Sie sind auf Seite 1von 7

Gas-Lift Valve Performance Testing

and Data Correlation


K.L. Decker, * SPE, Otis Engineering Corp.
Summary. This paper presents a method for analyzing data from gas-lift valve performance tests. This method relies on the ability
to determine the true dynamic stem position of the valve during flowing conditions. The types oftests required to obtain the data needed
by this analysis method are explained. The data analysis method is applicable to any pressure-operated gas-lift valve and yields predic-
tions within 10 % of actual flow rate.
, Introduction
Many criteria can be used to quantify valve performance. Temper-
ature sensitivity, corrosion and erosion resistance, stability of set
pressure, and vibration suppression are just a few. However, in
the context of this paper, gas-lift valve performance is defined as
the quantitative measure of a valve's flow rate to changes in casing
or tubing pressure for a given set pressure.
Accurate gas-lift valve performance data are requirements of the
valve spacing and port-sizing techniques. Some current spacing and
port-sizing techniques assume that, if the valve is open, it is full-
open. However, this is not always true. For most unloading and
lifting conditions, the port is considerably less than full-open; con-
sequently, the flow passage through the valve is considerably less
than anticipated.
The full-open assumption had allowed a simple method for cal-
culating flow passage through the valve-Le., the Thornhill-Craver
equation and charts. 1 The Thornhill-Craver equation was devel-
oped for full-open square-edge ports. A gas-lift valve rarely has
a full-open port; consequently, the gas flow pattern past the stem
and seat is considerably different than the flow through a square-
edge port. The gas flow pattern past the stem and seat of a gas-lift
valve resembles that of an annular venturi because the port is ob-
structed by the centrally located stem. The combination of reduced
stem travel and flow pattern significantly affect the performance
of a gas-lift valve and renders the use of the Thornl1ill-Craver equa-
tion and charts of little value in determining flow rate through a
gas-lift valve.
The need for accurate gas-lift valve performance data has been
pointed out in the literature both recently2,3 and in the past. 4,5 A
few gas-lift companies have responded to this need by beginning
tests to determine the flow characteristics of a gas-lift valve. 6
Researchers in an industry-sponsored research program at the U.
of Tulsa (TUALP) have been working since 1983 to clarify gas-
lift valve performance. This work thus far has resulted in several
papers that describe testing 7 and analysis techniques. 8,9
Because of the complicated nature of the flow patterns through
the valve, actual valve tests under simulated operating conditions
are required to quantify valve performance. This paper describes
three types of tests required to obtain the data needed to analyze
valve performance: load-rate, flow-capacity, and live-valve tests.
The basic goal of this analysis method is to determine the valve's
true dynamic stem position during flowing conditions.
Theory of Data Analysis
A gas-lift valve is a mechanical device designed to open and close
a port in response to pressure (Fig. 1). The valve stem is held on
a port by a preset load that may be either a nitrogen charge or a
spring. When the force of casing or tubing pressure acting on the
bellows exceeds this preset load, the valve stem begins to move
away from the seat. The amount of stem movement is a function
of a force balance on the bellows. The ability to predict this stem
position with consistent accuracy is directly related to the ability
to predict the pressures acting on the stem.
The flow pattern through the gas-lift valve resembles that of a
variable, annular venturi. The position of the valve stem with respect
Now a consultant.
Copyright 1993 Society of Petroleum Engineers
SPE Production & Facilities, May 1993
to the seat affects the pressures acting on the stem; conversely, the
pressures acting on the stem affect its position with respect to the
seat. Modeling this complex interaction is difficult. 10 The simpli-
fied case of tubing pressure acting on a stem area that is equivalent
to the port area is untrue. The magnitude of the tubing pressure
and of the areas worked upon is variable along the flow path be-
tween the stem and the seat.
As is true with any device that restricts flow, the magnitude of
the pressure field downstream of the restriction is less than the ap-
plied downstream pressure. Depending on the geometry and the
pressures applied, this reduced pressure field will recover to full
downstream pressure as the gas travels farther from the restriction.
For a gas-lift valve, this reduced pressure field exists at the sur-
face of the valve stem.
Another factor affecting the performance of a gas-lift valve is
the pressure ratio at which choking begins. With a simple square-
edge orifice, the prediction of the onset of choking is simple; with
an annular venturi, however, this prediction becomes much more
difficult. The pressure ratio at which choking occurs is a function
not only of the flow geometry of the stem and seat but also of the
travel distance of the stem with respect to the seat.
The difficulty of predicting the flow performance of a gas-lift
valve reduces to the problem of accurately predicting the dynamic
position of the valve stem. The force-balance equations used for
the last 40 years to calculate opening and set pressures are defined
in this paper as static-force-balance equations. These equations were
written to describe the forces acting on the bellows when the valve
stem is on the seat. Of course, this implies that flow is not occur-
ring and that the pressures acting on the bellows are nearly con-
stant. Application of the static-foree-balance equation to determine
the dynamic stem position during flowing conditions is inap-
propriate.
The foregoing discussion was included not to advocate the direct
solution of the dynamic-force-balance equation but to point out the
inadequacies of using the static-force-balance equation to determine
a dynamic condition. The chief problem with a dynamic-force-
balance equation is that determining the variable pressure field acting
on the stem is very difficult and requires at least a computer solution.
The dynamic stem position can be inferred by an analysis method
that uses data from three types of tests. First, the valve must be
tested for a load rate. The load-rate test determines both the maxi-
mum effective travel of the valve stem and the incremental amount
of force required to move the stem. The second test is performed
to determine the valve flow capacity. The test is per-
formed according to Instrument Soc. of America (ISA) standards 11
and yields a curve of the flow coefficient, C
v
, and critical-pressure-
ratio factor, Rep, as a function of the valve stem travel. The final
tests are the live-valve tests, which are conducted with a live gas-
lift valve operating under flowing conditions similar to those an-
ticipated downhole.
At each flowing pressure condition during the live-valve test, the
valve stem will seek a position so that the forces acting on the bel-
lows are in equilibrium. For each of those conditions, the flow
coefficient of the valve can be calculated by
C
v
= VI{ gZ)] 0.5 }. . ............... (1)
101
BELLOWS
LOAD SPRINGS
CHOKE PORTS
MAIN VALVE
AND SEAT
BACKCHECK
Fig. 1-Productlon-pressure-operated gas-11ft valve.
Knowing the flow coefficient at each of these flowing conditions .
and the relationship between flow coefficient and stem travel al-
lows one to infer the dynamic stem position of the valve during
flowing conditions. This analysis method allows one to determine
the dynamic stem position without knowing the details of the pres-
sure field surrounding the valve stem.
This information could be used to determine the opening forces
acting on the bellows:
Fopen =Fclose +kL
d
,
where Fopen = sum of forces acting to open the valve. These forces
include the casing pressure acting on some area of the bellows and
the average of the downstream pressure field acting on the stem.
Fclose is the sum of the forces acting to close the valve. For a
nitrogen-charged valve, this force would be equal to the charge pres-
102
:J
" In
In
:J
..
f<
..
In
LOAD RATE TEST

VALVE TYPE: 222RF1500
ALL TEST PRESSURES
300
000
o
I 200
:J
"
In
In
:J
..
f<
III
..
f<

"
..
100
o
2400 PSIG/INCH

0.00 0.02 0.04 0.08 0.10' 0.12 0.14 0.16
STEM TRAVEL
Fig. 2-Typical plot of load-rate test data.
Cv AND Xt DATA ANALYSIS
1.0..--------------------,
VALVE TYPBs 222RP'lSOO
CBOU SIZBI 1/'"
0.9
STEM TRAVELI 0.150
O.B
+51 LIMIT
0.7
0.6
.....-- ytev =O.667*Cv
Pressure Drop Ratio Factor I
0.5
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 O.B
PRESSURE RATIO (dp/p')
Fig. 3-Analysis of flow-capacity test data.
sure times the effective area of the bellows. For a spring-loaded
valve, this force would be equal to the preset load in the spring.
This equation is not particularly useful because it contains no vari-
ables to account for specific pressure conditions in the casing or
the tubing.
The analysis method used in this paper finds the ratio of dynam-
ic to static stem travel, LdlLs' and plots this function vs. the ratio
of differential pressure to pressure spread, Pdlps' These two
dimensionless ratios were selected after trial-and-error inspection
of hundreds of data sets. The correlation between these two ratios
is somewhat sensitive to casing pressure; as a result, additional
modification is required.
The basic premise of this analysis method is that the flow-path
geometry through the valve is the major factor affecting perform-
ance. The flow-path geometry is the result of valve mechanics and
pressure conditions. Analysis of the test data attempts to find a cor-
relation between the valve flow performance and valve mechanics.
Description of Tests
Load-Rate Test. The load-rate test is conducted to determine both
the spring rate and the maximum effective travel ofthe valve. This
test is conducted by applying pressure on the full area of the bel-
lows and measuring the stem displacement as a function of pres-
sure. Several devices capable of performing this test have been
described,12,13 so they are not discussed here.
The load-rate test is performed first by charging or setting the
valve to a known opening pressure. A test pressure somewhat greater
than the set pressure then is applied across the full area of the bel-
SPE Production & Facilities, May 1993
Fig. 4-Typical test apparatus.
lows, and the stem displacement is measured with a micrometer.
The test pressure then is increased by a nominal amount, and the
stem position again is measured. This procedure is continued until
the incremental stem movement decreases significantly when the
incremental pressure increases. When this condition is reached, the
stem has reached its maximum effective travel. The test pressures
then are plotted as a function of the stem inovement on linear coor-
dinate paper. The resulting plot is usually a straight line that
originates at the opening pressure and having a positive slope. The
valve load rate then is computed as the slope of this line (given
in psig/in.).
When the valve reaches its maximum effective travel, the slope
of this line increases sharply. Travel beyond this point, for all prac-
tical purposes, is not available during normal valve operation.
This test also can be performed with the valve stem at its maxi-
mum effective travel and with pressures decreased. When this is
done, the plot of the test pressure vs. stem travel will not follow
the same plot created when pressures were increased because of
bellows hysteresis (not discussed here). Even though these plots
are displaced from each other, their slopes are very similar. The
slope of the line is of primary interest; therefore, either test method
yields usable results. Fig. 2 is a typical plot of the load-rate test data.
Load-rate tests performed on nitrogen-charged valves will be de-
pendent on charge pressure. The higher the charge pressure, the
higher the load rate. Spring-loaded valves have a nearly constant
load rate, regardless of the set pressure. Spring-loaded valves gener-
ally have a much higher load rate than nitrogen-charged valves do.
The maximum effective travel of a valve defines that point at
which stem travel is no longer proportional to pressure increases.
This maximum effective travel is less than 0.200 in. for most I-in.
valves, and usually less than 0.300 in. for 1 V2-in. valves. A valve
with a greater maximum effective travel would not necessarily have
a greater flow capacity. The amount of stem travel required to get
a full-open port usually is less than the maximum effective travel.
Flow-Capacity Test. The second test determines the gas-lift valve
flow capacity, C
v
, and critical-pressure-ratio factor, Rep' as func-
tions of the stem displacement. The valve flow capacity is a meas-
SPE Production & Facilities, May 1993
ure of its resistance to flow through a specified restriction. Rep
defines the maximum pressure ratio, Pd/P', at which the valve can
be operated without choking. If the actual pressure ratio exceeds'
Rep' the valve is choked and flow is no longer proportional to Pd'
The Pd is measured with a differential-pressure meter at points re-
mote from the valve and seat. The upstream pressure is measured
on the pressure fixture at a point about 15 in. upstream of the test
valve. There are no significant restrictions between this point and
the test valve inlet ports. The downstream pressure is measured
on the pressure fixture at a point about 10 in. downstream of the
test valve. Again, no significant restrictions exist between the meas-
urement point and the test valve.
This test is performed according to the ISA standard developed
to determine flow capacities of variable-geometry valves. A test
valve is modified to include a stem that can be positioned accurate-
ly with respect to the seat without affecting the internal flow ge-
ometry of the valve. This usually is accomplished by replacing the
bellows with a threaded stem that can be positioned positively and
locked during the test. This test valve is then placed in a fixture
that will cause gas to flow through the valve in a manner similar
to that expected downhole. Several differential pressures are ap-
plied across the valve, and pressures and temperatures are recorded.
The data then are analyzed by plotting the product of flow coeffi-
cient and expansion factor, CvY, as a function of the pressure ra-
tio Pd/P'. A straight line is then fitted to this plot. The y axis
intercept is read as the valve's C
v
for that travel position. The crit-
ical pressure ratio factor, Rep' is found by (1) multiplying the C
v
by 0.667, (2) projecting a horizontal line from this value to the fit-
ted line, (3) projecting a vertical line from this intersection down
to the x axis, and (4) reading the Rep on the x axis. Fig. 3 shows
how this method would be used for a typical set of test data.
This procedure is then repeated for another stem position. Enough
tests are conducted that C
v
and Rep can be determined for the full
range of effective stem travel. The result of this test is a plot of
C
v
and Rep as functions of the stem travel.
C
v
initially rises very sharply with small amounts of travel, then
levels out and remains fairly constant with increased amounts of
103
FLOW CAPACITY DATA
2.50,.--------------------,-
2.25
2.00
1.75
1.50
1.25
1.00
0.75
0.50
0.25
VALVE TYPE: 222RF1500
CHOKE SIZE: 1 /4"
FLOW COEFFICIENT (C v)
"
/
CRITICA'L PRESSURE RATIO (Rep)
0.00
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16
STEM TRAVEL
fr

0
1=
-<
"


g;
:;!
u
1=
iii
u
Fig. 5-Flow coefficient and crltical-pressure-ratlo test data.
stem travel. The point at which the flow coefficient begins to re-
main fairly constant defines the maximum travel where increases
in stem travel will yield increases in flow capacity. This point does
not necessarily coincide with the maximum effective travel. In fact,
for small ports, the maximum C
v
usually is reached well before
the valve's maximum effective travel is reached.
Live-Valve Test. The final test is the live-valve test. This test is
performed on a live valve placed in a fixture simulating a sidepocket
mandrel. Additional fixturing is required to control and measure
the pressures and gas flow rates through the valve. Fig. 4 shows
a typical test apparatus.
Live-valve tests are conducted in one of two manners. The up-
stream pressure (casing pressure) can be held nearly constant while
the downstream pressure (tubing pressure) is dropped, or the down-
stream pressure can be held constant while the upstream pressure
is changed. Either method will cause the valve stem to move in
response to pressures and flow rates.
In the test with the constant casing pressure, the casing and tub-
ing pressures are equalized at some value less than the tested Plro
of the valve. The casing pressure is held constant, and the tubing
pressure is dropped in 25-psig increments while pressures and flow
rates are measured. Tubing pressures continue to be decreased un-
til either the valve closes or flow rates through the valve are no
longer proportional to differential pressure. This series of tests is
then repeated with the valve set at a different P Iro' The data from
these tests are then plotted as flow rate vs. tubing pressure.
The tests with constant tubing pressure are conducted by equaliz-
ing casing and tubing pressure at some value less than Plro and then
increasing casing pressure while maintaining constant tubing pres-
sure. Pressures and flow rates are recorded. As the casing pres-
sure is increased in small increments, the flow rate increases by
very large amounts. This test is repeated with the tubing pressure
held at a different value. The test data are then plotted with casing
pressure as a function of flow rate. This series oftests also is repeat-
ed with the valve adjusted to a different set pressure.
Correlation Development
The tests described above were performed on a 1 'h-in. production-
pressure-operated gas-lift valve (Fig. 1). The closing force is sup-
plied by a spring; therefore, the load rate is not sensitive to tem-
perature or pressure. The flow rate through the valve is controlled
by chokes upstream of the main valve and seat. The chokes range
in size from Vs to % in. in Y32-in. increments, and the main valve
and seat remain fixed for all choke sizes. The port diameter is 0.430
in., and the valve stem seals on this port at a diameter of 0.450
in. This valve exhibits a 19% casing sensitivity and an 81 % tubing
sensitivity regardless of choke size.
104
LIVE VALVE TESTS - CONSTANT Pcf METHOD
1000
900
VALVE TYPE: 222RF1500
CHOKE SIZE: 1/4"

800
PYd=829
700

Q
V 'V'V Pcf=90S
600
.jPOOoO '"
'"

t:. 500
Q 6. AD. 'V
'"
400 6 6 '"
'-'
'"
'!!aL::.OODO 6. 0
::>l
300
0 0 '"
.SJ 0 '"
..96lO 0 6 0
200
0
I!i eO
0
'"
100
6
'l",rq 6'"
0
600 700 800 900 1000 1100
FLOWING TUBING PRESSURE
Fig. 6-Typical live-valve test data from constant-easing-
pressure test.
The valve load rate was found to be about 2,400 psig/in. The
valve spring rate can be found by mUltiplying the load rate by the
effective bellows area. For the example valve, the spring rate was
computed to be about 550 Ibf/in. The maximum effective travel
of the valve was found to be roughly 0.190 in.
The C
v
and Rep were tested for Vs-, }k, IA-, and %-in. chokes.
Fig. 5 shows the results of the test of the lA-in. choke. A least-
squares curve fit was performed to find the coefficients of a third-
order polynomial equation that would describe C
v
and Rep func-
tions of the stem travel.
As noted, the valve reached its maximum flow capacity before
reaching its maximum effective travel. The curve-fit coefficients
were weighted to concentrate only on fitting the data up to the point
of maximum C
v
' For travel beyond this maximum, the C
v
was
considered constant. The range of the equations that describe C
v
and Rep as functions of stem travel is limited by this maximum
travel limit. The C
v
and Rep ratio equations are
Cv=AJLJ +A
2
LJ+A
3
L
d
+A
4
....................... (2)
and Rep=A5LJ +A
6
LJ +A7Ld+Ag . .................... (3)
Twenty-eight live-valve tests were performed with the valve set
at two different pressures and with four choke sizes. The live-valve
tests were performed with the constant-casing-pressure method be-
cause the valve was most sensitive to tubing pressure. Fig. 6 is a
plot of several live-valve tests for the 'A-in. choke.
At each recorded data point of the live-valve test, the static stem
travel was computed from
Ls =[0.193(Pif -Pvel )+0.037Pcf ]/k . .................. (4)
The numerical coefficients in this equation are specific to the de-
sign of the valve tested. These coefficients remain constant, regard-
less of the size choke tested. The static-force-balance equation for
an injection-pressure-operated valve would be different and would
vary with the port size. Whichever valve is tested, the static-force-
balance equation must consider the physical valve design and must
be written for the case of the stem on the seat with no flow.
The live-valve closing pressure, Pvcl, is the tubing pressure at
which the valve closes for the current casing pressure. The tested
valve has a small amount of casing sensitivity; consequently, the
live-valve closing pressure changes as a function of casing pres-
sure. For this valve, the live-valve closing pressure is
Pvcl=Pvet-0.192Pej' ............................... (5)
The pressure spread, P s' of the valve is the difference between
the casing pressure and the live-valve closing pressure:
Ps =Pej-Pvcl' .................................... (6)
The correlation pressure ratio, Reop is
Reor=Pd/Ps' ...................................... (7)
SPE Production & Facilities, May 1993
oJ
>

TRAVEL RATIO vs PRESSURE RATIO
1. 4 r--------------------------------------,
VALVE TYPE: 222RF1500
1.3 CHOKE SIZE: 1/4"
PYd =829
1.2 o 0 0
o 0 0
o
0.7
0.5
o
o
o
0.4L-__ L-__ L-__ L-__ L-__ L-__ L-__ L-__ L-__
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
PRESSURE RATIO (Pdips)
Fig. 7-Plot of correlation function for V4-in. choke.
The flowing pressure ratio, Rp is calculated as
...................................... (8)
As indicated earlier, the static stem travel is not the same as the
dynamic stem travel. The latter is inferred by use of the live-valve
test data to calculate C
v
for each of the pressure conditions with
Eq. 1. The expansion factor in Eq. I is calculated with
y= 1-Rpl(3FkRep). . ............................... (9)
The equation for the expansion factor includes a term for Rep.
At this point in the analysis, Rep for the dynamic stem position is
unknown; however, a very good estimate of the critical pressure
ratio factor can be made by use of the static stem travel in Eq. 4.
This estimate introduces an error of about 3 % to 4 % .
The dynamic stem travel is then found by solving Eq. 2 for L
d

In some cases, the computed flow coefficient from the live-valve
test is greater than the maximum C
v
obtained during the flow-
coefficient test because of data-collection errors and curve-fitting
approximations. In these cases, the dynamic stem travel is indeter-
minate. This correlation method assumes for these cases that the
dynamic stem travel and static stem travel are equal. This assump-
tion is false. However, for predicting a flow rate, the amount of
stem travel is not a factor when the flow coefficient is at its
maximum.
The goal of the analysis method is to find a correlation that pre-
dicts the dynamic stem travel as a function of known variables.
Preferably, the correlation includes known variables that have an
;;
t..
t.>
'"

OJ
....
""
0:

0
....l
t..
ACTUAL vs COMPUTED FLOW RATE
1000
VALVE TYPE: 222RF1500 '" COMPUTED
900 CHOKE SIZE: 1/4"
PVC1=829
800
700

0<il 00 600
... <WI!> Oy
500
o.
Qo o.
400
ii>9o '"' I!l 0(5
<il @ 15
300
Q
o
@<il <il
200
'"'
Q
9
100

0
.
__ L-__ L-__ L-__ L-__ L-__ L-__ L-__
600 650 700 750 800 850 900 950 1000 1050 1100
PRODUCTION PRESSURE (PSIG)
Fig. 9-Comparison of actual and computed flow rates.
SPE Production & Facilities, May 1993
MODIFIED TRAVEL RATIO vs PRESSURE RATIO
1.0 F'-"---------------------------,
VALVE TYPE: 222RF1500
0.9 CHOKE SIZE: 1/4"
PVc! =829
0.8
o
o
0.7
o
0.2
0.1
0.0 L-__ L-__ L-__ L-__ L-__ L-__ L-__ L-__ L-__ L-..--l
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
PRESSURE RATIO (NIps)
Fig. 8-Modified correlation ratio for V4-in. choke.
influence on valve flow performance. A function found to be rep-
resentative of valve dynamics over the full pressure range is
LdlLs =f(R
eor
)' .................................. (10)
Fig. 7 is a plot of this function for the example valve. The data
in this plot include all the live-valve test points for the lA-in. choke.
This plot function requires an additional modifier to account for
the small amount of casing sensitivity. For the data points where
the dynamic stem travel has been computed to be less than the travel
required to achieve maximum flow capacity, the travel ratio
(LdILs) is modified as follows:
R
tm
=(L
d
IL
s
)(P
s
IPvet)O.5 . .......................... (11)
This correlation function is then defined by performing a least-
squares curve fit to the data. Several forms of equations are used
to describe this function. The equations that best fit the data for
the example valve are
Rtm=AIReor+A2 ................................. (12)
for Reor <Rpmax and
Rtm=A3R2or+A4Reor+AS ......................... (13)
for Reor > Rpmax. Fig. 8 shows a plot of these equations, along with
the data for the lA-in. choke of the example valve.
Performance Calculation Procedure
The performance calculation procedure requires the following in-
formation: static-force-balance equation for the subject valve, tested
0:
o
0:
0:
OJ
....
Z
OJ
U
0:
PERCENT FLOW RATE ERROR
20 r----------------------------------------,
15
10
5
o
VALVE TYPE: 222RF1500
CHOKE SIZE: 1/4"
pvt:!""829 +10% ERROR
.1-
o
-5 0
-10 0
-15
_ _L _ _L _
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
PRESSURE RATIO (Pd/ps)
Fig. 10-Percent error of actual and computed flow rates.
105
TABLE 1-STATISTICAL ACCURACY OF THE CORRELATION
Choke size
Arithmetic Mean Standard Deviation Standard Error 95% Confidence 99% Confidence
(%) Error (%) Difference Error (%) Difference Error (%) Difference Error (%) Difference Error (%) Difference
Va -6.30 -12.10 17.40 18.7 1.23 1.32 2.42 2.60 3.20 3.43
3/16 -3.43 -11.70 15.30 28.5 1.12 2.08 2.20 4.10 2.90 5.41
V4 -3.58 -7.57 10.30 17.7 1.04 1.78 2.07 3.54 2.74 4.69
0/16 -11.47 -33.40 18.10 29.0 1.93 3.10 3.84 6.15 5.08 8.15
load rate for the subject valve, C
v
and Rep curve as a function of
travel, maximum C
v
travel for the subject valve, and equations of
correlation function.
Step 1. Solve the static-force-balance equation for stem travel
(Eq. 4) for the valve at the pressure conditions expected downhole.
Step 2. Find the live-valve closing pressure (Eq. 5), the valve
pressure spread (Eq. 6), and the correlation pressure ratio (Eq. 7)
for the pressure conditions expected downhole.
Step 3. Compute the modified travel ratio, R
tm
, as a function of
the correlation pressure ratio (Eqs. 12 and 13).
Step 4. Compute the dynamic stem travel as follows:
Ld=RtmLsf(PslPvet )0.5. . .......................... (14)
Step 5. Using the dynamic stem travel, compute C
v
and Rep
(Eqs. 2 and 3).
The Rep defines the maximum pressure ratio for which the flow
through the valve is proportional to differential pressure. If
Rp >R
ep
' then choked flow is occurring, and Rep must be used to
compute flow. Rp is computed as follows:
Rp =(Pcj-Ptj )/(Pef+ 14.7). . ...................... (15)
Step 6. Compute the flow rate for the given pressure conditions
and the specific valve and choke sizes using Eq. 9 to calculate the
expansion factor:
VMscflD gIZ)0.5 . ................ (16)
Accuracy
The performance calculation procedure was used to calculate gas
flow rate for all the tested choke sizes at the same pressure condi-
tions used during the live-valve test. Fig. 9 is a comparison plot
of the flow-performance data from an actual test and for the same
conditions for the 'A-in. choke.
The accuracy measurement must include a comparison of not only
the maximum flow through the valve but also of the "actual" vs.
"computed" flow over the full range of valve operation. The pres-
sure conditions at each live-valve test point were used to compute
a flow rate. The difference between the computed and the actual
flow rates then were plotted as the percent error in flow rate as
a function of the correlation pressure ratio, R
eor
' Fig. 10 is a graph
FLOW RATE COMPARISONS
600
o CORRELATION
500
THORNHILL-CRAVER PARTIAL
;:;-
r..
0 u
400
0
rn
0
6 0
0
0
'"
0
...
300
."
0::
0 0






0
200
0

...l
r..
0

0

100

0
0

i

0
750 800 850 900 950
FLOWING TUBING PRESSURE
Fig. 11-Comparison of this correlation to the Thornhill-Craver
equation.
106
of this plot for the example valve with lA-in. choke. Table 1 gives
a statistical comparison of the actual vs. computed flow rates for
all choke sizes.
The correlation accuracy has not been confirmed at pressures
> 1,500 psig; however, because of the method of development,
the correlation is assumed accurate at these pressures and beyond.
Comments on Correlation Development
The need to modify the travel ratio by Eq. 11 indicates that the
correlation function identified in Eq. 10 does not account for all
the flow phenomena during dynamic conditions. This analysis tech-
nique, however, does address the issue of paramount importance:
that the flow performance of a valve is a function of valve mechan-
ics. The best way to predict this performance is to include valve
mechanics in the model. The analysis method does not presuppose
a type of flow performance curve but instead uses the actual valve
performance to determine the dynamic stem position.
Comparison With the ThornhlllCraver Equation
The Thornhill-Craver equation was used to compute flow rates for
the same conditions (Fig. 9). A coefficient of discharge of 0.98
and a correction factor of 0.865 were used. Fig. 11 is a plot of
valve performance from this correlation and the Thornhill-Craver
results, assuming a partially restricted port. The amount of restric-
tion was determined by calculating the flow area between the stem
and seat for the computed static stem travel of the valve.
The Thornhill-Craver flow prediction with a full-open port greatly
exceeded actual performance. The Thornhill-Craver flow predic-
tion with partial restriction was closer to actual but still more than
20% in error over the full range of application.
Scope of Application
The correlation given here was developed with the test data ob-
tained from a production-pressure-operated valve; however, this
same analysis technique also can be used with an injection-pressure-
operated valve. To use the technique with an injection-pres sure-
operated valve, modify the static-force-balance equation and sub-
stitute the appropriate variables for load rate, flow coefficient, and
critical pressure ratio. Data obtained from TUALP on an injection-
pressure-operated valve were analyzed with this correlation tech-
nique and produced results similar to those reported for the
production-pressure-operated valve.
Conclusions
1. The flow-rate performance of any gas-lift valve can be deter-
mined for its full range of operation by performing three types of
tests and then using the analysis method described here to develop
the correlation.
2. The analysis method includes factors to account for valve
mechanics and thus is responsive to changes in valve characteristics.
3. The computed flow performance curves are accurate over the
entire range of valve operation, not just for the range of maximum
flow but also for the range of operation when the valve stem is ob-
structing the flow.
4. With the inclusion of the appropriate static-force-balance equa-
tion, the analysis method can be used for any type of valve.
5. Using this test method and correlation technique yields a sim-
ple set of equations that can be computer-programmed to calculate
valve performance for any operating conditions.
SPE Production & Facilities, May 1993
Nomenclature
Ai = curve-fit coefficients
C
v
= flow capacity per ISA test procedure
F = force acting to open a valve, Ibf
Fk = ratio of specific heats
k = spring rate, Ibf/in.
Ld = dynamic stem travel, in.
L
s
= static stem travel, in.
P' = absolute pressure, psia
P cj = casing flowing pressure, psig
~ j = casing flowing absolute pressure, psia
Pd = differential pressure, psig
P
s
= pressure spread, Pcj-Pvcl
Ptf = tubing flowing pressure, psig
Ptro = test-rack opening pressure, psig
Pvc/ = live-valve closing pressure, psig
PVC! = tested-valve closing pressure, psig
Rcor = correlation pressure ratio
Rep = critical pressure ratio
Rp = pressure ratio (Pdlp') per ISA standard
Rpmax = experimentally determined pressure ratio
R
tm
= modified travel ratio
T = temperature, OR
V = volume of flow, scf/hr
VMscf/D
= volume of flow, Mscf/D
Y = expansion factor
z = compressibility factor
'Y g = specific gravity of gas
Acknowledgments
I thank Otis Engineering Corp., a Halliburton company, for the
support and opportunity to perform and develop the analysis proce-
dure. Special thanks go to James Scott for his assistance during
the fabrication of the Gas Flow Test Stand apparatus.
References
1. Cook, H.L. Jr. and Dotterweich, F.H.: "Report on the Calibration of
Positive Flow Beans as Manufactured by Thornhill-Craver Company,"
final report, Texas A&I U., Kingsville (1946).
2. Laing, C.M.: "Gas Lift Design and Production Optimization Offshore
Trinidad," SPEPE (May 1989) 135-41.
3. Laing, C.M.: "Gas-Lift Design and Performance Analysis in the North
West Hutton Field," JPT (Jan. 1991) 96-102; Trans., AIME, 291.
4. Neely, A.B. and Vogel, J.V.: "A Field Test and Analytical Study of
Intermittent Gas Lift," paper SPE 4538 presented at the 1973 SPE An-
nual Meeting, Las Vegas, Sept. 30-0ct. 3.
SPE Production & Facilities, May 1993
Author
Kenneth L. Decker currently works as
a consultant. He was a member of the Ar-
tificial Lift Group of the R&D Dept. at Otis
Engineering Corp. In Dallas. He began
his career in the petroleum industry In
1980 as a project engineer for Teledyne
Merla, designing gas-11ft eqUipment. In
1986, he joined OtiS, again working
primarily in gas-lift equipment. Decker Is
chairman of an API work group that is
writing the Gas-Lift Valve Performance
Testing recommended procedure.
5. Steele, R.D.: "Application and Economics of Artificial Lift in the Judy
Creek Field, Alberta," paper SPE 6043 presented at the 1976 SPE An-
nual Technical Conference and Exhibition, New Orleans, Oct. 3-6.
6. DeMoss, E.E. and Tiemann, W.D.: "Gas-Lift Increases High Volume
Production From Claymore Field," paper SPE 9214 presented at the
1980 SPE Annual Technical Conference and Exhibition, Dallas, Sept.
21-24.
7. Nieberding, M.A. eta/.: "Normalization of Nitrogen-Loaded Gas-Lift-
Valve Performance Data," paper SPE 20673 presented at the 1990 SPE
Annual Technical Conference and Exhibition, New Orleans, Sept.
23-26.
8. Hepguler, G. et al. : "Dynamic Model of Gas-Lift Valve Performance, "
paper SPE 21637 presented at the 1991 SPE Annual Technical Con-
ference and Exhibition, Dallas, Oct. 6-9.
9. Acuna, H. et a/.: Modeling of Gas Lift Valve Performance Using the
Test Data, U. of Tulsa, TUALP (April 9, 1991).
10. Decker, K.L.: "Computer Modeling of Gas Lift Valve Performance,"
paper OTC 5246 presented at the 1986 Offshore Technology Confer-
ence, Houston, May 5-8.
11. Standard ISA-S75.02, Control Valve Test Procedure, ISA, Research
Triangle Park, NC (1975).
12. Eads, P.: "Load Rate Testing Device," paper presented at the 1990
ASME Gas Lift Workshop, New Orleans (Jan.).
13. Winkler, H.W. and Camp, G.F.: "Dynamic Performance Testing of
Single-Element Unbalanced Gas-Lift Valves," SPEPE (Aug. 1987)
183-90.
51 Metric Conversion Factors
OF (OF-32)/1.8
in. x 2.54*
Ibf x 4.448222
psi x 6.894757
OR OR 5/9
Conversion factor is exact.
E+OO
E+OO
E+OO
C
cm
N
kPa
K
SPEPF
Original SPE manuscript received for review Oct. 6, 1991. Revised manuscript received
June 22, 1992. Paper accepted for publication Jan. 4, 1993. Paper (SPE 22789) first present
ed at the 1991 SPE Annual Technical Conference and Exhibition held in Dallas, Oct. 6-9.
107

Das könnte Ihnen auch gefallen