Category: Gas Processing

In this tip of the month (TOTM) we will present the compressor calculations of a case study. We will compare the rigorous method results with the values from the short cut methods. The rigorous method is based on an equation of state like the Soave-Redlich-Kwong (SRK) for calculating the required enthalpies and entropies. The enthalpies and entropies are used to determine the power requirement and the discharge temperatures.  The results indicate that the accuracy of the shortcut method is sensitive to the value of heat capacity ratio, k.

Power Calculations

The theoretical power requirements are independent of compressor type; the actual power requirements vary with the compressor efficiency. In general the power is calculated by:

where  mass flow rate and h is specific enthalpy.

From a calculation viewpoint alone, the power calculation is particularly sensitive to the specification of flow rate, inlet temperature and pressure, and outlet pressure. Gas composition is important but a small error here is less important providing it does not involve the erroneous exclusion of corrosive components. A compressor is going to operate under varying values of the variables affecting its performance. Thus the most difficult part of a compressor calculation is specification of a reasonable range for each variable and not the calculation itself. Reference [1] emphasizes that using a single value for each variable is not the correct way to evaluate a compression system.

Normally, the thermodynamic calculations are performed for an ideal (reversible process). The results of a reversible process are then adapted to the real world through the use of an efficiency. In the compression process there are three ideal processes that can be visualized: 1) an isothermal process, 2) an isentropic process and 3) a polytropic process. Any one of these processes can be used suitably as a basis for evaluating compression power requirements by either hand or computer calculation. The isothermal process, however, is seldom used as a basis because the normal industrial compression process is not even approximately carried out at constant temperature.

For an isentropic (reversible and adiabatic) process, equation 1 can be written as:

and based on the polytropic process:

The isentropic head is calculated by equation 3A:

Similarly, the polytropic head is calculated by equation 3B:

The actual discharge temperature based on the isentropic path is calculated by equation 4A.

The actual discharge temperature based on the polytropic is calculated by equation 4B.

where η and η_P are the isentropic (or adiabatic) and polytropic efficiency, respectively, P₁ suction pressure, P₂ discharge pressure, T₁ and T₂ arethe suction and discharge temperatures, respectively, q is gas volume flow rate at standard condition of P_S and T_S, Z_a average gas compressibility factor, k heat capacity ratio, R the gas constant, and n is the polytropic path exponent. Equations 1 and 2 are equally correct theoretically. The practical choice depends on the available data, although it is somewhat arbitrary. The power calculation should be made per stage of compression and then summed for all stages connected to a single driver. For general planning purposes the graphical solutions shown in reference [2] produce results comparable to these equations.

Equation of State (EOS)

The heart of any commercial process flow simulation software is an equation of state. Due to their simplicity and relative accuracy, normally a cubic EOS such as Soave Redlich-Kwong (SRK) [3] or Peng-Robinson [4] is used. These equations are used to calculate phase behavior, enthalpy, and entropy. With proper binary interaction coefficients, the process simulation results of these two equations are practically the same. Therefore, only the SRK was used in this work.

Step-by-Step Computer Solution

For known gas rate, pressure (P₁), temperature (T₁), and composition at the inlet condition and discharge pressure (P_2­), computation of compressor power requirement is based on an EOS using a computer and involves two steps:

1. Determination of the ideal or isentropic (reversible and adiabatic) enthalpy change of the compression process. The ideal work requirement is obtained by multiplying mass rate by the isentropic enthalpy change.
2. Adjustment of the ideal work requirement for compressor efficiency.
   The step-by-step calculation based an EOS is outlined below.
   1. Assume steady state, i.e.   and the feed composition remain unchanged.
   2. Assume isentropic process, i.e. adiabatic and reversible
   3. Calculate enthalpy h₁=f(P₁, T₁, and composition) and suction entropy s₁=f(P₁, T₁, and composition) at the suction condition by EOS
   4. For the isentropic process . Note the * represents ideal value.
   5. Calculate the ideal enthalpy () at outlet condition for known composition, P₂ and .
   6. The ideal work is 
   7. The actual work is the ideal work divided by efficiency or 
   8. The actual enthalpy at the outlet condition is calculated by 
   9. The actual outlet temperature is calculated by EOS for known h₂, P₂, and composition.

The efficiency of the compressor, and hence, the compression process obviously depends on the method used to evaluate the work requirement. The isentropic efficiency is in the range of 0.70 to 0.90.

If the compressor head curve and efficiency curve are provided by the manufacturer,  the head is determined from the actual gas volume rate at the inlet condition. Second, from the head, the actual work, discharge pressure and finally the discharge temperature are calculated.

Case Study

The gas mixture with the composition shown in Table1 at 105 °F (40.6 °C) and 115 Psia (793 kPa) is compressed using a single-stage centrifugal compressor with the polytropic head and efficiency curves shown in Figures 1 and 2 at a speed of 7992 rpm. The total feed gas volumetric flow rate was 101 MMSCFD (2.86×10⁶ Sm³/d).

Table 1. Feed gas analysis

Figure 1. Compressor polytropic head and best efficiency point

Figure 2. Compressor polytropic efficiency

Results and Discussions

SRK (Rigorous Method): The feed composition, temperature, pressure, volumetric flow rate at standard condition along with the compressor polytropic head and efficiency curves data were entered into the ProMax  software [5] to perform the rigorous calculations based on the SRK EOS. The program calculated polytropic and isentropic efficiencies, heads, compression ratio (discharge pressure), discharge temperature and power. For the actual gas flow rate at the inlet condition, the polytropic efficiency is close to the compressor best efficiency point (BEP). The program also calculated the gas relative density, heat capacity ratio (k), and polytropic exponent (n). These calculated results are presented in the SRK columns of Table 2 (bold numbers with white background).

Table 2. Summary of the rigorous and shortcut calculated results

The bold numbers with white background are the calculated values

 

Short-1 (Shortcut Method): In this method, we used equations 2 through 4 to calculate the polytropic and isentropic heads, the discharge temperature and power. We used the ProMax calculated polytropic and isentropic efficiencies, compression ratio (P₂/P₁), heat capacity ratio (k) and polytropic exponent (n) to calculate head, power, and the discharge temperature. The results are presented in the short-1 columns of Table 2. Note the short-1 results (discharge temperature, adiabatic and polytropic heads and power) are very close to the SRK values. The calculated actual discharge temperature by equation 4A (isentropic path: 265.3˚F=129.6˚C) was slightly lower than by equation 4B (polytropic path: 265.9 ˚F=129.9 ˚C).

Short-2 (Shortcut Method): Similar to short-1 method, we used equations 2 through 4 to calculate the polytropic and isentropic heads, the actual discharge temperature and power. We used only the ProMax calculated values of polytropic efficiency (n_P), compression ratio (P₂/P₁), and relative density (y). The heat capacity ratio (k) was estimated by equation 5:

The polytropic exponent (n) was estimated by equation 6.

The isentropic (adiabatic) efficiency () was estimated by equation 7.

The results for this method are presented in the short-2 columns of Table 2. The calculated discharge temperature by equation 4A (isentropic path) was exactly the same as by equation 4B (polytropic). Note the short-2 results (discharge temperature, adiabatic and polytropic heads and power) are deviated from the SRK values.

The results in Table 2 indicate that an increase of 2.2% in k (from 1.224 to 1.251) results in power increase of 1.42%. The polytropic exponent (n) increased by 3% and isentropic efficiency () decreased by 0.5 %. The difference in the actual discharge temperatures of the SRK and short-2 values is 17.5 ˚F (9.7˚C).

With the exception of actual discharge temperature, these differences between the SRK and short-2 methods results for facilities calculations and planning purposes are negligible. Note that the accuracy of the shortcut method is dependent on the values of k and n. In Short-1 method in which we used the k and n values from the SRK method the results were identical to those of SRK method.

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. Maddox, R. N. and L. L. Lilly, “Gas conditioning and processing, Volume 3: Advanced Techniques and Applications,” John M. Campbell and Company, 2^nd Ed., Norman, Oklahoma, USA, 1990.
2. Campbell, J. M., “Gas Conditioning and Processing, Vol. 2, the Equipment Modules, 8^th Ed., Campbell Petroleum Series, Norman, Oklahoma, 2001
3. Soave, G., Chem. Eng. Sci., Vol. 27, pp. 1197-1203, 1972.
4. Peng, D. Y., and Robinson, D. B., Ind. Eng. Chem. Fundam., Vol. 15, p. 59, 1976.
5. ProMax 3.2, Bryan Research and Engineering, Inc, Bryan, Texas, 2011.

In the March 2011 tip of the month (TOTM) we studied a constant volume translation of liquid density method presented by Peneloux et al. [3] and demonstrated its application for pure components. Considerable improvements were obtained, specifically for the low temperature range (reduced temperature < 0.8), of saturated specific volume (or liquid density) predicted by Soave-Redlich-Kwong (SRK) [1] and Peng-Robinson (PR) [2], but, the constant volume shift fails near the critical temperature. In the July 2011 TOTM, we reviewed several temperature dependent volume correction methods [3-11]. They improved the accuracy of liquid density prediction considerably near the critical point region.

As the extension of the March and July 2011 Tips of the Month in this TOTM we will demonstrate application and accuracy of some of these methods to predict liquid density of light hydrocarbon mixtures encountered in gas processing. We will compare their accuracy against both experimental data and a few correlations.

Equations of state (EoS) are used in commercial simulation software for predicting phase behavior and thermodynamic properties. The cubic equations of state give relatively accurate results for predicting vapor-liquid equilibria, especially for non-polar or slightly polar systems. These equations can be used to accurately predict vapor densities, enthalpy and entropy. These advantages encourage the researchers to augment EoS ability more than before, especially liquid density, although their accuracy for liquid density prediction is generally poor. The popular EoSs such as SRK [1] and PR [2] predict liquid density with an average absolute error of about 8%, much higher than several good density correlations. This large magnitude of error is not acceptable by industry; therefore, they are not used for this purpose. In order to overcome this deficiency, volume translated methods have been developed.

The correlations and volume translated methods [3-11] used in this study are the same as those presented in the July 2011 TOTM. Only those methods, which gave the best results for the cases studied in this TOTM, are described briefly here.  More details about these methods can be found in the corresponding references.

1. Correlations
   The following correlations were used in this study.
   1. COSTALD, 1979: The COSTALD correlation by Hankinson and Thomson [12] requires two parameters: w_SRK, the optimized value of the acentric factor based on the SRK equation of state (EoS) and; V*, the pure component characteristic volume.
   2. RSD, 1972: Spencer and Danner [13] improved the liquid density correlation of Rackett [14]. The improved correlation for saturated liquid density calculation requires only Z_RA, the improved compressibility factor.
   3. NM, 1998: Nasrifar and Moshfeghian [15, 16] presented an equation and a set of mixing rules for predicting the liquid density of pure refrigerants and liquefied natural gas.
2. Volume Translated EoS Methods
   Equation 1 is the basic volume translated (shift) method proposed by Peneloux et al. [3] and used in this study. Equations 2 and 3 utilize the Kay’s rule to estimate mixture molecular weight (MW) and volume shift correction parameter (c).

In the above equations, is the corrected liquid specific volume, is the liquid specific volume calculated by the EoS, MW is the molecular weight, ρ^L is the liquid density, and the correction term or volume shift factor “c” is determined from experimentally measured liquid density. The volume shift factor is normally regressed against several experimental data points. The following methods were used to determine c for the mixtures.

1. Temperature Independent, PRF, 1982: Peneloux et al. [3] proposed the following expression to estimate the constant temperature volume shift correction for each component “i” in the mixture. In the absence of experimentally regressed value, it can be estimated as follows:
   where Z_RA, is the Rackett [15] parameter, R is the gas constant, and T_C and P_C are the critical temperature and pressure, respectively.
2. Temperature Dependent, AG, 2001: Ahlers and Gmehling [6] temperature dependent correction factor, c, is calculated as follows:(5)(6)

In the above equations, Tr is the reduced temperature, ω is acentric factor, T_C, P_C and Z_C are component i critical temperature, pressure and compressibility factor, respectively. The correction term, c, from the above methods is substituted into equation 1 to calculate the corrected density.

 

Results and Discussion:

We applied the preceding methods to several natural gas mixtures shown in Table 1. The experimentally measured temperature, pressure, composition and relative liquid density for each mixture are shown in this table [17]. These mixtures and corresponding conditions represent those encountered in the cryogenic processes. As an example, the condition and phase envelope for mixture number 1 of Table 1 are presented in Figure 1.

Table 1. Experimentally measured composition, temperature, pressure and relative density for the mixtures studied [17]

Table 2 presents the summary of the error analysis for the liquid density prediction by different methods for the natural gas mixtures shown in Table 1. As can be seen in Table 2, both SRK and PR EoSs give poor results; however, considerable improvements are observed by applying temperature-independent volume translated SRK (SRK-PRF) and temperature-dependent volume translated PR (PR-AG). The volume translated results for these mixtures are much closer to the results obtained by the three correlations of RSD, NM, and COSTALD.

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

Figure 1. Phase envelope and liquid density condition for mixture 1

Table 2: Comparison of acuracy of EoS, volume translated EoS, and correlations for predicting liquid density of mixtures prsented in Table 1

 

Reference:

1. Soave, G., Chem. Eng. Sci., Vol. 27, pp. 1197-1203, 1972.
2. Peng, D. Y., and Robinson, D. B., Ind. Eng. Chem. Fundam., Vol. 15, p. 59, 1976.
3. Peneloux, A. E., Rauzy, E., and Freze, R., Fluid Phase Equilib., Vol. 8, pp. 7-23, 1982.
4. Tsai, J. and Y.P. Chen, J. of Fluid Phase Equilibria, Vol. 145, pp. 193-215, 1998.
5. Ahlers, J. and J. Gmehling, J. of Fluid Phase Equilibria, Vol. 191, pp. 177-188, 2001.
6. Lin, H. and Y.Y. Duan, J. of Fluid Phase Equilibria, Vol. 233, pp. 194-203, 2005.
7. Ji, W.R. and D.A. Lempe, J. of Fluid Phase Equilibria, Vol. 130, pp. 49-63, 1997.
8. Pfohl, O., J. of Fluid Phase Equilibria, Vol. 163, pp. 157-159, 1999.
9. Magoulas, K. and D. Tassios, J. of Fluid Phase Equilibria, Vol. 56, pp. 119-140-445, 1990.
10. Frey, F., Augustine, C., Ciccolini, R.P., Paap, S., Modell, M., and J. Tester, , J. of Fluid Phase Equilibria, Vol. 260, pp. 316-325, 2007.
11. Frey, F., Modell, M., and J. Tester, J. of Fluid Phase Equilibria, Vol. 279, pp. 56-63, 2009.
12. Hankinson, R. W., Thomson, G. H., AIChE J., Vol. 25, no. 4, pp. 653-663, 1979.
13. Spancer, C. F., and Danner, R. P., J. Chem. Eng. Data, vol. 17, no. 2, pp. 236-241, 1972.
14. Rackett, H. G., J. Chem. Eng. Data, vol. 15, No. 4, pp. 514-517, 1970.
15. Nasrifar, Kh. and Moshfeghian, M., Fluid Phase equilibria Vol. 153, 231-242, 1998.
16. Nasrifar, Kh. and M. Moshfeghian, J. of Fluid Phase Equilibria, Vol. 158-160, pp. 437-445, 1998.
17. Haynes, W.M.J., J. Chem. Thermodyn., Vol 14, pp. 603–612, 1982.

As discussed in the May 2011 Tip Of The Month (TOTM), for hydrate inhibition, the most commonly used equilibrium inhibitors used in the upstream and midstream sectors of the oil and gas business are:  monoethylene or diethylene glycol (MEG or DEG) and methanol.  In general, glycols are more commonly used in systems requiring continuous inhibition. The glycol is typically recovered, regenerated and recirculated.  Methanol is more commonly used in systems that do not require continuous inhibition, i.e. systems only requiring inhibition during cold weather or upset conditions. Methanol is not commonly recovered and reused because of the difficulty of separation of the methanol from water. There are obviously exceptions to this. For example, methanol is used as a continuous inhibitor in a few offshore installations and in a handful of gas processing facilities. Another significant disadvantage of methanol relative to glycol is the high methanol losses to both the liquid hydrocarbon and vapor phase.

In the May TOTM, we reviewed experimental VLE data for methanol-hydrocarbon systems. We also revisited Figure 6.20 of reference [1] for methanol loss to the vapor phase using the experimental vapor liquid equilibrium data reported in the Gas Processors Association Research Report 117 (GPA RR 117) [2].

In this Tip of the Month, we will investigate further the methanol loss to the vapor phase and present a simple correlation to estimate methanol K-values. The K-value is defined as the mole fraction of methanol in vapor phase/mole fraction of methanol in aqueous liquid phase. Since the effect of gas composition is small, the K-values will be expressed only in terms of pressure and temperature. The estimated K-value can be used to calculate the rate of methanol loss to the vapor phase.  The objective of this TOTM is to develop a simple and generalized model for estimation of methanol losses in terms of K-values and present a generalized chart which is less dependent on the weight percent of methanol in the liquid aqueous phase.  A step-by-step calculation procedure using K-values will be presented as well.

Figure 1 (FPS). Effect of methanol concentration on methanol loss at -10 °F.

Figure 1 (SI). Effect of methanol concentration on methanol loss at -23.3 °C.

Methanol Loss in Terms of K-Value:

The polar version of Peng-Robinson [3] equation of state (PR EOS) in ProMax [4] was used to generate the required data in the following sections.

Figure 1 indicates that presenting charts for ratio of vapor to liquid composition in terms lb_m of methanol per MMSCF/(weight % methanol in aqueous phase) or kg of methanol per 10⁶ Sm³/(weight % methanol in aqueous phase) is very sensitive to the methanol concentration in liquid phase. Similar methanol concentration dependencies, as shown in Figure 1, were also observed for other isotherms. An alternative is to use the K-Values for the y-axis. However, we have found that the range of ratio of K-Values at 15 weight % to 25 weight % MeOH is from 0.95 to 1.0. Similarly, the range of ratio of K-Values at 50 weight% to 25 weight % MeOH is from 1.0 to 1.03. These ranges are much smaller than the corresponding ratios of methanol losses. Therefore, in the subsequent charts as well as for modeling we will use K-values as the dependent variables.

The K-Values over 25 weight % methanol solution are presented in Figure 2 as a function of pressure and temperature. It should be noted that Figure 2 can be used for methanol concentration up to 70 weight % in aqueous phase.

As shown in Figure 2, at lower temperatures considerable curvature is observed, which makes modeling rather difficult. However, if the same chart is replotted in terms of pressure times K-value, (P)(K) on the y-axis, as shown in Figure 3, much less curvature is observed.

In order to use Figure 2 to calculate the rate of methanol loss to vapor phase, the following steps should be followed:

1. At specified pressure (P) and temperature (T), read methanol K-Value from Figure 2.
2. Convert weight % of methanol (wt%) in aqueous phase to mole fraction (x) by: (1A)
3. Calculate methanol mole fraction in the gas phase (y) by y = (K)(x) (1B)
4. Calculate mass of methanol in vapor phase

FPS: (MeOH lb_m/MMSCF) = [(y lbmole MeOH)/(Total lbmole of gas)](32 lb_m/lbmole

MeOH) (lbmole of gas/379.5SCF) (10⁶ SCF/MMSCF) = 84 321.5 y ≈ 84 322 y    (1C)

SI: (MeOH kg/10⁶ Sm³) = [(y kmole MeOH)/(Total kmole of gas)](32 kg/kmole MeOH) (kmole of gas/23.64 Sm³) (10⁶) = 1 353 638 y ≈ 1 353 640 y     (1D)

A worked example is shown in Appendix A.

Figure 2 (FPS). Variation of methanol K-Values as a function of pressure and temperature.

Figure 2 (SI). Variation of methanol K-Values as a function of pressure and temperature.

Development of Mathematical Model

An alternative to Figure 2 is a simple and generalized correlation which can estimate the K-values easily. This will be handy when one intends to use spreadsheet calculations to estimate methanol vapor losses. A simple model presented by Wilson [5] for light hydrocarbon mixtures is shown in Equation 2.

(2)

where:

Tc_i critical temperature, °R or K

Pc_i critical pressure, psi, kPa or bar

ω_i acentric factor

P system pressure, psia, kPa or bar

T system temperature, °R or K

This correlation is applicable to low and moderate pressure, up to 500 Psia (3.5 MPa), and the K-values are assumed to be independent of composition.

We propose to use a similar equation in the form of equation 3 to estimate methanol K-value at specified pressure and temperature.

(3)

In the above equation, P^*, T^* and ω^* are the normalized pressure, temperature, and acentric factor; respectively. The same data that were used to generate Figure 2 were also used regress the above equation parameters using a non-linear program and the following expressions were developed.

P^* =  P/35  with P in Psia  (4)

T^* = T/615 with T in °R (5)

ω^* = 2.95-(0.02607)P^*+(8.92828×10^-5)(P^*)²-0.851257/ T^* (6)

We will refer to the above model as the “K-Value Express”. Spreadsheet estimation of methanol vapor losses can be developed by using equation 3 to determine methanol K values, and then applying the calculation method as discussed for the application of Figure 2.

Overall, 156 data points covering temperature range of -10 to 100 °F, and pressure range of 100 to 5000 psia for  25 weight % methanol solution were used. The overall Average Absolute Percent Error (AAPE) for the K-Values was 3.6% with a Maximum Absolute Percent Error (MAPE) for K was 19.7%. The MAPE occurred at -10 °F and 2000 psia where ProMax K was 0.001 and K-Value Express K was 0.0008.

Figure 3 presents the comparison between the methanol K-Values calculated by ProMax (solid lines in Figure 3) and those estimated by K-Value Express (identified by dashed line in Figure 3).

Results and Discussion:

The K-Value Express model with the parameters shown in Equations 4 to 6 were used without any further fitting to predict K-values over 15 and 50 weight % methanol for wide ranges of pressures and temperature. For the case of 15 weight % methanol with 149 data points, the AAPE for K was 5.2% with a MAPE of 14.7%. For the case 50 weight % methanol with 155 data points, the AAPE for K was 3.6% with a MAPE for K was 22.1%.

The maximum average absolute % error occurred at -10 °F and 2000 psia where ProMax K was 0.00098 and K-Value Express K was 0.00077 for the case of 50 weight % methanol.

Figure 4 presents the K-Value Express K-Values vs ProMax K-Values for more than 500 data points over 15, 25, and 50 weight % methanol solution. This figure indicates relatively good agreement between the two methods.

Figure 5 is a revised and extended version of Figure 6.20 in reference [1]. Notice y-axis and x-axis variables are switched.  In this way the dependent variable is on the y-axis and independent variable is on x-axis.

Conclusion:

ProMax was used to reproduce Figure 6.20 in reference 1 and presented here in this work as Figure 5.  This figure covers wider ranges of pressure, temperature, and methanol weight percent (up to 70 weight %). However, we suggest using Figure 2 as a better chart since it is less sensitive to methanol weight % in aqueous phase.  In addition, we developed a simple and generalized K-Value Express model that can be used to estimate methanol K-values for wide ranges of pressure, temperature, and methanol weight %. As shown in Figures 3 and 4, the proposed model is in good agreement with the results obtained from ProMax. The sample calculations in Appendix indicate good agreement between the methanol losses to vapor phase obtained from Figures 2, 3, 5, and the K-Values Express model.

Figure 3 (FPS). Comparison of predicted methanol K-Values by ProMax and the proposed Express K-Value model.

Figure 3 (SI). Comparison of predicted methanol K-Values by ProMax and the proposed Express K-Value model.

Figure 4. Accuracy of the proposed Express K-Value model against ProMax

Figure 5. Variation of methanol loss to vapor phase with pressure and temperature for methanol concentration of 25 weight % in the aqueous phase

Figure 5 (SI). Variation of methanol loss to vapor phase with pressure and temperature for methanol concentration of 25 weight % in the aqueous phase

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. Campbell, J. M. “Gas conditioning and processing, Volume 1: Fundamentals,” John M. Campbell and Company, Norman, Oklahoma, USA, 2001.
2. Ng, H. J., Chen, C. J., and D. B. Robinson, D.B.; RR-117, “The Solubility of Methanol or Glycol in Water – Hydrocarbon Systems,” Gas Processors Association (Mar. 1988).
3. Peng, D. Y., and Robinson, D. B., Ind. Eng. Chem. Fundam., Vol. 15, p. 59, 1976.
4. ProMax 3.1, Bryan Research and Engineering, Inc, Bryan, Texas, 2010.
5. Wilson, G., “A modified Redlich-Kwong equation of state applicable to general physical data calculations,” Paper No15C, 65^th AIChE National meeting, May, (1968).

 

 

 

Appendix A – Sample Calculations

Determine mass of methanol in vapor per MMSCF (10⁶ Sm³) at 1000 psia (6 897 kPa) and 10 °F (-12.2 °C) over a rich solution containing 25 weight % methanol.

Solution: Method 1 (Figure 5)

At 1000 psia (6 897 kPa) and 10 °F (-12.2 °C) using Figure 5;

FPS:

lbm MeOH/MMSCF/Wt%=0.5  or  lb_m MeOH/MMSCF = (0.5) (25) = 12.5

SI:

kg MeOH/10⁶ Sm³/Wt% = 8.2  or  MeOH/10⁶ Sm³ = (8.2) (25) = 205

Solution: Method 2 (Figure 2 or 3)

At 1000 psia (6 897 kPa) and 10 °F (-12.2 °C) using Figure 2; K=0.00095 or

Figure 3; PK=0.95 psia which gives K=0.00095

Convert 25 wt% to mole fraction by Eq 1A:

Calculate y by Eq 1B: y = (K)(x) = (0.00095)(0.1579) = 0.00015

FPS: Using Eq 1C: (MeOH lb_m/MMSCF) = 84 322 y = 84322(0.00015)  =  12.6

SI: Using Eq 1D: (MeOH kg/10⁶ Sm³) = 1 353 640 y = 1 353 640(0.00015) = 203

Solution: Method 3 (Express K-Value Model)

At 1000 psia (6 897 kPa) and 10 °F (-12.2 °C) using Eqs 3 through 6 calculate K.

Using Eq 4: P^* =  P/35 = 1000/35 = 28.57

Using Eq 5: T^* = T/615 = (460+10)/615 = 0.762

Using Equation 6:

ω^* = 2.95-(0.02607)P^*+(8.92828×10^-5)(P^*)²-0.851257/ T^*

ω^* = 2.95-(0.02607)( 28.57)+(8.92828×10^-5)( 28.57)²-0.851257/ 0.762

ω^* = 1.161

Using Eq 3:

Calculate methanol mole fraction in gas phase ( y) by Eq 1B:

y = (K)(x) = (0.00093)(0.1579) = 0.000147

FPS: Using Eq 1C: (MeOH lb_m/MMSCF) = 84 322 y = 84322(0.000147)  =  12.4

SI: Using Eq 1D: (MeOH kg/10⁶ Sm³) = 1 353 640 y = 1 353 640(0.000147)  = 199

Equations of state (EoS) are used in commercial simulation software for predicting phase behavior and thermodynamic properties. The cubic equations of state (EoS) give relatively accurate results for predicting vapor-liquid equilibria, especially for non-polar or slightly polar systems. Furthermore, these equations can be used to accurately predict vapor densities, enthalpy and entropy. These advantages encourage the researchers to augment EoS ability more than before, especially liquid density, although their accuracy for liquid density prediction is generally poor. The popular EoSs such as Soave-Redlich-Kwong (SRK) [1] and Peng-Robinson (PR) [2] predict liquid density with an average absolute error of about 8%, much higher than several good density correlations. This large magnitude of error is not acceptable by industry; therefore, they are not used for this purpose. In order to overcome this deficiency, volume translated methods have been developed.

In the March 2011 tip of the month (TOTM) we studied a constant volume translation of liquid density Method presented by Peneloux et al. [3] and demonstrated its application for hydrocarbons such as pure methane, n-pentane, decane, pentadecane and carbon dioxide. Considerable improvements, specifically for the low temperature range (T r < 0.8), of saturated specific volume (or liquid density) predicted by PR and SRK were obtained. On the other hand, the constant volume shift fails near the critical temperature, because the slope of volume with respect to temperature greatly increases in this region.

Since Peneloux et al. presented their constant volume translation (shift) method in 1982, several temperature dependent volume correction methods [4-11] have been proposed. In this TOTM we will demonstrate application and accuracy of some of these methods to predict liquid density of common hydrocarbons and non-hydrocarbons in gas processing. We will compare their accuracy against both experimental data and a few correlations.

1. CorrelationsThe following correlations were used in this study.
   1. COSTALD, 1979: The COSTALD correlation by Hankinson and Thomson [12] requires two parameters: SRK, the optimized value of the acentric factor based on the SRK equation of state (EoS) and; V*, the pure component characteristic volume.
   2. RSD, 1972: Spencer and Danner [13] improved the liquid density correlation of Rackett [14]. The improved correlation for saturated liquid density calculation requires only ZRA, the improved compressibility factor.
   3. NM, 1998: Nasrifar and Moshfeghian [15, 16] presented an equation and a set of mixing rules for predicting the liquid density of pure refrigerants and liquefied natural gas.
2. Volume Translated EoS Methods The following volume translated (shift) methods were used in this study.
   1. PRF, 1982: Peneloux et al. [3] proposed the following constant volume shift correction. (1)In the above equation, V^L is the corrected liquid specific volume, V^EoS is the liquid specific volume calculated by the EoS, MW is the molecular weight, p^L is the liquid density, and the correction term or volume shift factor “c” is determined from experimentally measured liquid density. The volume shift factor is normally regressed against several data points. In the absence of experimentally regressed value, it can be estimated as follows: (2)where Z_RA, is the Rackett [15] parameter, R is the gas constant, and T_C and P_C are the critical temperature and pressure, respectively.We determined the optimum value of “c” for each compound by the procedure described in
      the March 2011TOTM.
   2. MT, 1990: Magoulas and Tassios [4] temperature dependent correction factor is calculated as follows: (3) (4) (5) (6)We will refer to this method as MT-VTPR.
   3. TC, 1998: Tsai and Chen [5] temperature dependent correction factor is calculated as follows: (7) (8) (9) (10)
   4. AG, 2001: Ahlers and Gmehling [6] temperature dependent correction factor, c, is calculated as follows:
      (12) (13)

      (14)

      (15)

      (16)

      (17)

   5. LD, 2005: Lin and Duan [7] presented a temperature dependent factor, c, as follows: (18) (19) (20) (21)

In the above equations, Tr is the reduced temperature, ω is acentric factor, T_C, P_C and Z_C are critical temperature, pressure and compressibility factor, respectively. The correction term, c, from the above methods is substituted in equation 1 to calculate the corrected density.

Results and Discussion:

A simple MathCad program was written to perform all of the calculations based on the above methods. We applied the preceding methods to several pure compounds shown in Table 1. The reduced temperature (Tr) and number of points (N) for each compound are also shown in Table 1. This Table presents the summary of the error analysis for different methods for the pure compounds. As can be seen in Table 1, these generalized temperature dependent volume shift methods improve the accuracy but yet not as good as the generalized correlation methods shown in the last three column of Table 1.

Table 1. Summary of error analysis for different methods studied

Figures 1 through 4 present graphical comparisons between the predicted and experimental [17] liquid density values of methane, n-pentane, decane and pentadecane; respectively. Similar trends were observed for the other compounds shown in Table. For clarity, only the results for PR EoS, PRF-VTPR (constant volume shift) and LD-VTPR (temperature dependent volume shift) are presented in these figures. A much better accuracy is obtained near the critical region by applying the temperature dependent volume shift.

Figure 1. Comparison of predicted liquid density of CH₄ by PR EoS, volume translated PRF-VTPR and LD-VTPR against experimental data [17]

Figure 2. Comparison of predicted liquid density of C₅H₁₂ by PR EoS, volume translated PRF-VTPR and LD-VTPR against experimental data [17]

Figure 3. Comparison of predicted liquid density of C₁₀H₂₂ by PR EoS, volume translated PRF-VTPR and LD-VTPR against experimental data [17]

Figure 4. Comparison of predicted liquid density of C₁₅H₃₂ by PR EoS, volume translated PRF-VTPR and LD-VTPR against experimental data [17]

In order to show the sensitivity of the LD-VTPR method and the applicability of tuning its parameters, the Z_C value for pentadecane was changed from 0.243, represented by the solid black curve in Figure 4, to 0.231 which is represented by the dashed black curve in Figure 4. The curve for Z_C=0.231 is labeled as LD-VTPR*. This sensitivity is used for practical applications to tune the volume translated model parameters (e.g. Z_C) to match the predicted liquid density with the experimentally measured data.

Table 1 indicates that considerable improvements are obtained by applying temperature dependent volume shift corrections to the liquid specific volume (or liquid density) near the critical point region. However, the accuracy of the COSTALD, RSD and NM correlations are still by far much better than the volume translation applied to these two EoSs. As shown in Figure 4, further improvement of volume shift methods are obtained by tuning the parameters of volume shift methods with experimental measurement.

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.

A free copy of the MathCad Version 14 file showing the calculations steps for methane is available upon request.

By Dr. Mahmood Moshfeghian

Reference:

1. Soave, G., Chem. Eng. Sci., Vol. 27, pp. 1197-1203, 1972.
2. Peng, D. Y., and Robinson, D. B., Ind. Eng. Chem. Fundam., Vol. 15, p. 59, 1976.
3. Peneloux, A. E., Rauzy, E., and Freze, R., Fluid Phase Equilib., Vol. 8, pp. 7-23, 1982.
4. Magoulas, K. and D. Tassios, J. of Fluid Phase Equilibria, Vol. 56, pp. 119-140-445, 1990.
5. Tsai, J. and Y.P. Chen, J. of Fluid Phase Equilibria, Vol. 145, pp. 193-215, 1998.
6. Ahlers, J. and J. Gmehling, J. of Fluid Phase Equilibria, Vol. 191, pp. 177-188, 2001.
7. Lin, H. and Y.Y. Duan, J. of Fluid Phase Equilibria, Vol. 233, pp. 194-203, 2005.
8. Ji, W.R. and D.A. Lempe, J. of Fluid Phase Equilibria, Vol. 130, pp. 49-63, 1997.
9. Pfohl, O., J. of Fluid Phase Equilibria, Vol. 163, pp. 157-159, 1999.
10. Frey, F., Augustine, C., Ciccolini, R.P., Paap, S., Modell, M., and J. Tester, , J. of Fluid Phase Equilibria, Vol. 260, pp. 316-325, 2007.
11. Frey, F., Modell, M., and J. Tester, J. of Fluid Phase Equilibria, Vol. 279, pp. 56-63, 2009.
12. Hankinson, R. W., Thomson, G. H., AIChE J., Vol. 25, no. 4, pp. 653-663, 1979.
13. Spancer, C. F., and Danner, R. P., J. Chem. Eng. Data, vol. 17, no. 2, pp. 236-241, 1972.
14. Rackett, H. G., J. Chem. Eng. Data, vol. 15, No. 4, pp. 514-517, 1970.
15. Nasrifar, Kh. and Moshfeghian, M., Fluid Phase equilibria Vol. 153, 231-242, 1998.
16. Nasrifar, Kh. and M. Moshfeghian, J. of Fluid Phase Equilibria, Vol. 158-160, pp. 437-445, 1998.
17. Vargaftik, N.B., Handbook of Physical Properties of Liquids and Gases (Pure Substances and Mixtures), 2^nd ed., English Translation, Hemisphere Publication, 1975.

BTEX stands for benzene, toluene, ethylbenzene, and xylene, a group of compounds all that also belong to the broader category of Hazardous Air Pollutants (HAPs). Benzene is a known carcinogen, and has also been shown to cause blood disorders and to impact the central nervous system and the reproductive system.  Toluene may affect the reproductive and central nervous systems.  Ethylbenzene and xylene may have respiratory and neurological effects [1]. BTEX is present in natural gas streams and is being picked up in glycol dehydration and amine sweetening units.

In the United States HAP emissions from glycol dehydration units are regulated under 40 CFR, Part 63, Subpart HH.  Glycol dehydration units processing more than 3 MMscfd (0.85 10⁶ Sm³ per day) and having benzene emissions greater than 900 kg/year (1 ton/year) are required to control HAP emissions.

This problem is one which requires careful attention in the design phase. The purpose of this Tip of the Month (TOTM) is to discuss the primary factors affecting the absorption of BTEX components in glycol dehydration systems.

In gas dehydration service, triethylene glycol (TEG) will absorb limited quantities of BTEX from the gas. Based on the data from reference [2], predicted absorption levels for BTEX components vary from 5-10% for benzene to 20-30% for ethylbenzene and xylene. Figure 18.18 in reference [2] shows approximate ab­sorption percentages for BTEX components as a function of TEG circulation rate and contactor temperature at 6895 kPa (1000 psia). Absorption is fa­vored at lower temperatures, higher pressure, increasing TEG concentration and circulation rate.

The bulk of absorbed HAPs will be vented with the water vapor at the top of the regenera­tor. The most common emission mitigation strategies are to:

1) Condense the regenerator overhead vapor in a partial condenser and combust the remaining vapor.  The uncondensed vapors are typically routed to an incinerator or, if a direct-fired reboiler is used, routed to the reboiler fuel gas. The liquid hydrocarbons are collected and disposed of by blending into a crude oil or condensate stream.  The condensed water is typically routed to produced water disposal.

2) Route the regenerator overhead vapors to another process stream in the facility.   This is typically a low pressure stream such as flash vapors from the last stage of a crude or condensate stabilization system.

In this TOTM, we will revisit Figure 18.18 of reference [2] for estimating absorption of BTEX in the glycol dehydration systems using the experimental vapor-liquid equilibrium data reported in the Gas Processors Association Research Report 131 (GPA RR 131) [3]. The objective of this TOTM is to reproduce similar diagrams covering wider ranges of pressure and temperature. First we demonstrate the accuracy of ProMax [4] and the Peng-Robinson [5] equation of state (PR EOS) of the same software to generate the required data. Finally, for ease of use the generated results are presented graphically.

Verification of Thermodynamic Model:

A series of flash calculations for the reported experimentally measured pressures, temperatures and synthetic feed gas compositions were performed. The mixtures consisted of methane, benzene, toluene, ethylbenzene, o-xylene, TEG and water. The pressure ranged from 20 to 1000 psia (138 to 6895kPa) and temperature ranged from 77 to 400°F (25 to 204°C). These ranges cover the normal operating conditions of contactor, flash tank, and regenerator in a TEG dehydration plant. The calculated liquid (x) and vapor (y) phase compositions for the four BTEX components are compared with the corresponding experimental values and presented in Figure 1.

Figure 1. Comparison of calculated BTEX mole fractions in the liquid and vapor phases  by ProMax with the experimental values reported in GPA RR 131.

Results and Discussion:

For the purpose of this study, a contactor column with three theoretical stages and with the feed composition shown in Table 1 was simulated. The concentration of the lean TEG stream was 99.0 weight % TEG, and it was assumed the TEG temperature was 5°F (2.8°C) warmer than the feed gas. The feed gas was saturated with water at feed conditions. For each contactor pressure and temperature, the lean TEG circulation ratio was varied from 1 to 7 US gallon of TEG/lb_m of water removed (8.3 to 58.4 liters of TEG/kg of water removed).

Three temperatures and three pressures, covering typical contactor operation ranges were studied. Figures 2 to 5 present the results of simulations using ProMax. Absorption of BTEX components is plotted as a function of temperature, pressure and glycol circulation rate.

Table 1. Dry-basis composition of feed gas

Figure 2. Absorption of benzene as a function of temperature, pressure, and circulation ratio

In Figure 2, benzene absorption is plotted as a function of circulation ratio (liquid volume rate per gas standard volume rate) for two temperatures (77 and 122 °F or 25 and 50 °C) and two pressures (500 and 1000 psia or 3447 and 6895 kPa). Absorption increases with decreasing temperature and increasing circulation ratio. The effect of pressure on absorption is small but is more pronounced at 500 psia than at 1000 psia.  The likely reason for this is that at the lower pressure, the water content of the feed gas is higher and the heat of absorption effect increases the gas outlet temperature which, in turn, decreases the solubility of benzene in the TEG. This effect will be not as significant at higher pressures.

In TEG dehydration process, the common unit of circulation ratio is in gallons of TEG per pound of water absorbed (liters of TEG per kilogram of water absorbed). In Figures 3, 4, and 5 the circulation units on the x-axis were changed to these units.

Figures 3 to 5 can be used to estimate the absorption of BTEX components in a glycol dehydration system for a given pressure, temperature and circulation ratio.

Experimental solubility data for BTEX components in TEG at pressures greater than 1000 psia (6895 kPa) are not available in open literature. Figure 5, which presents BTEX absorption at 1500 psia (10344 kPa) has not been validated with experimental data. In addition, 1500 psia (10344 kPa) is above the cricondenbar of the feed gas used in this study and hence falls in the dense phase region. The solubility behavior of dilute vapor components in solvents such as TEG can be significantly different in the dense phase; therefore, caution should be taken in extrapolating these correlations above 1000 psia (6895 kPa).

Figure 3A. Absorption of benzene and toluene in TEG at 500 psia (3447 kPa)

Figure 3B. Absorption of ethylbenzene and o-xylene in TEG at 500 psia (3447 kPa)

Figure 4A. Absorption of benzene and toluene in TEG at 1000 psia (6895 kPa)

Figure 4B. Absorption of ethylbenzene and o-xylene in TEG at 1000 psia (6895 kPa)

Figure 5A. Absorption of benzene and toluene in TEG at 1500 psia (10342 kPa)

Figure 5B. Absorption of ethylbenzene and o-xylene in TEG at 1500 psia (10342 kPa)

Figure 6 shows the effect of pressure on the absorption of each BTEX component at 95°F (35°C) at 0.2 US GPM TEG/MMSCFD of gas (1.6 m³/h TEG/10⁶ Sm³/d of gas). Be reminded that high this work has not been experimentally validated at pressures above 1000 psia (6895 kPa).

 

Comparison with the GRI-GLYCalc Software:

GRI-GLYCalc [6] is a relatively simple and easy-to-use software package that is widely used by operators for the estimation of BTEX emissions from glycol units.  It is accepted by most state regulatory authorities.  Table 2 shows the ProMax results in this work compared to GLYCalc for each BTEX component at 3 different operating conditions.

 

Conclusions:

As shown in Figure 1, PR EOS can be used to estimate VLE of BTEX compounds in glycol systems.

In reviewing Figures 2 to 5, one can conclude that the absorption of the BTEX components decreases as:

1. Temperature increases
2. Circulation ratio decreases

For pressures between 500 (3450 kPa) and 1000 psia (6895 kPa), the effect of pressure on BTEX absorption is not large.

From operational point of view, minimizing circulation ratio is the most effective way of decreasing the absorption of BTEX components. This also minimizes reboiler duty and the size of the regeneration skid.  Lower TEG circulation rates require more theoretical stages in the contactor to meet outlet water content specifications, but the additional cost of a taller contactor is often offset by savings in the regeneration package.  Care should be taken that the glycol circulation rate is sufficient to ensure adequate liquid distribution over the packing.  Packing vendors can provide minimum circulation guidelines.

Finally, it should be noted that in the operation of a glycol dehydration unit, the desired outcome is to meet the water content specification for the outlet gas, e.g. 7 lbs H₂O/MMSCF (111 kg/10⁶ Sm³).   When using the graphs in this TOTM, different operating points (T, P and circ ratio) will produce different outlet water contents. Make sure that the operating points you are using to estimate BTEX absorption are can also meet the water specification.

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 Mahmood Moshfeghian and Robert A Hubbard

 

 

Figure 6. Impact of pressure on BTEX absorption at 95 F (35 C) and 0.2 US GPM TEG/MMSCFD of gas (1.6 m³/h TEG/10⁶ Sm³/d of gas)

Table 2. Comparison between GRI-GLYCalc and ProMax BTEX absorption at

1000 psia (6,895 kPa), 99.0 weight % lean TEG, and 3 theoretical trays

* gallons TEG/lb_m of water removed (liters TEG/kg of water removed)

 

Reference:

1. http://www.earthworksaction.org/BTEX.cfm, 2011.
2. Campbell, J. M. “Gas conditioning and processing, Volume 2: The Equipment Modules,” John M. Campbell and Company, Norman, Oklahoma, USA, 2001.
3. Ng, H. J., Chen, C. J., and Robinson, D.B.: RR-131, “The Solubility of Selected Aromatic Hydrocarbons in Triethylene Glycol,” Gas Processors Association (Dec. 1991).
4. ProMax 3.2, Bryan Research and Engineering, Inc, Bryan, Texas, 2011.
5. Peng, D. Y., and Robinson, D. B., Ind. Eng. Chem. Fundam., Vol. 15, p. 59, 1976.
6. GRI-GLYCalc 4.0, Gas Research Institute, Des Planes, Illinois, 2000

 

Three methods of preventing hydrate formation in pipelines and processing facilities are commonly used in our industry.  These are:

1)      Maintain the T & P of the system outside of the hydrate formation region.

2)      Dehydrate the gas to remove the water.

3)      Inhibit hydrate formation with chemical inhibitors.

Option 3 is commonly used when it is impractical or uneconomic to install dehydration facilities, typically glycol dehydration.

There are a number of chemicals used to inhibit hydrate formation, but generally fall into one of two types: equilibrium (sometimes called thermodynamic) inhibitors or kinetic inhibitors.  Equilibrium inhibitors lower the equilibrium hydrate formation point and include polar chemicals such as alcohols, glycols and salts.  Kinetic inhibitors (often referred to as Low Dosage Hydrate Inhibitors, LDHIs) do not change the equilibrium hydrate formation condition but instead modify the rate at which hydrates form or the ability of hydrate crystals to agglomerate into a plug that could block flow.  These will not be discussed in this article and more detail can be found in June and July 2010 tip of the month.

The most commonly used equilibrium inhibitors used in the upstream and midstream sectors of the oil and gas business are:  monoethylene or diethylene glycol (MEG or DEG) and methanol.  In general, glycols are more commonly used in systems requiring continuous inhibition.  The glycol is typically recovered, regenerated and recirculated.  Methanol is more commonly used in systems that do not require continuous inhibition, i.e. systems only requiring inhibition during cold weather or upset conditions.  In addition, methanol is not commonly recovered and reused. This is due to the difficulty of separation of the methanol from water. There are obviously exceptions to this, methanol is used as a continuous inhibitor in a few offshore installations and in a handful of gas processing facilities. Another significant disadvantage of methanol relative to glycol is the high methanol losses to both the liquid hydrocarbon and vapor phase.

The purpose of this article is to review experimental VLE data for methanol-hydrocarbon systems and to show correlations that may be used to estimate methanol losses to the vapor phase.

The total injection rate required to inhibit hydrate formation is the sum of the inhibitor in the

liquid water (aqueous phase) plus the inhibitor in the vapor phase plus the inhibitor in the liquid hydrocarbon phase, if any.

As described in Chapter 6, Volume of 1 of Reference [1], the Hammerschmidt [2], Nielsen and Bucklin [3], or Maddox et al. [4] equations can be used to estimate weight percent of methanol or glycol in the rich solution (aqueous phase) required to inhibit hydrate formation. The actual inhibitor injection rate to satisfy the aqueous phase inhibitor concentration needed is found by material balance and is a function of the amount of water to be inhibited as well as the lean inhibitor concentration.

Figure 6.20 on page 191 of reference [1] provides reliable estimates of vaporization loss for pressures less than about 4830 kPa (700 psia) and water phase methanol concentrations less than about 40 weight %. At higher pressures methanol losses to the vapor phase may be significantly higher than indicated in Figure 6.20, particularly at high methanol concentrations.

In this Tip Of The Month (TOTM), we will revisit Figure 6.20 of reference [1] for methanol loss to the vapor phase using the experimental vapor liquid equilibrium data reported in the Gas Processors Association Research Report 117 (GPA RR 117) [5]. The objective of this TOTM is to reproduce the same diagram covering wider ranges of pressure, temperature and weight percent of methanol in the aqueous phase. First we demonstrate the accuracy of ProMax [6] and then the polar version of Peng-Robinson [7] equation of state (PR EOS) of the same software to generate the required data. Finally, for ease of use the generated results are presented graphically.

 

Results and Discussion:

GPA RR 117 presents experimental equilibrium phase compositions for systems containing methane and n-heptane, methane and methylcyclohexane (MCH), and methane-toluene in the presence of 35 and 70 weight % methanol solutions. The experimental conditions for these data are shown in Table 1. In order to evaluate the accuracy of the ProMax software for these systems, we predicted the ratio of vapor to liquid composition in terms lb_m of methanol per MMSCF/(weight % methanol in aqueous phase) or kg of methanol per 10⁶ Sm³/(weight % methanol in aqueous phase). The results of this comparison are also shown in Table 1. The same comparison results are also shown graphically in Figures 1 and 2.

Table 1 indicates that the average absolute percent deviation for the 18 cases tested is about 15% with a maximum deviation of 23%. Considering the fact that the experimental data has some inherent error and some scatter in the data, the accuracy of ProMax is reasonably good for determination of methanol loss to the vapor phase.

 

Table 1. Accuracy of ProMax for calculating methanol loss to vapor phase

1. Methane-n-Heptane-Methanol-Water
2. Methane-Methylcyclohexane-Methanol-Water
3. Methane-Toluene-Methanol-Water

1. (lb_m of methanol per MMSCF)/(Weight % methanol in aqueous phase)
2. (kg of methanol per 10⁶ Sm³)/(Weight % methanol in aqueous phase)

 

Figure 1. Comparison of predicted methanol loss to the vapor phase by ProMax with the experimental values  reported in GPA RR 117.

Figure 1 (SI). Comparison of predicted methanol loss to the vapor phase by ProMax with the experimental values reported in GPA RR 117.

Figure 2. Comparison of predicted methanol loss to vapor phase by ProMax with the experimental values reported in GPA RR 117.

Figure 2 (SI). Comparison of predicted methanol loss to vapor phase by ProMax with the experimental values reported in GPA RR 117.

Figure 3 presents the effect of pressure and temperature on the methanol loss to the vapor phase. This diagram is generated for a mixture with total composition of 33.63 mol% methane, 22.42 mol% n-heptane, 24.95 Mol% methanol, and 19 mol % water. A three phase calculation on this mixture produces, at various pressure and temperature, vapor phases containing about 98 mol% methane and aqueous phases containing about 70 weight percent methanol. A similar diagram was generated for another mixture with composition of 29.43 mol% methane, 19.62 mol%, n-heptane, 11.86 Mol% methanol, and 39.09 mol % water. The later mixture produces an aqueous phase containing about 35 weight percent methanol. The calculated corresponding methanol losses were close for the two mixtures.

Figure 3. Variation of methanol loss to vapor phase with pressure and temperature over methanol concentration of up to 70 weight % methanol in the aqueous phase

Figure 3 (SI). Variation of methanol loss to vapor phase with pressure and temperature over methanol concentration of up to 70 weight % methanol in the aqueous phase

 

Conclusion:

As shown in Table 1 and Figures 1 and 2, the limited experimental data of GPA RR 117 indicate that ProMax can be used to estimate the methanol loss to the vapor phase. Therefore, this software was used to reproduce Figure 6.20 in reference 1 and presented here in this work as Figure 3.  This figure covers wider ranges of pressure, temperature, and methanol weight percent (up to 70 weight %). Even though the methanol loss to the vapor phase shown on the x-axis of Figure 3 depends on the gas composition, the effect of composition is small and can be negligible for the planning purposes and facilities calculations.

 

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. Campbell, J. M. “Gas conditioning and processing, Volume 1: Fundamentals,” John M. Campbell and Company, Norman, Oklahoma, USA, 2001.
2. Hammerschmidt, E. G., Ind. Engr. Chem., Vol. 26 (1934), p. 851.
3. Nielsen, R. B. and R. W. Bucklin, “Why not use methanol for hydrate control?,” Hyd. Proc., Vol. 62, No. 4 (Apr. 1983), p. 71
4. Maddox, R.N., M. Moshfeghian, C. H. Tu, A. Shariat, and A. J. Flying “Predicting Hydrate Temperature at High Inhibitor Concentration,” Proceeding of Laurence Reid Gas Conditioning Conference, March 4 – 6, 1991.
5. Ng, H. J., Chen, C. J., and D. B. Robinson, D.B.; RR-117, “The Solubility of Methanol or Glycol in Water – Hydrocarbon Systems,” Gas Processors Association (Mar. 1988).
6. ProMax 3.1, Bryan Research and Engineering, Inc, Bryan, Texas, 2010.
7. Peng, D. Y., and Robinson, D. B., Ind. Eng. Chem. Fundam., Vol. 15, p. 59, 1976.

Liquid density is needed for process simulation and equipment design. For example, accurate predictions of liquid density are needed for calculating the pressure drop in piping/pipeline and vessel sizing. Accurate liquid density is also essential for custody transfer.

In November 2006, December 2006 and January 2007 tips of the month (TOTM), we presented an overview of different methods and tools for predicting liquid densities. The methods for liquid density prediction include but are not limited to the following.

• Generalized Charts

There are several generalized charts for predicting the liquid density of petroleum fluids and hydrocarbons [1]. The charts normally present the relative density of paraffinic hydrocarbon mixtures at their boiling point or bubble point temperature and pressure. These charts apply to mixtures as well as pure components. Alignment points for paraffinic hydrocarbon mixtures and pure components are located according to their molecular weight. The accuracy of these charts is generally within 3 % of the measured values. However, the accuracy is somewhat less for mixtures with molecular weights less than 30 where temperature is low, and where the methane content is high or pseudo-reduced temperatures are above 0.9 [2].

• Correlations

In order to calculate liquid density reliably, several correlations such as: COrresponding STAte Liquid Density (COSTALD), modified Rackett equation by Spencer and Danner (RSD), and Nasrifar-Moshfeghian (NM) have been developed.

COSTALD: The COSTALD correlation by Hankinson and Thomson [3] requires two parameters: w_SRK, the optimized value of the acentric factor based on the SRK equation of state (EoS) and; V*, the pure component characteristic volume.

RSD: Spencer and Danner [4] improved the liquid density correlation of Rackett [5]. The improved correlation for saturated liquid density calculation requires only Z_RA, the improved compressibility factor.

NM: Nasrifar and Moshfeghian [6] presented an equation and a set of mixing rules for predicting the liquid density of pure refrigerants and liquefied natural gas.

• EoS Methods and Volume Translated

The equations of state are used in commercial simulation software for predicting phase behavior and thermodynamic properties. Generally, EoSs need a few parameters (usually two or three) that are normally obtained from critical properties. The cubic equations of state (EoS) give relatively accurate results for predicting vapor-liquid equilibria, especially for non-polar or slightly polar systems. Furthermore, these equations could be used to accurately predict vapor densities, enthalpy and entropy. These advantages encourage the researchers to augment EoS ability more than before, especially liquid density, although their accuracy for liquid density prediction is generally not as good as the correlations listed above. The popular EoSs such as SRK [7] and PR [8] predict liquid density with an average absolute error of about 8%, much higher than the correlations [9]. This large magnitude of error is not acceptable by industry; therefore they are not used for this purpose. In order to overcome this deficiency, a volume translated method has been developed by Peneloux et al. [10].

In this TOTM, we will present the method of volume translation for liquid density prediction and then demonstrate its application for liquid hydrocarbons such as pure methane, n-pentane, decane, pentadecane and carbon dioxide. In a future TOTM we will extend this procedure to multicomponent mixtures.

In order to improve the accuracy of EoSs for predicting liquid density, Peneloux et al. [10] proposed the following correction.

(1)

In the above equation,is the corrected liquid specific volume,  is the liquid specific volume calculated by SRK or PR EoS, MW is the molecular weight, ρ^L is the liquid density, and the correction term or volume shift factor “c” is determined from experimentally measured liquid density. It is normally regressed against several data points. In the absence of experimentally regressed value, it can be estimated as follows:

(2)

where Z_RA, is the Rackett [10] parameter, R is the gas constant and T_C and P_C are the critical temperature and pressure, respectively.

In this work, we determined the value of “c” by minimizing the absolute error between experimentally measured liquid densities [11] and the corresponding predicted values using MathCad software. To achieve this, the following equation was defined.

(3)

where NP is the number of data points, ρ_exp is the experimental liquid density. The value of “c” was determined by minimizing  f(c). In MathCad nomenclature the appropriate command is:

(4)

Results and Discussion:

We applied the preceding procedure to several pure compounds shown in Table 1. The temperature and pressure ranges, and the optimized values of “c” for both PR and SRK EoS as well Z_RA for the RSD correlation are shown in Table 1. Figures 1 and 2 present graphical comparisons between the predicted and experimental liquid density values of pentadecane. Similar trends were observed for the other compounds. Table 2 presents the summary of error analysis for different methods for the pure compounds shown in Table 1.

Table 1. Optimized volume translated parameter, c, for different compounds

Component	Temperature Range, °C	Pressure Range, kPa	NP	SRK            c x 106 m3/mol	SRK (c/b)	PR            c x 106 m3/mol	PR (c/b)	ZRA
Methane	-182 to -90	12 to 3464	55	-0.3506	-0.012	4.059	0.151	0.2891
n-Pentane	-121 to 124	0.0003 to 2548	21	-11.36	-0.113	2.04	0.023	0.2686
Decane	-13 to 344	0.006 to 2097	18	-41.44	-0.197	-13.99	-0.074	0.2522
Pentadecane	12 to 432	0.0002 to 1421	29	-92.89	-0.270	-48.25	-0.156	0.2385
CO2	-56 to 18	531 to 5463	63	-4.534	-0.153	0.955	0.036	0.2719

Figure 1. Comparison of predicted liquid density of C₁₅H₃₂ by PR and volume translated PR (VTPR) against experimental data [11]

Figure 1. Comparison of predicted liquid density of C₁₅H₃₂ by SRK and volume translated SRK (VTSRK) against experimental data [11]

Table 2 indicates that considerable improvements are obtained by applying volume translated correction to liquid specific volume (or liquid density) predicted by PR and SRK. However, the accuracy of the COSTALD, RSD and NM correlations are still by far much better than the volume translation applied to these two EoSs. The accuracy of EoS and its volume translated correction deteriorate as the critical point is approached.

Table 2. Summary of error analysis for different methods studied

Component	Average absolute Error %
	PR	VTPR	SRK	VTSRK	RSD	NM	COSTALD
Methane	9.65	2.13	3.96	3.68	0.12	0.20	0.12
n-Pentane	2.64	2.14	11.03	3.07	0.14	0.15	0.11
Decane	7.62	3.09	18.16	4.43	1.04	0.70	0.95
Pentadecane	14.08	3.63	23.71	4.61	0.20	1.13	0.25
Carbon Dioxide	2.68	2.24	11.25	3.48	0.28	0.12	0.31
Overall Average	7.33	2.65	13.62	3.85	0.36	0.46	0.35

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:

• Campbell, J. M. “Gas conditioning and processing, Volume 1: Fundamentals,” John M. Campbell and Company, Norman, Oklahoma, USA, 2001.
• Engineering Data Book, 12^th Editions, Gas Processors and Suppliers Association Data Book, Tulsa, Oklahoma, 2004.
• Hankinson, R. W., Thomson, G. H., AIChE J., Vol. 25, no. 4, pp. 653-663, 1979.
• Spancer, C. F., and Danner, R. P., J. Chem. Eng. Data, vol. 17, no. 2, pp. 236-241, 1972.
• Rackett, H. G., J. Chem. Eng. Data, vol. 15, No. 4, pp. 514-517, 1970.
• Nasrifar, Kh. and Moshfeghian, M., Fluid Phase equilibria Vol. 153, 231-242, 1998.
• Soave, G., Chem. Eng. Sci., Vol. 27, pp. 1197-1203, 1972.
• Peng, D. Y., and Robinson, D. B., Ind. Eng. Chem. Fundam., Vol. 15, p. 59, 1976.
• Nasrifar, Kh. and M. Moshfeghian, J. of Fluid Phase Equilibria, Vol. 158-160, pp. 437-445, 1998.
• Peneloux, A. E., Rauzy, E., and Freze, R., Fluid Phase Equilib., Vol. 8, pp. 7-23, 1982
• Vargaftik, N.B., Handbook of Physical Properties of Liquids and Gases (Pure Substances and Mixtures), 2^nd ed., English Translation, Hemisphere Publication, 1975.

In a past Tip Of The Month (TOTM), we have shown that one of the first issues to be resolved by a facilities engineer working in a gas plant or gas production facility is where the process is operating with respect to the phase diagram.  A general knowledge, if not a detailed knowledge, will allow the design engineer and the facilities operator to make intelligent decisions that have significant impact on the profitability of a gas production facility.

The best way to prevent hydrate formation (and corrosion) is to keep the pipelines, tubing and equipment dry of liquid water. In this TOTM we will demonstrate how the water dew point and hydrate formation curves are shifted along a conventional phase envelope as natural gas is dehydrated.

Case Study:

In order to demonstrate the phase behavior of natural gases containing water and the impact of water content on the water dew point and hydrate formation temperatures, let’s consider the natural gas shown in Table 1. To generate the diagrams in this TOTM, we used ProMax [1] based on the Peng-Robinson equation of state (PR EOS) [2].

Table 1. Dry gas composition

Component	Mole %
C1	80.0
C2	10.0
C3	4.0
iC4	3.0
nC4	3.0
Sum	100.0

Results and Discussion:

Figure 1 presents the phase envelope, hydrate formation and water dew point curves of this gas with a water content of 0.06 mole percent, equivalent to 28.5 lb_m/MMSCF (456 kg/10⁶ Sm³). Notice that up to a pressure of about 414 psia (2854 kPa), the water dew point curve is slightly to the left of the hydrate formation curve. This indicates that the gas is under-saturated with water at pressures below this point. This also means that it is thermodynamically unstable and will not form a free aqueous phase. All the water is converted to hydrate and this state is referred to as “meta-stable” equilibrium. For more detail on this meta-stable state, see December 2010 TOTM. Similar behavior is demonstrated in Figure 2 for which the water content was reduced to 0.0427 mole percent, equivalent to 20.3 lb_m/MMSCF (324.6 kg/10⁶ Sm³). In this case the water dew point and hydrate formation curves intersect at a higher pressure of 1000 psia (6895 kPa). Below this pressure, the gas is under-saturated and has a meta-stable equilibrium state. Therefore, the water dew point curve is to the left of the hydrate formation curve, but above the intersection pressure it moves to the right of the hydrate formation curve where the water content is above saturation.

Figure 3 presents the superimposition of Figures 1 and 2 having water dew point and hydrate formation curves for two different water contents (0.06 and 0.0427 mole%). Notice the hydrate formation curves for both cases coincide with each other for pressures of 1000 psia (6895 kPa) and higher.

Figure 4 presents the phase envelope along with the water dew point and hydrate formation curves for the same gas as the water content was reduced to 0.0427, 0.03, 0.0148, and 0.00422 mole % corresponding to 20.3, 14.2, 7, 2 lb_m/MMSCF (324.6, 228, 112, 32 kg/10⁶ Sm³), respectively. Notice for all the cases where the gas is under-saturated with water, the water dew point curves are located to the left of the corresponding hydrate formation curves. Under these conditions the equilibrium state is thermodynamically unstable (meta-stable) and will not form a free aqueous phase. However, if the water content is above saturation point, then the water dew point will position to the right of the corresponding hydrate formation curve and free water will form under stable condition.

Conclusions:

We have demonstrated the impact of the water content on the phase behavior of a natural gas. The emphasis was placed on the interaction of the water dew point and hydrate formation curves. It was shown that the relative location of the water dew point and hydrate curves with respect to each other is a strong function of the water content. It was also shown for the cases where water content is above saturation point, the water dew point curve locates to the right of the hydrate curve. Under this condition free water forms and then hydrates may form if conditions are right. This is what is normally expected and shown in text books. However, if the water content is under-saturated, the water dew point curve will be located to the left of the hydrate formation curve and the equilibrium state is thermodynamically unstable (meta-stable) and will not form a free aqueous phase.

As discussed in last month’s TOTM, facility engineers have to determine how this behavior affects their operations.  These phase envelopes suggest that, at low water concentrations, hydrates may form even though free water is not present.  Indeed, this phenomenon has been observed.  At cryogenic conditions, when the water is removed by molecular sieves, the amount of metastable water is so small it should not cause operational issues.

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. ProMax 3.2, Bryan Research and Engineering, Inc, Bryan, Texas, 2010.
2. Peng, D. Y. and Robinson, D. B., I. and E. C. Fund, Vol. 15, p. 59, 1976.

Glycol dehydration is the most common dehydration process used to meet pipeline sales specifications and field requirements (gas lift, fuel, etc.). Triethylene glycol (TEG) is the most common glycol used in absorption systems. Chapter 18, Gas Conditioning and Processing [1] presents the process flow diagram and basics of glycol units. A key parameter in sizing the TEG dehydration unit is the water dew point temperature of dry gas leaving the contactor tower. Once the dry gas water dew point temperature and contactor pressure are specified, water content charts similar to Figure 1 in reference [2] can be used to estimate the water content of lean sweet dry gas. The required lean TEG concentration is thermodynamically related to the dry gas water content which influences the operating (OPEX) and capital (CAPEX) costs. The lower dry gas water content requires a higher lean TEG concentration. This parameter sets the lean TEG concentration entering the top of contactor and the required number of trays (or height of packing) in the contactor tower.

The rich TEG solution is normally regenerated at low pressure and high temperature. Maximum concentrations achievable in an atmospheric regenerator operating at a decomposition temperature of 404 °F (206°C) is 98.7 weight %. The corresponding dry gas water dew point temperature for this lean TEG weight % and contactor temperature of 100°F (38°C) is 18°F  (-8°C).

If the lean glycol concentration required at the absorber to meet the dew point specification is higher than the above maximum concentrations, then some method of further increasing the glycol concentration at the regenerator must be incorporated in the unit. Virtually all of these methods involve lowering the partial pressure of the glycol solution either by pulling a vacuum on the regenerator or by introducing stripping gas into the regenerator.

For water saturated gases, the water dew point temperature is either above or at the hydrate formation temperature. However, if the gas is water under-saturated, the hydrate formation temperature will be higher than water dew point. This means at a given specified water dew point temperature, there are two water content values; the lower value will be at the hydrate formation temperature and the higher value will be at the water dew point temperature. Therefore, the designer has to choose one of these two values. Which value should be chosen? The answer to this question is “It depends”! The lower value of water content means higher lean TEG concentration and consequently higher CAPEX and OPEX.

In this TOTM we will attempt to answer the question by studying a case in which the specified water dew point temperature is below the hydrate formation temperature. For this purpose, we will discuss the water content of natural gas in equilibrium with hydrate and when the condensed water phase is liquid.

The water content chart of Figure 6.1 in reference [2] is based on the assumption that the condensed water phase is a liquid. However, at temperatures below the hydrate temperature of the gas, the “condensed” phase will be a solid (hydrate). The water content of a gas in equilibrium with a hydrate will be lower than equilibrium with a metastable liquid.

Hydrate formation is a time dependent process. The rate at which hydrate crystals form depends upon several factors including gas composition, presence of crystal nucleation sites in the liquid phase, degree of agitation, etc. During this transient “hydrate formation period” the liquid water present is termed “metastable liquid.” Metastable water is liquid water which, at equilibrium, will exist as a hydrate.

Reference [3] presents experimental data showing equilibrium water contents of gases above hydrates. Data from Reference [3] are presented in Figure 6.5 of reference [2] and plotted here as rotated square in Figure 1 at 1000 Psia (6,897 kPa). For comparative purposes, the “metastable” water content of the gas (dashed line) as well as the hydrate formation temperature (solid line) calculated by ProMax [4] using the Peng-Robinson [5] equation of state are also shown. The water content of gases in the hydrate region is a strong function of composition. Figure 1 should not be applied to other gas compositions.

Figure 1. Water content of 94.69 mole % methane and 5.31 mole % propane – gas in equilibrium with hydrate at 1000 Psia (6,897 kPa)

Case Study:

To demonstrate, the effect of water content of a dried gas in equilibrium with hydrate on the required lean TEG concentration, let’s consider the gas mixture presented in Figure 1. This gas enters a contactor tower at 1000 Psia (6,897 kPa) and 100 °F (37.8°C) with a rate of 144 MMSCFD (4.077 10⁶ Sm³/d). At this condition, the water content of the wet gas is 57.6 lb/MMSCF (922.4 kg/10⁶ Sm³). It is desired to dehydrate the gas to a water dew point temperature of 5°F (-15°C) using a TEG dehydration unit.

Results and Discussion:

According to Figure 1, at a temperature of 5°F (-15°C) the water content is 1.2 lb/MMSCF (19.2 kg/10⁶ Sm³) and 1.97 lb/MMSCF (31.5 kg/10⁶ Sm³) in equilibrium with metastable water and hydrate phase, respectively. ProMax was used to simulate this TEG dehydration unit for the case of three theoretical trays in the contactor tower.  The simulation results for these two water content cases are shown in Table 1. This table clearly indicates that the required lean TEG concentrations are not the same and consequently will impact the regeneration requirements of the rich TEG solution. The difference between the lean TEG concentrations will be even more at a lower dry gas water dew point specification.

The simulation results clearly indicate that the choice of water content for a specified dry gas water dew point as the basis for design affects the required lean TEG concentration and consequently the rich TEG solution regeneration requirements.

Table 1. Comparison of simulation results for two different water content specifications

Simulation Results Using ProMax	Based on Water Dew Point Temperature of 5 °F (-15°C)	Based on Hydrate Formation Temperature of 5 °F (-15°C)
Water Dew Point Temperature , °F (°C)	5.0 (-15.0)	-6.2 (-21.2)
Hydrate Formation Temperature, °F (°C)	14.7 (-9.6)	5.0 (-15.0)
Water Content, lb/MMSCF (kg/106 Sm3)	1.97 (31.5)	1.20 (19.2)
Gallon/lb of Water Removed (liter/kg of Water Removed)	3.95 (32.9)	3.90 (32.4)
Lean TEG Weight %	99.45	99.72

Conclusions:

When designing dehydration systems, particularly TEG systems to meet extremely low water dew point specifications, it is necessary to determine the water content of the dried gas in equilibrium with a hydrate using a correlation like that presented in Figure 1. If a metastable correlation is used, one will overestimate the saturated water content of the gas at the dew point specification. This, in turn, may result in a dehydration design which is unable to meet the required water removal. Where experimental data is unavailable, utilization of an EOS-based correlation which has been tuned to empirical data can provide an estimate of water content in equilibrium with hydrates.

To meet pipeline sales specifications, it is normally acceptable to use the water content in equilibrium with the metastable phase (the dashed line in Figure 1) because the difference in the water contents is not that high. However, for extremely low water dew point specifications where there is a cryogenic process downstream, it is recommended to use the water content in equilibrium with hydrate (the solid line in Figure 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. 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
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. Song, K.Y. and Kobayashi, R, “Measurement & Interpretation of the Water Content of a Methane-5.31 Mol% Propane Mixture in the Gaseous State in Equilibrium With Hydrate,” Research Report RR-50, Gas Processors Association, Tula, Oklahoma, 1982
4. ProMax 3.1, Bryan Research and Engineering, Inc, Bryan, Texas, 2010.
5. Peng, D. Y. and Robinson, D. B., I. and E. C. Fund, Vol. 15, p. 59, 1976.

Hydrogen blistering is a type of hydrogen-induced failure produced when hydrogen atoms enter low-strength steels that have macroscopic defects, such as laminations. The defects in the steel (void spaces) provide places for hydrogen atoms to combine, forming gaseous molecular hydrogen (H₂) that can build enough pressure to produce blistering. Hydrogen blistering is a problem mainly in sour environments. It does not cause a brittle failure, but it can produce rupture or leakage [1]. Description and mechanisms of hydrogen blistering can be found in literature [2]. Hydrogen sulfide concentration, temperature and thickness of material affect hydrogen blistering.

In this TOTM we will consider the quantitative effect of temperature and hydrogen sulfide mole fraction causing hydrogen damage in the fractionation columns of an operating natural gas liquid (NGL) Plant [3]. The fractionation unit was designed to process a broad-cut of NGL, which is an off-product from crude oil production units and produces essentially propane, butane, and natural gasoline. The feed to the process is introduced into the fractionation unit where propane, butane and gasoline are separated by three distillation columns. In the first column, which is a deethanizer, ethane and lighter compounds are separated from the feed stream. In the second column, a depropanizer, propane is fractionated and sent to the amine treater for further processing to meet market specifications. The bottoms of the depropanizer are fed to the third column, a debutanizer, in which butane is distilled over and sent to a Merox unit for further treating. The bottoms of the debutanizer column, essentially gasoline, are also sent to Merox for treating. More information on NGL production technologies can be found in reference [4].

During the overhaul of this NGL Plant, the inspection team found that the deethanizer-reflux-accumulator had been damaged due to severe hydrogen blistering in the shell and bottom plate, and the vessel was rejected. Four years later, during an inspection of the Plant, the deethanizer rectifying section and depropanizer rectifying section were found to have been severely damaged by hydrogen blistering.

In order to study the effects of hydrogen sulfide and local temperature quantitatively and more closely, the three distillation columns in a fractionation unit were simulated. In this simulation, which could assist one to thoroughly understand the causes of hydrogen attack, the values of temperature and hydrogen sulfide mole fraction along each column were determined by performing tray-by-tray calculations.

 

Case Study:

The operating NGL Plant consists of fractionation, treating, drying, refrigeration, utility, storage and loading facilities to process approximately 57,700 barrels (9172 m³) of broad-cut NGL per day. The charge to this plant is essentially Natural Gasoline Liquid which is condensed out of oil-field gas and off-product from several crude oil production units. The broad-cut is processed to produce propane, butane and light gasoline (Pentane Plus Product). The feed stream also contains some impurities such as hydrogen sulfide, carbon dioxide and mercaptan, which are removed by treating the products after fractionation.

Since hydrogen blistering occurred only in the fractionation unit, a brief description of this unit is given in the following section [3].

A schematic flow diagram for this unit is given in Figure 1. The 40-tray deethanizer tower receives raw feed from NGL recovery plants, fractionates out ethane and lighter products and delivers essentially ethane-free NGL to the depropanizer column. The feed to the deethanizer is introduced between the 27th and 28th trays at 135°F and 362 Psig (57.2 °C and 2497 kPag). Bottoms product from the deethanizer is charged to tray 22 of a 45-tray depropanizer column. The distillate product, which is essentially propane, is sent to the amine treater unit for further processing. The depropanizer bottoms are charged to tray 20 of the 40-tray debutanizer column. The debutanizer distillate product, which comprises the net butane product, is sent to the Merox plant for treating. The debutanizer bottom (pentane and heavier products essentially free of butane) is also sent to the Merox plant for further processing.

Figure 1. Flow diagram of fractionation unit

 

The following information was specified for simulation of the fractionation unit:

(i)   Flow diagram as shown in Figure 1

(ii)  Feed stream condition and composition as shown in Table 1

(iii) Column specification as presented in Table 2

Other specifications such as a desired percentage recovery of a component in any product stream, could have been used instead of the reflux ratio or bottoms product  ratio. In the course of simulation, tray by tray calculations were performed to calculate temperature, pressure, vapor and liquid compositions, and vapor and liquid traffics for each tray in each column. In addition, distillate and bottoms rates, temperature, pressure, composition, reboiler and condenser duties were also calculated, as were height and diameter of the columns.

To perform the simulation, Vapor Liquid Equilibrium K-values, liquid and vapor enthalpies were computed by the Peng-Robinson equation of state [5].  In the tray-by-tray calculation it was assumed that the trays performed ideally (100 % efficiency). The simulation was carried out by UniSim simulation software [6]

 

Table 1. Feed stream composition and specification

Component	Mole %
CO2	1.167
H2S	0.325
Methane	5.625
Ethane	15.724
Propane	28.190
i-Butane	6.724
n-Butane	17.812
i-Pentane	5.812
n-Pentane	6.846
n-Hexane	5.998
C7+	5.777
T, °F (°C)	135.0 (57.2)
P, Psig (kPag)	362.0 (2497)
Rate, lbmole/hr (kmole/h)	8619 (3909)

 

 

Table 2. Fractionation towers specifications

Column	Pressure, Psig (kPag)			No of Trays	Feed Tray from Bottom	Reflux Ratio, L/F	Bottoms Ratio, B/F	Condenser Type
	Feed	Condenser	Reboiler
Deethanizer	362 (2497)	347(2393)	360 (2483)	40	27	0.4438	0.7749	Partial
Depropnizer	300(2069)	290(2000)	300 (2069)	45	22	1.0709	0.6415	Total
Debutanizer	95 (655)	85(586)	95 (655)	40	20	1.0082	0.4889	Total

Results and Discussion:

Performing a simulation, a great deal of information is produced. However, only information of interest in regard to hydrogen blistering is presented here. To test the validity of the simulation results, composition and condition of the key process streams are compared with those supplied by the designer of the plant [7] and presented in Table 3. In most cases the results compare favorably. In addition, condenser and reboiler duties for each column are compared with the original design values in Table 4. This comparison of the two sets of results shows a maximum deviation of –13.3% for the depropanizer reboiler. With the exception of the deethanizer boiler, all of the design heat exchange duties are higher than those obtained in this simulation, which is, of course, a normal safeguard in plant design.

Table 3 shows hydrogen sulfide is fractionated in the first two columns and does not reach the debutanizer column. Since hydrogen blistering occurred in the first two columns, only these results were examined closely. To study the variation of hydrogen sulfide composition (in both liquid and vapor phases) along each column, its composition is plotted as a function of tray number. This is shown in Figure 2 for the deethanizer and Figure 3 for the depropanizer.

 

Table 3. Comparison of simulation results and design data for process streams leaving fractionation towers

Component	Stream 5		Stream 13		Stream 21		Stream 25
	Simulation	Design	Simulation	Design	Simulation	Design	Simulation	Design
CO2	5.183	5.184	0.000	0.000	0.000	0.000	0.000	0.000
H2S	1.213	1.149	0.185	0.208	0.000	0.000	0.000	0.000
Methane	24.983	24.991	0.000	0.000	0.000	0.000	0.000	0.000
Ethane	65.138	66.171	3.812	2.994	0.000	0.000	0.000	0.000
Propane	3.483	2.505	92.460	95.808	6.779	3.987	0.000	0.000
i-Butane	0.000	0.000	3.244	0.890	22.918	25.484	0.002	0.011
n-Butane	0.000	0.000	0.298	0.100	69.276	69.542	0.529	0.477
i-Pentane	0.000	0.000	0.000	0.000	0.990	0.923	22.880	22.951
n-Pentane	0.000	0.000	0.000	0.000	0.037	0.064	28.133	28.105
n-Hexane	0.000	0.000	0.000	0.000	0.000	0.000	24.682	24.682
C7+	0.000	0.000	0.000	0.000	0.000	0.000	23.773	23.774
Total	100.0	100.0	100.0	100.0	100.0	100.0	100.0	100.0
T, F	16.4	20.0	134.5	141.0	131.2	144.0	264.6	273.0
T, C	-8.7	-6.7	56.9	60.6	55.1	62.2	129.2	133.9
P, psig	347		290		85		95
P, kPa(g)	2393		2000		586		655
Rate, lbmole/hr	1940.4	1939.8	2394.1	2394.1	2189.6	2189.6	2094.5	2094.5
Rate, kmole/h	880.2	879.9	1086.0	1085.9	993.2	993.2	950.1	950.0

 

Similarly, in Figure 4, the temperature variation along these two columns is plotted as a function of tray number, and it can be seen that the temperature profiles decrease smoothly from bottom to top except in the feed zone, which is to be expected in a distillation column with no side draw or inter-stage reboiler/cooler.

Figure 2 indicates that the maximum mole fraction of hydrogen sulfide occurred on tray 11 in the stripping section of the deethanizer while hydrogen blistering occurred in the rectifying section. Therefore, other factors such as temperature must be influencing the hydrogen damage. In the stripping section where no hydrogen blistering occurred, the temperature was higher than in the rectifying section where hydrogen blistering was detected. Another region where hydrogen blistering was found is the top part of the depropanizer rectifying section. In this section of the column, the hydrogen sulfide mole fraction is almost the same as in the stripping section of the deethanizer; however the temperatures for these two sections are not the same. The temperature range for the deethanizer stripping section is 142° to 240°F (61.1 to 115.6°C), and for the troubled region of the depropanizer, it is 142° to 134° F (61.1 to 56.6°C), trays 44, 45 and the condenser. Again, it can be seen how temperature influences the hydrogen blistering damage process. In this case, the hydrogen blistering was occurring at lower temperatures. Simulation results also indicate that carbon dioxide does not reach the depropanizer and debutanizer.

 

Conclusions:

Based on the simulation results and preceding discussion, the following conclusions can be made:

1-   Hydrogen blistering can occur where hydrogen sulfide is present. In the case studied a mole fraction of as low as 0.002 for hydrogen sulfide caused hydrogen damage.

2-   With the presence of hydrogen sulfide, temperature is the important factor promoting hydrogen blistering. In the case studied a temperature of less than 142°F (61.1°C) caused hydrogen damage. Higher temperature drives hydrogen out of the wall to atmosphere.

 

There are probably other factors governing hydrogen damage such as microstructure of materials, thickness of material, presence of CO₂, etc. Even though the simulation was performed based on a dry feed, the actual feed to the plant contained some water.

The above results are consistent and the same as those reported by the author in 1985 [3]. In the original work, the simulation was carried out by a computer package named Process Analysis System (PAS) developed by Erbar and Maddox [8]. At that time, the computations were made on an IBM 370 main frame at Shiraz University Computing Center. The Soave-Redlich-Kwong [9] equation of state was used in the original work.

 

A heat exchanger failure at the Tesoro Anacortes refinery was determined to have experienced a form of hydrogen blistering. That failure led to the deaths of seven workers and the refinery was shut down for over six months to repair the damage.  It was determined that the root cause of the failure was hydrogen blistering in the steel of the heat exchanger which resulted in rupture.  These types of hydrogen attacks can be discovered during scheduled inspections.  If there is a concern that conditions are conducive to hydrogen blistering, one can use a hydrogen patch probe to measure hydrogen activity within metals.  If hydrogen activity is found in the metal, then additional testing can be completed to determine if any internal cracks have developed.

 

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. http://www.glossary.oilfield.slb.com/Display.cfm?Term=hydrogen%20blistering
2. Mostert, R., and Sharp, W.R., “Low Temperature Hydrogen Damage Assessment in the Gas and Refining Industries,” 3^rd Middle East Nondestructive Testing Conference & Exhibition – Bahrain, Manama, 27-30 Nov 2005.
3. Moshfeghian, M., “Hydrogen damage (Blistering) case study: Mahshahr NGL Plant”, Iranian J. of Science & Technology, Vol 11, No.1, 1985.
4. 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
5. Peng, D. Y. and Robinson, D. B., I. and E. C. Fund, Vol. 15, p. 59, 1976.
6. UniSim Design R390.1, Honeywell International, Inc., Calgary, Canada, 2010.
7. Parsons, R. M., NGL Fractionation Facilities, Operation Manual Bandar Mahshahr, The Ralph M. Parsons Company U. K. Ltd.
8. Erbar, J. H., and Maddox, R. N., Process Analysis System, Documentation, Oklahoma State University, Stillwater OK., 1978.
9. Soave, G., Chem. Eng. Sci. Vol. 27, No. 6, p. 1197, 1972.