AUTHOR
Alton Brown � 1603 WaterviewDrive, Richardson, Texas, 75080;[email protected]
Alton Brown worked as a research geologistat ARCO’s Plano, Texas, Research Centerfrom 1980 until ARCO’s merger with BP.Since then, he has been an independentconsultant. His research interests includepetroleum migration, carbonate sedimentologyand diagenesis, basin analysis, and gasgeochemistry.
ACKNOWLEDGEMENTS
I thank Russell Davies for constructivecomments early in manuscript preparationand for later reviews and suggestions. I alsothank AAPG reviewers Quentin Fisher andJim Handschy for additional comments andsuggestions.
Capillary effects onfault-fill sealingAlton Brown
ABSTRACT
Capillary-pressure models and concepts were used to evaluate the
effects of excess pressure, capillary hysteresis, and relative perme-
ability on fault-fill sealing. Overpressured fault fill (fault water pres-
sure higher than reservoir water pressure) always increases the height
of the sealed petroleum column. The sealing interface moves into
the overpressured fault fill where water flows from the fault into
the reservoir. Underpressured fault fill decreases petroleum column
height only where cross-fault water flow is absent. If water flows
across the fault, column height is unaffected. Water cannot flow
across faults where the reservoir is at irreducible water saturation.
Relative permeability smoothes the transition from membrane
sealing to leakage, and thus, hydraulic-resistance sealing is possible
after membrane-seal failure. The height of membrane sealing by
homogeneous, water-wet fault fill exceeds the height of hydraulic-
resistance sealing at geological leakage rates. Hydraulic-resistance
sealing becomes more significant when charge and leakage are both
high, when trap life is short, and during production.
Trap leakage rate through a water-wet, fault-fill pore network
cannot exceed trap charge rate during initial trap charging. If charg-
ing slows, leakage exceeds charge until a new equilibrium column
height develops. If charge stops, the seal continues to leak until the
petroleum column height is reduced substantially below its original
height. Membrane sealing is reestablished at low capillary pressure.
Theoretically, restored seal capacity is close to the original capacity.
Cross-fault pressure and petroleum column height cannot be
converted to seal capacities because charge history and seal type in-
fluence sealing. Cross-fault pressure data should be analyzed in light
of the charge and pressure history so the different controls on fault-
fill sealing can be assessed.
INTRODUCTION
Fault-sealing behavior controls distribution and production charac-
teristics of fault traps. Two fault-fill–sealing mechanisms have been
Copyright #2003. The American Association of Petroleum Geologists. All rights reserved.
Manuscript received November 16, 2001; provisional acceptance April 16, 2002; revised manuscriptreceived July 15, 2002; final acceptance August 1, 2002.
DOI:10.1306/08010201127
AAPG Bulletin, v. 87, no. 3 (March 2003), pp. 381–395 381
proposed: membrane and hydraulic-resistance seal-
ing (Heum, 1996). The relative importance of the two
mechanisms remains controversial (Sales, 1993; Heum,
1996; Fulljames et al., 1997). The influence of cross-
fault water-pressure differences on fault-fill sealing is
also controversial (Myers, 1968; Heum, 1996; Bjorkum
et al., 1998).
The purpose of this article is to analyze the in-
fluence of capillary-pressure and relative permeability
variations on fault-fill sealing. Capillary-pressure varia-
tions within the fault fill control membrane sealing in
the presence of cross-fault water-pressure differences. Ef-
fective permeability changes with capillary pressure; thus,
the relative importance of membrane and hydraulic-
resistance sealing in homogeneous, water-wet faults can
be modeled as a function of capillary pressure. Hystere-
sis effects on fault-seal leakage are also evaluated.
The sealing concepts presented in this article are
based on homogeneous fault-fill properties. Real fault
fill is heterogeneous (Knipe et al., 1998). Homogene-
ity is assumed so that the article can focus on mecha-
nisms and rates, not heterogeneity characterization. The
sealing concepts developed here can be applied to well-
characterized heterogeneous fault fill in much the same
way that capillary pressure controls on saturation and
relative permeability developed for homogeneous res-
ervoirs can be applied to real, heterogeneous reservoirs.
BASIC CAPILLARY-PRESSURE CONCEPTS
Petroleum sealing is controlled by the interaction of
petroleum with water in the pore system. It is therefore
difficult to discuss matrix-sealing properties of the fault
fill without also discussing the capillary properties of
the system. The capillary pressure (Pc) is the difference
between the petroleum pressure (Pp) and water pressure
(Pw) at the same position. Natural capillary pressure
commonly develops from buoyancy of a static petro-
leum column in water (Figure 1). The elevation where
capillary pressure is zero is called the free-water level.
The petroleum and water pressures relative to the free-
water level are calculated from the height (H ) above the
free-water level multiplied by the difference of petro-
leum and water fluid densities (Up and Uw, respectively)
and the hydrostatic gradient (g).
Pc ¼ Pp � Pw ¼ rpgH � rwgH ¼ �rgH ð1Þ
Pore systems can be characterized by total porosity
and injection capillary-pressure curves. The injection (or
drainage) capillary-pressure curve is a plot of fluid satu-
ration (horizontal axis) as a function of capillary pressure
(vertical axis) as the wetting fluid drains from the rock
(Jennings, 1987). This mimics filling accumulations.
Injection curves have three important characteristics:
382 Capillary Effects on Fault-Fill Sealing
01
Pe
PdPlateau
W-Rhysteresisloops
Pt
Si
Smax
FWL
Water + oil production;transition zone
Water-free production
Water production;spotty stain
Water production
PWC
Fractional Hg saturation
Log
(Cap
illar
y Pr
essu
re)
Figure 1. Capillary-pressurecurves. Equilibrium mercuryinjection curve is shown by theheavy solid line. Withdrawal-reinjection (W-R) hysteresisloops (mercury withdrawalfollowed by reinjection) areshown by dashed lines witharrows showing the direction ofsaturation change. Equilibriuminjection does not developirreducible saturation, and Smax
is the maximum mercury satura-tion attained at the maximumapplied capillary pressure.Air-brine tests (thin solid line)develop an irreducible watersaturation (S i). Productivezones are shown on the right.Zone thickness is distorted bylogarithmic capillary-pressurescale. Symbols are identifiedin Table 1.
the threshold pressure, the pressure at which water
saturation becomes irreducible, and curve shape factor
(Figure 1). The threshold pressure, P t, is the pressure at
which a continuous thread of nonwetting fluid extends
across the sample and petroleum permeability is great-
er than zero at higher capillary pressure. Threshold pres-
sure through an infinite medium is interpreted at the
inflection between the concave downward and concave
upward curvature in the lower part of the injection pla-
teau measured on small samples (Katz and Thompson,
1987). In practice, threshold pressure is approximated
by the displacement pressure (Pd), the extrapolation of
the plateau to 100% wetting-phase saturation (Jennings,
1987; Figure 1). At irreducible saturation (S i), wetting-
fluid permeability is zero; thus, fluid saturations remain
constant with increasing capillary pressure. Pressure at
initiation of irreducible saturation is approximated by
the pressure just above the plateau. The shape factor (l)
describes the slope of the plateau and abruptness of
change from plateau to irreducible water saturation.
Saturation as capillary pressure decreases is de-
scribed by the withdrawal curve (Figure 1). Withdrawal
mimics natural leakage from a petroleum accumulation.
Withdrawal curves are characterized by residual petro-
leum saturation at zero capillary pressure and petro-
leum saturation higher than that of the injection curve.
High saturation is caused by snap-off and isolation of
petroleum in large pores with small throats (Wardlaw
and Taylor, 1976).
Capillary-pressure curves are commonly measured
with the mercury vacuum system. This has the advan-
tage of rapid analysis and high reproducibility. Mercury-
injection, capillary-pressure tests have two major disad-
vantages. First, mercury vacuum capillary pressures have
to be converted to reservoir capillary pressure. The res-
ervoir surface tension and wettability necessary for con-
version are poorly known in most reservoir and explora-
tion settings (Schowalter, 1979; O’Connor, 2000).
Second, mercury vacuum tests do not provide a good
estimate of irreducible saturation or pressure at which
irreducible saturation develops, because mercury does
not displace a fluid as capillary pressure increases (Mor-
row, 1971; Wardlaw and Taylor, 1976).
FAULT-FILL SEALING
Upward-increasing capillary pressure divides potential
seal behavior of water-wet, homogeneous, fault fill into
three zones: membrane sealing, hydraulic-resistance seal-
ing, and seal failure (Figure 2). Membrane sealing occurs
when petroleum has insufficient capillary pressure to
invade the seal. This zone occurs in all water-wet rocks.
Higher in the petroleum column, capillary pressure is
sufficient to invade the fault fill; thus, petroleum leaks.
Low in the zone of leakage, effective permeability is in-
sufficient for geologically significant leakage; thus, the
fault seals by hydraulic resistance (Heum, 1996). Effec-
tive fault-fill permeability and cross-fault pressure gradient
Brown 383
Table 1. Symbols Used in Equations and Figures*
Symbol Meaning
FWL free-water level, the elevation where
capillary pressure is zero
H height above free-water level
h height of truncated pyramidal trap geometry
used for calculating trap leakage
H si height above FWL where irreducible
saturation develops in reservoir
H t height above FWL where capillary pressure
equals threshold pressure
k a absolute permeability
k ro relative permeability to petroleum
k w water relative permeability
L length of edge of truncated pyramidal trap geometry
PWC petroleum-water contact (either oil or gas)
P c capillary pressure
Pd displacement pressure
P e entry capillary pressure
Pp petroleum pressure
P t threshold pressure
P tf threshold pressure in fault fill
P tr threshold pressure in reservoir
Pw water pressure
S i irreducible water saturation
Smax maximum mercury saturation during
mercury-injection, capillary-pressure test
Sw water saturation
DP excess pressure at sealing interface
DP t total excess pressure (water pressure at aquifer
side of fault-fill minus water pressure
in reservoir at same elevation)
DUg capillary-pressure gradient, water density minus
petroleum density (DU) times
local hydrostatic gradient (g)
l capillary-pressure curve shape factor
Up in-situ petroleum density
Uw in-situ water density
*Symbols used only in the Appendix are not listed.
increase upward; thus, leakage increases upward until it
becomes geologically significant. This is the zone of fault-
fill seal failure. Seal-failure threshold is controlled by the
charge rate, timescale of interest, and area of leakage.
Exceptionally high petroleum columns are needed to
cause failure of low-permeability, fault-fill seals. The
absence of this highest zone allows faults to seal.
Relative permeability smoothes the transition from
membrane sealing through hydraulic-resistance seal-
ing to seal failure. If permeability abruptly increased
to the absolute permeability once the threshold pres-
sure was exceeded, then it would be possible for some
rocks to form a membrane seal but not a hydraulic-
resistance seal. With relative permeability effects, pe-
troleum columns exceeding membrane-seal capacity
will always have a zone of hydraulic-resistance seal-
ing if fault fill is homogeneous. Likewise, all water-
wet hydraulic-resistance seals have an underlying
zone of membrane sealing.
Membrane Sealing
Membrane Sealing under Hydrostatic Conditions
Membrane sealing is sealing by surface tension between
water and petroleum (Watts, 1987). Petroleum perme-
ability is zero when capillary pressure is less than the
384 Capillary Effects on Fault-Fill Sealing
Fau
lt fil
l
Res
ervo
ir
Pc = 0 (FWL)
Mem
bran
ese
alH
ydra
ulic
-re
sist
ance
sea
lN
ot a
seal
Geologicallysignificantleakage
Pc
= ∆
ρgH
H, P
c
Pc = Ptf
Pc = Ptr (PWC)
(B)
C
Slowleakage
OWC
Si
Fault fill
Res
ervo
ir
Top Seal
Ptf
FWL
)
Fault fill ReservoirAquifer
Pw
Pp
Pc
Distance
pres
sure
Pw+Pt(C)
Membrane sealFWL
Not a seal
seal
Hydraulic-resistance
(AFigure 2. Fault-seal mecha-nism zonation caused bycapillary-pressure variationunder hydrostatic conditions.(A) Block diagram showingreservoir and fault-fill seal zo-nation. Free-water level (FWL)is elevation with zero capillarypressure. Petroleum-watercontact (PWC) occurs wherecapillary pressure equals reser-voir threshold pressure (P tr). S i
is the base of the irreduciblewater zone in the reservoir. Theboundary between membraneand hydraulic-resistance sealingis elevation where capillarypressure equals the fault-fillthreshold pressure (P tf). (B)Section along fault fill andadjacent reservoir showing dis-tribution of petroleum (hori-zontal line pattern) in reservoir(right column) and fault fill (leftcolumn) as a function of height(H ). Hydraulic-resistance seal-ing fails where cross-fault leak-age becomes geologicallysignificant. (C) Pressure crosssection of reservoir-fault fill-aquifer at level ‘‘C’’ labeled inB. Fault-fill seals because capil-lary pressure is less than fault-fill threshold pressure at thereservoir interface. Thresholdpressure in aquifer is assumedto be small, thus, the aquifer isnot shown in A and B. Symbolsare identified in Table 1.
displacement pressure, and the rock acts as a perfect seal.
Petroleum-wet rocks (such as fault fill containing mature
source rocks) have no membrane-seal potential under
hydrostatic conditions; thus, petroleum can only be
sealed by hydraulic resistance.
In water-wet rocks, the lowest capillary pressure at
which petroleum permeability rises above zero is the
threshold pressure (P t); thus, membrane-seal capacity is
equal to the capillary threshold pressure. Under hydro-
static conditions, the maximum petroleum column
height, H t, can be calculated from the threshold
pressure and the density difference using
Ht ¼ Pt=�Ug ð1Þ
The threshold pressure is controlled by the pore net-
work (Watts, 1987), saturation history (Wardlaw and
Taylor, 1976), wettability, and surface tension (Ibra-
him et al., 1970).
Excess Pressure Effects on Membrane Sealing
For purposes of faults-seal analysis, excess pressure (DP)
is defined as water pressure on the aquifer side of the
sealing interface minus water pressure in the petroleum-
bearing reservoir at the same elevation. Total excess
pressure (DP t) is the aquifer water pressure minus the
reservoir water pressure at the same elevation. ‘‘Over-
pressure’’ indicates positive excess pressure (higher water
Brown 385
Figure 3. Fault-seal mechanisms with positive fault-fill excess pressure. (A) Section along fault fill (left) and adjacent reservoir (right)showing distribution of petroleum saturation as a function of height (H ). The base of irreducible water saturation is below H = P tf/DUg,thus, sealing interface is entirely at the reservoir-seal boundary. (B) Same as A, with base of reservoir irreducible water saturation aboveH = P tf /(DUg). Sealing interface extends into the fault fill, but seal failure does not occur because DUgH < P tf + DP t. Membrane-sealfailure occurs where H = (P tf + DP t)/DUg. (C) Pressure cross sections across reservoir-fault fill-aquifer at constant elevation. Crosssection pressure behavior is referenced in A and B by the numbers at the left of each cross section. Symbols are identified in Table 1.
pressure on the aquifer side), and ‘‘underpressure’’ in-
dicates negative excess pressure (higher water pressure
on the reservoir side). Excess pressure at the sealing
interface depends on its location and the presence or
absence of cross-fault water flow. When the reservoir
water saturation equals irreducible water saturation, it
is impermeable to water. Water cannot flow across the
fault; thus, fault-fill excess pressure is uniform and equal
to the total excess pressure (pressure trend 1, Figures 3C,
5C). If the reservoir is permeable to water (water
saturation is greater than irreducible water saturation),
water flows across the fault and excess pressure varies
across the fault fill (pressure trends 2 and 3 in Figure 3C
and pressure trend 2 in Figure 5C). Excess pressure is less
than the total excess pressure except at the aquifer-fill
interface.
Capillary pressure varies with elevation (H) and
excess pressure (DP):
Pc ¼ �rgH ��P ð2Þ
A membrane seal fails when the capillary pressure
equals the fault-fill threshold pressure (Pc = P tf). From
equation 2, the maximum membrane-sealed petro-
leum column height (H t) in the presence of excess
pressure is the following:
Ht ¼ ðPtf þ�PÞ=�rg ð3Þ
Threshold pressure remains constant with variable
excess pressure, but the height equilibrated to the thresh-
old pressure varies. From equation 3, seal overpressure
enhances the petroleum column height, whereas seal un-
derpressure reduces the petroleum column height.
Fault-fill overpressure will always help seal a taller
petroleum column, as described by equation 3, with
DP = DP t. Cross-fault flow causes the sealing interface
to move into the fault fill at some elevations above the
free-water level (Table 2, Figure 3). The sealing inter-
face moves back to the fill-reservoir contact at the
elevation when the reservoir is at irreducible water sat-
uration. If reservoir’s irreducible water saturation
develops in the zone of hydraulic-resistance sealing,
the transition from membrane to hydraulic-resistance
sealing occurs in the fault fill and the upper part of the
membrane seal acts like a hydrodynamic seal (Figure 4).
Hydrodynamic sealing is caused by bending the
free-water level and petroleum-water contact in re-
sponse to change in potentiometric gradient (Hubbert,
1953). Fault fill has a lower effective water permeabil-
ity than the reservoir; thus, potentiometric gradient and
hydrodynamic tilt are steeper in the fault fill than in
the reservoir (Figure 4). Contact tilting is independent
of rock wettability; thus, hydrodynamic membrane
seals can develop in petroleum- and mixed-wettability
rocks. The intersection of the tilted petroleum-water
contact in the fault fill with the aquifer marks the
transition from hydrodynamic sealing to hydraulic-
resistance seal. If the petroleum column is too thin for
hydraulic-resistance sealing to develop, hydrodyna-
mic sealing prevents cross-fault leakage. Petroleum
will be lost up the fault unless fault fill above the
reservoir acts as a hydraulic-resistance or membrane
seal.
386 Capillary Effects on Fault-Fill Sealing
Table 2. Effects of Excess Pressure on Membrane Fault-Fill Sealing
Sealing interface position and effect
of excess pressure on sealing
Valid elevation range,
overpressured fault fill (DP t > 0)
Valid elevation range,
underpressured fault fill (DP t < 0)
Seal at reservoir-fault interface:
column height is affected by excess pressure:
H t = (P tf + DP t )/DUg
H si < P tf /DUg H si < (P tf + DP t )/DUg
Seal within fault fill:
column height is affected by excess pressure:
H t = (P tf + DP t )/DUg
P tf /DUg < H si < (P tf + DP t )/DUg not possible
Hydrodynamic sealing within fault fill:
column height is affected by excess pressure:
H t = (P tf + DP t )/DUg
H si > (P tf + DP t )/DUg not possible
Seal at reservoir-fault interface:
column height is unaffected by excess pressure:
H t = P tf /DUg
not possible H swi > (P tf + DP t )/DUg
Underpressured fault fill may or may not affect seal
capacity, depending on presence of cross-fault flow (Table
2). Where water does not flow across the fault, column
height is described by equation 3, withDP =DP t (Figure
5C, case 1). Sealing is always at the reservoir interface
if fault fill is homogeneous. Cross-fault flow causes
infinitesimal excess pressure at the fault-reservoir in-
terface; thus, seal capacity is unaffected by flow (Figure
5C, case 2). Fault-fill hydrodynamic sealing is impos-
sible where the fault fill is underpressured.
Hydraulic-Resistance Sealing
Basic Concepts for Hydraulic-Resistance Sealing
Hydraulic-resistance sealing occurs where rocks are
permeable to petroleum, but petroleum permeabil-
ity is so low that petroleum can be trapped for geo-
logical lengths of time (Heum, 1996). The base of the
hydraulic-resistance sealing zone is the elevation where
capillary pressure equals the threshold capillary pres-
sure and petroleum permeability is zero. Excess pres-
sure changes the elevation of this threshold, as discussed
in the previous section. Petroleum-wet rocks (such as
fault fill containing mature source rocks) spontaneously
imbibe petroleum; thus, hydraulic-resistance sealing
begins at negative or zero capillary pressure.
Hydraulic-resistance seals fail when leakage be-
comes geologically significant. Leakage rate must be
less than charge rate; otherwise, the accumulation will
not form. When leakage rate exceeds charge rate after
formation of the accumulation, the leakage must be
sufficiently slow to preserve the accumulation since
trapping, a time span measured in millions to hun-
dreds of millions of years. Sealing, therefore, depends
on timescale of interest, area of leakage, and effective
petroleum permeability of the seal.
Effective petroleum permeability is the product of
absolute permeability and relative petroleum perme-
ability. No siliciclastic rock has been identified with
zero absolute permeability (Neuzil, 1994); thus, all
siliciclastic rocks (including fault fill) are potential
hydraulic-resistance seals where capillary pressure is
sufficiently large. Absolute permeability is controlled
by the pore network. Absolute permeability can be
estimated from the threshold pressure (Smith, 1966;
Jennings, 1987). Relative permeability is best consid-
ered a function of capillary pressure for sealing studies
because saturation is a function of capillary pressure
(Figure 6). Petroleum saturation increases during trap
charging; thus, injection relative permeability is appli-
cable during charging. Injection relative permeability
can be calculated from the injection capillary-pressure
curve (Jennings, 1987; see Appendix). When leakage
rate exceeds charge rate after the accumulation forms,
Brown 387
Res
ervo
ir
Fau
lt fil
l
H = Ptr/(∆ρg)(Tilted reservoir
PWC)
Lines ofequal waterexcesspressure
H = Ptf/(∆ρg)
Waterflow
Tilted fault-fillPWC
H = (Ptf + ∆P)/∆ρgHyd
raul
ic-
resi
stan
cese
al
Mem
bran
e se
al
Figure 4. Section along the fault fill (left) and adjacent res-ervoir (right), showing distribution of petroleum with hydro-dynamic sealing in the upper part of membrane-sealing zone.Capillary pressure varies laterally at the same elevation, andthus, all fluid contacts tilt. Low fault-fill permeability amplifies thehydrodynamic tilt, and thus, the oil-water contact tilts steeply.The tilted oil-water contact intersects the juxtaposed permeablebed at the transition to hydraulic-resistance sealing. Symbols areidentified in Table 1.
the capillary pressure decreases; thus, relative perme-
ability follows withdrawal curves (Figure 6).
Modeling Hydraulic-Resistance Sealing
Cross-fault petroleum leakage is modeled as a function
of height above the free-water level. Details of the mod-
eling are presented in the Appendix, and properties of the
modeled petroleum fluids are given in Table 3. Fault fill is
assumed homogeneous and water wet. Water pressures
are assumed hydrostatic. The height of the fault through
which petroleum flows (transmissive height of the
fault) is normalized to the height of membrane sealing.
Leakage is calculated as petroleum transmissivity, the
volumetric flow of petroleum across the entire height
of the homogeneous fault above the free-water level per
lateral unit length of fault (Figure 7B). Transmissivity
multiplied by the fault length gives the petroleum
leakage across the fault.
Transmissivity is found to be a function of normal-
ized transmissive height, capillary-pressure curve shape
factor (l), and petroleum fluid characteristics (Figure 7A).
Normalization of the petroleum column height removes
dependence on threshold pressure. This simplifies anal-
ysis of hydraulic-resistance sealing. The four trends
on Figure 7A show the range of oil and gas transmis-
sivity for fault fill with a reasonable range of l and
typical oil and gas fluid properties (Table 3). The shape
factor affects transmissivity by a factor of about two.
Gas and oil have transmissivities differing by about an
order of magnitude due to different viscosity, density,
388 Capillary Effects on Fault-Fill Sealing
Figure 5. Fault-seal mechanisms with negative fault-fill excess pressure. (A) Section along fault fill (left) and adjacent reservoir(right) showing distribution of petroleum saturation as a function of height (H). Reservoir irreducible water develops below H = (P tf +DP)/(DUg). Trapped petroleum column is reduced by H = (DP)/(DUg). (B) Same as A, with reservoir irreducible water developingabove H = (P tf + DP)/DUg. Membrane-seal capacity is unaffected by pressure difference. (C) Horizontal pressure cross sectionsacross the reservoir/fault fill/aquifer applicable to underpressured faults. Cross section pressure behavior is referenced in A and B bythe numbers at the left of each cross section. Symbols are identified in Table 1.
and interfacial tension. Hydraulic-resistance sealing is
evaluated for three settings: during charge, after charge,
and during production.
The maximum possible leakage rate during charge
is the charge rate. Charge rates are estimated by dividing
the petroleum-in-place by the charge duration for
traps that have little documented leakage. The highest
observed charge rates are approximately 500 m3
petroleum per year, calculated for giant Pleistocene
accumulations with minor leakage. Assuming sealing
faults 1 km long, leakage equals charge rate when
transmissivity is 0.5 m3 petroleum per meter of fault
length per year (m3/m/yr). At this transmissivity, the
ratio of height of hydraulic-resistance sealing to height
of membrane sealing for gas accumulations is about
0.2; that is, about 15% of the total sealing height is
caused by hydraulic-resistance sealing (Figure 7A).
Height ratio for oil accumulations at this transmissiv-
ity is about 0.5; that is, hydraulic-resistance sealing
accounts for about one-third of the trapping height. The
charge rate used here is the maximum realistic possi-
ble rate. It is at least an order of magnitude higher
than that of the vast majority of accumulations. Slow-
er charged accumulations with lower petroleum vol-
umes will have a much smaller fraction of hydraulic-
resistance sealing, in the range of 1–10% of the
trapping height caused by hydraulic-resistance sealing.
To evaluate dominant sealing type after charge
stops, trap geometry and fault length must be assumed.
Trap geometry and fault length were scaled to a
truncated pyramid with variable height to edge-length
ratio (Figure 7D, Appendix). Different trap lifetimes
were assumed to determine a reasonable range of
maximum leakage rates for accumulations with the
assumed lifetime (Appendix). Leakage rate is the
moveable petroleum divided by life span. To estimate
maximum leakage rates, consider the shortest reason-
able lifetime (3 m.y.) and the largest reasonable accu-
mulation (2 � 109 m3 petroleum). With this trap size
and lifetime, transmissivity is 0.03 for trap height/side
length ratio (h/L) of 0.01 and 0.1 for h/L of 0.1. These
transmissivities correspond with gas height ratio near
0.1 (9% hydraulic-resistance sealing) and oil height
ratio near 0.2 (16% hydraulic-resistance sealing). The
assumed leakage rate is much higher than those ex-
pected for accumulations with longer life and smaller
reserves; thus, the estimated height ratios are near the
maximum that could be expected for homogeneous,
water-wet fault seals.
Production rates can be used to determine signifi-
cance of hydraulic-resistance sealing during production.
Assume a field production rate of 1,000,000 m3/yr,
with one-tenth of this volume crossing a 1-km fault bar-
rier within the field. Transmissivity is about 100 m3
petroleum per meter of fault length per year. Height
ratio for gas accumulations at this transmissivity is
Brown 389
Figure 6. Relative permeability vs. capillary pressure. Gaspermeability increases during injection (thin solid line) as waterpermeability (heavy solid line) decreases. Gas withdrawal-secondary injection hysteresis loops are shown by dashed lineswith arrows showing direction of permeability change. Non-wetting phase permeability is lower during withdrawal thanduring injection. Secondary and primary injection have similarthreshold pressures (P t). Gas relative permeability data arefrom Colonna et al. (1972). The water relative permeabilitycurve is generic.
Table 3. Model Parameters
Property Oil Gas
Capillary-pressure gradient, 0.003 (0.13) 0.008 (0.347)
MPa/m (psi/ft)
g cos u, dyn-cm 26 33
Viscosity, cp 1 0.02
Fault width, m (ft) 1 (3.3) 1 (3.3)
about 1; that is, about 50% of the petroleum-water
contact elevation difference developing across the bar-
rier during production is caused by hydraulic-resistance
sealing. Height ratio for oil accumulations at this
transmissivity is about 4; that is, hydraulic-resistance
sealing accounts for about 80% of oil-column height
differences that develop during production.
Capillary Hysteresis and Fault Sealing
Capillary pressure decreases during production or nat-
ural trap leakage. Relative permeability changes with
decreasing capillary pressure are different from those
with increasing capillary pressure (Figure 6). At first,
petroleum relative permeability does not decrease sig-
nificantly as capillary pressure decreases. As capillary
pressure approaches the threshold pressure, petroleum
relative permeability drops abruptly with decreasing
capillary pressure. Permeability drops to zero (mem-
brane sealing) at a withdrawal capillary pressure about
half to a quarter of the injection threshold pressure.
This matches withdrawal seal capacity predicted from
theoretical analysis of the snap-off process (Petroleum
Research Corporation, 1959). Mercury, secondary-
injection entry pressure on hysteresis loops indicates
that the original seal capacity is restored once the
390 Capillary Effects on Fault-Fill Sealing
Figure 7. Hydraulic-resistance–sealing model. (A) Calculated normalized transmissivity as a function of height of transmissive faultnormalized to height of membrane-sealing zone. This is a function of the capillary shape factor, l, fluid density, surface tension, andwettability (Table 3). (B) Expected transmissivity as a function of accumulation size and accumulation life, based on trap geometricmodel shown in D. Dot shows example discussed in text. (C) Definition of transmissivity. Total petroleum loss rate is calculated bymultiplying the transmissivity by the length of the fault. (D) Trap geometry assumed to estimate transmissivity from accumulationlifetime and size. See Appendix.
capillary pressure has dropped sufficiently. Healing of
seal capacity is also reported in other studies (e.g., Co-
lonna et al., 1972; Schowalter, 1979).
Relative permeability hysteresis affects petroleum
column height when charge rate to a leaking accumu-
lation slows. This can be illustrated with a simple,
hydrostatic fill-and-leak charge history (Figure 8). As
the trap fills, it is initially membrane sealed. Leakage is
zero and the petroleum column increases until the
capillary pressure exceeds the seal threshold pressure.
Initial leakage rate is slower than charge, and thus, the
petroleum column and capillary pressure continue to
increase. As capillary pressure increases, effective seal
permeability increases until leakage rate equals charge
rate (Figure 8A). Leakage rate cannot exceed charge
rate as long as charge rate, petroleum type, seal phys-
ical properties, and hydrodynamic setting remain
constant.
As the charge rate begins to slow, the seal’s effec-
tive permeability remains about the same (Figure 6);
thus, leakage rate now exceeds the diminished charge
rate. The free-water level rises and capillary pressure
decreases (Figure 8B). Leakage rate decreases in re-
sponse to the lower capillary pressure, until leakage
equilibrates with the lower charge rate at a thinner
petroleum column. If charge stops, leakage will con-
tinue until membrane sealing is reestablished at a cap-
illary pressure one-half to one-quarter of the original
membrane-sealing height. Most of the petroleum ac-
cumulation will be lost and the thinned accumulation
will have a thick, residual-oil–saturation zone (Figure
8C). Once membrane sealing is reestablished, later
Brown 391
Figure 8. Effect of petroleum perme-ability hysteresis on seal failure. The leftfigures show fluid contacts and trap con-figuration; right figures are effective per-meability of seal and reservoir (horizontalaxis) plotted against capillary pressurescaled for height (vertical). (A) Leakagerate equals charge rate. Seal permeabilityis on injection curve (i). (B) As chargerates slows, the oil column shrinks andcapillary pressure decreases to a newequilibrium. Oil relative permeabilityis on the withdrawal curve (w), andinitial decrease in permeability is modest.(C) If charge stops, the FWL rises untilwithdrawal-oil-permeability of the sealdrops to zero. This reduces the columnto about one-half to one-fourth of itsoriginal height before membrane sealingis reestablished. (D) Threshold pressure isrestored to approximately the originalthreshold pressure and additional chargecan build petroleum column to a heightnear the original column without leaking.
charge can refill the trap to its original membrane-seal
capacity (Figure 8D).
DISCUSSION
Cross-Fault Water-Pressure Differences and Sealing
The conflicting interpretations by Heum (1996) and
Bjorkum et al. (1998) arise in overpressured reservoirs
(‘‘water-drive leakage’’ of Heum, 1996, and ‘‘under-
pressured seals’’ using the terminology of this article).
Heum (1996) assumes no flow through the seal, and
thus, total water-pressure difference is concentrated at
the sealing interface, as shown in Figure 5C (case 1).
Bjorkum et al. (1998) assume that water will always
flow through the entire oil-saturated reservoir under
hydrodynamic conditions; thus, the pressure drop is
distributed across the seal (Figure 5C, case 2). Water
only flows through petroleum reservoirs that have not
reached irreducible water saturation. Tall petroleum
columns in a good-quality reservoir are generally docu-
mented to reach irreducible water saturation (Mor-
row, 1971) and the Bjorkum et al. (1998) analysis
only applies to thin columns in poor-quality reser-
voirs. Although both interpretations are correct in the
proper setting, settings favoring reduction of petro-
leum columns by underpressured seals are more likely
in large petroleum accumulations.
Interaction of Capillary Pressures across Faults
Seal capacity has been interpreted to equal the total
cross-fault pressure difference where petroleum col-
umns lie on both sides of the fault (e.g., Yielding et al.,
1997; Grauls et al., 2000). This interpretation under-
estimates injection seal capacity. As pointed out by
Fisher et al. (2001), once seal effective permeability to
petroleum is greater than zero, petroleum columns will
communicate across the fault fill. Pressures equalize if
capillary pressure does not drop below the withdrawal
threshold pressure of the fault fill at the communication
point. Older works (e.g., Smith, 1980) have long rec-
ognized this fact. Where petroleum-water contacts are
different, the fault is sealing by either membrane or
hydraulic resistance. Seal capacity is estimated from the
capillary pressure calculated for the tallest petroleum
column.
This misinterpretation is based on Watts’s (1987)
conclusion that free-water level elevation differences
across a leaking fault will be maintained during cross-
fault leakage at a value where the difference in capillary
pressure equals the threshold pressure of the fault.
Watts’s (1987) interpretation is incorrectly cited to
Smith (1966, 1980). Smith’s (1966) figure 7, upon
which Watts’s (1987) figure 11 is based, clearly states
that differences in petroleum-water contact are caused
by differences in threshold pressure of sandstones
across the fault, not the entry pressure of the fault
fill.
Although the effects of differences in drainage and
imbibition relative permeability were discussed by
Watts (1987), the effects of permeability hysteresis
on reservoir behavior differences across a leaking fault
were not discussed. Free-water levels equalize once
capillary pressure exceeds fault threshold pressure,
assuming that capillary pressure does not drop below
the withdrawal threshold pressure. The reservoir that
leaked petroleum will have effective petroleum perme-
ability that lies along the withdrawal curve, whereas
effective petroleum permeability in the reservoir that
was charged by the leakage lies along the injection
curve. The reservoir that leaked petroleum has higher
effective petroleum permeability than that across the
fault at the same elevation, assuming equivalent rock
properties. The reservoir that leaked petroleum will
also have water-free petroleum production lower in the
accumulation and a thicker nonproductive residual oil
zone below the petroleum-water contact. Residual oil
zones at the base of the accumulation can be used to
identify which reservoir has leaked across the fault. If
trap compartments were charged separately, the fault
seal may not have leaked. This can be confirmed by
evaluating the thickness of the transition zone where
reservoir rock properties are similar on both sides of
the fault.
Membrane vs. Hydraulic-Resistance Seals
Hydraulic-resistance sealing in homogeneous, water-wet
faults does not substantially increase the membrane-
sealed height before leakage becomes geologically
significant. In most natural settings, fault fill will leak
at geologically significant rates once the petroleum
column is a few percent greater than the height of
the membrane-sealing zone (Figure 7). Hydraulic-
resistance sealing does not double the membrane-
sealed height even under worst-case natural scenarios,
given homogeneous, water-wet seals. This result is
consistent with caprock sealing models by Ingram
et al. (1997) and consistent with the conclusion that
hydraulic resistance is not significant for sealing fault
392 Capillary Effects on Fault-Fill Sealing
traps (Fulljames et al., 1997). Hydraulic-resistance
sealing of water-wet, homogenous, fault-fill seals is
uncommonly important because the transmissivity
required for geologically significant leakage is small.
Most faults that leak at geological flow rates will act
as hydraulic-resistance seals during production, because
rates sufficient for geological leakage are much lower
than production rates (Fulljames et al., 1997). Prepro-
duction membrane seals may also become hydraulic-
resistance seals during production unless their displace-
ment pressures are very high and field water-drive is
strong. As water pressure in the fault fill drops with
production, capillary pressure at the interface between
the fault fill and an undrained compartment increases
until petroleum invades the fault and the fill leaks.
Models to date are relatively primitive, including
the one presented here, and the significance of model-
ing results may change with further investigation. Cur-
rent models do not account for fault-fill heterogeneity,
which is certainly significant. Fault-fill heterogeneity
will cause leakage rates to vary independent of elevation.
The thickness of the zone of membrane sealing will
thin, and the zone of hydraulic-resistance sealing may
thicken or thin. Leakage rate is also affected by the
threshold pressure of the aquifer into which petroleum
leaks. Fault-fill transmissivity may also be lower in het-
erogeneous faults than that modeled for homogeneous
faults, because the vertical fraction of the fault actually
leaking is smaller than modeled.
Pulsed Seal Leakage
Petroleum permeability hysteresis dampens pulsed
leakage through intact fault-fill and matrix seals. Leak-
age rate is relatively insensitive to minor changes in cap-
illary pressure. Petroleum permeability at membrane-
seal failure is infinitesimal and increases only if capillary
pressure increases significantly above the threshold
pressure (Figure 6). Petroleum permeability remains
almost unchanged during the initial stages of decreas-
ing capillary pressure (Figure 6). The leakage rate of an
intact membrane seal exceeds the charge rate only if
the charge rate drops (Figure 8). Once the leakage rate
exceeds the charge rate, the trap will drain to a low
capillary pressure before initial sealing capacity is
reestablished. Although hysteresis may seem to en-
hance pulsed leakage, the charge-rate variation causes
variable leakage rate. Overall, hysteresis dampens mi-
grating petroleum flux variations and causes leakage-
rate variation to lag behind charge-rate variation. These
effects push petroleum migration through porous media
such as fault fill or sedimentary strata toward steady flow.
Pulsed leakage does occur in nature. The best example
is leakage by natural hydraulic fracture (e.g., Nunn and
Meulbroek, 2002). Pulsing is caused by gravitational
instability of individual hydraulic fractures. This flow
mechanism does not involve matrix porosity; thus,
matrix capillary properties do not dampen pulsing.
Harris et al. (1999) propose that pulsed water-
pressure changes in the reservoir will cause pulsed
leakage through matrix pore systems. They assumed an
abrupt water-pressure change at the fault-reservoir
interface, incompressible fluid behavior, and abrupt
change in permeability at the threshold pressure. These
assumptions favor cross-fault leakage and leakage rate
independent from pulse duration. Relative permeabil-
ity variation and fluid and frame compressibility sig-
nificantly lower and smooth fault leakage rates. If water
pressure changes are caused by dilation or compression
of the rock pore space (during earthquakes, for exam-
ple), then water-pressure differences between the res-
ervoir and fault fill will be modest or absent, because
both pore systems will compress.
CONCLUSIONS
Capillary-pressure concepts can be used to evaluate
sealing in homogeneous water-wet fault fill. Membrane
sealing is controlled by the capillary threshold pressure,
which can be determined from capillary-pressure tests.
Absolute permeability and relative permeability can
also be estimated from the capillary-pressure curve
and threshold pressure (Appendix). Results from this
theoretical evaluation include the following:
� Relative permeability smoothes the transition from
membrane sealing to geologically significant leakage.
All homogeneous, water-wet faults have a range of
petroleum column height where leakage is slow
enough to seal by hydraulic resistance.� Higher excess pressure in faults increases the petro-
leum column regardless of cross-fault water flow, as
estimated by equation 3. Sealing occurs within the
fault fill where water flows across the fault fill; other-
wise, sealing occurs at the interface between the fault
fill and the reservoir.� Lower excess pressure in sealing faults decreases the
petroleum column where water does not flow across
the fault. When water flows across the fault, under-
pressure does not decrease the petroleum column.
Brown 393
Water flows across faults only if the reservoir is perme-
able to water; that is, water can flow when reservoir
water saturation exceeds irreducible water saturation.� Absolute permeability and relative permeability are
scaled to the capillary-pressure curve; thus, the leak-
age rate across a water-wet, homogeneous fault is a
function of the ratio of transmissive fault height to
membrane-sealing height. This ratio is independent
from the threshold pressure. The leakage rate for
normalized fault height is controlled by the capillary
shape factor and fluid properties.� Hydraulic-resistance sealing adds a relatively modest
sealing height to water-wet, homogeneous, membrane-
sealing faults. Leakage through the fault abruptly
increases with increasing column height; thus, leak-
age becomes geologically significant at a relatively
short elevation above the height of membrane seal-
ing. Membrane-seal capacity of water-wet, homo-
geneous faults can probably be estimated from the
pressure differences across hydraulic-resistance
fault seals, especially where charge rate is slow.� Charge history is an essential variable for interpret-
ing cross-fault pressure differences where petroleum
leaks. Active charge offsets the effects of leakage;
thus, petroleum column height slightly exceeds
membrane-sealing height. However, once the charg-
ing rate starts to decrease, the petroleum column
height may drop as much as 75% from its original
height, especially in settings where charge has com-
pletely stopped. Thin fault-sealed petroleum col-
umns may result from low-withdrawal threshold
pressure after an earlier stage of intense leakage, not
low-injection-threshold pressure.
APPENDIX: FAULT LEAKAGE MODEL
Estimation of Cross-Fault Transmissivity
Hydraulic-resistance leakage was modeled in the following steps. (1)Absolute permeability of a homogeneous fault fill was calculatedfrom the threshold pressure and relative permeability was calculatedfrom the pore-throat distribution. (2) Cross-fault flow at eachelevation was calculated with Darcy’s law, the effective permeability,a fixed fault width, and capillary pressure as the pressure differencedriving petroleum flow. (3) Cumulative cross-fault petroleum flowwas calculated from the elevation of the threshold pressure to anelevation of interest.
1. Absolute water permeability (in millidarcys) was estimatedfrom an empirical power law relationship between absolute perme-ability (K a) and threshold pore-throat radius (r t, in microns):
ka ¼ 0:46 � r2:2t ð4Þ
The exponent describing this relationship is approximately 2,consistent with the Kozeny-Carmen relationship. Empirical perme-ability of small threshold pore throats is lower than predicted by thismodel. This is caused by increasing tortuosity with decreasing thresh-old radius. This effect is also responsible for a power law slope greaterthan 2. Except for using an exponent greater than 2, there is nocorrection for tortuosity in the model. Absolute permeability issomewhat overestimated for samples with very small throat radii.This overestimates the significance of hydraulic-resistance sealing.
Relative petroleum permeability (K ro) was calculated from theeffective water saturation (Jennings, 1987):
Kro ¼ ð1 � Sw* Þð1 � Sw*ð2þlÞ=lÞ: ð5Þ
Lambda, l, is the capillary curve shape factor (Jennings, 1987). Thefractional effective water saturation, Sw* , is normally calculated fromempirical fractionalwater saturation (S w) using the relationship S w* =(S w � S i)/(1 � S i). This study used model capillary-pressurecurves in which capillary pressure is assumed to have a power lawrelation to S w* . It was also assumed that S w* = S w. Effective watersaturation was directly calculated from the model power lawrelationship between capillary pressure and fractional effective sat-uration (modified from Jennings, 1987):
Sw* ¼Pc
ffiffiffiffika
f
q
0:07g cos �
0@
1A
�l
ð6Þ
where P c is the capillary pressure for the system of interest, F is thefractional porosity, g is the surface tension, and u is the angle ofwettability. The shape factor, l, was varied between 1 and 2 to in-clude most unimodal pore-throat distributions. Capillary pressureis calculated from the height above the free-water level and thecapillary-pressure gradient. Table 3 summarizes parameters used inthe four models shown in Figure 7.
2. Cross-fault volumetric flow per unit area of fault at each heightwas calculated from the effective permeability (k ak ro), fault thickness(T ), and fluid viscosity (A) using Darcy’s law with constant assumedrelative permeability across the fault thickness. The pressure-dropdriving flow is the difference between the capillary pressure andthe fault-fill, threshold capillary pressure. Flow per unit area (Q) is
Q ¼ kakro
mPc � Ptf
Tð7Þ
Equation 7 is an approximation. First, relative permeability isassumed constant across the fault fill, but it actually drops across thefault as capillary pressure decreases.Thisassumptionoverestimates flow.The pressure-difference driving flow is probably underestimated byequation 7. The pressure change is the maximum possible pressurechange across the fault fill, because petroleum pressure less than thethreshold pressure is not possible under steady flow on the injectioncurve. The cross-fault pressure drop does not include the effects of thecapillary threshold pressure at the discharge face of the fault fill norchanges in relative permeability resulting from such changes. Thefault-fill, threshold capillary pressure exceeds that of the reservoirinto which petroleum flows; thus, the steep capillary-pressure gradi-ent partially offsets the lower relative permeability near this interface.
3. Transmissivity (volumetric flow per unit length of fault mea-sured transverse to the flow) was calculated by numerically integratingthe volumetric flow from the threshold elevation to the elevation ofinterest. Volumetric flow per unit area increases with height becauserelative permeability and capillary pressure increase upward.
394 Capillary Effects on Fault-Fill Sealing
Correlating Transmissivity with Leakage Rate
Transmissivity multiplied by the fault length gives the total leakage.Total leakage can be estimated from the trap size and the age of atrap. It is assumed that half of the petroleum will remain as residualsaturation. Accumulation life is modeled as half the accumulationpetroleum volume divided by the leakage rate. This simplisticapproach underestimates accumulation age for a given initialtransmissivity because leakage rate will decrease as the accumulationis leaked due to decreasing capillary pressure.
To correlate fault transmissivity to trap leakage rates, trap-and-fault geometry must be assumed. For this study, traps are assumed tobe truncated pyramids with the top area one-quarter the area of thebase (Figure 7D). One side of the pyramid is a leaking fault. Assumedheights of the truncated pyramid are one-tenth and one-hundredththe length of the base of the pyramid. Petroleum was assumed tooccupy 20% of the total trap volume (equals 20% porosity with 100%oil saturation). Accumulation size was varied from 0.1 million to 10billion m3. Leakage rate is calculated by multiplying the fault trans-missivity by fault length, approximated by 0.6 times the pyramid base(because most leakage occurs in the upper part of the fault).
The magnitude of geologically significant leakage depends on thelifetime of the accumulation. A model accumulation life of 3 m.y. waschosen for estimating leakage rate and transmissivity. This is about anorder of magnitude less than reported median petroleum accumulationage (Macgregor, 1996); thus, median transmissivity is about an order ofmagnitude lower than those estimated here. The lower age and highertransmissivity were chosen for the following reasons. First, the assumedlinear leakage rate underestimates accumulation age. Transmissivityestimates might err due to assumed trap-and-fault geometry; thus, ahigh threshold is likely to include geometries more favorable for leakage.Finally, the short accumulation age encompasses just about all eco-nomic accumulations, not just the median case.
REFERENCES CITED
Bjorkum, P. A., O. Walderhaug, and P. H. Nadeau, 1998, Physicalconstraints on hydrocarbon leakage and trapping revisited:Petroleum Geoscience, v. 43, p. 237–239.
Colonna, J., F. Brissaun, and J. L. Millet, 1972, Evolution ofcapillarity and relative permeability hysteresis: Society ofPetroleum Engineers Journal, v. 12 (February), p. 28–38.
Fisher, Q. J., S. D. Harris, E. McAllister, R. J. Knipe, and A. J.Bolton, 2001, Hydrocarbon flow across faults by capillaryleakage revisited: Marine and Petroleum Geology, v. 187,p. 251–257.
Fulljames, J. R., L. J. J. Zijerveld, and R. C. M. W. Franssen, 1997,Fault seal processes: systematic analysis of fault seals overgeological and production time scales, in P. Moller-Petersenand A. G. Koestler, eds., Hydrocarbon seals: importance forexploration and production: Norwegian Petroleum SocietySpecial Publication 7, p. 51–79.
Grauls, D., F. Pascaud, and T. Rives, 2000, Fault seal quantitativeassessment in hydrocarbon-compartmentalized structure usingfluid pressure data (abs.), in A. G. Koestler and R. Hunsdale,eds., Hydrocarbon seal quantification: Stavanger, Norway,16–18 October 2000, Norwegian Petroleum Society ExtendedAbstracts, p. 63–71.
Harris, S. D., L. Elliott, and R. J. Knipe, 1999, The pulsed migrationof hydrocarbons across inactive faults: Hydrology and EarthSystem Sciences, v. 3, p. 151–175.
Heum, O. R., 1996, A fluid dynamic classification of hydrocarbonentrapment: Petroleum Geoscience, v. 2, p. 145–158.
Hubbert, M. K., 1953, Entrapment of petroleum under hydro-dynamic conditions: AAPG Bulletin, v. 37, p. 1954–2026.
Ibrahim, M. A., M. R. Tek, and D. L. Katz, 1970, Threshold pressurein gas storage: Arlington, Virginia, American Gas Association,Pipeline Research Committee Project Report 26–47, 309 p.
Ingram, G. M., J. L. Urai, and M. A. Naylor, 1997, Sealing processes andtop seal assessment, in P. Moller-Petersen and A. G. Koestler, eds.,Hydrocarbon seals: importance for exploration and production:Norwegian Petroleum Society Special Publication 7, p. 165–174.
Jennings, J. B., 1987, Capillary pressure techniques: application toexploration and development geology: AAPG Bulletin, v. 71,p. 1196–1209.
Katz, A., and A. H. Thompson, 1987, Prediction of rock electricalconductivity from mercury injection measurements: Journal ofGeophysical Research, v. 92, p. 599–607.
Knipe, R. J., G. Jones, and Q. J. Fisher, 1998, Faulting, fault sealingand fluid flow in hydrocarbon reservoirs: an introduction, in G.Jones, Q. Fisher, and R. Knipe, eds., Faulting, fault sealing andfluid flow in hydrocarbon reservoirs: Geological Society(London) Special Publication 147, p. vii –xxi.
Macgregor, D. S., 1996, Factors controlling the destruction andpreservation of giant light oil fields: Petroleum Geoscience,v. 2, p. 197–217.
Morrow, N. R., 1971, The retention of connate water inhydrocarbon reservoirs— part 1: a review of basic principles:Journal of Canadian Petroleum Technology, v. 10, no. 1(January–March), p. 38–55.
Myers, J., 1968, Differential pressures— a trapping mechanism inGulf Coast oil and gas fields: Transactions of the Gulf CoastAssociation of Geological Societies, v. 18, p. 56–80.
Neuzil, C. E., 1994, How permeable are clays and shales?: WaterResources Research, v. 30, p. 145–150.
Nunn, J. A., and P. Meulbroek, 2002, Kilometer-scale upwardmigration of hydrocarbon in geopressured sediments bybuoyancy-driven propagation of methane-filled fractures:AAPG Bulletin, v. 86, p. 907–918.
O’Connor, S. J., 2000, Hydrocarbon-water interfacial tension valuesat reservoir conditions: inconsistencies in the technicalliterature and the impact on maximum oil and gas columnheight calculations: AAPG Bulletin, v. 84, p. 1537–1541.
Petroleum Research Corporation, 1959, Reservoir pinchouts—sieves or seals?: Denver, Colorado, Petroleum ResearchCorporation Report A-5, September 1, 1959, 58 p.
Sales, J. K., 1993, Closure vs. seal capacity— a fundamental controlon the distribution of oil and gas, in A. Dore, J. H. Augustson,C. Hermanrud, D. J. Stewart, Ø. Sylta, eds., Basin modelling:advances and applications: Norwegian Petroleum SocietySpecial Publication 3, p. 399–414.
Schowalter, T. T., 1979, Mechanics of secondary hydrocarbonmigration and entrapment: AAPG Bulletin, v. 63, p. 723–760.
Smith, D. A., 1966, Theoretical considerations of sealing and non-sealing faults: AAPG Bulletin, v. 50, p. 363–374.
Smith, D. A., 1980, Sealing and nonsealing faults in Louisiana GulfCoast salt basin: AAPG Bulletin, v. 64, p. 145–172.
Wardlaw, N. C., and R. P. Taylor, 1976, Mercury capillary pressurecurves and the interpretation of pore structure and capillarybehavior in reservoir rocks: Bulletin of Canadian PetroleumGeology, v. 24, p. 225–262.
Watts, N. L., 1987, Theoretical aspects of cap-rock and fault sealsfor single- and two-phase hydrocarbon columns: Marine andPetroleum Geology, v. 4, p. 275–307.
Yielding, G., B. Freeman, and D. T. Needham, 1997, Quantitativefault seal prediction: AAPG Bulletin, v. 81, p. 897–917.
Brown 395