Gas-Lift Valve Testing and ModellingPage 1

Testing and Modelling Gas-Lift Valves

Prepared for Presentation at the

2001 ASME/API Gas-Lift Workshop

February 6 - 7, 2001

By

Ken Decker, Decker Technology

Cleon Dunham, Oilfield Automation Consulting

Introduction

The artificial lift process of gas-lift is responsible for producing many millions of barrels of oil per day, worldwide. To be successful, most gas-lift wells depend on gas-lift valves to allow gas injection at the optimum depth and rate. Yet in many cases, the performance of gas-lift valves is not well understood. Consequently, the designs that employ these valves are often less than optimum, and in some cases are completely ineffective.

Why Are Gas-Lift Valves Important

Most gas-lift wells require a set or "string" of two or more gas-lift valves. When the gas-lift process is first started, the annulus is normally filled with liquid, e.g. completion fluid. This liquid must be removed for the gas-lift process to work. As gas is injected into the annulus at the surface, the increased pressure forces the liquid through the open gas-lift valves. This displacement process is called unloading.

Typically, a string of "unloading" gas-lift valves is needed to reach the desired depth of gas-lift injection. The valves are “spaced” at specified depths so that relatively low surface injection pressure can be used to sequentially inject gas through each lower valve. For unloading to occur, the annular completion fluid must flow through the gas-lift valves and be "u-tubed" to the surface. As each gas-lift valve in the tubing string is "uncovered," it must remain open and inject gas into the tubing string long enough to reduce the flowing tubing pressure at the next lower valveso the gas-lift injection process can continue on down the well.

The purpose of each unloading gas-lift valve is to support the unloading process, inject enough gas to allow transfer to the next lower valve, and then close and remain closed so gas-lifting can occur on a continuous basis through a lower orifice or special gas-lift valve from the desired depth of gas injection, which is normally just above the production packer.

For unloading to occur effectively, several factors are important:

  • Each valve must be open at the right time. It must be open initially to allow the unloading process to begin and proceed. It must be able to re-open if the well is temporarily stopped and needs to be re-started. (This process is called "kick off.") The gas-lift designer must be able to accurately predict when each valve will be open and when it will re-open.
  • Each valve must open enough to transmit the desired amount of gas. Gas-lift valves have a load rate that may vary from a few tens to many hundreds of pounds per inch depending on the physical design and construction of the valve. The gas-lift system designer must be able to predict how far each valve will open under each condition of upstream and downstream pressure, and how much gas it will transmit under each condition. A valve that is not fully open is in the "throttling" state. Some valves will open slightly, inject a little gas (that is, "throttle"), close, and then re-open to repeat the process. This phenomenon is called cycling or heading and may be very detrimental to the performance of the well and the gas-lift system.
  • Each valve must close when it is no longer needed for the unloading or kick-off process. The designer must be able to predict when each valve will close and what upstream and downstream pressure conditions will be required to keep it closed.

Until the development of accurate gas-lift valve testing and modelling procedures (see next section), gas-lift valve open and close calculations were based on a steady-state force/balance equation. Gas passage through the valve was based on the Thornhill-Craver1 equation for calculating gas flow through a square-edged orifice. Neither of these methods accurately represented reality in the dynamic environment of an operating gas-lift valve in a "real" oil well. The lack of tested performance data on valves caused designers to use many “rules of thumb” which were learned experiences from previous successes and failures. Unfortunately, these “rules of thumb” usually worked only for the oil field in which they were learned. Thus, gas-lift designs often were significantly in error and as a result, many gas-lift wells did not unload and/or perform effectively.

API Recommended Practice 11V2

In January 1995, the American Petroleum Institute (API) published the first edition of API Recommended Practice (RP), 11V2, "Gas Lift Valve Performance Testing."2 A 2nd edition will be released in early 2001. This document provides recommended procedures for flow performance testing of wireline-retrievable and tubing-retrievable IPO (injection pressure operated) and PPO (production pressure operated) gas-lift valves.

Prior to the publication of this recommended practice, there was no standard for gas-lift valve performance testing. And, there was no standard, scientific method for developing gas-lift valve performance models. That is, there was no published, accepted way to accurately predict when a gas-lift valve in "real world," dynamic service, would open, how far it would open, when it would close, and how much gas could be transmitted through the valve at various upstream and downstream pressures.

With the publication of API RP 11V2, the gas-lift industry now had an accepted way to test and model gas-lift valve performance. However, there was still no specific company or process for actually placing this recommended practice into use so that Producing Companies could have full access to actual performance test data and models on their gas-lift valves.

Gas-Lift Valve Performance Clearinghouse (VPC)

In 1996, several companies came together to form a joint industry project called the Valve Performance Clearinghouse (VPC)3. (The original and current members of the VPC are listed in Table I.) The purpose of the VPC is to test gas-lift valves according to the API RP 11V2 procedure, to evaluate their performance, and to develop accurate models that may be used to accurately predict:

  • When each valve will be open, and when it will re-open, under dynamic operating conditions.
  • How far each valve will be open (how far the stem will travel off seat), during various conditions of upstream and downstream pressure.
  • How much gas can be transmitted through the valve at each opening, based on an accurate flow model and characteristic for the valve.
  • When each valve will close, under dynamic operating conditions.

To develop these models, the VPC determines:

  • The accurate opening pressure of the valve, at operating temperature.
  • The accurate closing pressure of the valve, at operating conditions.
  • The flow characteristic (Cv) of the valve.
  • The load rate of the spring and/or bellows.
  • The maximum effective stem travel.

None of this information is normally published by the gas-lift valve companies.

The normal VPC process is to conduct a full suite of tests on one valve of a given model, per API RP 11V2, and to determine the gas-lift performance parameters and model based on these test results. However, the VPC has also used gas-lift valve test data that has been donated by its member companies, and data that is in the public domain, such as that generated by the Tulsa University Artificial Lift Project (TUALP)4. The TUALP project did not generate all of the test data that is specified in the API process, so the models that are based on the TUALP data are more limited than the "full VPC" models.

VPC Results

Table 2 lists the gas-lift valves and orifices that have so far been tested and/or modelled by the VPC. For the valves that have been tested by the VPC, qualitative performance information is also available. For instance, some valves had very limited gas passage rates; some could not close under normal operating conditions, etc. This information is propriety to the members of the VPC and is not included in this paper.

Companies that are not members of the VPC can use models for specific valves by licensing them from the VPC. Several companies that tend to use only one or two types or models of valves have chosen this method to access the specific gas-lift performance models they need.

Using VPC Results

The VPC produces a small personal computer program that can be used by any member company to estimate the flow performance of a modelled gas-lift valve for any pressure and temperature conditions. A typical plot from the program is shown in Figure 1.

Figure 2 shows an interesting comparison of the predicted performance of a typical 1-inch IPO gas-lift valve vs. the performance of that same valve as it would be predicted using the Thornhill-Craver model.

How Member Companies Use VPC Results

The member companies have made significant use of the VPC test results and models. For instance, in Shell International:

  • Only gas-lift valves that have been tested and/or modelled by the VPC are recommended for use by the Shell Operating Companies around the world.
  • Those gas-lift valves that have been found to be inadequate in design and/or performance are specifically not recommended for use.
  • The gas-lift valve models developed by the VPC have been incorporated into Shell's gas-lift design, analysis, and optimization program, WinGLUE5. These models are available for use by all Shell gas-lift personnel for both design and analysis purposes.
  • In some cases, the poor performance of a valve under VPC testing has led the maker of that valve to redesign it and come up with a far superior product. (Clearly this is a win/win situation.)
  • Shell developed a model for choked gas-lift valves6 based on the VPC model for the valve performance and the Thornhill-Craver model for the choke. Shell routinely places chokes in unloading gas-lift valves to help in the unloading process.

Future Plans

As seen in Table 2, many of the most common and frequently used gas-lift valves have already been tested and modelled. For the future, the VPC has the following plans:

  • Continue to test those valves and orifices that the member companies wish to have tested.
  • Test selected new valves as they are offered to the marketplace. (For instance, Shell will not recommend a new valve until it has been tested by the VPC.)
  • Test several valves of a given model to evaluate the consistency of the manufacturing process.
  • Evaluate the potential erosion damage that may occur during the unloading process.
  • Evaluate the impact of chokes on gas-lift valve performance and erodability.
  • Evaluate the potential life cycle of gas-lift valve bellows.
  • Evaluate the impact of various metallurgies on the performance and life of gas-lift valves.

Other Members Welcome

The VPC is a joint industry project. Membership is open to all gas-lift users and suppliers. The member companies have found great value in having reliable, accurate information and performance models of their gas-lift valves. And, there has been high value in being able to screen out valves that don’t provide acceptable performance.

Conclusions

For the gas-lift process to be successful in most wells, gas-lift valves must be fully understood and they must work properly. We know that gas-lift valves do not behave as previously thought and characterized with the steady-state force/balance equation for open/close calculations and the Thornhill-Craver model for gas passage.

To effectively design and operate gas-lift wells, it is essential to know the performance of the gas-lift valves being used, and to design the gas-lift process accordingly.

The API RP 11V2 recommended practice for gas-lift valve testing and modelling now provides a tool for the gas-lift industry that lets us know how gas-lift valves should perform and to accurately predict their actual performance in each well under dynamic operating conditions.

The Valve Performance Clearinghouse (VPC), a joint industry project, is an effective vehicle for employing the API technique to provide accurate gas-left test results and performance models to the member companies.

Acknowledgements

The authors wish to thank Shell International and the member companies of the Valve Performance Clearinghouse for permission to prepare and present this paper.

References

  1. Thornhill-Craver is a standard, published method for calculating the gas flow rate through a "square edge" orifice.
  1. API Recommended Practice 11V2, "Gas Lift Valve Performance Testing," 1st Edition, January 1995. This document covers the test procedures for flow performance testing of wireline-retrievable and tubing-retrievable IPO (injection pressure operated), and PPO (production pressure operated) gas lift valves. Pages: 36.
  1. The Valve Performance Clearinghouse (VPC) is a joint industry project with the purpose of testing and modelling gas-lift valves. For more information, see the VPC web site
  1. TUALP, the Tulsa University Artificial Lift Project, is a consortium of companies that sponsor artificial lift research at the University of Tulsa, in Tulsa, Oklahoma, USA. For more information, see the TUALP web site
  1. WinGLUE is Shell International's gas-lift design, analysis, and optimization program. For more on the WinGLUE, see the WinGLUE web site at
  1. Ken Decker, Cleon Dunham, and Burney Waring, "Using Chokes in Unloading Gas-Lift Valves."Prepared for presentation at the 2001 ASME/API Gas-Lift Workshop, February 6 - 7, 2001.

Tables

VPC Member Companies

Table 1

Initial VPC Member Companies2001 VPC Member Companies

ChevronChevron

Edinburgh Petroleum ServicesExxon/Mobil (pending)

Exxon PDVSA (pending)

Shell InternationalShell International

Weatherford Weatherford

Gas-Lift Valves and Orifices Tested/Modelled by the VPC Through 2000

Table 2

1" IPO (Injection Pressure Operated) Gas-Lift Valves

Altec AT1-CF with 10, 12, 14, and 16/64" choke sizes. 1" wireline retrievable.

Camco BK with 1/8", 3/16", 1/4", and 5/16" ports. 1" wireline retrievable.

Weatherford R-1 with 10, 12, 16, and 20 port sizes. 1" wireline retrievable.

Weatherford R-1BL with 10, 12, 16, and 20 port sizes. 1" bottom latch wireline retrievable.

Camco/Merla NM-16R with 1/8", 3/16", 1/4", and 5/16" ports. 1" wireline retrievable.

McMurry Macco R-1D with 1/8", 3/16", 1/4", and 5/16" ports. 1" wireline retrievable.

McMurry Macco JR-STDN with 1/8", 3/16", 1/4", and 5/16" ports. 1" wireline retrievable.

McMurry Macco C-1 with 1/8", 3/16", and 1/4" ports. 1" tubing retrievable.

1" PPO (Production Pressure Operated) Gas-Lift Valves

Camco BKF-6 with 1/8", 3/16", and 1/4" ports. 1" wireline retrievable with spring and crossover seat.

Camco BKR-5 with 1/8", 3/16", and 1/4" ports. 1" wireline retrievable with nitrogen charge and crossover seat.

Camco SRF-10 with 1/8", 3/16", and 1/4" ports. 1" wireline retrievable with spring and crossover seat.

McMurry Macco RF-1 with 10, 12, and 16 choke size. 1" wireline retrievable with spring load.

Weatherford RF-1BL with 8, 10,12, and 16 choke size. 1" wireline retrievable with spring load. Bottom latch.

1" (Alternative Operation) Gas-Lift Valves

Altec AT1-VL with 1/8", 3/16", and 1/4" port sizes. 1" wireline retrievable differential with independent re-open.

Weatherford RPDV-2 with 10, 12, and 14 choke size. 1-1/2" wireline retrievable differential.

Camco/Merla LNM-31R with 21, 25, and 29/64ths ports. 1" wireline retrievable proportional response.

Camco/Merla WFM-14R with 8, and 10 port size. 1" wireline retrievable pilot operated valve.

1.5" IPO Gas-Lift Valves

Camco R-20 with 3/16", 1/4", 5/16", 3/8", 7/16", and 1/2" ports. 1-1/2" wireline retrievable.

Weatherford R-2 with 12, 16, 20, 24, and 28 64th-inch port size. 1-1/2" wireline retrievable.

Camco/Merla N-17R with 1/8", 3/16", 1/4", 5/16", and 3/8" ports. 1-1/2" wireline retrievable.

McMurry Macco C2 with 1/8", 3/16", and 1/4" ports sizes. 1-1/2" tubing retrievable.

1" Gas-Lift Orifice

Camco NOVA with 13/64ths ports. 1" orifice valve with verturi choke.

Camco DCR-DK with 24 and 32 choke sizes. 1-1/2" wireline retrievable dump/kill valve

Typical VPC Gas-Lift Valve Performance Plot

Fig. 1

Comparison of Gas-Lift Valve Performance Based on VPC Model

Vs. Performance Based on Thornhill-Craver Model

Fig. 2