Sei sulla pagina 1di 7

SPE 113076

A New Method for Vertical Leak Detection in Low to Moderate Permeability


Flooded Reservoirs
D.W. Walser, Pinnacle Technologies Inc., SPE, D.K. Astakhov, Pinnacle Technologies Inc., SPE, and G.R.
Stanley, Pinnacle Technologies Inc., SPE
Copyright 2008, Society of Petroleum Engineers

This paper was prepared for presentation at the 2008 SPE/DOE Improved Oil Recovery Symposium held in Tulsa, Oklahoma, U.S.A., 1923April2008.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.


Abstract
Rising commodity prices have resulted in an increase in secondary recovery projects that are associated with lower
permeability reservoirs. These processes often inject fluids above the parting pressure for the duration of the life of the flood,
and pressurization takes place as fluids leak off from the exposed surface area of the induced fracture into the surrounding
moderate or low-permeability reservoir matrix. The primary difference between these situations and conventional flooding is
that matrix fluid migration is initiating off an induced fracturing plane, as opposed to point-source initiation.

During the early history of a flood, mass-balance issues are often ignored, as average reservoir pressures are typically low,
and injection rates can dramatically exceed production. As pressures rise, some localized portions of the reservoir not only
can exceed virgin static pressures, but volumes near existing induced fractures can approach or equal the pressure in those
fractures. When this occurs, vertical fracturing can be initiated, either above or below the reservoir, depending upon the
location of the closest lower stress interval.

A new method to detect the location of vertical leakage has been developed. This method utilizes existing Surface
Tiltmeter (STM) technologies and a new surface deformation calculation regime to zero in on volumetric changes that occur
above or below the limits of the given reservoir. This method uses measured surface deformation from observed tilt to
extrapolate the volumetric change at different specified depths. With a number of constraints, and using a linear geophysical
model based on poroelastic equations, an inversion routine is used to find the reservoir compaction or expansion at different
depths. When volumetric change best fits the measured deformation data (or tilt) at the surface, the depth and aerial location
is then correlated to the area near a specific well or wells.

Introduction
Surface Tiltmeter Mapping has been utilized for many years to determine induced hydraulic fracture (or fracture network)
azimuth, induced fracture dip, and the relative volumetric component of the fractures that are either generally vertical or
horizontal
1-8
. More recently, the process has been applied to a variety of long-term reservoir monitoring projects in which
water or steam injection is tracked in an effort to determine the relative locations of reservoir dilation or compaction
9-12
.

High commodity prices have led to a renewed focus on secondary recovery projects in reservoirs of lower and lower average
permeability. It is no longer uncommon to inject water above parting pressure in all injectors for the entire life of the
flooding operation, and for matrix flow to initiate along and normal to the induced fracture face, rather than at a single point
on a wellbore. Studies have revealed that these induced fractures can be inordinately long
13
, sometimes even exceeding the
spacing radius of the injectors. The implication is, for these cases, exposed surface areas of the induced fracture(s) can be
enormous. For moderately low permeability and favorable induced fracture azimuth direction, this can be beneficial, in that
the huge surface areas can assist in repressurizing larger reservoir volumes in a shorter period of time.

Fractures induced by injection should reach an equilibrium growth rate once sufficient area has been created such that the
leakoff rate equals the injection rate. Though that can change somewhat if sludge, solids, and other fluid phases begin to
limit matrix flow out the fracture face (due to the creation of a filter cake or near-fracture-face plugging), it is hoped that
equilibrium geometries are relatively contained within the injection spacing radius, and hopefully within the bounding layers
2 SPE 113076

above and below the reservoir. If matrix permeability is so low that induced injection fracturing does not reach equilibrium
in terms of geometry, severe leakage above or below the reservoir of interest then becomes a real possibility. This situation
may be exacerbated if stress or fracture gradients above or below the reservoir of interest are substantially lower than in the
reservoir itself.

History
Recently, a project was undertaken in a low to moderate-permeability waterflood, in which it was desired to diagnose why
producers were not responding to substantial line-drive injection. Primarily, it was desired to measure the injection fluid
induced fracture azimuth in order to confirm suspicions that reservoir pressurization was (by coincidence) occurring only
along the azimuth of the line drive. Unfortunately, issues related to matters of confidentiality prevented the use of the
specifics of the project as a case history. Material balance calculations relating to injection volumes and produced fluids
suggested that either matrix flow to the producers was extremely slow, or a huge percentage of fluid was exiting the reservoir
by a means other than producing wells. A surface tiltmeter array was constructed across a large swath of injectors and
producers, and measurements were taken during both flowback and injection operations. Evaluation of tiltmeter response
revealed two points:
A moderately marginal match was attained between theoretical tilt and the observed tilt, assuming the actual
reservoir thickness and depth. Confidence in the analysis was relatively low at this point in time.
A volumetric response somewhat consistent with fluid movement along the same azimuth as the line of injectors
was noted in some cases, and oblique to the injector line azimuth in other cases.

Due to difficulties associated with uniqueness of potential solutions (described later in this paper), tiltmeter evaluations are
normally constrained by as many known parameters as possible; two of which are reservoir depth and thickness. In an effort
to evaluate multiple possible scenarios for this particular situation, different depths and reservoir thicknesses were assumed,
and 100% of the injected fluid was assumed to be creating fracture(s) at these different depths. One particular depth
(different from the actual reservoir depth) gave a better match between theoretical tilt and actual measured tilt, which led to
increased scrutiny of the calculation process as a potential method for better pinpointing potential reservoir leakage. In this
particular case, though the exact percentage of injected volume that was leaking was unknown (as well as what percentage
volume was simply repressurizing low matrix pressures near injectors), it was suspected by the operator that the leakage
percentage was very high. Therefore, an assumption of 100% leakage was used, and served as a good starting point. Initial
analysis could not only determine the elevation of leakage, but also x-y positioning from a map view perspective. Corrective
operational action could then be taken, and as time progressed, the relative ratios between leaked volume and volume
remaining in matrix flow in the reservoir of interest could be determined by mass balance procedures. Once this ratio was
determined, volumetric surface tilt calculations could be re-examined in order to boost confidence in the elevation of the
primary leak.

Theoretical volumetric strain was calculated utilizing a poroelastic model structured after the assumed depth and reservoir
thickness. Actual surface tilt measurements were inverted for the reservoir depth, utilizing a model incorporating a
smoothness-constrained penalty function, and a best-fit solution was found by varying the volumetric strain in the poroelastic
model until theoretical tilt best matched the actual tilt that was observed at each site.

The reservoir is divided into a number of rectangular blocks and constant volumetric change is assumed in each block.
Equations (1) and (2) below are numerically integrated over each block to obtain the theoretical tilt. Total tilt on the surface
is calculated as a sum of tilts generated by all reservoir blocks.

Poroelastic Forward Model
The forward poroelastic model used in this work was described and developed over a number of years
14-18
. The theoretical
displacement (u) in any reservoir grid block generated by the volumetric changes is

=
V
i
u
i
dV x g m
B K
x u


) , ( ) (
0
(1)
where is the shear modulus, K
u
is the undrained bulk modulus, B is the Skempton pore pressure coefficient,
0
is the
reference fluid density, m is the mass change per unit of bulk volume, and ) , ( x g
i
is the Greens function for a point
source inflation in a half-space
15
. Integration goes over the entire reservoir volume V by steps through the reservoir blocks

V .

Two components of displacement gradient measured by tiltmeters can be calculated as:
SPE 113076 3
( )


dV
x
x g
m
B K
x
x u
x T
V
i u i
i

=
3 0 3
) , ( ) (
(2)
where i can be 1 or 2, and x is a vector (x
1
, x
2
, x
3
).

Inverse Problem
Since the volumetric deformation is assumed to be constant in each reservoir block, the smaller the grid [or the higher the
block count], the better heterogeneity in the reservoir may be addressed. The number of tiltmeter sites is limited, however.
The vector

d = G
S
(3)

is the measured tilt on the surface. It has 2n elements, where n is the number of sites, and G
S
is the discrete Greens function
relating deformation on the surface with volume changes in the reservoir. Since the number of measurements (the site count)
is quite limited, the non-uniqueness of the solution can be a major hurdle. To circumvent this problem, some constraints are
introduced. Normally, the primary goal is to define fluid movement in the reservoir (due to injection or production), and it is
natural to limit volumetric changes to the reservoir extent, and assume no volumetric change outside the reservoir. In that
case, the most reasonable solution can be found by tradeoff between the roughness of the volumetric changes and the match
between the observed and modeled tilt responses
14
. If the goal, however, is to detect a vertical leak from the flooded
reservoir, the location and depth of the volumetric changes are generally unknown. A precise solution would require detailed
geological information about the formation properties above and below the reservoir, and in particular, the depth and
thickness of all high permeability or low stress layers where the leaking fluid could escape. Even with this information, the
problem still has multiple solutions if the percentage of injected fluid staying in-zone is not known.

It may not be possible to exactly locate the source (or area) of leakage, but it is possible to develop a strategy that will allow
narrowing possible sources of the leak to some specific portion(s) of the reservoir. Remedial operations could then be limited
to a much narrower range of options. If the flood has progressed to the point where injection/production material balances
indicate that an equilibrium has been reached, then a ratio of in-zone to out-of-zone migration may be established, and the
overall inversion may be constrained further.

In the case of measurement of surface vertical displacement, upward movement practically always corresponds to positive
volumetric changes in the reservoir below (dilation/injection), and downward movement reflects negative volumetric changes
(compaction/production). Tiltmeters, however, measure the displacement gradient (slope) rather than the actual displacement
itself, and it is generally possible to match the observations with positive or negative (or some combination of both)
volumetric changes. However, in a leak detection search, one more constraint needs to be applied. All deformations have to
be matched by only negative (in the case of shut-in monitoring) or only positive (in the case of injection start monitoring)
volumetric changes. This practice can narrow the range of possible solutions, but may sometimes lead to false leaks that
appear to match the deformation response in another location.

Example Showing Reservoir Leak
To better understand how suspected leaks (or vertical fluid migration) are found using this tilt monitoring method, an
example is presented. In this case, a leak is occurring at ~200 ft in depth, and an array of 22 tiltmeters is used to measure the
surface deformation. For each site, there are two vectors shown. The black vectors represent the measured tilt, and the red
vectors represent the theoretical tilt. The theoretical vector is based on a series of input parameters, which then become the
solution after many inversions are performed to find the global minimum. A perfect fit would be where two vectors at each
site are exactly the same. In Figure 1, the red area corresponds to the region of expansion at a depth of ~200 ft. Notice that
there are two sites exhibiting a significantly high tilt response, suggesting a shallow leak where the depth and x-y location
can be determined.











4 SPE 113076


Figure 1 Example showing measured tilt (black vectors) and theoretical tilt (red vectors) for a leak mapped at ~200 ft in depth. This
shows a few sites exhibiting very high tilt compared to the rest of the sites; a strong sign that a leak is occurring much shallower
than the injection zone.


Vertical Leakage Detection
There are two methods presented: the first uses the Inverse Problem with the extracted tilt data at shut-in, and thus detects
reservoir compaction and vertical leakage; and the second uses the extracted tilt data, starting with injection, to detect
reservoir dilation and vertical leakage.

Shut-in Monitoring
In this method, the extracted tilt is used to measure the change in surface deformation starting at the time of injector shut-
in. It can be from a single well, or (as in the example shown) a row of injector wells. The method involves monitoring for a
period of 12 to 24 hours after shut-in, and measurement of the surface deformation change resulting from reservoir
compaction and fluid migration as the fluid leaks off. If there is significant vertical leakage, these areas appear as
compaction areas, which are shown in blue in Figure 2. To differentiate vertical leakage from compaction at reservoir depth,
the inverse problem of matching volumetric changes to the measured tilt data is solved multiple times by constraining (each
time) the depth of the volumetric expansion to individual known lower stress gradient layers above and below the producing
reservoir. Closely spaced layers may be combined in stacks. Since the depth of the volumetric change is unknown, and
many negative changes occur at reservoir depth, hopefully only few occur at the depth above or below the reservoir. The
situation is analogous to the well known gravimetric problem: the same gravity field on the surface can be produced by
different mass distribution at different depths. If the real volumetric compaction occurs at one depth and the Inverse Problem
has been solved constraining it to another depth, the location of such volumetric change may not be correct. To resolve this,
it may take several injection cycles for inverse problem solutions to start grouping the compaction areas in particular
locations that correlate to lower stress layers, or vertical leakage. To verify the results from an injection shut-in, or to find
other problem portions of the reservoir, the second method is used: monitoring at the start of injection.

SPE 113076 5

Figure 2 Inverted volumetric strain distribution showing localized compaction area (vertical leakage) in a low stress zone above
the reservoir.


Injection Monitoring
In the second method, all injectors are shut down for several days to allow time for existing fractures to completely close
and create a baseline. Then a row of injectors are started simultaneously. The injection cycle can be 4, 8 or 12 hours;
different durations may be necessary for different reservoirs, depending upon Darcy parameters and differential between
stresses in the zone of interest and stresses above or below. After each injection cycle, a particular group of injectors are shut
down, and after a few more days, a second group is turned on, etc. The inverse problem of matching volumetric changes to
the measured tilt data is solved multiple times, constraining (each time) the depth of the volumetric expansion to individual
lower stress layers above and below the producing reservoir. Similar depths are input using the Inverse Problem to locate
areas of expansion. These results can then be correlated to the same locations that compaction occurred when the same well
group was shut in. Figure 3 shows a typical volumetric strain distribution for the injection monitoring method. Here a
dominant red area (expansion) corresponds to a depth and location above the producing reservoir.


6 SPE 113076



Figure 3 Inverted volumetric strain distribution showing localized expansion areas (vertical leakage) in a low stress zone above
the reservoir.

Solution Confidence and Reliability
The same surface deformation may be induced by volumetric changes at different depths. This means that without identifying
the elevation of target leaking zone(s), one can not reliably focus on the x-y location of the leak. Layers of high
permeability or low stresses above or below the producing reservoir should be identified in advance of analysis as potential
targets for the investigation of possible injection-induced leaks, in order to constrain the total number of possibilities to the
greatest extent possible. Relatively detailed geological and stress (or fracture gradient) information about the lithological
column above and below is needed in order to accomplish this. Since the 100% leaking volume assumption quoted above
is rarely accurate, it is always possible to have a few suspected leak areas that in reality correspond to volumetric changes in
the reservoir itself. It is important to note that if the actual leaks elevation is correctly identified, the actual leaks x-y
location will not go undetected. The solution may identify several x-y areas as possible leak locations, but only one or two of
them may be the real leaks. In other words, it may not be true that all identified areas are leaks, but all existing leaks are
within the list of identified areas. Therefore, whatever the case, the number of areas with possible leaks is still substantially
reduced, and can be further studied for problems. For example, in the initial discovery project referred to in this paper, an
area of expansion was identified in an x-y extent right next to an injector. Upon learning that that particular injector had
massive plugging problems and was only placing 5 10 bpd of fluid into the reservoir, confidence in the placement of the
deformation depth in an identified out-of-zone low-stress interval was dramatically increased.

Though the focus of this paper has been the identification of leaks associated with induced fractures away from individual
wellbores, there is no reason why the same process(es) could not be applied to the identification of [out of zone] leaks in
individual wellbores, as long as those leaks resulted in creation of induced fractures in a low-permeability lithology near the
point of fluid exit. An example of this would be a well with casing damage, or shear failure at a very shallow depth, thus
resulting in a leak above the reservoir in a zone of low stress.

SPE 113076 7
Conclusions
A method to narrow in on the possible or probable locations of induced injection fractures above or below low and moderate
permeability reservoirs under secondary recovery has been developed. Though there are limitations to the process, primarily
associated with not having enough constraints, it is often possible to utilize information other than the surface deformation
data itself to sufficiently constrain the calculations, and identify the depth and x-y coordinates of possible leaks. In addition,
careful design of multiple injection and shut-in experiments can (under certain circumstances) introduce constraints sufficient
to remove a substantial degree of uncertainty associated with the location of suspected leaks. With the growing development
of tilt based technology and its applications, reservoir management can be improved on secondary recovery projects.

Acknowledgements
The authors are grateful to the management of Pinnacle Technologies Inc. for granting permission to publish this work and to
Jing Du for substantial assistance in the development of the process.

References

1. Holzhausen, G.R., and Egan, H.N., Detection and Control of Hydraulic Fractures in Water Injection Wells, paper SPE 16362
presented at the 1987 SPE California Regional Meeting, Ventura, CA, April 8-10.
2. Davis, E., Astakhov, D., Wright, C., Precise Deformation Monitoring by High Resolution Tiltmeters, Butsuri-Tsana, Society of
Exploration Geophysicists of Japan, (2001), Vol. 54, No. 6, pp. 425-432.
3. Wright, C.A., Davis, E.J., Minner, W.A., Ward, J.F., Weijers, L., Pinnacle Technologies, Inc.; Schell, E.J., Union Pacific Resources;
Hunter, S.P., Lawrence Livermore National Laboratories, Surface Tiltmeter Fracture Mapping Reaches New Depths - 10,000 Feet
and Beyond?, paper SPE 39919 presented at the 1999 SPE Rocky Mountain Regional/Low-Permeability Reservoirs Symposium, 5-8
April, Denver, CO.
4. Weijers, L., Wright, C.A., Demetrius, S.L., Wang, G., Davis, E.J., Pinnacle Technologies Inc.; Emanuele, M.A., Broussard, J.B.,
Chevron U.S.A. Production Co.; Golich, G.M. Aera Energy LLC, Fracture Growth and Reorientation in Steam Injection Wells,
paper 54079 presented at the 1999 International Thermal Operations/Heavy Oil Symposium, 17-19 March, Bakersfield, CA.
5. Mayerhofer, M., Demetrius, S.L., and Griffin, L.G., SPE, Pinnacle Technologies; Bezant, R.B., TotalFina; Nevans, J., SPE, First
Permian LLC; and Doublet, L., SPE, Gaither Petroleum, Tiltmeter Hydraulic Fracture Mapping in the North Robertson Field, West
Texas, paper SPE 59715 presented at the 2000 SPE Permian Basin Oil and Gas Recovery Conference, 21-23 March, Midland, TX.
6. Cipolla, C.L., Wright, C.A., Pinnacle Technologies, Diagnostic Techniques to Understand Hydraulic Fracturing: What? Why? and
How?, paper SPE 59735 presented at the 2000 SPE/CERI Gas Technology Symposium, 3-5 April, Calgary, Alberta, Canada.
7. Griffin, L.G., Wright, C.A., Davis, E.J., Pinnacle Technologies; Wolhart, S.L., Gas Research Institute; Moschovidis, Z.A., PCM
Technical Inc., Surface and Downhole Tiltmeter Mapping: An Effective Tool for Monitoring Downhole Drill Cuttings Disposal,
paper SPE 63032 presented at the 2000 SPE Annual Technical Conference and Exhibition, 1-4 October, Dallas, TX.
8. Minner, W.A. Wright, C.A., Stanley, G.R., de Pater, C.J., Pinnacle Technologies, Inc; Gorham, T.L., Eckerfield, L.D., Hejl, K.A.,
ChevronTexaco Corp., Waterflood and Production-Induced Stress Changes Dramatically Affect Hydraulic Fracture Behavior in Lost
Hills Infill Wells, paper SPE 77536 presented at the 2002 SPE Annual Technical Conference and Exhibition, 29 September-2
October, San Antonio, TX.
9. Bruno, M. S. and Bilak, R. A.: Cost-Effective Monitoring of Injected Steam Migration Using Surface Deformation Analysis, paper
SPE 27888 presented at the 1994 SPE Western Regional Meeting, Long Beach, CA, March 23-25.
10. Davis, E.J., Wright, C.A., Demetrius, S.L., Pinnacle Technologies; Choi, J., Chevron USA, Craley, G., Nuevo Energy, Precise
Tiltmeter Subsidence Monitoring Enhances Reservoir Management, paper SPE 62577 presented at the 2000 SPE/AAPG Western
Regional Meeting, 19-22 June, Long Beach, CA.
11. Brink, J.L., Chevron USA Production Company; Patzek, T.W., Lawrence Berkeley National Laboratory and University of California,
Berkeley; Silin, D.B., Lawrence Berkeley National Laboratory; Fielding, E.J., Jet Propulsion Laboratory, California Institute of
Technology, Lost Hills Field Trial - Incorporating New Technology for Reservoir Management, paper SPE 77646 presented at the
2002 SPE Annual Technical Conference and Exhibition, 29 September-2 October, San Antonio, TX.
12. Kramm, R.C., Griffin, L.G., Apruzzese, J., and Stanley, G.R., Pinnacle Technologies; Flaten, D.W., Bradshaw, M.K., and Lucas, B.C.,
Devon Energy Corp., Mapping Preferential Flow in a West Texas Waterflood Using Surface Tiltmeters, paper SPE 95566 presented
at the 2005 SPE Annual Technical Conference and Exhibition, 9-12 October, Dallas, TX.
13. Griffin, L.G., Wright, C.A., Demetrius, S.L., Pinnacle Technologies; Blackburn, B.D., and Price, D.G., Marathon Oil Company,
Identification and Implications of Induced Hydraulic Fractures in Waterfloods: Case History HGEU, paper SPE 59525 presented at
the 2000 SPE Permian Basin Oil and Gas Recovery Conference, 21-23 March, Midland, TX.
14. Du, J., Wolhart, S.L., Brissenden, S.J., McGillivray, P., Davis, E.J., Roadarmel, W.H., Wright, C.A., Mapping Fluid Flow in a
Reservoir Using Tiltmeter-Based Surface Deformation Measurements, paper SPE 96897 presented at the 2005 SPE Annual
Technical Conference and Exhibition, Oct. 9-12, Dallas, TX,
15. Segall, P., Stress and Subsidence Resulting from Subsurface Fluid Withdrawal in the Epicentral Region of the 1983 Coalinga
Earthquake, J. Geophys. Res., (1985), 90, 6801-6816.
16. Vasco, D.W., Karasaki, K. and Myer, L., Monitoring of Fluid Injection and Soil Consolidation Using Surface Tilt Measurements,
Journal of Geotech. and Geoenvironmental Eng., (January 1998), 124, 29-37.
17. Okada, Y., Internal Deformation Due to Shear and Tensile Faults in a Half-Space, Bull. of the Seism. Soc. of America, (1992), Vol.
82, N 2, pp.1018 1040.
18. Du, J., and Olson, J. E., A Poroelastic Reservoir Model for Predicting Subsidence and Mapping Subsurface Pressure Fronts, J. of
Petroleum Science and Engineering, (2001), 30, 181-197.

Potrebbero piacerti anche