30 Chapter two ( new study in overburden pressure )

2.1) PRESSURE (P), OVERBURDEN (Sv), AND MINIMUM HORIZONTAL STRESS (Shmin) IN
EUGENE ISLAND BLOCK 330, OFFSHORE GULF OF MEXICO :

Benefits: 1. There are few detailed published analyses examining in situ stress data
from both sands and shales. Results from this analysis may yield insight into:

a) optimizing drilling programs (i.e. casing points, mud programs);

b) minimizing fracture completion costs (through an understanding of expected

stress state in sands and shales);

c) predicting seal integrity in exploration settings.

2. Overburden gradients based on wireline density logs are significantly less (~0.93
psi/ft) than the commonly-assumed 1 psi/ft. We recommend calculating it for each
area of study. 3. Precisely-recorded leak-off and fracture completion test data may
be used to characterize minimum horizontal stress and pore pressure. 4. The data
presented in this report can be used to define a field-specific (EI-330) fracture
gradient model.

2.2 Technical Perspective:

Fracture completions have become popular in poorly consolidated sediments in order

to maximize borehole stability and production rates. Few fields contain extensive
fracture completions and thus few data compilations exist which constrain pressure
(Pp), overburden stress (Sv), and minimum principal stress (Shmin). The EI-330 field,
the subject of this study, contains 22 fracture completions to date. Careful
characterization of this dataset has the potential to: 1) Develop a stress model to
31 Chapter two ( new study in overburden pressure )
reduce the cost of fracture completions (by eliminating the need for a pre-frac
stress measurement); 2) better constrain the fracture gradient to assure borehole
stability and minimize drilling costs; and 3) risk fault seal ahead of the drill bit
through an understanding of the least principal stress behavior.

2.3 Technical Approach:

We characterize minimum principal stress (Shmin) and pore pressure (Pp) within
reservoir sands (based on fracture completion data) and shales (based on leak-off
data) within the Eugene Island South Addition (Figure 1). The availability of minifrac
data provides an important advantage over data used in previous studies as accurate
least principal stress and pore pressure measurements are available from within the
same sand intervals. Leak-off test data, in contrast, are relatively abundant and have
frequently been used in attempts to make pore pressure and least horizontal stress
predictions. Given the two different types of data (LOTs and minifracs) available to us,
we try to assess how these can be characterized and compared with each other, how
they fit previously published fracture gradient models, how they vary with pore
pressure and geologic structures, and how they might influence hydrocarbon
migration and accumulation. We evaluated pore pressure (Pp), minimum principal
stress (Shmin), and overburden stress (Sv) at each minimum principal stress
measurement, and then assessed the errors of these estimates. Next, we
characterized relationships between Pp, Sv, and Shmin, and compared these results
to previous work. We also compared stress results for sands to those for shales.
Lastly, we developed a preliminary analysis of stress evolution during production.
32 Chapter two ( new study in overburden pressure )
2.4 Project Implications:

1. Examination of Pp, Shmin, and Sv in one location reveals that regional empirical
fracture gradients cannot accurately predict local fracture gradients. Fracture
gradient models should be locally derived where possible.

2. For a given pore pressure, closure stresses (Shmin) are slightly greater in shales
than in sands which should lead to some level of fracture containment.

3. The large scatter in pressure and stress data indicates that structure,
stratigraphy, and rock properties lead to large spatial variation in stress behavior.

2.5 Introduction in the project :

Analysis of fracture completion reports and leak-off tests (LOTs) from different wells
in the South Eugene Island 330 field (EI-330) demonstrates strong variation in pore
pressure (Pp) and minimum horizontal stress (S3 = Shmin). Shmin values within
sands and shales range from half of the overburden stress (Sv) to roughly 90% of
Sv. Pore pressures vary from sub-hydrostatic to 90% of Sv. Despite these
variations, the data can be characterized in the following manner. An effective stress
ratio (K=(Shmin-Pp)/(Sv-Pp)) of approximately 1/3 acts as a lower bound for the
observed data. Most of the data record a stress ratio greater than one third which
implies that many areas have lower differential stresses than are implied by
frictional faulting theory. Sands show a lower mean value for K (0.54) than shales
(0.70), which implies higher differential stresses in the sands than in the shales.
Analysis of these data with respect to individual reservoir sands reveals that there
33 Chapter two ( new study in overburden pressure )
may be an important relation between differential stress behavior and age,
composition, burial history, and spatial position. Analysis of poroelastic reservoir
behavior in response to fluid production suggests that there are significant changes
in stress that accompany production, but it is not supported by the limited data
available.

2. 6 Work Performed :

Geologic overview The data used in this study were acquired in the Eugene Island 330
(EI-330) field, a PlioPleistocene hydrocarbon reservoir contained within a salt-
withdrawal minibasin. The geologic evolution of this basin is described by Alexander
and Flemings (1995) and the discovery and development of the EI-330 field is
described by Holland et al. (1992). The Eugene Island minibasin is bounded to the
north by a regional (down to the south) fault system and to the south by an antithetic
fault system (Figure 2.1).
34 Chapter two ( new study in overburden pressure )

Hydrocarbons are produced in over 25 different sands which are themselves

segmented into at least 100 structurally or stratigraphically distinct reservoirs. The
major producers include the GA through OI sandstone (Figures 2.2). There is a
characteristic increase in overpressure with depth that is stratigraphically
controlled. The GA and shallower sands are near hydrostatic; the JD through LF are
moderately overpressured; and the OI and deeper sands reach near lithostatic
pressures (Figure 3A and B) ,Because overpressures are closely correlated to
stratigraphy, offset along normal faults in the minibasin results in abrupt lateral
contrasts in fluid pressure. Analysis of the normal fault system with respect to the
pore pressure distribution shows that isobars are highly discontinuous across fault
segments suggesting the compartmentalization of individual reservoir pockets
(Gordon and Flemings, in review).
35 Chapter two ( new study in overburden pressure )

Figure2.2 Dip-line well log cross section (marked C-C' in figure 1) in the South Eugene Island field
showing the minibasin and footwall separated by normal growth fault system. Sand intervals are shown
in gray; dashed lines are the corresponding flooding surfaces. Displayed well logs are either gamma ray
or spontaneous potential logs (left) and resistivity logs (right) (from Alexander and Flemings, 1995).
36 Chapter two ( new study in overburden pressure )

2.7 ) Fracture completion reports and leak-off test data :

Twenty-two fracture completions and nineteen leak-off test measurements from
production wells within the EI-330 field (see basemap, Figure 2.1) provided pore
pressure and least principal stress data used in this study. In the case of the
fracture completions, detailed reports allowed us to verify and extract accurate
pressure values to assess both least principal stresses and pore pressures. We
considered the reported fracture closure pressure , to be closest to the least
principal stress. The FCP is calculated from the break in slope of the pressure-time
curve after the well has been shutin and the pressure has bled off to the critical
point where it can no longer sustain the least principal stress to keep the
minifracture open ,At this stage the fracture has propagated far enough into the
formation that fracture initiation, near wellbore and fluid-related friction effects
during pumping have dissipated, and the pressure recorded should more or less
reflect the least principal stress in the reservoir. At least two closure pressure
values were generally obtained per test, using G-function (Nolte, 1979) and square
root of time plots. In cases where a step-rate test was also run, two more values for
Shmin were available. We used the minimum reported closure pressure as the lower
bound on Shmin and the highest reported closure pressure as the upper bound. The
average Shmin was the arithmetic mean of all of the closure pressure values and the
errors were calculated as the difference between mean value and upper/lower
bound value. Pore pressure values were usually obtained at the latest stages of each
minifracture test, when the well is shut-in, the fracture is closed, and the pressure
37 Chapter two ( new study in overburden pressure )
asymptotically reaches the pore pressure. Again, at least two values were available,
allowing us to define upper and lower bounds for subsequent uncertainty calculations

Information about the leak-off tests, by contrast, is not nearly as complete. Although
a LOT is run every time a section of well is cased, often the driller concludes the test
before an actual leak-off occurs. We refer to these tests as Formation Integrity Tests
(FITs) and Bottomhole
Figure2.4 disregarded these
pressure dataflow
and pump in this study,
rate vs. time inbecause of the uncertainty
a typical minifracture of the
test (Gaarenstroom et al., 1993).

data. We only considered tests in which fluid was lost into the hydraulically fractured
formation. In these cases, two pressure values were generally reported: (1) the
pressure at which the formation had broken down and a fracture was propagated; (2)
the bleed-off pressure after shut-in and time dependent pressure decay. Shut-in
pressures which are commonly used to approximate the least principal stress were
not available to us. Hence, we decided to use the maximum pressure, which we
believe reflects the fracture propagation pressure. Although this pressure is
different from the real Shmin value, it is sufficiently close to be used as an upper
38 Chapter two ( new study in overburden pressure )
bound value to Shmin (Figure 2.4). The bleed-off value was used as a lower bound, but
involves larger uncertainties because it was somewhat arbitrarily chosen after shut-
in. Also, fluid loss due to diffusion is a major concern, especially in shallow LOTs. This
would contribute to further error. Drilling engineers are aware of this problem and
are working to adjust pumping rates in areas where this has been shown to be a
problem (personal communication, L.J. McClure, Pennzoil, 1996)

2.8 Pore pressure calculations

for LOT Because of their drastically varying permeability and compaction
characteristics, we consider sand and shale LOT data separately (Tables A2 and A3).
Lithologic well log data (i.e. gamma ray, spontaneous potential) helped to distinguish
between them. For LOTs conducted in permeable reservoir sand units, mud weight
data provide an upper bound value for pore pressure because wells in this region are
drilled overbalanced with respect to pore pressures in sands to prevent well kicks
and blowouts. Pressure surveys within neighboring sand units indicate that on
average mud weight overbalances pore pressure by approximately 10%. We used this
mean difference to calculate a lower bound for pore pressure in sands by simply
deducting 10% from the reported mud weight. Pore pressures in shales may be
greater than in nearby sands because the shales have much lower permeabilities and
cannot release fluids at sufficient rates during compaction. Since it is nearly
impossible to measure pore pressure in shales because of the extremely low
permeability, we estimate Pp using mud weight data and porosity-effective stress
analysis. We employed a technique used by Hart et al. (1995) to calculate Pp in shales
from porosity. Porosity (f) was calculated from wireline sonic logs (Issler, 1992), and
b (compressibility) and f0 (reference porosity) are empirical constants. This method
calculates overpressure generated by compaction disequilibrium, but may
underestimate actual pore pressure because it does not account for late-stage
pressure generation mechanisms such as smectite dehydration (Hart et al., 1995)
Therefore, we use porosityderived pressures as a lower bound for Pp in
overpressured shales, and, for lack of any more accurate data, we use mud weight-
39 Chapter two ( new study in overburden pressure )
derived pressure as an upper-bound for Pp in overpressured shales. LOTs at shallow
depths were considered to be within the hydrostatic zone. The volumetric fraction of
shale at shallow depths in the EI-330 field is not larger than 30% implying only small
shale lenses amongst much larger sand bodies. Therefore, these small lenses are
assumed to drain hydrostatically with the sands despite the lithologic differences
between them. Since the porosity-effective stress methodology is generally believed
to give reasonable estimates for hydrostatic pore pressures (Hart et al., 1995), we
used it to obtain the lower bound value for Pp. For the upper bound estimate we
chose the mud weight equivalent pressure which is expected to overbalance the
pressures in the shales under hydrostatic conditions.
Uncertainties were calculated by estimating absolute errors for each parameter.
This is the most conservative approach since errors are cumulative when
considering quantities such as effective stresses or effective stress ratios.
2.9 Overburden stress :
A necessary step in this analysis is accurate calculation of the overburden stress.
Since we expect undercompaction effects at depth resulting in overburden gradients
that are considerably lower than the commonly assumed gradient of 1 psi/ft (22.65
MPa/km), the overburden stress was calculated by integrating wireline density logs
over depth. Four wells were chosen for this analysis based on hole condition and
quality of available well log data - primarily density, sonic, resistivity, and caliper
logs. These wells are located in different fault blocks of the growth fault system and
would allow us to not only see variations of overburden gradients with depth but also
see how they vary laterally. In order to obtain reasonable and continuous density
values with depth, caliper logs helped to identify washed-out depth intervals for
which the density log was replaced with interpolated values. The weight of the water
column (75.6 m deep in this area) was also taken into account. The continuous
density profile was integrated, multiplied by the gravitational constant, and then
divided by depth to yield the overburden gradient. Figure 2.4 shows the differences
between the results of this approach and the constant (1 psi/ft) overburden gradient
40 Chapter two ( new study in overburden pressure )
assumption. Obviously, the latter method grossly overestimates the true overburden
values.

2.10 Analysis :
Figure2.51Overburden gradients for four wells used to calculate the overburden stress at fracture completion and LOT

Stress and pore pressure variations versus depth The values for pore pressure and
depths in the EI-330 field. Note for comparison the often-assumed gradient of 1 psi/ft (22.65 MPa/km).

minimum horizontal stress (Table A1) are shown in composite figures of pressure and
stress vs. depth for sands (Figures 6A, 6B) and shales (Figures 7A, 7B). Large
variations in pressure and stress magnitudes are apparent at all depths. This results
from the fact that despite their spatial proximity (within or near the same minibasin),
different individual reservoirs have varying pressure and stress states. In general,
the least principal stress varies from 65% to 100% of lithostatic (Sv) and pore
pressures range from sub-hydrostatic to 90% of Sv. As observed in previous
compilations of such data (e.g. Anderson et al., 1973; Althaus, 1977; Breckels and van
Eekelen, 1981) both pore pressures and least principal stresses (S3 = Shmin)
approach the overburden stress with depth. This simultaneously decreases
differential and effective stresses with depth. Notice the large uncertainties in pore
41 Chapter two ( new study in overburden pressure )
pressures for overpressured shales (Figure 7A) based on the fact that mud weight
data poorly constrain Pp of shales under these conditions.

Sub-hydrostatic pore pressure conditions near 6900 ft (2100m) and 7700 ft

(2350m) depth within reservoir sands (Figure 6A) reflect pressure drawdown due to
hydrocarbon production.
Figure 2.6 Average Pressure
pore pressure (A) and surveys available
minimum principal stressto us (B)
Shmin show a pore
for sands pressure
as extracted from
declinesymbol).
for the KE-1 sand at 6200 ft (1900m) by 580-870 psi (4-6 MPa) (fault block
fracture completion and LOT reports. Error bars denote upper and lower bounds (mostly masked by
Triangles represent fracture completions and squares are LOTs. Lithostatic stress (dashed line)
dependent), for the LFsand at 6900 ft (2100m) by 1450 psi (10 MPa), and for the OI-
was derived from integrated density logs (331 #1 well) and hydrostatic pressure (solid line) was calculated
for brine (0.465 psi/ft = 10.53 MPa/km). The seawater depth in this area is 75.6m.
sand at 7700 ft (2350m) by 2600 psi (18 MPa). Similar numbers are probably
realistic for the IC-4 sand at 5600 ft (1700m), although no drawdown information for
this sand is available at present. This observation shows that pore pressure prior to
well completion was considerably higher in these reservoir sands and sub-
hydrostatic pressure conditions imply a transient pressure response due to
hydrocarbon production. We believe that this effect also has a strong influence on the
stress state and that Shmin at the time of fracture completion may not represent its
original value since stress changes due to fluid extraction are superimposed. The
issue of restoring the stress state to its original value will be discussed further
below.
42 Chapter two ( new study in overburden pressure )

Least principal effective stresses (shmin = Shmin - Pp) are shown in Figure 8,
separated again for sands and shales. Effective stresses show the same scatter as
noted for pore pressure and Shmin in Figure 7 varying from nearly 2900 psi (20
MPa) to almost zero (in which case the sand would be expected to be close to natural
hydraulic fracturing). However, the observation of larger effective stresses at
greater depth for sands (Figure 8A) is counterintuitive since the gradient of pore
pressure increase in the overpressured zones is generally believed to be greater
than the gradient of the least principal stress, hence, reducing effective stresses. It
is not surprising, therefore, that the sands with high effective stresses correlate
with sub-hydrostatic pore pressure conditions being attributed to hydrocarbon
production as described above. Restoration of original pore pressures and stresses
will show that effective stresses prior to fluid withdrawal were lower than these
show. Although obscured by large uncertainties, the decrease in effective stress in
overpressure at greater depth seems to be apparent for the shales (Figure 8B).
43 Chapter two ( new study in overburden pressure )

Figure 2.8 shows the effective stress ratio, K, versus depth, for sands (Figure 9A)
and for shales (Figure 9B). K is calculated by simply taking the ratio of the effective
least horizontal stress shmin to the effective overburden sv (where sv = Sv - Pp): K
is used to characterize the fracture gradient as it describes the behavior of least
principal stress and the overburden as a function of pore pressure. We observe on
average lower K values for sands (mean = 0.54, std. dev. = 0.25) compared to shales
(mean = 0.70, std. dev. = 0.33) further emphasizing a difference between lithologies
in terms of pore pressure and stress.

Figure 2.9 Effective stress ratio K vs. depth including error bars for upper and lower bounds. Dashed line shows the
effective stress ratio for Coulomb failure with a coefficient of friction of 0.6. Solid line represents a zero differential
stress state. (A) K for sands from fracture completion reports (triangles) and LOTs (squares). (B) K for overpressured
(triangles) and hydrostatically pressured (squares) shales.
44 Chapter two ( new study in overburden pressure )

A frictional coefficient (m) of 0.6 implies a K of 0.32 (solid line in Figure 10). Neither
category - sands or shales - appear to fall on this line despite the fact that sands
show a distinctly lower mean value for K. Additionally, the large error bars result
from the uncertainty analysis and the sensitivity of Equation 2 with respect to small
changes of its parameters. Previous studies (e.g. Brennan and Annis, 1984; Matthews
and Kelly, 1967; Pilkington, 1978), have derived empirical formulas for K as a function
of depth. Most of these assumed that differential stresses would eventually approach
zero (e.g. K would approach 1) in overpressure at great depth. In contrast to these
previous studies, our data do not show an explicit increase of K with depth. This may
be due to either large uncertainties in K for both sands and shales or because of the
fact that we primarily focused on data from within sands rather than shales.

2.11 Results :
We have compared stress measurements made through leak-off tests with those
made from fracture completions. Leak-off tests yield the same general trend in
stress data as the fracture completion data. This is a significant finding if it will allow
companies to spare the expense of the pre-frac test and use leak-off test data to
determine completion parameters. B. We have shown that sands maintain higher
stress differentials than shales implying a different stress state for sands than
shales. C. We have also demonstrated a correlation of stress behavior to
stratigraphic location. Different subsets of sands appear to have different stress
characteristics.
2.12 Recommendations for Industry :
A. Since our calculated overburden was significantly less than the commonly-
assumed 1 psi/ft, we would recommend calculating this stress for each area of
study. B. If leak-off test data were more precisely reported it could be used in the
45 Chapter two ( new study in overburden pressure )
place of prefrac tests to assess minimum horizontal stress and pore pressure prior
to fracture completion. C. It is recommended that these data be used to define a
field-specific (EI-330) fracture gradient model to be used in designing drilling
programs.