You are on page 1of 7

OTC 17206

Fracture Pressure Prediction With Improved Poisson’s Ratio Estimation


E. Kozlov, Paradigm Geophysical Services; I. Garagash, United Inst. of the Earth’s Physics; J. Wang, N. Baransky,
D. Dopkin, and N. Ivanova, Paradigm Geophysical; and A. Lowrie, Consultant
Copyright 2005, Offshore Technology Conference
is the leak-off test (LOT) data and the mud weight profile
This paper was prepared for presentation at the 2005 Offshore Technology Conference held in established experimentally at neighbouring wells, so in the
Houston, TX, U.S.A., 2–5 May 2005.
areas with a dense network of wells, the common practice is
This paper was selected for presentation by an OTC Program Committee following review of
information contained in a proposal submitted by the author(s). Contents of the paper, as either to rely on the experimental data and disregard the
presented, have not been reviewed by the Offshore Technology Conference and are subject to Eaton’s formula-based predictions, or to calibrate the latter
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Offshore Technology Conference, its officers, or members. Papers presented at with account for the LOT data.
OTC are subject to publication review by Sponsor Society Committees of the Offshore
Technology Conference. Electronic reproduction, distribution, or storage of any part of this When implementing such a calibration, the estimates
paper for commercial purposes without the written consent of the Offshore Technology derived from formula (1) occur to be deviating from the LOT
Conference is prohibited. Permission to reproduce n i print is restricted to a proposal of not
more than 300 words; illustrations may not be copied. The proposal must contain conspicuous data more often than not, with the deviation increasing with
acknowledgment of where and by whom the paper was presented. Write Librarian, OTC, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435. depth, if for the ν values in (1), the laboratory or sonic log
measurements of the Poisson’s ratio are taken. Indeed, in
Abstract practically all kinds of wet rocks, such measurements show the
decrease of ν values with the differential pressure Plitho –
The commonly used model for fracture pressure determination Pfluid [2, 3, 4, 5]. And since the differential pressure is
makes use of the ratio of the horizontal effective stress and the increasing with depth as long as pore pressure gradient
vertical stress as a function of the Poisson’s ratio, the latter remains less than the overburden pressure gradient, these log-
being inferred from seismic velocities. It has been documented or laboratory-derived ν values are commonly decreasing with
that fracture pressure predicted in such a way deviates depth, as, e.g., in Figure 1. However, to make the formula (1)-
systematically from what is derived from LOT data, so the based predictions consistent with the LOT data and respective
predictions provide an unreliable basis for well planning. In
this paper, we analyze the reasons for the deviations and
present a new methodology to predict Poisson’s ratio and
fracture pressure using seismic , well log and laboratory data.
Examples confirm the superiority of the new methodology.

Introduction

As exploration for hydrocarbons is shifted to more and more


difficult environment (deep water, under salt, overpressure,
etc.), the drilling gets more costly, and drilling hazards are
more unwanted than ever. A pre-requisite of successful
drilling is careful planning of well trajectory, casing, and
drilling mud weight. Essential part of this planning consists in
prediction of rock properties, pore pressure and fracture
pressure profiles.
For such a prediction, a well established technology exists.
An essential part of the technology consists in calculation of
pore pressure, minimum horizontal stress and fracture pressure
from seismic velocities VP and VS and rock density ρ . The Fig. 1. Sonic log P-wave (a) and S-wave (b) velocities and Poisson’s
simplest, yet widely used approach to conversion of these data ratio curve (c) calculated from the velocities (a) and (b) (reproduced
into the sought-for parameters is the Eaton’s [1] formula from paper [ 6 ]), solid lines; Eaton’s [ 1 ] curve ν = a + b·loge z (dashed
ν line). Note inconsistency between Poisson’s ratio dependencies
Pfrac = Pfluid + ( Plitho − Pfluid ) . (1) represented by solid and dashed lines.
1 −ν
Here Pfrac is the fracture pressure, Pfluid is the pore fluid mud weight profiles, the ν values should increase with depth.
pressure, Plitho is the overburden stress, and ν is the Poisson’s E.g., for a particular dataset, Eaton [1] have shown that the
ratio. An alternative source of information on fracture pressure consistence is observed if ν = a + b·loge z with a = - 0.0673
2 OTC 17206

and b = 0.0645, which makes ν increasing with depth z repose angle. Obviously, the equation (3) cannot be universal
rather fast, especially at small depths, Figure 1. [10]: the effect of fluid saturation strongly depends on whether
In undrilled fields, there is no LOT or mud weight profile there is a possibility for free fluid flow (drained case). The
data to calibrate the estimates of fracture pressure obtained equation (3) is applicable to the drained state of the rock, when
using formula (1)- based technology, so it is of practical
interest to find out what’s wrong with the routine technology,
and how one can get a more reliable prediction of fracture
pressure. With the aim to answer these questions, we will
consider first a theoretical background of the formula (1), then
substantiate the technique of getting more appropriate
estimates for ν and Pfrac, and finally demonstrate some
examples.

A theoretical background

Formula (1)-based technology of fracture pressure estimation


needs updating on two basic reasons: (i) the formula itself is
not quite adequate, and (ii) the Poisson’s ratio estimates
derived from sonic velocities or lab measurements are
commonly not applicable directly to estimation of the fracture
pressure. Consider first the formula itself.
Formula (1) update. There are two reasons to update
formula (1): it does not account for the rock cohesion, and it
oversimplifies the relation between horizontal stress Phoriz and
vertical stress Pvert .
The role of cohesion. Fractures on the uncased borehole
walls are assumed to appear as soon as the drilling mud
pressure from inside the borehole exceeds minimum
compressive stress directed from outside the borehole. The
minimum compressive stress in (1) is represented by the sum
of pore pressure Pfluid and minimum horizontal stress
expressed as differential (vertical) stress Pd = (Plitho - Pfluid )
factored by the ratio ν/(1−ν). First of all, there is every reason
to replace differential stress by vertical effective stress defined
as
Pvert = Plitho + Pcoh - nPfluid , (2)
where Pcoh is the rock matrix cohesion and n is the effective
stress coefficient [7, 8]. Usually,
0.75 < n = 1. The necessity of introducing the Pcoh term is Fig. 2. Schematic relations between pressures and stresses
obvious: well cemented sandstone or monolithic carbonate involved in fracture pressure prediction, for onshore or shallow
would preserve its integrity at much higher pressure from water offshore (a) and 1 km deep water offshore (b) fields, P horiz
horiz
inside the borehole than, say, loose sand or naturally fractured and Plitho stresses defined from seismic velocities using factor
rock, Figure 2. ν /(1 – ν) to convert vertical stresses into horizontal stresses. The
Relation between Phoriz and Pvert.. Expression ν/(1−ν) in (1) light- and dark-shaded two-sided arrows indicate fracture
serves as a factor converting vertical differential stress Pd into pressure P frac , estimated, respectively, with and without taking
into account the rock cohesion P coh. The intervals from 2.5 to 3.0
horizontal stress Phoriz assuming the latter to be azimuthally
km in shallow water and downward from 2.2 km in deep water
invariant. The factor ν/(1−ν) had been deduced by Dinnik [9] horiz
considering a loose, uncemented, laterally confined rock would be defined as unstable if Plitho is estimated using the
sample subjected to uniaxial vertical stress. For such rocks, the Poisson’s ratio derived directly from seismic velocities. Note the
shallow water flow (SWF)-causing overpressure at about 1.2 km
law of the angle of repose (internal friction angle) α states that deep, and fault-related zone of decreased P coh at about 2.2.km –
(i) this angle is independent of the saturating fluid, and (ii) both these intervals mean trouble.
Phoriz /Pd = (1-sin α )/(1+sin α), hence,
ν/(1 – ν) = (1-sin α )/(1+sin α). (3) the liquid pore fluid is free to leave pores, as is characteristic
for hydrocarbon reservoirs. In a perfectly permeable reservoir,
Here, the left hand side is strongly dependent on the type of
the pore fluid cannot provide any support to the external load,
saturating fluid – this is shown both theoretically and
so mechanically the rock behaves as if it is dry. And in dry
experimentally [2,3,4,5]. The right side, however, is claimed
rocks, Poisson’s ratio ν behaves as prescribed by the Eaton’s
to be independent of the saturating fluid by definition of the
OTC 17206 3

equation, i.e., increases with effective stress and, in the an instant vertical loading-unloading action. In other words,
absence of overpressure, also increases with depth [2,4,5]. In considering the definition of the Poisson’s ratio, it is natural to
contrast, the equation (3) is not applicable to the undrained expect inequality νgeol > νinst to hold true. The point is just
case, when liquid pore fill is sealed in the pores, i.e., in the what is the strength of this inequality, and whether it is
case of shaly rocks which comprise about 75% of the whole possible to calculate νgeol from νinst with the aim of predicting
sedimentary rockmass. Note: it is exactly the kind of rock Phoriz and Pfrac from Plitho and pore pressure Pfluid .
where undrained state can lead to overpressure. As shown in
Appendix A, for such a case As shown in [11], the problem of predicting νgeol from νinst
can be solved using equation
ν ρ 1 − 2ν 1− 2νinst Einst Einst + Egeol
P horiz = P vert (1 + m water ), (4) ν geol =νinst{1+ [1 −exp(− t)]}, (5)
1−ν ρ ν 2νinst Einst + Egeol 3η
where ρ water and ρ are bulk densities of, respectively, (liquid) derived in Appendix B. The “long lasting” elastic moduli are
pore fill and rock on the whole, m = Pfluid/Phydro , and Phydro is known to be by up to 30% less than respective “instantaneous”
the hydrostatic pore pressure at a current depth. Replacement moduli, so for average νinst = 0.25, the νgeol value calculated
of the ratio ν/(1 − ν) by equation (6) makes Phoriz much nearer using (5) amounts to 0.397. The use of νgeol instead of νinst
to Pvert, Figure 3, as is characteristic for combined use of shifts estimates of Phoriz to essentially larger values thus
equation (1) and Eaton’s empirical expression ν = a + b·loge z. increasing the Pfrac estimates, Figure 4.

Fig. 3. Dependence of relation P horiz/Pvert on the Poisson’s ratio ν and


relative pore pressure m = Pfluid/Phydro prescribed by equation (4).
Note: equations (1) and (4) combined together can serve
also for prediction of shallow water flow (SWF) of
unconsolidated sands. Of course, this time the value estimated
as Pfrac has nothing to do with forming fractures – no such
phenomena can occur in the unconsolidated sands. However,
the sand flow into the well is stimulated by abnormally high
pore pressure and marginal difference between Pfluid and Plitho ,
with Pcoh this time being about zero just below the mud line.
Adjustment of routinely derived ν values. There are at least
two issues to consider: (i) deviation of sonic log- or lab
measurement-derived “instantaneous” Poisson’s ratio ν = νinst
from “geological” ν = νgeol, and (ii) effect of anisotropy.
Fig. 4. Schematic result of converting the seismic velocity-derived
The deviation of νgeol from νinst . Rocks are known to be estimates of ν inst into geologically justified values ν geol for the relations
seismically elastic and geologically plastic. It means that between pressures and stresses, shown in Figure 2. This conversion
rocks, subject to lithostatic stress during millions and millions made the depth dependence of the Poisson’s ratio much nearer to what
of years, behave as a slab of tar subject to a week-long stress: was established by Eaton from LOT data (Figure 1) and shifts the P horiz
after removing the stress, they wouldn’t return to their initial (z) curve much nearer to P vert(z) curve for both shallow water (a) and
state. Hence, it is natural to expect that in a rock in situ, deep water (b) cases, as shown by one-sided arrows paving the way to
subject to lithostatic stress for millions of years, a higher a mode reliable prediction of fracture pressure P frac (shaded arrows,
horizontal stress is developed than it would have resulted from designations as in Figure 2).
4 OTC 17206

Anisotropy of stress, normal to well trajectory. Even in the and taking into account, if necessary, VTI anisotropy, equation
isotropic rock, the distribution of stresses normal to the well (5);
trajectory, is anisotropic for wells making an essentially non- - Establishing normal velocity Vnorm cube, e.g., using Willie’s
zero angle θ with vertical. For such wells, minimum stress Time Average equation for clastic (shaly) rocks;
normal to the well trajectory is always Phoriz , so variation of θ - Transfer to the cube of pore pressure in shales using equation
does not affect fracture pressure Pfrac. However, maximum Pfluid = Plitho − ( Plitho − Phydro ) ⋅ (Vint /Vnorm ) 3 , (7)
stress normal to the well trajectory amounts to [(Pvert)2 sin2 θ
proposed in [16].
+ (Phoriz )2 cos2 θ )1/2 , which shall be taken into account when
Further, proceed to estimate the fracture pressure Pfrac. For
designing the borehole casing.
this,
- Create “seismic” Poisson’s ratio νinst cube, e.g., by the use of
VTI anisotropy of rocks. In anisotropic rocks, the effect
one of the options of AVO analysis followed by pseudo-
depends on the type of anisotropy. In shales, velocity of
acoustic inversion;
horizontally propagating waves commonly exceeds velocity of
- Convert νinst cube into νgeol cube using equation (5);
vertically traveling waves of the same mode with no
- Define Pcoh cube with account for available information on
dependence on azimuth. Such a variation is described as
the zones of increased fracturing, faulting, and sedimentation
transverse isotropy with vertical axis of symmetry (“vertical
hiatuses;
transverse isotropy” – VTI). Starting from Dinnik’s equation
- Calculate Pfrac cube by combining νgeol cube, Pfluid cube, Pcoh
Phoriz /Pvert = ν/(1−ν) for an isotropic case, Banik et al. [12]
cube and Plitho profile in accord with equations (2) and (4).
have shown that in the homogeneous VTI medium, Phoriz /Pvert
= [ν/(1−ν) + δ], where δ is one of the three Thomsen’s
Some results of ν geol, Pfrac, and SWF prediction
anisotropy coefficients [13]. As shown above, in the case of a
layer with arbitrarily varying pore pressure overlain by
Numerous and quite convincing examples of successful pore
inhomogeneous isotropic overburden, the Dinnik’s equation
shall be replaced by equation (4). Using the same approach as pressure Pfluid and Poisson’s ratio νinst prediction using seismic
in Appendix A for the VTI medium considered in [12], we data are published recently (e.g., [15,17]), so we concentrate
arrived at equation here on the conversion of νinst into νgeol by the use of equation
(5) and further transfer to Pfrac and SWF modeling. In Figure
5, a section of interpreted seismic data cube for an oil field in
ν ρ 1 − 2ν Timan-Pechora basin is shown. The νinst cube had been created
P horiz = P vert (1 + m water +δ ). (6)
1 −ν ρ rock ν by inversion of the AVO Poisson’s reflectivity cube calibrated
using Poisson’s ratio pseudo logs obtained from sonic log VP ,
HTI anisotropy of rocks. Sometimes seismic velocity is clay content and rock density to convert VP into VS in the
azimuthally dependent. The most common reason for this is available wells. The calibration confirms reliability of the
vertical, azimuthally aligned fracturing. The simplest AVO-derived νinst values, Figure 6. Resulting νgeol cube
theoretical model, approximating azimuthally variant velocity (Figure 7b) has essentially higher Poisson’s ratio values
distribution, is transversely isotropic medium with horizontal concentrated within a much narrower range 0.35 to 0.44 as
axis of symmetry (“horizontally transverse isotropy” – HTI).
In such a medium, horizontal stress Phoriz is azimuthally
variant, so fracture pressure Pfrac is defined by minimum
horizontal stress Pminhoriz . The direction of horiz coincides
Pmin
with the direction of the HTI axis of symmetry; this is also the
direction of minimum P-wave velocity V P(min) . So the only
way to predict fracture pressure in such a medium is to
measure V P(min) by means of azimuthal velocity analysis (e.g.,
DRMA [14]).

Note: within the fractured interval, irrespective of whether


fractures are aligned or not, the rock cohesion Pcoh tends to
zero, which shall be accounted for when using equation (2).

The workflow
The most appropriate starting part of the workflow is the one
discussed in [15]:
- Estimation of interval velocity Vint cube from either DMO or
migration velocity refinement, using semblance-based or Fig. 5. Migrated stack with interpretation in the background, 1 –
carbonates and bioherms, 2 – carbonate shales, 3 - shales
DRMA-based [14] interval velocity estimates, including, if
necessary, azimuthal (HTI) velocity analysis to find V P(min) ,
OTC 17206 5

3/5; pore pressure is set to vary from 66% to 83% of the


minimum horizontal stress reaching 100% of it at the well
bottom.

Fig. 6. Gamma ray (GR) log (a), sonic (V P) log (c)and clay content data
were used to create V S curve (b); from V P and V S , the log-derived ν isnt =
ν isnt, log (d) were calculated using standard approach. In parallel, AVO-
derived Poisson reflectivity cube was inverted to ν isnt = ν isnt, AVO (f).
Comparison (e) of ν isnt, AVO with ν isnt, log confirms high quality of ν isnt, AVO
estimates.

Fig. 8. Pore pressure (a) and fracture pressure (b) gradients.

Two stages of the disequilibrium behavior of the sandy layer


penetrated by a well are simulated: first, the transfer of the
rocks state from elastic to elasto-plastic one (Figure 9), and
second, origination and evolution of a cavern at the well
bottom zone in the course of drilling (no casing installed),
Figure 10. As seen, the modeling predicts unstable situation
Fig. 7. Vertical slices of ν inst (a) and ν geol (b) cubes and inflow of a sand suspension into the borehole at the pore
pressure reaching 0.83 of the minimum horizontal stress.
compared to the range of 0.19 to 0.38 inherent in the νinst cube,
Figure 7a. The target carbonate reservoir is enclosed into Conclusion
shale-dominating formations. No significant anisotropy had
been detected at the survey area, so finally, fracture pressure Prediction of fracture pressure from seismic and well log data
gradient was calculated by combining equations (1) and (4), can be made more accurate if rock cohesion and plasticity is
Figure 8. The resultant Pfrac values occur to be much lower taken into account. For this, a simple technique is proposed.
than it could have been if the raw νinst values were used. The technique is based on the notion of diagenetic evolution of
Since no SWF phenomenon could be in the onshore field rocks as elasto-plastic bodies subject to long lasting stresses
under study, the example imitating SWF-prone environment is leading to rock cohesion, anisotropy, local disintegration
taken from another region [18]. A case of a sandy layer (faults, fractures, SWF) and deviation of the seismic -derived
embedded between two impermeable halfspaces is modeled rock moduli values from those of “geological” rock moduli
[19]. Principal horizontal components of lithostatic stress in formed in the environment of long-lasting stress. The
the plane normal to the well trajectory are assumed to relate as plausibility of the technique is illustrated by a field example.
6 OTC 17206

νgeol = “geologic” Poisson’s ratio inherent in


rocks subject to long-lasting stress
Pcoh = rock matrix cohesion
Pd = differential stress
Pfluid = pore pressure, MPa
Pfraq = fracture pressure, MPa
Phoriz = horizontal effective stress, MPa
horiz
Plitho = horizontal component of the lithostatic
stress, MPa
horiz
Plitho
, inst = horizontal component of lithostatic stress
calculated using νinst value, MPa
horiz
Plitho, geol= horizontal component of lithostatic stress
Fig. 9. Plastic volume around the well at pore pressure equal to 0.83 of
minimum horizontal component of lithostatic stress, for the drill bit Plitho = overburden stress, MPa
penetrating 0.20 (left) and 0.70 (right) of the reservoir layer thickness. Pvert = vertical effective stress, MPa
ρ = rock density, kg/m3
ρ water = in situ liquid pore fill density, kg/m3
calculated using νgeol value, MPa
t = geological time, years
Vint = observed P-wave interval velocity,
m/s
Vnorm = P-wave interval velocity inherent to
normally compacted shale
VP = longitudinal (P)wave velocity, m/s
VS = shear (S)wave velocity, m/s
z = depth

References
1. Eaton, B.A., “Graphical method predicting pressure worldwide”.
World Oil, 185, 1972, 51-56.
Fig. 10. A cavern within the sandy layer, the drill bit penetrating the 2. Castagna, J.P., Batzle, M.L., and Eastwood, R.L., “Relationships
reservoir layer by 0.2 of its thickness (left) and 0.7 of the thickness between compressional wave and shear wave velocities in clastic
(right). silicate rocks”, Geophysics, 1984, 50, 571-579.
3. Carcione J.M. and Cavallini F., “Poisson’s ratio at high pore
Acknowledgements pressure”, Geophysical prospecting, 2002, 50, 97-106.
4. Dvorkin, J., and Walls, L., “Poisson’s ratio and pore pressure
We are grateful to Paradigm Geophysical for permission to estimation”, EAGE 64 th Conference and Exhibition, Expanded
publish this paper, and to Robert Gullco, Paradigm, who Abstracts, 2002, B-37, Florence
5. Lo T-W., Coyner K.B., and Toksoz, M.N., “Experimental
brought to our attention the difference between dependencies determination of elastic anisotropy in Berea sandstone, Chicopee
ν(z) – the one established by Eaton and those derived from shale, and Chelmsford granite”. Geophysics, 1986, 51-1, 164-171.
sonic log data. The work was partially supported by RFBR 6. Eissa, M.A., and Castagna, J.P., “Case study: AVO analysis in a
(03-05-65015). high-impedance Atoka sandstone (Pennsilvanian), North Arkoma
Basin, McIntosh County, Oklahoma”. The Leading Edge, October,
Nomenclature 2003, 988-996.
7. Gangi A.F. and Carlson R.L., “An asperity-deformation model for
effective pressure.” Tectonophysics, 1996, 256, 241-251.
α = angle of repose
8. Kozlov E.A., “Pressure-dependent seismic response of fractured
δ = the Thomsen’s anisotropy coefficient rock”, Geophysics, 2004, 69, 885-897.
Einst = the Young’s modulus defined for 9. Dinnik A.N., “On the rock stress and calculation of the mining
“instantaneous” stress, MPa segmental lining. Engineer Operator”, 1925, ? 7. (In Russian).
Egeol = the Young’s modulus defined for 10. Thomsen, L., “Poisson was not a rock physicist, either!” The
long lasting stress, MPa Leading Edge, July 1996, 852-855
η = the rock viscosity, MPa×year 11. Garagash I.A. “Determination of critical time for creep
n = the effective stress coefficient deformation”. Bulletin of Kazakhstan Academy of Sciences,
1981, ? 4, p.43-51
ν = the Poisson’s ratio 12. Banik N.C., Banic A., Wool, G., Schulz, G., Dutta, N., Marple,
νinst = seismic-derived “instantaneous” R., Casper, T., and Repar, N., “Application of anisotropy in pore
Poisson’s ratio pressure prediction”. EAGE 64th Conference and Exhibition,
νinst, AVO = the amplitude versus offset (AVO)- Expanded Abstracts 2002, B-37, Florence
derived value of the Poisson’s ratio 13. Thomsen, L., “Weak elastic anisotropy”, Geophysics, 1986, 51,
OTC 17206 7

1954-1966. pore pressure, re-defining σ33 = - Pvert and σ11 = - Phoriz ,


14. Kozlov E.A. and Varivoda D.Y., “Dense 3D residual moveout expressions (A-2) and (A-3) are easily reduced to equation
analysis as a tool for HTI parameter estimation”. Geophysical (4) in the main text.
Prospecting, 2005, 53, 131-148.
15. Snijder J.H., Dickson D., Hillier A., Litvin A., Gregory C and
Crookall P., “3D pore pressure prediction in the Columbus basin, Appendix B. Derivation of equation (5)
offshore Trinidad and Nobago”. EAGE 63rd Conference & Generally, elasto-plastic hereditary behavior of an
Technical Exhibition, Expanded Abstracts, 2001, O-36. isotropic rock subject to long lasting stress can be defined
16. Eaton B., and Eaton, T., “Fracture gradient prediction for the new
generation”, World Oil, 1997, October issue, 93-100 [11] by equations
17. Wang J., Dopkin D., and Huw J., 2004, “Vizualization for pore Egeol 2ν geol
pressure prediction”. First Break, 23, March 2004, 57-62.
σ ij = [(δ ikδ jl+δ ilδ kj) + δ klδ ij]ε kl ,(B-1a)
2(1+ν geol) 1− 2ν geol
18. Kozlov E., Garagash I., and Lowrie. A. “Seismic-based
Geomechanical Modeling Provides Attributes for Directional Well E geol = E inst (1 − Q∗ ) , (B-1b)
Planning”. AAPG Annual International Conference, Expanded

Abstracts, 2003. ν geol = ν inst (1 + R ) . (B-1c)
19. Garagash I.A., Nikolaevski V.N., Dudley J.W. “FLAC simulation
of triaxial and compaction tests in unconsolidated reservoir sand”, Here, σij and εkl are stresses and strains, δij are
in: FLAC and Numerical Modeling in Geomechanics. Proceedings Kronecker’s deltas, E is the rock Young modulus, and
of the international symposium, 1999, Balkema, 505-510. values marked by the “geol” and “inst” subscript are
properties describing the rock response to a long lasting
Appendix A. Derivation of equation (4) stress and an instantaneous stress, respectively. The
symbols Q* and R* in (B-1b) and (B-1c) designate integral
Assuming the rock skeleton to be subject to effective operators of the Volterra’s type
portion σ ij + pδij of total stresses σ ij related to the rock
t

element facets, respective deformations are defined as


Qt* [ f ] = ∫ Q(t − τ ) f (τ )dτ , (B-2a)
−∞
ε ij = (σ kl + p δ kl ) Π ijkl , (A-1) t
where Π ijkl = Π klij , Π ijkl = Π jikl , and Πijkl = Πijlk are Rt*[ f ] = ∫ R (t − τ ) f (τ )dτ , (B-2b)
−∞
elements of the symmetric compliance tensor containing, in
Here, t is geologic time, and f(t) is any smooth function.
the case of arbitrary symmetry, 21 independent components;
When describing the hereditary properties of real
compressive stresses here are set negative.
media, it is the shear aftereffects that shall be taken into
According to [19], for orthotropic medium with axes of
account first of all, while the bulk aftereffects can be
symmetry parallel to coordinate axes,
regarded as negligible. On this ground, the confining
1 ν ν
ε 11 = (σ 11 + p ) − 21 (σ 22 + p ) − 31 (σ 33 + p ) , (A-2a) stress operator can be set invariant, i.e.,
E1 E2 E3
E geol E inst
ν 1 ν = = const . (B-3)
ε 22 = − 12 (σ 11 + p) + (σ 22 + p) − 32 (σ 33 + p) , (A-2b)
1 − 2ν geol 1 − 2ν inst
E1 E2 E3
ν ν 1 This leads [11] to
ε 33 = − 13 (σ 11 + p) − 23 (σ 22 + p) + (σ 33 + p) , (A-2c)
1 − 2ν inst ∗
E1 E2 E3 ν geol = ν inst (1 + Q ). (B-4)
1 2ν inst
ε 23 = σ 23 , (A-2d)
2 G 23 In the case of an elasto-plastic Voigt solid (Figure B-1),
1 the integral operator Q* is reduced to
ε 13 = σ 13 , (A-2e)
2 G 13 E
t Einst + Elong
Q∗[ f ] = inst ∫ f (τ ) exp[− (t −τ )] dτ , (B-5)
1 3η 3η
ε 12 = σ 12 , (A-2f) −∞
2 G 12 where η is the rock viscosity. Further simplification,
where E1ν21 = E2ν12, E2ν 32 = E3ν 23 , E3ν 13 = E1ν 31 , and Ei are setting strains to zero at t < 0 and definition of f(t)
elements of the Young’s modulus vector. included, leads to equation (5) in the main text [11].
The density, elastic constants, stresses and strains are
assumed to be dependent on the vertical co-ordinate
(depth) x3 only. In such a case, vertical stress σ 33 at the
depth x3 = z is the maximum stress and is defined by the
weight of the overburden as
z
σ 33 = − g ∫ ρ ( x 3 ) dx 3 , (A-3)
0
where ρ is average rock density, and g is gravity
acceleration. For pure uniaxial external stress and variable Fig. B-1. Schematic mehanical analogue of the Voigt’s solid.

You might also like