LSU/MMS WELL CONTROL WORKSHOP SESSION 4

NOVEMBER 19-20, 1996 PRESENTATION 21

DRILL STRING SAFETY YALYE TEST PROGRAM

by

Adam T. Bourgoyne, Jr.

Elliot D. Coleman

Thomas T. Core

Petroleum Engineering Department

Louisiana State University

Baton Rouge, Louisiana 70803-6417

OBJECTIVE
The objective of this task was to experimentally measure the torque required to close and
open drill string safety valves for various flow rates, back pressures, and valve designs.
ABSTRACT
As a primary component of the drillpipe blowout protection system, drill string safety
valves should be very reliable. The drill string safety valve's reliability is questionable in its
current design configuration. The Petroleum Engineering Research and Technology Transfer
Laboratory (PERTTL) under grants from the U.S. Department of the Interior's Minerals
Management Service has conducted research to investigate the mechanism of failure associated
with the common failure modes. The research also intends to make recommendations for
designs that will solve the reliability problems associated VI-1th these valves.
INTRODUCTION
A study of blowout preventer pressure test results by the Minerals Management Service
(MMS) for the U.S. Outer Continental Shelf during 1993 and 1994 identified drill string safety
valves (DSSV's) as one of the least reliable components of the well control system [Hauser,
1995]. Figure 1 details the results. Note that the pressure test failure rate for drill string safety
valves and inside blowout preventers was about 25%. This was especially troublesome, since the
level of redundant protection for
Annular Connections & Other blowouts through the inside of the
19% drill string is much less than for
Choke flow through the annulus. Note
Manifold also the choke manifold had a high
Choke Valves pressure test failure rate. A failure
& Kill 29% in this component is not as serious
Valves because these valves are not a
11% primary blowout barrier. Failure of
one of these valves generally
Orillstring Valves & I-BOP
would not lead to a blowout.
25% After Hauser. 1995 Because it is a primary blowout
Figure 1: Results compiled from blowout preventer component barrier for the drill string, failure
pressure tests for the U. S. Outer Continental Shelf during 1993 of the drill string safety valve
and 1994. could have devastating results.

LSU/MMS WELL CONTROL WORKSHOP SESSION 4

NOVEMBER 19-20, 1996 PRESENTATION 21

In 1994, Mobil conducted an industry survey which identified 29 safety valve failures
during well control operations over an unspecified period. The survey was conducted after Mobil
experienced a number of problems in 1993 with stabbing valves leaking after being stripped into
a well in a threatened blowout situation. The survey findings, as listed below [Tarr, 1996),
identify several common failure modes for safety valves that point to problems inherent to the
basic design of the DSSV's.
• Failure to seal against pressure from below
• Failure to open when under pressure due to high torque
• Failure to seal against pressure from above
• Failure to seal against outside pressure when stripped into a well
• Failure to close due to high torque when throttling mud back.flow
• Failure to seal due to erosion from abrasive flow

Brian Tarr, one of the authors of the study and a Mobil employee, is also chairing an AP!
Task Group Subcommittee to recommend changes to AP! Specification 7, Section 2 for Safety
Valves. The subcommittee is recommending a new classification scheme for safety valves based
on performance testing of valve prototypes. A project jointly sponsored by Mobil and the Gas
Research Institute was funding tests of two new prototype valves at the University of Clausthal
in Germany. The new prototypes being tested were from German and Canadian manufacturers.
The test protocol being followed were the draft procedures being considered by the AP! Task
Group Subcommittee.
In 1995, MMS sponsored a project at LSU to study the failures of DSSV's and
recommend improved designs for these valves to help prevent blowouts through drillpipe.
The following topics will be discussed in this report: (1) a review of the basic drill string
safety valve terminology and function, (2) common failure modes of DSSV's, (3) identification
of alternative devices that can be used with a safety valve to improve reliability, (4) the
problems associated with the design of DSSV's that are being addressed by the MMS/LSU
project, (5) the experimental test apparatus and procedures, (6) DSSV test results from industry
and the results from the experiments at PERTTL, and (7) the recommendations and conclusions
drawn from this test data.

DRILL STRING SAFETY VALVES (DSSV'S)

Drill string safety valves are ball valves used to stop flow through the drill string. Shown
in Figure 2 is a photograph of a traditional TIW drill string safety valve. The patent has expired
on this simple design which is now available from several manufacturers in addition to Texas
Iron Works (TIW) from which it took its name. The name TIW Valve is often used as the generic
name for a drill string safety valve. This photograph was taken during a visit to a valve
manufacturing facility. The valve has been disassembled here to show the main working
components.

 LSU/MMS WELL CONTROL WORKSHOP SESSION 4

NOVEMBER 19-20, 1996 PRESENTATION 21

When rotated 180 degrees, the portion of the

safety valve shown on the right side of Figure 2
would accept the upper valve seat and spring and
screw down over the ball. After assembly, the ball
"floats" between the upper and lower seats and seals
when pressure is applied against the ball. The spring
assists in providing a low pressure seal. The valve
stem fits into a circular hole in the valve body. The
valve is operated by means of a wrench that is
inserted into the valve stem and turned one quarter
turn.
Displayed in Figure 3 is a photograph of a
safety valve made-up on top of a section of drillpipe.
The valve has been cutaway so that the ball and seats
may be observed. This particular safety valve is a
one piece valve design that eliminates the need for
threads in the valve body area. This not only
decreases the number of possible leak paths, but also
eliminates the problem of the ball locking due to
excessive make-up torque. The basic design remains
Figure 2: Photograph of the traditional TIW with a floating ball in a cage which houses the fixed
drill string safety valve. upper and lower seats.
Shown in Figure 4 are the traditional
locations of safety valves. Government regulations
require that a safety valve, with an operating wrench,
for each size drillpipe be maintained on the rig floor
at all times.

Swivel
Upp er kelly cock

Kell

Figure 3: Photograph of a safety valve which

bas been cutaway and made-up on top or a
section of drillpipe.
Lower kelly cock

S•m '"' f DSSV

(S1'bbi"g "'•l

Figure 4: Schematic showing traditional locations

or safety valves.

LSU/MMS WELL CONTROL WORKSHOP SESSION 4

NOVEMBER 19-20, 1996 PRESENTATION 21

COMMON FAILURE MODES

During fishing operations in the J. W. Goldsby No. I, observations began to indicate that
the 18.0 ppg mud in the hole was insufficient to maintain well control. After backing off the
pipe in preparation to sidetrack the well, it began to flow up the drillpipe. The well would not
flow with the kelly attached, but flowed when the kelly was removed. It was decided that the
kelly saver sub and the DSSV would be left on the drillpipe in the closed position in order to rig
up chicksan to the trip tank. After the chicksan was rigged up, the DSSV was opened and it was
noted that the well was flowing. The DSSV was closed but failed to seal. In the time it took to
ready a second DSSV, the well flowed 25 bbls. Stabbing the valve on a joint of drillpipe to
overcome the flow, the second valve would not seal when closed. A third valve was stabbed
using the same technique and also would not seal when closed. Attempts to close the valve
included rigging the valve wrench to the catline to try to force the valve closed. This resulted in
bent and sheared wrenches. Figure 5 is a photograph of the well taken during the blowout. The
well was estimated to be flowing at I 000 BPH with a measured flowing pressure of 3800 psi and
a shut in pressure of 7300 psi.

Amoco conducted a series of

safety valve tests at their research lab
after their Goldsby Blowout in 1990.
The results of this unpublished study
provides information on common
failure modes for safety valves. The
Goldsby blowout let the high
pressure, high flow rate fluid move
from below the valve, past the ball
and seats, and out of the top. A
similar failure occurs when pressure
testing equipment is installed on top
of a faulty safety valve which allows
flow from above the valve, past the
ball and seats, and into the drillpipe
below. This prevents a valid pressure
test from being performed.
Eroded balls, seats and seals
are common. The erosion is due to
flow of mud solids through the valve
as it is being closed. These failures
are caused by a partially closed or
over rotated valve. If high flow rates
Figure 5: Photograph of Amoco Goldsby Blowout. are going to be stopped, the valve
must be shut completely and quickly.
If the valve is not completely closed
in one quick motion, a narrow flow

4
LSU/MMS WELL CONTROL WORKSHOP SESSION 4
NOVEMBER 19-20, 1996 PRESENTATION 21

path is created between the ball and the seat, eroding the closing side of the seal in a very short
time. If the valve is slammed shut there is a possibility of a permanent deformation in the valve
stem stop. This deformation allows the ball to be over rotated causing a flow path to erode the
seal on the opposite side. However, if the ball is not aligned perfectly in the open position,
erosion in an upper or lower kelly valve will also occur during normal drilling operations. In
addition, erosion is caused by wireline work done through the valve.
After stripping a stabbing valve into the well, a failed safety valve can let pressure move
from the annular space around the valve, in through the valve stem, and into the drillpipe.
Surface pressure readings will be irregular or misleading and could cause mistakes to be made
during the well control operations. This is caused when the stem is eroded by an unintentional
flow path or is damaged by stress cracks. Failed elastomers can also cause this type of failure.
Failure of the valve to close within the available torque limits is another significant
failure mode. About 400 ft-lbs is generally regarded as an upper limit of torque that can be
applied manually with an operating wrench. If the torque required to completely close the valve
is exceeded before the valve is fully closed, the one of the failures associated with partially
closed valves can occur. High torque is caused by the build up of pressure in the valve as the
valve begins to restrict the flow. The pressure pushes the valve stem further into and against the
valve body and the ball is forced against the upper seat. These two actions create friction forces
that can not be overcome. If the ball and stem are put under too much pressure, local stress
deformations create metal to metal contacts with the associated high friction surfaces. Poor
dimensional tolerances also allow metal to metal contact. The ball of a two-piece valve often
locks if too much make-up torque is applied across the valve body. Tong placement is critical
when tightening across this type of valve.
Failure of the valve to open on a
pressure differential or even after pressures
are equalized across the ball is also a failure
mode. When the torque required to open the
valve to start well control operations is too
high, the valve has completely failed. It is
sometimes necessary to freeze a plug of ice­
mud below the safety valve so that the valve
can be replaced while there is pressure on the
drillpipe. Higher torque values occur during Caused by Human Error
opening yet are caused by the same actions (Partially Closed Valve)
associated with high torque values during
Figure 6: Photograph of ball and seat that bas been
closing. eroded by mud Rowing through a partially closed
lower kelly valve.
Shown in Figure 6 through Figure 12 are photographs of failed safety valve
components. These photographs were taken during a visit to a safety valve manufacturer and at
PERTTL. They illustrate some of the types of failures that have been discussed. The
backgrounds of the photographs have been cleaned up electronically to better show the
components of interest.

LSU/MMS WELL CONTROL WORKSHOP SESSION 4

NOVEMBER 19-20, 1996 PRESENTATION 21

Shovvn in Figure 6 is a photograph of a ball and seat

that has been eroded by mud flowing through a partially closed
lower kelly valve. The valve was erroneously left in this
position during drilling operations and would not seal during a
well control event.
An example of a safety valve ball cut by wireline work
being done through the valve is illustrated in Figure 7. In order
to achieve as large a bore as possible, there is not much extra
sealing area on the spherical surface near the ID of the ball.
This type of wear can open a leak path that can then be further
eroded by flow of mud.
A valve seat cut by fluid erosion due to a slightly over
closed valve is depicted in Figure 8. Wear on the valve stem Figure 7: Safety valve ball cut by
wire line.
stop can sometimes allow too much rotation of the ball. The
design of the valve stem stop is very important. A photograph
illustrating a failure in the valve stem is shown in Figure 9.

Valve Stem
Stop
Eroded
Hole

Side View

Interior
View

Figure 8: Valve seat cut by Ouid Figure 9: Photograph illustrating valve stem
erosion caused by over-rotation failure.
of the ball valve.

Figure 10: Ball cage deformed Figure ll: Seal erosion Figure 12: Valve stem wear due
around stem opening by caused by over rotation of to ball cage deformation.
excessive torque. the ball.

6
LSU!MMS WELL CONTROL WORKSHOP SESSION 4
NOVEMBER 19-20, 1996 PRESENTATION 21

Figure 10 shows a valve component that has been subject to excessive torque, which
caused permanent deformation in the ball cage and valve stem stops. The resulting deformation
allowed over rotation of the ball which caused seal erosion (shown in Figure 11) and metal to
metal contact between the ball cage and the valve stem. This contact is apparent from the wear
shov.n in Figure 12 on the valve stem.
AUXILIARY DEVICES
Patent searches have supplied good coverage of devices to prevent blowouts through the
drillpipe. After 23 patents were reviewed, it was found that a number of alternatives to ball
valves have been tried. However, ball valves appear to be best suited to the need for full opening
valves ..vith a small outside diameter that can be stripped into the well under pressure. Therefore,
auxiliary equipment that compliments the use of safety valves and increases the number of
barriers to a blowout through the drill string is preferred. Much of this auxiliary equipment has
been identified through discussions with industry experts. The auxiliary equipment identified for
added blowout barriers included shear rams, floats or check valve placed in the drill collars near
the bottom of the drill string, a drop-in check valve, a velocity triggered check valve, and a
double valve assembly.
Shear rams can be used to cut through the drillpipe and close the well on top of the
drillpipe if the safety valve fails. The disadvantage of shearing the drillpipe and dropping it to
bottom is that it can make it more difficult to eventually circulate kill mud to the bottom of the
well.
Floats or drill collars are widely used by some operators to make it easier to stab and
close safety valves at the surface. Both flapper and dart type check valves are available. Even if
the check valve leaks, the flow rate is generally reduced enough so that the safety valve can be
successfully closed without cutting out the valve.
Operators may not want to use floats for the
following reasons: (1) extra time is needed to fill
the inside of the pipe when lowering pipe into the
well, (2) higher surge pressures occur when pipe is
lowered into the well, and (3) the shut-in drillpipe
pressure is more difficult to read after taking a
kick.
The drop-in check valve overcomes many
of the objections to a float in the drill collars.
Figure 13 is a schematic of a drop-in check valve.
A sub that will accept a check valve is run in the
drill string near bottom. Just before it becomes
necessary to pull the drill string from the well, the
check valve assembly is dropped into an open
drillpipe connection and pumped to bottom where
it latches into the sub. If the well tries to blowout
during tripping operations, the check valve will
stop the flow and make it easy to stab and close the Figure 13: Schematic of a drop-in check valve.

LSU/MMS WELL CONTROL WORKSHOP SESSION 4

NOVEMBER 19-20, 1996 PRESENTATION 21

surface safety valve as part of the shut-in procedure. In the event wireline work below the check
valve becomes necessary, the drop-in check valve is wireline retrievable.
An example of a velocity triggered check valve is shown in Figure 14. This valve was
designed and tested to a limited extent during the late ?O's by Hughes Tool Company for Shell. It
was lost in the shuffle of buy-outs during the 80's. Prototype valves are again being built by a
new company. Future research will test this valve as part of the MMS project at LSU.
In the double valve assembly,
as seen in Figure 15, we assume that
the lower ball may cut out for high
Threads flow rates but that the flow rate should
be reduced enough to allow the upper
valve to be successfully closed if it is
closed before the bottom valve totally
fails. The bottom valve can also be
Ball
used as a mud saver valve since a back­
Venturi up valve is available.
The problem with this
Passage approach is that it is not well suited to
stabbing valves because of the extra
) weight that must be handled. A single
! stabbing valve for 4.5-in. or 5-in.
drillpipe weighs more than I 00 lbs. To
Figure 14: Velocity triggered Figure IS: Double minimize the weight of a double valve,
check valve. ball valve assembly. one manufacturer is currently working
on a double ball, single body design.
This new valve design is currently being field tested by Amoco near Baton Rouge in the
Tuscalousa trend.

TEST APPARATUS
The test apparatus designed for the data
acquisition associated with testing the DSSV' s is
shown in Figure 16 and Figure 17. The torque
sensor is the primary data generating device used
in the testing of the DSSV's. The sensor was
chosen over a torque wrench because the
information from the sensor is much easier to
incorporate with other data taken during the
experimental tests. The torque sensor is
manufactured in such a way that it is simple to put
the apparatus together quickly. A pneumatic
actuator is used to open and close the DSSV's Figure 16: Test apparatus wltb pump in
with the torque sensor fixed between the valve and background.
the operator. The actuator is designed to be used

 LSU/MMS WELL CONTROL WORKSHOP SESSION 4

NOVEMBER 19-20, 1996 PRESENTATION 21

with valves that open and close through

ninety degrees. The force generated by the
actuator is supplied by air pressure coming
in through a low pressure regulator. The
actuator is easily activated using a shuttle
valve located downstream from the pressure
regulator. The position of the valves is
determined from a signal generated by a
resistance potentiometer fixed to the
actuator. A check system is utiliz.ed to tell if
the valve is closing completely. A
microphone is fixed to the valve and the
Figure 17: Test apparatus torque sensor, operator, and flow noise is amplified and displayed on an
potentiometer oscilloscope next to the valve. The operator
can easily see when the valve has complete closure by looking at the noise generated by the
microphone.
The data is acquired through an analog to digital PC board and stored using LabView
software. Additional sensors to record pressure in the test string also generate signals recorded
by the software during the tests.

TEST PROCEDURES
The testing of the DSSV's was done in two different ways: (I) a static pressure test, and
(2) closing on flow. The static pressure test consists of putting the test piping and equipment in
·the configuration shown in Figure 18. When the test string is pressured to the test pressure set at
the choke, the drill string safety valve is subjected to static pressure. The valve is then closed on
this static pressure and the torque and other data is recorded. The next test point is taken by
increasing the set point of the choke to a higher pressure setting.
The flow test configuration is shown in Figure 19. To test the valve under flowing
conditions, circulation through the test piping is established at the test rate. The valve is closed
on the flow and the torque and other data is recorded. To move to the next test point, the flow is
increased to the next desirable level.
Pneumatic Actuator

\ '\~
v

~hoke .... Choke ..... HT-400

..ii
",,
\\
\ ' .
'\~etyVaM:

00 ­
Computer Computer

Plug Valve

Figure 18: Static test diagram. Figure 19: Flow test diagram.

LSU/MMS WELL CONTROL WORKSHOP SESSION 4

NOVEMBER 19-20, 1996 PRESENTATION 21

FINDINGS
Using the test apparatus and the testing procedures, test results for two commercially
available valves were obtained from two different experiments. The static pressure test was
performed on a TIW two-piece valve and an M&M LiteTorque valve. The flow test was also
performed on these two valves. The static pressure test was performed at l ,000 psi on each of
the valves. The collected data for the two valves are shown side by side in Figure 20 to make a
comparison between the two valve designs. Closing values of twenty-five to thirty-five foot­
pounds of torque for the Lite- Torque valve are three to four times smaller than the 110 to 115
foot-pounds of torque for the two-piece (TIW) valve. Figure 21, the graphs for the 2,000 psi
static tests, shows the LiteTorque valve torque values ranging from twenty to forty foot-pounds
and the two-piece (TIW) valve torque values exceeding 300 foot-pounds. At 3,000 psi, the
LiteTorque valve has torque values that do not exceed fifty-five foot-pounds and the two-piece
(TIW) valve exceed 500 foot-pounds. These graphs are shown in Figure 22. The static pressure
tests of the different valves makes the design differences of the two valves more apparent. The
LiteTorque valve contains a bearing between the stem and the valve body. This bearing reduces
the frictional forces between the valve stem and the valve casing. The two-piece valve based on
a more conventional TIW design does not have the bearing between the stem and the casing and
the frictional forces in this area cause increased torque values to be obtained.
The flow test was performed using flow rates that started at I 00 gallons per minute (gpm)
and increased by 50 gpm up to 350 gpm. Three closing cycles were recorded at each of the flow
rates for the M&M LiteTorque valve and the M&M two-piece (TIW) valve. The results for the
LiteTorque valve and for the two-piece valve are shown in Figure 23 and Figure 24. Although
the flow test data for the two valves differs significantly in value, the condition of the two valves
also varies significantly. The LiteTorque valve was flow tested after being used to calibrate the
test apparatus in a variety of configurations. This particular valve had been used extensively as
the "set up" valve for all of the testing procedures for many months. The two-piece TIW valve
was rebuilt with completely new elastomer seals around the stem and new teflon seals in the
seats. The past use the LiteTorque valve and the recent rebuild of the two-piece TIW valve make
up for the difference in the torque values that were recorded in the data.
CONCLUSION AND RECOMMENDATIONS

The following conclusions can be drawn based on the results obtained in this study to date:

I. Some of the DSSV's tested in this study would not close above 180 gpm with 600 ft.
lbs of torque. A significant chance of valve failure has been observed both in this
study and in the field. Since valve failure and a lack of redundancy corresponds to a
lack of protection for the drillpipe, auxiliary devices should be available in case of
safety valve failure.
2. The results observed for each valve proved to be a function not only of its design and
condition, but also the closing technique of the operator in the test stand.
3. Preparation of a training tape to instruct personnel on the common causes of valve
failure and on the correct valve closing technique is recommended.
4. Additional testing of the current DSSV designs and the refinement of current designs
or the development of additional designs is recommended.

10

LSU/MMS WELL CONTROL WORKSHOP SESSION 4

NOVEMBER 19-20, 1996 PRESENTATION 21

M&M LiteTorque Valve 1000 psi Static Test

35 -- -- --------------- - -- -- - --- -------­

-
1!
30 -------­
25 _ ________.. .. -------­
• •
~ 20 • • ____.._ • •
°'
:I ••
15 _________.__ _:_____ _ _ _ _ __
------.­
• •
-
!:
{:. 10

5 --­
0
0 0.2 0.4 0.6 0.8 1

Percent Closed

M&M 2-Piece Valve (TIW) 1000 psi Static Test

120 ­

100 --r;:<------
rr•tA'5t ••++• +• + +•• + •
____________ +

--------------------------­ ---- ­ -----­ ---------~~--~------ --------­

•
0 - - - - - - - - - - - - - - - - - - - _ _ _ ; ______ ___J

0 0.2 0.4 0.6 0.8 1

Percent Closed

Figure 20: I 000 psi Static test results.

II
LSU/MMS WELL CONTROL WORKSHOP SESSION 4 .
NOVEMBER 19-20, 1996 PRESENTATION 21

M&M LiteTorque Valve 2000 psi Static Test

45 --------------­
40 ----~------- -----------------------------------­

j
35 - - - - - - ­
•
l
• --------+
• • •
!
~!: 15_____,,______________________________________
10 - - - - - ­
5 ------------------------­
0
0 0.2 0.4 0.6 0.8 1
Percent Closed

M&M 2-Piece Valve (TIW) 2000 psi Static Test

...............-.-.. •
• ••• ••
350 -----·

300 ___ •
-
.!
250 •
• '
-~=II>
200

!:
~ 100
150
•
50
o.
•
0 0.2 0.4 0.6 0.8 1
Percent Closed

Figure 21 :2000 psi Static test results.

12

LSUfMMS WELL CONTROL WORKSHOP SESSION 4

NOVEMBER 19-20, 1996 PRESENTATION 21

M&M LiteTorque Valve 3000 psi Static Test

•
-­ ···- -··-·-·----­ .••
••
• •• • ·.-- - - ­
•.-----~-
•
10 ~

0.2 0.4 0.6 0.8 1

Percent Closed

M&M 2-Piece Valve (TIW) 3000 psi Static Test

600

•• ~-·······
500 --- ··---,.

j 400

-
~
300 --·
•
200
•
100
•
0 •
0 0.2 0.4 0.6 0.8 1
Percent Closed

Figure 22: 3000 psi Static test results.

13
LSU!MMS WELL CONTROL WORKSHOP SESSION 4
NOVEMBER 19-20, 1996 PRESENTATION 21

M&M UteTorque Valve Flow Test

•• •

-.­
100 _______________ _..._ ___ _J_I_

0 -----~------
•
0 50 100 150 200 250 300 350
Flow Rate (gpm)

Figure 23: LiteTorque flow test results.

M&M 2-Piece (TIW) Valve Flow Test

200

-.. .a
180
160
140
120
•
­
al:
Ill
:I
100
80
!! 60
{!.
40
20
0
0 50 100 150 200 250 300 350

Flow Rate (gpm)

Figure 24: Two-piece flow test results.

14
LSU/MMS WELL CONTROL WORKSHOP SESSION 4
NOVEMBER 19-20, 1996 PRESENTATION 21

REFERENCES
I. Hauser, William: "Minerals Management Service (MMS) Review of Blowout
Preventer (BOP) Testing and Maintenance Requirements for Drilling Activities on
the Outer Continental Shelf (OCS)," paper presented at the !ADC Well Control
Conference of the Americas, 1994.
2. Tarr, Brian A.: "Research Targets Drillstring Safety Valve Improvements," Gas Tips
(Spring, 1996).

15