Methods of Detecting and Locating

Tubing and Packer Leaks in the Western

Operating Area of the Prudhoe Bay Field
C.M. Michel, SPE, BP Exploration

Summary Tool Selection

Evaluation methods have been developed to detect cases of tubing/ The optimal tool string depends in part on the magnitude of the leak.
annulus communication. Temperature, spinner and noise logs, as At the lower range of leak rates (or leaks behind another string of
well as fluid level detection equipment, are used under a variety of pipe) the noise log is often the most sensitive and the best tool for
flow conditions. Step-wise procedures are provided. pinpointing a leak. The temperature log works over a broad range
of leak rates, butis usually not as sensitive as the noise log. A spinner
Introduction can be used for rates above the detection limit of the tool (usually
about 6 feet/minute velocity). A philosophical approach to tool
Production in the Prudhoe Bay Unit Western Operating Area selection is provided in Fig. 2.
(WOA) began in June 1977. Wells now flow naturally or with the
aid of gas lift. Rates vary from 100 to 10,000 BOPD or more and gas/ Naturally Flowing Well Leak Detection
oil ratios (GOR) from 600-10,000 scf/STB. Water cuts have in-
Annulus pressure readings are taken on a daily basis. Any unex-
creased mainly in the waterflood areas and can approach 100%. The
plained increase in annulus pressure is cause for investigation. An
combination of water and 12 % carbon dioxide in the dissolved solu-
attempt is made to bleed off the pressure. The initial and final pres-
tion gas forms carbonic acid, corroding tubulars despite corrosion sures along with the amount and type of fluid bled is documented.
inhibition treatments. A typical completion is shown in Fig. 1. Tub- If the pressure returns, then troubleshooting efforts begin.
ing sizes range from 3- 1/2" to 7".
In the early years offield life few workovers were required. Most Leak Detection Method. An acoustic sounding of the liquid level
of these were to replace defective packers and thermocase tubing. in the annulus is the first step in leak detection. After non-gas lift
An increasing number of mechanical failures of tubular components wells are completed they are left with a full liquid column in the an-
as well as worsening corrosivity of produced fluids has significantly nulus. Assuming no wireline work has been subsequently done to
increased the occurrences of tubing - annulus communication and allow wellbore gas to enter the annulus since the completion, a liq-
corresponding workover requirements. In 1988-1989, 50 cases of uid level significantly below the wellhead is further confirmation of
tubing - annulus communication were found in the WOA. Of these, a leak. Further, because the leak must be at or below the liquid level,
5 were permanently repaired with wireline techniques and 3 were the level may provide an indication of the location of the leak. In one
temporarily repaired until a workover could be implemented. The case the entire annular fluid volume to the packer was voided.
remaining 42 wells were shut in and worked over without attempt-
ing remedial measures. Determination of the Leak Rate by the Non-Ideal Gas Law. The
This paper discusses methods used at Prudhoe Bay for identifying rate of increase in the annulus pressure alone is not sufficient in de-
problem wells and determining the exact location of the leak in natu- termining the severity of the leak as it is a function of the leak rate
and the compressibilty of the annular fluids. For example, even a
rally flowing, gas lifted, and injection wells.
very slow transfer of wellbore fluids into a liquid-packed annulus
results in a dramatic increase in annulus pressure. On the other hand,
Traditional Methods for Leak Detection the pressure of an annulus having a deep liquid level responds de-
Numerous articles are available related to temperature logging for ceptively slow to a significant influx of wellbore fluids,
production logging purposes.l· 2 Noise logging articles have fo- With the well shut-in, the annulus pressure is bled after noting the
cused on production logging or finding leaks behind casing. 3 ,4 This initial fluid level. For low GOR wells: the fluid entering the well is
work is useful to the extent that the basic concepts of tool response mostly liquid. The rise in annulus fluid level with time can be con-
still apply in tubing-annulus communication troubleshooting. verted to barrels per minute (BPM). For high GOR wells: mostly gas
The technical literature on identifying tubing leaks consists most- enters the annulus and the liquid level remains fairly constant. The
ly of mechanical devices run on wireline. The devices form a seal system can be modeled as a concentric cylinder of constant volume
to the tubing wall and are pumped down to the leak. Two such exam- V bounded by the wellhead at the top, the liquid level at the bottom,
the tubing on the inner radius, and the casing on the outer radius.
ples are provided in the references. 5 ,6 It is unlikely these devices
The influx of gas into the inner annulus from initial to final condi-
would work in North Slope wells due to the restriction at the sub sur-
tions can be quantified by using the non-ideal gas law:
face safety valve.
_ 188 (P/zr- P/zi)V
Leak Detection Methods at Prudhoe Bay Qleak - (460 + T) ~t
Leak determination techniques for Prudhoe Bay wells have been de- with P expressed in psi, V in barrels, ~t in minutes, T in OF, and Qleak
veloped which primarily employ electric line logging. The strategy in standard cubic feet per minute (SCFM).
used depends on the condition of the well and whether the well is
naturally flowing, on gas lift, or is an injector. Each case is discussed Annular flow leak determination method. Gas flashing across the
individually. The underlying strategy in almost all cases is to obtain leak provides Joule-Thompson effect cooling easily detectable on
baseline results with the well in near static thermal equilibrium, then the temperature log. This process has the advantage of being consis-
alter conditions to induce temperature and time transients. tent with the conditions under which the leak is known to occur: that
is, hydrocarbons flowing from the tubing to annulus side. Based on
Copyright 1995 Society of Petroleum Engineers experience to date, approximately 10 SCFM is considered the mini-
mum rate detectable by electric line logging methods. Leaks near
Original SPE manuscript received for review April7, 1991. Revised manuscript received Nov,
13. 1992. Paper accepted for publication Dec. 6, 1994. Paper (SPE 21727) first presented
surface. where higher differential pressure can be established, are
at the 1991 SPE Production Operations Symposium held in Oklahoma City, April 7-9. more pronounced.

124 SPE Production & Facilities, May 1995

 ALL DEPTHS ARE MEASURED DEPTHS UNLESS
INDICATED OTHERWISE II)
II)

J-24 UJ
Z
UJ
>
SUBSURFACE SAFETY i=
u
VALVE UJ
LL
2087 FEET LL
UJ
13-3/8", 72#/FT ...J
2701 a
4-_ _ 3 TO 15 GAS LIFT
a
I-
MANDRELS

4-1/2" SLIDING
SLEEVE
INCREASING LEAK RATE ..
10420 FT Fig. 2-Philosophical view of tool selection.
PBR
9-5/8" PACKER 10516 An example of the temperature log while flowing the annulus is
shown in Fig. 3.
4-1/2" "X" NIPPLE
TOP OF 7"
LINER 10557 Annular injection leak determination method. For wells that leak
4-1/2" "XN" NIPPLE at high liquid rates (greater than 0.25 BPM), an alternative method
is to pump a liquid down the annulus and through the leak. A plug
9-5/8" 47#/FT
10843
is set first in the tubing tail. If pumping can be sustained down the
annulus with barrel for barrel returns up the tubing, then the packer
is ruled out as the communication problem. Compared to the base-
line pass, the dynamic passes will exhibit the following behavior:
1) Identical temperature below the leak (static fluid).
PERFORATIONS 2) A temperature spike at the leak due to friction heating (unless
no restriction exists).
11325-11355
3) Either cooler or warmer temperature above the leak, depend-
(8890-8913 TVD) ing on the amount of friction heating, the pump rate, and tempera-
ture of fluid pumped (relative to original temperature of fluids
downhole).
A temperature log from troubleshooting a large leak by pumping
7", 26#/FT down the annulus is shown in Fig. 4_
11700 For wells with packer failures, there will be no returns up the tub-
ing. Progressively cooler temperatures will be measured from sur-
Fig. 1-Typical Prudhoe Bay WOA well completion. face to the plug depth.
The troubleshooting can be done without initially setting a plug
Prior to electric-line logging, tubing and annulus are both shut in in the tailpipe, but can be more difficult. If sufficient annular injec-
for at least eight hours. This allows the wellbore to approach static tion rate is obtained such that a high pressure drop is developed
thermal equilibrium for baseline conditions. Electric line is rigged across the leak, then friction heating is detected. Alternatively, if the
up and a continuous temperature log made from surface to a point leak rate is sufficient such that the velocity in the pipe exceeds
at least 200 feet below the tubing tail. Baseline data may also be ac- approximately 6 feet/minute, a spinner tool can aid in detection.
quired with the noise tool, which is included on the tool string if the The annular flow method is preferred to the annular injection
leakage rate is low and ambiguity anticipated with the temperature method, since it duplicates conditions under which the well is
logging alone.* The tool is then positioned at a point at least 200 feet known to leak. No plug is necessary, and plugs are best avoided due
below the suspected leak while monitoring temperature. to evaluation complications if they leak and due to potential removal
The annulus is bled off in 400 psi stages, interspersed with tem- problems if they become stuck. ARCO Alaska, Inc. (the operator of
perature passes. After the initial bleed off, one expects to see in- the Eastern Operating Area at Prudhoe Bay), however, has reported
successes using the annular injection method at rates down to 0.25
creasing temperature at the stationary positioning point below the
gallons per minute'?
leak. This is due to warmer (deeper) wellbore liquids gradually
flowing upwards past the tool and toward the leak. Thus, when the
"dynamic" (after bleed off) temperature passes are made they show Leak Detection in Gas Lifted Wells
progressively warmer temperatures below the leak. This is usually Gas lift wells are the easiest cases to troubleshoot. The investigation
on the order of a 0.25 - 2°F. is normally completed with the well flowing at steady-state condi-
The dynamic passes diverge from being warmer than the static tions on gas lift.
pass below the leak, to being colder near the leak where Joule- As a tubing leak develops, lift gas will pass through the leak. A
Thompson and/or latent heat of vaporization effects are occurring. decrease in casing pressure is usually experienced. This is due to 1)
For deep leaks where low tubing - annulus ilp was induced, this an increase in the total annulus-to-tubing flow area and 2) the shal-
cooling effect may be only 0.5 to 3 oF. For shallow leaks with greater lower "lifting point" if the leak develops above the normal lifting
ilp the effect will be far greater. point. For the latter locations, a decrease in gross fluid production
Above the leak, the dynamic temperature passes remain progres- associated with inefficient gas lift operation is also common. Table
sively cooler than the static pass as the cooled liquids travel up the 1 provides an example.
annulus. At some distance above the leak, typically 200-300 feet, An acoustic fluid level measurement in the annulus with the well
the dynamic passes usually approach the static pass temperatures as on lift is an important first step in troubleshooting. Various scenarios
thermal exchange with the surrounding formation dominates. with the corresponding acoustic sounding results are as follows:
The intermediate temperature passes while bleeding offthe annu- 1) Large hole, shallow leak: The liquid level can be just below the
lus in stages help pinpoint and confirm the leak. leak, since little or no differential pressure can be developed across
the leak to unload annular fluids (which can accumulate after a shut-
• A discussion of noise logging techniques as it relates to leak determination is given in the
in period). The actual distance below the leak of the annulus fluid
appendix. level will be equal to the pressure drop across the leak divided by the

SPE Production & Facilities, May 1995 125

 TEMPERATURF. DEG F TEMPERATURF. DEG F

,,
210 215 220 225 140 145 150 15S
9200 8300

BASELINE
PASS ,, t
,
9300 - 8400 - GEOTHERMAL

t
GRADIENT

9400 - 8500 -

9500 -
,,
,, t 8600 - LEAKING
t
GAS LIFT
MA:-.JDREL
DEJYfH \ DEJYfH
IN Ff

9600 -
LEAKIt\G
GAS UFT
t 1:-': Ff

MANDREL 8700 -

GEOTHERMAL
DYNAMIC GRADIENT
9700 - PASS

t 8800 -

9800 -
\ 8900 -

Fig. 3-ldentification of tubing leak using the annular flow meth- Fig. 4-ldentification of tubing leak using the annular injection
od. method.

difference in the casing gas gradient and the flowing tubing gradient uncertainty remains. Examples of tubing and packer leaks are
below the leak. shown as Figs. 5 and 6, respectively.
If the leak is higher than the shut-in tubing liquid level, then the A screening procedure similar to the above has been adopted as
annulus may stay dry, even to the bottom operating gas lift valve. part of production logging work. The well bore temperature is con-
2) Small hole, shallow leak: The annular fluid level can still be at tinually recorded while running in the hole. This information is also
the normal lifting point. (There must be significant pressure drop useful for gas lift valve redesign and troubleshooting.
across the leak for this situation to occur).
3) Leak below the normal lifting point, small operating gas lift
Injection Well Leak Detection
valve port size: The annulus will have unloaded below the normal
lifting point all the way to the leak. Injection well leaks usually present a particular challenge. Typical-
4) Leak below the normal lifting point, large operating gas lift ly, the leak rate is low. Because of the higher bottomhole pressure
valve port size: Because virtually no pressure drop is taken across (compared to producing wells) even a small leak can. over time,
the gas lift valve, the fluid will not unload significantly below this cause an annulus pressure approaching the wellhead injection pres-
point. Ifthis is suspected. a smaller port size or dummy gas lift valve sure. The leaks can sometimes be temperature sensitive, leaking
should be installed. only while the well is on injection with warm fluids.
In most cases a suspected leak is easily verified by temperature
logging. The tool string used consists of a temperature tool and cas-
TABLE 1-EFFECT ON CASING PRESSURE AND
ing collar log. A lift gas rate of over 3 MMscflD, or enough to ensure
PRODUCTION RATE OF A HOLE DEVELOPING
a significant pressure drop across the leak, is preferred. IN A TUBING STRING

Leak determination method. A log of the entire tubing string to a Well M-10 Casing
point 200 feet below the tailpipe is made. Where the leak is encoun- Test Gross Fluid Watercut Lift Gas Rate Pressure
tered, a general shift in the temperature gradient is noted. The gas Date Rate (BLPD) (%) (MMscf/D) (psi)
entry results in a flowing tubing temperature decrease above the 1/3/88 7200 62 3.7 1785
leak anywhere from 0.25 to 6 degrees, depending on lift gas rate,
1/26/88 6500 63 4.1 1570
fluid rates and composition, and hole size. Logging out through the
2/4/88 6200 63 4.1 1320
tailpipe is done to investigate any leaks at the packer, which then
show up as a cooling at the tubing tail. However, logging much be- 2/18/88 5800 62 4.1 1235
low the annular liquid level is unnecessary since no lift gas could be 3/22/88 5800 63 4.1 1240
encountered. Any suspicious anomalies are repeated. The lift gas 4/6/88 5800 63 4.1 1210
can also be shut-in and a "baseline" pass made without gas lift if any 5/16/88 5800 63 4.1 1070

126 SPE Production & Facilities. May 1995

 TEMPIDTURf, DEG F
TEMPFRATURE. DEG F 9850 _ 222 114 12<> 228

10100- GAS LIFT

MANDREL

(f)
0
Z
8z
0 9900 - o
r
10150-
0
c;")
§
t ~
nn
0
C
9950 -
SLIDING
SLEEVE ~
(")
o
c
(f)
..... ~
R R
r r
~
LEAKING

10250- GAS LIFT

~ lOCOO- PACKER
o
MANDREL 0 r
r DFPTH m
m
DEPTH <
!NIT
<
m
I~ IT m r
r 10050-
(f)
tn I
DOWN DIP I
0
o
AT HOLE
..... .....

10100-
10350-

10150-

10400-

Fig. 5-E-line and acoustic fluid level response for a gas lift well Fig. 6-Temperature log and acoustic fluid level sounding on a
with tubing leak. gas lift well with a packer failure.

Injection wells at Prudhoe Bay are either water, water-alternat- multiple sets made while running the tool into the hole. injectivity
ing-miscible gas (WAG), or gas. Many injection wells are converted was zero and no annular returns were apparent. After setting the tool
producers. Wells experiencing little or no communication problems below the PBR, slow but definitive leak injectivity was apparent.
on water injection can have a much greater communication problem
on gas injection. Thus, troubleshooting of these wells is done while Recommendations
on the gas injection cycle if possible.
I) Do not rely heavily on hydrostatic head calculations to deter-
Monitoring, problem detection, and leak quantification processes
mine the leak location. If anything, annulus pressure tends to be
for injection wells is analogous to those described for naturally
higher than what would be calculated for a given depth of tubing/an-
flowing production wells.
nulus communication.
2) Prior to investigating with electric-line logging, attempt to du-
Gas Injection Well Leak Detection
plicate the conditions under which the logging will be done. For ex-
For wells with significant leaks an identical strategy to that used for ample, shut the well in first for 6-8 hours. Then bleed off some annu-
naturally flowing producing wells is used: a baseline temperature lus pressure and note the rate of annular pressure/liquid level
pass under near static conditions is made, followed by bleeding the build-up. Some leaks are thermally related and cease after the well
annulus in stages interspersed with dynamic temperature passes. In is shut in (necessary for a baseline pass).
one (extreme) example, a well with a leak at 1,021 feet exhibited 56 3) There are numerous individual ways to pinpoint the leak loca-
degrees of cooling compared to the baseline pass. Most other wells tion, many of which involve combinations of the aforementioned
have not been as easy to identify because the leaks were deeper and techniques. It is important to determine ahead of time what type of
slower. For troubleshooting these wells a noise tool is usually in- log response is anticipated. This can impact the tools to be selected
cluded in the tool string. and the sequence of actions planned.
If the above method is unsuccessful, a plug is set in the tubing tail. 4) In most cases it is best to use the reservoir as the pressure
The annulus is allowed to reach an equilibrium pressure. The tubing source (annular flow method) rather than pumping liquid down the
is pressurized with gas, and the annulus pressure is bled off. If the annulus (annular injection method).
annulus pressure returns to its original value and the tubing pressure 5) When possible, get baseline measurements prior to inducing
does not change, a packer leak is indicated. If tubing pressure drops tubing/annulus communication. This will provide a greater degree
as annulus pressure increases, then the leak is somewhere in the tub- of confidence in the results.
ing string. If the leak is in the tubing string, then a slug of liquid is
pumped into the tubing and allowed to fall. Once in place, its top can Nomenclature
be verified with an acoustic liquid level device. The tubing is pres-
surized with gas and then shut in. If technique is successful, the liq- Pi = initial annulus pressure, psi
uid level will slowly move to the location of the leak and stop. The Pf= final annulus pressure, psi
exact location of the liquid top can be verified by using a fluid identi- Zi = Z factor, initial conditions
fication device such as a density or capacitance type electric line Zf= Z factor, final conditions
tool. T= annulus temperature, OF
V= volume of annulus from wellhead to liquid level, bbls
Water Injection Well Leak Detection L'l.t = elapsed time in minutes
Qleak = leak rate in standard cubic feet per minute (SCFM).
A plug can be set in the tail pipe and pumping done down the tubing
or annulus, similar to the method described for naturally flowing
producing wells. Acknowledgments
In one 7" completion with no nipple profiles that leaked at a very The contribution of the members of the BP Exploration (Alaska)
slow rate, a modified Baker-Lynes inflatable packer set with coiled North Slope Production Engineering department are gratefully ac-
tubing was used to confirm a leak at the PBR. (The poppet valve was knowledged. Julie Heusser and David Smith of ARCO Alaska, Inc.
removed allowing multiple sets with the same packer.) During the provided me with further insights and examples.

SPE Production & Facilities, May 1995 127

 Mlll.I\I ()L· I~
(WG S(JIJ.t:)
R. M. McKinley (Exxon Production and Research Company) has
published much of the research on noise logging, much of which has
been oriented toward its uses as a production logging tool and iden-
HICH 't~ NOISJ:: ~IUQUI::J"l:Y IY\"'DWIUlH ~
(SCHI.UMII~R(;t:I(1
tifying channels behind casing. He reports that single phase fluids
produce higher noise levels in the 1000-2000 Hz range (4). Gas ex-
panding into a water-filled channel produces increased noise in the
200-600 Hz range.
Typical noise logging equipment filters the signal into various
frequency windows (Fig. 7 from Schlumberger (8». Field results to
date have found increases in noise levels in all windows in the vicin-
ity of the leak. An example from a packer leak is provided (Fig. 7).
Noise logging is a slow process. Discrete stops must be made,
each requiring nearly one minute. Noise attenuation in liquid is low,
so stops can be widely spaced (10 - 500 feet). Attenuation in gas
filled tubing, however, is high and stops should be made only two
feet apart.
The noise tool is typically run on the same string as the tempera-
ture tool. When in the noise data acquisition mode, no temperature
Fig. 7-Noise log for a well with a packer failure. or casing collar log data is available. All possible extraneous surface
noise should be eliminated when noise logging.
The techniques and/or conclusions are those of the authoring
company and may not be shared by the other Prudhoe Bay Unit 51 Metric Conversion Factors
Working Interest Owners.
bbl x 1589 873 E-OI = m3
ftx3.048 * E-OI=m
References
ft 3 x 2.831 685 E -02= m3
I. Smith, R. C., and Steffensen. R. J., "Improved Interpretation Guidelines psi x6.894757 E+OO=kPa
for Temperature Profiles in Water Injection Wells," SPE Paper 4649. Soci- OF (OF-32)/\.8 = °C
ety of Petroleum Engineers. Richardson. Texas, 1973.
2. Curtis, M. R.. and Witterholt, E. J., "Useofthe Temperature Log for Deter- ·Conversion factor is exact. SPEPF
mining Flow Rates in Producing Wells," SPE Paper 4637, Society of Pe-
troleum Engineers, Richardson, Texas,1973.
3. McKinley, R. M. and Bower, F. M. "Noise Logging: Theory, Art of Inter- C. M. Michel is a Senior Production Engineer for BP Exploration
pretation, and Operational Procedures," July 1976. (Alaska). He is currently involved in hydraulic fracturing. He re-
4. McKinley, R. M., Bower, FM, Rumble, R.C.: "The Structure and Inter- ceived a BS degree in chemical engineeri ng 1978 and MBA in
pretation of Noise From Flow Behind Cemented Casing," 1. Pet. Tech. P. 1982. both from Oregon State U., and worked as a process engi-
329-338 March 1973. neer in the pulp and paper industry between degrees. He
5. Norris, JD. Tubing Leak De tector for Wells, and Method of Operating joined Sohio Petroleum (later BP Exploration) in 1982 and has as-
Same. US Patent No. 3,342.06 1. sumed various production engineering assignments. Michel is a
6. Hubbard. Glen O. Locating Holes in Tubing. US Patent No. 3,696,660. registered petroleum engineer in Alaska.
7. Huesser. Juli e and Smith, David, ARCO Alaska, Inc. Personal conversa-
tion.
8. Schlumberger software manual.

Appendix-Noise Logging
Noise logging can provide additional information on the location of
the leak.

128 SPE Production & Facilities, May 1995