Sie sind auf Seite 1von 4

IPO Gas Lift Design With

Valve Performance
Kenneth L. Decker, Decker Technology

Summary
Although gas lift design has been practiced for many years with
the use of the Thornhill Craver (TC) equation (Cook and Dotterweich 1946) to predict the rate of gas passage through a gas-liftvalve port, the equation and charts were never intended for use
with live gas lift valves.
It is now possible to obtain tested valve-gas-passageperformance data for any pressure and temperature conditions. To
date, a method has not been provided for the use of this information during the design stage.
This paper will show how tested valve-performance data can be
used in the design of 1-in. (25.4-mm) injection-pressure-operated
(IPO) gas-lift valves.
Introduction
The American Petroleum Institute (APIs) recommended method
of spacing and sizing gas-lift-valve ports is published in APIs
Recommended Practice (RP) 11V6 (RP 11V6, Recommended
Practice for Design of Continuous Flow Gas Lift Installations
Using Injection Pressure Operated Valves. 1992). This RP describes the design technique that has been used for many years
with considerable success and uses the TC equations and charts as
the principal methods of sizing gas-lift-valve ports.
There are several commensurate assumptions with the use of
the TC chart. First, it assumes that the valve port is fully open,
and second, it assumes an unobstructed flow path through the
valve. Both of these assumptions could be incorrect, depending
on the type of gas lift valve used and the pressure being applied to
the bellows.
Every IPO gas lift valve has a property called loadrate. This
property refers to the amount of opening a valve will achieve for
a given annulus and tubing pressure. In most cases, a gas lift valve
is rarely fully open when passing gas. Secondly, the valve and/or
stem, downstream restrictions, and the reverse-flow check usually
obstruct the flow passage through a gas lift valve.
As a consequence of these exceptions, the flow rate through the
gas lift valve is considerably less than that of the TC predictions.
In many cases, the design engineer will apply a safety factor to
the TC predictions. The value of the safety factor is a result of
experience, but it usually varies from 50 to 80% of the value given
by the TC correlation.
It is now possible to predict accurately the rate of gas passage
through a gas lift valve for any pressure and temperature conditions. This is possible as a result of correlations developed from
actual tests of the valves at pressures similar to those encountered
in service, correlations made available by license through the
Valve Performance Clearinghouse (VPC).
Background
The purpose of gas lift is to lighten the flowing-production gradient by injecting gas into the production string. There are two
phases to a gas lift operation: (1) unloading and (2) operating. The
objective of unloading is to leverage the injection pressure by
sequentially injecting gas through deeper unloading valves until
the operating valve is reached. The operating phase should be a

Copyright 2008 Society of Petroleum Engineers


This paper (SPE 109694) was accepted for presentation at the 2007 SPE Annual Technical
Conference and Exhibition, Anaheim, California, U.S.A., 1114 November 2007, and revised for publication. Original manuscript received for review 9 August 2007. Revised manuscript received for review 26 December 2007. Paper peer approved 8 January 2008.

464

single-point injection through the operating valve over a range of


injection rates.
The API design method recommends that each lower unloading
valve should use an injection pressure that is approximately 138 to
345 kPa (20 to 50 psi) lower than the upper valve. The theory is
that as each lower valve is uncovered and begins injecting, the
injection pressure would drop by this amount. This technique
helped to close upper valves and also gave the operator a mechanism
(wellhead-injection pressure) to determine which valve was open.
This theory is based on the assumption that the amount of gas
being injected at the surface is equal to approximately 50 to 80%
of the combined total of the injecting valve and the next lower
valve when it uncovers. If this is true, the wellhead injection
pressure will drop 138 to 345 kPa (20 to 50 psi) as each lower
valve is uncovered. If the amount of gas injected at the surface is
greater than the combined total of the upper valve and the next
lower valve, the injection pressure will not drop but will continue
to increase.
This paper will suggest a modification to RP 11V6 (1992). The
modification will involve the use of tested valve-gas-flow rates to
determine port sizes. The valve-performance predictions will be
supplied by the VPC correlation. These correlations provide
gas-passage-flow-rate predictions that are within 15% of actual
flow for any pressure and temperature conditions.
Inflow/Outflow Analysis
The first step in designing a gas lift well is to determine the
theoretical lifting point and production rate on the basis of the
reservoir, tubing completion, and available gas-injection pressure
and rate. This step is well documented in RP 11V6 (1992) and
should be followed. It has been practiced for many years on a wide
variety of wells and has proved to be reliable.
Wellhead-Injection Pressure
The next step in spacing the unloading valves is to determine the
surface-gas-injection pressure. Several common terms are used to
describe this pressure: available wellhead-injection pressure, operating pressure, and kickoff pressure. Because the injection pressure is such an important variable in the design, a definition of
these terms is in order.
Available wellhead-injection pressure is the pressure upstream
of the gas-injection-control choke during normal operations. This
choke may be placed at the wellhead or may be placed at a manifold quite distant from the wellhead.
Operating pressure is the surface annulus injection pressure
when lifting at the operating valve. For successful gas lift operation, this pressure must be lower than the available wellheadinjection pressure.
Kickoff pressure is the pressure available at the wellhead during the unloading process. The design engineer chooses this pressure, and this is normally the same as the available wellheadinjection pressure.
It is advantageous to the IPO-design technique to select the
highest kickoff pressure possible. This will allow the design engineer to achieve injection at the greatest depth. However, it is not
advisable to use a kickoff pressure equal to the available wellheadinjection pressure, for several reasons.
The design of a new well uses reservoir information that may
not be accurate. The flowing pressure gradient could be heavier
than anticipated, requiring more pressure and gas volume to reach
the operating point.
In most cases, the gas lift valve test-rack pressure (TRO) can
be set to 35 kPa (5 psi) of the desired pressure. This assumes
November 2008 SPE Production & Operations

It is advantageous to have a near-steady flow rate through the


orifice for small changes in differential pressure. The design engineer plays a significant role in this choice of differential pressure
at the orifice. If the injection-pressure and production-pressure
systems are stable, using the minimum differential is a good decision. If either is subjected to fluctuations or the well has a history
of slugging or surging, the wise choice is to use a larger differential
pressure at the expense of being unable to inject deeper. The
primary concern is to ensure that the well is stable when injecting
through the operating valve. An unstable well is less productive,
causes surface-collection problems, and could induce instabilities
at other wells. Additionally, unstable flow could result in nearwellbore damage affecting sand control and water cut.

Fig. 1Flow performance of top valve.

good gas lift shop practices. It is common to find that the actual
valve set pressures are as much as 138 kPa (20 psi) from the
desired pressure.
Finally, it will be seen that additional injection pressure is
required to achieve the desired gas-flow rate through the unloading
valves. For these reasons, it is recommended to use a kickoff
pressure that is at least 345 kPa (50 psi) less than the available
wellhead-injection pressure as the design kickoff pressure. This
may cause the operating valve to be positioned shallower than
desired, but the alternative may be a well that cannot reach the
operating valve.
Valve Performance
Current design method assumes that when the opening pressure of
a valve (Pvot) is reached, the valve will be open fully. It is this
assumption that has encouraged the use of the TC equation as the
means of determining the rate of gas that will flow through the
valve. If in fact this were true, the use of the TC equation would
be justified. However, when the opening pressure of the valve is
reached, the valve does not open fully.
Every IPO valve has a property referred to as a loadrate. This
is a measure of the amount of increase in pressure above the dome
pressure required to open the valve. In the case of 1-in. IPO valves,
the amount of pressure increase required to open the valve fully
may be as much as 345 to 550 kPa (50 to 80 psi) greater than dome
pressure. In the case of 1-in. IPO valves, the amount of increase
could be 138 to 345 kPa (20 to 50 psi).
Current design method does not make a distinction between 1and 1-in. valves when sizing the ports. For the same set pressure
and port size, the 1-in. valve will flow a much higher rate than
the 1-in. valve. The principal reason for the higher flow rates of the
1-in. valves is the lower loadrate and the much larger bellows/
port ratio.
In the majority of cases, a 1-in. unloading valve will be throttling the flow during unloading. As shown in Fig. 1, a 1-in. IPO
valve will throttle to the closed position as the downstream pressure drops, even though the injection pressure at depth (Piod) is
equal to the opening pressure at depth (Pvot).
Operating Valve
The inflow/outflow analysis yielded the theoretical operating
depth and production rate. It is the purpose of spacing the unloading valves and sizing the ports to ensure that this operating depth
is reached. Typically, in continous-flow gas lift wells, the operating valve is an orifice. The use of the TC correlation to size the
port is reasonable for an orifice valve, even though the prediction
could be as much as 20 to 30% higher than the actual flow capacity.
The port in the orifice valve should be sized to pass the required
amount of gas with a minimum 343-kPa (50-psi) differential. The
orifice will be operating in the proportional flow range, and small
changes in either the upstream or the downstream pressure will
have a significant impact on the flow rate through the orifice. The
change in flow rate as a function of differential pressure decreases
when the differential pressure increases.
November 2008 SPE Production & Operations

Sizing the Unloading Valves


The current method of spacing the unloading valves as described
in RP 11V6 (1992) is valid. Many decisions are made for each
unloading valve, including the port size and the pressure drop to
apply for the next lower valve. Both of these decisions are functions of valve performance.
When the unloading process begins, it is assumed that the
transfer of liquid from the annulus to the production string will
cause the liquid level in the production string to rise to the surface
and flow into the collection system at the pressure of the collection
system. This may not always be the case, depending on the injectivity of the reservoir, the reservoir static pressure, and the depth of
liquid in the production string. We will assume the liquid rises to
the surface when the top valve is uncovered.
The surface injection pressure (Pinj) will be at kickoff pressure,
and the production pressure at the valve (Ptf) will be 138 kPa (20
psi) less than Piod when the first valve uncovers. For these conditions with a 1-in. IPO valve with 4.7 mm (12/64ths-in.) port, the
Ptro of the valve will be 7888 kPag (1,144 psig). Fig. 1 shows the
performance of this valve under these conditions for a range of
Piod pressures.
Note the maximum injection rate through the valve at the design Piod will be only 250 Mscf/D of gas. It was estimated that 19
825 m3/d (700 Mscf/D) of gas would be required to achieve a total
gas/liquid ratio (TGLR) of 1,200:1 ft3/bbl (213 m3/m3). The TC
charts indicate that a 4.7-mm (12/64ths-in.) port is more than sufficient to deliver the required rate of gas, but actual performance is
much lower.
An inspection of the flowing-pressure-gradient charts for this
condition will show, at this depth, that the difference in flowing
production pressure for a TGLR of 1,200:1 ft3/bbl (213 m3/m3)
and a TGLR of 200:1 ft3/bbl (35.6 m3/m3) is less than 345 kPa (50
psi). It was assumed that the flow rate would be 127 m3/d (800
BLPD), but this is only an estimate. Even if the flow rate were 127
m3/d (800 BLPD) of gas, the rate of gas required could be reduced
by a factor of 6 and the production pressure at the valve would be
only 345 kPa (50 psi) higher than anticipated.
Existing gas-lift-design methods assign a single flow rate to the
valve, regardless of production pressure. This would be valid if the
valve were in critical flow. Because this is rarely achieved with
1-in. IPO valves, the flow rate through the valve is changing constantly as the production pressure is dropping. Which rate should
be usedthe maximum, the average, or the rate at the transfer
production pressure? For purposes of picking a port size, the best
strategy is to take the average of the maximum flow rate and the
flow rate at the transfer production pressure.
In this case, at Piod8790 kPag (1,275 psig) and Pvot9080
kPag (1,317 psig), the maximum flow rate would be approximately
7080 m3/d (250 Mscf/D) of gas. Assume the flowing production
pressure at which transfer will take place to the next lower unloading valve is 2758 kPag (400 psig). The valve is closed at this
pressure. The average would be 3540 m3/d (125 Mscf/D) of gas.
The top valve with a 4.7-mm (12/64ths-in.) port at the design
injection pressure will flow much less than the design required,
but it will flow enough to reduce the flowing production pressure to within 345 kPa (50 psi) of the transfer pressure. The transfer pressure could be moved 345 kPa (50 psi) higher than the
final flowing-production pressure. This would be safe but will
465

Fig. 2Flow performance of second valve.

result in a closer spacing of the unloading valves and will not solve
the problem.
The wellhead-injection pressure (Pinj) during unloading will be
a function of valve set pressures (Pvot), surface gas-injection rate
(Qinj), and valve performance for each of the unloading valves. It
is recommended that the gas-injection rate at the surface be adjusted to the final injection rate when the first valve uncovers. If
this is practiced on this well, Pinj will rise to the kickoff pressure
8275 kPag (1,200 psig) and higher, if possible. The valve can flow
only 7080 m3/d (250 Mscf/D) of gas at maximum flow when at
least 19 825 to 21 240 m3/d (700 to 750 Mscf/D) of gas is being
injected at the surface. This is more than the valve can pass for the
pressure conditions. The result will be that Qinj will decrease to the
amount the valve can pass while Pinj increases. This would be a
near-stymie condition if not for the fact that unloading is continuing to occur.
The second valve is spaced by reducing Piod by 138 kPa (20
psi). The Pvot of the valve will be 9465 kPag (1,373 psig), with
Piod9000 kPag (1,305 psig). Fig. 2 shows the performance for
this valve for a range of Piod pressures. The Piod would have to rise
655 kPa (95 psi) higher than the design pressure for the valve to
pass 19 825 m3/d (700 Mscf/D) of gas. This is not possible.
The designer has several options: (1) Increase the port size of
the valve, (2) lower the set pressure of the valve, or (3) allow
for the wellhead-injection pressure to increase beyond the design pressure. Fig. 3 shows the performance of this valve for these
three options.
For the 4.7-mm (12/64ths-in.) ported valves, Piod was allowed
to increase 138 kPa (20 psi) higher than the design pressure for one
valve, and in the other, the Pvot was decreased 138 kPa (20 psi)
below the calculated Pvot. Note that the performance for these two
options is nearly the same. None of the options is capable of
flowing 19 825 m3/d (700 Mscf/D) of gas.
Using the 4.7-mm (12/64ths-in.) ported options means this
valve will be open and flowing approximately 4250 m3/d (150
Mscf/D) of gas when the third valve uncovers and begins passing
gas. If more than 8500 m3/d (300 Mscf/D) of gas is being injected

Fig. 4Remaining valves at design conditions.


466

Fig. 3Options for second valve.

at the surface, both the top valve and the second valve will remain
open when the third valve uncovers. This is not a problem as
long as these valves close when the operating valve is reached.
The 6.3-mm (16/64ths-in.) ported valve will flow slightly more gas
for the pressure conditions but still will not pass the desired rate
of gas.
Because the top two valves are at relatively shallow depths and
we do not want them to use all the available injection volume, the
best option is to use the 4.7-mm (12/64ths-in.) ported valve and
allow Piod to rise to the available wellhead pressure. This means
that both valves will be open and injecting and that Pinj will not
drop the expected 138 kPa (20 psi).
Fig. 4 shows the performance of unloading Valves 3 through 6
at the design conditions. As noted, none of the valves is able to
pass the required amount of gas. The performance of these valves
is shown assuming that the injection pressure would drop 138 kPa
(20 psig) at each unloading valve. This will not occur if the injection rate at the surface exceeds the combined gas-flow rate through
all the valves. Assuming 19 825 m3/d (700 Mscf/D) of gas is being
injected at the surface, the combined flow rate through all the
remaining valves will not equal the injection rate at the surface. As
a consequence, the actual injection pressure acting at these valves
will be much higher than anticipated and all valves could be open.
Fig. 5 shows the remaining valves with the 4.7-mm (12/64thsin.) port when full injection pressure is available. The assumption
is that the 138-kPa (20-psi) pressure drop did not occur at each
valve and Piod remained at or near the available wellhead-injection
pressure. The combined flow rate through the valves at these
higher injection pressures will be in excess of the injection rate
at the surface. The actual injection pressure will be less than available wellhead-injection pressure but not as low as was anticipated
in the design.
Fig. 6 shows the remaining valves with the 6.3-mm (16/64thsin.) port at design conditions. Once again, the combined injection
rate through these valves is less than required and is less than that
being injected at the surface. The Piod at each valve will be higher
than anticipated.

Fig. 5Remaining valves with 12/64ths-in. port and full annulus


pressure available.
November 2008 SPE Production & Operations

Fig. 6Remaining valves with 16/64ths-in. ports.

Fig. 7 shows the remaining valves with the 6.3-mm (16/64thsin.) port when full injection pressure is available. The combined
total of all valves is well in excess of the rate injected at the
surface, and, as a consequence, the Piod at each valve will be less
than available wellhead-injection pressure. It is possible that Pinj
could drop slightly for each valve as it uncovers, but these valves
will probably reopen multiple times.
Summary
If the desire is to achieve single-point sequential injection during
the unloading process with a discernable pressure drop at each
valve, then a 345-kPa (50-psi) pressure drop will have to be taken
at each unloading valve. However, the use of this much injectionpressure drop at each unloading valve has the undesirable effect of
limiting the depth of injection. The other alternative is to use a
kickoff pressure that is at least 345 kPa (50 psi) less than the
available wellhead-injection pressure. This will have the effect of
reducing the Ptro of each valve and will allow Piod at each valve to
rise higher than the Pvot of the valve sufficiently to allow the
desired gas passage.
The main benefit of sequential single-point injection during
unloading is to record the Pinj pressure drop as each valve begins
injection. This is a valuable troubleshooting tool and allows the
operator to determine the point of injection. However, if this
troubleshooting tool is not used during unloading, designing for
single-point sequential injection has little value.
In the above example, use of the 4.7-mm (12/64ths-in.) ports at
the top two valves is justified. At shallow depths, the amount of
gas required to achieve the objective production flowing pressure
can be considerably less than the design anticipates with only a
small increase in production pressure. At greater depths, this is not
the case
The remaining unloading valves should use 6.3-mm (16/64thsin.) ports. Even with larger ports, the gas flow rate through the
valves is less than that required at anticipated injection pressure. Pinj will rise, and multiple valves will be open during unloading; these valves could close and reopen many times. As injection takes place at greater depths, the well will begin inflow.
The reservoir gas/oil ratio will eventually help to provide the
additional gas required to achieve the objective production flowing pressure, and the rise in temperature will help to close the
upper valves.
Conclusions
The use of the TC equations and charts for 1-in. IPO valves will
result in gas-passage predictions that could be as much as 3 to 4
times the amount the valve actually can pass.

November 2008 SPE Production & Operations

Fig. 7Remaining valves with 16/64ths-in. port and full annulus


pressure available.

Every IPO valve has a property called loadrate. Because of this


property, IPO valves do not become fully open when injection
pressure reaches opening pressure. It is recommended that the
kickoff pressure used for designing with 1-in. IPO valves be at
least 345 kPa (50 psi) less than available pressure.
The design technique leads the engineer to believe that singlepoint sequential injection will occur during unloading. Depending
on valve performance, this may not be the case. Multiple valves
could be injecting and could close and reopen many times.
Multipoint injection during unloading is not a problem as long
as the unloading valves close and remain closed when the operating valve is reached. This can be assured if Piod pressure drops are
taken at each valve.
Nomenclature
Pcf
Pinj
Piod
Ptf
Ptro
Pvot
Qinj

annulus pressure at depth


injection pressure at wellhead
injection pressure at depth
flowing production pressure at depth
opening pressure at reference temperature
valve opening pressure at depth
gas flow rate

Acknowledgments
The author wishes to thank the members of the VPC for their
support in acquiring the data needed to provide the correlations
and graphs used in this paper.
References
Cook, H.L. and Dotterweich, F.H. 1946. Report on the Calibration of
Positive Flow Beans Manufactured by Thornhill-Craver Company,
Inc., Houston, Texas, Aug. 10, 1946. Kingsville, Texas: Dept. of Engineering, Texas College of Arts and Industries.
RP 11V6, Recommended Practice for Design of Continuous Flow Gas Lift
Installations Using Injection Pressure Operated Valves. 1992. Washington, DC: API.

Ken Decker is CEO of Decker Technology, Inc. His company


specializes in consulting for gas lift, marketing the AVT-II, and
administrating the VPC. Decker has published several SPE
papers during his 27-year career and continues to participate
and support the Gaslift Workshop.

467

Das könnte Ihnen auch gefallen