Sie sind auf Seite 1von 11

PAPER 2000-29

Aquifer Influx Modelling for


Gas Reservoirs
H.K. Lies
HyPex Consulting Ltd.

This paper is to be presented at the Petroleum Society’s Canadian International Petroleum Conference 2000, Calgary, Alberta,
Canada, June 4 – 8, 2000. Discussion of this paper is invited and may be presented at the meeting if filed in writing with the technical
program chairman prior to the conclusion of the meeting. This paper and any discussion filed will be considered for publication in
Petroleum Society journals. Publication rights are reserved. This is a pre-print and subject to correction.

application of the models can be extended to any gas


ABSTRACT reservoir.
Aquifer water influx is an important mechanism which
effects the performance of natural gas reservoirs. The
influx of water has serious consequences in the estimation P/Z V. CUMULATIVE PRODUCTION
of gas reserves through material balance and in the PLOTS
prediction of future pressures. The effect of water influx The computation of the initial gas-in-place for
from either a finite or infinite aquifer always exists to constant volume reservoir requires the following data:
some extent, however these effects are not always
recognized. • Initial reservoir pressure
Methods of modeling water influx include the Hurst • Cumulative gas volume
modified steady-state method, and various unsteady state • Stabilized shut-in reservoir pressures
methods such as those of van Everdingen-Hurst, and • Gas deviation factor
Carter-Tracy. More recently, the four methods introduced Although this method is currently used, it is not
by Leung for finite and infinite aquifers provide relatively applicable to water-drive gas reservoirs. With pressure
simple models to use with substantial improvements in reduction, when water enters the space occupied by the
accuracy and efficiency over the previous methods. gas, the pressures are maintained in part depending on the
Using the methods introduced by Leung, combined nature of the water drive.
with the volumetric's of the gas reservoir, aquifer
parameters such as size, total compressibility, influx
angle, permeability, porosity and thickness can be
AQUIFER SUPPORT
determined. With adequate pressure data, the models Pressure support occurs due to a variety of
converge to a unique solution for all aquifer parameters. mechanisms - including an infinite acting aquifer,
The development of the aquifer influx models was the unsteady state finite aquifer, shale water influx, water
result of the need to better understand the behavior of the expansion and rock compressibility/compaction. Through
McMurray aquifer system in north east Alberta. The 7-5- the analysis of DSTs, AOFs, Logs, pressure and
81-6W4M well is used as an example, though the production data, the mechanism and future performance
of the aquifer can be forecast.
P/Z plots of gas pools attached to a medium to slowly use of a Laplace transformation. This still required the use
responding aquifers usually follow the straight line of a of a complicated superposition procedure.
depletion drive reservoir and then gradually rise above it. In 1960, Carter and Tracy (7) introduced the
This is shown in Figure-1. convolution integral to simplify the solution to water
Extrapolation of the P/Z will overestimate the gas-in- influx to an explicit step-wise calculation.
place. The gas wells water out as the true gas-in-place is Fetkovitch (12) 1971, introduced the Pseudo Steady-
approached and while the reservoir pressure remains State model for finite aquifers to explain the late-time
relatively high (ie. 50% of original reservoir pressure). transient effects of water influx.
The P/Z plot for gas pools influenced by strong Leung (13) in 1986, introduced three different finite
aquifers can initially follow a straight line and looks and an infinite aquifer models for unsteady-state water-
similar to a depletion gas drive reservoir, as long as influx calculations. They are relatively simple to use and
constant gas production rates are maintained. The represent a substantial improvement in the accuracy and
extrapolation of the P/Z plot will severely overestimate efficiency over previous methods.
the gas-in-place in these cases. Again, these gas wells
water out as the true gas-in-place is approached and the The four models introduced by Leung are
reservoir pressure as before remains relatively high (ie. recommended. The reader is directed to the original
~75% of original reservoir pressure). paper for the mathematical derivations and application
procedure. The discussion here will be limited to the
application of these models, and in particular, the infinite
PERFORMANCE OF WELLS AFFECTED acting aquifer model.
BY AQUIFER Essentially all aquifer influx models are a solution to
Gas production from wells that are experiencing water the diffusivity equation. The driving force for water
influx will initially produce gas with no decline in the influx comes from the total compressibility of the rock
case of strong aquifers, and moderate decline in the case and water system combined. Theoretically, the model
of medium and slow acting aquifers. In all cases does not account for recharge which would be expected in
production declines rapidly when the water production a hydrogeological framework. It could be argued that an
increases. A typical production profile is shown in Figure infinite aquifer model does represent a recharge system.
2. The basic input parameters for an aquifer influx model
are pressure through time at the influx boundary,
permeability, thickness, total compressibility, and influx
ALTERNATE INTERPRETATIONS angle (in the case of a radial system). The model predicts
The apparent bending of the material balance plot may the rate at which water will encroach as the pressure
be interpreted as gas migration from edge or tight areas of transient propagates out into to the aquifer.
the reservoir, or pressure support from an underlying The pressure at boundary of the gas pool is calculated
aquifer. by material balance.
There are multiple unknowns in the material balance
equation, for water drive reservoirs, and calculations Vp = Ve + We + BwWp
generally involve several assumptions on the reservoir
description. Consequently, material balance predictions
The volume of gas produced Vp equals the change in
are often unreliable when a detailed understanding of the
volume of the initial gas Ve due to the pressure drop, plus
reservoir and supporting aquifer does not exist.
the volume of water influx We minus the volume of water
History of Water Influx Models produced BwWp.
The first reference to water influx was by Schilthius
(1) in 1936. Influx is proportional to the difference in Production = Expansion + Water Influx - Water
pressure between the original formation pressure and the Production
reservoir pressure at a given time. In the author’s opinion,
this method should not be used as it is inaccurate. Combination of the water influx model with the
material balance equation, enables the gas pools pressure
Next, Hurst (2) in 1942, determined water influx
to be predicted versus time and cumulative production.
based on the solution of the diffusivity equation. The
Hard data inputs include:
mathematics offer a certain amount of latitude in fixing
the constants associated with water influx. However, what Initial reservoir pressure Pi
was most critical in these analysis was the history of the Initial aquifer pressure Pi
reservoir pressure as it varied with time.
Temperature T
Everdingen and Hurst (3) in 1949, improved and
simplified the solution of the diffusivity equation by the Gas properties PVT
Gas zone thickness hg
Gas zone porosity φg times the pool size had to be used (again beyond the valid
Water saturation Sw range) to approach a model match.
Gas zone permeability kg
Production history Gp
MODIFIED PSEUDO STEADY STATE
MODEL
Most of the above parameter can be determine with
reasonable certainty. Model match parameters are as Leung third model is the modified pseudo steady state
follows: (MPSS) model. The MPSS model similar to the TPSS
Gas drainage radius re model extends the range of validity to an aquifer radius of
50 times the gas pool radius.
Aquifer inner radius re
When applied to a selection of wells producing gas
Water viscosity µw from the McMurray, aquifer sizes much larger than 50
Total compressibility ct times the pool size had to be used (again beyond the valid
range) to approach a model match.
Aquifer thickness ha
In the case of all three of the finite aquifer models the
Aquifer porosity φa
high rate of water influx needed to match the pressure
Aquifer permeability ka history required aquifer sizes beyond the valid range of
Influx angle θa the model.
Aquifer radius / Gas pool radius R
The model match parameters are varied until a INFINITE AQUIFER MODEL
reasonable match is achieved with the observed pressure The infinite aquifer model is suggested as the model
data. At least three pressure points are needed to be able of choice for almost all situations, due to the fact that
to determine the aquifer parameters. It will be shown later water zones/aquifers are interconnected within a
that the infinite aquifer water influx model converges to a hydrogeological framework and rarely isolated even
unique solution for all of the match parameters. across aquitards.
The infinite aquifer model was applied to a selection
PSEUDO STEADY STATE MODEL of wells producing gas from the McMurray. The model
provided excellent matches in various single well cases
The pseudo steady state model (PSS) is the first model
primarily in the early years. The addition of gas
described by Leung. This is the easiest of the models to
production within the a given wells original drainage
apply and is valid for finite aquifers with a radius less
radius has a significant effect. In addition, the effect of
than 5 (R<5) times that of the gas pool.
increased regional gas withdrawal over time is observed
The PSS model was initially chosen for the as having an influence on the regional aquifer.
McMurray aquifer based on the evidence presented by the
gas producers at the public hearings, stating that the
water zones underlying the gas pools were limited water PRESSURE DATA
pods. The use of the aquifer influx model will assist in
The PSS model was applied to a sampling of wells determining the original gas in place. Unfortunately,
from various areas. In a great majority of cases, where without accurate pressure data, modeling will be of little
adequate pressure data was available, the aquifer radius assistance.
had to be set at over five times the gas pool radius to The AOF test, well logs and gas analysis are the
approach a model match, beyond the range of accuracy source of the hard input data in addition to the production
for the PSS model. Obviously, the McMurray aquifers are history. Accurate pressure data, in particular the initial
not small limited pods. pressure point, is critical in providing accuracy and
confidence in the final model match. Without pressure
TRANSIENT PSEUDO STEADY STATE data through time there is no way of being able to
understand or manage the reservoir, bad reservoir and
MODEL economic decisions follow. Without sufficient pressure
The transient (TPSS) model is the second model data showing a straight line on the P/Z versus cumulative
described by Leung. It accounts for early time effects that production plot it cannot be proven that water influx is
are neglected in the PSS model. It is more difficult to not occurring.
apply but is valid for finite aquifers with a radius up to 10 The initial testing of a gas well usually includes
times that of the gas pool. conducting an absolute open flow potential test. Ideally
When applied to a selection of wells producing gas the well is left shut-in after the initial completion and
from the McMurray, aquifer sizes much larger than 10 clean up flow period (one gas zone only) until the
wellhead pressure stabilizes. Once the wellhead pressure
has stabilized, a static gradient survey to base of the elevation. At 1000 meters elevation the average
perforations should be conducted. Immediately after the atmospheric pressure is 89.9 kPaa down from the average
initial static gradient is run, the flow and build up sea level value of 101.3 kPaa (Source: International Civil
recorders should be run and landed above the mid point of Aviation Organization). Surface gauge readings will
perforation, but within 10 meters. The extended flow therefore increase by the pressure difference (11.4 kPa)
period can be started immediately. Use as high initial rate between the two elevations while measuring a constant
and reduce to a rate sufficient to lift any water. Flow for pressure source.
12 hours and shut-in. Monitor pressure build up until To adjust to the absolute pressure, the barometric
the wellhead pressure returns to within 99% of the pressure, at the well head, at the time of the test must be
original pressure or stabilizes. Finally, retrieve the flow known. This can be estimated based on the surface
and build up recorders and run a final static. elevation, realizing that the normal atmospheric
The above procedure will ensure that there is a 99.9% variations of +/-2 kPa is occurring. With proper care,
probability of obtaining an accurate initial reservoir pressures can be determine to an accuracy greater than
pressure. Long extended gas flow times are not needed.
recommended. The Para Chard et at 100/07-05-081-06W4M well was
It is critical that static gradient survey report the chosen as it provides an example of a well which has
surface casing and tubing pressure using a clean and well experienced limited effects from gas production within
maintained dead weight gauge. The surface dead weight its drainage radius or from any apparent change in the
gauge reading will assist in determining the accuracy of regional aquifer. The following are the hard input
the bottomhole recorders. In addition, the current parameters used for the 7-5 well:
barometric pressure should be recorded if an accurate Imperial Metric
conversion to the absolute pressure is to be made.
Initial reservoir pressure 229.6 psia 1583 kPaa
Bottomhole gauge are not as accurate as claimed. Initial aquifer pressure 229.6 psia 1583 kPaa
Mechanical gauges in the field are outside of the claimed
accuracy more than fifty percent of the time. The same Temperature 44.6 F 7.0 C
hold true for electronic gauges. It is common practice to Gas deviation factor 0.966
used a different set of gauges for the initial static gradient, Gas zone thickness 3.0m
flow and build up and the final static. Rarely, do these Gas zone porosity 33%
recorders agree within the specified accuracy limits. Water saturation 20%
A well maintained dead weight gauge will accurately Gas permeability 1686md
measure the gauge pressure (kPag) to the resolution of
the weight set. The surface dead weight gauge reading The production history of the 7-5 well was obtained
must be used to calibrate the recorders. from the public records and input into the model. Records
It is foolish the use the pressure reported in a absolute show that the 7-5 well went on production in April of
open flow test or on a static gradient without ensuring the 1989 at an average calendar rate of 1 Mmcf/d and
recorders are calibrated to the dead weight gauge reading. maintained this rate until early 1995 at which time the rate
Accurate pressure data is important in being able to dropped to approximately 400 mcf/d. This drop is
predict aquifer performance. associated with an increase of water production from 10
Bbls/Mmcf to 60 Bbls/Mmcf. Water production continued
to increase and since 1997 has fluctuated between 100 to
ABSOLUTE PRESSURE CONVERSION 300 Bbls/Mmcf (Figure 2). Also in 1999 the 7-5 well
It is common industry practice to use varying appears to have water out while still at a relatively high
barometric pressure conversions ranging from 85 to reservoir pressure.
101.35 kPa to determine the absolute pressure. This Cumulative Production to Jan 1, 2000
results in the introduction of unnecessary additional error
Gas 2.893 Mmcf 81.9 E6M3
into the pressure measurements. In the case of high
pressures, this error will be relatively small. However at Water 91.7 Mstb 14.6 E3M3
lower pressures, this error can becomes significant.
The simplest method to minimize the error introduced Well logs from the 7-5 well indicate 9 meters of clean
by the barometric conversion is to ensure that a sand , 3 are gas over 6 of water. Below the upper clean
consistent value is used. It is recommended, however that sand section are two more water bearing sands separated
the conversion value always be the standard sea level by shale before encountering a bitumen pay zone of 20
pressure of 101.325 kPaa. meters thickness (Figure 3).

The pressure at the land surface decreases with


increasing
FIELD EXAMPLE from a total of 146 wells producing from the McMurray
Formation in the Surmont area of north east Alberta. The
The author, has chosen an area of topical debate, shut-in request was to protect the bitumen resource
which is in the northeastern part of the Province of recoverable through the steam assist gravity drainage
Alberta. The example well is in the immediate vicinity (SAGD) process. The pressure depletion of the associated
of the Surmont oil sands lease. gas production is expected to reduce recovery, and in
some cases, result in the sterilization of the bitumen
INFINITE AQUIFER MODEL MATCH resource.
In a recent decision of the AEUB, it was concluded
The following pressures from the 7-5 well were used
that pressure support, due to water influx would be, in
as history match parameters.
general, weak and insignificant. As outlined, the author
Date Method psia kPaa believes that most of the wells in this area could be
interpreted as having water influx with significant
Mar-88 BU 229.6 1583
pressure support.
Feb-93 StGr 169.5 1169
The moderate water influx example shown in Figure 1
Jun-97 Dwg 146.8 610 represents the Chard 7-05-81-06W4M wells infinite
aquifer model match. Water influx rates are estimated to
The model match parameters (ie. re, µw, ct, ha, φa & be 7000-8000 BWPD (1100 - 1300 M3/D). Recognizing
θa) are varied until the best match to the observed the effect of water influx may alter the perception of the
pressures is achieved. With the use of the computer it is effect that gas production has on potential SAGD
easy to determine the match parameters which provided production.
the minimum deviation from the observed pressure data
(Figure 4). In addition, the solution is convergent to one
unique solution (Figure 5). The following match represent ECONOMIC IMPLICATIONS
a .8% total pressure deviation for the 7-5 well. Take the example of a gas property acquisition in
which a strong water influx is occurring. The material
Gas pool radius 5560 feet 1695 meters balance calculation and decline analysis would indicate
Water viscosity 1.00cp 1.00cp substantial remaining reserves. In the worst case the
Total compressibility 3.5E-06 v/v/psi financial loss could be 100% of the purchase price, if the
Aquifer thickness 20.54 feet 6.38 meters wells were to start watering out soon after the acquisition.
Aquifer porosity 20% 20% Without an accurate series of pressures over time it
Aquifer permeability 1650md 1650md can not be determined whether gas or water influx is
Aquifer influx angle 94.9° 94.9° occurring. Instead it is traditionally assumed that the gas
reservoir behaves volumetrically with no supporting
evidence. This highlights a common mistake, in material
The author has found through the use of the infinite balance analysis.
aquifer model that it has impressive accuracy, making it
an exceptional analysis tool to see far beyond the
wellbore. In the case of the 7-5 well the thickness of the CONCLUSIONS
aquifer is exactly as seen on the log. While the AOF test
1. The newer method of Leung provides superior
determined a gas permeability of 1686 md, compared to
accuracy in the water influx analysis described,
the models aquifer prediction of 1650 md. The influx
compared to older techniques.
angle requires a review of the geology to confirm. The
porosity represents the bulk porosity of the infinite aquifer 2. Volumetric behavior must be proven not assumed.
system which includes both reservoir and non reservoir Pressure history is critical in this regard.
rock. The total compressibility is completely reasonable 3. The material balance is commonly misused:
and is probably the best measure of that value. production performance must be consistent with the
Water influx is estimated to have reached a maximum material balance technique used.
of 8250 Bwpd (1300 M3/D) in 1994 before starting to 4. Pressure data must be carefully collected and screened
drop off. When the gas rate dropped to 400 mcf/d the gas in the case of low pressure reservoirs.
pool pressure stabilized at approximately 145 psia (1000
5. In the author’s opinion, existing analyses presented in
kPaa), the rate of water influx and gas voidage were in
the northeastern portion of Alberta may be in
balance.
considerable error, due to inadequate pressure data
collection, screening and interpretation.
RECENT AEUB DECISION 6. Since the price crash of 1985 the phrase “lean and
Effective May 1, 2000 the Alberta Energy Utilities mean” became popular. One of the implications of
Board requested the shut-in of associated gas production lean, was that reservoir pressures were no longer
obtained regularly on producing properties. In 12 FETKOVICH, M.J.: “A Simplified Approach to
hindsight, this may prove to have been no saving at Water Influx Calculations-Finite Aquifer Systems,”
all. Jour. Pet. Tech. (July 1971) 814.
7. Correct interpretation of production performance and 13 LEUNG, W.F.: “A Fast Convolution Method for
material balance is important in property evaluation. Implementing Single-Porosity Finite/Infinite Aquifer
8. The use of the interpretation techniques outlined in Models for Water-Influx Calculations”, SPE Res.Eng.
this paper may substantially affect the ongoing dispute (September 1986) 490.
between operators of SAGD oil sand projects and 14 LEUNG, W.F.: “A New Pseudosteady-State Model
shallow gas projects. for Dual-Porosity/Dual-Permeability Aquifers and
Two Interconnected Single-Porosity Aquifers”, SPE
Res.Eng. (September 1986) 511.
ACKNOWLEDGMENTS
15 VOGT, J.P. and WANG, B.: “Accurate Formulas for
The author is grateful for the assistance from his Calculating the Water Influx Superposition Integral”,
daughter, April Lies and Mike Carlson of Applied SPE Eastern Regional Meeting, Pittsburgh, (October
Reservoir Engineering Ltd. in the preparation of this 1987) SPE 17066.
paper.
16 KLINS, M.A., BOUCHARD, A.J. and CABLE, C.L.:
“A Polynomial approach to the van Everdingen-Hurst
REFERENCES Dimensionless Variables for Water Encroachment”,
SPE Res.Eng. (February 1988) 320.
1 SCHITHUIS, R.J.: “Active Oil and Reservoir
Energy”, Trans., AIME (1936) 118, 37. 17 PETROLEUM SOCIETY MONOGRAPH NO 1.:
Determination of Oil and Gas Reserves, Petroleum
2 HURST, W.: “Water Influx into a Reservoir and its
Society of the CIM, Calgary Section (1994) 18,149.
Application to the Equation of Volumetric Balance”,
Trans., AIME (1943) 151, 57.
3 VAN EVERDINGEN, A.F. and HURST, W.: “The AUTHOR BIOGRAPHY
Application of the Laplace Transformation to Flow H.K.Lies is the president of HyPex Consulting Ltd.
Problems in Reservoirs”, Trans., AIME (1949) 186, HyPex provides specialized consulting on the Geologic
305. Fluid System.
4 VAN EVERDINGEN, A.F., TIMMERMAN, E.H. The Geologic Fluid System is an integrated analysis of
and MCMAHON, J.J.: “Application of the Material Geology, Hydrogeology, Geochemistry and Reservoir
Balance Equation to a Partial Water-Drive Reservoir”, Engineering. He has over 25 years of varied petroleum
Trans., AIME (1953) 198, 51. industry experience which include positions in Special
5 HURST, W.: “The Simplification of the Material Projects, Evaluations, Banking , Operations, Reservoir
Balance Formulas by the Laplace Transformation”, Engineering and as president of a Petroleum
Trans., AIME (1958) 213, 292. Hydrogeological research and consulting company. Mr.
Lies hold a B.Sc. degree in electrical engineering from the
6 CRAFT, B.C. and HAWKINS, M.F.:Applied
University of Alberta (1976). He is a registered
Petroleum Reservoir Engineering, Prentice-Hall,
professional engineer in Alberta and is a member of the
Englemood Cliffs, N.J. (1959
CSPG and the Petroleum Society of CIM.
7 CARTER, R.D. and TRACY, G.W.: “An Improved
Method for Calculating Water Influx”, Trans., AIME
(1960) 219, 415.
8 HAVENLA, D. and ODEH, A.S.: “The Material
Balance as an Equation of a Straight Line”, Jour. Pet.
Tech. (August 1963) 896.
9 HUBBARD, R.M. and ELENBAAS, J.R.:
“Determining Gas-Filled Pore Volume in a Water-
Drive Gas-Storage Reservoir”, Jour. Pet. Tech. (April
1964) 383
10 BRUNS, J.R., FETKOVICH, M.J. and MEITZEN,
V.C.: “The Effect of water on p/Z-Cumulative Gas
Production Curves”, Jour. Pet. Tech. (March 1965)
287.
11 AGARWAL, R.G., AL-HUSSAINY, R. and RAMEY,
JR.H.J.: “The Importance of Water Influx in Gas
Reservoirs”, Jour. Pet. Tech. (November 1965) 1336.

Das könnte Ihnen auch gefallen