Gas-sampling accuracy unaffected by in-line separator

V.C. Ting
Chevron Petroleum Technology Co.
La Habra, Calif.

In experiments on three pipelines, the in-line separator caused no statistically significant condensation of unprocessed gas. Entrained NGL in the pipeline was observed, however, and gas liquid collected in the separator.

The U.S. petroleum industry follows Gas Processors Association (GPA) Standard 2166-86 for natural-gas sampling. In the standard, a sampling probe and an in-line separator are recommended as part of the sampling equipment system.

The in-line separator is used to remove any liquid entrainment in the sample gas stream to protect the gas chromatograph system. There is a concern, however, that thermodynamic conditions in the in-line separator cause the heavy, gaseous-phase hydrocarbon components to condense.

If condensation occurs while a sample is drawn from a pipeline, a lower than true gas-heating value can be expected.

The current gas-sampling method and analysis technique only account for the amount of gas phase flowing in the pipeline. In many cases, neglecting any amount of liquid condensate in the pipeline means valuable products are not measured.

In addition, liquid condensate in the line will affect the accuracy of the volumetric flow rate measurement of unprocessed natural gas.

The tests conducted led to a recommendation for use of an in-line separator for unprocessed gas sampling. Careful monitoring and recording of the amount of liquid collected in the separator are necessary. Small amounts of liquid accumulation in the separator also indicate flow metering error.

Measurement accuracy

Two types of natural gas are transported and measured in the pipelines: processed and unprocessed.

Processed natural gas consists of a high percentage of methane and is typically measured at the custody-transfer stations leaving a gas plant. Unprocessed gas measurement points, on the other hand, can be found near the wellhead, gas-gathering location, plant inlet, and offshore production platform.

The accuracy of processed and unprocessed gas measurement is of great importance because of the monetary value of the large volume of gas being transferred.

Chevron Petroleum Technology Co. has investigated the metering accuracy of processed natural gas. For a processed-gas system, flow-rate measurements are guided by measurement standards such as API 2530 or AGA Report No. 3.¹

In addition, thermodynamic and physical properties are defined in the AGA Report No. 8 standard.² The overall flow-measurement uncertainty is 1% for processed natural gas.

Unprocessed gas measurement is more complex. Heavy hydrocarbon components condense or evaporate when the gas flowing temperature and pressure change in pipelines.

Sampling and analysis of the flowing fluid is key in determining the correct heating value and density of the flowing gas at the point of sampling. If liquid is in the flowing stream, additional flow-measurement errors are introduced.^3-6

The flow measurement uncertainty for unprocessed gas flow can be higher than 1%, depending on compositions, liquid fraction, and the flowing conditions.

For obtaining natural-gas samples for analysis, GPA Standard 2166-86 is recommended.⁷ In this standard, a sample probe and an in-line separator must be used to obtain dry or wet natural-gas samples.

It is possible that the heavy gaseous-phase hydrocarbon components condense in the separator while the sample is drawn from a pipeline. If the condensation occurs in the separator, a lower-than-actual gas heating value can be expected.

Test locations

The gas-sampling tests were conducted at Chevron's Sand Hills gas plant. Unprocessed natural gas was gathered from the fields, compressed, and then processed in the gas plant.

Field samples were collected from three different pipelines at different flowing conditions. These represent unprocessed gas at a field-gathering location with high, intermediate, and low-pressure gas. The nominal conditions and typical gas compositions are listed in Tables 1 [6681 bytes] and 2 [11495 bytes].

High-pressure gas sampling was first tested at the North Line. This 12-in. line, located at the inlet of the Sand Hills gas plant, consists of a nominal 1,200 BTU/standard cu ft (scf) unprocessed gas flowing at a rate of 30 MMscf/day (scfd) with pressure of 700 psig.

Gas sampling tests at intermediate pressure were conducted at the L&H gas compressor station. This 8-in. pipeline is located at the inlet of the compressors with a gas flowing rate of 5 MMscfd at 80 psig.

The low-pressure sampling test was conducted at the McElroy Block 42 field-gathering station on a 4-in. line. The gas flowing pressure is 10 psig with a varying gas flow rate less than 500 Mcfd.

Sampling, analysis systems

Fig. 1 [12095 bytes] shows the experimental setup to determine the effect of an in-line separator. The sampling system and sampling probe were set up according to GPA Standard 2166-86 recommendations.

Sample gas was collected into two 1-liter sample cylinders, A and B. Teflon-coated stainless steel sample cylinders were used. Cylinder A was connected downstream of a 250-ml GPA Type B separator, while Cylinder B bypassed the separator.

A 1/4-in. insertion sampling probe with a full open gate valve was used for the sampling project. The probe, with a flat probe tip, was placed at the center of the pipe. The size of the gas sampling line was 1/4 in.

The sample probe was selected to collect unprocessed gas samples only. This gas-sampling system was not designed to collect representative gas/liquid samples.

The sampling line, separator, and sample cylinders were heated with heating jackets and tapes. The system temperature was maintained at 10° F. greater than line temperature during sampling. Pipeline pressure and ambient temperature were also measured.

A Hewlett Packard 5880 gas chromatograph (GC) analyzer was installed in the field.

The HP5880 GC was configured with two single-filament thermal conductivity detectors to determine constituents from methane through normal undecane, nitrogen, air, carbon dioxide, hydrogen sulfide, and water. The accuracy of the GC is calibrated within 1 BTU/scf.

Three types of standard gas (Table 3 [7865 bytes]) were used for calibration:

• Custom blend calibration gas

• Philip NGPA calibration gas

• H2S and CH4 mixture gas.

The custom blend standard gas and H2S/CH4 mixture were used at the initial calibration to determine the response factors for components C5-C9 and H2S. During normal daily operation, NGPA standard gas is used.

In the laboratory, the sample cylinder and sampling line connected to the GC were heated to 50° F. greater than the line temperature. The sample cylinder was laid horizontally on the table, and the sample line was sloped upward to the GC.

The GC oven temperature was programmed to operate in the range of 140-338° F.

Sampling methods

Three GPA sampling procedures were selected for testing:

• The fill-and-empty method

• The reduced pressure method

• The evacuated container method.

Two samples, A and B, were obtained according to GPA 2166 sampling procedures; Sample A was located downstream of the in-line separator and Sample B bypassed the separator.

A step-by-step procedure is listed in the standard and will not be duplicated here.

A total of 154 gas samples was collected during summer 1989 and winter 1990 using all three sampling procedures. Gas samples from Cylinders A and B were analyzed within 1 hr after being collected, and the corresponding heating value (BTU) was computed.

Properties and the heating value of gas components at 60° F. and 14.69 psia were selected from GPA Standard 2172-86.⁸

If the in-line separator caused the gaseous phase of the heavy hydrocarbon components to condense during sampling, the gas heating value collected in sample Cylinder A is expected to be lower than Cylinder B.

Therefore, we characterize the effect of the in-line separator by taking two samples, one with separator and another without, and compare the change of the gas heating value.

The following equation expresses the deviation of heating value from two sample cylinders:

Heating-value deviation, % 5

100 ( )

where:

BTUB = Gas heating value of Sample B without using a separator, BTU/scf

BTUA = Gas heating value of Sample A using a separator, BTU/scf

The effect of an in-line separator is expressed and plotted here as a heating-value deviation (Equation 1 [79509 bytes]). Discussion of test results is grouped in the order of sampling methods.

Pipeline gas temperature varied within 10° F. over the summer and winter periods when the test was conducted, and no significant ambient temperature effect was detected.

Fill and empty

The fill-and-empty meth od is commonly used in the field for spot-gas sampling because it requires the least preparation time.

For unprocessed natural gas sampling, this method is applicable when the sample container temperature equals or is greater than the source temperature.

Gas is purged through the clean sample cylinder for several cycles before a sample is collected. The number of purge cycles depends on the pipeline pressure. The final sample pressure equals the line pressure.

The effect of the separator in the fill-and-empty method, in terms of change of heating value at three different flowing conditions, is plotted in Figs. 2 [36098 bytes], 3 [15910 bytes],4 [9843 bytes]. The heating value deviation (Equation 1 [79509 bytes]) and gas heating value for each flowing condition are shown separately.

The left vertical axis represents the heating-value deviation, and the right vertical axis shows the gas-heating value. The dotted line is the mean heating-value deviation.

The majority of heating-value deviation data points was bounded within 0.4% (Figs. 2-4).

Each data point represents the difference of gas-analysis results (Samples A and B) at a set of conditions over 1 year. The mean heating value deviation of each data set represents the average value of many samples taken over different days in winter and summer periods.

Each data point in the set should be considered as an individual event, and the set of data should not be viewed as repeatable sample runs.

Hence, the use of statistical analysis, such as standard deviation, to treat the data set may not be meaningful. If any statistical analysis is to be applied, one must control such variables as gas compositions, line pressure, and gas temperature. One should view these figures as trend charts and not dwell on the fluctuation of the data points.

A bar chart indicating the mean heating value deviation for each flowing condition is presented in Fig. 5 [11344 bytes]. At the L&H compressor station (nominal pressure = 80 psig), the mean heating-value deviation is 0.17%.

In terms of heating value, samples collected without using a separator measured an average of 2 BTU/scf higher than those samples collected with a separator. The effect at North Line (high pressure) and Block 42 are 0.05% (0.6 BTU/scf) and 0.03% (0.4 BTU/scf) respectively. Test results indicated that there is no statistically significant effect of an in-line separator.

During the sampling tests at the North Line, yellowish NGL accumulated in the separator and Sample Cylinder B. The liquid samples were analyzed at Chevron Petroleum Technology Co. by a distillation method to determine the distribution of the distillates over the boiling-point range.

Test results are shown in Fig. 6 [32560 bytes]. Ninety percent of the liquid distillates had boiling points higher than 160° F. The composition of the liquid was analyzed at Chevron Research Technology Co. by a chromatography technique, and the results are presented in Table 4 [5230 bytes].

The NGL sample contained more than 87 mole % of C6+. This can be explained from a typical phase-equilibrium diagram (Fig. 7) for the high-gravity gas.

The pipeline flowing conditions are denoted in the diagram. The North Line is operating near the envelope of the two-phase region where NGL can easily drop out of the gaseous phase.

When NGL was found in the separator and sample Cylinder B, gas heating-value deviation measured from Cylinders A and B was similar to other tests. This suggested that liquid collected in the separator was free-liquid droplets flowing along with the gas in the pipeline, and liquid was accumulated in the separator during the purge cycles.

Although the natural-gas heating value can be determined with the fill-and-empty method, NGL in the pipeline was not measured with the current sampling technique.

Evacuation

The evacuated method was selected for testing because it is a recommended procedure for spot sampling when cylinder temperature exceeds or is less than the line temperature.

An evacuated cylinder can reduce the possibility of heavy hydrocarbon components condensing in the cylinder. But one must be sure that the valves, fittings, and sample lines are leak-free.

Since Chevron's objective was to investigate the effect of the in-line separator, the cylinder and line temperatures were maintained at greater than the line temperature to avoid condensation.

The evacuated method was conducted in the North Line and L&H Station.

The sample cylinders were evacuated to a pressure of 1 mm Hg or less in air. A vacuum gauge was used to ensure that the sample cylinders were not leaking. In addition, no oxygen was detected in the GC analysis in all the samples.

Figs. 8 and 9 show the effect of the separator in terms of gas-heating value deviation (Equation 1) at the North Line and L&H station. The average deviations are 0.08% and -0.02% (0.9 and 20.2 BTU/scf), respectively.

Again, no significant changes were detected. NGL was also found in the separator in the North Line.

Reduced pressure

The purpose of using the reduced-pressure method is to reduce the possibility of heavy hydrocarbon components condensing in the cylinder. It is designed to avoid liquid condensation by charging the sample cylinder to a lower pressure.

This method is similar to the evacuation method, except that the sample pressure must be slowly controlled to a desired pressure.

A sampling test was conducted at the North Line only because the reduced pressure method is recommended for sampling when line pressure is above 100 psig.

Evacuated cylinders are used to collect gas samples up to 200 psig from a line pressure of 800 psig. Fig. 10 shows the results of the effects of the separator using the reduced-pressure method.

Test results from seven sets of samples indicated that the separator has little effect on the gas samples. The mean deviation is calculated to be 0.047% (0.5 BTU/scf). Like other test methods, NGL was also found in the separator in the North Line.

A summary of the effect of the separator for the methods tested is presented in Table 5 [5884 bytes] in terms of heating value deviation and change of heating value.

The following conclusions can be derived from the field test:

• The in-line separator recommended by the GPA standard for use in sampling of unprocessed natural gas was tested at three conditions in the field similar to an oil and gas production operation.

• Results indicated no statistically significant effects on the natural gas heating value-measurement.

• Monitoring the amount of NGL in an in-line separator during unprocessed gas sampling is strongly recommended.

• Accumulation of gas condensate in the separator is an indication of NGL flowing in the pipeline. Any amount of entrained liquid in a gas pipeline affects the accuracy of flow rate measurement and liquid allocation procedure.

• Currently, no accurate procedure exists to collect a representative gas-liquid sample in a pipeline.

The effect of entrained liquid on gas-flow measurement and the accuracy of the liquid allocation method are not fully understood. Further research on gas-liquid measurement and sampling procedures is required.

Reference

1. API, "Manual of Petroleum Measurement Standards Chapter 14, Natural Gas Fluids Measurement, Section 3, Concentric, Square-Edged Orifice Meters, Part 1-General Equations and Uncertainty Guidelines," AGA Report No. 3, September 1990.

2. AGA, Transmission Measurement Committee, Report No. 8, "Compressibility Factors of Natural Gas and other Related Hydrocarbon Gases," Second Ed., November 1992.

3. Ting, V.C., and Shen, J.J.S., "Field Flow Calibration of Natural Gas Orifice Meters," ASME Journal of Energy Resources Technology, Vol. 111, No. 1, pp. 22-33, 1989.

4. Shen, J.J.S., and Ting, V.C., "Accuracy in Field Flow Measurement of Unprocessed Natural Gas," presented at the 1992 International Gas Research Conference, Orlando, Nov. 16-19, 1992.

5. Ting, V.C., and Corpron, G.P. "Effect of Liquid Entrainment on the Accuracy of Orifice Meters for Gas Flow Measurement," 1995 International Gas Research Conference, Cannes, Nov. 6-9, 1995.

6. Ting, V.C., "Effect of Orifice Meter Orientation on Wet Gas Flow Measurement Accuracy," SPE Gas Technology Symposium, Calgary, June 28-30, 1993.

7. Gas Processors Association, "Obtaining Natural Gas Samples for Analysis by Gas Chromatography," GPA Standard 2166-86, 1986.

8. Gas Processors Association, "Calculation of Gross Heating Value, Relative Density, and Compressibility Factor for Natural Gas Mixtures from Compositional Analysis," GPA Standard 2172-86.

The Author