Journal of Petroleum Science and Engineering 96–97 (2012) 37–48
Contents lists available at SciVerse ScienceDirect
Journal of Petroleum Science and Engineering
journal homepage: www.elsevier.com/locate/petrol
In-situ stress and fault reactivation associated with LNG injection
in the Tiechanshan gas field, fold-thrust belt of Western Taiwan
Jih-hao Hung a,n, Jong-chang Wu b
a
b
Institute of Geophysics, National Central University, 300 Junda Road, Jhongli City 32001, Taiwan
Exploration and Production Business Division, CPC Corporation, Miaoli City 36001,Taiwan
a r t i c l e i n f o
abstract
Article history:
Received 14 May 2011
Accepted 6 August 2012
Available online 1 September 2012
The Tiechanshan gas field located in the fold-thrust belt of western Taiwan was depleted and converted
for underground storage of Liquefied Natural Gas (LNG) decades ago. Recently, CO2 sequestration has
been planned at shallower depths of this structure. We characterize the in-situ stresses from over 40
wells and assess the leakage potential through fault reactivation in response to LNG-injection increased
pore-pressure. The formation pore pressure (Pf), vertical stress (Sv), and minimum horizontal stress
(Shmin) are measured from repeated formation tests, density logs, and hydrofrac data including leak-off
tests and fluid injection. Formation pore pressures are hydrostatic above depths of 2 km, and increase
with local gradients of 14.02 and 21.26 MPa/km above and below 3.2 km, respectively. Extremely high
pore pressures (lp ¼ 0.8) are observed at depths below 3.8 km. Lower than normal pressures (average
9.47 MPa/km) are observed in the gas-bearing reservoir of the Talu A-sand. The gradient of Shmin is
17.46 MPa/km or equivalent to 0.74 of Sv ( 23.60 MPa/km). A detailed structure contour map of the
top of the A-sand, combined with the measured Shmin and Sv, show that the stress state in the
Tiechanshan field is predominantly strike-slip stress regime (SHmax 4SV 4 Shmin). An upper-bound value
of the maximum horizontal stress (SHmax) constrained by frictional limits and the coefficient of friction
(m ¼ 0.6) is about 27.36 MPa/km. Caliper logs from two wells show that the mean azimuth of preferred
orientation of borehole breakouts are in 0281N. Consequently, the maximum horizontal stress axis
tends 1181N, which is sub-parallel to the far-field plate-convergence direction. Geomechanical analyses
of the reactivation of pre-existing faults at the depths of the LNG reservoir sand indicate that all faults
are stable at the present stress state. Sensitivity analyses indicate that 5.9 MPa excess pore pressure
would be required to cause the optimal oriented f1 fault to reactivate. This corresponds to LNG column
heights of 760 m (density ¼ 790 kg/m3), whereas the height of structural closure of the A-sand does
not exceed 400 m. Therefore, it is unlikely that LNG injection will reactivate f1 fault.
& 2012 Elsevier B.V. All rights reserved.
Keywords:
subsurface LNG storage
in-situ stress
fault stability
Tiechanshan gas field
1. Introduction
Depleted oil and gas reservoirs hold great promise as sequestration sites. The fact that hydrocarbons have been held in them on a
geological time scale indicates the presence of effective trapping and
sealing mechanisms. One of the main issues with Liquefied Natural
Gas (LNG) or CO2 geo-sequestration is the potential risk of fluid
leakage. Fluid injection causes changes in the pore pressure and
stress field that could potentially alter the fluid retention properties
of the reservoir either by hydraulically fracturing the cap rock or by
triggering slip on pre-existing faults. From a geomechanical viewpoint, one of the key steps in the evaluation of any potential site
n
Corresponding author. Tel.: þ886 3 4275564; fax: þ886 3 4222044.
E-mail addresses: [email protected], [email protected] (J.-h. Hung).
0920-4105/$ - see front matter & 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.petrol.2012.08.002
considered for geologic sequestration is to determine the subsurface
in-situ stress and the strength of both reservoir and cap rocks.
Depleted gas fields in western Taiwan are potential candidates
for CO2 sequestration and LNG injection. The Tiechanshan gas field
(called TCS field hereafter) was an example of converting from a
depleted gas field into underground LNG storage 20 years ago, and a
pilot test of CO2 sequestration is planned at shallow depths. Owing
to limited downhole hydraulic fracturing measurements in the gas
fields of western Taiwan, previous studies (e.g., Tung et al., 1976;
Chen, 1981, 1982; Wang et al., 2001) have applied Eaton’s method
(Eaton, 1972) and Poisson’s ratio (n) derived from logging or
laboratory experiments to predict the minimum horizontal stress
(or fracture) gradient, upon which the regional fracture gradient (or
equivalent mud circulation weight) curves at various formations can
be established. Conversely, Huang (1991) and Hsieh et al. (2005)
utilized the empirical equations of Hubbert and Willis (1957), Coates
and Denoo (1981), respectively, to estimate the horizontal stresses
38
J. Hung, J. Wu / Journal of Petroleum Science and Engineering 96–97 (2012) 37–48
Fig. 1. Location of the Tiechanshan field and structural depth contour map at the top of Talu A-sand based on 3D seismic survey and well data. The numbers indicate well
locations. The A-sand ranges in depth between 2500 and 2900 m below sea level (mbsl) and marked contour line is 2750 m. Dashed lines with numbers are faults, while
arrows indicate the strike-slip movement of the fault blocks. Red curves are sealing fault segments that compartmentalize the field into four production blocks with varied
colors. A-A0 is the location of the cross-section in Fig. 2 (adapted from Tzeng et al., 2003) (For interpretation of the references to color in this figure legend, the reader is
referred to the web version of this article.).
J. Hung, J. Wu / Journal of Petroleum Science and Engineering 96–97 (2012) 37–48
in the TCS field. An additional coefficient, the Biot’s constant (a ¼1),
was applied in the latter calculation. Normal fault stress regime and
uniform horizontal stresses (a certain ratio to the vertical stress)
were assumed in above studies. Considering the significant difference in results obtained from the various models and the lack of
analysis of leakage potential, the volume of injected LNG was
maintained below the hydrocarbon column heights originally stored
in the reservoir to ensure the integrity of both the seal and reservoir
rocks. Nevertheless, provided that an accurate in-situ stresses and
rock strength can be determined, an optimum storage volume can
be better estimated to achieve a balance between economical
operation and risk potential. In this study we intend to quantify
the in-situ stress field using available data and geomechanical
analyses of the reactivation of pre-existing faults at reservoir depth.
Results from the study of the reservoir at the hydrocarbon depth can
be directly applied to evaluate the capacity of the shallow sandstones where CO2 is planned to be stored.
2. Geology of the Tiechanshan field
The TCS field is located in the north-central corner of the
western Foothills of Taiwan and is part of the Tertiary basin
developed along the passive Euroasian Plate margin (Fig. 1). The
TCS field evolved into a gentle anticline during Luzon arc-Asian
continental collision since Early Pliocene time and is composed of
two segments of anticline separated by a dextral strike fault (f6 in
Fig. 1). The pre-orogenic stratigraphy in the TCS field includes
clastic sequences of sandstone and shale deposits from Early
Oligocene (Wuchihshan Formation) to Early Pliocene (Kueichulin
Formation). The passive continental margin sequence was
conformably overlain by syn-orogenic Pliocene Chinshui Shale
to the Pleistocene Toukoshan Formation. The conglomerate in the
upper member of the Toukoshan Formation represents peak
arc-continental collision. The TCS anticline is a western verging,
gentle fault-bend fold bounded to the west by the Futoken thrust
39
fault dipping 50–601 to the east (see Figs. 1 and 2). The reservoir
(Talu A-sand) is cut by several faults including: (1) the main
longitudinal, high-angle reverse fault (f2) with a 60-761 dip
toward the west;(2) NW-SE (f3, f6, f9) to E-W (f1, f5) trending
high-angle oblique-slip faults; and (3) minor NE-SW tending
normal faults (f4, f7, f8). These faults are well constrained from
3-D seismic surveys and subsurface geology from more than 40
wells. The larger faults extend longer along strike and deeper into
the Oligocene volcanic basement judging from the cross section
and depth contour map of the top of the A-sand (Fig. 1). The
contour map also indicates that NW-SE trending faults are mainly
sinistral with a strike-slip component about an order magnitude
greater than normal slip. The TCS field is compartmentalized into
four blocks bounded by the f2 and f5 faults and portions of the f6
and f8 faults based on a comparison of historical pressure
depletion curves, stratigraphic throw and production interference
tests among wells across the faults (Tzeng et al., 2003). Partial
sealing of f6 and f8 faults (red segments in Fig. 2) is due to that
juxtaposition of the permeable sands across these faults occurs at
elevations below 2700 mbsl and above which cross-fault flow was
prohibited. The separate pressure decline curves at a late stage of
gas production from wells across these sealing segments also
substantiate the conclusion that juxtaposition of sandstones
against low permeable shale is the primary mechanism controlling hydrocarbon accumulation and fault sealing. None of the
mapped faults at the reservoir depth ranges are at the verge of
failure under current in-situ stress state with a coefficient of
m ¼0.6 (or described as being critically stressed). This will be
further discussed below.
The sedimentary strata encountered in the wells are composed
of continuous Upper Oligocene to recent sediments of cyclic,
shallow marine to shoreline clastic deposits of more than 5 km
thick. The Mid-Miocene Talu Shale is composed of mainly gray
shale and siltstone, within which the Talu sandstone of deltaic
deposits is the primary hydrocarbon producing interval in the
anticline. The Talu sandstone comprises four intervals of sands,
Fig. 2. Structural interpretation of A-A0 cross section shows that the Tiehchanshan anticline is a ramp anticline formed above the Futoken thrust (left). Secondary reverse
fault (f2) and normal faults (f5 and f8) are developed in the hanging-wall associated with bending of the thrust. Stratigraphy separation across the faults indicates
relatively small displacements along these faults. Stratigraphic sequence encountered in drilling is on the right.
40
J. Hung, J. Wu / Journal of Petroleum Science and Engineering 96–97 (2012) 37–48
Fig. 3. (a) Sandstones within the Talu Shale shown by SP (spontaneous potential) and LN (resistivity) log curves in the TCS-6 well; the uppermost A-sand is the main
producing zone. (b) Semi-log plots of sonic P-wave travel-time values with depth in shale intervals of well 39. Note the four zones of inferred constant gradients with
breaks at depths of 2.0, 3.2 and 3.8 km. Red dashed lines are the best linear fit of travel time curves in individual zones. (c) Measured wellbore pressure values from leak-off
test (LOT), cement injection (CMT), fluid intake and gas cut (Flow in), loss of circulation (Mud loss), and mud weights. Predicted pore-pressures from (a) following the Eaton
(1972) method are shown by the gray curves. Pore pressures in the Talu A-sand (apparent gradient 9.47 MPa/km) are lower than the hydrostatic pressure. The combined
results indicate that pore pressure increases with a local the gradient (thin solid line) within each compartment: hydrostatic pore pressure to the depth of 2 km, two
transition zones (apparent gradients of 14.24 MPa/km in zone 3 and 43.87 MPa/km in zone 4), and extremely high pore pressures (zone 4) at depths greater than
3.8 km. Pore pressure differences (Dp) in each compartment boundary are also estimated. Lines of hydrostatic pressure (blue) and measured minimum horizontal (green)
and vertical stresses (orange) are also shown (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).
among which the uppermost A-sand is the main producing and
LNG storage reservoir (Fig. 3a). The average porosity of A-sand is
18% (15–22% range), and the average permeability is 75 mD
(50–100 mD range). The Talu Shale and overlying Chinshui Shale
provides regional top seals of the A-sand throughout the northwestern Taiwan Neogene basin.
measure the in-situ stresses and fault strength. The validity of a
geomechanical model is dependent on the quality of the model
data. The parameters needed to define the in-situ stress are
described in the following section along with the data sources.
2.1. Geomechanical characterization
A total of 53 pore pressure values were recorded from the stable
flow during repeat formation tests (RFTs) in 34 wells at depths of
1680–4700 m below sea level (mbsl). The majority of the data are
measured at the depths of A-sand (Fig. 3c). Mud weights can also
In order to quantitatively assess which faults will reactivate if
pressurized fluids are injected in the reservoir, it is necessary to
2.2. Pore pressure
J. Hung, J. Wu / Journal of Petroleum Science and Engineering 96–97 (2012) 37–48
be used to estimate pore pressure in permeable formations as they
tend to take in drilling mud (or loss of circulation) if the mud
pressure is significantly in excess of the pore pressure or produce
fluids or gas (gas cut) into to the well if the converse is true. To
prevent boreholes from caving or breakouts, the pressure of mud
used in drilling is commonly higher than the pore pressure (i.e.,
over-balanced drilling). Therefore both accumulative mud pressure
and mud loss pressure provide upper-bound values of the pore
pressure, and pressures measured from RFTs and where there is
fluid (gas) flow are used to calculate local pore pressure gradients
(as shown in Fig. 3c).
Direct measurement of the pore pressure in impermeable
shale is very difficult, and several alternative methods have been
developed. For example, shale porosity (f) shows a characteristic
of exponential decay with depth due to compaction by increase of
effective overburden stress (sv). Athy (1930) presented a shale
normal compaction curve f–sv) showing decrease in porosity
with overburden stress. Qualitatively, the normal compaction
trend can be used to infer the presence of overpressure.
For example, in the semi-log plots of the P-wave travel time
within shale intervals with depth, an anomaly separates zones
with normal compaction with a higher value of linear gradient
above it and under-compacted zones with a lower gradient value
below (e.g., Fertl, 1976; Magara, 1978). The sonic P-wave travel
time in shale from well #39 in the TC field indicates that the onset
of overpressure occurs at 2 km, below which three overpressure
zones can be identified with varied gradients at depths of 3.2 km
and 3.8 km where the linear trends terminate (Fig. 3b). Other
geophysical data such as density, resistivity and porosity logs
measured from several wells down to 3 km depth also indicate
the existence of overpressure zones at depths below 2 km. Sonic
log data may be inverted for pore pressure using the empirical
method of Eaton (1972)
P f =z ¼ Sv =z Sv =zP p =z ðDt n =Dt o Þ3
ð1Þ
where Pf/z is the pore pressure gradient (MPa/km); Sv/z the
vertical stress gradient (MPa/km); Pp/z the hydrostatic pore
pressure gradient (MPa/km); Dtn: normal sonic log value (ms/ft),
and Dto: observed sonic log value (ms/ft). The predicted pore
pressure (Fig. 3c) is higher than the measured values in the
overpressure zones and deviates with depth. Therefore the trend,
rather than the absolute value, is used to constrain the boundaries
of each overpressure zone.
Combined direct measurement and indirect estimates from
the sonic log can better illustrate the variation of pore pressures
with depth. Above 2 km depth (Zone 1), the pore pressure is
essentially hydrostatic (saline water, 10.51 MPa/km). A linear
increase in pore pressures with a gradient of 14.24 MPa/km
may exist from 2 km to 3.2 km (Fig. 3b, Zone 2). Although
subdivisions can be made within Zone 2 based on minor
discontinuities in sonic log data and the derived pore pressure,
insufficient measurements and continuous mud weights make it
difficult to further compartmentalize pore pressures. Within Zone
2, measured formation pressures at intervals of A-sand (2400–
2900 mbsl) are lower than hydrostatic, which could be a
combined result of insufficient accumulation of gas volume and
lower density than the saline water within the ambient shale.
Pore pressure increases drastically with depth within Zone 3
(an apparent gradient of 21.26 MPa/km) and extremely high
pore pressures are observed at depths greater than 3.8 km. In
Zone 4, the ratio of excess pore pressure gradient to the vertical
stress or lithostatic pressure gradient (lp¼Pf/Sv) reaches a value
of 0.8. This value is greater than that obtained in previous
estimation (lp¼0.7) from drilling mud weight (Suppe and
Wittke, 1977). The abnormal pressure gradient may continue
undiminished below 5 km.
41
2.3. Minimum horizontal stress
The magnitude of the minimum horizontal stress (Shmin) can
be estimated from hydraulic fracturing tests (minifrac or microfrac) with cycles of fluid injection at desired intervals through
open holes or perforations after the casing is set. Injection of
cement slurry to fix poorly cemented intervals is a common
practice in the drilling of oil and gas wells. In this case, the
measured formation breakdown pressures can be viewed as a
type of hydraulic fracturing if the perforated section is less than a
few meters, and the injection leads to formation fracturing.
Alternatively, leak-off tests (LOTs) at the bottom of each casing
shoe, can be used to estimate the Shmin. The fluid leak-off
represents the point of fracture initiation and corresponds to
the departure from linearity on the pressure versus time plot. In
the TCS field, LOTs are performed at open-hole intervals of
5–30 m below the casing shoe with only one cycle rather than
extended leak-off tests (XLOTs). This, coupled with the pressure
and flow rate measured at the surface, makes leak-off pressure
results less reliable for estimating the magnitude of Shmin than the
results determined from minifrac tests. Nonetheless, it is widely
accepted that the lower bound to leak-off pressures in vertical
wells gives a reasonable estimate of the Shmin (Bell, 2003; Zoback
et al., 2003; Reynolds et al., 2006). Lastly, mud weights at the
depth where loss of circulation occurs can also provide an
estimate of Shmin. A correction of mud gelation pressure was
made during cement injection to obtain a value close to the true
formation strength (Gin and Huang, 1983).
A total of 25 combined measurements (4 from LOTs, 11 from
cement injection and 10 from mud loss) are collected from
vertical wells ranging from 302 to 4805 m. Most of the LOTs
conducted in the TCS field are shallower than 2 Km in depth,
whereas the cement injection data are obtained at greater depths
(Fig. 3c). The Shmin is estimated to be 17.46 MPa/km from the
best-fit linear trend of the data. Except in the shallower part, the
majority of the measured Shmin are less than vertical stress
measurements (Sv), with a ratio (Shmin/Sv) of being 0.74 at the
same depth.
2.4. Vertical stress
The magnitude of the vertical stress (Sv) is obtained by
integration of rock bulk densities of overlying mass taken from
density logs from the surface to the depth of interests. Care is
taken to remove measured outliers with density corrections (Dr)
greater than 70.05 g/cm3 and spurious measurements from
irregular borehole sections with caliper greater than 75% of the
bit size (Schlumberger, 2001; White and Hills, 2004). The average
linear vertical stress gradient obtained from the density logs of
three deep wells (38, 41, and Y-1) is about 23.60 Mpa/km (Fig. 3c).
2.5. Maximum horizontal stress
The maximum horizontal stress (SHmax) can only be estimated
from a calculation of critically resolved shear stress on existing
fault planes because neither XLOT nor borehole-wall image data
are available in the TCS field. Anderson’s (1951) frictional limits
theory states that the ratio of the principal maximum (s1) to
minimal effective stress (s3) cannot exceed the magnitude
required to cause faulting on an optimally oriented, pre-existing,
cohesionless fault plane (Sibson, 1974; Zoback and Healy, 1984).
On a critically stressed fault, the frictional limit to stress is given
by:
qffiffiffiffiffiffiffiffiffiffiffiffiffi
ðs1 =s3 Þ ¼ S1 P f = S3 Pf %ð 1þ m2 þ mÞ2
ð2Þ
42
J. Hung, J. Wu / Journal of Petroleum Science and Engineering 96–97 (2012) 37–48
where m is the coefficient of friction, Pf is the pore pressure, S1
and s1 are the total and effective maximum principal stresses,
respectively, and S3 and s3 are the total and effective minimum
principal stresses, respectively. Eq. (2) provides an upper bound
value of s1. The stress ratio is 3.1, assuming that m ¼ 0.6. In the
shallow part of the crust, the measured in-situ vertical and
horizontal stress axes are close to those of the principal stresses
(Angelier, 1984). Consequently, the calculated stresses (Sv, SHmax
and Shmin) can be related to the principal stresses (S1, S2 and S3) in
three (i.e., normal, strike-slip, and reverse) faulting environments.
The major slip components on the oblique-slip faults have
either a reverse or strike-slip motion from the mapped faults at
the top of the A-sand (Fig. 1). This indicates that the in-situ stress
state near the reservoir depths could cause either reverse or
strike-slip faulting. Measured stress magnitudes near the depth of
the A-sand further show that Sv is greater than Shmin (Fig. 3c).
Focal mechanisms determined from earthquakes (ML ¼2 5)
that occurred from 1973–1989 in west-central Taiwan (Yeh et al.,
1991) show that the maximal principal stress (S1) is nearly
horizontal, and the minimal principal stress axis (S3) has a higher
dipping angle than that of the S2, and displays a reverse-slip on
the NE-SW trending nodal planes or at left-lateral slip on the NWSE nodal planes. A total of 12 most recent seismic events
associated with the main shock (ML ¼5.2) occurred in 1992
around the TCS field (Fig. 4) is dominated by NW-SE trending,
high-angle dipping strike-slip faults at depths of 8–10 km.
Despite similar orientations between faults mapped in Talu
A-sand and at deeper level, the stress state may not be exactly
Fig. 4. (a) Earthquake sequences (within the large rectangle) occurred in 1992/04/22 around the Tiechanshan field. The largest earthquake magnitude is ML ¼ 5.2. A total of
12 focal spheres (in small rectangle) grouped in three colors is projected on NE-SW trending A-A0 (b) and E-W trending B-B0 (c) section. Note the focal depth ranges
between 8 and 10 km, and corresponding fault-plane solutions of colored earthquakes are shown in their relative locations and temporally sequential numbers. Majority of
focal spheres shows strike-slip motion, and the NW-SE trending fault plane is conformable with that of mapped faults in the Talu A-sand (data from Central Weather
Bureau, 2010).
J. Hung, J. Wu / Journal of Petroleum Science and Engineering 96–97 (2012) 37–48
the same because few normal faults are also observed at seismic
depths (Fig. 4c).
A coherent result among the various scales of stress data
mentioned above indicates that in-situ stresses near A-sand are
SHmax corresponding to S1, while Sv has a higher magnitude than
Shmin, being sub-parallel to S2 and S3. The best-fit linear regression
of the calculated upper bound value of SHmax (Eq. (2)) in a strikeslip stress regime and coefficient of friction, m ¼0.6, has a gradient
of 27.36 MPa/km (Fig. 5), and the stress ratio f ¼(S2–S3)/(S1–S3) is
about 0.84. In a reverse fault stress regime, the SHmax gradient is
43.87 MPa/km calculating using the same value of frictional
coefficient.
2.6. Stress orientation from caliper logs
Failure could occur around the borehole wall if the unequal
horizontal stresses reach the strength of the rock at a particular
depth. Borehole breakouts (BOs) are compressional shear failure
in the area of maximal compressive circumferential stress (at the
azimuth of Shmin), which results in spalling of the borehole wall
and enlargement of borehole diameter in the zone of failure. The
borehole geometry can be recorded from four-arm caliper tool.
BOs identified from two vertical wells, #41 (near the fold crest)
and Y-1 (southern part) show that the azimuths of SHmax is
predominantly oriented 1407101N and 1107101N, respectively,
with an average azimuth of 1251N (Fig. 6), following the criteria of
the World Stress Map Project (Reinecker et al., 2003) and previous
studies (Plumb and Hickman, 1985; Suppe, 1985; Hung et al.,
2009; Zoback, 2007).
43
3. Fault slip potential
LNG injection of into the subsurface Talu A-sand may result in
an increase in reservoir pore pressure, which could lead to brittle
failure of the rocks if the stress acting on the rocks exceeds their
strength (e.g., Hillis and Reynolds, 2003; Streit and Hillis, 2004;
Zoback, 2007). In the case of the A-sand, an increase in the
injection-induced local pore pressure in the vicinity of one of the
inactive bounding faults will reduce the effective normal stress
acting on the fault plane thereby reducing the strength of the fault.
Brittle failure will take place if the increase in pore pressure (DPf)
reduces the effective normal stress such that the Coulomb frictional criterion is met (Fig. 7). Optimally oriented faults with
respect to the current stress state are most at risk of reactivation
and require a smaller DPf to induce movement. To determine the
risk of leakage potential for all faults in the TCS field, structure
contours on the top of the A-sand based on seismic profiles and
subsurface drilling data are used to constrain the locations of
3-dimensional non-planar fault surfaces from 2500 to 4000 m.
Breaking up these boundary faults into small ( 100 m 100 m)
triangular elements of individual planar fault planes allow us to
calculate the shear and normal stress on each part of the fault, and
determine the pore pressure at which fault segment is expected
to slip.
Geomechanical calculation requires a regional pore pressure
profile and stress tensor, together with the fault information. Since
available data is insufficient to determine the stresses in individual
compartmentalized blocks, we can only define one stress tensor for
the entire field using one-dimensional model that varies with depth
TKS
CL
CS
YTP
SLF
KTS
SFC
TK
KYS
TL
A-Sand
PL
CHK
PLIN
MS
Fig. 5. Best-fit linear regression of SHmax calculated using frictional limit of the faults and coefficient of friction m ¼ 0.6 for strike-slip (red dashed line) and reverse fault
(blue dashed line) stress regimes. Stratigraphic column is on the right (For interpretation of the references to color in this figure legend, the reader is referred to the web
version of this article.).
44
J. Hung, J. Wu / Journal of Petroleum Science and Engineering 96–97 (2012) 37–48
Fig. 6. (a) Caliper logs of bit size, C1, and C2 for well YL1. (b) Locations of borehole breakouts determined from the borehole calipers. (c) Length-weighted SHmax orientation
in the Talu Shale showing a total of 108 m breakouts with a predominant azimuth of 1251. Dots indicate the poles to bedding, which shows average horizontal fold axis
(b-axis) and bedding.
Table 1
Principal stress gradients (shown in Fig. 3c) and orientations, pore pressure and coefficient of friction used in
geomechanical modeling of fault reactivation risk for
strike-slip fault stress regime.
Stress
Gradient (MPa/km)
Orientation (1N)
Sv
SHmax
Shmin
Pp
23.60
27.47
17.46
9.47
0.6
Vertical
0.125
0.035
Vertical
–
m
Fig. 7. Mohr circle representation of the stress state and reactivation of a fault
plane using the Mohr-Coulomb failure envelope. The black dot is the pole to the
fault plane. The horizontal distance between the fault plane and the failure
envelope (DPf) describes the required increase in pore pressure needed to cause
slipping on this cohesionless fault.
(shown in Fig. 3c). Each fault surface is modeled as a 3-dimensional
grid. The principal axis of the grid is parallel to the strike and dip of
the fault. The shear and effective normal stress acting on each grid of
the fault plane are calculated from previously defined stress tensor
and pore pressure profiles (summarized in Table 1) using the
method of Wiprut and Zoback (2002). The commercial software
package TrapTester developed by Badley Geoscience Ltd. (www.
badleys.co.uk) can also help to calculate and display the required
increase of pore pressure or pore pressure differences (DPf) to cause
slip along the fault surface.
Model results of DPf on each fault segment from 2500 to 4000 m
are represented by color-coded fault surfaces assuming faults are all
cohesionless with a uniform coefficient of friction m ¼0.6 (Fig. 8).
Fault slip potential is, in general, inversely proportional to the depth
if faults maintaining similar trend and dip. This is because the
gradient of maximum differential stress (SHmax Shmin ¼ 10 MPa/
km) is less that of Shmin (17 MPa/km; see Fig. 5), which drives the
Mohr circle away from the failure envelope as depth increases.
Among all faults, f1 fault is optimally orientated for reactivation
(Fig. 9) therefore requires the least amount of excessive pore
pressure to reactivate. At the average depth ( 2750 m) for the
reservoir interval, a pressure of approximately 11 MPa is required to
reactivate the f1 fault. This corresponds to an LNG column height of
approximately 1400 m (density¼ 790 kg/m3). The f1 fault is not at
risk of reactivation because the height of structure closure of the
A-sand is no more than 400 m. Consequently this fault will not be an
upward leakage pathway for the LNG. Nevertheless, f1 fault is nonsealing in horizontal migration of hydrocarbons due to the juxtaposition of permeable sandstones across the f1 fault.
LNG injection may cause changes in pore pressure that are
pervasive throughout the reservoir if not localized in the vicinity
J. Hung, J. Wu / Journal of Petroleum Science and Engineering 96–97 (2012) 37–48
of a fault. The magnitude of the total vertical stress (Sv) will not
change during injection, whereas horizontal stresses could change
in proportion to the magnitude of pore pressure changes
45
as follows:
DSHor ¼ a ð12nÞ=ð1nÞ DP p
ð3Þ
where SHor is the horizontal stress; a is the Biot’s coefficient and
n is the Poisson’s ratio. This equation is derived for an isotropic,
porous and elastic reservoir of infinite lateral extent. Segall and
Fitzgerald (1996) showed that this relationship is valid as long as
the ratio of the lateral extent to the thickness of the reservoir is
greater than 10:1. To evaluate the poroelastic effects on fault
stability DPf is estimated as in the previous analysis of the f1 fault
under the condition of DPp ¼5 MPa, a ¼1, and n ¼0.25 in the strikeslip fault stress regime. As shown in Fig. 10, poroelastic effects will
further decrease the amount of extra pressure needed to cause
slipping on the f1 fault, since the increase in pore pressure has the
same effect on both SHmax and Shmin. This means the effective
differential stress (sHmax shmin) does not change. At depth
2750 m, an approximately additional 2.6 MPa or total of 7.6 MPa
compared to 11 MPa of excess pressure (as obtained in the
previous calculation) would be required to cause the f1 fault to
reactivate.
3.1. Sensitivity and scenario analysis of parameters
Fig. 8. Perspective view of color-coded fault surfaces showing the pore pressure
values required for reactivation of T-field faults from 2500 to 4000 mbsl. Faults
trending E-W (f1 and f5) have the highest slip potential, whereas more excess pore
pressure is require to cause the NE-SW trending (f2, f7, and f8) faults to slip.
The risk of fault reactivation decreases with increase of depth.
To evaluate how the parameters (Table 1) including the
horizontal stress magnitudes, coefficient of friction, and pore
pressure affect the slip potential of the f1 fault, both sensitivity
53 (Mpa)
f6
f3
f4
Low risk
Cohesionless
faults
μ=0.6
f7
ΔPf
f8
f9
f5
f2
High risk
f1
11
σhmin
σv
σHmax
Effective Normal Stress (Mpa)
Fig. 9. (a) Polar stereographic projection of the increase of pore pressures required to cause slipping of all fault orientations (poles to fault planes) at average reservoir
depth ( 2750 m). The red lines in vertical faults with 301 to SHmax indicate two optimal orientations (951 and 1551) for reactivation. (b) Mohr diagram of the stress state
and risk of reactivation of cohesionless fault planes using the Coulomb failure envelope of m ¼ 0.6. Note f1 fault is the most susceptible to reactivation.
46
J. Hung, J. Wu / Journal of Petroleum Science and Engineering 96–97 (2012) 37–48
Fig. 10. Fault surface color-coded with DPf values indicating the slip potential of the f1 fault considering poroelastic effect. At depth of 2750 m, approximately 2.6 MPa
excess pore pressure would be required to cause the fault to slip.
SHmax
Shmin
Sv
Pp
fri. coe.
25
Table 2
Values of parameters (within one standard error) at depth of 2750 m used in the
scenario analysis.
20
ΔPf (MPa)
15
10
Parameter
Minimum
Mean
Maximum
SHmax
Sv
Shmin
Fault azimuth
Fault dip
73.67
64.63
48.08
761N
661
0.60
24.43
82.25
64.90
51.43
911N
761
0.72
26.04
90.83
65.17
54.77
1061N
861
0.85
27.66
m
Pore pressure
5
-15%
-10%
0
-5%
0%
5%
Percentage change of parameters
10%
15%
Fig. 11. Pore pressure (DPf) required to cause the f1 fault to slip as a function of
percentage changes of parameters. Shmin and SHmax have higher effects (with
greater slope) on the DPf values.
and scenario analyses of those parameters are performed. Similar
to previous analysis, an incremental change of 75% of the
parameter values at depth 2750 m is applied during calculating
the pore pressure differences (DPf) needed to cause the slipping of
the f1 fault. Two kinds of sensitivity analyses are applied. One is
to vary one parameter at a time whereas others remain fixed at
their average value, and the result is shown in Fig. 9. The other is
to vary both independent and dependent parameters together
such that SHmax value changes accordingly with the change of
either frictional coefficient or pore pressure. Both tests show
similar results except flipping the order of frictional coefficient
and pore pressure. That is, stresses (Shmin and SHmax) are the two
main parameters controlling the DPf values (Fig. 11).
Scenario analysis was carried out using over 10,000 Monte
Carlo simulations in strike-slip fault stress regime by random
sampling of all independent and dependent parameters and
varying one standard error value of all parameters estimated
from wells in the TCS field. Fault attitude is also added as one
parameter (Table 2). Fig. 12a shows the fault slip potential
probability as a function of reservoir pressure for variations of
indicated components of the stress tensor. From the scenario
tests, in 99% of the cases, DPf values of more than 12 MPa would
be required to cause slipping on the f1 fault. To account for the
uncertainties of fault geometry, we evaluate fault slip probability
as a function of variations in fault azimuth (Fig. 12b, left) and dip
(right). In 99% of the test scenarios, comparable DPf values of
above 15 MPa are required for both azimuth and dip. Similarly,
DPf values above 13 MPa and 15 MPa (not shown) are observed
for variations in frictional coefficient and pore pressure, respectively. For tests of random sampling of the above parameters, a
minimum DPf of 5.9 MPa (Fig. 12c) would be necessary to induce
slipping on the f1 fault. Thus even in a pessimistic risk scenario,
an LNG column of approximately 760 m in height (with a density
of 790 kg/m3) would be required to reach the lowest estimated
DPf value ( 5.9 MPa). Assuming that the injection well is located
in the vicinity of f1 fault and injectivity is permissible, the pore
pressure of 5.9 MPa will not be reached because the structural
closure of A-Sand controls the accumulation of LNG of 400 m in
height or a pressure of 3.1 MPa.
4. Discussion
The absolute values of DPf calculated in this study are subject
to large errors due to uncertainties in the data used for the
geomechanical model and limitations of the methodology.
In particular, the maximum horizontal stress and the fault
strength data are poorly constrained. For example, the maximum
horizontal stress probably has more uncertainty compared to
other parameters because it is calculated from Shmin and by
assuming that faults are cohesionless with frictional coefficient
m ¼0.6. As expected, fault reactivation risk for cohesionless or
low-friction faults will require lower values of SHmax and DPf than
for healed or strong faults. Further measurements of fault
strength and in-situ stresses must be undertaken to evaluate
CO2 storage in shallow reservoirs.
Injection of fluid into a reservoir will increase pore pressure and
reduce the effective normal stresses. This is the usual cause of fault
reactivation, as it drives the Mohr circle towards the failure
envelope. Geomechanical modeling shows that fault reactivation
in the TCS field is unlikely since most faults are non-optimally
J. Hung, J. Wu / Journal of Petroleum Science and Engineering 96–97 (2012) 37–48
47
Fig. 12. Fault slip potential probability for the strike-slip environment as a function of: (a) varying each component of the stress tensor; (b) variation of fault azimuth (left)
and dip (right), and (c) random sampling of all parameters.
48
J. Hung, J. Wu / Journal of Petroleum Science and Engineering 96–97 (2012) 37–48
oriented and require excess pore pressure greater than the hydrocarbon column pressure that the structure can accommodate. There
is no documented evidence of casing failure due to fault reactivation
or any reported leakage indicators from surface monitors deployed
in the TCS field. In addition, there has been no induced seismicity
associated with fluid injection over the past 20 years.
In the above analysis we considered the case where the buoyancy pressure or column height of the hydrocarbon in an anticlinal
reservoir bounded by faults is dynamically balanced by the pore
pressure difference on optimally oriented faults. Other dynamic
mechanisms such as capillary pressure and hydraulic fracturing
limit of the cap rock need to be taken into account. They would be
useful to evaluate the risk of leakage and provide other constraints
on the maximum hydrocarbon column that the reservoir could
contain before the column reaches the spill point of the anticline.
Hydrocarbon can flow through the cap rock if the buoyant pressure
of the gas and oil column exceeds the capillary entry pressure of the
cap rock. On the other hand, hydraulic fracturing of the overlying
units could occur if the fluid pressure in the reservoir exceeded the
magnitude of the least principal stress in the cap rock. In the TCS
field, the natural gas in the A-sand is not filled-to-spill, and the
amount of injected LNG never exceeding the original hydrocarbon
column heights; therefore, in the absence of other data, particularly
the petrophysical and mechanical data of the cap rock, it is difficult
to distinguish which of above mechanisms act to limit the amount
of hydrocarbon in the A-sand.
Pore-pressure/stress coupling (or poroelastic effect) is
assumed to have equal effects on horizontal stresses in axialsymmetrical media (Segall and Fitzgerald, 1996; Streit and Hillis,
2004). That is, the maximal differential stress (s1 s3) will
remain fixed before and after the change of fluid pressure in
strike-slip stress regime. Unfortunately, there is no measurement
of changes in SHor with respect to DPp during LNG injection.
Therefore, this hypothesis is speculative. Furthermore, our analyses of the fault stability consider only the mechanics of LNG
injection. Other long-term effects, such as geochemical weakening and changes in frictional coefficients with time due to
interaction between the hydrocarbon fluid and fault zone material, should be incorporated into future study.
5. Conclusion
We use petroleum exploration data to constrain the magnitude
of in-situ stresses down to 5 km in depth in the Tiechanshan field.
The stress regime is determined to be a strike-slip judging from both
slip components of the faults at the depth of Talu A-sand reservoir
and downhole stress measurements. Pore pressure is hydrostatic
above 2 km in depth, below which there are three distinct anomalous, compartmentalized high pore pressure zones with varied
gradients, as determined from repeated measurements of formation
pressure and the sonic logs. The minimum and maximum horizontal
stresses are the two main factors controlling fault reactivation in the
TCS filed. Even a conservative estimate of the DP value for the
optimally oriented f1 fault indicates that it is at low risk of
reactivation due to LNG injection given current stress tensor.
Acknowledgments
The work was financial supported by the National Science
Council (NSC), Taiwan, ROC, under the contracts of 98-ET-E-008002-ET and 99-ET-E-008-003-ET. The authors are grateful to
Taiwan Petroleum Corporation for permission to use the data
and publish the results. Badley Geoscience Ltd. kindly provided
TrapTester software.
References
Anderson, E.M., 1951. The Dynamics of Faulting and Dyke Formation with
Applications to Britain. Oliver and Boyd, Edinburgh.
Angelier, J., 1984. Tectonic analysis of fault slip data sets. J. Geophys. Res. 89,
5835–5848.
Athy, K.F., 1930. Density, porosity and compaction of sedimentary rocks. AAPG
Bull. 14, 1–24.
Bell, J.S., 2003. Practical methods for estimating in situ stresses for borehole
stability applications in sedimentary basins. J. Pet. Sci. Eng. 38, 111–119.
Central Weather Bureau, 2010. Seismicity catalog /http://www.cwb.gov.tw/eng/
index.htmS.
Chen, H.T., 1981. Study of formation fracture gradient in the sandstones of the
Chinshui and Chuhuangkeng area of western Taiwan. J. Pet. Eng. 22, 177–191,
in Chinese with English abstract.
Chen, H.T, 1982. Establishment and application of the of formation fracture
gradient curves in the north and southern Kutingkeng Formation of Taiwan.
J. Pet. Eng. 23, 62–74, in Chinese with English abstract.
Coates, G.R., Denoo, S.A., 1981. Mechanical properties program using borehole
analysis and Mohr circle. Proceedings of the SPWLA 22nd Annual Logging
Symposium, Mexico City, June 23–26, paper DD.
Eaton, B.A., 1972. Graphical method predicts geopressure worldwide. World Oil
182, 51–56.
Fertl, W.H., 1976. Abnormal Formation Pressures. Elsevier, New York 382 pp.
Gin, Y.S., Huang, Y.H., 1983. Using leak-off tests to estimate formation fracture
pressure. J. Pet. Eng. 24, 1–10, in Chinese with English abstract.
Hillis, R.R., Reynolds, S.D., 2003. In situ stress field, fault reactivation and seal
integrity in the Bight basin. South Australia Department of Primary Industries
and Resources. Report Book 2003/2, 4–21.
Hsieh, B.Z, Lin, Z.S., Fang, C.H, 2005. Stress distribution and wellbore stability
analysis in poroelastic gas storage reservoir. J. Pet. Eng. 41, 43–61, in Chinese
with English abstract.
Huang, R.H., 1991. Implementation of rock mechanics method in initial formation
fracture pressure prediction. J. Pet. Eng. 32, 173–178, in Chinese with English
abstract.
Hubbert, M.K., Willis, D.G., 1957. Mechanics of hydraulic fracturing. Pet. Trans.
AIME 210, 153–163.
Hung, J.H., Ma, K.F., Wu, Y.H., Wu, H.Y., Ito, H, Lin, W., Yeh, E.C., 2009. Structure
geology, physical properties, fault zone characteristics and stress state in
scientific drill holes of Taiwan Chelungpu fault drilling project. Tectonophysics
466, 307–321.
Magara, K., 1978. Compaction and Fluid Migration. Elsevier, New York 320 pp.
Plumb, R.A., Hickman, S.H., 1985. Stress-induced borehole enlargement: a
comparison between the four-arm dipmeter and the borehole televiewer in
the Auburn geothermal well. J. Geophys. Res. 90, 5513–5521.
Reinecker, J., Tingay, M., Müller, B., 2003. Borehole breakout analysis from fourarm caliper logs. World Stress Map Project, 1–5.
Reynolds, S.D., Mildren, S.D., Hillis, R.R., Meyer, J.J., 2006. Constraining stress
magnitudes using petroleum exploration data in the Cooper-Eromanga Basins,
Australia. Tectonophysics 415, 123–140.
Schlumberger, 2001. Platform Express SMP-5177, 15.
Segall, P., Fitzgerald, S.D., 1996. A note on induced stress changes in hydrocarbon
and geothermal reservoirs. Tectonophysics 289, 117–128.
Sibson, R.H., 1974. Frictional constraints on thrust, wrench and normal faults.
Nature 249, 542–544.
Streit, J.E., Hillis, R.R., 2004. Estimating fault stability and sustainable fluid
pressure in underground storage of CO2 in porous rocks. Energy 29,
1445–1456.
Suppe, J., 1985. Present-day stress directions in western Taiwan inferred from
borehole elongation. Pet. Geol. Taiwan 21, 1–12.
Suppe, J., Wittke, J.H., 1977. Abnormal pore-fluid pressures in relation to
stratigraphy and structure in the active fold-and-thrust belt of northwestern
Taiwan. Pet. Geol. Taiwan 14, 11–24.
Tung, S.H., Pang, W.H., Chen, H.T, 1976. Prediction and application of formation
fracture gradient. J. Pet. Eng. 17, 1–32, in Chinese with English abstract.
Tzeng, J.J., Wu, W.J., Chen, T.L., 2003. Study of fault sealing problems in the
structure of TCS. Petroleum Geology of Taiwan 36, 215–240.
Wang, W.L., Huang, R.H., Fang, C.H., Hsiuan, T.H., 2001. Prediction and application
of formation pore pressure and fracture gradient. J. Pet. Eng. 42, 21–32
in Chinese with English abstract.
White, A., Hills, R., 2004. In-situ stress field and fault reactivation in the Mutineer
and Exeter Fields, Australian North West Shelf. Explor. Geophys. 35, 217–223.
Wiprut, D., Zoback, M., 2002. Fault reactivation, leakage potential, and hydrocarbon
column heights in the northern North Sea. In: Koestler, A.G., Hunsdale, R. (Eds.),
Hydrocarbon Seal Quantification, 11. Elsevier Sciences, B.V. Amsterdam, NPF
Spec. Pub, pp. 203–219.
Yeh, Y.H., Lin, C.H., Barrier, E., Angelier, J., 1991. Stress tensor analysis in the Taiwan
area from focal mechanisms of earthquake. Tectonophysics 200, 267–280.
Zoback, M., Healy, J.H., 1984. Frictional faulting and in situ stress. Annal. Geophys.
2 (6), 689–698.
Zoback, M., Barton, C.A., Brudy, M., Castillom, D.A., Finkbeiner, T., Grollimund, B.R.,
Moos, D.B., Peska, P., Ward, C.D., Wiprut, D.J., 2003. Determination of stress
orientation and magnitude in deep wells. Int. J. Rock Mech. Min. Sci. 40,
1049–1076.
Zoback, M., 2007. Reservoir Geomechanics. Cambridge University Press.