276426 Calculating Dew Points for Natural Gas Containing Water and/or Selected Production Chemicals

Wednesday, October 31, 2012: 8:48 AM
322 (Convention Center )
Eirini K. Karakatsani, Chemical and Biochemical Engineering(CERE), Technical University of Denmark, Kgs. Lyngby, Denmark and Georgios Kontogeorgis, Chemical and Biochemical Engineering (CERE), Technical University of Denmark, Kgs. Lyngby, Denmark

Calculating Dew Points for Natural Gas containing Water and/or selected Production Chemicals

Eirini K.Karakatsani1 and Georgios M. Kontogeorgis1

1Technical University of Denmark, Department of Chemical and Biochemical Engineering, Center for Energy Resources Engineering (CERE)

Corresponding author e-mail: eirka@kt.dtu.dk

1.    Introduction

The water content of natural gas (NG) often poses problems during the production, transportation and distribution of the gas. Small quantities of undesired dissolved water may condense leading to the formation of condensed water, hydrates and/or ice.  Such condensed phases may result in corrosion, two-phase flow problems, safety hazards and flow assurance issues, slugging of the flow lines, valves and instrumentation resulting in reduced capacity and shutdowns, and reduction of the oil recovery efficiency because of reduction of the reservoir permeability [1]. Accurate thermodynamic models able to calculate the water vapor concentration in equilibrium with hydrate, ice and water in natural gas at pipeline operating conditions (253-323K and up to 250bar) are necessary both because experimental data are limited and difficult to obtain, and because the limits of dehydration techniques (physical adsorption and condensation) need to be defined. The latter techniques use chemicals (eg. glycols and alcohols) which also condensate, adding one more level of challenge when it comes to design of pipelines and process equipment and thermodynamic modeling of formed mixtures.

Aim of this work is to evaluate and compare the performance of different carefully selected thermodynamic models vis-à-vis their capacity to calculate dew points of natural gas mixtures with and without chemicals. A main focus is put on the further development of the CPA (Cubic-Plus-Association) equation of state (EoS) [2] for the applications under consideration and its comparison to GERG-water  calculation method, an ISO-standard model specifically designed to correlate water content and dew points of natural gas [3].

2.    Results and discussion

In this work appropriate ice and hydrate models have been combined with the predictive CPA model for calculating the equilibrium conditions of all possible fluid and condensed phases in a natural gas mixture.

More specifically, the water fugacity in the ice phase at the desired pressure of the system P is given by the following equation:

which corrects the saturation fugacity at the same temperature by the Poynting factor, while for the hydrate phase modeling the well established statistical model proposed by van der Waals and Platteeuw [4] (vdW-P) was used together with the simplified approach suggested by Parrish and Prausnitz [5] for the Langmuir constants. According to vdW-P theory the chemical potential of the hydrate phase is given by the expression: 

where R is the universal gas constant, νi  is the number of type i cavities per water molecule (which are:  ν1 = 1/23 and ν2 = 3/23 for structure I hydrate and ν1 = 2/17 and ν2 = 1/17 for type II hydrates) and the summation is over all cavity types (both 1 and 2). Finally, the occupancy of cavity m by a component i, Θmi, is calculated as follow:

Here fk is the fugacity of a component k in the equilibrium vapor phase obtained from an equation of state, the summation is over all hydrate-forming components while Cmi are the Langmuir constants. Following the simplified approach, suggested by Parrish and Prausnitz: 

                                                                                                                  

where and are fitted parameters.

In this work new and  parameters were derived for four main natural gas components (CH4 , C2H6, C3H8, CO2) using the most recent available phase equilibrium experimental data (both Hydrate-Ice-Vapour and Hydrate-Liquie-Vapour data), only single hydrate data where possible, the most recent CPA parameters for NG components and checking the internal consistency of experimental data before using them (f.ex. Bakker pointed inconsistencies between the data sets previous used for methane parameters estimation [6]). The new parameters (Table 1) have been tested under many different conditions (low temperatures, hydrate structure transitions, HVE, HLE,..), where HLE and structural transitions have been found to be the biggest challenges for the model (see f.ex. Figure 1 for HLE calculations where results of GERG-water, SRK, PR and VPT EoS are also included). It is obvious that the CPA EoS performs qualitatively better both with and without binary interaction parameters. It also seems that different thermodynamic models' results scatter more at high pressures and low temperatures (HPLT), where hydrates usually form (see f.ex. Figure 2).     

Small cavity

Large cavity

Component

Structure

Ami x 103 (K/bar)

Bmi (K)

Ami x 103 (K/bar)

Bmi (K)

CH4

I

9.99997

1969.77

316.434

2048.66

II

4.05

2637

295.2

900

C2H6

I

0

0

57.0184

3040.91

II

0

0

9.54095

4000

C3H8

II

0

0

79.6303

3912.8

CO2

I

0

0

148.949

2547.53

II

0

0

136.917

3500

Table 1. Optimized values of Ami and Bmi for calculating the Langmuir constants.

Figure 1 (left). Water content of a binary hydrocarbon liquid mixture (0.646 C2H6- 0.354 C3H8 mole fraction) mixture in equilibrium with hydrate.

Figure 2 (right).  Water content of a binary hydrocarbon gas mixture (0.9469 CH4- 0.0531 C3H8 mole fraction) mixture in equilibrium with hydrate.

When inhibitors are included in the NG mixture, results produced by different thermodynamic models tend to be more similar to each other at low inhibitor concentrations, while the scatter is more pronounced at high concentrations, which are often the case nowadays as exploration and production activities move into colder and deeper regions (see Figure 3). About Figure 3, it should be noted that the phase behavior of NG mixtures is qualitatively similar to that of pure methane. Based on the results of this Figure, CPA is found in better quantitative agreement with experimental data than any other model.  

Figure 3 (left). Methane hydrate formation in the presence of triethylene glycol (TEG) as inhibitor.

Figure 4 (right).  Water content of CO2 in equilibrium with hydrates at 137.9bar and different temperatures.

Recently, the accuracy and reliability of different experimental measurements for the highly asymmetric CO2-H2O system at HPLT conditions have been questioned. F.ex. Haghighi et al.[7] measured water content for pure CO2 in equilibrium with hydrates at 137.9 bar and found it substantially smaller compared to the previously reported values in GPA RR-99. Although our results do not coincide with those new measurements, they are in better agreement with them than with the older results reported by GPA, especially when it comes to CPA results (see Figure 4).

In another case, Seo et al.[8] measured the solubility of water in liquid CO2 at 6.1 and 10.1MPa in the presence of hydrate and observed a weak pressure dependence, contrary to the previously reported data of Song and Kobayashi. Indeed our models' results also suggest a weaker pressure dependence than the originally suggested but we believe that the high temperature data are still correct, referring to VHE conditions instead of LHE conditions as originally thought (taking into account the existence of a three-phase line (VLL) which ends at the UCEP) (see. Figure 5). As seen in the Figure, CPA EoS result almost coincide with the experimentally reported value of water vapor composition along the 3-phase line.

Finally, Eslamimanesh et al.[9] performed a thermodynamic consistency test based on an area approach and studied the reliability of experimental data of solubility in CO2-H2O system. They found 3 thermodynamically inconsistent data series [10-11] at 298.20K, 333.20K and 353.10K which we compared against our model correlations. VPT EoS was found the one in worse agreement with the specific data. 

Figure 5. Solubility of water in the liquid CO2 for the LCO2-H and LCO2-Lw phase between 5.992MPa and 10.34MPa.

Conclusions

The results reveal that CPA is a versatile model that can –in most cases- capture the complicated phase behavior of systems with NG mixtures and/or production chemicals at pipeline operating conditions, when combined with the vdW-P theory. Even when CPA is purely predictive (i.e. all binary interaction parameters are set equal to 0) it provides qualitatively correct results and can be used for calculating the thermodynamically stable phase, which is often not known in advance. 

Comparison between different models' results and experimental data shows that there is still a lot to be done both in terms of experimental data consistency tests and model development.

 References

[1] A.Eslamimanesh, A.H.Mohammadi, D.Richon, AIChE J. 57(9), 2566-2573 (2011).

[2] G.K.Folas, E.W.Froyna, J.Lovland, G.M.Kontogeorgis, E.Solbraa, Fluid Phase Equilib. 252, 162-174 (2007).

[3] ISO 18453, Natural Gas – Correlation between water content and water dew point.

[4] J.H.van der Walls, J.C.Platteeuw, Adv. Chem. Phys. 2, 1-56 (1959).

[5] W.R.Parrish, J.M.Prausnitz, Ind. Eng. Chem.. Process Des. Develop., 11, 26-34 (1972).

[6] R.J.Bakker, Geochimica et Cosmochimica Acta, 60(10), 1657-1681 (1996).

[7] H.Haghighi, A.Chapoy, R.Burgass, B.Tohidi, Proceedings of the 7th International Conference on Gas Hydrates 2011.

[8] M.D.Seo, J.W.Kang, C.S.Lee, J. Chem. Eng. Data, 56, 2626-2629 (2011).

[9] A.Eslamimanesh, A.H.Mohammadi, D.Richon, J. Chem. Eng. Data, 56, 1573-1586 (2011).

[10] T.Nakayama, H.Sagara, K.Arai, S.Saito, Fluid Phase Equilib. 38, 109-127 (1987).

[11] A.Bamberger, G.Sieder, G.Maurer, J.Supercrit. Fluids 17, 97-110 (2000).


Extended Abstract: File Uploaded