Improved interpretation of AUTHOR

Alton Brown
Consultant, 1603 Waterview

wireline pressure data Drive, Richardson, Texas, 75080;
[email protected]
Alton Brown Alton Brown worked as a research geologist at
ARCO’s Research Center in Plano, Texas, from
1980 until ARCO’s merger with BP Amoco. Since
then, he has been an independent consultant.
ABSTRACT Research interests include petroleum migration,
carbonate sedimentology and diagenesis, basin
Modern wireline pressure data can have resolution and reproduc- analysis, and gas geochemistry.
ibility sufficient to detect small fluid-density changes and pressure
barriers, yet these features are commonly overlooked on conven-
tional pressure-depth plots. The large pressure variation caused ACKNOWLEDGEMENTS
by weight of subsurface fluids hides these subtle features. Excess This work was completed at the ARCO
pressure is the pressure left after subtracting the weight of a fluid Research Center in Plano, Texas. I thank ARCO
from the total pressure. This concept is applied to wireline pressure and VASTAR for permission to release this
data to remove effects of weight and emphasize subtle pressure study. ARCO and VASTAR have subsequently
differences caused by density variations and pressure barriers. Fluid- become part of BP Amoco, which is also
density changes of 0.02 g/cm3 or less can be resolved, and within- acknowledged for its cooperation. AGIP and
well pressure barriers in the order of 5 kPa (0.7 psi) can be detected. Petroecuador are gratefully acknowledged for
Using good-quality data, effects of reservoir capillary-displacement releasing Villano field pressure data. Paul
pressure can be detected by offset of the free-water level from the Willette, Lee Russell, and Jim Twyman
reviewed earlier drafts of the manuscript.
petroleum-water contact. This effect can be used to estimate reser-
AAPG reviewers Jim Puckette and Alain Huc
voir wettability. Subsurface fluid-density measurements can also
are also acknowledged. David Novak, Andy
be used to evaluate oil or gas quality on a bed-by-bed scale in traps Harper, Paul Willette, and Herb Vickers helped
having variable oil or gas composition, to detect compartmental- with the information-release process. A. F.
ization by small petroleum density differences, to verify quality of Veneruso kindly provided unpublished updates
samples for PVT (pressure, volume, temperature) analysis, and esti- to his pressure-gauge response model.
mate salinity or temperature of unsampled water zones. Reference to any tool or gauge model or
Data quality limits barrier and fluid-contact resolution; thus, manufacturer is not an endorsement or
quality control is essential. Pressure measurement errors on the recommendation for that product.
3-kPa (0.5-psi) scale can be detected from behavior of the buildup
pressure. Tests having the potential for small amounts of super-
charge are identified from the overbalance and formation mobility.
Examples illustrate identification of free-water levels and fluid con-
tacts, fluid identification, supercharge identification, and water-zone
compartmentalization.

INTRODUCTION

Pressure-depth plots have been used for the last quarter century
to evaluate fluid density, fluid contacts, and pressure compart-

Copyright #2003. The American Association of Petroleum Geologists. All rights reserved.
Manuscript received August 16, 2001; provisional acceptance March 22, 2002; revised manuscript
received July 8, 2002; final acceptance August 22, 2002.
DOI:10.1306/08220201022

AAPG Bulletin, v. 87, no. 2 (February 2003), pp. 295 – 311 295
mentalization from wireline pressure surveys (Pelissier- is normally so small that the pressure stabilizes within
Combescure et al., 1979). Over the last 10 years or so, a a few minutes. In good tests, pressure stabilizes at the
new generation of temperature-compensated quartz formation pressure and the pretest ends. The mud pres-
pressure gauges have increased within-well, wireline sure at the test depth is recorded prior to setting the
pressure test resolution and repeatability to about 1 kPa probe and after withdrawal of the probe. These are re-
(0.2 psi; Veneruso et al., 1991). In many wells, total ported as hydrostatic or mud pressures. The other re-
pressure range of a wireline pressure survey is so large ported pretest result is the drawdown mobility (formation
that pressure-depth plots cannot take advantage of the permeability/filtrate viscosity). It is calculated from the
high resolution of modern pressure gauges. pressure drop during drawdown.
This article uses a new interpretation technique The most commonly used wireline pressure–
based on the concept of excess pressure. Data are trans- interpretation technique is the pressure-depth diagram,
formed to remove the effects of the weight of the static a plot of stabilized formation pressure against true ver-
fluid; thereby, small pressure differences can be visu- tical depth (Figure 1). If the total pressure variation is
alized. This technique enhances the measurement of large, pressure-depth diagrams do not have resolution
fluid densities and resolves small density changes and sufficient to take advantage of the resolution of mod-
pressure barriers that are not likely to be recognized on ern wireline pressure gauges. For example, the pres-
standard pressure-depth plots. Poorly documented phe- sure data in Figure 1 appear to be of quite high quality
nomena can also be detected, such as effects of capillary- (low scatter), but the fluid contact is hard to identify,
displacement pressures near fluid contacts. The high even where contact elevation is identified. Water and
resolution also allows new applications for wireline oil in this example have a relatively small density dif-
pressure data. This technique was briefly described ference, and thus, the pressure-depth trends of the two
on an earlier poster (Brown and Loucks, 2000). This fluids are nearly parallel. One way to visualize small
article presents the concept in more detail using examples density differences is to expand the pressure scale. The
to illustrate its application. Wireline pressure data col- slope difference is greater, but the contact may still
lected after production indicates differential depletion;
thus, interpretation techniques are different from those
presented here. Pressure (MPa)
High-resolution analysis requires tighter quality con-
33.8 34 34.2 34.4 34. 6
trol, because small pressure-measurement errors can –3040
20 psi
greatly reduce interpretation strength. Established quality-
control techniques (e.g., Dewan, 1983) are adapted to re-
solve more subtle test problems. Supercharged tests (tests oil oil-water contact
having anomalously high reservoir pressures) can be iden- –3060
tified by new simplified relationships to overbalance,
free-water level
filter-cake properties, and formation permeability.
Subsea depth (m)

–3080
PRESSURE ANALYSIS METHODOLOGY
water
Dewan (1983) and other wireline-log-analysis textbooks
present basic wireline pressure collection, quality con- –3100
trol, and interpretation methods. The wireline pres-
sures discussed in this article are ‘‘pretest’’ pressures;
that is, the static formation pressures are collected be-
fore wireline sampling. Data are collected in the fol- –3120
lowing manner (Pelissier-Combescure et al., 1979). The Figure 1. Conventional pressure-depth plot for the Villano
tool probe is pressed through the filter cake to the oil accumulation, Ecuador. The diagonal line fits the water
borehole wall. A small volume of fluid is withdrawn pressures from the lower part of the survey. Data in the upper
from the formation, and thus, the pressure drops (draw- part of the section deviate from the line owing to the presence
down). Pressure then builds as fluids in the formation of oil. Horizontal lines show elevations of the free-water level
flow toward the borehole (buildup). Drawdown volume and oil-water contact.

296 Improved Interpretation of Wireline Pressure Data

 (A) Pressure (B) Excess pressure surement (Figure 2A). The quantitative form of this
relationship is the following (Hubbert, 1956):
pressure
analyses excess pressure ¼ rgz þ Pm
excess pressure ¼ 0:4335rz þ Pm
ðft; g=cm3 ; and psiÞ
Depth

excess pressure ¼ 9:8067E  6rz þ Pm

Depth
3
excess ðm; kg=m ; and MPaÞ ð1Þ
pressure

where P m is the measured pressure at depth z relative

to the datum (negative downward), U is the density of
the fluid at reservoir conditions, and g is the pressure
assumed fluid-pressure gradient for fluid having a density of 1 g/cm3. Excess
trend
pressure can be calculated using any datum. The mag-
nitude of the excess pressure has less meaning than
excess-pressure differences calculated using the same
Figure 2. Excess-pressure concept. (A) Pressure-depth diagram, datum and fluid density. Excess pressure is easiest to
showing pressure analyses (dots), pressure trend assumed for interpret if the chosen fluid density is the dominant
excess-pressure calculation (diagonal line), and excess pressure reservoir fluid density. A single static fluid having con-
(difference between expected and measured pressure, horizontal stant density and free communication with itself (no
lines). (B) Excess-pressure –depth diagram. The vertical data
barriers) has the same excess pressure at all elevations
trend in the lower part of the survey indicates that data match
if density is chosen correctly (Figure 2B). Excess pressure
the assumed fluid density. Excess pressures for the shallower
points (horizontal lines) increase up section, indicative of a fluid is constant because fluid potential is uniform (Hubbert,
having lower density. 1956).
Excess-pressure plots are constructed by identify-
ing the density that equalizes excess pressure of the
be difficult to recognize. In addition, scale expansion fluid of interest at all depths. Start by choosing a depth
increases the size of the diagram, and large diagrams are interval in the pressure survey that has a single fluid
cumbersome. and no potential sealing lithologies. Excess pressures
are calculated and plotted against depth using an arbi-
Excess-Pressure Definition and Construction of trary fluid density. If the excess-pressure-vs.-depth trend
Excess-Pressure Plots is rotated clockwise from vertical, the chosen density
is too high and a lower density value is substituted.
Much of the pressure variations in pressure-depth plots The assumed density is iterated until excess-pressure
are caused by the weight of the fluids themselves. By variance is minimized and the excess-pressure trend is
removing effects of the weight of one of the fluids on vertical.
pressure, small pressure differences caused by density Data in Figure 1 are used as an example. The water-
variations and pressure barriers can be enhanced. This saturated zone below the shale was chosen for analysis
approach is referred to as the ‘‘excess-pressure’’ method and 1 g/cm3 density was assumed. This excess-pressure
(Brown and Loucks, 2000). Excess-pressure estimation trend slopes clockwise from vertical and the density
is a common technique used elsewhere to analyze basin- is slowly reduced until the excess-pressure trend is
scale water flow and geopressure development (e.g., over- vertical at 0.966 g/cm3 (Figure 3). Tilt to the excess-
pressure of Mann and Mackenzie, 1990). In hydrologic pressure slope is evident with only 0.006-g/cm3 change
applications, freshwater or native-water density is used in assumed water density (Figure 4). Excess pressure
for excess-pressure calculation. For wireline pressure in the water column ranges from 5023 to 5026 kPa
analysis, the density of any fluid in the reservoir is used. (728 to 728.5 psi), a range of 3 kPa (0.5 psi). In com-
Excess pressure is calculated from an assumed fluid parison, formation pressure over the same interval
density, gauge depth, and measured pressure. Excess ranges from about 34,160 to 34,560 kPa (4954.5 to
pressure is the difference between the measured pressure 5012.5 psi), a range of 400 kPa (58 psi). Pressure bar-
and the pressure expected from the weight of a fluid rier and fluid contact are evident above the analyzed
between the datum and the depth of pressure mea- interval, whereas these features are difficult to recognize

Brown 297
 Excess pressure (kPa) Excess pressure (kPa)
5020 5030 5040 5050 6705 6710 6715
–3040 −3040
0.5 psi oil gradient
oil
Subsea depth (m)

–3060
free-water level
−3050

Subsea depth (m)

–3080 shale bed
(pressure barrier)
oil mobile
−3060
OWC oil immobile
–3100 water

1 psi
–3120 FWL
water gradient
Figure 3. Excess pressure vs. depth plotted for the Villano −3070
field data shown in Figure 1. A density of 0.966 g/cm3 was
used to calculate excess pressure. Three trends are evident: a Figure 5. Excess pressure calculated using oil density (0.91
vertical trend below the shale bed (main aquifer), a short g/cm3), using data from Figure 1 shallower than  3072 m.
vertical trend above the shale bed (aquifer having slightly Intersection of oil and water trends is the free-water level
higher pressure), and a diagonal trend (oil column). The (FWL), the elevation where capillary pressure is zero. Oil-water
shaded zone is a shale bed that acts as a barrier between contact (OWC) elevation lies between the highest data on the
the aquifers with different pressure. The rest of the reservoir water trend and lowest data on the oil trend. The free-water
is sandstone. level is lower than the oil-water contact due to water-wet
conditions in the reservoir.

on Figure 1. The shallower data can be analyzed by using

excess pressure calculated from the oil density of 0.91 g/
cm3 (Figure 5).
Excess pressure (kPa)
4840 4850 4860 5210 5220 5030 5040
−3040 Interpretation Using Excess Pressure
(A) ( B) (C)
Fluid density, fluid contacts, and pressure barriers can be
−3060 interpreted from excess-pressure plots. Fluid density is
Subsea depth (m)

shale bed estimated by rotating the excess-pressure trend to vertical,

as discussed previously. Selection of fluid density is an
iterative process; thereby, barriers and slope changes can
−3080 0.972 g/cm3 0.960 g/cm3 0.966 g/cm 3
be detected during the density-estimation process. If a
possible barrier or contact is identified, the depth range
of analyzed samples is narrowed so that only a single
−3100
fluid is evaluated. In contrast, fluid density is calculated
from pressure-depth plots by regression. Pressure-barrier
or small density changes may not be noticed before
−3120 regression; thus, the density calculated from the trend
Figure 4. Sensitivity of excess-pressure density estimation. may not represent the actual fluid density.
Excess-pressure-vs.-depth plot for assumed fluid densities of Slope change indicates fluid-density change. Fluid-
(A) 0.972, (B) 0.960, and (C) 0.966 g/cm3. Fluid-density change density changes at fluid contacts and across petroleum
of 0.006 g/cm3 is evident in the lower water-bearing sandstone seals (Figure 6). On excess-pressure plots, clockwise tilt
from the excess-pressure slope. Same data are shown in from vertical indicates a density that is lower than mod-
Figures 1 and 3. Shaded zone is the shale bed. eled. Expanding the scale increases the excess-pressure

298 Improved Interpretation of Wireline Pressure Data

 (A) (B)
Pressure Pressure
Water with lower density

seal seal

gas
fluid contacts fluid contacts
GOC GOC

Depth
Depth

oil
pressure barrier pressure barrier
oil
OWC OWC

FWL FWL

water

Figure 6. Identification of fluid contacts and pressure barriers using pressure plots. (A) Pressure-depth plot showing characteristics of
free-water level (FWL), oil-water contact (OWC), gas-oil contact (GOC), a pressure barrier, and a seal. (B) Excess-pressure plot calculated
for oil density showing excess-pressure characteristics of free-water level (FWL), oil-water contact (OWC), gas-oil contact (GOC), a
pressure barrier, and a seal. Gas trend is rotated clockwise from vertical (lower density), whereas water trend is rotated counter-
clockwise (density greater than oil). Water above seal has lower density than water below the OWC; thus, its slope is different. Excess-
pressure scale is expanded by a factor of about seven relative to the pressure scale; thus, contacts and barriers are more obvious.

slope of fluids having a density different from modeled the contact estimated from wireline pressure data. If
density, but vertical excess-pressure trends do not change porosity-resistivity logging indicates a deeper petroleum
as the scale expands. Scale can be expanded as much as contact than estimated from wireline pressure data,
needed to detect small density changes. Once a different the petroleum-water contact has probably moved up-
slope indicates a different fluid, fluid density can be cal- ward since trapping. The deeper petroleum is residual,
culated iteratively, just like the density of the first fluid and the permeability-saturation relationship may fall
(Figure 5). In contrast, all trends are tilted on pressure- on the imbibition curve higher in the reservoir.
depth plots, and thus, expanding the pressure axis changes Abrupt offsets of pressure-depth trends indicate
the slopes of both lines. Even after expanding the pres- pressure seals. Pressure seals plot as offsets between
sure scale, minor slope changes may not be recognized. tilted trends on pressure-depth diagrams (Figure 6A).
Pressure-depth plots of most data lack sufficient res- These offsets may not be recognized where the magni-
olution to differentiate between free-water level (elevation tude of the offset is small compared to the total pressure
where capillary pressure is zero) and petroleum-water change across the barrier. Excess-pressure plots remove
contact (elevation with lowest moveable petroleum), most of the total pressure change across the barrier, and
but these surfaces can be distinguished using excess- thus, excess-pressure scale can be expanded to visualize
pressure plots. Intersection of the petroleum and water the small excess-pressure difference (Figure 6B). If fluid-
trends is the free-water level, because at this eleva- density changes across a pressure barrier (such as the
tion, the petroleum and water pressures are the same. top seal), the excess-pressure slope as well as the mag-
Petroleum-water contact occurs at or below the lowest nitude of the excess pressure differs.
test that lies on the petroleum-density trend. The dif-
ference in petroleum-water contact elevation and free-
water level indicates wetting conditions in the reservoir DATA-QUALITY CONTROL
(Desbrandes and Gualdron, 1987). Reservoir-saturation
history can be evaluated by comparing petroleum-water The standard deviation of water-leg excess-pressure data
contact estimated from porosity-resistivity logging to in Figure 3 is 0.65 kPa (0.09 psi). This is comparable

Brown 299
to the within-well reproducibility of the temperature- Strain-
Full-scale gauge CQG
compensated quartz-gauge response (Veneruso et al., pressure pressure
pressure
1991). Many data sets collected under good logging con-
ditions show low scatter, but some surveys show sig- Mud
480 pressure
nificantly larger pressure scatter. For example, Fraisse
et al. (1987) report that 32% of repeated quartz-gauge
formation pressures have a pressure difference of 50 End
kPa (7 psi) or greater. The quality of the interpretation buildup
is only as good as the quality of the data, and thus, data- 360
quality evaluation becomes essential.
Pressure-measurement problems, supercharging,

Time (s)
or depth errors may cause bad data. In most cases, bad
data cannot be corrected. Thus, the best strategy is the
identification of bad or suspect data and its elimina- Buildup
240
tion from the data set. The data normally supplied to
the geologist is a table of summary pretest formation
pressures, their depths, hydrostatic pressures, and draw-
down mobilities (formation permeability/fluid viscos-
Drawdown
ity). Tests with suspect pressures may also be identified.
120
These data are insufficient to detect subtle data prob-
lems. Quality must be assessed from the transient pres-
sure data and other data available on the pressure-test Probe set
logs. Mud
pressure
0
Pressure-Measurement Errors
0 10 10 psi 1 psi
thousand psi
Pressure-measurement problems have been recognized Figure 7. Wireline pretest pressure variation for a good
since the introduction of multiple-testing tools (e.g., test. Time increases upward. The left track shows events and
Dewan, 1983). Traditional criteria identify data with total pressure variation, with pressure increasing toward the
tens to hundreds of psi errors. These buildup criteria right. Shading shows the period of drawdown. The center
have been modified to detect problems in the psi range and right tracks show high-resolution pressure variation dur-
ing the latter part of the buildup for the strain-gauge and
desired for high-resolution pressure analysis.
temperature-compensated quartz gauge (CQG), respectively.
The pressure buildup during a good test is smooth,
The quartz-gauge pressure track is only 1 psi wide; thus, rising
with the rate of pressure increase decreasing with time pressure wraps to the left side of the track after leaving the
(Figure 7). The pressure appears stabilized at the end right edge. Quartz-gauge resolution is indicated by waviness
of a good test; thus, the final pressure is very close to during the latter part of the buildup. The test was terminated
the formation pressure. Random pressure fluctuations after quartz-gauge pressure stabilized. The strain-gauge pres-
during the latter part of the buildup are very small (<0.2 sure appears to stabilize sooner because the pressure range
kPa [<0.03 psi]; Figure 7). is larger. These data were collected using the Schlumberger
Pressure builds slowly in low-permeability rocks. MDT tool.
Where reservoir permeability is very low, tight tests are
identified and the tests are aborted. Low-permeability
tests that approach static pressure are sometimes ter- where supercharging is likely. Early test termination
minated prematurely to save rig time and prevent tool offsets some of the supercharging effects.
sticking. Final pressures of these tests are not stabilized. Where the probe is completely plugged or the seal
Static pressure can be determined by extrapolating pres- is completely lost, major pressure differences from res-
sure data using a Horner plot or spherical-flow plot (see ervoir pressure are quickly noted and the test is aborted
Dewan, 1983, for methods). Extrapolated pressures of and noted on the summary table. In some tests, probe-
incomplete tests should be used with caution. Most tests seal leakage and probe plugging are minor and the test
with incomplete buildup occur in low-permeability rock is completed. Leakage and periodic plugging may occur

300 Improved Interpretation of Wireline Pressure Data

during the entire buildup period or during the early or
Strain-
late parts of the buildup. Pressure spikes or drops on the gauge CQG
Full-scale
buildup curve identify subtle seal leakage (Figure 8). In pressure pressure pressure
some tests, the rate of pressure increase increases with 360
time, opposite from normal test behavior (Figure 8). A End Static pressure
plugged probe may be interpreted as equilibrated pres- buildup
sure, because pressure buildup stops. Late-test plugging
causes an abrupt flattening of the pressure plotted against
Horner- or spherically normalized time. Most tests with 240

Time (s)
plugging or leaking during buildup must be discarded.
Stabilized pressure can be interpreted from tests where Buildup
probe plugging or seal leakage only affects the last part of
buildup. Horner- or spherically normalized time plots Anomalous
from the earlier parts of the buildup project to the sta- 120 pressure
bilized pressure. Likewise, tests with good probe sealing
during the later part of the buildup may indicate static
reservoir pressure even where earlier parts of the test
Drawdown
were affected by probe-seal leakage.
Probe set
0
0 10 10 psi 1 psi
Strain -
Full-scale gauge CQG thousand psi
pressure pressure pressure
Figure 9. Wireline pretest pressure variation having anom-
480
alous high pressure in the early part of buildup. Tracks and
scales are the same as those in Figure 7. Quartz-gauge pres-
sure builds to a value exceeding the static pressure before
stabilizing 5 min into the test. This effect causes up to about
2-kPa (0.3-psi) error. If data are extrapolated downward to
360 a false equilibrium pressure using pressure-transient tech-
niques, reservoir pressure may be underestimated by up
Buildup to 1 psi. These data were collected using the Schlumberger
Pressure
MDT tool.
Time (s)

drop

240
In some settings, the temperature-compensated
quartz gauge shows an anomalous pressure response.
Pressure Pressure rises above the formation pressure during the
spike
middle part of the buildup and then asymptotically
120
decreases with time (Figure 9). The cause of this phe-
Drawdown nomenon is not clear. If the test is terminated too soon,
the pressure decrease is not recorded and the reported
pressure is slightly high (0.7– 2 kPa [0.1–0.3 psi]) to
Probe set
0 the static pressure. Where the pressure decline is re-
0 10 10 psi 1 psi corded, interpreters may extrapolate reservoir pres-
thousand psi sure from the part of the buildup curve where pressure
is falling. This extrapolation is not theoretically jus-
Figure 8. Wireline pretest pressure variation for a test having
a small amount of seal leakage. Tracks and scales are the same tified. It may lead to extrapolated equilibrium pres-
as those in Figure 7. Seal leakage is indicated by small ( <1 kPa sure as much as 7 kPa (1 psi) lower than actual static
[< 0.2 psi]) pressure spikes and drops on quartz-gauge pres- pressure. If this effect is observed during data collec-
sure and by an increasing rate of pressure buildup late in the tion, the best approach is to allow longer buildup
test. Test termination is not shown on this display. These data periods to allow the gauge to stabilize at static for-
were collected using the Schlumberger MDT tool. mation pressure.

Brown 301
Depth Errors (A)
A depth error of 0.3 m will result in approximately
Filter cake
3 kPa (0.4 psi) excess-pressure error in water-bearing
sections; thus, depth errors decrease excess-pressure data
Borehole formation
quality. Depths must be adjusted to true vertical depth
for proper analysis. If the depth datum is adjusted during
the pressure logging run, pressure tests before and after
depth adjustment should be compared to see if there is
a systematic pressure difference caused by the depth
adjustment. Pulling stuck tools is likely to stretch the
cable, and logging runs with tool sticking may have higher
scatter than other data. (B)
Within-well depth errors are difficult to detect or
Mud pressure
correct. Theoretically, the mud pressure can be used
to correct the depth, but this has not proved useful Pressure drop
unless depth errors are great. Hydrostatic (mud) pres- across filter cake

Pressure
sure measurements are rarely allowed to stabilize be- Pressure drop in
fore or after the pretest; thus, reported before- and formation
Supercharge
after-test hydrostatic pressures may differ by as much Static
formation
as 10 kPa (1.5 psi). Mud density changes during log- pressure
ging as mud changes temperature; thus, a slight drift Borehole Formation
Radial distance
to the mud pressure at a fixed depth is present. Mud
pressure also changes as mud level in the borehole varies Figure 10. Supercharge development. (A) Supercharge re-
while logging. sults from radial filtrate flow from the borehole (left) into
the formation (right) through the permeable filter cake. Pres-
sure near the borehole must exceed static formation pres-
Supercharging sure to accommodate flow. (B) Pressure profile across the
borehole, filter cake, and formation. Where the filter cake has
Supercharging results from leakage of mud filtrate high permeability relative to the formation, pressure next
to the borehole is significantly higher than static forma-
through the filter cake (Figure 10). All filter cakes
tion pressure. Pressure is measured at the borehole wall;
that developed from water-based muds are perme-
thus, the static test pressure is higher than the static forma-
able; thus, filtrate from overbalanced mud leaks into tion pressure. This is the supercharge. Supercharge increases
the formation. If the filter cake has high permeability with increasing mud overbalance and increasing filter-cake
or if the formation has low permeability, leakage into permeability.
the formation is faster than dispersion into the for-
mation. Pressure rises above the formation pressure
near the borehole wall. The probe measures pressure pressure difference. Numerical and analytical solutions
at the borehole wall; thus, tests have high pressures to supercharging provide a basis for predicting super-
unrepresentative of the formation. All wireline pres- charged tests (Pelissier-Combescure et al., 1979; Ste-
sure tests in water-based muds are supercharged because wart and Wittmann, 1979; Phelps et al., 1984; Waid
filtration through the filter cake always occurs. Under et al., 1992). These models are rarely used with field
good logging conditions, supercharging is too small to data.
measure. Instead of these complex models, a simple model
Where supercharging is hundreds to thousands is used to approximate conditions where supercharg-
of kilopascals in excess of formation pressure, super- ing is significant. During filtrate loss, water flows radially
charging can be identified solely on the basis of its high through two concentric zones of differing permeability:
pressure (e.g., Pelissier-Combescure et al., 1979). Super- the filter cake and the formation. In each radial zone,
charging on the 1 –10-kPa (0.1 – 1.5-psi) scale cannot the pressure drop can be calculated as a function of the
be reliably identified by higher pressure because den- permeability and the radial distance, assuming the Dupuit
sity changes or compartmentalization may cause this assumptions of steady, incompressible, radial flow into

302 Improved Interpretation of Wireline Pressure Data

an infinite confined reservoir (modified from Viessman from a borehole that has good mud properties, the
et al., 1977): radius of influence is probably in the order of 3 m
(10 ft) or so, unless the well has serious fluid-loss prob-
Qm r lems or a highly compressible fluid. Doubling this

p ¼ Pi  P ¼ ln ð2Þ
2pgrk ri radius will increase the calculated supercharge by about
30%. An uncertain radius of influence and a poorly con-
where
p is the pressure difference measured at radius strained filter-cake permeability limit the supercharge
r from the reference pressure P i measured at radius r i, approximation to order-of-magnitude estimates. Pre-
Q is the fluid flux through the borehole wall per unit diction quality is insufficient to correct supercharged
depth, k is the permeability, g is the gravitational ac- tests, but it is sufficient to predict settings where super-
celeration, U is the fluid density, and A is the viscosity. charge may be significant.
Equation 2 can be used to describe the flow through If average drilling conditions are assumed, super-
both the filter cake and the formation near the borehole. charge can be estimated as a function of formation and
The supercharge-plus-pressure drop across the filter cake filter-cake permeability (Figure 11). For example, a su-
is the overbalance; that is, the mud pressure in excess of percharge of about 3 kPa per 1000 kPa mud overbal-
the formation pressure. The ratio of supercharge to over- ance (3 psi/1000 psi overbalance) can be expected in a
balance can be determined by dividing the pressure drop reservoir having a permeability of 1 md under typical
in the formation by the total pressure drop (over- drilling conditions (30 cm borehole diameter, filter
balance). By taking the ratio, Q, A, g, 2, and U cancel: cake 1.2 cm thick having permeability of 0.1 Ad).
  If supercharging is relatively small compared to over-
supercharge kfc ln rrbr balance, the first term in the denominator of equation 3
¼     ð3Þ
overbalance k ln rr þ k ln rb
fc rb f rb xmc

where permeability (k) subscripts fc and f refer to filter

cake and formation, respectively, and radius (r) sub-
scripts b and r refer to borehole and radius of influence,
respectively. The filter-cake thickness is X fc. The flow
model assumes that filtration has reached a steady flow,
and thus, filtration history is ignored (except for calcu-
lation of filter-cake properties). The variables in equa-
tion 3 can be evaluated from well data. Borehole radius
is determined from caliper logs. Filter-cake properties
can be calculated from high-temperature, high-pressure
mud-filtration test, mud-solids test, and time since the
last wiper trip, all of which are in daily drilling and
mud reports. The filter-cake properties are calculated
by assuming static filtration since the last wiper trip.
Formulas for calculating the filter-cake properties from
test data are described in textbooks on drilling fluids
(e.g., Gray and Darley, 1980). Formation permeability
is estimated from the mobility (permeability/viscos-
ity) measured by wireline pressure tests.
Because equations 2 and 3 describe incompres-
sible, steady flow, the radial distance to undisturbed
Figure 11. Supercharge modeled as a function of rock per-
formation pressure is infinite and the natural log is
meability and overbalance using equation 3. Vertical axis is
therefore not definable. Real reservoirs and fluids are supercharge normalized to overbalance (supercharge in kPa/
compressible, and the approximation represented by 1000 kPa is equal to psi/1000 psi). Assumed borehole con-
the Dupuit assumption can be solved with a radius ditions are indicated on the figure. The range of the modeled
equal to the distance with undisturbed pressure (radius filter-cake permeability is that expected in water-based mud
of influence). For the small volumes of filtrate leaking with modern mud systems.

Brown 303
 10000
1000 tool type and service company are the same. Absolute-
gauge accuracy is about 14 kPa (2 psi) for typical
operating depths and pressures (Joseph et al., 1992).
1000 Most examples of between-well error are less than
100
this, near 10 kPa (1.5 psi; Figure 13). I have also seen
Supercharge (kPa)

Supercharge (psi)
cases where static excess pressures differ by as much
as 35 kPa (5 psi) between nearby wells where the
100
10 only reasonable cause for pressure difference is gauge-
calibration error. Error this high has not been observed

10
1

Excess pressure
1 10 kPa
0.01 0.1 1 10 100 1000
Horner slope (MPa)/log(normalized time) 1 psi

Figure 12. Plot of logarithm of supercharge (static formation

pressure minus pressure expected at test depth) vs. logarithm
of the inverse of buildup permeability, as indicated by the Hor-
ner slope. Horner slope is the regression of pressure against
the log of Horner-normalized time. A trend having a loga-
True vertical depth
rithmic slope near 1 indicates that data are consistent with
supercharge origin. The slope flattens at high Horner slopes
because the maximum supercharge is the overbalance. Data
are from Temane area, Mozambique.
100 m

can be dropped. Supercharge becomes linearly propor-

tional to the inverse of formation permeability where
overbalance, filter-cake properties, and radius of in-
fluence are similar. This leads to the familiar linear rela-
tionship between supercharge and inverse of formation
mobility (e.g., Pelissier-Combescure et al., 1979). The Well
1
correlation is extended to low supercharge by plotting 2
both axes on logarithmic scales (Figure 12). If the data 3
form a trend with a logarithmic slope near  1, then the 4
5
pressures are consistent with supercharge. Buildup mea-
6
sures low-formation mobility better than drawdown; 7
thus, the inverse of the Horner or spherical slope against
pressure can be used instead of drawdown mobility,
but the correlation slope should lie near 1. The highest Figure 13. Example between-well absolute pressure accuracy
supercharged tests are those with the lowest forma- indicated by excess pressure vs. depth. Perfect accuracy and
tion mobility; thus, tests with the most severe super- reservoir communication would plot as a single vertical trend.
charge are likely to have incomplete buildup. Tests should The data show both within-well and between-well differences,
having a total variation in the order of 12 kPa (1.8 psi). Varia-
be extrapolated to static pressures before analyzing for
tion includes effects of both depth uncertainty and pressure-
supercharging.
gauge absolute accuracy. Higher data scatter in some wells
indicates poorer logging conditions and tool stability. Data were
collected from a large anticlinal gas accumulation having a high
BETWEEN-WELL PRESSURE DIFFERENCES permeability, sheet-sandstone reservoir, and excellent lateral
communication. Wells were located up to tens of kilometers
Absolute-gauge and depth accuracy limits the inter- apart. Pressures were collected using different Schlumberger
pretation of data from multiple wells, even where the MDT tools.

304 Improved Interpretation of Wireline Pressure Data

using the latest generation of tools. Occasionally, there to the excess-pressure error than the gauge error. Typical
is an excess-pressure difference close to 100 kPa (14.5 absolute-depth error in vertical wells is in the order of
psi) between runs using different tools in the same well 0.03% where cable stretch has been corrected for tension,
or nearby wells with good pressure communication. temperature, and pressure (Sollie and Rodgers, 1994).
This pressure difference is probably caused by absolute Excess-pressure uncertainty caused by depth uncer-
pressure mistakenly reported as gauge pressure. In my tainty at 3 km depth in vertical wells is at best near
experience, most examples of different runs using the 10 kPa (1.5 psi), similar to the 14-kPa (2 psi) absolute-
same tool in the same well have between-test repro- gauge error for temperature-compensated quartz gauges
ducibility similar to that within each run. This is not reported by Veneruso et al. (1991). Highly deviated wells
true where the pressure tool has been fished or roughly will have a greater depth error (Wolff and deWardt,
handled. 1981; Brooks and Wilson, 1996).
Pressure differences caused by errors in absolute- Between-well excess-pressure differences can be
depth accuracy between wells may cause apparent pres- corrected by adjusting absolute pressure, absolute depth,
sure differences between wells. In some wells, especially or both. The effect of correcting a depth error is dif-
deviated wells, absolute-depth error may contribute more ferent from that of correcting a gauge error. This is
best illustrated with an example (Figure 14). A gas
accumulation over water is penetrated and tested by
two wells, and well 2 has either a depth or pressure
Pressure error. If the depth of well 2 is shifted to align the water
trends, the free-water level of well 2 becomes lower
Pressure than well 1. If the pressure of well 2 is shifted to align
correction
the water trends, the free-water level in well 2 is
pressure shift
Well 2 after

higher than well 1. If the free-water level elevation

is assumed to be the same, then both the depth and
Well 2
Well 1

pressure are adjusted (Rodgers, 1998). If the same free-

Well 2 FWL
water level is assumed, then the data cannot test com-
Well 2 FWL after
pressure shift munication or tilting; thus, the between-well pressure
Depth

comparison does not provide any new information. New

depth surveys may resolve ambiguity if they reduce
Well 1 FWL depth uncertainty. In addition, if either the depth or
pressure correction necessary to align the data exceeds
reported accuracy, then that correction is probably not
Well 2 FWL after reasonable.
correction
Depth

depth shift
W pth
de
ell sh
2 a ift

EXAMPLES
fte
r

Villano Field, Oriente Basin, Ecuador

Figure 14. Ambiguity of correcting between-well pressure- Villano field is a heavy oil accumulation in the south-
trend differences. Wells show different pressure trends caused western Oriente basin, Ecuador. The Albian Lower Hol-
by between-well pressure gauge or depth error (solid lines). If lin Sandstone reservoir has a few shale beds high in
water is hydrostatic, the two water-leg pressure trends should the section, but most of the unit is high-permeability,
be the same. There are three ways to align the water trends:
moderate-porosity, fluvial sandstone. The trap is a fault-
pressures can be adjusted, depth can be adjusted, or both
bend fold anticline with no major faulting within the
depth and pressure can be adjusted to align the free-water
levels (FWL). If well 2 pressure is shifted, the FWL of well 2 lies main part of the reservoir. The oil and water densities
above that of well 1 (dashed line). If well 2 depth is shifted, the are similar; thus, the free-water level is barely evident
FWL of well 2 lies below that of well 1 (long-short dash). With- on the pressure-depth plot (Figure 1). The slope inflec-
out independent evidence for the cause of the pressure-trend tion on the excess-pressure diagram constructed using
difference, any correction is ambiguous. water density (0.966 g/cm3) readily identifies the free-

Brown 305
water level (Figure 3). Water-leg excess-pressure trends Despite excellent borehole environmental condi-
are offset by about 1 kPa (0.2 psi) across a minor shale tions, some pressures are quite anomalous (Figure 15).
layer. This shale acts as a barrier separating water zones Upper parts of the sandstone reservoir have a gradient
with slightly different pressures. This barrier is below indicating gas density near 0.1 g/cm3, a value expected
the oil-water contact, but it may affect water coning at these depths and pressures. Deeper in the sand-
during field production. stone, the pressure increases rapidly with depth as if
Oil excess-pressure data (calculated with 0.91 g/cm3) the fluid was becoming denser. This appears at first
seem to have considerable scatter, but this is caused glance to be a fluid contact; however, the density of the
by the extreme magnification of the excess-pressure deeper fluid would have to be approximately 5 g/cm3
scale (Figure 5). Oil excess pressure ranges from about to account for the slope of the deeper interval. The
6709 to 6714 kPa (972.3 to 973 psi), a range of 5 kPa pressure data were analyzed to determine the cause of
(0.7 psi). No pressure barriers can be identified in the this effect and the true gas-water contact elevation.
oil column. The free-water level is the intersection of Upon inspection of the buildup curves, it became
the vertical oil pressure trend and the water trend. Its apparent that most of the pressure tests in the lower
elevation is  3068.6 m subsea, as determined by part of the sandstone were incompletely built up. The
algebraic intersection of the excess-pressure trends of first step was extrapolation to static pressure. This causes
Figures 3 and 5. Some tests above the free-water level the pressure trend to steepen even more (Figure 15).
fall on the water excess-pressure trend. No physical Because the pressures appeared too high, the possibility
barrier separating the oil tests from the water tests of supercharging was investigated. Buildup mobilities
is present; thus, this distribution is interpreted as a
capillary-threshold effect. Oil pressures are not
measured deeper in the borehole because oil is im-
mobile below this depth and the gauge measures only Pressure (MPa)
mobile fluid pressure. The oil-water contact lies be- 13.3 13.4 13.5 13.6 13.7 13.8 13.9
1304
tween the deepest pressure test falling on the oil trend shale
and the shallowest pressure test falling on the water
trend. The oil-water contact depth derived from the
wireline pressure data corresponds to an abrupt upward Gradient due to
increase in oil saturation determined by wireline log 1305 density (g/cm3)
analysis.
Depth (m)

The oil-water pressure difference measured at the

oil-water contact corresponds to the native oil-brine
1306 01 5
capillary displacement pressure at reservoir conditions,
using the concept of Desbrandes and Gualdron (1987).
If so, the capillary entry pressure of Hollin sandstone
at this depth is about 5 kPa (0.7 psi). Typical mercury
capillary displacement pressures of Hollin sandstones 1307
are on the order of 56 kPa; thus, calculated product of
oil surface tension and wettability is about 0.032 N/m shale
(32 dyn/cm). This is a fairly high product, indicating Uncorrected
Extrapolated
that conditions in the field are strongly water wet. 1308

Temane Area, Mozambique Figure 15. Pressure-depth plot showing an apparent in-
creasing density in the lower part of the pressure survey from a
well in the Tamane area in Mozambique. Uncorrected data
The Temane gas discoveries in Mozambique occur as
(crosses) indicate density changing from a gas gradient (0.1
separate gas pools in thin, coarsening-upward, Creta- g/cm3) down section to a gradient exceeding 5 g/cm3. Most of
ceous sandstones. The traps are simple dip-closures at the lower tests were incompletely built up; thus, static pressures
relatively shallow depth. The low gas density causes gas were extrapolated (diamonds). Extrapolation-corrected data in-
pressures to plot as an almost vertical trend with depth; dicate even greater density change. The reservoir is coarsening-
thus, there is little advantage to using excess pressure. upward sandstone. Shading indicates bounding shales. No shale
This example illustrates the detection of supercharging. barriers are present in this reservoir.

306 Improved Interpretation of Wireline Pressure Data

 10000
if possible. Wireline pressures were measured, but in-
terpretations of the raw data were ambiguous due to
bad tests, multiple thin-reservoir zones, and multiple
1000
model fluids. Bad data are caused by seal leakage or a tight,
supercharged reservoir (Figure 17A). The permeabil-
ity threshold for supercharging was estimated from
100 well data, and tests with permeability below the thresh-
old were edited out. Tests with leaky probe seals were
Supercharge (kPa)

identified from buildup characteristics. If leaks devel-

10
oped late in the test, earlier test data were extrapolated
to static pressure; otherwise, leaky tests were omitted.
Average correction of acceptable data is about 4 kPa
(0.6 psi; Figure 17B). The edited data showed much less
1 scatter (Figure 17A). The uppermost tested zone had too
few reliable tests to estimate fluid density. Petroleum
zones in the edited data were identified by clockwise
0.1 rotation of excess-pressure trends from vertical (zones 1
0.0001 0.001 0.01 0.1 1 10 100 1000 and 3, Figure 17C).
Formation permeability (md)
Figure 16. Comparison of modeled supercharge against for-
mation permeability (line) having observed supercharge (points), (A) (B) (C)
Pressure (MPa) Pressure correction (kPa) Excess pressure (MPa)
assuming the section is entirely gas saturated. Close fit of the −× 650
60 65 70 75 –3 0 5 10 27.8 28.3 28.8

model to data indicates that the high static pressures in the lower 500 psi 50 psi

part of the survey are supercharged (same data as Figure 15).

−× 700

1
Subsea depth (m)

were estimated from buildup slopes for each test. Filter- −× 750
cake properties were estimated from mud tests reported
on daily well reports. From these data, supercharge was 2
−× 800
calculated using equation 3. In general, there is a good
comparison between predicted supercharge and ob-
−× 850
served supercharge (Figure 16). This strongly supports 3
Unedited data
the hypothesis that supercharging is responsible for Edited data 1.07 g/cm3
the high measured pressures. The sandstone has a dis- −× 900

tinct coarsening-upward texture that is reflected by Figure 17. Gulf of Mexico example. Shaded zones indicate
upward-increasing permeability in the sandstone. Su- shale. (A) Pressure-depth plot having both unedited (open dia-
percharge increases systematically down the section monds) and edited (filled diamonds) wireline pressure data.
as reservoir permeability decreases. Once the cause of Petroleum-bearing zones are hard to identify because of the
anomalous pressures was identified, the intersection thin, multiple reservoirs and spurious data. (B) Magnitude of
of the pressure trend of the good gas tests and water pressure corrections for acceptable data. Supercharged tests,
tests in nearby wells indicated the actual free-water tight tests, and tests with significant probe-seal leaks were
level. removed from the data set. (C) Excess-pressure-vs.-depth plot
of edited data from the lower two-thirds of the run. Excess
pressure is calculated using water density calculated in zone 2
(1.07 g/cm3). Petroleum-bearing zones can now be clearly dis-
Gulf of Mexico
tinguished from water zones by difference slopes. The water
zone in the middle of the section has greater pressure than
An offshore Louisiana shelf, Gulf of Mexico well pen- the water legs of overlying and underlying petroleum-bearing
etrated several petroleum- and water-bearing zones. The intervals; thus, these data cannot be used to estimate the free-
presence of petroleum was known from conventional water level for the penetrated accumulations. Data from upper-
wireline logs; thus, the pressure survey was run to de- most zone are not plotted because they are off the scale.
termine petroleum fluid type and predict fluid contacts Numbers refer to the zones analyzed in Figure 18.

Brown 307
 Zone 1 with a fluid density of 0.46 g/cm3 is con- communicate with permeable zones shallower than
sistent with a liquid condensate (Figure 18A). No excess- zone 2; otherwise, water pressure could not drop below
pressure offset or density change across the upper shale that of zone 2. Zone 3 probably intersects a fault along
in this reservoir is present; thus, the shale is not a geo- which fluid leaks, whereas the sandstones in zone 2 do
logical pressure barrier. The single test below the lower not intersect a transmissive fault. If this interpretation
shale has slightly higher excess pressure, but the differ- is correct, the spillpoint for zone 3 may occur at the fault
ence lies within expected data scatter, and thus, the lower intersection with the reservoir. This also explains why
shale is not interpreted as a barrier. zone 2 sandstones are not charged with petroleum.
Zone 2 is entirely water bearing with a density of No oil tests occur below the shale in zone 3 (Figure
1.07 g/cm3, consistent with the high salinity of the pore 18C). This may be caused by either capillary effects
water. Excess pressures in the thicker, medial sandstone or by shale sealing the base of the oil accumulation.
are constant, indicating vertical pressure communication Drawdown mobilities of these tests are about 49 md/
over geological time (Figure 18B). The excess pressures cp, or about 25 md with assumed filtrate viscosity. Well-
decrease up section, indicating upward cross-formational sorted, fine-grained sandstones with a permeability of
water flow. 25 md have a displacement pressure of 14–26 kPa
Zone 3 is petroleum- and water-saturated sand- (2–4 psi) under reservoir conditions (Smith, 1966). Cap-
stone. Petroleum fluid density is 0.68 g/cm3, consis- illary pressure in the uppermost water test is 10 kPa
tent with a high-volatile oil or black oil with a high (1.5 psi). This is less than the capillary-displacement
gas-oil ratio (GOR; Figure 18C). Water has a density pressure expected for the lower sandstone; thus, mobile
similar to that of zone 2, but the excess pressure is oil saturation would not be expected in this zone,
much lower (75 kPa [11 psi] less than the deepest zone even if the shale did not seal.
2 test; Figure 17). This indicates that fluids in zone 3

DISCUSSION
(A) Zone 1 (B) Zone 2 (C) Zone 3
Excess pressure (kPa) Excess pressure (kPa)
× 620 × 650 × 680
Excess pressure (kPa)
× 410 × 420 × 430
Suitable Data and Settings for Excess-Pressure Analysis
× 175 × 185
−× 716 0.5 psi
−× 760 1 psi 1 psi

−× 850 High-resolution analysis of wireline pressure data re-

quires good-quality data. Test buildup curves should
always be evaluated for data-quality control. The end
Subsea depth (m)

Subsea depth (m)

Subsea depth (m)

−× 720 −× 780
user rarely examines the paper logs that contain build-
−× 854 up data; they are usually filed and forgotten. Digital
files are rarely even archived. Sometimes, test depth or
−× 724 −× 800 final buildup pressure is incorrectly transcribed to the
summary tables given to the end user. I recommend
−× 858
that the end user at least qualitatively examine build-
0.46 g/cm 3 1.07 g/cm 3 0.68 g/cm3 up curves for all tests prior to data analysis.
−× 728 −× 820
Strain-gauge pressure data do not have the repro-
Figure 18. Detailed analysis of three zones in the Gulf of ducibility necessary for meaningful results, given the
Mexico example. Shaded zones indicate shale. (A) Zone 1 small excess-pressure variations in most petroleum ac-
excess-pressure-vs.-depth plot (fluid density of 0.46 g/cm3, cumulations. Uncompensated quartz gauges measure
consistent with condensate). Shales do not act as geological pressure reliably if the gauge is allowed to stabilize to
barriers, because data form a single excess-pressure trend. (B)
reservoir temperatures before beginning the pretest. Most
Zone 2 excess-pressure-vs.-depth plot (fluid density of 1.07
temperature-compensated quartz-gauge data have suffi-
g /cm3, salt water). Shales act as pressure barriers. Offsets of
excess-pressure trend indicate upward flow of water. (C) Zone 3 cient resolution and reproducibility to apply the excess-
excess-pressure-vs.-depth plot is calculated with a fluid density pressure approach.
of 0.68 g/cm3 (high-GOR oil). The shale might act as a seal, Even temperature-compensated quartz pressure
but the lack of petroleum in the lower zone is more likely gauges will give poor results under bad logging condi-
caused by insufficient capillary pressure. Zone numbers are tions. Some settings always give poor results because
shown on Figure 17. supercharging and probe-seal leakage are common. Most

308 Improved Interpretation of Wireline Pressure Data

pressure tests in highly fractured reservoirs show probe- tests, slope changes in thin reservoirs may be difficult to
seal leakage during buildup. If all of the leaking tests differentiate from excess-pressure offset across a barrier.
are omitted from the data set, few data are left to ana- Possible barriers should always be verified by in-
lyze. Pressure buildup data can be used to rank tests by tegrating pressure analysis with other data. A pressure
quality, and the best-quality tests can be used for in- barrier must be associated with some lithological fea-
terpretations. Many tests in low-permeability reservoirs ture laterally extensive enough to isolate parts of the
will be supercharged. The reservoir zones with high per- reservoir. In most reservoirs, this is an evaporite bed,
meability will be closest to actual reservoir pressure. mudrock bed, or clay-rich fault zone in the depth range
Minimizing mud overbalance during testing will also of the expected seal. If a small pressure offset is asso-
minimize supercharging. Neither fractured nor low- ciated with the same stratigraphic horizon in nearby
permeability reservoirs can be analyzed to the level wells, then the barrier is probably valid.
shown in the examples, but the process of quality con- Comparing results from nearby wells can also vali-
trol can usually change totally meaningless data into date small fluid-density changes. Within-well density
data showing approximate fluid contacts and major estimates are not affected by absolute pressure errors
pressure barriers. between wells. Fluid density in the same compart-
Pressures measured in wells having nearby pro- ments or zone should be similar in nearby wells, and
duction are probably affected by production. Measured the fluid contact should occur at approximately the
excess-pressure variation may be caused by differential same elevation.
flow instead of static pressure variations. Pressure varia-
tions in wells affected by production have to be evalu-
ated for lateral connectivity to producing wells as well Using High-Quality Pressure Data
as pressure variations caused by cross-formation flow.
Shallow gas has a density so low that the gas pres- Fluid-density estimates from good-quality surveys are
sure is almost uniform in the reservoir. Where gas den- accurate enough to estimate gas gravity and oil type,
sity is low, there is no advantage to the excess-pressure not just general fluid type (water, oil, and gas). High-
approach. Standard pressure-depth plots can be used to resolution fluid-density estimates can be used to address
evaluate fluid contacts and barriers. other exploration and development problems besides
the identification of general fluid type, fluid contact, and
barrier identification. The following are some example
Limits to Barrier and Fluid Contact Identification applications of fluid-density estimates that I have used.

The excess-pressure scale can be expanded sufficiently 1. High-CO2 methane gas can be distinguished from
to display small, random excess-pressure variation. Ran- low CO2 methane gas by their subsurface density.
dom pressure variations will cause excess-pressure con- This technique works best where gas is dry because
figurations similar to barriers or fluid contacts if few ethane and other higher hydrocarbons also increase gas
tests are available over the reservoir interval. The data density. CO2 concentrations can be quantified where
can be misinterpreted, unless statistical guidelines are gas density is modeled from well-constrained pres-
used to guide interpretation. sure and temperature data.
Like any statistical problem, the confidence in the 2. Reservoir compartmentalization can be identified
slope of a data trend or change of the mean between by small petroleum density differences across poten-
two populations is controlled on the number of data tial barriers. Some flow barriers are permeable enough
points, the data variance, and (for confidence of slope for pressures to equilibrate over geological time,
estimate) the depth range over which the slope is mea- but insufficiently permeable to allow free mixing
sured. Increasing the number of valid tests and test quality of oils. Without mixing, small density differences are
control increases interpretation confidence. The thick- preserved for geological lengths of time. This ap-
ness of the petroleum-bearing reservoir (depth range) plication is similar to geochemical reservoir com-
is fixed. Using a given data variance, fluid-density reso- partmentalization detection. Density-stratified oils
lution can be increased only by taking more valid tests. (oil density decreasing upward) are gravitationally
The confidence interval for the mean excess pressure stable. Mixing is geologically slow, even where bar-
decreases with increasing sample size as predicted by riers are absent. Oil density must be different at the
the t distribution. Even when using a large number of same elevation in different wells or oil density must

Brown 309
 decrease downward across a barrier in the same well CONCLUSIONS
to indicate compartmentalization.
3. Zones with heavy oil can be distinguished from zones Conventional pressure-depth plots cannot fully dis-
having light oil in accumulations where oil quality play the resolution of modern wireline pressure data.
varies. Heavy oils have subsurface density heavier Excess-pressure plots show many subtle features in
than the density of associated light oils. Once high the pressure data that can be easily overlooked on
density identifies zones with heavy oil, completion pressure-depth plots. Excess pressure is the pressure
strategies can be designed to maximize the econom- left after subtracting the weight of the fluid from the
ic mix of produced petroleum. total pressure. The excess pressure of static, homoge-
4. Accurate prediction of phase behavior depends on neous fluid in good pressure communication will not
good samples, but collection of samples represent- change with depth; thus, excess-pressure variations
ative of the subsurface petroleum is not always suc- with depth indicate barriers and fluid contacts. The
cessful. In addition to collecting excellent-quality excess-pressure scale can be expanded as much as nec-
samples in difficult settings (e.g., Reignier and Jo- essary to evaluate minor pressure barriers and density
seph, 1992), wireline pressure tools can collect pre- changes. Using good data, within-well systematic excess-
test data to test the quality of the samples. Reservoir pressure differences of less than 5 kPa (0.7 psi) can be
fluid density predicted by numerical or experimental interpreted in terms of pressure barriers and fluid-
PVT models can be compared to the in-situ density density changes. Examples demonstrate the utility of
determined from pressure data. A large density dif- these techniques.
ference between predicted and observed density at Data-quality evaluation is essential. Small anoma-
reservoir conditions may indicate that GOR was in- lies in the buildup-pressure curve indicate pressure
correctly estimated for the fluid modeling. errors on the psi scale caused by leaking probe seals,
5. Pore-water salinity can be estimated where temper- probe plugging, and gauge problems. Suspected super-
ature is known, and in areas with known low salinity, charged samples can be identified from equation 3 or
the static reservoir temperature can be estimated by plotting the logarithm of supercharge against the
from the water density. Quantitative water salinity logarithm of test mobility. Most bad tests have to be dis-
or temperature prediction requires good models for carded, but a few can be corrected if problems are minor.
water density as a function of composition, tempera- Small excess-pressure differences between wells can-
ture, and pressure. Water density models developed not be detected as easily as within-well excess-pressure
by Batzle and Wang (1992) have proven accurate differences, because absolute-depth and pressure calibra-
over the range of normal reservoir pressures and tion between wells is poorer than within-well pressure
temperatures. resolution. Between-well pressure corrections involving
simple pressure or depth shifts are ambiguous.
The significance of pressure barriers on field produc- Fluid-density resolution is sufficiently high to use
tion behavior has sometimes been questioned because for new applications. These include petroleum quality
many barriers affecting production are not pressure evaluation, barrier detection by small density differences,
barriers. The persistence of small excess-pressure off- and validation of PVT models where sample quality is
sets across barriers in a petroleum column indicates that questionable.
petroleum pressure has not equilibrated over geological
time. Other barriers have permeability high enough to
allow pressure equilibration, but too low for geochemical REFERENCES CITED
equilibration. Geochemical means (Kaufman et al., 1990)
or small petroleum-density differences can detect these Batzle, M., and Z. Wang, 1992, Seismic properties of pore fluids:
barriers. At the high rates of flow during production, Geophysics, v. 57, p. 1396 – 1408.
all of these barriers affect production. In many fields, Brooks, A. G., and H. Wilson, 1996, An improved method for
computing wellbore position uncertainty and its application
insufficient uncontaminated oil samples are available
to collision and target intersection probability analysis: 1996
for geochemical analysis on the scale of the pressure European Petroleum Conference of the Society of Petro-
sampling. Pressure detection of barriers should be used leum Engineers, Milan, October 1996, SPE Preprint 36863,
p. 411 – 420.
with geochemical detection methodologies wherever
Brown, A., and R. G. Loucks, 2000, Evaluation of anhydrite seals in
possible, because both methods have their strengths, the Arab Formation, Al Rayyan field, Qatar (abs.): GeoArabia,
and integrated interpretations are superior. v. 5, p. 62 – 63.

310 Improved Interpretation of Wireline Pressure Data

Desbrandes, R., and J. Gualdron, 1987, In situ rock wettability of Petroleum Engineers, London, October 1984, SPE Preprint
determination with formation pressure data: Society of Pro- 12962, p. 59 – 74.
fessional Well Log Analysts 28th Annual Logging Symposium Reignier, P. J., and J. A. Joseph, 1992, Management of a North Sea
Transactions June – July 1987, v. 1, p. Z1 – Z20. reservoir containing near-critical fluids using new generation sam-
Dewan, J. T., 1983, Essentials of modern open-hole log interpreta- pling technology for wireline formation testers: European Petro-
tion: Tulsa, Oklahoma, Pennwell Publishing, 361 p. leum Conference of the Society of Petroleum Engineers, Cannes,
Fraisse, C., R. Simon, B. Seto, D. Nardio, J. Pouzet, and Y. France, November 1992, SPE Preprint 25014, p. 541–555.
Grosjean, 1987, Problem areas in interpretation of pressure Rodgers, S., 1998, A protocol for interwell pressure comparison:
measurements in thin and multilayered reservoirs: Proceedings Log Analyst, v. 39, March – April, p. 10.
of the Indonesian Petroleum Association, 16th Annual Con- Smith, D. A., 1966, Theoretical considerations for sealing and non-
vention, October 1987, p. 71 – 85. sealing faults: AAPG Bulletin, v. 50, p. 363 – 374.
Gray, G. R., and H. Darley, 1980, Composition and properties of Sollie, F., and S. Rodgers, 1994, Towards better measurements of
oil well drilling fluids, (4th ed.): Houston, Texas, Gulf logging depth: Society of Professional Well Log Analysts 35th
Publishing, 630 p. Annual Logging Symposium Transactions, v. 1, p. D1 – D15.
Hubbert, M. K., 1956, Darcy’s law and the field equations of the Stewart, G., and M. Wittmann, 1979, Interpretation of the pressure
flow of underground fluids: Petroleum Transactions of the response of the repeat formation tester: 54th Annual Fall
American Institute of Mining Engineers, v. 207, p. 222 – 239. Technology Conference and Exhibition of the Society of
Joseph, J., T. Zimmerman, T. Ireland, N. Colley, S. Richardson, and Petroleum Engineers, Las Vegas, Nevada, September 1979,
P. Reignier, 1992, The MDT tool: a wireline testing break- SPE Preprint 8362, p. 23 – 26.
through: Oilfield Review, v. 4, April, p. 58 – 65. Veneruso, A. F., C. Erlig-Economides, and L. Petijean, 1991,
Kaufman, R. L., A. S. Ahmed, and R. J. Elsinger, 1990, Gas Pressure gauge specification considerations in practical well
chromatography as a development and production tool for testing: 66th Annual Technical Conference and Exhibition of
fingerprinting oils from individual reservoirs: applications in the Society of Petroleum Engineers, Dallas, Texas, October
the Gulf of Mexico, in D. Schumaker and B. F. Perkins, eds., 1991, SPE Preprint 22752, v. , p. 865 – 878.
Gulf Coast Oils and Gases: Proceedings of the 9th Annual Viessman, W., J. W. Knapp, G. L. Lewis, and T. Harbaugh, 1977,
Research Conference of the SEPM, October 1990, p. 263 – 282. Introduction to hydrology, 2d ed.: New York, Harper & Row,
Mann, D. M., and A. Mackenzie, 1990, Prediction of pore fluid 704 p.
pressures in sedimentary basins: Marine and Petroleum Waid, M. C., M. Proett, R. Vasquez, C. Chen, and W. Ford, 1992,
Geology, v. 7, p. 55 – 65. Interpretation of wireline formation tester pressure response
Pelissier-Combescure, J., D. Pollock, and M. Wittmann, 1979, with combined flowline and chamber storage, mud super-
Application of repeat formation tester pressure measurements charging, and mud invasion effects: Society of Professional
in the Middle East: Middle East Oil Technical Conference of Well Log Analysts 33d Annual Logging Symposium Transac-
the Society of Petroleum Engineers, Manama, Bahrain, March tions, June 1992, v. 2, p. HH1 – HH25.
1979, SPE Preprint 7775, p. 199 – 231. Wolff, C. J. M., and J. P. Dewardt, 1981, Borehole position
Phelps, G., G. Stewart, and J. Peden, 1984, The effect of filtrate uncertainty — analysis of measuring methods and derivation of
invasion and formation wettability on repeat formation tester systematic error model: Journal of Petroleum Technology,
measurements: European Petroleum Conference of the Society v. 33, p. 2339 – 2350.

Brown 311