You are on page 1of 64

PRESSURE DROP AND CLOSURE FORCES IN VELOCITY

TYPE SUBSURFACE SAFETY VALVES

H. D. Beggs, J. P. Brill, E. A. Proaiio, C. E. Roman-Lazo * I

INTRODUCTION

Subsurface safety valves (SSSVs) a r e required by law m most offshore producing


wells. The purpose of the valves is to shut off well flow in the production htbing below the
mudline in the event disasters, such a s explosions o r fires, disable surface shutdown de -
vices. Several types of SSSVs a r e used, including those which a r e controlled from the sur-
face by hydraulic fluids, pressure sensing valves, and differential pressure or fluid velocity
actuated valves. This report deals only with velocity actuated SSSVs.

Actuation of the velocity-type SSSV is based on a simple force balance princjplc.


Loss of pressure above a valve increases the flow rate through the valve and also the
pressure drop across the valve. For subcritical flow, the pressure loss across a restric-
tion, ,such a s the choke o r bean used in a safety valve, is proportional to the flow rate of
fluids through the restriction'. 'The safety valve i s held open by spring and seal gripplng
forces which together a r e greater than the opposing resultant well fluid forces generated by
normal production rates. However, for higher than normal production rates corresponding
to loss of tubinghead back-pressure, the net well fluid forces become great enough to over-
come the spring and seal gripping forces and to actuate valve closure. The consequences of
incorrect valve sizing a r e either premature closures, which result in lost production and
operator expense, or loss of protection from using a valve which cannot be closed by well
flow rates corresponding to disaster conditions.

Before recent API safety valve standards were written, functional testing pro-
cedures and selection of manufacturing tolerances on c r i t i c p valve components were left to
2
the discretion of valve manufacturers. New API standards and recommended practices
have been written by the API Committee on Standardization of Offshore Safety and Anti-
pollution Equipment (OSAPE). These documents provide manufacturing tolerances and a
formal procedure for functional and perforniance testing.

The design procedure to be followed when selecting a velocity-type SSSV for a


particular well is illustrat d in Fig. 1.1, which is a logic diagram of the computer program
described in API RP-14B.
5 Steps 1-3 a r e based on measured well flow data and a r e required
to predict the productivity o r inflow performance of the well. Steps 4-6 a r e used to determine
the bean size required to,produce the desired pressure drop across the valve for the
selected closure flow rate. Once the bean size and pressure drop a r e determined, ~t is then
necessary to select the valve components which will allow the SSSV to close at the pressure
drop calculated in Step 6. This i s illustrated in Step 7 and involves selecting the spring
force which balances the pressure force tending to close the SSSV at the selected flow rate.

Current recommendations for valve type and spring and choke size for each well
condition a r e made using technology based on single-phase flow theory. Slnce most valves
operate under gas-liquid flow conditions, the development of improved multiphase flow pre-
'
dictions was recognized a s a high potential area for safety valve improvement. As a result,
the API Offshore Safety and Anti-Pollution Research Committee (OSAPR) awarded a research
grant to The University of Tulsa to study multiphase flow through safety valves and chokes,
The purpose of this research was to develop correlations for predicting pressure drop across
a SSSV occurring during multiphase flow a s a function of variables such a s gas and liquid flow
rates, bean o r choke size, gas-liquid ratio and average pressure. The contracted study was
performed specifically for 2-3/8 in. nominal Otis J and Camco A-3 valves. The resul
this study, officially named OSAPR Project No. 1, were reported in September, 1976, and
P
can be used to perform the calculations highlighted in Steps 1 and 6 of Fig. 1.1.

A very important step in the total design procedure, Step 7, was not considered
in Project No. 1. This involves predicting the pressure drop across the valve at which
closure will occur when the valve is equipped with a particular bean and spring combination.
Valve closure occurs when the fluid forces a r e sufficient to overcome spring forces and
seal gripping forces o r friction. Spring forces can be estimated from a knowledge of the
spring constant and a s a function of the number of spacers used to impart an initial com-
pression in the spring. Very little is known about the nature of the seal gripping forces o r
friction.

Very little closure test data is available and,furthermore, the data available was
obtained using single-phase liquid o r gas a s the flowing fluid. Consequently, in order to im-
prove this step in design procedure, a proposal was submitted by The University of Tulsa to
the API OSAPR committee ,in November, 1975. This proposal was funded in February, 1976
and the existing experimental facility was used to obtain multiphase flow valve closure data
to solve the problem. The SSSVs selected for testing were 2-3/8 in. nominal Otis J and
Camco A-3 valves. The valves used were furnished by the manufacturers. Several com-
binations of bean size and spring force were utilized for each valve. Since actual valves
were used, the pressure drop occurring across the valve and its locking mandrel was
measured and used to develop the design equations.

As a consequence of the procedure used to conduct a closure test, many addi-


tional multiphase flowing pressure drop data points were obtained for each bean size for
both valves. These data points were combined with those obtained in OSAPR Project No. 1
to develop improved pressure drop correlations. Since these data were obtained using
actual valves, it is felt that the improved pressure drop correlations combined with the
closure equations developed in this study represent a significant improvement in the overall
SSSV design procedure.

The purpose of this report is to present final results of the OSAPR Project No. 5
research project. Future sections will describe the experimental equipment and testing pro-
cedure (Section TI), the experimental data obtained (Section m), the development and evalua-
tion of the closure correlation (Section IV), the revised pressure drop correlations (Section
V ) , and conclusions and recommendations (Section VI). Example design problems to illus-
trate the application of the results a r e presented a s an appendix.

EXPERIMENTAL EQUIPMENT AND TESTING PROCEDURE

The experimental test facility which was constructed for Project No. 1 was used
to obtaln data for this study. Although the facility was described in detail in the final report
for Project No. 1, this description is repeated here order to make this report self con-
tained. Also, several modifications and additions were made to the equipment to adapt it for
making valve closure tests.
P a r t of this project requlred determmaiion of the mechanical force required to
overcome spring force and friction o r seal gripping forces to move the sliding bean section.
This required construction of equipment to perform these measurements for both the Camco
and Otis valves.

The following sections describe the major components of the system and the
testing procedure followed to obtain the data. A schematic dlagram of the test facility is
shown in Fig. 2.1.

Liquid System

The liquid used in the tests is fresh water and i s stored in a 750 gallon stainless
steel tank. The liquid pumping system consists of a Kobe model RB 50 and a Kobe model
RD 150 connected in parallel. The RB 50 is capable of pumping 2400 B/D at 1000 psig. The
RD 150 has an output capacity of 4000 B/D at 650 psig.

The liquid metering system consists of two Camco turbine meters, 3/4 in. and
1-1/2 in. , and a 1/4 m. ITT Barton turbine meter, all connected in parallel. The maximum
capacities of the three meters a r e 982, 5580 and 95 B/D respectively. The mlnimum rates
a r e approximately 10% of the maximum capacities. The liquid flow rate i s controlled by a
Masoneilan Camflex flow controller which is pneumatically actuated. The flow rate can be
remotely controlled from the instrumentation building and 1s continuously recorded on a
strip chart recorder. Thermometers a r e placed at several locations m order to monitor
the liquid temperature.

Gas System

At the beginning of Project No. 5 the gas system was converted from natural gas
to air. A 3-stage W orthington compressor compresses the a i r from atmospheric pressure
up to the desired charging pressure in the system. The second compressor is a 2-stage
Knight with a capacity of 4 M M s c f / ~ ,a maxlmum discharge pressure of 960 psig whsch
operates on a closed system at discharge pressure levels between 600 and 900 psig. ' Two
coded receiver tanks a r e installed at the suction and discharge sides of the Knight com-
pressor, in order to eliminate pressure fluctuations due to pulsations, and also to ensure
sufficient gas supply for the compressor in case the system i s disturbed. The inlet receiver
tank also protects agalnst liquid overflow comlng from the gas-liquid separator. A by-pass
system with an automatic valve was installed from the outlet receiver to the inlet receiver
to permit lower gas rates than the minlmum compressor output to flow through the system,
to permit better control on the gas system and also to provlde proper feedback lnto the
suction receiver. The gas outlet from the Knight compressor is connected to a manifold for
distribution.

A line directs the gas from the manifold to the gas meterlng system. The meter-
ing system consists of two Daniel orifice meter runs connected in parallel, 3/4 In. and 3 in.,
respectively, which can be used alternately for a desired flow rate. Downstream of the
metering system, a s was the case for the liquid system, valves isolate the gas system when
flowing single-phase liquid.

The gas flow rate i s controlled by a Kobe constant rate flow controller. The
static and differential pressures from the meter runs a r e recorded using two methods. A
Barton circular chart recorder i s used to continuously record the two pressures and a
Camco 464C Flow Computer automatically calculates the gas flow rate and sends a signal to
a strip chart recorder.

Two-Phase System

The liquid and gas converge at a mixing tee, from which they flow into a 65 ft.
deep well, around a U-tube connection, back up the well and through the test assembly to a
height of 50 ft above the ground. T o prevent end effects, the fluids flow approximately 75 ft
vertically before reaching the test assembly.

After leaving the test assembly the fluids flow through a section of vertlcal pipe
approximately 15 ft long and then through a high-pressure flexible hose which loops back and
is connected to a pipe returning to the separator. A by-pass system is operated with an
electropneumatically actuated valve, and allows instantaneous variation in the direction of
flow either through the test assembly o r around it. Similar quick-closing valves a r e install-
ed in the vertical section above the ground, two below the test assembly and two above it.

Two thermometers a r e installed in the two-phase system, one before the by-pass
and one after it. When the-by-pass is closed and fluid is flowing through the test section, it
is possible to estimate the temperatures upstream and downstream of the choke.

The return line is connected to a C. E. Natco 1000 psig WOG (21 in. x 10 ft)
horizontal separator which can handle high liquid and gas rates. The liquid and gas outlets
from the separator a r e equipped with a manual gate valve and a 2 in. Camflex series 3500
regulator valve. Both regulator valves will close on instrument air failure to prevent loss
of gas through the liquid return line, and to avoid liquid flow into the compressor inlet re-
ceiver. The liquid return line is connected to a filter before reaching the storage tank.
Both liquid and gas systems a r e closed systems.

Testing Procedure

The experimental procedure consisted of flowing fluids through the SSSV at rates
sufficient to generate a pressure drop across the bean which would cause the SSSV to close.
A detailed testing procedure is outlined for both single-phase liquid tests and two-phase air-
water tests. Refer to Figs. 2.1 and 2.2 for the location of the various components cited in
the description.

Smgle-Phase Liquid Tests

1. Shut off the gas system. Close the flow controller in the liquid meter run.
Start the centrifugal pump and zero the strip chart recorder R-4. Open the
flow controller slightly to give a flow rate which is insufficient to close the
SSSV.

2. Close the bypass valve in the test section to force the liquid to flow through
the SSSV. Activate the pressure measuring system and zero the recorders
R-1, R-2 and R-3. P r e s s u r e transducer T-1 measures absolute pressure in
the system and transducers T-2 and T-3 measure pressure drop across the
SSSV.
3. Select a starting point and chart speed and activate all recorders simultane-
ously.

4. Using the flow controller, increase the liquid flow rate m small steps untll
closure conditions a r e approached. Then increase the flow rate smoothly
until the SSSV closes. As soon a s the absolute pressure upstream of the
SSSV reaches a predetermined value, the bypass will open.

5. Repeat Steps 2-4 at least once to verify the results.

6. Change the SSSV, the bean size, the spring o r the number of spacers and go to
Step 1 for another test.

Two-Phase Tests

1. With the SSSV installed, pressurize the system to a pressure of 400-600 psig
with the 3-stage compressor. Start the Knight two-stage compressor.

2. Open the gas flow controller and start gas flow through the bypass valve.
Verlfy that the same gas r a t e i s being recorded by the circular chart recorder
and the gas flow computer.

3. Close the bypass valve and force the gas to flow through the SSSV.

4. With the liquid flow controller closed, start the centrifugal pump. Slowly open
the flow controller until a small amount of liquid is flowing.

5. Activate the pressure recording system, zero all recorders and start the re-
corders simultaneously on the same chart speed.

6. Increase the gas rate in steps untll closure conditions a r e approached.

7 . Increase the liquid rate smoothly until the SSSV closes and the bypass valve
opens.

8. Repeat Steps 2-7 to verify the results.

9. Make the necessary changes in the SSSV and go to Step 1 for another test.

Test A ssemblg

The test assembly consists of a section of 2 in. line pipe (I.D. = 1.995 in.) in
which a locking mandrel has been installed to mount the SSSVfs. The locking mandrel will
accept either the Otis o r the Camco Valve and seals the annulus to isolate the upstream and
downstream pressures. The assembly 1s equipped with four pressure taps. A schematic of
the test assembly showlng the pressure tap locations 1s shown in Frg. 2.3. Taps 1 and 4 were
located to correspond to Taps 1 and 1 8 in Project No. 1 in order to compare measurements
made on the simulated valves with those made using actual SSSVfs. Taps 2 and 3 were
located a s close a s possible to the upstream and downstream ends of the SSSVfs, respectively.
Slnce the Otls and Camco valves a r e of slightly different lengths, two locations were prov~ded
for Tap 2.
Spring Force Measurement Assembly

In order to accurately determine the mechanical forces acting to oppose valve


closure, it was necessary to measure force o r weight applied to the spring versus axial de-
flection of the spring. A simple apparatus to accomplish this was constructed and is shown
schematically in Fig. 2.4. The measurements were performed for several combinations of
springs and spacers for each SSSV. Because of the different design of the valves, a slightly
different arrangement was used for the measurements. The construction of the Otis valve is
such that it closes instantaneously when the bean assembly moves. It was found that the
weight causing closure varied depending on how long the weight was applied. This tendency
was also observed in the closure flow tests. That is, the force at which the valve closes
depends on the rate at which the closure force is applied. Consequently, in conducting both
the spring force and closure tests the rate a t which the force was applied, o r the rate at
which the flow rates were increased, was held constant.

EXPERIMENTAL DATA

In order to develop more accurate methods to design SSSVs, it was necessary to


obtain considerable experimental data under conditions similar to those under which the
valves operate in the field. It was also necessary to accurately measure the spring forces
acting to hold the valve open and to determine the forces tending to close the valve, which
a r e generated by a pressure drop across the bean. In the process of measuring the
pressure drop at closure, several data points of flow rate versus pressure drop prior to
closure were obtained. These data were used to improve the pressure drop prediction
method developed in OSAPR Project No. 1. The following sections describe the data obtained,
the method of obtaining the data and the ranges of the data.

Spring Force Data

As discussed previously, a velocity-type SSSV remains open until the well flow
rate increases to the point that its associated pressure drop, acting on the piston area, over-
comes the forces holding the valve open. The forces acting to hold the valve open a r e
created by a mechanical spring and the magnitude of the force can be increased o r decreased
by changing the length of the spring. This is accomplished by adding o r removing spacers
o r washers of various thicknesses to change the degree of preloading of the spring. The
force generated by the spring can of course be calculated if the number of spacers, thickness
of the spacers and spring constant a r e accurately known. These data a r e published by the
valve manufacturer, but a r e subject to e r r o r because of manufacturing tolerances. To
alleviate these e r r o r s measurements of force versus spring deflection were performed for
each assembled valve used in the study.

Springs of two sizes a r e available for the Camco valve. These springs have con-
stants of approximately 12 and 30 lbf/in. of deflection. The spring force opposing closure in
the Camco valve can be calculated from the following equation:

F = K (D. + D c + NT)
s 1
where F = spring force opposing valve closure, Ib
S f'
K = spriqg constant, lb in. ,
'f
D.1 = compression of spring with no spacers, in.,
D = distance which the bean must move before c l o h r e , in.,
C
N = number of spacers, and
T = thickness of spacers, in.
Manufacturer's specifications for these variables are: ,

K = 12 o r 30 Ib in.,
D. = 0.875 in. ,
4
1
N = 0, 1, 2, 3 o r 4, and
T=0.5in.

Exammation of the design and operation of the Camco valve revealed that after the
bean has moved a distance of 0.953 inches, the flapper has moved into the flow stream and the
safety valve closes. Therefore, the value of D is 0.953 in.
C

Comparison of the measured spring forces and those calculated using the above
equation and the manufacturer's specifications resulted in substantial differences. The
measurements were made using a 12 lb in. spring (Spring m) and a 30 lb in. spring
S
(Spring 11) while the springs were inst led in the safety valves. The tests'f involved measure-
ment of bean deflection versus force o r weight for various combinations of spring and
spacers. Results of these measurements a r e presented graphically in Figs. 3.1 and 3.2.
The average slope of the lines represents the actual spring constant and the intercept at zero
.
deflection represents the value of M D. The force a t a deflection of 0.953 m. represents the
value of K Dc. Using the experirnenta data, and rearranging Eq. (3. I ) , the following
equations for the two springs can be written:

F =A+BN
s
where
A = K (D.+ D )
1 C

For the two springs supplied wlth this valve, the following values were obtamed:

Accurate spring forces at closure conditions can be calculated for a particular


spring and spacer combination from:

Spring III: Fs = 13.51 + 5.84 N


Spring 11: F = 41.63 + 13.5 N
s
The effect of lubrication was not considered when performing the force-deflection
measurements used to obtain Eqs. (3.3) and (3.4).

The values of A and B calculated from manufacturer's specifications a r e 21.90


f' 4
and 6 respectively for the 1 2 lb in. spring and 54.84 and 15 respectively for the 30 lb in.
spring. These differences can possibly be attributed to the fact that thevalves supplied by
Camco had previously been in use, that is, the springs could have been over-stressed during
prior testing.

The differences in the calculated and measured values of A and B suggest that it
would be advisable to perform similar measurements on each safety valve which is to be in-
stalled in a well. The values of A and B could be determined from only two force-deflection
measurements made on an assembled safety valve. These measurements could be supplied
by the manufacturer for each valve sold.

The closing mechanism in the Otis valve is designed so that closure occurs a s
soon a s the bean moves. Therefore, the D term in Eq. (3.1) is zero. Only one spring is
available and according to the specification; given in the API RP-14B report the spring con-
f'
stant is 75.1 lb in. Several measurements of force necessary to close the valve with vary-
ing numbers of spacers were made on the assembled valve supplied by Otis. The results
a r e shown in Fig. 3 . 3 . The equation for the Otis valve is:

F =A+BN
where
s

~ F o r the particular Otis valve tested, the following values were obtained:

Therefore, spring force acting to hold the valve open can be calculated from

As was the case for the Camco valve, if the manufacturer's specifications listed in API
RP-14B were used, different values for these parameters were obtained. These values a r e

As was pointed out in the discussion of the Camco valve, the measurements were
performed on the assembled valve, whereas the manufacturer's specifications a r e given for
the individual components of the valve. The differences a r e shown merely to reinforce the
suggestion that these measurements should be made on each individual assembled valve at
the factory.

A step by step procedure and example of determining the spring force equation
from two measurements follows.

Procedure to Determine Spring Force Equation

1 With the spring and N spacers installed 1n the SSSV, measure the
defleclion of the bean o r plslon, X , at two d~ffercntvalues of
applied force, F
S
2 Calculate I i = A I? /AX
S
3. Calculate D = ( F -I<TN)/I<-s using one value o l 17 and 1ts comes-
1 S S

4. Calculate A = K ( D + D ) . D = 0. 953 m. for the Canlco A-3 valve.


1 C C
D = 0 for the Otrs J valve.
C
5. Calculate B = K T
6. The sprlng force equation 1s then
F =A+BN
s
Example: D e t e r m ~ n ethe sprlng force equallon for a Camco A-3 valve using lhc
following data:
N=2
1. FS1 = 20 lbf X = 0. 5G5 in.
1

D = 0.176 in.
1

Valve Closure Data


F o r each of the SSSVs tested, the widest ranges of flow rates, bean sizes and
spring forces possible were covered. Table 3 . 1 shows the ranges for each variable for
both the Otls J valve and the Camco A-3 valve. The number of closure tests obtained for
each valve and bean size 1s glven 1n Table 3.2.
Data measured for each test consisted of gas and 1iquid flow rates, upstream
pressure at Tap 1, pressure drop between Taps 1 and 4, and pressure drop between Taps 2
and 3. Figure 3.4 shows typical gas and liquid flow rate recordings, and typical absolute
pressure and pressure drop recordings. The instant of closure can easily be seen on the
recordings.

In order to facilitate identification of individual data points a coded numbering


system was devised. The number for each data point consists of 7 digits. The first digit
identifies the SSSV used, the second digit identifies the fluid o r fluids, the third digit speci-
fies the bean size, the fourth digit specifies the spring installed in the valve, the fifth digit
specifies the number of spacers in the valve and the remaining 2 digits specify the data
point run number. As an example, a test with the number 4321011 would identify the eleventh
test performed using a Camco valve for an air-water test through a 12/64 bean. A 12 ib(in.
spring with no spacers is used to oppose closure. The coding system i s outlined in Fig. 3.5.

Pressure Drop Data

The testing procedure used to obtain the valvel closure data resulted in many data
points of gas and liquid flow rates and pressure drop across the SSSV prior to closure. As
outlined in the testing procedure, the flow rates were increased in steps and allowed to
stabilize before another increase. Since the flow rates and pressure drop were continuously
recorded, at each flow rate increase step a data point was obtained to be used in improving
the correlations developed in Project No. 1. Approximately 800 and 460 data points were
obtained for the Camco and Otis valves, respectively. The application of this data to develop
the revised pressure drop correlations i s described in Section 5.

Table 3.1 Ranges of Flow Rates


CAMCO A-3 VALVE

OTIS J VALVE

Bean Size, Llquld Rate, b b l / ~ Gas Rate, nlscl/d


~n. M h ~ l m u m M a x l m u n ~ Mrn~nlum R'Iaximuin
12/64 146.0 515.0 56.4 155.4
16/64 142.0 839.0 134.0 2G3.5
20/64 172.0 1320.0 187.4 698.3
32/64 456.0 870. 0 518.8 l(;:;o. 7
able 3.2 N u n ~ b e rof Data Points

CAMCO A-3 VALVE

Bcnll S ~ L ; C , Numbcr or Tesls

111. Llcluld T~\~o-I)ll'tse


Told

12/64 17 13 30

20/64 27 20 47

24/64 14 0 14

a 8/24 10 o 10
32/64 16 13 29

Total 84 46 130
-.

OTIS J VALVE
DEVELOPMENT OF CLOSURE CORRELATION

An important step in the procedure for selecting a SSSV to install in a particular


well involves determining the spring force necessary to hold the valve open until a selected
pressure drop across the valve is achieved. This requires an equation o r correlation of
closing pressure drop a s a function of spring force. This section describes the development
of an equation of this type for both the Otis J and Camco A-3 SSSV1s.

Theory

An analysis of forces acting on the movable bean in a SSSV when closure is


imminent can be expressed by
ApA=F + F (4 1)
s g
where
Ap = pressure drop across the SSSV bean,
A = effective a r e a on which Ap acts,
F = spring force opposing bean movement, and
S

F = additional forces opposing bean movement, such a s friction and


g seal gripping forces.
The pressure crop can be calculated using equations presented in Section 5 and
the spring force can be calculated using methods described in Section 3. There remain two
unknown factors in Eq. (4.1) - the effective a r e a A, and the seal gripping forces, F
g'
Solving Eq. (4.1) for Ap gives

A plot of Ap versus F on coordinate axes will result in a straight line of slope


~ intercept at Ap = 0 of% = F /A. The reciprocal of the slope of the line is then
a = 1 / and
the effective area of the bean. g

As the bean size is increased, the actual annular area exposed to the flow de-
creases. Therefore, it would seem that a larger pressure drop would be required to cause
closure a s the actual a r e a is decreased by installing a larger bean. However, measured
closure pressure drop data revealed that there is no definite trend toward larger Ap a s bean
size is increased. That is, for all practical purposes, the effective a r e a is constant, re-
gardless of bean size. consultation with Dr. Peter Griffith of MIT supported this observation.
A control volume analysis performed by Dr. Griffith to reinforce his argument is attached
a s Appendix B.

Data Analysis

A number of closure pressure drop t e s t s were made for several bean sizes on
each SSSV. For each bean size the spring force was changed by adding spacers to increase
the spring force opposing closure. A separate plot of Ap versus F was made for each bean
S
size and then all the data were combined to produce one equation applicable for any bean size
for each SSSV. A least squares fit of the data produced values of a and b for use in Eq. (4.2)
f o r each bean size and also for the combined data. Plots of the experimental data a r e shown
in Figs. 4.1 through 4.13. It should be pointed out that each data point on the graph may
represent several actual tests. That is, regardless of the gas and liquid flow rates, the
SSSV will close at the same o r nearly the same Ap.

Tables 4.1 and 4.2 list the values of a and b for each SSSV by bean size and also
for the combined data.
Table 4.1 - Constants for the 0 tis J Valve
I3c;~nSlzc, in. 3 11

12/64 1.3735 6.1490


1(~/(;JI 1 :l(iS(; 1 :I. OO,l I
20/64 1.2268 30.0396
24/64 1.5401 1.6673
2 8/64 1.3895 10.5294
32/64 1.4808 23.8254
C omblned 1.3338 20.5526

Table 4.2 - Constants for the Canlco A-3 Valve

Bean Slze, in. a b

12/64 2.0287 16.1058


20/64 1.8581 17.6183
24/64 2.3096 1.GS17
2 8/64 1.8870 18.9425
32/64 2.1127 19.8987
C omblned 2.0107 16.4073

Fig. 4.14 shows plots of bean size versus effective area calculated from
Tables 4.1 and 4.2. The scatter of the data suggests that no effect of bean size on the
closure pressure drop can be established. Therefore, it was decided to use the values of
a and b determined by using the combined data to predlct closure pressure drop a s a function
of spring force independent of bean size.
In order to test the feasibility of using a single equation for each SSSV, a com-
parison was made of the accuracy and precision of the calculated closure pressure drops
using two methods. In the first method the pressure drop was calculated using different
values of a and b in Eq. (4.2) for each bean size. In the second method single values of a and
b obtained from the combined data analysis were used. The results a r e summarized in
Table 4 . 3 . Tables 4 . 4 through 4 . 7 list the results for the individual data points.

Table 4 . 3 - Statistical Comparison of Methods

Comblned Data

Based on the samll increase in the accuracy of the closure pressure drop pre-
diction accomplished by using a different equation for each bean size, the following equations
a r e recommended for determining the relationship between spring force and closure pressure
drop:
2 in. Otis J Valve: Ap = 1.3338 Fs+ 20.5526 (4-3)
2 in. Camco A-3 Valve: Ap = 2.0107 F + 16.4073
s (4.4)
Example calculations utilizing these equations to select a SSSV for a particular
well a r e given in Appendix A.
REVISED PRESSURE DROP CORRELATIONS

The same procedure a s outlined in Section 5 of the OSAPR Project No. 1 final
report was used to develop improved empirical discharge coefficient correlations. The
pressure drop was calculated with a C = 1.0 using the homogeneous Model No. 2 equation:
D

The discharge coefficient used in the regression analysis was defined from Eq. (5.1) as:

CD = / A 1, ineasurcd
ApTp

In OSAPR Project No. 1 the discharge coefficient was correlated with Reynolds number,
NRe, Froude number, N density ratio, P and diameter ratio, R .
Analysis of the
larger data base obtaine$g this study using?he actual SSSV1s r e v e a f d that only two para-
meters were highly correlated with the discharge coefficient. These parameters a r e
diameter ratio and velocity ratio, V The velocity ratio i s essentially an in situ gas-liquid
D'
ratio since it i s the ratio of the superficial gas velocity to the superficial liquid velocity, that
is VD = v /V . These findings agree to a certain extent to single-phase flow theory,
since for %ngf&-Phase flow through an orifice the discharge coefficient becomes a function of
diameter ratio only for large values of Reynolds number.

The final form of the discharge coefficient correlation for the Camco A-3 and
Otis J 2 in. SSSV's is:

where

Rn = d I)/d 1'
VD = vsg/vsL,

d = b e a n clinmctcr,
B

C = empirical constanls given in Table 5.1. For single-phase llquld


< flow,
Statistical Evaluation of Parameters

The statistical parameters used to evaluate the calculated pressure drop using
the developed discharge coefficient correlations w e r e average percent e r r o r , I?, and stand-
ard deviation of percent e r r o r s from the average percent e r r o r , S, which were defined as:

1 where

- A P measured
1
*'calculated
meamircd
)I
x 100

N = number of obse=-vatior-E
and

Table 5.2 gives a summary of the statistical results of applying the individual correlations to
the experimental data from which they w e r e developed. Comparisons between the measured
and calculated values for pressure drop a r e shown graphically in Figs. 5.1 through 5.4.

Table 5.1
I<nipiric:nl C o c f f i c i c n t ~for Ori I i c c *
Discharge Coefficlcnt Correlations

Cnmco 0 tis
CamcoLlquid Two-Phase Ot~sLlquld Two-Phase

C 0.2815 .5417 1. 5247 1.1819


o
9.4691 3.8749 -13.9697 -1.8761
C1
-25.5689 -10.4536 51.0889 ,9922
C2

C3
-0- -0- -0- - .0119
4
Table 5.2 - Sumnlary of Statistical Results

No -
Bean Srze Tests 1;: S

2 12/64 90 1.21 14.82


s
.-I
GI 16/64 128 - 0.14 15.27
(II
.-4

5 20/64 90 0. 68 10.25

Q) 12/64 61 3.56 18.48


: -
PC 16/64 62 - 6.39 16.14
I
- - _ ____ .. ---
E-c
S
(II
20/64 15 22.29 9. GO
r
.-4
4.2

0 15 - 1.33 17.53
32/64

z3 12/64 76 - 0.11 3.56


2' ------- -- -.
d
o 2 0/64 91 0.26 4. G7
2
cd
32/64 181 0.13 8.35
U
8

Q)

.G
: 12/64 137 0.40 12.98
PC
I
o 20/64 119 - 2.68 19.40
3
E-c
o
0
20/64* 35 13.59 (3.20 .
B
CJ
U 32/64 172 0.21 9.77 ,

*
Gas and Water Tests

Overall Statistical Results

No. -
SSSV Tests E S

O t i s Llquld 308 0.49 15.25

0 tis Two-Phase 153 - .88 18.61

C amco Llquid 348 -0.11 G. 68

- -54
Camco Two-Phase 463 14.12
.
CONCLUSIONS AND RECOMMENDATIONS

As a result of the study described in the previous sections, the following con-
clusions have been reached.

1. Methods have been developed to predict the pressure drop at which a SSSV
will close a s a function of spring force.

2. For all practical purposes, the closure pressure drop is indepenent of bean
size.

3. Calculations of spring force opposing closure a r e subject to considerable


e r r o r if friction and variations in spring length among individual SSSV1s a r e
ignored.

4. A procedure is suggested for accurately predicting effective spring force for


individual SSSV1s.

5. Improved correlations a r e presented for predicting pressure drop across


SSSV1soccurring during two-phase flow as a function of bean size.

6. Combining the procedures described in the above conclusions, an accurate


method now exists for selecting a 2-in. velocity actuated SSSV for preventing
blowouts in producing wells.

Several other areas which warrent additional study were made evident by the re-
search performed in OSAPR Project No. 5. Some of these recommendations a r e listed
below.

1. The same type of study should be extended to other brands of velocity actuated,
2-in. SSSV1swhich a r e available commercially.

2. The study whould be extended to larger size SSSV1s, especially the 2.5 in. size.

3. An attempt should be made to develop a single pressure drop prediction method


which will apply to all SSSV1s. This will require extensive analysis of the
experimental data obtained in OSAPR Projects No. 1 and 5, and also the data
which might be obtained should recommendations 1 and 2 be implemented.

ACKNOWLEDGMENTS

The authors a r e grateful to the API OSAPR Committee for the financial support
and technical guidance provided during the project. The support of the University of Tulsa is
also recognized.

Special thanks a r e due the advisory subcommittee for their valuable suggestions.
The committee consists of chairman Phillip D. Pattillo, Amoco Production Co. Research
Center; Robert Goodwin, Gulf Oil Corporation; and T. R . Sifferman, Continental Oil Co.
Assistance from Mark Malinowski of Continental Oil Co. and Peter Griffith of the Massachu-
setts Institute of Technology is also appreciated.
Numerous companies made equipment donations to the project which helped
minimize expenditures. These include Bettis Corp. , Camco, Inc. , Daniel Industries,
Kobe Inc., McMurry Oil Tools Inc., Otis Engineering Corp., T oK. Valve and Mfg. Co.,
and Willis Oil Tool Co. Equipment donated included test fixtures, quick unions, orifice
meters, flow rate controllers, liquid pump, ball valves and pneumatic actuators, turbine
meters, check valves and recording equipment.

Special recognition is due the graduate students who performed most of the
actual design and construction of the test facilities, obtained the experimental data, evaluated
existing models, and developed the orifice coefficient correlations. Work of this caliber can
only be done by out standing students. Accordingly, the authors express sincere appreciation
to Mr. Eduardo Proaiio and Mr. Carlos Roman-Lazo. A project of this magnitude also re-
quires special efforts from support staff. Mr. Ted Hope, Research Technician, and Mrs.
Betty Pence, Research Secretary, played vital roles throughout the project.

NOMENCLATURE
Synlbol

Cross sectronal area

Regression coefflcrent

Discharge coefflcrent

Drameter

Percent e r r o r

Average percent error

Acccleration of gravrty

Number of observations

Precsure

Volumelrlc flow rate a1 111-srtu condllrons

Volumetric flow rate at standard condl tions

In-situ gadliquid ratio

Standard deviation from average percent e r r o r

Vcloc1ly

Mass flow rate

Pressure ratio, and lndepel~dentvariables


NOMENCLATURE (Cont. )

Grcelr Sbymbols

A Difference

P Dynamic viscosity

X No-slip holdup

P Denslty a t m-sihl conditions

u Surface tensloll

Subscripts

B Bean (choke)

C Critical

est Estimated

6 Gas

L Liquid

M Measured

m No-slip o r mixture

011s Observed

S Superficial

TP Two-Phase

t Tubmg, o r upstream flow tube

Dlmenslonless Groups
2
N Froude number, v /gd
Fr
Gas-velocity number, v ( p /ga)'l
sg g
1
L ~ q u i d v e l o c i t y n u m b e r ,v ( p /gu)'
N ~ v SL L
Velocity ratio, v /v
I7D s g SL
NOMENCLA'TURE (Con1 )

Rcynolds number, p v d/p

Dlametcr ralro, d /d
B t
D e n s ~ t yratlo, p /p
g TJ
NOMENCLATURE (Con1 )

l<eynolds number, p v d//p

REFERENCES

1 A P I Speclflcatlon f o r 'lSubsurEace S a f c l ~ V:ilvesM,


i API Spec 14A, F ~ r s El d ~ t ~ o n ,
October, 1973

2 API U s e r ' s Manual f o r "API l 4 B Subsurface Controlled Sirbslrrface Safety Vnive


Slzlng- Computer Program11, API Manual 14 BM, F ~ r s Edlllon, t June, 1974,
Sectlon B4; and API Recommended P r x c t i c e f o r "Desigl, Inslallalion and
Operatron of Subsurface Safety Valve Systems1', API R P 111-B, F ~ r s I<dlllon,
l
O c t o l ~ e r ,1973

3 Brlll, J P , Beggs, H D and Sylvester, N D : " O r ~ f i c eCoefficrcnts for Two-


P h a s e Flow Through Veloclty Conlrollcd Subsurface Safet.y Valvcsl', Rcport
to API OSAPR Research Committee, September, 1976

APPENDIX A
APPLICATION EXAMPLES

In order to demonstrate the application of the correlations and equations re-


sulting from this study, examples a r e worked for the selection of bean size and number of
spacers for both the Camco and Otis valves. A hypothetical well 1s chosen which has the
following characteristics:
0
Well depth = 10000 ft - Bottom hole temgerature = 270 F
Tubing diameter = 1.995 in. I. D. Oil gravity = 35 API
0
Surface temperature = 110 Gas gravity = 0.65
Avg. reservoir pressure = 3000 psia

The well is presently equipped with a SSSV located at 3000 ft below the surface
containing a 20/64 in. bean. A recent production test on the well resulted in the following
data:
q = 700 STB/D q = 280 Mscfd Flowing tubing pressure = 300 psia
0 g
It is desired to equip the well with a SSSV which will allow a normal production
rate of 800 STB/D at the Same GOR. The procedure outlined in the flow chart (Fig. 1.1)is
to be followed. The assumption is made that the Productivity Index is independent of flow
rate.
Solution for the C amco A-3 Valve

STEP 1. q =~OOSTB/D q = 280 Mscfd Ptf = 300 psia


0 g
E P 2. Using the Orkiszewski two-phase vertical flow correlation, the pressure down-
stream of the SSSV (3000 ft) is 688.4 psia. Calculation of the pressure drop a c r o s s
the SSSV presently installed is an iterative procedure, since the discharge coeffi-
cient correlation i s based on conditions upstream of the SSSV.

.
'I?r la1 1 ]<slim:ilc All = lo psi.

2
AU -j 0.0005:12 I1

v = 9 3 . 7 3 fl/scc
sLB
v - 97.08 ~ I . / S P C
sg 13
v 191.41 fl/scc
m I3
3
0 = 25.33 lb / f t
m m

R~
- 20/64
1.995
-
- 0.3125
1.995
= 0.156F -
v D = 1.042
2
c = C + C R +C2RD+C3VD
D 0 1 D

From Table 5 . 1

2 2
Ap = p v / 2 gc C d = 125.77 psi
m mB

Thls i s not close enough to the estimated Ap of 1 0 psi. Using the calculated Ap a s
the next estimated Ap resulted in convergence at Ap = 113.3 psi m two more ~ t e r a -
tions. The pressure upstream of the SSSV i s then 688.4+113.3= 801.7 psla. The
calculated p is 2528.2 psia at 10000 ft.
wf
STEP 3.

STEP 4.
Select a pressure differential across the new SSSV to be installed to be
s 150 psi.

STEP 5.
The desired normal production rate is to be 800 STB/D. The closure
rate must be between 110 and 15w0 of the normal production rate, therefore the
pressure drop calculations were made using

STEP 6 .
For qo = 1000 STB/D, pwf= p --
qo =
looo = 2324 psia
ws PI 3000 - 1.48

Using the Orkiszewski method, the SSSV upstream pressure i s 707 psia. A bean
size must be selected such that the pressure drop across the bean at a flow rate of 1000
STB/D is l e s s than 150 psi. Using the discharge coefficient equation, it was found that a
bean size of 24/64 in. would give a C = 0.900 and Ap = 1 2 0 . 3 psi. The flowing wellhead
D
pressure is then calculated to be 178 psia, which is greater than 50 psi and is therefore
satisfactory.

STEP 7.
F o r the 2 in. Camco valve available, the spring force data obtained by
making two measurements of force vs deflection resulted in the following equations:

For the 12 lbf/in. spring: F = 1 3 . 5 1 + 5 . 8 4 N


s

For the 30 lb in. spring: F = 41.63 + 1 3 . 5 N


'f S

The spring iorce required to allow the SSSV to close at a Ap of 1 2 0 . 3 psi is calcu-
lated from Eq. (4.4) for the Camco A-3 valve.

Selecting a 30 lb spring, the number of spacers required is


f
F
S
41.63--
-
51.67 4 1 . 6 3
N= - = 0.744
13.5 13.5
Therefore
N = 1 spacer.

Using 1 spacer will require a pressure drop to close of


I
Ap = 127 psi

Therefore, the SSSV will not close at a producing r a t e of exactly 1000 STB/D. Using
a trial and e r r o r procedure, ~t was determined that the producing r a t e required to produce a
Ap of 127 psi is 1030 STB/D. This r a t e 1s within the limits required a s it is only
(100) (1030)/800 = 129% of the desired producing rate.

The completed deslgn then calls f o r a 2 in. Camco A-3 valve equipped with a 24/64 in.
bean and a 30 lb m. spring with 1 spacer to be installed. The SSSV will then close when the
'f
wells producing r a t e exceeds 1030 STB/D.

Soliltion for the 0 tis J Valve

STEP 1.
q = 700 STB/D q = 280 Mscfd
0 g
P = 300 psia
tf
STEP 2.
P r e s s u r e downstream of the SSSV is 688.4 psia. Calculation of Ap a c r o s s the valve
is iterative.
This is not close enough to the estimated Ap of 10 psi. After 3 iterations,
Ap was found to be 111 psi. The pressure upstream of the SSSV is 688.4+111=799.4 psia.
The calculated p at 10000 ft is 2519 psia.
wf
STEP 3.

STEP 4.
Select a pressure differential across the new SSSV to be s150 psi.
STEP 5.
Select the closure producing rate to be 1000 STB/D.
STEP 6.
For qo = 1000 S?B/D, P = P
wf ws
- -= 3000 - 1000 = 2315 psia
PI 1.46

Using the Orkiszewski method, the SSSV upstream pressure is 700 psia. A bean
size must be selected such that the pressure drop across the bean at a flow rate of 1000 STB/D
is less than 150 psi. Using the Otis two-phase discharge coefficient equation, it was found
that a bean size of 24/64 in. would give a C = 0.852 and Ap = 135.2 psi. The flowing well
D
head pressure is then calculated to be 163 psia, which is satisfactory.

m E P 7.
F o r the 2 in. Otis valve available, the spring force data obtained by making
two measurements of force vs deflection resulted in the following equation:

The spring force required to allow the SSSV to close at a Ap of 135.2 psi is
calculated from Eq. (4.3) for the Otis J valve.

The number of spacers required is

Therefore
N = 7 spacers.
Using 7 spacers will require a pressure drop t o close of

Ap = 137 psi

Therefore the SSSV will not close a t a producing rate of exactly 1000 STB/D. Using a
trial and e r r o r procedure it was determined that the producing rate required to produce a Ap
of 137 psi is 1017 STB/D. This rate is within the limits required a s it i s only (100)(1017)/800
= 127% of the desired p r o d ~ ~ c i nrate.
g - $.,

The completed design then calls for a 2 in. Otis J valve eqiipped with a 24/64 in.
bean with 7 spacers installed. The SSSV will then close when the wells producing rate exceeds
1017 STB/D.
APPENDIX B
Control Volume Analysis of a Camco A-3 Safety Valve

Introduction

The purpose in doing a control volume analysis of the safety valves i s to see if the
experimental results a r e reasonable and suggest an appropriate way to handle the data. Figure
1 shows a schematic of the valve. Flow through the orifice (bean) causes a large pressure drop
across the valve, large enough to cause the valve to close. In this analysis we shall relate the
forces felt by the valve to the pressure difference across it.

Figure 2 shows the control volume used in this analysis. It i-ncludes the sliding sleeve,
a piece of fluid on each end of the valve and the fluid contained in the valve. It is not the obvious
control volume, but is the easiest one to compute and yields some interesting 'results. Let u s go
step by step through the analysis and list and justify each of the assumptions that must be made.
Let u s begin with the momentum equation.

A
out

If we place the control volume boundaries at some distance from the top and bottom
surfaces of the valve, the velocity i s one dimensional and normal to the control volume
boundary. If, in addition, AP/P is small, the volume change of the mixture a s it passes
through the valve is negligible and V is equal to V The continuity equation requires that
ut'
the flow rates in ( p AV) and out to behequal and equa?lon (1)reduces to

If the control volume were moved to the top and bottom surfaces of the sliding sleeve,
the momentum fluxes in and out would not be the same, and the right hand side of (1)would
have to be retaned. If, alternatively, we choose the control volume to be just the sleeve, we
would have to include the shear force due to the fluid on the inside surface of the sleeve and
the pressure force on the internal horizontal surfaces. We would determine this by making a
control volume of the fluld contained in the valve. The shear force and pressure forces at
each end would have to be included. This control volume would also have some unknown
momentum fluxes and pressure forces on the step too, so this is not a convenient control
volume and i s not recommended.

Returning to the control volume chosen, there are, in general, pressure forces on
the horizontal surfaces (PA), fluid shear forces on the vertical surfaces, a spring force on
the shoulder, (F ) and gap shear forces on the two seals (F .
The shear forces on the out-
s
d
slde of t k seals a r e assumed 0 because the seals reduce th flow to negligible proportions.
The fluid shear forces before and after the sleeve a r e also small because the pipe i s large
enough so that the shear stresses a r e small and the corresponding shear forces a r e negligible.
Equation (2) then becomes

A plot of the pressure difference necessary to shut the valve versus the spring force should
, an intercept of (F /A). The data shows this.
then yield a straight line with a slope of 1 / ~and
g
This analysis also shows that the valve closes at the same pressure difference no
matter what the bean size is. We would also expect the slope ( 1 / ~to) be closely related to
the cross sectional a r e a of the sliding sleeve and constant, independent of flow rate, O/G ratio
and spring o r bean size. To the extent that the friction forces remain constant, we would ex-
pect the intercept to remain constant too. All these trends a r e evident in the data.
STEP 1
-

USE
--
v(AE;;L )
CORRCLATION

CALCIJLATF
PRESSURC

STEP 3
- CAI C U L A T E I WCI L
INrLOY
PERFORMANCE
I

STEP 4 --
SELECT

STEP 5 SELCCT ? l l O Y ~f

STEP 6 CALCULATE
1 BOTTrJI.1 H O L E
PRESSUPF 1-

PRESSURE
SSCSV

DROP OR
BEAN S I Z E
.-

STEP 7 CALCULATE

0--
OESlGIl
COMPLETE
OK

Flgure 1.1 - Loglc D~agranlof A P I Veloclly Type


SSSV Slzl~lgCornpuler Program
NO. 1 2(2A) 3 4 D - DEMODULATOR
E - EQUALIZING VALVE
MURPHY G-GAGE
SWITCH M-MANIFOLDVALVEFOR
EACH PRESSURE TAP

S - SAFETY EQUALIZING

T - TRANSDUCER

I I I
I
I I I
I I I t

GAS
COMPUTER

R -1 R -2 R -3 R -4 R -5
F l g x ~ r e2.2 - Pressure and F l o w Rate Recording System
2" LINE PIPE
(1.995 I ' I D)

LOCI(ING MANDREL

sssv

at-
0
W
0
Y

F~gllrc2 . 3 - Tcsl A s s e l l ~ b l y
I BLY

Flgure 2 . 4 - Spring Force Measurel~lclltAssembl~cs


I AVG. K = 27.0 L B f / IN

10 - I

0 I I I I I I

0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6


DEFLECTION - INCHES

Flgure 3 . 1 - Sprlng F o r c e Measurements - 30 lb Sprlng


0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
DEFLECTIOI\I - INCHES

F ~ g u r e3 . 2 - Spring Force Measurements - 1 2 lh Spring

B-41
Frgure 3. Fi Tcsl Numbcrmg System

x x x x x l x x l

Run Number

-0- 0
1- 1
2 - 2 Cainco Valve
3 - 3
4- 4
Number of Spacers
t - 5 - 1

8 - 20
L
9 - 25 I

Spring
-
1 Camco 12 Spring
* -
2 Camco 30 Spring
-3 -
Otis Sprsng

C
- 8/64
Bean Size
2 - 12/64
3 - 20/64
4 - 24/64
5 - 32/64
G - 16/64

1 - air
2 - water
Fluids 3 - a i r and water
d

4 - gas
- gas and wntcr
\
Valve Type 2 - Otls
-4 - Cainco
100
SPRING FORCE, FS
Fig. 4.1 12/64 Combllled O t i s Data
M E W R E Q PRESSURE QRGP,PSI

Fig. 5.1 - C o m p a r i s o n o l C:~lculalcclv s Mcasurcd Prcssurc DI-O~J-


O l ~ sTwo-Ptlasc 'l'c:sl,s
Fig. 5.2 - Comparison of Calculated vs Measured P r e s s u r e Drop -
Otis Liquid T e s t s
Flp;. 5.3 - Comparison or Calculatcd vs R/lc3sured P ~ ~ c s s i lDrop
rc -
Camco TWO-13h3hcT C S ~ S
MEClSUREO PRESSURE ORW,PSI

r i g . 5.4 - Comparison of Calculated vs Measured P r e s s u r e Drop -


Camco Liquid T e s t s
Figure 2 - Conlrol volulne uscd in analysis prcscntcd in l l ~ i s:q)pcndlx.'l'llcrc 1s 110
net pressure o r shear forces acting on the flapper and no lnomeiltum flux
o r sllcnr rorces acting on tllc sides o l lhc liquid plug shown at thc lop and
bollom of lllc conlrol volumc.

You might also like