Вы находитесь на странице: 1из 15

SPE 81009

P/Z Analysis of a Mature Gas Condensate Field, Offshore Trinidad


Brian J Baptiste, SPE, EOG Resources Trinidad Limited; Tennyson Jagai, SPE, The University of the West Indies
Copyright 2003, Society of Petroleum Engineers Inc.
This paper was prepared for presentation at the SPE Latin American and Caribbean Petroleum
Engineering Conference held in Port-of-Spain, Trinidad, West Indies, 2730 April 2003.
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, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes 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 where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836 U.S.A., fax 01-972-952-9435.

Abstract
Graphical application of the gas Material Balance Equation
(MBE), P/Z plots, have typically been used to predict the gas
in place (GIP) of volumetric reservoirs. Few reservoirs
however, are truly volumetric and several workers have
developed procedures to correct such P/Z plots, particularly in
over pressured regimes, for the pressure maintaining effects of
rock collapse and/or shale water influx.
The paper examines and discusses the applicability of
eight (8) such procedures to the mildly overpressured,
relatively low yield, gas condensate Kiskadee Field in which
the original GIP is known from late life P/Z data and in which
pressure maintaining effects on the early life P/Z data were not
previously recognised.
Production, bottom hole pressure, PVT and pore volume
compressibility data are used to analyse the early and late life
P/Z data of each identified fault block and to determine the
extent and causes of apparent early life pressure maintenance
as well as observed variations in the degree of pressure
maintenance in individual blocks. A modification is proposed
to the Bourgoyne procedure that accurately corrects the early
life P/Z data of mildly overpressured reservoirs. A further
modification is proposed to correct late life P/Z data for the
effects of decreasing shale water influx and increasing water
saturation. This further modification is successfully applied
and results in GIP estimates that are within 3 % of the known
volume.
Introduction
An estimate of the Original Gas In Place (OGIP or Gi) for a
volumetric gas reservoir can be obtained from volumetric gas
material balance considerations and yields:
P Z = Pi Z i (Pi Z i Gi )G p ..(1)
This linear relationship is the expression of a constant volume
reservoir and assumes that rock and water expansion are
negligible and that there is no net movement of gas into or out

of the reservoir volume of interest. A material balance plot of


P/Z vs. Gp for a volumetric, depletion drive gas reservoir
generates a straight line of slope (Pi / ZiGi) with an intercept
of Pi / Zi for Gp = 0. Extrapolation of the straight line to the Gp
axis i.e. P/Z = 0, yields the OGIP.
Many gas or low yield gas condensate reservoirs do not
yield linear plots but instead may curve upwards or
downwards2. The constant volume assumption is then not
valid. In addition to bad data, an upwards or downwards
curving P/Z plot may be due to a number of reasons inclusive
of water influx, drainage/leakage into the reservoir,
subsidence/compaction drive, expanding oil rim (upwards
curving) or retrograde condensation, drainage/leakage out of
the reservoir, competitor overproduction and/or overpressured
reservoir (downwards curving). Retrograde condensation is
usually not significant enough to result in downwards
curvature13.
Downward curving plots usually plot more accurately as a
continuous curve but are generally approximated by two (2)
linear segments with distinct slopes that yield a good fit to the
end points of the data set with a poor match in the area of
intersection9. The slope of the early life data (the first slope)
extrapolates to an Apparent Gas In Place (AGIP). The late life
data plots with a higher slope and extrapolates to the OGIP.
Dependent on the severity of the downward curvature,
extrapolation of early life data may yield grossly inflated
estimates of the GIP with consequent deleterious effects on the
ability of the Operator to efficiently manage and produce the
field. This is particularly important where long-term
contractual arrangements are in force for the supply of gas
from the field as pertains for the majority of Trinidads East
Coast gas reservoirs.
Kiskadee Field. The Kiskadee Field is located 40 kilometres
off the eastern coast of Trinidad in approximately 270 of
water. The field was discovered in 1977 with the drilling of
two (2) exploration wells that tested gas-condensate from the
primary reservoir sands at 15,500 subsea (Kiskadee Sand).
The sands are mildly overpressure (0.55 psi/ft), Pliocene age,
shallow marine, deltaic deposits with an average porosity of
25% and permeabilities within the cleaner sands of between
50 to 100 mD. The field lies along the gently dipping,
northwest plunging structural nose of a major (Cassia)
anticlinal structure. Primary structural elements are northwest
trending normal synthetic faults, downthrown to the northeast,
that divide the field into four (4) productive fault blocks (FBs)
that are as shown in Figure 1.
Fault throws vary from one hundred to several hundred
feet. Secondary structural elements are northwest trending

SPE 81009

antithetic faults with western hanging walls. Hydrocarbon


trapping is both structural and stratigraphic. Structural traps
consist of updip closures against the synthetic and antithetic
faults. The main stratigraphic trap is the shale-out of the
Kiskadee sand to the southeast.
The first development well was drilled in 1993 and field
development was completed in 1995. Of the eight (8) wells
drilled, one encountered the objective sands wet in a
downthrown and isolated fault block. The remaining seven
wells encountered full hydrocarbon columns with no gas water
contacts. Cumulative production to mid 2002 was 400 BCF
and 12.3 MMBC.
Previous Work
The only documented study of the Kiskadee Field that utilised
P/Z plots for the prediction of GIP was a proprietary study of
1996, which utilised bottomhole pressure data and production
data available to end 1995. That data set consisted entirely of
early life P/Z values and were therefore straight lines, which
showed no change in slope. The extrapolation of these data
yielded inflated estimates of OGIP, which were explained by
the invocation of strong water drives. The current P/Z plots
are based on pressure and production data to January 2000 and
indicate a marked change in slope (downward curving)
beyond end 1995 (late life data) for FBs 1, 3 and 4. The early
life, straight line P/Z plot of FB 2 has continued with the late
life data with only a minor change in slope relative to the other
fault blocks.
Various methods2 have been proposed to correct the slope
of early life data to yield straight line P/Z plots to the OGIP.
Hammerlindls 1971 study7 of abnormally pressured
depletion-type gas reservoirs indicated that the initial early life
slope is due to pressure maintenance as a result of formation
compaction, crystal expansion and water expansion which act
to reduce the available gas pore volume. Early life reservoir
performance is therefore dictated by the effective
compressibility of the system, which may be several times
greater than the gas compressibility. Effective compressibility,
Ceff, is given by:
Ceff = C g S g + Cw S w + C f / S g ..(2)

The second, later life slope begins at approximately normal


reservoir pressure by which time formation compaction is
complete and the reservoir behaves as a normal depletion type
reservoir where gas compressibility is approximately equal to
the effective compressibility.
Hammerlindl proposed two (2) methods to correct the
AGIP estimates derived from extrapolation of P/Z curve of the
early life data. His first method, referred to in this paper as
Method A, is a P/Z correction procedure where OGIP is given
by
..(3)
OGIP = AGIP Ceff Cg

where

(C

eff

avg

Cg )avg = (Ceff Cg )i + (Ceff Cg )h 2 .(4)

and h = a pressure above the normal pressure gradient.


AGIP is obtained by straight line extrapolation of the early life
P/Z plot and
C g = (1 P ) (1 Z )(Z P ) ..(5)

Hammerlindls second method, referred to herein as Method


B, incorporates Cf in the expansion term of the MBE and the
OGIP is calculated from the equation
OGIP = (AGIP )(G pr )(6)
where Gpr is a correction factor derived from the division of
the MBE used for normally pressured reservoirs by the MBE
used for abnormally pressured reservoirs i.e.
(B Bgh )
(7)
G = gi

[(B

pr

gi

Bgh ) + (Bgh P(C f + S si C w )) (1 S wi )


4

Bourgoynes modification of the depletion type MBE for


an abnormally pressured gas reservoir in equilibrium with a
small aquifer, the shale immediately surrounding the reservoir
and aquifer and the shale stringers within the reservoir yields
Pi

C e P 2 (8)
C e Pi P

G = m
1

Z i

1 S Z

1 S w Z

The constant m is obtained from the observed values of P and


Gp and can be used to estimate the OGIP.
Bourgoyne calculated shale compressibility from the
relationship
Depth P ..(9)
0 .1
Cs =
log10
Pi P
Depth Pi
Effective compressibility in Equation (3.1.7) is given by
f
C = C (1 + R ) + C (S + R ) + C (1 + R ) .(10)
e

where f refers to the shale fraction and R is the aquifer to gas


reservoir ratio.
Ramagost and Farshad10 rearranged the MBE of an
abnormally pressured, depletion type reservoir to the form
P P (S w C w +C f ) Pi G p .(11)

1
= 1
1 Sw
Z
G
Zi
which differs from the straight line equation MBE of a
normally pressured depletion type gas reservoir by the P/Z
Adjustment Factor
P(S w C w +C f ) (12)

1
1 Sw

which is used to correct early life P/Z data to yield a straight


line P/Z plot to the OGIP.
Roach11 independently derived an equation identical to
Equation (11) of Ramagost and Farshad but referred to their
P/Z adjustment factor as Ct and rearranged the equation to
yield
1
= Er (13)
G
where
[(P Z )i (P Z )] 1 , = [(P Z )i (P Z )]G and
=
p
P
P

Er =

S wC w + C f
1 Sw

A plot of versus therefore results in a straight line of slope


1/G and a y-axis intercept of Er. The latter term, as for the
P/Z Adjustment Factor of Ramagost and Farshad, accounts for
the reservoir drive supplied by rock and water
compressibilities that are additional to that supplied by gas
expansion.

SPE 81009

Bernard3 also proposed a modification similar to that of


Roach and of Ramagost and Farshad to the classic depletion
drive equation i.e.
G
P
(1 C P ) = Pi 1 p .(14)
Zi
G
Z
Bernard however, suggested that the term (1-C.P) is a
catch-all approximation for the effects of rock and water
compressibilities, steady state water influx from a small,
active aquifer and from shales. In the absence of any water
influx, C would then be equal to (SwCw + Cf)/(1-Sw) i.e.

(1-C P) of Bernard = 1 P(S w C w +C f ) of Ramagost and

1 Sw

Farshad = Ct of Roach.
Bernard acknowledged that this method yields meaningful
results only with late life P/Z data. Based on a statistical study
of OGIP/AGIP ratios calculated for 13 US Gulf Coast
reservoirs, Bernard therefore proposed that a correction factor
of 0.7 applied to AGIPs obtained by extrapolation of early life
P/Z data would yield a reliable estimate of OGIP.
Begland and Whitehead1 presented a computer program model
that accounted for the pressure dependence of pore volume
and water compressibilities and the effect of their changing
values with decreasing reservoir pressure on OGIP estimates.
They concluded that ignoring pressure dependence could
result in up to a 25% overestimate of OGIP.
Data
Kiskadee wells and fault blocks have had fairly consistent and
adequate pressure sampling to date. The pressure data set
consists of 53 pressure measurements, of which 29 are from
static tests and 34 are from buildup tests. A total of 37 tests
have been recorded with high quality, quartz memory gauges.
An initial reservoir pressure was recorded for each well at the
time of initial completion.
The Gas Deviation Factor, Z, is derived from a tuned,
constant volume depletion simulation at 238 oF of a Kiskadee
composite fluid whose composition was derived from the
arithmetic average of four (4) wells in three (3) fault blocks
and is given by Equation 15:

Z = (1.00065) + 3.40115 101 (Pi Psat )

0.862

+ 5.60428 101 (Pi Psat )

1.862

Cumulative gas production, Gp includes the gas


equivalent of condensate production, GE and the gas
equivalent of condensed water production, GEw which are
obtained from he following relationships:.
GE = 133000( o M o ) scf / STB

GEw = 7390 scf of gas/STB of water


P/Z analyses that do not consider Cf may significantly
overestimate the gas in place8. Pore volume compressibility
data for the subject reservoir do not exist but are available for
a petrophysically similar reservoir at a depth approximately
1,500 feet above the subject reservoir. The data, with a
compensation factor of 100%, are shown in Figure 2, which
indicates that the variation of pore volume compressibility
with net confining pressure is a continuous function that can
be described by the following relationships:
Lower limit at 78 oF: C f = 28075(NCP )0.9704 . (16)
Upper limit at 78 oF: C f = 33009(NCP )0.9659 ..(17)

Little literature exists on the variation of pore volume


compressibility with temperature. Von Gonten and Choudary14
observed that pore volume compressibility at 400 oF averaged
21% higher than at room temperature. A prorated temperature
correction factor of 1.11 is applied to the pore volume
compressibility to account for the known average reservoir
temperature of 240 oF.
The variation in depth of 1,500 between the actual data
and the subject reservoir is accounted for by the variation in
net confining pressure with depth and no further depth
correction is necessary. An overburden pressure gradient of 1
psi per foot is assumed. Water compressibility is assumed to
be constant at 3 x 10-6 psi-1.
All pressure compartments are currently below the Dew
Point, but were initially within 1,500 psi above the Dew Point.
Reservoir fluid studies predict that retrograde liquid
condensate (volatile oil) volume in the reservoir pore spaces
will not exceed a maximum of 5 % which will be insufficient
to achieve liquid mobility. The effect of this retrograde liquid
condensation on pressure maintenance and relative
permeability is considered to be negligible. Initial and current
condensate yields are 48 and 22 barrels / MMCF respectively.
Correction of P/Z plots.
P/Z plots were generated for each of the four (4) fault blocks
of the field. Figures 3 and 4 are examples of these plots for
FBs 1 and 2 respectively. FBs 1, 3 and 4 show a marked
downward curvature in their P/Z plots. The P/Z plot of FB 2
can be closely approximated by a straight line, but the
overpressured and normal pressured segments when plotted
separately, indicate a minor change in slope between both
segments relative to the other fault blocks.
Given the relative maturity of the Kiskadee Field as
evidenced by the number of late life data points, the OGIP
obtained from the P/Z plots can be assumed to be reasonably
accurate and is referred to as the known GIP. It is within 3 %
of that calculated from reservoir simulation models. A
comparison of the AGIPs and OGIPs (known GIP) calculated
from the P/Z plots is shown in Table 1.
The significant overestimate of GIP from the early life
data of FBs 1, 3 and 4 are indicative of a pressure support
mechanism that is active during the early life of the reservoir
and which is decreasing in intensity or effect with time i.e. the
pressure support mechanism is either inactive or
insignificantly active during the late life of the reservoir as
defined by the steeper P/Z slope.
The severity of the early life overestimate of gas in place
varies for each of fault blocks and indicates that the early life
pressure support mechanism is not equally active or effective
for each block.
Application of Hammerlindls Methods to P/Z plots. The
P/Z and Material Balance corrections of Hammerlindl were
applied to the early life P/Z data of each Fault Block. Three
(3) variations were used within the Hammerlindl P/Z Method
A and two (2) variations were applied to the Material Balance
Method B. Variation 1 in both methods utilises a constant pore
volume compressibility, Cf, in accordance with the original
procedure proposed by Hammerlindl. Variation 2 within the
P/Z Method A utilises Equation (17) to recognise and account
for the pressure dependence of Cf in accordance with the

method of Begland and Whitehead with the exception that Cw


is kept constant. The change of Cf with pressure (Variation 2)
has only a marginal effect and yields a maximum increase
over Variation 1 in calculated OGIP of only 1.6%. This is
expected because of the initial low confining pressure which
results in values of Cf which lie within the lower portion of the
pore volume compressibility curve of Figure 4.3.3 over the
investigated pressure range. The results obtained therefore did
not warrant the application of Variation 2 to the Material
Balance Method B. Variation 3 of the P/Z Method A and
Variation 2 of the Material Balance Method B determines by
trial and error, the value of Cf required to correct early life
data to the known GIP. The results are shown in Table 2.
The known Cf (variation 1) as a percentage of the Cf that is
required to be applied to the early life data in order to match
the known GIP were calculated. For FBs 2 and 3, where early
life pressure maintenance appears to be least evident from the
P/Z plots, pore volume compressibility as determined from
Equation (17) is within 2 microsips of that required to correct
the early life AGIP to the known GIP. Pore volume
compressibilities are however, only 23% and 36% respectively
of that which would be required to fully account for early life
pressure maintenance in FBs 1 and 4.
While Hammerlindls Methods do not fully account for
the observed pressure maintenance in all blocks, it does result
in a substantial reduction in the over estimates of GIP when
applied to early life P/Z data.
However, it is apparent that Hammerlindls Methods are
most suitable for application to FB 2 which is the block that
exhibits the lowest magnitude of early life pressure
maintenance and are most inaccurate for FBs 1 and 4 which
both exhibit the highest level of pressure maintenance based
on the P/Z plots.
Application of the Method of Ramagost and Farshad to
P/Z plots. The P/Z Adjustment Factor of Ramagost and
Farshad was applied to the early life P/Z data of each fault
block. The use of this Adjustment Factor should yield a
corrected straight line P/Z plot to the OGIP. Pore volume
compressibility was calculated as a function of pressure.
Water compressibility and water saturation were however,
assumed constant in accordance with the calculation methods
of Ramagost and Farshad.
Figure 5 is an example of the P/Z plot for FB 4, which
shows both the uncorrected and the Ramagost and Farshad corrected early life data.
The Ramagost and Farshad correction is more accurate in
correcting the AGIP obtained from extrapolation of the early
life data where the uncorrected P/Z plots indicate relatively
low early life pressure maintenance (FB 2), whereas the
correction is most inadequate for those blocks that exhibit
marked early life pressure maintenance (FBs 1 and 4).
The Ramagost and Farshad corrected estimates are
contained in Table 1. The results illustrate the marked
similarity between the Ramagost and Farshad and the
Hammerlindl methods and indicate that with the exception of
FB 2, the Ramagost and Farshad method is only marginally
less accurate than that of Hammerlindl.
Application of the Method of Roach to P/Z plots. The P/Z
correction procedure proposed by Roach was modified to
allow pore volume compressibility to be calculated as a

SPE 81009

function of pressure. Water compressibility is assumed


constant. An example of a plot of alpha (A) versus beta (B) is
shown in Figure 6 for FB 1. Roachs alpha-beta plot of the
Anderson L reservoir of the south Texas Mobil-David Field
yielded early time data scatter identical to that observed in this
paper. Early data scatter is particularly severe for FB 1 and
incorporates all of the early life data. The scatter is however,
limited to only the first data point for FBs 3 and 4 and to the
first two (2) points of FB 2. Roach attributed the scatter to the
procedures sensitivity to errors in pressure measurements
early in the reservoir life.
When the scatter of early life data is neglected in
accordance with Roachs procedure, the remaining points
yield good straight line relationships. These are utilised to
calculate the original GIP and the water and pore volume
compressibility correction factor, Er, as proposed by Roach.
These are summarised in Table 1.
The results are particularly interesting and are at a major
variance with the results of the Hammerlindl and the
Ramagost and Farshad procedures in that the calculated GIP is
consistently underestimated relative to the known GIP. This
method therefore overcorrects for pressure maintenance in
each fault block. As for the previous methods however, the
major variance is observed for FBs 1 and 4 (underestimates of
27% and 37% respectively) where early life pressure
maintenance is most pronounced. Similarly, the procedure is
most accurate for FB 2 followed by FB 3 where the
underestimates of GIP are 3.5% and 9% respectively. Fault
blocks that show the largest variances between the known and
calculated GIP also exhibit the largest variances between the
Required Cf of the Hammerlindls methods and Roachs Er
and vice versa. This implies that a large variance in gas
estimates is directly correlatable with too large a correction
factor.
Whereas the Hammerlindl and the Ramagost and Farshad
methods utilise the early life data for correction, the method of
Roach utilises either the late life data or both the early and late
life data dependent on the severity of the early life data scatter.
The different procedures all employ the rock collapse theory
of pressure maintenance and the same or similar general
parameters but yield markedly different results. It is therefore
plausible that the reasons for the significant difference in
results between the methods of Hammerlindl and Ramagost
and Farshad and that of Roach are the use of late life data in
the latter method and the overcorrection of that data i.e.
corrections are being applied to the late life data for pressure
maintenance effects that are not present or are negligible.
Application of the Methods of Bernard to P/Z plots.
Bernards procedure for estimation of OGIP (G) and C
(catch-all term) were calculated on the basis of late life data.
As for the Roach procedure above, Table 1 indicates that the
Bernard correction to the standard MBE (procedure B1 of
Table 1) results in a calculated GIP that is significantly lower
than the known GIP.
The estimates of the GIP and the correction factors of
Roach and those of Bernard are remarkably similar for FBs 1,
2 and 3. This is expected as both procedures utilise late life
data. While the Roach procedure, unlike that of Bernard, also
incorporates early life data, the early life data points scatter for
FB 1 eliminated all those points from analysis with less

SPE 81009

scatter, elimination (and therefore greater discrepancies) for


the other blocks. The largest discrepancy is observed for FB 4
and is due to the greater early life pressure maintenance of that
block.
Bernards statistical correction factor of 0.7 applied to
AGIPs obtained by extrapolation of early life P/Z data yields
estimated OGIPs (Bernards Statistical GIP) as listed in Table
1as B2. While the statistical correction is reasonably accurate
(approximately 11% variance) for FBs 3 and 4, the wide
variance observed for FBs 1 and 2 (184% and 82%
overestimates respectively) indicates that the application of a
statistical correction factor is unreliable.
Application of the Method of Bourgoyne to P/Z plots. The
Bourgoyne relationship shown in Equation (3.1.7) above is
rearranged to yield
P P
C P C P
G p = m i 1 e i + e (18)
Zi Z 1 S w 1 S w
If the various P/Z adjustment or correction factors of
Ramagost and Farshad, Roach and Bernard are denoted by X,
then their modification to the depletion type MBE can be
written as
G
P
[X ] = Pi 1 p ..(19)
Z
Zi
G
and rearranged as

Gp =

Zi
G
Pi

Pi P

( X ) (20)
Zi Z

Comparing equations (18) and (20) yields


ZG
m = i .(21)
Pi
and
CP
CP
C P ....(22)
X = 1 e i + e = 1 e
1 Sw 1 Sw
1 Sw
The Bourgoyne equation can therefore be written in a
modified form that is identical to those of Ramagost, Roach
and Bernard i.e.

P Ce P Pi G p ..(23)

1
= 1
Z 1 S w Zi
G

A plot of (P/Z)X versus Gp should yield a straight line that


extrapolates to the OGIP.
This modification to the Bourgoyne procedure to allow
the use of his effective compressibility term for the generation
of a straight line P/Z plot represents a significant departure
from Bourgoynes use of the term in a polynomial equation to
generate a continuous P/Z convex curve. The modification is
referred to in this paper as Bourgoynes Variation 1. The use
of Equation (8) to generate a continuous curve as described by
Bourgoyne, 1972, is referred in this paper as Bourgoynes
Variation 2. Both variations to the Bourgoyne procedure are
further adjusted to allow Cf to be a function of pressure in
accordance with Equation (17) and to be corrected to reservoir
temperature. Shale compressibility is a function of pressure in
accordance with Equation (9). Porosity, water compressibility
and the shale fraction, f, of the reservoir and aquifer
contributing to water influx are assumed constant in
accordance with Bourgoynes example calculation.

Application of Bourgoynes Variation 1 (BG1) to P/Z plots.


Variation 1 utilises Equation (23) above and is dependent on
the selection of a value of R that is in conformance with the
volumetric and geologic considerations of the individual Fault
Blocks as well as the production and pressure data. R is the
dominant element of Equation (10) that determines Ce and has
significant impact on the calculated OGIP.
A value of R in conformance with volumetric and
geologic considerations was determined for each fault block
by calculating the ratios of the hydrocarbon reservoir to the
areal extent of the contiguous aquifer on a full-scale map
(1:50,000) of Figure 1. The results suggest an R value of
between 1 and 2 for FBs 1 and 4, a value of less than 1 for FB
2 and approximately 2.5 for FB 3 if it is assumed that the
effective aquifer extends to the regional synclinal axis north of
the Kiskadee Field. An R value of less than 5 is considered to
be indicative of a weak water drive.
The late life P/Z plots and the relative absence of saline
water production may be considered to be indicative of the
absence of a strong water drive However, individual fault
blocks have produced cumulatives of between 29% to 44% of
their GIP to the time of the last pressure measurements and
flattening of the respective P/Z curves indicative of water
support may yet occur. Additionally, lack of water production
is not necessarily indicative of the absence of a strong water
drive.
Since Equation (23) dictates that an appropriate value of
R should yield a straight line, corrected P/Z plot, it is inferred
that the value of R which yields the closest straight line fit to
the corrected P/Z data would be indicative of an appropriate R
value for the subject fault block.
Figures 7 is an example of a P/Z plot based on this
modification to the Bourgoyne procedure for FB 1, with R
values of 0, 1.5, 2.5, 5, 7.5 and 10. Also shown for comparison
in these Figures are the uncorrected P/Z plots.
A value of R = 0 is the case of no aquifer effect i.e. the
effective compressibility of Equation (10) is given by

Ce = Cr + Cw S w + C s

f
..(24)

An increase in R implies an increase in aquifer effect and


pressure maintenance. Extrapolation of the corrected P/Z data
will therefore yield decreasing estimates of GIP due to over
correction and a gradual flattening of the curves (Figure 7).
These curves are in conformance with, but show curvatures
opposite to, those of Bruns, Fetkovich and Meitzen5 for
increasing R as those workers solved pressures implicitly for
differing encroachment factors, reserves and geometries. Early
life pressure maintenance of FB 1 (and similarly for FB 4) is
illustrated in Figure 7. At R = 0, the P/Z data is corrected for
rock, water and shale compressibility but downward curvature
is still evident for both blocks. This implies that the above
compressibility corrections are insufficient to account for the
early life pressure maintenance of these fault blocks.
The above and similar plots for FBs 1, 2 and 3 suggest
that appropriate approximate values of R for FBs 1, 2, 3 and 4
are 1.5 to 2.5, 0, 0 and 1.5 respectively.
Figure 8 is an example of a P/Z plot of FB 1, which
illustrates how all fault blocks are corrected by application of

the Bourgoyne Variation 1 modification to the MBE with the


above R values. Also shown for comparison is the uncorrected
P/Z data. Figure 8 for FB 1 is based on an R value of 1.6. This
value was selected on the basis of the GIP predicted by the
early life data and by both the early and late life data for R
values of 1.5, 1.6, 1.7 and 1.8. The value of R = 1.6 most
accurately corrects the early life data to within 1% of the
known GIP and is accepted as the most appropriate value of R
for FB 1 on the basis that pressure support mechanisms that
affect the P/Z curves (rock, water, shale compressibilities and
shale water influx) are most predominant during the early life
of the reservoir (Wallace, 1969). At R = 1.6 however, the GIP
obtained by extrapolation of both the late and early life data is
20% lower than the known GIP. It is therefore evident for FB
4, as for FB 1, that the Bourgoyne Variation 1 method is
adequately correcting the early life data but is over-correcting
those of the late life. These observations are further discussed
below.
Results for each of the fault blocks of the application of
Bourgoyne Variation 1are summarized in Table 1 (BG1) and
indicates that application of this method to the early life data
results in estimates that are between 91% - 99% of the known
GIP.
Application of Bourgoynes Variation 2 (BG2) to P/Z plots.
Bourgoyne utilised the known gas in place (m*Pi / Zi) of the
NS2B reservoir of the North Ossum Field (1972) and a
computed Ce to show that the resultant P/Z versus Gp plot is
represented by a continuous curve from Pi / Zi to the OGIP (no
linear segments) which conforms to Bourgoynes Equation (8)
above. This equation can be rearranged to the polynomial
form
2
P
C P P C Z P
G p = m i 1 e i e ...(25)
Z i 1 S w Z 1 S w Z
This second variation of the Bourgoyne procedure was tested.
Since the gas in place for each fault block is known, a second
order polynomial curve of the form shown in Equation (25)
fitted to the P/Z data of each block should extrapolate to the
GIP.
Figure 9 is an example of the P/Z plots of FB 1, which
illustrated how the data of each block is best fitted with a
second order polynomial curve and the curve is extrapolated to
a P/Z value of zero. This curve is shown as a solid line. The
curve fits to the data are excellent. The equation for FB 1 is
shown. The polynomial curves of each fault block conform to
the Bourgoyne Equation (8). The extrapolation of these P/Z
polynomial curves to P/Z = 0 yields estimates of the GIP for
each fault block which are summarized in Table 1.
Also shown in Figure 9 is a dashed straight line that
connects the initial P/Z to the known GIP. This line would
conform to the P/Z curve of a volumetric gas reservoir
corrected for early life pressure maintenance or of a truly
volumetric gas reservoir with no pressure maintenance.
Extrapolation of the polynomial curve to P/Z = 0 assumes
a continued rate of decrease of pressure maintenance that is in
excess of that which occurs or the existence of continued
pressure support that is actually no longer active and results in
an underestimate of the volumetric GIP.
The initial position of the dashed straight line below the
polynomial curve indicates that pressure support mechanisms

SPE 81009

are active in the reservoir. The area between both curves may
be considered to be indicative of the level of uncorrected
pressure support. Similarly, the later position of the
polynomial curve below the dashed straight line in Figure 9 is
indicative of over-correction of the data and the area contained
between both curves is qualitatively indicative of the extent of
over-correction for pressure maintenance that is no longer
active or that is only minimally active.
Of note is that while this Variation 2 of the Bourgoyne
procedure results in an underestimate of the GIP for FBs 1, 3
and 4, which exhibit a change in their P/Z slopes, the
procedure marginally overestimates the GIP for FB 2. This is
not unexpected as the uncorrected P/Z data of that fault block
plots as a relatively straight line indicative of reservoir
behaviour that is close to volumetric.
Discussion of P/Z Correction Procedures.
Table 1 summarises the results of the eight (8) different
procedures, inclusive of their variations, investigated above
for the estimation of GIP from P/Z plots corrected for rock and
water compressibilities and shale water influx. Variances to
the known gas in place are shown in italics. Negative
variances are indicated by brackets. Also indicated by the
letters EPS are the fault blocks that show evidence of
significant early life pressure maintenance based on P/Z plot
curvatures.
Bernards statistical procedure (B2), which was shown to
be unreliable, is omitted from further discussion.
Immediately noticeable is that the rock collapse theories of
Hammerlindl and Roach et al result in significant and
unacceptable overestimates of the gas in place when applied to
the early life data of EPS type FBs 1 and 4. However, when
these theories are applied to the early life data of FBs 2 and 3,
which show minimal early life support, the results indicate
relatively small and acceptable variances to the known GIP.
These variances are the lowest of the eight (8) procedures
studied for FBs 2 and 3 and indicate that rock and water
compressibility corrections are sufficient to account for all of
the pressure maintenance mechanisms additional to gas
expansion for FB 2 and for the majority of the early life
pressure maintenance of FB 3.
The negative variance of each fault block with the
application of Roachs rock collapse theory confirms that the
corrections for rock and water compressibilities that are
applied to the early life P/Z data are not required to be applied
to the late life data. This would be in accordance with the
theory of Hammerlindl that envisages formation compaction
to be essentially complete at normal reservoir pressure with
normal depletion type behaviour below that pressure. The
same applies for the negative variances to the known GIP,
which result from the catch-all procedure of Bernard.
Bourgoynes Variation 1, when applied to the early life
data of EPS FBs 1 and 4, yields the lowest variance to the
known GIP of all methods investigated. These variances are
approximately 2 % and 3 % for FBs 1 and 4 respectively and
are excellent approximations to the GIP. Variation 1 of the
Bourgoyne procedure therefore, adequately corrects the early
life data but when applied to both the early and late life data of
the EPS fault blocks the procedure results in significant
negative variances of 18 % and 28 % for FBs 1 and 4

SPE 81009

respectively. This is indicative that, as also pertains for the


rock collapse procedures of Roach and the catch-all method of
Bernard, the early life pressure support mechanisms of EPS
type fault blocks are inactive or not as active in late life and
that the Bourgoynes Variation 1 procedure which is used to
correct the early life data cannot be extended to the late life
period.
It is noted as well, that of the procedures investigated,
those that utilise the catch-all correction result in the largest
variances to the known GIP of FBs 2 and 3. Bourgoynes
Variation 1 in particular yields the largest variance (8 %) to
the known GIP for FB 2 and is among the largest negative
variances (13 %) for FB 3.
Of interest is that application of the eight (8) procedures
results in a fairly substantial range of variances for EPS FBs 1
and 4 with a clearly defined best-fit procedure. The absolute
variance of 2 % for Bourgoynes Variation 1 of FB 1 is
significantly lower than the next closest absolute variance of
18 % (Bourgoynes Variation 2) and is far superior to the
maximum absolute variance of 107 % for the Ramagost and
Farshad procedure. Similarly, the absolute variance of 3 % for
Bourgoynes Variation 1 of FB 4 compares very favorably
with the next closest absolute variance of 15 % (Bernards
Material Balance correction) and the maximum absolute
variance of 47 % for the procedure of Ramagost and Farshad.
It is therefore evident that the Bourgoyne Variation 1
modification to the classical material balance equation for a
volumetric gas reservoir is the most applicable procedure for
the estimation of GIP from the early life data of mildly
overpressured, relatively low yield gas condensate reservoirs
with weak aquifer support. This weak aquifer support (R =
1.5), the inadequacy of solely rock collapse mechanisms to
compensate for early life pressure maintenance and the
applicability of the Bourgoyne method indicates that a
probable and significant early life pressure maintenance
mechanism of FBs 1 and 4 is that of shale water influx from
interbedded and surrounding shales. This is supported by the
results of a detailed review of the water production and
chlorinity analysis of the Field (in press).
Application of the eight (8) procedures above to the P/Z data
is however, not as definitive for FBs 2 and 3 as for the EPS
type blocks. If Bernards statistical approach is excluded, the
absolute variances for FB 2 range from 1 % to a maximum of
8 %. The rock collapse procedures of Hammerlindl and
Ramagost et al clearly yield the most accurate estimate of the
GIP. But a variance to the GIP of less than 10 % as is
observed for all procedures applied to the P/Z data of FB 2,
would be considered by most operating companies to be very
acceptable. These low variances are attributed to one primary
cause i.e. that the reservoir of FB 2 is behaving almost entirely
as a volumetric reservoir. Formation volume compaction,
aquifer support and shale water influx are therefore, not totally
effective pressure support mechanisms in this compartment.
FB 2 is an updip, isolated compartment in partial contact with
an areally limited aquifer. Bourgoynes evaluation suggests an
appropriate R value of zero, which is indicative of a
practically non-existent water drive support. However, initial
reservoir pressure gradient was 0.52 psi/ft, which is slightly
less pressured than the initial, although mild, 0.55 psi/ft of FB
1. It is proposed that the formation compaction (pore volume

compressibility), water compressibility and shale water influx


mechanisms of early life pressure maintenance were in fact
active in Fault Block 2 and may have been almost as active as
for FBs 1 and 4. FB 2 was produced initially at a sustained,
average rate of 105 MMCF/D compared to average rates of 65
MMCF/D, 25 MMCF/D and 30 MMCF/D for FBs 1, 3 and 4
respectively. Of the two (2) wells in FB 2, one was produced
at approximately 70 % of the average rate for the fault block
through 4 tubing compared to the 2 -7/8 tubing in place in
the second well. The inference is drawn that production rates
of FB 2 were sufficiently high that pore space compaction
with decreasing reservoir pressure and the rate of pore space
occupation by shale water influx were insufficient, relative to
the rate of pore space voidage by production, to act as pressure
maintenance mechanisms.
Application of the Hammerlindl and the Ramagost and
Farshad methods to FB 3 most closely approximate the GIP of
that block with a positive variance of 7 % and 9 %
respectively. As for FB 2, most of the remaining pressure
support is assumed to be by shale water influx. Excluding
Bernards statistical correction procedure, the next closest
approximation based on early life data is that of Bourgoynes
Variation 1 which yields a negative variance of 13 %. The
rock collapse procedures of Hammerlindl and of Ramagost
and Farshad do not fully correct for the early life pressure
maintenance, but they do account for the majority of that
pressure support. With aquifer support considered as
negligible (R = 0), the additional correction for shale water
influx with Bourgoynes Variation 1 slightly overestimates the
early life pressure support. It should be noted however, that
this is a relatively small, isolated block. The lowest observed
variance of 7% (rock collapse corrections only) translates in
terms of gas volume to a variance of only 4 BCF to the known
reserves of 56 BCF. Similarly, the 13 % negative variance
observed on application of the Bourgoyne Variation 1
procedure to correct for shale water influx translates to a gas
volume variance of only 7 BCF. This is in good agreement
with the gas volume variances observed for the best fit, early
life procedures of FBs 1, 2 and 4 which are 5 BCF, 4 BCF and
7 BCF respectively. In general, therefore, procedures have
been identified that are capable of correcting the early life data
of each fault block to within 10 % and to within 10 BCF of the
GIP.
The phenomena of the underestimation of GIP (negative
variances) when late life P/Z data are corrected for pressure
maintenance can be explained on the basis of shale water
influx into the producing reservoir. The mechanics of water
flow in shales is a function of several shale properties
inclusive of shale porosity and permeability, shale
compressibility and shale water viscosity. Bourgoyne et al,
1972, considers shale permeability to be the most important
factor in the ability of a shale to allow water influx. He
concluded on the basis of his study that shales with an initial
permeability of the order of 10-6 mD and a compressibility of
the order of 40 microsips could significantly affect reservoir
performance by shale water influx with such influx becoming
negligible as shale permeability decreased to 10-7 mD.
Additionally, Bourgoyne proposed that pressure support by
shale water influx was sensitive to reservoir production rates.
Further, such pressure support would also decrease with

reservoir depletion as shale permeability decreased with


increasing effective overburden pressure and exsolution of
dissolved shale water gas. This is in agreement with the study
by Wallace15 of overpressured South Louisiana gas reservoirs,
in which he demonstrated that while maximum shale water is
available at high pressures, compaction results in shale water
influx at pressures less than hydrostatic and continues with
declining pressure at proportionately lower rates. Shale
permeabilities and compressibilities in the subject area are
currently unknown but may be as high as 10-5 mD in
accordance with Bourgoynes listing of permeability data on
the basis of log character and core description and are
sufficiently permeable to allow water flow. It can therefore be
expected that shale water influx in mildly overpressured
reservoirs would occur early in the life of the reservoir and
would decrease substantially and eventually cease as pressures
decline to, and below, normal.
Entry of shale water from adjacent shales into the
reservoir will have the effect of increasing the water saturation
within the first few feet adjacent to the contributing shales but
may have a greater impact on the entire bed for thinner
sandstone units.
It is generally accepted that the relative permeability to
water in water wet rocks is negligible below a saturation of
2Swi which would imply that a water saturation of 50% is
required to allow water flow within the reservoir. While it is
conceivable that water saturations in the sandstones adjacent
to the contributing shales may approach these saturations,
capillary pressure curves of the reservoir12 indicate that for an
average reservoir permeability of 50 mD, free water will exist
at marginal increases in water saturation above the initial
saturation of 25%.
It is evident therefore, that if shale water influx is
assumed to be an active process then the early life data of the
P/Z plots should be corrected for the consequent pressure
maintaining effects. If shale water influx can be expected to
decrease with decreasing reservoir pressure then P/Z
corrections for shale water influx should not remain constant
throughout the entire productive life of the reservoir. Indeed,
such corrections should gradually decrease to zero at some
time beyond the decline of reservoir pressure to hydrostatic.
Similarly, if it is accepted that shale water influx into reservoir
sands will have the effect of an average increase in the water
saturation of the reservoir, then that increase in water
saturation should be accounted for in the procedures used to
correct the P/Z data of the reservoir. Failure to do so will
result in an overcorrection of the P/Z data and an
underestimation of GIP when the procedures are applied to
late life data.
Modification to late life data correction for the prediction
of gas in place.
The application of Bourgoynes Variation 1 to the P/Z data
was repeated and the parameters of Sw and f were decreased
with time to allow a calculation of GIP that is close to the
known value. Adjustment of the shale content factor was
accomplished by consistently decreasing the factor with time
in order to simulate decreasing shale water influx due to
reduced permeability with decreasing pressure.
The results are not meant to accurately model the behaviour of
the reservoir but to demonstrate that the effect of increasing

SPE 81009

water saturation and decreasing shale permeability on the


Bourgoyne corrected late life P/Z data of Variation 1 can be
corrected by accounting for those effects. An example of the
results is shown in Figure 10 for FB 1 and the results are
summarised in Table 3.
Although the water saturations required to match the data do
not exceed 27 %, localised saturations may significantly
exceed that value and the 27 % saturation is interpreted to be
an average reservoir wide value.
The estimated GIP is however, more sensitive to variations in
the shale content and by extension, to variations in the influx
of shale water. Reasonable matches to the known GIP are
achieved for all blocks by assuming that the effect of shale
water influx is negligible at or just below normal pressure.
This modification is required to account for increasing water
saturation, where applicable, but moreso to account for
decreasing shale water influx effect. The modification yields a
resulting GIP that is within 3 % or less of the known GIP.
Unfortunately, it is practically impossible to predict or
determine the variation of shale water influx and the variation
in reservoir water saturation with decreasing reservoir
pressure. However, the following guidelines, in accordance
with the above procedures, can be utilised in the field
application of the Bourgoyne Variation 1 procedure to early
and late life P/Z data:
1. Estimate R based on geologic data and field analogues
2. Correct the early life P/Z data in accordance with
Bourgoynes Variation 1 procedure
3. Obtain the best fit straight line through the early life data
up to the hydrostatic P/Z equivalent and project to the GIP
4. Adjust subsequent P/Z data points that depart
significantly from the straight line established above back
unto the straight line by adjusting the water saturation
and/or shale values. For small reservoirs and/or high
production rates assume a constant water saturation.
The above guidelines may minimise under estimates of the
GIP that will occur when Bourgoynes Variation 1 is applied
to acquired late life P/Z data and the early life estimates are
revised to accommodate the latter data.
Conclusions.
1. Shale water influx from interbedded and surrounding
shales is a probable and significant early life pressure
maintenance mechanism of FBs 1 and 4.
2. Bourgoynes Variation 1, applied to the early life data of
these blocks, estimates the GIP to within 3% of the
known volume. Selection of an appropriate R is of
particular importance.
3. Production rates of FB 2 were sufficiently high that pore
space compaction with decreasing reservoir pressure and
the rate of pore space occupation by shale water influx
were insufficient, relative to the rate of pore space
voidage by production, to act as pressure maintenance
mechanisms. The Hammerlindl and the Ramagost and
Farshad procedures, applied to the early life data of this
block, estimates the GIP to within 2% of the known
volume.
4. Application of the Hammerlindl and the Ramagost and
Farshad procedures to FB 3 most closely approximate the
GIP with a positive variance of 7 % and 9 % respectively.

SPE 81009

5.

6.

7.

Bourgoynes Variation 1 yields a negative variance of 13


%. These variances represent less than 10 BCF of the
GIP.
The modifications of Hammerlindl and Ramagost and
Farshad to the classical MBE for a volumetric gas
reservoir will yield reliable estimates of GIP from the
early life data of mildly overpressured, relatively low
yield gas condensate reservoirs with negligible aquifer
support but active shale water influx if such reservoirs are
being produced at a sufficiently high rate to negate the
pressure maintaining effects of shale water influx.
At lower production rates, a shale water influx correction
is required and use of the Bourgoyne Variation 1
procedure is recommended as the most applicable
procedure for the estimation of GIP from early life data.
The implication of this observation is that the rock
collapse corrections would have been inadequate for
estimation of the GIP in FB 2 had that block been
produced at similarly constrained gas rates as the other
fault blocks. The shape of the P/Z curve is rate dependent.
An estimation of GIP of all fault blocks can be accurately
made by modifying the application of Bourgoynes
Variation 1 to the late life portion of the P/Z data. The
modification is required to account for increasing water
saturation, where applicable, but moreso to account for
decreasing shale water influx effect. The modification
yields a resulting GIP that is within 3 % or less of the
known GIP.

Recommendations for future work.


1. At current reservoirs pressures, the shales may have
already contributed the majority of their waters. Pressure
increases due to shale water influx may be measured in
currently shut in wells and for wells that are shut in for
extended periods during future development in this or
other fields by daily static pressure measurements from
downhole, high quality pressure gauges15.
2. A study of East Coast shales with emphasis on shale
chemistry, shale water chemistry and shale porosity and
permeability and their variation with depth and pore
pressure.
3. Full hole core analyses of existing or future cores in the
area should investigate the variation of pore volume
compressibility with temperature.
4. A Deliverability Projection study of the field to compare
actual performance with that predicted by deliverability
equations which incorporate the effect of gas permeability
reduction due to pore volume decrease with increasing net
overburden pressure and liquid saturations increase below
the dew point. This study should incorporate the analysis of
backpressure-test curves, which should indicate a shift to
the left as liquid content increases due to shale water influx
and retrograde condensate drop out below the dew point6.
5. A reservoir simulation study is required to determine the
volume and extent of the aquifer acting on FBs 1 and 4 in
order to confirm that only that volume of the aquifer
downdip from Kiskadee to the syncline north of Kiskadee
is acting on the Kiskadee Field. Such a result would
confirm the selected values of R determined in this paper
for those fault blocks.

Acknowledgements.
The authors acknowledge the approvals of EOG Resources
Trinidad Limited, Petroleum Company of Trinidad and
Tobago Limited and The National Gas Company of Trinidad
and Tobago Limited .to submit this paper.
Nomenclature.
= Alpha, P/Z parameter after Roach, 1981, microsips
= Beta, P/Z parameter after Roach, 1981, MMCF/psi
AGIP = Apparent Gas In Place, BCF
Bgi = Initial Gas Formation Volume Factor (scf/ ft3)
Bgh = Gas Formation Volume Factor at a pressure above
hydrostatic (scf/ ft3)
Ct = P/Z Adjustment Factor of Roach, 1981, microsips
C = P/Z Adjustment Factor of Bernard, microsips
Ce = Effective Compressibility of Bourgoyne, 1972,
microsips
Ceff = Effective Compressibility of Hammerlindl, 1971,
microsips
Cf = Pore Volume Compressibility. Also Bourgoynes Cr,
microsips
Cg = Gas Compressibility, microsips Cs
Cw = Water Compressibility, microsips
Cs = Shale Compressibility, microsips
dZ = Change in Gas Compressibilty Factor. Also Z,
microsips
dP = Change in Gas Pressure. Also P, psi
-Er = P/Z Adjustment Factor of Roach, 1981, microsips
f = Shale Fraction
GE = Gas equivalent of liquid, scf/STB
Gi = Initial Gas In Place. Also OGIP, BCF
GEw = Gas equivalent of water, scf/STB
GIP = Gas In Place, BCF
Gpr = Hammerlindl Correction Factor for GIP estimate
Gp = Produced Gas, BCF
k = Permeability, Millidarcies
MBE = Material Balance Equation
Microsip = 10-6 psi-1
m = Slope of P/Z versus Gp, rock and water
compressibilities neglected, Bourgoyne, 1972.
NCP = Net Confining Pressure, psi
P = Pressure, psi
Pi = Initial Pressure, psi
Pavg = Average Pressure, psi
Psat = Saturation Pressure, psig
R = Aquifer to Gas Ratio
Sgi = Initial Gas Saturation
Sg = Gas Saturation
Sw = Connate Water Saturation, fraction
Z = Gas Deviation Factor
Zi = Initial Gas Deviation Factor
References.
1.
2.

Begland, T. and Whitehead, W.: Depletion Performance


of Volumetric High-Pressured Gas Reservoirs, SPE
Reservoir Engineering (August 1989): 279-282.
Bernard, W.: Reservoir Engineering Concepts in
Abnormal Formation Environments. In Abnormal
Pressures. Developments in Petroleum Science, 38. ed., W.

10

SPE 81009

3.

4.

5.
6.

7.

8.

9.

10.

11.
12.

13.
14.

15.

H. Fertl, R.E. Chapman and R.F. Hotz, 93-105. The


Netherlands: (1994) Elsevier Science B.V.
Bernard, W. J.: Gulf Coast Geopressured Gas Reservoirs:
Drive Mechanism and Performance Prediction, paper SPE
14362 presented at the 60th Annual Technical Conference
and Exhibition of the Society of Petroleum Engineers of
AIME, Las Vegas, California, September 1985.
Bourgoyne, A.; Hawkins, M; Lavaquial, F. and
Wickenhauser, T.: Shale Water as a Pressure Support
Mechanism in Superpressured Reservoirs, paper SPE
3851 presented at the 3rd Symposium on Abnormal
Subsurface Pore Pressure, Louisiana State University, May
1972.
Bruns, J.; Fetkovich, M. and Meitzen, V.: The Effect of
Water Influx on p/z Cumulative Gas Production Curves,
JPT (March 1965) 287.
Duggan, J.: The Anderson L An Abnormally pressured
Gas Reservoir in South Texas, paper SPE 2938 presented
at the 45th SPE Fall Meeting, Houston, Texas, October
1970.
Hammerlindl, D.: Predicting Gas Reserves in Abnormally
Pressured Reservoirs, paper SPE 3479 presented at the
46th Annual Fall Technical Conference and Exhibition of
the Society of Petroleum Engineers of AIME, New
Orleans, Louisiana, October 1971.
Harville, D. and Hawkins, M.: Rock Compressibility and
Failure as Reservoir Mechanisms in Geopressured gas
Reservoirs. paper presented at the First Annual
Symposium on Abnormal Sub-surface Pressures, Baton
Rouge, La, April 1967.
Prasad, R. and Rogers, L.: Superpressured Gas Reservoirs:
Case Studies and a Generalised Tank Model, paper SPE
16861 presented at the 62nd Annual Fall Technical
Conference and Exhibition of the Society of Petroleum
Engineers of AIME, Dallas, Texas, September 1987.
Ramagost, B. and Farshad, F.: P/Z Abnormally Pressured
Gas Reservoirs, paper SPE 10125 presented at the 56th
Annual Fall Technical Conference and Exhibition of the
Society of Petroleum Engineers of AIME, San Antonio,
Texas, October 1981.
Roach, R. H.: Analysing Geopressured Reservoirs A
Material Balance Technique, paper SPE 9968, available at
SPE, Richardson, TX (1981).
Rosen, M.: Geology and Development of the Kiskadee
Field, Offshore Trinidad. Paper presented at the 1997
Offshore Technology Conference, Houston, Texas, May
1997.
Smith, C.; Tracy, G. and Farrar, R.: Applied Reservoir
Engineering, Volume 1. Oil & Gas Consultants,
International, Inc. (1992).
Von Gonten, W. and Choudhary, B.: The Effect of
Pressure
and
Temperature
on
Pore
Volume
Compressibility, paper SPE 2526, presented at the 44th
Annual Fall Meeting of the Society of Petroleum Engineers
of AIME, Denver, Colorado, September 1969.
Wallace, W.: Water Production from Abnormally
Pressured Gas Reservoirs in South Louisiana, Journal of
Petroleum Technology, (1969) Volume 21: 969 983.

SPE 81009

11

FB 4

FB 1
FB 3

FB: Fault Block


LKG: Lowest Known Gas

FB 2

Productive well

Figure 1: Kiskadee Fault Block Map

50

Pore Compressibility, microsips

40

30

20

10

0
0
Figure 2: Kiskadee Pore Volume Compressibility

2000

4000
6000
8000
10000
Net Confining Pressure, psi

12000

14000

12

SPE 81009

700
600
500
400
AGIP 894 BCF

300
200

Overpressure

100

OGIP 340 BCF

0
0

10

15

20

25

30

35

40

45

50

Gas Production, Gp,


Figure 3: P/Z versus Cumulative Production, Fault Block 1

700

Notes:
Overpressured section can be curve fitted by
2
P/Z = -17.065Gp + 6155.3, R = 0.9919
to yield an AGIP of 361 BCF.

600
500

Normal pressured section can be curve fitted by


2
P/Z = -20.117Gp + 6217.9, R = 0.9904 to yield an OGIP of
309 BCF.

400
300
Overpressure

200
100
0
0

10

15
20
25
Gas Production, Gp,

Figure 4: P/Z versus Cumulative Production, Fault Block 2

30

35

40

SPE 81009

13

6400

120

6300
100

GIP = 249

P/Z, P/Z * X

80

6200
Uncorrected
P/Z AGIP = 373
BCF

A, microsips

60

6100

Early life
data

40
20
0
0

6000

10

15

20

30

40

B, MMSCF/psi

20

25

Cumulative Gas Production, Gp, BCF

-60

Er = 61

-80

Figure 5: P/Z and Ramagost and Farshad correctd P/Z Plot, Fault
Block 1

Figure 6: Plot of Roach correctd P/Z data, Fault Block 1

7000
6000
5000
P/Z * X

Uncorrected P/Z

4000

R=0
R = 1.5

3000

R = 2.5
R=5

2000

R = 7.5

1000
R = 10

0
0

50

-40

5900
0

10

-20

Ramagost
GIP
= 301 BCF

50

100

Cumulative Gas Production, Gp, BCF


Figure 7: Effect of R on Bourgoynes Variation 1 Application to P/Z: Fault Block 1

150

14

SPE 81009

7000

Uncorrected Late Life


data. OGIP = 340 BCF

6000

P/Z, P/Z*X

5000

R = 1.6

Corrected Early Life


data. AGIP = 343 BCF

4000
3000
2000

Corrected Early & Late Life


data. AGIP = 274 BCF

1000
0
0

100

200

300

400

500

Cumulative Gas Production, Gp, BCF


Figure 8: Bourgoynes Variation 1 corrected P/Z Data, Fault Block 1

7000

Best Fit Polynomial:

6000

2
P
P
P
G p = i (6.3326 ) (0.0788 )
Z
Z
Z i

5000

P/Z

4000
3000

OGIP
= 340 BCF

2000

AGIP = 248 BCF


1000
0
0

50

100

150

200

250

300

350

Cumulative Gas Production, Gp, BCF


Figure 9: Bourgoynes Variation 2 corrected P/Z Data, Fault Block 1

7000
Variable Sw, f

6000

Uncorrected Late Life data


OGIP = 340 BCF

P/Z, P/Z*X'

5000

Modified Bourgoyne Variation 1


AGIP = 340 BCF

4000
3000
2000
1000
0
0

100

200

300

400

Cum ula tive Ga s Production, Gp, BCF


Figure 10: Modification of Bourgoyne correction to late life data, Fault Block 1

500

SPE 81009

15

Table 1: Comparison of Results of P/Z Correction Procedures

FB
1
(EPS)
2
3

Rock Collapse1
Catch-All2
Data Life
AGIP (P/Z),,
Known GIP, BCF
BCF
894
340
361

309

83

56

H
X

R&F
X

Roach
X

E&L

644
89
304
(2)
60
7
287
40

703
107
313
1
61
9
301
47

249
(27)
298
(4)
51
(9)
130
(37)

4
373
205
(EPS)
1
Includes rock and water compressibilities
2
Includes rock and water compressibilities and shale water influx.
H - Hammerlindl
R & F - Ramagost & Farshad
B1 Bernards Material Balance correction
B2 Bernards Statistical correction
BG1 Bourgoynes Variation 1
BG2 Bourgoynes Variation 2
E Early life data
E & L Early and Late life data

Table 2: Comparison of Hammerlindls GIP, AGIP and


known GIP
Fault
Block

AGIP
(BCF)

Known
GIP
(BCF)

Hammerlindl GIP
(BCF)
P/Z Method A

Mat Bal
Method B

Variation
1
2
3
4

894
361
83
373

340
309
56
205

1
640
302
60
285

2
650
304
60
289

1
644
304
60
287

B1

B2

BG1

X
X
L
E
E
Computed OGIP, BCF
(Variance to known GIP, %)
249
626
335
(27)
84
(2)
298
253
284
(4)
(18)
(8)
47
50
49
(16)
(11)
(13)
174
224
198
(15)
9
(3)

BG2

X
E&L

X
E/E & L

278
(18)
284
(8)
51
(9)
147
(28)

248
(27)
321
4
42
(25)
149
(27)

Table 3: Modified Bourgoyne corrected Early & Late


Life Data
Fault
Block

Known
GIP
(BCF)

1
2
3
4

340
309
56
205

Modified Bourgoynes
Corrected
Early & Late Life
GIP (BCF)
330
302
56
201

Variance
%
(3)
(2)
(0)
(2)

Вам также может понравиться