Liquid contaminants in natural gas pipelines have recently become an increasing concern. Typical liquid contaminants can include water and hydrocarbons condensed from the gas stream, glycol carryover from inefficient equipment or process upsets, methanol injected to avoid hydrate formation, and compressor oils.
These contaminants are usually separated and filtered from the gas stream before delivery to a pipeline. However, separators and filters are never 100% efficient, so there will always be traces of contaminants that can enter the pipeline, and ambient temperatures below the dew point of the gas stream can cause condensation in the pipeline itself. Only representative samples of the gas phase are required for custody transfer purposes, so the ability to reject liquids at the sample probe is of interest to natural gas pipelines.
Industry practice requires that natural gas sample probes be inserted vertically from the top of a straight, horizontal pipeline run, with the tip of the probe in the center one-third of the pipe cross-section. Both of these requirements were created to keep the probe from capturing liquids or contaminants that may be migrating along the pipe walls.
However, the probe must also be short enough to prevent flow-induced resonant vibration from breaking the probe inside the pipe. Besides leading to contaminated samples, broken probes and debris in the pipeline can damage downstream equipment.[, , ] The latest API and GPA sampling standards include a formula to calculate the maximum allowable probe length for avoiding flow-induced vibration, but in some cases, the probe lengths specified by this formula will not reach the center one-third of the pipe cross‑section.
To help resolve these conflicting probe length requirements, Pipeline Research Council International (PRCI) funded a series of projects at Southwest Research Institute (SwRI).[, ] This article describes the findings of one part of that research, which experimentally assessed the effects of probe insertion length on the accuracy of gas samples drawn from gas‑liquid hydrocarbon pipe flows.
The ultimate goal was to recommend minimum probe insertion depths for collecting representative natural gas samples where contaminants may be present. The tests described here evaluated the performance of a straight-cut sample probe in annular flows and annular-mist flows of methane and hydrocarbon liquids. The findings can be useful to meter station designers and others whose work involves collecting samples from gas pipelines with liquid contaminants.
SwRI’s Multiphase Flow Facility (MFF) is a closed flow loop capable of supplying various combined streams of gas and liquid to a test section for meters and other equipment. The MFF can operate with natural gas, crude or refined oil, hydrocarbon condensates, water and combinations of these fluids at pressures from 100-3,600 psig and at liquid volume flow fractions (LVFs), ranging from 0-100%. For the sample probe tests, a heptane distillate was chosen to represent a generic liquid contaminant in a gas pipeline because its presence could easily be identified through sample analysis. A supply of methane of 99.7% purity represented the natural gas stream.
Figure 1 shows the test section layout. After the gas and liquid phases were measured separately, the phases were recombined in a 6-inch-diameter line and passed through a reducer to a 3-inch-diameter flow development section. The flow development section had a length of 244 pipe diameters, long enough to produce a fully developed annular flow at an isokinetic reference probe and the sample probe under test. Downstream of the probe, an axial flow visualization spool allowed video recordings of the flow.
Figure 1: MFF layout for the sample probe tests.
Evaluating the performance of the sample test probe at different insertion depths required reference measurements of the gas and liquid distributions in the pipe flow. MFF staff developed a liquid entrainment measurement system (LEMS) to study the properties of multiphase flows, particularly liquid entrainment behavior in high-pressure hydrocarbon fluids (Figure 2). The LEMS uses a Pitot-style isokinetic probe mounted on a traverse mechanism to collect reference samples of the mixed gas-liquid flow at different elevations in the pipe.
The reference samples are drawn into a collection bottle where the gas and liquid are separated by gravity. The gas phase flows through a Coriolis meter that measures the gas flow rate of the reference sample. A differential pressure transmitter measures the rate of change of the liquid level in the bottle, and this measurement is used to determine the liquid flow rate of the reference sample.
Figure 2: LEMS schematic and view of isokinetic probe tip.
A+ Corporation (A+) provided a customized Genie Model 760 Direct Drive Probe™ for the tests. This probe has a threaded shaft and an NPT process connection that allows the probe depth to be adjusted without opening the pipeline to atmosphere. The tip of the customized straight-cut probe was reduced from the standard diameter of 0.7 inches to 0.25 inches.
This change reduced the blockage in the 3-inch Schedule 160 pipe used for the test run and provided closer geometric similarity to sampling arrangements at pipeline field sites. To avoid flow interference between the probes, the isokinetic reference probe and the test probe were separated by 71.75 inches, or 287 times the diameter of the upstream probe.
To assess the performance of the test probe, a sample conditioning system (SCS) was used to vaporize any liquid heptane ingested by the probe during gas sampling. This approach ensured that the liquid contaminant would be delivered along with the extracted gas sample to a gas chromatograph (GC) for analysis. The analysis of each vaporized sample determined whether any liquid had been ingested by the test probe and quantified the heating value bias due to the liquid contamination.
The SCS incorporated a pair of regulators to reduce the sample stream pressure in two stages from 900 psig to 20 psig. The complete sample line to the GC inlet was wrapped in heat trace set to a temperature of 150°F to maintain the samples as vapor in transit to the GC. A 150-watt heated regulator was used as the second-stage regulator to offset Joule-Thomson cooling. The heated samples were transported to an ABB NGC 8209 C9+ GC on extended loan to PRCI and SwRI.
To accommodate the flow capacity of the heated regulator, samples were taken at a flow rate consistent with the use of the GC or a field analyzer. A “fast loop” was also installed at the GC inlet to maintain the sample flow rate and avoid stream stagnation between samples. As a precaution against slugs of liquid reaching the GC, a Welker® Analyzer Liquid Shut-Off Valve model ALS-1 was installed in the SCS between the regulators and the GC inlet.
In a fully developed gas-liquid flow, the two phases will be in equilibrium, and the gas phase will be saturated with components from the liquid phase. This saturated gas composition – not the pure methane used as the gas supply – represents the gas phase that a sample probe would collect if it perfectly rejected all the free liquids from the stream.
At the start of the tests, the methane gas was saturated with the heptane distillate, circulated through the loop, and the saturated gas was sampled and analyzed. These initial samples provided reference heating values for comparison to the test probe samples in the gas-liquid flows. During the gas-liquid tests, the gas portions of the gas-liquid reference samples collected by the LEMS were also transferred by the SCS to the GC and analyzed to check the saturated gas heating values. As protection against liquid carryover from the LEMS reaching the GC, a custom high-pressure Welker Fluid Sentinel sample conditioner was installed on the gas outlet of the LEMS collection bottle.
The arrangement of the phases in a gas-liquid pipe flow, known as the flow regime, depends on variables such as the flow rates of the separate phases, the pipe diameter, and the phase densities. Figure 3 shows several possible gas-liquid flow regimes in a horizontal pipe. For this project, it was decided to test the straight-cut probe in annular or annular-mist flows with low LVFs of no more than 2%. It was also decided to vary the thickness of the liquid annulus since more liquid at the pipe wall could make collecting a gas-only sample more difficult.
Figure 3: Examples of gas-liquid flow regimes in a horizontal pipe.
The chosen test conditions corresponded to flow regimes known as API Type I and API Type II flows. Physically, API Type I flows generally have the lowest liquid flow rates and least liquid momentum of all API flow types. In most oil and gas applications, API Type I flows have a bulk LVF below 0.2%, and the phases are distributed in either a stratified flow or, as observed through the MFF sight glass, a mist flow with a very thin liquid annulus (Figure 4). Type I flows may occur in a gas stream just below its hydrocarbon dew point. API Type II flows typically have higher liquid flow rates and a thicker liquid annulus than Type I flows and can represent a flow condition due to a process upset.
The flow regime map (Figure 5) demonstrates how the final flow conditions were chosen. Flow regime maps such as these are commonly used to predict the approximate transitions between flow regimes and flow patterns. The dashed curve marks the MFF upper operating limit for methane‑heptane flows in 3-inch Schedule 160 pipe. This pipe size was chosen for the test section as the best compromise between strong annular-mist flow conditions and a practical scale for the test equipment. Table 1 lists the nominal flow conditions chosen for the tests.
Figure 4: Observed liquid distributions in API type I flows (left) and Type II flows (right).
Figure 5: Oil-gas flow regime map showing superficial velocities of chosen test conditions.
Table 1: Nominal flow conditions for the sample probe tests.
Figure 6 compares the reference LVFs measured by the LEMS in the Type I flow to the LVFs of the samples from the straight-cut test probe. The horizontal axis shows the relative distance from the pipe centerline to the pipe wall (r/R). The local LVFs measured by the LEMS increased toward the bottom of the pipe flow, confirming that gravity pulled the liquid mist in the gas core downward.
The LVFs computed from the straight-cut probe samples were lower than the reference LEMS values, indicating that the straight-cut probe was rejecting some of the entrained liquids. This suggests that placing the probe tip closer to the upper wall, but not within the liquid annulus, is more likely to produce a representative gas-only sample.
Figure 6: Comparison of Local LVFs in API Type I flow to LVFs of samples from the straight-cut test probe.
For the same flow, Figure 7 shows that the heating values of the straight-cut test probe samples fell between the heating values of the saturated gas only and the local heating values of the gas-liquid mist flow. This confirmed that the straight-cut probe ingested some liquids, causing a bias in the sample heating value between 5-7 Btu/scf.
As with the LVF data, the heating value errors are slightly less near the upper pipe wall, suggesting this is a better sample location to avoid liquids in the gas core. Therefore, to avoid liquid contaminants that may be present in a pipeline, it may be preferable to place the probe tip above the center one-third of the pipe cross-section.
Figure 7: Comparison of local gas-liquid heating values in API type I flow to heating values of samples from the straight-cut test probe.
Figure 8 compares LVF sample data from the Type II flow on a logarithmic scale. The local LVFs measured by the isokinetic probe in the core during Type II flow were an order of magnitude less than in the Type I flow. This indicates that in the Type II flow, the mist concentration in the gas core was much lower than in the Type I flow and that most of the liquid was along the walls.
The straight-cut probe ingested more liquids from the Type II flow than the reference isokinetic probe. It was conjectured that sprays of liquid from the bottom of the pipe introduced extra liquids into the downward-facing test probe while the forward-facing isokinetic probe only captured mist from the gas core.
Figure 8: Comparison of local LVFs in API Type II flow to LVFs of samples from two test probes.
Figure 9 compares the heating values of the samples from the Type II flow. The error bars on the test probe data indicate large variations between consecutive samples at the same flow conditions, indicating a dynamic two-phase flow.
At the pipe centerline, the straight probe samples had a higher average heating value than the isokinetic probe samples, suggesting that the dynamic flow conditions or the probe geometry may be biasing the sample. In this type of flow, which can occur after a significant process upset, obtaining a representative sample of only the gas phase is highly unlikely without active separation of the liquids. Therefore, there is no preferred location for the probe tip to obtain gas-only samples from API Type II flow.
From these tests, recommendations are possible on probe insertion depths in gas pipelines that may carry small amounts of liquid contaminants. A key observation is that samples from the straight-cut probe in the API Type I flow were repeatable. This suggests that the Type I gas-liquid flow patterns were stable and that the straight-cut probe can block some, but not all, of the mist from entering the sample. The Type I tests also showed that the local LVF due to mist in the gas core decreases toward the top of the pipe. Therefore, to minimize liquid ingestion from Type I flows, it is preferable to insert the probe tip to a depth above the center one-third of the pipe cross‑section.
The minimum probe insertion depth in Type I flows must avoid liquid ingestion from the annulus. Unfortunately, the annular thickness depends on many variables, including the pipe diameter, the LVF, the Reynolds numbers of the gas and liquid phases, the pipe wall friction factor, and the upstream pipe geometry. Further work is required to identify a practical minimum probe insertion depth that will avoid ingestion of liquid pipe wall contaminants in all flow scenarios.
In the API Type II flows, much of the liquid occupied the annulus around the pipe wall. Samples from the straight probe were less repeatable in these tests, likely due to ingestion of liquid sprays from the annulus. The Type II results indicated that the straight-cut probe could not collect a representative gas-only sample in these flows, so there would be no advantage or disadvantage to inserting the sample probe inside or outside the center one-third of the pipe cross‑section.
From the standpoint of PRCI’s operating company members, the prime benefit of being a PRCI member is the ability to conduct research in a collaborative environment that allows leveraging other members’ resource contributions. This project is one example where 13 operating companies came together to fund experimental work in SwRI’s MFF to determine the minimum required insertion depth of gas sample probes to avoid the ingestion of liquid contaminants that may be present in the gas flow.
API Manual of Petroleum Measurement Standards, Chapter 14 – Natural Gas Fluids Measurement, Section 1 – Collecting and Handling of Natural Gas Samples for Custody Transfer, 7th Edition, American Petroleum Institute, Washington, D.C., May 2016.
GPA Midstream Standard 2166-05, Obtaining Natural Gas Samples for Analysis by Gas Chromatography, GPA Midstream, Tulsa, OK, October 2005 (reaffirmed 2017).
ISO 10715:1997, Natural gas – Sampling guidelines, International Organization for Standardization, Geneva, Switzerland, June 1997.
EEMUA Publication 138, Design and Installation of On-Line Analyser Systems, Edition 2, Engineering Equipment and Materials Users’ Association, London, UK, 2010.
George, D., and Thorson, J., “Sample Probe Insertion Depth Testing,” final report to Pipeline Research Council International, Inc., Catalog No. PR-015-15608-R01, Chantilly, VA, December 2016.
George, D., and Grant, C., “Study of Sample Probe Minimum Insertion Depth Requirements,” final report to Pipeline Research Council International, Inc., Catalog No. PR-015-14609-R01, Chantilly, VA, May 2015.
Viana, F., Mantilla, I., Mohan, R., and Shoham, O., “Liquid Entrainment in Gas at High Pressure, Part 1: Experimental Approach and Initial Testing,” Proceedings of the 10th North American Conference on Multiphase Technology, Banff, Canada, June 2016.
Welker Incorporated, Welker Analyzer Liquid Shut-Off, Sugar Land, TX, www.welker.com, 2015.
Welker Incorporated, Welker Fluid Sentinel Sample Conditioning System, Sugar Land, TX, www.welker.com, 2015.
ASME Technical Report MFC-19G-2008, Wet Gas Flowmetering Guide, American Society of Mechanical Engineers, New York, NY, July 2008.