of NaCl + MEG and KC l+ MEG, respectively.

Table 2 presents a comparison between the accuracy of the proposed method with Javanmardi et al. method against the experimental data for methane gas in the presence of mixed inhibitors. Table 2 also indicates an average absolute temperature difference of 4.7 and 3.5 °C for the proposed method and Javanmardi et al. method, respectively.

In summary, a simple procedure is proposed for estimation of the hydrate formation temperature in the presence of mixed THIs such as MEG plus a salt solution. This procedure can be used for a mixture of glycol and electrolyte solutions. The procedure is relatively simple and its accuracy is good enough for facility calculations. For more accurate prediction of hydrate formation temperature in the presence of electrolytes, the readers should refer to the papers presented by Javanmardi et al. [1] and Masoudi et al. [5].

To learn more about similar cases and how to minimize operational problems, we suggest attending our G4 (Gas Conditioning and Processing) and G5 (Gas Conditioning and Processing – Special) courses.

John M. Campbell Consulting (JMCC) offers consulting expertise on this subject and many others. For more information about the services JMCC provides, visit our website at www.jmcampbellconsulting.com, or email your consulting needs to consulting@jmcampbell.com.

By: Dr. Mahmood Moshfeghian

Reference:

1. Javanmardi, J., Moshfeghian, M. and R. N. Maddox, “An Accurate Model for Prediction of Gas Hydrate Formation Conditions in Mixture of Aqueous Electrolyte Solutions and Alcohol,”Canadian J. of Chemical Engineering, 79, 367-373, (2001).
2. Szymczak, S., Sanders, K., Pakulski, M., Higgins, T.; “Chemical Compromise: A Thermodynamic and Low-Dose Hydrate-Inhibitor Solution for Hydrate Control in the Gulf of Mexico,” SPE Projects, Facilities & Construction, (Dec 2006).
3. Campbell, J. M., “Gas Conditioning and Processing”, Vol. 1, The Basic Principles, 8th Ed., Second Printing, J. M. Campbell and Company, Norman, Oklahoma, (2002).
4. Hammerschmidt, E. G. “Formation of Gas Hydrate in Natural Gas Transmission Lines”, Ind. Eng. Chem., 26, 851-855, (1934).
5. Masoudi, R., Tohidi, B., Anderson, R., Burgass, R., and Yang, J. “Experimental Measurement and Thermodynamic Modelling of Clathrate Hydrate Equilibria and Salt Solubility in Aqueous Ethylene Glycol and Electrolyte Solutions,” Fluid Phase Equilibria, 219, 157-163 (2004).

In the February 2010 tip of the month (TOTM) we presented the distribution and concentration of sulfur-containing compounds in an NGL Fractionation (NF) plant using HYSYS [1] with the Peng-Robinson equation of state (PR EOS) [2]. In this TOTM we will present the distribution and concentration of the sulfur-containing compounds in the same NF plant using ProMax [3] and VMGSim [4] both using the PR EoS. These two simulation results will be compared with the HYSYS [1] results. The software’s built-in binary interaction parameters were used in this study. The NF plant is the same as the one described by Alsayegh et al. [5]. The feed composition, rate, condition, and product specifications are shown in Tables 1 and 2 and the plant process flow diagram is shown in Figure 1 of the February 2010 TOTM. An overall tray efficiency of 90 percent was used for all columns.

Expected Product Distribution: Figure 1, reproduced from Figure 9 of a paper published by Likins and Hix [6], shows a descending order log scale bar-graph of the pure compounds vapor pressure for the components of interest to this study. This figure shows that COS should distribute to both the ethane and the propane streams. MeSH, with a vapor pressure close to n-butane should distribute primarily with the butanes with a small amount distributing to the pentane stream. EtSH, having a vapor pressure between butane and pentane, should distribute primarily with butane and pentane. CS₂ should distribute primarily to the pentane and the C₆⁺ streams with only minor distribution to the butane stream. The heavier sulfur compounds should end up almost entirely in the C₆⁺ stream.

Results of Computer Simulation:

The NF plant described in the previous section was simulated using HYSYS [1], ProMax and VMGSim based on the PR EOS [2]. In this study, the respective software built-in (library) binary interaction parameters were used even though we recommend evaluating the accuracy of VLE results against experimental data and if necessary the insertion of VLE data regression into the EOS interaction parameters. This regression may be required to adequately model the systems dealing with mercaptans.

1. Table 1. Concentration (PPM, mole) of sulfur containing compounds in the gas and product streams

The focus of this study is on the distribution (% recovery) and concentration (PPM) of the sulfur-containing compounds in the product streams. Table 1 presents the PPM concentration of sulfur-containing compounds in the feed and product streams. Figures 2 through 8 present bar-graphs of the recovery of each sulfur-containing compound in the gas and product streams. The mole percent recovery is defined as the number of moles of a component in the product stream divided by the moles of the same component in the feed stream (Stream 5). In these figures, the gas and product streams are followed by letters H, P, and V representing HYSYS, ProMax, and VMGSim results, respectively.

H₂S: Figure 2 shows the distribution and recovery of H₂S in the gas, C₂ and C₃ product streams. As expected, the majority of the H₂S distributes in the gas and the C₂ product streams. As can be seen in this figure, the results of the simulators are the same.

COS: Figure 3 shows the distribution and recovery of COS in the gas, C₂, and C₃. As expected, the majority of the COS ends up in the C₃ product stream. As can be seen in this figure, the results of the three simulators are almost the same.

MeSH: Figure 4 shows the distribution and recovery of MeSH in the gas, C₃, and C₄ product streams. For HYSYS and VMGSim, contrary to the data presented in Figure 1, the majority of the MeSH distributes to the C₃ stream rather than to the C₄ stream. However, the ProMax result follows the same trend as in Figure 1 and the majority of MeSH distributes to the C₄ stream.

EtSH: Figure 5 shows the distribution and recovery of EtSH in the C₃, C₄, and C₅ streams. Unexpectedly, HYSYS predicts that the majority of the EtSH ends up in the C₄ stream rather than the C5 product as would be expected based on the data of Figure 1. However, the results of ProMax and VMGSim are closer to the Figure 1 data.

CS₂: Figure 6 shows the distribution and recovery of CS₂ in the C₄ and C₅ product streams. Contrary to the Figure 1 pure CS₂ behavior the results of HYSYS and VMGSim show that the majority of the CS₂ ends up in the C₄ stream. However, based on the ProMax results, the majority of the CS₂ ends up in the C₅ stream which is consistent with data in Figure 1.

iC₃SH: Figure 7 shows the distribution and recovery of iC₃SH in the C₄, C₅ and C₆₊ product streams. As expected, iC₃SH ends up in the C₅ and C₆₊ streams. Notice that ProMax shows a higher concentration of iC₃SH in the C₅ product stream while HYSYS and VMGSim predict lower but nearly the same recovery of iC₃SH.

iC₄SH: Figure 8 shows recovery of iC₄SH in the C₆₊ product stream. All of the iC₄SH ends up in the C₆₊ stream as expected when the Figure 1 data is analyzed.

Conclusions:

The calculation results presented and discussed here are specific to the NGL fractionation plant studied here, but there are some general conclusions that can be drawn from this study.

The results indicate that the highest concentration of methyl mercaptan (MeSH) is present in the C₃ product (stream 15) based on HYSYS and VMGSim but its highest concentration is in the C₄ product (stream 20) based on the ProMax results.

The results of HYSYS indicate that the highest concentration of ethyl mercaptan (EtSH) is present in the C₄ product (stream 20) but ProMax and VMGSim results indicate that its highest concentration occurs in the C₅ Product (stream 23).

The highest concentration of carbon disulfide (CS₂) is present in C₅ Product (stream 23) according to the three simulator results.

The binary interaction parameters used in the EOS play an important role in the VLE behavior of the system under study, and affect the distribution of the sulfur-containing compounds present in the feed. Use of improper or incorrect binary interaction parameters may generate erroneous results. Care must be taken to use correct values of binary interaction parameters. In this study, the simulator library values of the binary interaction parameters were used.

The predictions by HYSYS, ProMax, and VMGSim in Figures 4 through 7 (showing the distribution of MeSH, EtSH, CS₂, and iCH₃SH respectively) contain some disagreements. The results also indicate that these compounds were not distributed among the hydrocarbon products in the same way one would expect from their volatilities and concentrations. This may be explained by the conclusion reported by Harryman and Smith [7, 8] who wrote “iC₃SH is formed during fractionation within the depropanizer and the deethanizer.” Therefore, further evaluation should be conducted to arrive at a concrete decision. In an upcoming TOTM, we will investigate the VLE behavior of the theses systems using experimental data. This should be a good reason to perform laboratory tests and detailed thermodynamic calculations to determine process flow rates and composition. Detailed process analysis shouldalways be made to justify and prove correct decisions as to selection of process flow schemes.

To learn more about similar cases and how to minimize operational problems, we suggest attending the John M. Campbell courses; G4 (Gas Conditioning and Processing), G5 (Gas Conditioning and Processing – Special) and G-6 Gas Treating and Sulfur Recovery.

John M. Campbell Consulting (JMCC) offers consulting expertise on this subject and many others. For more information about the services JMCC provides, visit our website at www.jmcampbellconsulting.com, or email your consulting needs to consulting@jmcampbell.com.

By: Dr. Mahmood Moshfeghian

Reference:

1. ASPENone, Engineering Suite, HYSYS Version 7.0, Aspen Technology, Inc., Cambridge, Massachusetts U.S.A., 2009.
2. Peng, D.,Y. and D. B. Robinson, Ind. Eng. Chem. Fundam. 15, 59-64, 1976.
3. ProMax 3.1, Bryan Research and Engineering, Inc, Bryan, Texas, 2009.
4. VMGSim 5.0.5, Virtual materials Group, Inc, Calgary, Alberta, 2010.
5. Al-Sayegh, A.R., Moshfeghian, M.  Abbszadeh, M.R., Johannes, A. H. and R. N. Maddox, “Computer simulation accurately  determines volatile sulfur compounds,” Oil and Gas J., Oct 21, 2002.
6. Likins, W. and M. Hix, “Sulfur Distribution Prediction with Commercial Simulators,” the 46th Annual Laurance Reid Gas Conditioning Conference Norman, OK 3 – 6 March, 1996.
7. Harryman, J.M. and B. Smith, “Sulfur Compounds Distribution in NGL’s; Plant Test Data – GPA Section A Committee, Plant design,“ Proceedings 73^rd GPA Annual Convention, New Orleans, Louisiana, March, 1994.
8. Harryman, J.M. and B. Smith, “Update on Sulfur Compounds Distribution in NGL’s; Plant Test Data – GPA Section A Committee, Plant design,“ Proceedings 75^th GPA Annual Convention, Denver, Colorado, March, 1996.

Because phase envelope generation and its impact on design and performance of gas processing plants is so important it has been the topic of several Tips Of The Month (TOTM). As emphasized by Rusten et al. [1], there are several challenges that have to be addressed in order to succeed with the phase envelope modeling of real natural gases. The most important are:

1. Sampling procedures
2. Sample preparations
3. Chromatographic gas analysis. A detailed composition is required for satisfactory input to thermodynamic models
4. Thermodynamic models that correctly predict the phase envelope

In this TOTM we will demonstrate the impact of thermodynamic modeling for rich gases in the dense phase region. For a discussion on the dense phase, please see the January 2010 TOTM. The value of the dense phase viscosity is very similar to gas phase viscosity. The dense phase density is closer to the liquid phase density. Therefore, it has become attractive to transport rich natural gas in the dense phase region. In October 2005 we discussed several methods of C₇₊ (heavy ends) characterization and checked the accuracy of several methods and presented tips to improve the accuracy of each method. These methods are presented briefly below. For more detail, please refer to Gas Conditioning and Processing, Volume 3, Advanced Techniques and Applications [2].

Method A: The C₇₊ is treated as a single hypothetical component based on its molecular weight (MW) and specific gravity (SG). The normal boiling point is predicted; the critical temperature, critical pressure, and acentric factor are also predicted using correlations similar to the ones by Riazi and Duabert [3].

Adjusting MW (or Tc) in Method A: By adjusting the molecular weight of the C₇₊ fraction we can closely match the measured dew point. The critical temperature (Tc) can also be adjusted to make the phase envelope curve pass through the measured dew point. The Tc adjustment is preferred because less work is involved to match the calculated and experimental values.

Method B: The C₇₊ is broken into Single Carbon Numbers (SCN) ranging from SCN 7 to SCN 17+ using the exponential decay procedure presented by Katz [4] and applied by others [5-7].

Method C: The large number of SCN components of Method B may be lumped into 4 cuts. The properties of the lumped cuts are estimated from the individual SCN components.

Method D: This method is similar to Method B except that 12 normal parafins (alkanes) are used to represent the C₇₊instead of SCN components. The advantage of this method is that n-alkane components are readily available in many commercial software packages but the SCNs may not be.

Tuning MW in Method D: The distribution (i.e. mole %) of the alkane part of the C₇₊ depends on the assumed value of the C₇₊ MW.

Tuning the binary interaction parameters, k_ij, in Methods B and C: A common correlation to estimate the binary interaction parameter is:

In the above equation, ν_ci and ν_cj represent the critical volumes of components i and j, respectively. The default value of exponent n is normally set to 1.2 but it can be used as a tuning parameter to match the experimentally measured dew point.

In this TOTM we will generate the dew point curve for the rich gas shown in Table 1 using the C₇₊ characterization methods described above. The dew point curve portion of the phase envelope for this gas was generated using both HYSYS [8] and ProMax [9] simulation software by the Soave-Redlich-Kwong (SRK) [10] (Figure 1) and Peng-Robinson (PR) [11] (Figure 2). The experimentally measured dew point pressure [12] is also show in these two figures as a red triangle.

Figures 1 and 2 were generated using a single C₇₊ cut with the relative density and molecular weight shown in Table 1. Other required properities were estimated using the default options of the simulation software. As can be seen in these figures using the PR Equation of State with ProMax gives the closest prediction of the experimentally measured dew point. As decribed above the MW can be adjusted to match experimantal and calculated data.

The single carbon number (SCN) analysis as described in Method B above was used for further tuning of the thermodynamic model, The predidicted dew point pressures for the different cases studied here are shown in Table 2. Figure 3 demonstrates the same information graphically.

Using Method B, the experimental dew point is most closely represented using four SCNs with a combined molecular weight of 118.2. The properties and mole percent distribution of these four SCN components for the optimum case are given in Table 3.

Table 4 shows the improvement made in the dew point prediction by using four SCNs with a modified molecular weight of 118.2 instead of a single C₇₊ cut. The ProMax PR EOS is used for both cases. The predicted dew point curves for these two cases can be seen in Figure 4.

As can be seen in Figure 4, proper characterization of the heavy components (see Tables 3 and 4) can improve the quality of the phase envelope and match the experimentally measured dew point in the dense phase region. For a detailed discussion of this topic, the readers may refer to the Rusten et al. paper [1].

To learn more about similar cases and how to minimize operational problems, we suggest attending the John M. Campbell courses; G4 (Gas Conditioning and Processing) and G5 (Gas Conditioning and Processing – Special).

John M. Campbell Consulting (JMCC) offers consulting expertise on this subject and many others. For more information about the services JMCC provides, visit our website at www.jmcampbellconsulting.com, or email your consulting needs to consulting@jmcampbell.com.

By: Dr. Mahmood Moshfeghian

Reference:

1. Rusten, B.H., Gjertsen, L.H., Solbraa, E.,  Kirkerød, T., Haugum, T. and Puntervold, s., “Determination of the phase envelope – crucial for process design and problem solving,” presented at the 87^th GPA National Convention, Grapevine, 2008
2. Maddox, R. N. and L. Lilly, “Gas Conditioning and Processing, Computer Applications for Production/Processing Facilities,” John M. Campbell and Company, Norman, Oklahoma, 1995.
3. Riazi, M.R. and T.E. Daubert, Hydr. Proc. P. 115, (March) 1980
4. Katz, D. J. Petrol. Technol., 1205-1214, (June) 1983.
5. Whitson, C. H. SPE J., 683-694, (August) 1983
6. Starling, K. E. Presented at the American Gas Association Operations Conference, Orlando, FL, April 27-30, 2003
7. Moshfeghian, M., Maddox, R.N., and A.H. Johannes, “Application of Exponential Decay Distribution of C₆₊ Cut for Lean Natural Gas Phase Envelope,” J. of Chem. Engr. Japan, Vol  39, No 4, pp.375-382, 2006
8. ASPENone, Engineering Suite, HYSYS Version 7.0, Aspen Technology, Inc., Cambridge, Massachusetts U.S.A., 2009.
9. ProMax^®, Bryan Research & Engineering Inc, Version 3.2, Bryan, Texas, 2009
10. Soave, G., Chem. Eng. Sci. 27, 1197-1203, 1972.
11. Peng, D.,Y. and D. B. Robinson, Ind. Eng. Chem. Fundam. 15, 59-64, 1976.
12. Sage, B.H, and R.H. Olds, AIME 170, 156–173, 1947.

Natural gas liquids (NGLs) consist of the hydrocarbon components in a produced gas stream that can be extracted and sold. Common NGL products are ethane (C₂H₆), propane (C₃H₈), butanes (iC₄H₁₀ and nC₄H₁₀) and natural gasoline (C₅₊).  Ethane is the lightest NGL and its recovery can be justified in those areas where a ready petrochemical market and a viable transportation network exist. Ethane is mainly used as a petrochemical feedstock. Propane is used for petrochemical feedstock, and also finds wide application as a domestic and industrial fuel. Propane is frequently sold as a mixture of propane and butane called LPG (Liquefied Petroleum Gas).

The market for butanes is primarily as a petrochemical feedstock, fuel and/or for gasoline blending when vapor pressure requirements allow it. Isobutane (iC₄) is the most valuable of the NGLs. Its primary use is as refinery feedstock for manufacture of high octane blending components for motor gasoline. Normal butane can be used as a feedstock to olefin plants where it is converted to mono-olefins (ethylene and propylene) and the diolefin, butadiene as well as other by-products. The largest use for isobutane is as a gasoline blending component for octane number and vapor pressure control. Natural gasoline refers to the pentanes and heavier components in a gas stream and they are also commonly referred to as condensate or naphtha; it usually consists primarily of straight and branched chain paraffins. Natural gasoline is most commonly used as refinery feedstock, although it can also be used as a petrochemical feedstock. The details of the processes required, and the principles of their operation are discussed in Maddox and Lilly [1], and Maddox and Morgan [2]. A summary about the distribution of sulfur-containing compounds is presented on pages of 287-291 [2].  Specifically, page 290 presents the conclusions from the papers presented by Harryman and Smith [3, 4] which highlight the complexity of sulfur-containing distribution in the NGL product streams.

Raw NGL feed to an NGL fractionation (NF) plant may contain sulfur-containing compounds such as carbonyl sulfide (COS), methyl mercaptan (MeSH), ethyl mercaptan (EtSH), carbon disulfide (CS₂), isopropyl mercaptan (iC₃SH), isobutyl mercaptan (iC₄SH), etc. For the purpose of meeting NGL products specification, it is important to accurately determine the distribution and concentration of the various mercaptans during NF process.

Likins and Hix [5] evaluated the accuracy of four commercial simulation programs by comparing their predicted K-values with the experimentally measured values. They concluded that “In this limited evaluation against laboratory VLE data, no one program can be claimed to be an outstanding winner. Although simulator D does an excellent job with one system, it poorly predicts behavior in the second system and is surpassed by simulator B. Simulator C behaves erratically in that its predictions range from excellent to horrible (dimethyl sulfide) depending on the component.” They also simulated two different NF plants using commercial simulation programs and compared the distribution and concentration of mercaptans in different product streams with field data.  Again, they concluded that none of the simulators do a good job modeling the sulfur distribution overall.

In order to improve the accuracy of commercial simulators, Alsayegh et al. [6] presented a procedure to determine the binary interaction parameters between mercaptans and hydrocarbons using experimentally measured vapor-liquid equilibria (VLE).

In this tip of the month (TOTM), we will determine the distribution and concentration of different mercaptans in an NGL fractionation plant using HYSYS [7] Peng-Robinson [8] equation of state. The built-in HYSYS binary interaction parameters were used in this study. The NF plant is the same as the one described by Alsayegh et al. [6]. The feed composition, rate, and condition are shown in Table 1 [6] and the plant process flow diagram is shown in Figure 1 [6].

The column specifications are shown in Table 2 [6].  An overall tray efficiency of 90 percent was used for all columns. In the last column of Table 2, DV and D represent the vapor and the total rate of the overhead stream, respectively. Therefore, the D_V/D is the vapor fraction in the overhead product stream. In addition, reflux ratio (L/D) is defined as the reflux rate (L) divided by the total overhead stream rate.

Expected Product Distribution: Figure 2, reproduced from Figure 9 of Likins and Hix paper [5], shows a descending order log scale bar-graph of the pure compounds vapor pressure for the components of interest to this study. This figure shows that COS should distribute to both the ethane and the propane streams. MeSH, with a vapor pressure close to n-butane should distribute primarily with the butanes with a small amount distributing to the pentane stream. EtSH, having a vapor pressure between butane and pentane, should distribute primarily with butane and pentane. CS₂should distribute primarily to the pentane and the C₆⁺ streams with only minor distribution to the butane stream. The heavier sulfur compounds should end up almost entirely in the C₆⁺ stream.

Results of Computer Simulation:

The NF plant described in the previous section was simulated using HYSYS [7] based on the Peng-Robinson equation of state (EOS) [8]. In this study, the HYSYS built-in binary interaction parameters were used even though we recommend insertion of VLE data regression into the EOS interaction parameters. This regression is required to adequately model the systems dealing with mercaptans. Table 3 presents the mole percent recovery of each component in the product and gas streams predicted by HYSYS. The mole percent recovery is defined as the number of moles of a component in the product stream divided by the moles of the same component in the feed stream (Stream 5). Table 3 also presents the vapor fraction, temperature, pressure, and flow rate of each stream. The focus of this study is on the distribution (% recovery) and concentration (PPM) of the sulfur-containing compounds in the product streams. Table 4 presents the PPM concentration of sulfur-containing compounds in the product streams.

Figures 3 through 9 present bar-graphs of the recovery of each sulfur-containing compound in the product streams.

H₂S: Figure 3 shows the distribution and recovery of H₂S in the gas, C₂ and C₃ streams. As expected, the majority of the H₂S distributes in the gas and the C₂ streams.

COS: Figure 4 shows the distribution and recovery of COS in the gas, C₂, and C₃. As expected, the majority of the COS ends up in the C₃ stream.

MeSH: Figure 5 shows the distribution and recovery of MeSH in the gas, C₃, and C₄ streams. Contrary to the data presented in Figure 2, the majority of the MeSH distributes to the C₃ stream rather than to the C4 stream.

EtSH: Figure 6 shows the distribution and recovery of EtSH in the C3, C₄, and C₅ streams. Unexpectedly, the majority of the EtSH ends up in the C₄ stream rather than C5 as would be expected in Figure 2.

CS₂: Figure 7 shows the distribution and recovery of CS₂ in the C4, and C5 streams. Contrary to the pure CS₂ behavior (Figure 2), the majority of the CS₂ ends up in C4 stream.

iC₃SH: Figure 8 shows the distribution and recovery of iC₃SH in the C4, C5 and C6+. As expected, iC₃SH ends up in C5 and C6+ streams.

iC₄SH: Figure 9 shows recovery of iC₄SH in the C6+ stream. As expected, all of the iC₄SH ends up in the C6+ stream.

Conclusions:

The calculation results presented and discussed here are specific to the liquid fractionation plant studied here, but there are some general conclusions that can be drawn from this study.

The results indicate that the highest concentration of ethyl mercaptan (EtSH) and carbon disulfide (CS₂) are present in the C₄ product (stream 20) and C5 Product (stream 23), respectively. The highest concentration of methyl mercaptan (MeSH) is present in the C₃ product (stream 15).

The binary interaction parameters used in the EOS play an important role in the VLE behavior of the system under study, and affect the distribution of the sulfur-containing compounds present in the feed. Use of improper or incorrect binary interaction parameters may generate erroneous results. Care must be taken to use correct values of binary interaction parameters. In this study, the HYSYS library values of the binary interaction parameters were used.

Some of the sulfur-containing compounds (i.e. MeSH, EtSH, and CS₂) were not distributed among the hydrocarbon products in the same the way one would expect from their volatilities and concentrations. This may be explained by the conclusion reported by Harryman and Smith who wrote “iC3SH is formed during fractionation within the depropanizer and the deethanizer”.  This should be a good reason to perform laboratory tests and detailed thermodynamic tray calculations to determine process flow rates and composition. Detailed process analysis should always be made to justify and prove correct decisions as to selection of process flow schemes.

To learn more about similar cases and how to minimize operational problems, we suggest attending the John M. Campbell courses; G4 (Gas Conditioning and Processing), G5 (Gas Conditioning and Processing-Special) and G6 (Gas Treating and Sulfur Recovery).

John M. Campbell Consulting (JMCC) offers consulting expertise on this subject and many others. For more information about the services JMCC provides, visit our website at www.jmcampbellconsulting.com, or email your consulting needs to consulting@jmcampbell.com.

By: Dr. Mahmood Moshfeghian

Reference:

1. Maddox, R. N. and L. Lilly, “Gas Conditioning and Processing, Computer Applications for Production/Processing Facilities,” John M. Campbell and Company, Norman, Oklahoma, 1995.
2. Maddox, R. N. and D. J. Morgan, “Gas Conditioning and Processing, Gas Treating and Sulfur Recovery Vol. 4,” John M. Campbell and Company, Norman, Oklahoma, 2006.
3. Harryman, J.M. and B. Smith, “Sulfur Compounds Distribution in NGL’s; Plant Test Data – GPA Section A Committee, Plant design,“ Proceedings 73^rd GPA Annual Convention, New Orleans, Louisiana, March, 1994.
4. Harryman, J.M. and B. Smith, “Update on Sulfur Compounds Distribution in NGL’s; Plant Test Data – GPA Section A Committee, Plant design,“ Proceedings 75^th GPA Annual Convention, Denver, Colorado, March, 1996.
5. Likins, W. and M. Hix, “Sulfur Distribution Prediction with Commercial Simulators,” the 46th Annual Laurance Reid Gas Conditioning Conference Norman, OK 3 – 6 March, 1996.
6. Al-Sayegh, A.R., Moshfeghian, M.  Abbszadeh, M.R., Johannes, A. H. and R. N. Maddox, “Computer simulation accurately  determines volatile sulfur compounds,” Oil and Gas J., Oct 21, 2002.
7. ASPENone, Engineering Suite, HYSYS Version 7.0, Aspen Technology, Inc., Cambridge, Massachusetts U.S.A., 2009.
8. Peng, D.,Y. and D. B. Robinson, Ind. Eng. Chem. Fundam. 15, 59-64, 1976.

Process simulation computer programs are excellent tools for designing or evaluating gas processing plants, chemical plants, oil refineries or pipelines. In these simulation programs, most of the thermodynamic properties are calculated by an equation of state (EOS). The cubic equations of state can be regarded as the heart of these programs for generating the required properties. However, none of the equations of state is perfect and often some sort of tuning must be done prior to their applications. Some tuning is already done by researchers and has been embedded in the data base of these simulation programs. In dealing with non-standard or complex systems, the user should check the validity and accuracy of the selected thermodynamic package (i.e. EOS) in the simulation programs prior to attempting to run the desired simulation. Often the users find that tuning is required. This can be done by performing a series of vapor liquid equilibria (VLE) calculations such as dew point, bubble point or flash calculations and comparing the results with the field data or experimental data. If the accuracy is not within acceptable range, then the EOS should be tuned to improve its accuracy. The tuning can be done in several ways but the one most often used is adjusting/regressing the binary interaction parameters between binary pairs in the mixture using the experimental PVT or VLE data.

In this tip of the month (TOTM), we will demonstrate how the binary interaction parameters are tuned in a simulation program to improve the accuracy of a selected EOS. For this purpose, we will demonstrate how the accuracy of the bubble point pressure prediction of a ternary system of carbon dioxide, pentadecane, and hexadecane can be improved. We will use the Peng-Robinson (PR) [1] equation of state in ProMax [2] and the experimental VLE data published in the literature [3]. The same procedure can be used with any EOS in other simulation programs.

The PR EOS
The PR EOS [2] in terms of pressure (P), volume (v) and temperature (T) is defined as:

The values of the parameters a and b must be determined in a special way for mixtures. Any equation, or series of equations, used to obtain mixture parameters is called a combination rule or mixing rule. The calculation, regardless of its exact form, is based on the premise that the properties of a mixture are some kind of weighted average summation of the properties of the individual molecules comprising that mixture.
The mixing rules used in cubic equations of state (i.e., Peng-Robinson, Soave-Redlich-Kwong, and van der Waals) are:

Where: a and b = the interaction energy and molecular size parameters for the mixture
a_i, b_i = a and b parameters for component i in the mixture
x_i = composition (mol fraction) for component i in the mixture
k_ij = binary interaction parameter
n = number of component in the mixture
R = Universal gas constant
The a_i and b_i for each component in the mixture are calculated in terms of critical temperature (Tc_i), pressure (Pc_i), and  acentric factor (?_i) as presented in equations 4 and 5.

Once a and b have been determined, the equation of state computations proceed as though a and b were for a pure component. With cubic equations of state the mixing rules sum the properties based on binary pairs.

The binary interaction parameter, k_ij, has no theoretical basis. It is empirical and is used to overcome deficiencies in the corresponding states theory or the basic model (equation of state). Binary interaction parameters are regressed from experimental data for a specific model and should be applied in that model only. In addition, k_ij’s can be determined from regression of PVT data or VLE data. This will result in different k_ij’s for the same binary mixture.

The Effect of k_ij on Bubble Point Pressure Prediction
To study the effect of the k_ij, the bubble point pressure for a binary mixture of CO₂ (1) and pentadecane (2) at 40 °C for a series of CO₂ mole % in the liquid phase were predicted using the PR EOS in ProMax. First, the default value of the binary interaction in the data base (DB) of ProMax in which k₁₂=0.0 was used.  The predicted results were compared with the experimental values and the average absolute percent deviation (AAPD) for eight data points calculated to be 41.06%. This AAPD was reduced to 1.64% when the binary interaction parameter of k₁₂=0.112 was used. Figure 1 presents the effect of k₁₂ on the predicted bubble point pressure of CO2 and pentadecane mixture. This figure demonstrates clearly the role of k_ij in improving the accuracy for bubble point pressure calculations. The improvement is substantial and the accuracy now is as good as the experimental data.

Similar improvement is observed when the binary interaction parameter, k₁₂, was changed from zero, and the default value in data base (k₁₂=DB) of ProMax, to 0.112 for the binary mixture of CO₂ (1) and hexadecane (2) at 40 °C. For this case the AAPDs were 40.65%, 3.64% and 1.26% for k₁₂=0.0, k₁₂=DB, and k₁₂=0.112; respectively.

For these two systems the liquid densities were also predicted and compared with the experimental values. For CO₂and pentadecane binary system, the calculated AAPD for liquid densities were 6.10% and 6.36% for k₁₂=0.0 and k₁₂=0.112; respectively. Similar AAPD changes were observed for CO₂ and hexadecane binary mixture.

Normally, the binary interaction parameters obtained from regressing binary mixture VLE data work well in multicomponent systems. This is demonstrated by using the same obtained kijs in a ternary mixture. The obtained binary interaction parameters of CO₂ + pentadecane and CO₂ + hexadecane were used without any further change to predict the bubble point pressure of the ternary mixtures of CO₂ (1) + pentadecane (2) + hexadecane (3). Figure 3 indicates these binary interaction parameters obtained from the individual binary mixtures improve the accuracy of EOS considerably. Similar to the case of binary mixtures, when the binary interaction parameters, k₁₂, k₁₃, were changed from zero, and the default value of ProMax data base (kijs=DB), to 0.112 for the ternary mixture of CO₂ (1) + pentadecane (2) + hexadecane (3) at 40 °C, the AAPDs were reduced from 40.99%  and 25.16% to 1.75%, respectively.

Discussion and Conclusions
It was shown that the binary interaction parameters of an EOS can be adjusted/tuned/regressed to improve the accuracy of VLE calculations considerably. It was also shown that when the regressed binary interaction parameters based on the binary experimental VLE data used without further changes in a multicomponent system considerable improvement in accuracy may be obtained.

It is a sound practice to check the accuracy of a selected thermodynamic package prior to running any simulation. However, experimental or field data are required to fulfill this task.

To learn more about similar cases and how to run process simulations, we suggest attending our G40(Process/Facility Fundamentals), G4 (Gas Conditioning and Processing) and G5 (Gas Conditioning and Processing – Special) courses.

By: Dr. Mahmood Moshfeghian

References:

1. Peng, D.Y. and Robinson, D.B., “A New Two-Constant Equation of State,” Ind. Eng. Chem., Fundam., Vol. 15, No. 1, P. 59, 1976.
2. ProMax, V. 3.0, Bryan, Tex.: Bryan Research & Engineering Inc, 2009.
3. Tanaka, H., Yamaki, Y. and Kato, M., “Solubility of Carbon Dioxide in Pentadecane, Hexadecane, and Pentadecane + Hexadecane,” J. Chem. Eng. Data,38, 386-388,1993.