Professional Documents
Culture Documents
Stage:
Page: 78
Total Pages: 10
Summary
An inflow (commonly known as negative pressure) test is an important step in proving well integrity for some well operations.
Although the concept is straightforward, there are no standard
procedures for conducting and interpreting inflow tests. Recent
industry experience has shown the potential for these tests to be
misinterpreted, with disastrous results. This paper describes proper
quantitative interpretation of both successful and unsuccessful tests
demonstrated on two full-scale test wells.
Tests with and without leaks are simulated in two vertical wells
with 2,971- and 5,884-ft depths. Test results are interpreted quantitatively to distinguish successes from failures. In addition, the
impacts of leak rate and the differential pressure between the leak
source and well are investigated. The impacts of gas as a leak fluid,
different well volumes, different surface-piping arrangements, and
measuring with different-sensitivity tanks are also shown.
The success of a test should be verified with both flow and
pressure checks. For the flow check, the trend between bleed-off
pressure and volume of bleedoff can be predicted on the basis of
fluid compressibility, tubing expansion, and drainage from surface
piping. Deviations from that trend will identify the onset of a
leak. Early detection and control of a high-rate leak can reduce
the formation-fluid volume entering the well and the consequent
risks. Detecting a low-rate leak is difficult if the test duration is
too short. Therefore, a 30-minute test duration was evaluated for
detection of low-rate leaks.
The fundamental differences between a leak and no leak are
demonstrated conclusively with quantitative results. Approaches
for overcoming complications that interfere with conclusive test
results are explained and demonstrated. The demonstration test
results and the proposed interpretation methods are applicable to
vertical wells. Drilling-fluid properties, multiple fluids, and thermal expansion can also influence results, but were not investigated in this study. Nevertheless, these results should provide a
basis for improving field procedures and minimizing the risk of
misinterpreting a leak as a successful test.
Introduction
An inflow test is now mandatory before unlatching the blowout
preventer at any point in the well for wells in US federal waters
that use a subsea blowout-preventer stack or wells with mudlinesuspension equipment (US Securities and Exchange Commission
2012). However, a particular design for running and interpreting
the tests is not specified in the regulations. Nelson and Guillout
(1990) have briefly explained the concepts for interpreting a negative pressure test. A document on the International Association of
Drilling Contractors website is focused on purposes of running a
negative pressure test and planning issues and concerns rather
than the test procedure and interpretation (Armistead 2011). The
American Petroleum Institute (API) recently published a fairly
Experimental Setup
Leak and no-leak scenarios were tested in two vertical wells.
Fresh water was used as the drilling fluid inside the wellbore and
also as a leak fluid. The other leak fluid used was natural gas. In
these controlled systems, any potential trapped gas from the previous experiments was flushed out before running the tests. However, an elevated surface line was added to the LSU#1 well to
investigate the effect of trapped gas in complicating test results
for some of the experiments, as explained later in this paper. Fig.
1 is a schematic of the experimental setup on the LSU#1 well.
This well was used to run the majority of the tests with liquid
leaks and also to run both gas-leak tests. The well is 2,971 ft deep
and has been completed with an 85/8-in. casing and two concentric
27/8- and 51=2-in. tubing strings. The bottoms of both strings are
closed to the 85/8-in. casing annulus. The leak source was
designed to be at the top of the 51=2-in. tubing string at the surface.
The 27/8-in. internal tubing string was used as the test string to
bleed off the wellbore fluid during the tests. A check valve is
located in a side-pocket mandrel near the bottom of the 27/8-in. internal tubing string, and it allows fluids to enter the tubing string
from the 51=2-in. tubing string, but does not allow fluid movement
**
This paper (SPE 163442) was accepted for presentation at the SPE/IADC Drilling Conference
and Exhibition, Amsterdam, 57 March 2013, and revised for publication. Original manuscript
received for review 8 October 2013. Revised manuscript received for review 18 February
2014. Paper peer approved 26 February 2014.
78
Stage:
Page: 79
Total Pages: 10
SMALL
PUMP
PRESSURE GAUGE
PLUG VALVE
MEASURING
TANK
2 -in.
Tubing
78
P2
AUTOMATIC
CHOKE
5-in.
Tubing
P3
CONSTANTPRESSURE
LOOP
85 8-in.
Casing
MD = 2,750 ft
TRIPLEX
PUMP
MUD
PITS
MD = 2,791 ft
Fig. 1Schematic of the experimental setup on the LSU#1 well. (MD 5 measured depth.)
Measuring tank
LSU#1 well
Experimental Procedure
The inflow-test experiments on LSU#1 were performed with the
following procedure:
1. Close the bleed-off valve and start pumping water into the
27/8-in. tubing (test string) at a slow rate to build up to approximately 2,000 psi of trapped pressure. This will simulate the
trapped pressure caused by displacement of the drilling fluid from
the test string by a less-dense fluid in a real case.
2. Shut down the pump and wait for pressures to stabilize.
3. For tests conducted with water as the leak fluid, start the triplex
pump and adjust the automatic choke to provide 300- to 700-psi pressure upstream of the choke (P2 in Fig. 1). Isolate this pressure from
the well with the gate valve on the 51/2-in. tubing (annulus) for a noleak test. For a test simulating a downhole leak, open the gate valve
and adjust the needle valve to simulate the leak path. Because the
79
Stage:
Page: 80
Total Pages: 10
PLUG VALVE
SMALL
PUMP
1-in.
Tubing
PERFORATED
TUBING
3-in.
Tubing
95 8-in.
Casing
TRIPLEX
PUMP
MUD
PITS
MD = 5,816 ft
MD = 5,852 ft
P1
MEASURING
TANK
MD = 5,882 ft
MD = 5,884 ft
hydrostatic pressures in the test string and the annulus are similar for
the tests in which water is used as the leak fluid, the surface pressure
on the annulus will act as the pressure in excess of the hydrostatic
pressure in the test string that would cause flow through a leak. For
tests in which the leak fluid was gas, the annulus is filled with gas
and the surface pressure applied to the annulus is increased to provide a downhole pressure at the check valve greater than the hydrostatic pressure in the test string.
4. Slowly open the bleed-off valve to the measuring tank and
continuously monitor and record the volume of bleedoff vs. the
pressure on the test string (P1 in Fig. 1) until the pressure drops to
essentially zero (flow check). Compare the actual volume of
bleedoff with the expected bleed-off volume caused by fluid compressibility and surface-line drainage.
Successful flow test: If the measured bleed-off volume is
as expected and if the flow stops accordingly, it is an indication of
a successful flow check. Continue the flow check for an additional
30 minutes after the flow has stopped to make sure that there is no
additional flow. Then, proceed with Step 5.
Failed flow test: If the return flow does not stop and/or
exceeds the expected volume, then there is an indication of a leak,
the flow check is a failure, and Step 5 should be implemented.
5. Fully close the bleed-off valve on the return line and observe
for a pressure buildup (pressure check) for 30 minutes.
Successful pressure check: After a successful flow check
has been performed, if there is no pressure buildup, then the test is
confirmed as successful.
Measuring tank
LSU#2 well
location
1
dh da . . . . . . . . . . . . . . . . . . . . . . . . . . 2
E
1
da dh ; . . . . . . . . . . . . . . . . . . . . . . . . . . 3
E
qD d
; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2h
and
da
qD d
; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
4h
where pD is differential pressure across the pipe wall, d is the average pipe diameter, and h is the pipe thickness. The volume
May 2014 SPE Production & Operations
Stage:
2,500
2,500
2,000
2,000
Page: 81
Total Pages: 10
20
1,500
1,000
10
Volume, gal
1,000
Pressure, psi
1,500
Volume, gal
Pressure, psi
15
5
500
500
0
0
10
20
30
40
50
60
70
Time, minutes
0
0
10
20
30
40
50
60
0
70
Time, minutes
Fig. 6No-leak test on the 27/8- and 51=2-in. tubing strings of the
LSU#1 well.
Calculations performed for this test are provided as an example in Appendix B. Calculated fluid expansion for the test in Fig.
5, where only the tubing was pressurized, was 4 gal, which was
based on a constant water-compressibility factor (cf) of 3106
psi1. Furthermore, the volume change because of hoop strain was
calculated to be 0.48 gal, while the axial change in volume was
found to be only 0.05 gal. Therefore, the total bleed-off volume
was expected to be approximately 4.5 gal. The actual bleed-off
volume (pit gain) for this test was slightly more than 4.6 gal. Fluctuations on the order of 0.4 gal were observed in the pit-gain plot
during the flow check, whereas there was actually no flow during
that period, indicating that this difference was within the repeatability and sensitivity limits of the pit-level indicator used for this
test. The bleed-off volume for another test with similar setup was
4.3 gal, again within the measurement limits. It should be noted
that the fluid was bled off into a 50-gal measuring tank for both of
these tests, and, therefore, these measurements were more sensitive than would be possible with the use of regular rig equipment.
Effect of Wellbore Volume on Bleed-Off Volume. A no-leak
test was also conducted with a larger wellbore volume, where fluid
in both the 27/8- and 51/2-in. tubing strings was compressed to
2,020 psi (Fig. 6). The estimated fluid compressibility in this case
was 15.35 gal, proportionately larger than for the test of the tubing
only. The total return volume, including pipe-expansion effects,
was estimated to be approximately 19.1 gal. The actual bleed-off
volume (pit gain) for this test was 18.34 gal. A 30-minute flow
check followed by a 30-minute pressure check confirmed that there
was no leak. Note that the step changes in the actual pit-gain plot
were because the minimum change that the pit-level sensor could
detect was on the order of 1 gal in the 50-gal measuring tank. The
accuracy of the readings decreased substantially by switching to a
larger-sized measuring tank. A similar test measured 4 gal less
return volume than expected with a 10-bbl measuring tank, which
reinforces the advantage of selecting a measuring-tank size that
allows accurate measurement of the predicted volume.
Effect of Surface Piping on Bleed-Off Volume. The effect of
a longer, elevated return line (Fig. 2) on the test results was investigated in a second set of tests. The total return-hose volume was
13.7 gal, which was much smaller than the 60.7-bbl wellbore volume. Therefore, the expected compressibility effect of the surface
return line was very small. However, 28 ft of the return hose, with
a 4.5-gal capacity, was inclined downward toward the measuring
tank (Fig. 2). It was likely that this part of the hose would drain
fully during a flow check. Water in another 10 ft of the hose
upstream of this part could also potentially drain another 1.6 gal.
Therefore, to obtain an accurate estimation of the effect of surface
lines on the test results, only the return line was pressured up to
the test pressure, keeping the gate valve on the tree closed. The
water volume used to fill and pressure up the hose was measured
accurately and compared with pit gain after opening the bleed-off
valves. The test started by injecting 6.25 gal of water until the pressure on the line reached 2,140 psi. Once the pressure stabilized,
DVh
1 eh 2 1di2
L . . . . . . . . . . . . . . . . . . . . 6
1; 029:41
and
DVa
ea di2
L; . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1; 029:41
where DVh is volume change because of hoop strain, DVa is volume change because of axial strain, di is tubing internal diameter,
and L is tubing length. Therefore, the total volume change
because of fluid and metal expansions (DVT) was calculated as
DVT cf VDp e2h 2eh ea
di2
L: . . . . . . . 8
1; 029:41
81
30
Stage:
Page: 82
Total Pages: 10
180
2,500
2,000
160
15
1,000
10
Tubing pressure (psi)
Leak differential pressure (psi)
Actual pit gain (gal)
Fluid expansion (gal)
Fluid and pipe expansion (gal)
Surface line effect (gal)
500
10
15
20
100
Tubing pressure (psi)
Injected volume (gal)
Actual pit gain (gal)
Fluid expansion (gal)
Fluid and pipe expansion (gal)
1,000
80
60
40
0
0
120
1,500
500
140
25
20
0
Time, minutes
Volume, gal
20
Pressure, psi
1,500
2,000
Volume, gal
Pressure, psi
25
50
100
150
Time, minutes
Fig. 7No-leak test on the 2 /8- and 51=2-in. tubing strings of the
LSU#1 well with elevated return line.
verifying that there were no leaks, the bleed-off valve was opened
and the pressure bled off rapidly. The final volume of bleedoff was
5.9 gal. The difference between the volume pumped and the bleedoff was approximately 0.35 gal, which was larger than expected,
but still within previously measured repeatability limits. This test
was used to demonstrate that the effect of this longer, elevated surface line would be approximately 6 gal more bleed-off volume than
expected on the basis of system compressibility. This effect was
considered in the interpretation of the subsequent tests on the same
well with a similar surface lineup. It should be noted that the pressure vs. volume pumped for these tests was very nonlinear because
of the need to fill the line before pressure increased and to drain
fluid after bleeding off the pressure. Although the pressure vs. volume pumped in was not linear, the volume bled off during a successful test still equaled the volume pumped.
Fig. 7 shows results of a no-leak test in which the fluid in the
27/8- and 51=2-in. tubings and in the return hose was compressed to
2,000 psi. The 6-gal expected effect of draining the surface line
was added to the estimated wellbore fluid and tubing compressibility of 19 gal to predict a total pit gain of 25 gal for this case.
This expected bleed-off volume matched well with the actual
bleed-off volume (pit gain) of 24 gal. The 1-gal difference is
larger than the previous measurement sensitivity, but is not considered significant. A flow check confirmed that there was no leak, but
no attempt was made to conduct a pressure check because the purpose of this test was primarily to see how return-line volume and
height can complicate the test interpretation.
The last no-leak test was performed on the LSU#2 well, which
is deeper and has a larger casing, as shown in Fig. 3. The volume
of the surface lines for this well was more than 75 gal. Therefore,
the effect of the surface lines on interpretation of the inflow-test
results on this well was expected to be significant. However, as a
result of a leaky gate valve on the tree of this well at the time these
tests were conducted, it was not possible to pressure test the surface
35
2,500
30
2,000
1,500
20
15
1,000
500
Volume, gal
Pressure, psi
25
10
5
0
0
10
20
30
40
50
60
Time, minutes
Fig. 9Leak test on the 27/8-in. tubing string of the LSU#1 well.
82
2,500
Stage:
30
Page: 83
Total Pages: 10
10
2,500
1,000
500
10
Pressure, psi
15
Volume, gal
Pressure, psi
20
1,500
2,000
1,500
8
7
6
5
4
1,000
Volume, gal
25
2,000
3
Return valve closed, gradual change
500
2
1
0
0
10
15
20
25
30
20
40
Time, minutes
60
80
Time, minutes
Fig. 10Leak test on the 27/8- and 51=2-in. tubing strings of the
LSU#1 well.
system demonstrate the basic expected leak behavior (i.e., a detectable and continuing flow when pressure is bled off and a pressure
buildup on the test string when it is shut in).
Detection of Small, Low-Rate Leaks. The low, but still significant, flow rates relative to the wellbore volumes in the preceding
examples made it easy to differentiate a leak from no leak. However, interpretation of the test in the case of a very small leak is
more difficult. Fig. 11 shows an example of a low-rate leak
designed to be on the order of 0.1 gal/min. The test started by partially opening the return line, which caused a rapid decrease in the
tubing pressure. The flow rate, measured manually with a quartsized mud cup 3 minutes after the start of the test, was 0.13 gal/
min. The bleed-off valve was fully opened 6 minutes into the test,
but the return flow was more like drops than a continuous flow.
Flow declined to 0.06 gal/min after 13 minutes and to less than
0.04 gal/min at 33 minutes from the start of the test. The well was
shut in after 42 minutes when the flow was barely noticeable.
Except for a sharp increase in pit level when pressure was initially
bled off the well, the pit-level indicator did not show any change
during the 40-minute duration of the flow check. After the well
was shut in, the pressure started to build up at a very slow rate.
The pressure built to approximately 50 psi over a period of
approximately 40 minutes. Despite the low leak rate, the volume
of bleedoff was twice as large (9.6 gal) as expected (4.4 gal), and
the 30-minute pressure check indicated a leak. It is notable that
shorter test periods might have allowed a false conclusion that no
leak was occurring. Conversely, comparison of the actual to the
calculated bleedback volume, very careful visual monitoring of
the return flow, use of an appropriately sized measuring tank, and
use of minimal surface-return piping all contributed to the detection of this leak.
Fig. 12 shows the results of a leak test on the 27/8-in. tubing
string by use of the longer, elevated return line. The well was shut
20
1,500
15
1,000
10
500
0
0
10
20
30
40
25
2,000
50
Time, minutes
Fig. 12Leak test on the 27=8-in. tubing string of the LSU#1 well
with elevated return line.
Pressure, psi
Volume, gal
Pressure, psi
2,000
30
2,500
20
1,500
15
Tubing pressure (psi)
Leak differential pressure (psi)
Actual pit gain (gal)
Injected volume (gal)
Fluid expansion (gal)
Fluid and pipe expansion (gal)
Surface line effect (gal)
1,000
500
Volume, gal
25
2,500
10
5
0
0
0
20
40
60
80
100
120
Time, minutes
Fig. 13Leak test on the 27/8- and 51=2-in. tubings of the LSU#1
well with elevated return line.
83
Stage:
2,500
250
2,500
2,000
200
2,000
1,500
150
Page: 84
Total Pages: 10
35
500
0
0
50
100
50
25
1,500
20
Tubing pressure (psi)
Leak pressure (psi)
Actual pit gain (gal)
Fluid expansion (gal)
Fluid and pipe expansion (gal)
1,000
15
Volume, gal
Pressure, psi
100
1,000
Volume, gal
Pressure, psi
30
10
500
5
0
150
Time, minutes
0
0
10
20
30
40
50
60
70
Time, minutes
the 27=8-in. tubing was evidently related to the leak rate. The leak
rate in this test was slightly less than 0.2 gal/min compared with
the average leak rate of 0.75 gal/min in the previous test. Presumably, this slower leak rate required a longer time to refill the return
line. Approximately 15 minutes after starting the pressure check,
the pressure started to build up in this test at a very low rate,
reaching roughly 100 psi in 35 minutes while continuing to build
up toward the leak differential pressure.
Comparison of leak tests for cases with and without a surface
line indicates that the pressure-check phase of an inflow test is
more sensitive and conclusive if the pressure gauge is located on
the well tubing and the valve used to shut in the well is on the
well close to the gauge, as was the case for Test 6 (Fig. 10) on
well LSU#1. In contrast, conducting a similar test with elevated
surface lines (Test 9, Fig. 13) significantly reduced both the magnitude and the rate of pressure buildup.
A liquid leak was also simulated on the LSU#2 well (Fig. 14).
As already explained, the large well and surface volume of this
system makes it more difficult to detect a leak in an inflow test.
The leak rate for this test was designed to be 0.6 gal/min, which is
a fairly low rate relative to the well volume of more than 400 bbl.
The test preparation began by pressuring up the well to more
than 2,000 psi with 155 gal of water. After the surface pressure
stabilized, the bleed-off valve was opened to begin the inflow test
by bleeding off pressure for a flow check. At a time of 54 minutes,
the pressure was almost completely bled off, but the volume that
bled back exceeded the total volume required to pressurize the
system, indicating the likelihood of a leak. The return flow continued at a rate of at least 0.6 gal/min, which was equal to the known
leak rate, and the total pit gain at 90 minutes was 199 gal, which
was approximately 1 bbl (28%) more than the expected return
volume. Consequently, the flow check was considered a failure,
and the well was closed in to do a pressure check. However, it
took an additional 38 minutes to see a pressure change. After that,
the pressure started to increase gradually up to 100 psi at 58
minutes after closing in the well. The 38-minute elapsed time
without change in pressure is likely a result of the drained volume
of the surface lines that had to be filled before pressure would
build up. It shows that for wells with large surface volume, even a
30-minute pressure check may not be long enough to observe a
pressure buildup. However, a warning from a failed flow check in
this test justified the longer pressure-check period. Additional evidence for occurrence of a leak was the difference between the
injected volume during the positive pressure test, equivalent to
the bleed-off volume following a positive pressure test, and the
pit gain during the inflow-test flow check. A time-saving alternative to extending a pressure check for an unknown time could be
to install a pressure sensor below a valve at the top of the test
string to allow pressure to be monitored in a liquid-filled system
that is isolated from the surface piping.
potential for gas to leak into the tubing during an inflow test on
the 27/8-in. tubing string (Fig. 15). The check valve between the
injection annulus and the tubing allowed for maintaining the gas
pressure below the wellbore pressure at the start of the test without water from the tubing entering the annulus. Similar to the liquid tests, trapped pressure on the tubing was bled off during the
flow-check portion of the inflow test until the tubing became
underbalanced relative to the annulus at 2 minutes into the test.
This allowed gas flow through the check valve, simulating a gas
leak near the bottom of the tubing. The relatively rapid gas flow
once an underbalanced condition was reached caused the leaksource pressure, equivalent to a formation pressure, to decrease
during the flow check, as seen in Fig. 15. It was not practical to
reduce the wellbore pressure to zero in this case because the
return flow rate, and therefore the leak rate, would be excessive.
The return flow rate (approximately 2.7 gal/min) and the pit gain
(25 gal more than expected) were both clearly indicative of a substantial leak within a few minutes after reaching the underbalanced condition. A gradual increase in wellbore pressure was
observed even during the flow check. This was likely caused by
rapid influx and subsequent expansion of gas in the small wellbore
volume that caused additional backpressure at the surface because
of increased flow through the fixed opening of the return valve.
Surprisingly, when the return valve was closed just before 10
minutes, the pressure-buildup rate decreased. A repeat gas-leak
test with the same setup confirmed occurrence of this behavior.
Subsequently, surface pressure began increasing more rapidly,
apparently as the gas-migration velocity increased. Detection of
the gas leak at a moderate rate into the small tubing volume with
minimal surface piping was relatively easy from both the flow
check and the pressure check.
Stage:
Total Pages: 10
2,000
2,500
Leak test
No-leak test
Expected expansion
Expected expansion
2,000
Page: 85
1,500
1,000
Return valve closed
0
5
10
15
20
25
No-leak test
1,000
m surface lines
Drainage from
500
500
Leak test
1,500
30
Fig. 16Expected and actual tubing pressure vs. pit gain in the
LSU#1 well (27/8- and 51=2-in. tubing strings).
underbalanced, and the leak started in the leak test when the tubing pressure dropped below 700 psi. Therefore, similar trends for
the leak and no-leak tests and the prediction were expected at
pressures greater than 700 psi. Although these trends were not
identical, they had similar slopes, indicating that the compressibility and well volume in each test was as expected. In addition, the
total-recorded pit gains at 700 psi for both tests were almost identical and were close to the expected bleed-off volume. At lower
pressures, the trend for leak test deviates from the expected trend,
the leak volume exceeds the predicted volume, and, ultimately,
the pressure builds up after the well is shut in on the leak. As seen
in this plot, the leak test deviated from the expected plot at
approximately 19-gal pit gain, but the well was kept open until it
was finally shut in after a 29-gal pit gain. This example shows
how deviation from the expected trend could help in early detection of the leak before having an excessive pit gain, and how a
pressure buildup after shut-in can also be shown as a leak indication on this same plot. Fig. 17 shows the same approach applied
to a test on the LSU#1 well with an elevated return hose. The leak
differential pressure in this case was 300 psi, the leak rate was
only 0.2 gal/min, and drainage from the elevated return hose
exceeded the volume predicted solely on thebasis of the well volume (dashed blue line) for both the leak (red line) and no-leak
(green line) tests. However, the volume bled in the no-leak test
matched closely with the prediction (solid blue line) that
accounted for the total volume because of both well compressibility and surface lines. Nevertheless, the drainage from the surface
lines masked the early flow from the leak. The leak test did not
deviate from the no-leak test or the prediction until the volume of
bleedoff exceeded the predicted total bleed-off volume. However,
the leak bleed-off rate continued, whereas the bleedoff from the
no-leak test stopped when the line had drained. Consequently, the
leak bleed-off volume exceeded the prediction. The pressure
increase at the right side of the plot is indicative of the pressure
buildup after shut-in, confirming the presence of a leak. Similar
behavior was observed for the leak test on the LSU#2 well, and
applying this approach to higher-rate leaks gave even more conclusive early detection of the existence of the leak. This example
also shows a potential practical advantage of conducting a positive pressure test and measuring the bleed-off volume vs. tubing
pressure, if volumes and fluid compressibilities are similar or corrected for, as a basis for predicting the bleed-off behavior of a noleak test.
Conclusions
1. The traditional understanding of a successful inflow test (i.e.,
that the return flow stops after pressure is bled off during the
flow-check period and that no pressure buildup is observed during a subsequent shut-in and pressure check) is exactly correct
for a test under ideal conditions (i.e., negligible surface-piping
volumes and adequate test duration).
0
0
10
15
20
25
30
Fig. 17Expected and actual tubing pressure vs. pit gain in the
LSU#1 well (27/8- and 51=2-in. tubing strings with elevated return
line).
85
should be terminated with the well shut in as soon as a potential leak is detected.
10. The pressure-check phase of an inflow test is most conclusive
if the pressure gauge is located on the well tubing and the
valve used to shut in the well is on the well close to the gauge.
This minimizes the volume of the surface lines that a leak
would have to refill and minimizes the time required to
observe a pressure buildup.
11. Instrument errors and inaccuracy, especially in pit-gain measurements of bleed-off volumes, can be a significant source of
uncertainty in interpreting test results. The return flow should
be visually monitored continuously during the flow-check period as the most-positive indication of whether a leak is
occurring.
Nomenclature
cf fluid compressibility factor, Lt2/m, psi1
d pipe diameter, L, in.
di internal diameter, L, in.
E elasticity modulus, m/Lt2, psi
h pipe thickness, L, in.
L length, L, ft
p pressure, m/Lt2, psi
pD differential pressure, m/Lt2, psi
Dp pressure change, m/Lt2, psi
V volume, L3, bbl [gal]
DV volume change, L3, bbl [gal]
DVa axial volume change, L3, bbl [gal]
DVT total volume change, L3, bbl [gal]
DVh hoop volume change, L3, bbl [gal]
ea axial strain, dimensionless
eh hoop strain, dimensionless
da axial stress, m/Lt2, psi
dh hoop stress, m/Lt2, psi
t Poissons ratio, dimensionless
Acknowledgments
The authors would like to acknowledge the Louisiana State University (LSU) Petroleum Engineering Research and Technology
Transfer (PERTT) laboratory for allowing and supporting these
full-scale experiments. We would especially like to thank the student workers, colleagues, and staff who helped in conducting the
experiments. We also acknowledge the contribution of Brian F.
Towler, professor at the University of Wyoming, in advocating
the relevance of understanding and demonstrating the proper
interpretation of negative pressure tests.
References
API RP 96, Deepwater Well Design and Construction. 2013. Washington,
DC: API.
Armistead, G. 2011. Negative Pressure Test. IADC Well Control Committee Meeting Minutes, Chevron, Houston, Texas (10 February 2011).
Craft, B.C., Hawkins, M., and Terry, R.E. 1991. Applied Petroleum Reservoir Engineering, second edition. Englewood Cliffs, New Jersey: Prentice-Hall.
Nelson, E.B. and Guillout, D. ed. 1990. Well Cementing. Sugar Land,
Texas: Schlumberger.
Peng, L.-C. 2003. An Interpretation on Pressure Elongation in Piping Systems, Peng Engineering. http://www.pipestress.com/papers/PressureElong.pdf (accessed 18 October 2012).
Roylance, D. 2001. Pressure Vessels. Mechanics of Materials course module (Sec. 311, Fall 1999), Massachusetts Institute of Technology,
Cambridge, Massachusetts (23 August 2001).
US Securities and Exchange Commission. 2012. Oil and Gas and Sulphur
Operations in the Outer Continental Shelf. Final Rule, 30 CFR Part
250, US SEC, Washington, DC (22 August 2012). Federal Register 77
(163): 50891.
86
Stage:
Page: 86
Total Pages: 10
Appendix A
See Table A-1.
TABLE A-1SUMMARY OF THE EXPERIMENTAL
CONDITIONS
Leak
Test Figure(s) Well Fluid
1
2
3
4
5
6
7
8
9
10
11
5
6, 16
7, 17
8
9
10, 16
11
12
13, 17
14
15
1
1
1
2
1
1
1
1
1
2
1
Test
Type
Water
No-Leak
Water
No-Leak
Water
No-Leak
Water
No-Leak
Water
Leak
Water
Leak
Water Low-rate leak
Water
Leak
Water
Leak
Water
Leak
Gas
Leak
Strings Elevated
Tested Return
(in.)
Line
27=8
27=8; 51=2
27=8; 51=2
All
27=8
27=8; 51=2
27=8
27=8
27=8; 51=2
All
27=8
No
No
Yes
Yes
No
No
No
Yes
Yes
Yes
No
Appendix B
Tubing external and internal dimeters were 2.875 and 2.441 in.,
resepectively, with a total lentgh of 2,750 ft. Therefore, the total
compressible fluid volume was calculated as 668.5 gal. Fluid
compressibility according to Eq. 1 was then calculated as
DV 3 106 668:5 0 2000 4:0 gal:
B-1
The hoop and axial expansion of the tubing was also calculated with Eqs. 6 and 7, respectively, considering a conversion
factor from barrels to gallons as
DVh
and
DVa
projects, and assisted with the development of internal wellcontrol policies and procedures. Bourgoyne holds a BS degree
and an MS degree in petroleum engineering from LSU.
John Rogers Smith is a Campanile Professor and an associate
professor in the Craft and Hawkins Department of Petroleum
Engineering at LSU and is a registered, professional petroleum
engineer in Louisiana. He has more than 20 years of industry
experience in drilling, production, research, and reservoir
activities. At LSU, Smith teaches courses on drilling engineering
Stage:
Page: 87
Total Pages: 10
and prevention of oil and gas blowouts as well as other drillingand production-related topics, and his research interests are
focused on drilling and well control. Smith holds a BS degree in
electrical engineering from the University of Texas at Austin
and an MS degree and a PhD degree in petroleum engineering from LSU. He is a Lifetime Member of SPE, having volunteered to the Society for more than 40 years. Smith was an SPE
Distinguished Lecturer on the challenges of deep drilling, and
continues to be active as a volunteer, speaker, presenter, and
author.
87