You are on page 1of 10

.

. .
SPEYISRM 47201

Virgin damage surface of a synthetic analogue to a reservoir sandstone


Tor Eding Unander, [KU Petroleum Research, and Rune M. HoIt, SPE, IKU / NTNU

LWYfi@t 1SSS, %c+ety of Petrc4eum Englc&ars, Inc. overcome. Typically, the zero stress state is not within the
This paper wes Pn$pemd for presemtsdica a! the SPE.ilSRM Eurock 90 held In TrcWhelm, virgin damage surface of a reservoir rock. Per definition, any
Norway, S-10 July 199&
cored sample is then damaged, and the virginity of the
TM paper was sakted for presentation by an SPE Prcgram Committee folfoting rwfew of damage surface is lost. We have circumvented this problem
tnfumatbn wotained in m abstract sutnnffted by the author(s) Ccmtents of the paper, as
presented have nol been reviewd by rho Soclery of Petroleum Enginears and am subied 10 by making a synthetic rock sample under simulated in situ
c0w3f0n by the amfw(s). Tfw material, as pmsemed, *S nm rily reftecl any
pesltkm of ffw SOcWy of Petroleum S@7e.sm, Rs ofticws, or membsrs Pspws presented at
conditions and tested it directly from that state. We have
SPE meetings are sub@t to pubiicalicm rwlew by Editorial Committees of the Society of previously used this technique in an extensive study of core
Perro!wm Engineers. t3acrrcmlc rep+cdudcm, dlstrMutfon, or stwe.ga of any pan of this papw
fw @mmerckl purposes wfrhoul me wrlrren con%mt of the Society of Petroleum Engineers is damage investigations3456and the specific material used in
prdbiied. Pennlsslcm to reproduca fn mfnf Is reshfcted to en abstract of not mare than S03 this study is further described elsewhere in these proceedings.
wdq iffustrakW may not be w+xed. The abstract must ccalain cuwptcuous
acknawfsdgmem of where and by Wwn the psiw wes presented. Write tibrari~, SpE, p.o. The design goal of this synthetic material was to make it
W Ss5SSS, Ricfurdsco, T% 75CBS-Z336. U.S A,, fax 01-972-952-94S5.
mechanically and petrophysically similar to a sandstone from a
North Sea gas field.
Abstract Studies of the virgin darnage surface are important for two
This paper is based on laboratory investigations of synthetic reasons. First, it is useful in studies of core damage. Here we
sandstone cemented under stress making it possibIe to perform show that we cannot unload the sample to zero stress without
mechanical tests directly from the simulated in situ stress state. crossing the damage surface. This is a manifestation of coring
We have mapped the acoustic emission damage surface, or damage. Second, it is important to know the virgin damage
Holcomb surface as we have called it here, around this stress surface of a reservoir rock when trying to model the
state. The results show that this surface does not coincide with mechanical behaviour of a reservoir. This paper will show
the true damage surface as damage is induced in the material that even small stress changes will permanently deform the
even for very small load deviations from the forming stress rock. This means that it is inappropriate to assume that the
state. The Holcomb surface mapping is done with a new reservoir deforms elastically as response to depletion or
technique that is able to better describe and quantify onset of pressurisation.
acoustic emission.
The AE damage surface or Holcomb surface
Introduction In the acoustic emission (AE) context, the damage surface is a
We have in this work tried to irwestigate how much in situ surface in stress or strain space that encloses all stress,
stresses in a sandstone reservoir can be changed from the respectively, strain states that a sample can be loaded to
original stress state without damaging the reservoir rock. without inducing AE activity. This concept is primarily
Damage means here irreversible alterations of the rock. . This developed by HoIcomb and Costin8. It is analogous to a yield
has been done by mapping the so-called virgin damage surface surface, which encloses all stress states a sample can attain
of the reservoir rock. This is the surface in stress space that without being plastified. As mentioned, we do not know a
encloses aII stress states the material may be loaded to, starting priori if the AE damage surface is the same as the true
from the original stress state, without damaging the material. damage surface as defined in the introduction. Thus, to
We have wanted to do this mapping by using acoustic distinguish these two surfaces, we will refer to the AE damage
emission (AE) measurements. This is because we have surface as the Holcomb surface in the following.
assumed that this technique is most sensitive to
micromechanical damage, and due to a general interest in AE Detection of the Holcomb surface. Unfortunately, the
damage surfaces. The AE damage surface is an important detection of the Holcomb surface has the same problems as
concept in studies of how to determine in situ stress from AE yield surface detection: A good criterion is needed to
measurements*, and for AE mechanism studies in generalz. determine the position of the surface. The problem is to
Note that it is not given that the AE damage surface is the detect a transition from no AE activity to some activity, the so-
same as the true damage surface as defined above. Thus, we called onset. The difficult point is that in the case of AE, zero
have checked this assumption. activity probably does not exist. It is always a cme of
h this study, one fundamental problem had to be

23
.
IL

2 T.E. UNANDER, R.M. HOLT SPE 47201

listening cIosely enough. To circumvent the problem of step and to is the onset time. The resulting model is
identifying zero activity, some activity must be accepted, but it

saw=%
should conform to a certain pattern. In a recent pape?, a
general method for onset determination is presented. Two
different criteria are also presented there, and both will be
presented and used here. The general idea is that a model for
We have called it the Step model, and an example of this

the time evohrtion of the AE activity is fitted to the measured
data. The onset of AE activity is then a parameter of the model can be seen in Fig. 3.
model that can be retrieved after the fitting The second model describes a process that at a certain time
After determining the AE onset from several testes where starts to dominate a background process. Both these
the stress is changed in various directions from the in situ processes are described by Eq. 1, but if the constant b of the
stress, the virgin Holcomb surface may be constructed. second process is larger than the b of the first process, the
According to the definition of the Holcomb surface, the onset second process will eventually become dominant, The model
is the point of time when the stress, or strain, crosses the can be written as
surface. Then, the 10CUSof alI stress, or strain, states at onset Ymwe(t)=exp(al +~t)+exp(a, +~t) ............................
is the measured Holcomb surface.
Onset models. The two criteria that will be used here The model is called the Emerge model as it may be used to
correspond to two different models. They describe two describe activity from a new AE mechanism emerging out of a
different ways onset may occur, and consequently they cannot background activity level. An example of this model can be
be directly compared. Note that the two models are not seen in Fig. 6. The onset point in this model has been defined
alternative in the sense that they may be fitted to the same data as the expected time for the first hit to occur due to a process
set. The situation should be compared to the different failure given by Eq. 1 with parameters a2 and & As the second
mechanisms that are responsible for the total failure surface of process is the interesting one, it is reasonable that only these
a material. The idea that various parts of the Holcomb two parameters should be involved. The expectational value
surface are caused by different damage mechanisms has and variance can be found numerically by the expressions
recently been considered by Pestman and van Munsterl. given by
The general assumption underlying both models is that the
normrd behaviour is an AE activity increasing exponentially ~ = J ~f,(t)d............................................................ (4)
when a sample is loaded at a constant rate. The activity can be o
described by
y(t) = e ...................................................................(l) Crz= (t-p)*.fr(r)dt................................................(5)
J
o
where yis a measure of the AE activity, tis time and a and b
where the probability density function fr is given by
are constants. Thk Kind of behaviour is generally observed,
and an onset deviates from this behaviour. In this study, AE
activity is quantified by the hit rate, that is the number of
acoustic emissions recorded by one sensor per unit time. The
proper variable in Eq. 1 is not time, but the co-ordinate in the
r(t)=+exp(-ly(z)h))
20(6)
The derivation of these expressions can be found elsewhere.
space where the Holcomb surface is mapped. For practical f.t is defined as the onset time, but note that & cannot be
reasons, we will use time, and when stress and strain change directly interpreted as the variance in the onset time estimate.
linearly with time, the representations are equivalent. Nevertheless, c? gives an indication of the onset sharpness. In
The two models differ in how deviations in the general
many cases, this definition of onset gives onset points that are
behaviour may occur. The deviations described are localized
placed much before the visual impression of an onset.
in time such that an onset may be defined. The first model
Note that the Step model can be viewed as the case where
describes the I@iser step2. Tlris is the behaviour seen in cyclic
the exponential slope of the process decreases from b+ I/t= to
loading experiments where the Kaiser effect is marked. Due
b as opposed to the Emerge model where the slope increases
to a previous Ioad cycle, no AE activity occurs before the
from b, to b2. It means that if the estimated logarithm of the
previous maximum stress is exceeded. At that point, the
rate versus time is either convex or concave within a certain
virgin behaviour described by Eq. 1 resumes. This can be
interval, it will be reasonably modelled by one of the models.
modelled by multiplying Eq. 1 with a step function h. Usually,
Moreover, there is never any doubt about which model to use.
the step is not perfect and the function that will be used here is
Fig. 1 to Fig. 9 show how these two models have been fitted to
...............................................(2) the measurements.
h(t)
., = ~ As an aside, note that evolution of AE activity with time
l+WJI-W1
[ (.3 )
has received very little attention in the AE literature, but
already Scholz presented a model for this. His model
where t= is a transition time describing the sharpness of the assumed that the activity was governed by a Gaussian

24
SPE 47201 VIRGIN DAMAGE SURFACE OF A SYNTHETIC ANALOGUE TO A RESERVOIR SANDSTONE 3

variation in the stress field in a material with constant source location, and in those cases, there is no end surface
microscopic strength. This gives a different normai AE effect. In the other cases, this is not checked.
behaviour than the one described by Eq. 1. The rationale
behind our form may be that potential AE sources is Resuite
distributed in the sample according to a Boltzmann-like model The time history plots of aIl the tests are shown in Fig. 1 to
where stress has the role of temperature. When the stress Fig. 9. The plots show the lapse of the stresses with the AE hit
increases, the number of the potential sources that crack rate and the fitted onset model. The estimated rate is shown as
increases exponentially a step-like curve, and note that the rate is plotted on a
logarithmic scale. This method for rate estimation is presented
Experimented method in a recent publication. In tests GRONOO1 and GRONO03 the
The experiments presented here have been run in a standard activity was measured in I s intervals with ionger pauses in
triaxial testing machine from TerraTek Inc., Utah, on between. Then the inherent stochastic nature of the AE
cylindrical samples with diameter 1.5 (38 mm) and length 3 process becomes more apparent, and this accounts for the
(76 mm). In the standard triaxial configuration, two of the noisy rate curve in these two cases.
three principal stresses are equal, and there is no provision for The fitted onset model is shown as a thick line in the
rotating the principal stresses. This means that we are only figures. The model parameters have been found by calculating
able to map the damage surface in a two-dimensionai plane in the maximum likelihood estimates of each of the parameters.
stress space with crl = cr. and Crz= rs3= p., where a, is the axial The hits are distributed in equally long bins, typicaily 5 or 10s
stress and pc is the confining pressure. p. is not necessarily long, and the number of hits per bin is counted. The likelihood
less than IJWThe axial and lateral deformation of tire sample is of the outcome, given the model, is calculated for each bin.
also measured. This enables mapping of the damage surface The maximum likelihood estimates are then the parameter
in strain space. values that maximize the sum of the likelihoods. Note that in
Ilk investigation has been performed on a synthetic the tests GRON023 (Eg. 6) and lT96- 12 @lg. 7) the Emerge
materird named Gronstone. This materiai is described in more model has been used. In the other cases, the Step model has
detail in an accompanying paper7. The unconfined been used. This is solely determined by the shape of the onset,
compressive strength of the material is around 10 MPa, and not due to any assumptions about mechanism etc. In most
porosity 20%, and mean grain size is 180 pm. The important cases, the models fit the measurements well. The behaviour of
point in the manufacturing procedure is that the mixture of uniaxial strain test GRONOO1 deviates from the identical tests
sand and cement is loaded to 30 MPa axiai stress and 15 MPa GRONO03 and GRON014, and in GRONO19, the unloading
lateral stress and cured at this stress state. This stress state is was stopped before a clear second trend was established.
then the simulated in siw stress state, and the actual tests are The stresses and strains at onset are tabulated in Table 2.
run along different stress paths starting from this state. The stress values are plotted in Fig. 10. It is also possible to
Table 1 gives an overview of the tests used in this study. plot the virgin Holcomb surface in strain space. As we have
Two of the tests(TT96-11 and TT96- 12) were started with defined the material to be undamaged at the forming stress,
a series of smaiI stress cycles before the main test. The idea this state shotdd also be the zero point for the strain
of these cycles was to check that no damage was induced calculation. Because the tests are both loading and unloading,
when the stress was cycled within the Holcomb surface. For it is convenient to use octahedral strains. The relations
titk purpose acoustic pulse transmission velocity was between the radial and axial strain, e, and G, and the
11
measured during the cycling. These measurements are octahedral normal and shear strains, %~ and ywi are
performed using pairs of piezo-electric transducers. One of
2e +& ............................................................. (7)
the transducers is exited with a 300 Vw, one cycle sine wave, Em = ~
and the arrival of the acoustic pulse is measured by the @her 3
transducer and recorded with a digital oscilloscope.
The equipment we have used for the acoustic emission
measurements is a LOCAN320 from Physical Acoustics
Corp., New Jersey (PAC). This is a hit description based The virgin Holcomb surface in strain space using this
device, which means that the exact time of each hit is representation is shown in Fig. 11.
recorded. This is necessary for performing the analyses that Damage verification. In the tests TT96-11 and TT96- 12, the
we have done here, We have recorded the activity with only main loading was preceded with some small load cycles to
one transducer, a Micro 30 from PAC mounted outside the investigate whether damage was induced for load cycles inside
confining pressure sleeve. Note that one transducer is the Holcomb surface. In TT96- 11, the shear stress was first
generally not sufficient because there is no way to know what increased, then unloaded to the forming stress. In the second
is the source of the activity. Typicaliy, crushing of grains cycle, the shear was decreased by the same amount, and the
between the sample and pistons is the dominant source of reloaded back to the forming stress. The mean stress was kept
acoustic emission in sandstone, a phenomenon that will be constant ali the time. Between each load segment, the acoustic
documented in a future publication. We have checked the velocity was measured. TT96- 12 was cycled in a similar way,
source of the activity in the case of the uniaxial strain tests by

25
.

u. \

4 T.E. UNANDER, R.M. HOLT SPE 47201

but in this case, the mean stress was changed with constant may even be fitted to a two-segment line as proposed by
shear stress. The loadinghutloading was done in the direction Pestman and van Munster*. The load paths used with
of pure shear and pure mean stress, respectively, due to the GRONOI 9 and GRON015 do not approach such a criterion. -
special role of these quantities in linear, isotropic elasticity. These two tests are unloading, and indicate that the zero stress
Given a linear, isotropic material pure shear loading will not state is not within our Holcomb surface as expected. While
affect the volumetric strain, and similarly, mean stress loading the shape of the Holcomb surface in stress space (Fig. IO) is
will not affect the shear strain. Note that this material is not relatively regular, the surface in strain space @lg. I 1) is more
completely isotropic as it is made under anisotropic stresses complex. This may partly be attributed to experimental
and compacted in an anisotropic fashion. Fig. 12 and 13g. 13 scatter, but we think that this reflects a variety of damage
show the stress versus strain curves for these to series of mechanisms. Fttst, the AE behaviour in tests GRON023 and
cycles. Note the large permanent deformation due to the shear TT96-12 is fundamentally different from the rest. There, the
loading cycle, which clearly show that it was not within the Emerge model was used as opposed to the Step model
true damage surface. Some permanent deformation can be elsewhere. Strictly, those points cannot be compared to the
seen in the other load directions, too. other. Still, it makes sense to define a damage surface that
The assumption that these small cycles were not damaging represents different damage mechanisms in different
the sample was further checked by acoustic emission and directions. Second, with some imagination it is possible to see
velocity measurements. These measurements are shown in Flg a trend for the sIightly dilative to compactive tests, TT96- 11 to
14 for TT96-I 1. The sensitivity of [he AE measurements was GRON014. Interestingly enough, the unloading test
12 dB higher in these measurements than what is seen in Fig. GRON015 coincides with the pure shear loading test TT96- 11.
5. Wkh the lower sensitivity, the recorded activity would be Grain rotation may be the mechanism at work in these tests.
negligible. The increase in axial velocity is 20 m/s. The mean Finally, there are three strongly dilatant cases, GRON023,
stress cycling in TT96- 12 gave negligible AE activity even TT96-12 and GRON019. Here, the cement may give the AE
with the increased sensitivity, and the axial velocity was activity when local tensile stresses break cement bonds. These
approximately constant, but the radial velocity increased with broken bonds may start to coalesce, creating axial extensile
30 mk. Incremental shear and bulk modulus of the material cracks that in turn leads to global failure.
may be directly found from these stress cycles, and are This material is very close or at the virgin damage surface
2.3 GPa and 4.7 GPa, respectively. in the direction of increased shear. Note that the material has a
Fig. 15 compares the AE activity recorded during the first significant unconfined compressive strength (10 MPa), and
shear loading in test TT96- 11 with the first part of the main then most probably a significant cohesion. The reason for this
shear loading. When the hit rate is corrected for the difference negligible micromechanical strength may be a wedge effect.
in load rate in the two cases, the activity behaves identically. Even minute macroscopic stress increases may be amplified to
Normally, a Kaiser effect would have been expected here. overcome the cohesion locally. In other directions than the
This observation is intriguing in Iight of the Step model used shear direction, the material seems to endure the load change
in this case. The model gives an onset at a stress level beyond more. It is unreasonable to assume that the damage surface is
what is plotted in Fig. 15. That means that the first cycle did not completely within the Holcomb surface. That means that
not cross the Holcomb surface, and consequently the Holcomb the AE activity seen during unloading indicates that the
surface was not distorted by this load cycle. Then, it is as damage surface has been crossed.
expected that the AE behaviour is identical in the two cases The two onset models that we have used here have nicely
because the HoIcomb surface is the same in the two cases. been able to describe and quantify the AE onset. It is
encouraging that the quantified sharpness of the onset is
Discussion good, that is, the transition period is well inside the test
Clearly, the assumption Holcomb surface is the same as the segment. Furthermore, there does not seem to be any need of
darnage surface for this material is not valid. It is interesting other models.
to note that the damage is most clearly seen from the
mechanical measurements. Acoustic methods are usually Conclusions
assumed more sensitive to micromechanical damage. 1, This study does not support an assumption that a
Consequently, it would have been natural to construct a reservoir rock has a safe region within the stress may be
damage surface, or actualIy a yield surface, from the varied elastically without damage being induced in the rock.
mechanical measurements. This is of course possible, but the This means that it is not sufficient to use simple, linearized
procedure is much more involved including a multitude of models when simulating deformation of a reservoir.
load cycles. One nice example of such mapping is the work of 2. For a normally consolidated material, the microscopic
Tatsuoka and Ishihara12. We can here only speculate about strength may be close to zero for shear loading even if it is
what the Holcomb surface in this case represent. It may be cemented.
that what we measure is a reminiscence of the behaviour of the 3. The AE measurements have given a closed virgin
sand before cementing. Holcomb surface that does not include the zero stress state.
Generally, the onset points, except GRONO19 and However, it is clear that in this case the Holcomb surface does
GRONO15, follow a kind of Mohr-Coulomb criterion. They not coincide with the true damage surface.

26
. .

SPE 47201 VIRGIN DAMAGE SURFACE OF A SYNTHETIC ANALOGUE TO A RESERVOIR SANDSTONE 5

4. The two models for describing AE onset have given 5. Holt R. M., Unander T. E., Kenter C. J. and SantarelIi, F.J.:
new possibilities for quantifying onset. This increases the UnIoading effects on mechanical properties of a very weak
possibility to use AE activity to identify damage mechanisms. artificial sandstone: Applications to corirrg: Geotechnica/
Engineering of Hard Soils-Sofi Rocks, Anagnastopouhrs et al.

Acknowledgements
(eds),A.A.Bafkema,Rotterdam (1993), pp. 1609-1614.
6. HoIt, R.M., Brigrroli, M., Fjax, E., Kenter, C.J. and Unander,
This work has been performed at IKU Petroleum Research. It T.E.: Core damage effects on compaction behaviour~ paper
has been sponsored partly through the Core quality SPE 28027, Eurock 94, Balkema, Rotterdam, (1994), pp. 55-
programme at IKU and pwtIy through Mr. Unanders Ph. D.- 62.
study. The Ph.D. study is sponsored by Statoils scholarship 7. HoIt, R.M., Brignoli, M., Kenter, C.J., Meij, R. and Schutjens,
programme, VISTA, and SheIl International Exploration and P.M.T.M.: From core compaction to reservoir compaction:
Production B.V., RTS, Rijswijk, The Netherlands. Correction for core damage effects; paper SPE 47263 presented
The authors thank Lasse Stamnes for running the tests. at the SPWLSRM Eurock 98, Trondheim, Norway, 8-10 July
1998.
8. Holcomb, D.J. and Costin, L.S.: Detecting Damage Surfaces in
References
Brittle Materials Using Acoustic Emissions: J Appl. &tech.
(1986) 108, pp. 536-544.
1. Pestrnan, B.J, and van Munster, J.G.: An Acoustic Emission 9. Unarrder, T. E. and Elvebakk, G.: Tools for Analysis of AE
Study of Damage Development and Stress-Memory Effects in Time Behavior, Acoustic EmissiotiMicroseismic Activi& in
Sandstone,n Int. J. Rock. iUech. Min. Sci. & Geomech. Abstr. Geologic Structures and Materials Proceedings of the Sixth
(1996) 33, No. 6, pp. 585-593. Conference, Trarrs Tech Publications, C1austal-ZelIerfeld,
2. Holcomb, D.J.: Observations of the Kaiser Effect Under Germany (1998), pp. 557-570.
Multiaxial Stress States: Implications for Its Use in Determining 10. Sholz, C.H.: Microfracturing and the Inelastic Deformation of
In Situ Stressfl Geophys. Res. Lett. (1993) 20, No. 19, pp. 2119- Rock in Compression, J. Geophys Res. (1968) 73, No. 4,
2122. pp. 1417-1432.
3. HoIt, R.M. and Kenter, C.J.: Laboratory simulation of core 11. Chen, W.F. end Han, D.J.: Plastici~ for Structural Engineers,
damage induced by stress release, Rock Mechanics, Tlllerson & Springer-Verlag, New York, (1988).
Waversik (eds), Balkema, Rotterdam (1992), pp. 959-968. 12. Tatsuoka, F. and Ishihara, K.: Yielding of Sand in Triaxial
4. Holt R. M., Unander T. E. and Kenter C. J.: Constitutive Comvressiorr, Soils and Foundations (1 974) 14, No. 2, pp. 63-
mechanical behaviour of synthetic sandstone formed under 76.
stress, Int. J. Rock Mech. Min. Sci. & Geomech. Abstr. (1993)
30, No.7, PP.719-722.

TABLE I-OVERVIEW OF THE TESTS.

Tvoe of test ~rese.kit rain increment Qnset model used


GRONOOI Uniaxial strain compaaon (Ea >0,6, = o) Step
GRONO03 Uniaxial strain compaction (E. > o,E, = o) Step
GRON014 Uniaxial strain cornpatilon (Ea >0, E, = o) Step
GRON022 Axial loading (u. >0, u, = o) Step
TT96-11 Shear loading (6./6, = -2,6. > o) Step
GRON023 Radial unloading (u. = o, d, < o) Emerge
lT96-12 Isotropic unloading (ua/ur = 1,6= < o) Emerge
GRON019 Prcymrtionalunloading (6./6, =2, & co) Step
GRONOI 5 Axial unloading (Ua <0,6, = o) Step

I TABLE 2.-STRESS AND STRAIN AT AE ONSET

~st name %.l!&81 m &JD@JllkJL?@ml


GRONOOI 37.8 17.0 0.37 1.01
GRONO03 40.1 17.2 0.48 1.38
GRON014 41.5 18.1 0.57 1.48
GRON022 38.4 14.9 -0.05 1.72
TT96-11 33.9 13.1 -0<28 2.00
GRON023 29.9 li.3 -0.31 0.74
TT96-12 23.5 8.5 -0.90 0.70
GFtoNoi9 10.2 5.0 -1.24 1.55
GRON015 13.8 14.9 -0.33 2.17

27
T.E. UNANDER, R.M. HOLT SPE 47201
R

II ! 1, 1
L 0.01
o
1....,,...,.,..:. i 0.1 o~
530Jsso353w-73@J ~
ssmssoomco-~ ~
Time [s] GRONOI 4
Time [s] GRONOOI

Fig. I.AE rate andthefittad model together with theaxlal and Fig. 3. AE rate and the fitted model together with the axial and
radial stress for test GRONOOI. The solid vertical lines show the radial stress for teat GRON014. The solid vertical lines show the
estimated onset point. The broken Iineeindicate thesharpness estimated onset point The broken lines indicate the sharpness
of the onset. Sss comment in text regarding the noisy rate curve. of the onset.

lW
SOT -1

10

1 0,
0.1

i
4 I
I 1 0.1 o~
1, ,1,.1 ,+___--.--l 0.01
o ! . .-A -
5w303aI --- 1Om
s3cosews3m-~ -
Time [s] GRON022
Time [s] GRONO03

Fig. 2. AE rate and the fitted model together with the axial and Fig. 4. AE rate and the fitted model together with the axial and
radial stress for test GRON022. The solid vertical lines show the
radial stress for test GRONO03. The solid vertical lines show the
estimated onset point. The broken lines indicate the sharpneas
estimated onset po~nt. The broken lines indicate the sharpness
of the onset.
of the onset. Sea comment in text regarding the noisy rate cuwe.
.

w..

7
SPE 47201 VIRGIN tIAk4AGE suFIFAcE OF A sYNTI-imc ANALOGUE To A FIESERVOIR SANDsTONE

25. ~ 100

10

0.1

I.. l...:.,..:,,. 0.01


0 1 , ;0.01
?4xlms400smJ~ ~ I04C0
7m7e327@XlsotQm -~
Time [s] l-r!wl 1
Time [s] Trs6-12

Fig. 5. AE rate and the fitted model together with the axial and Fig. 7. AE rate and the fitted model together with the axial and
radial stress for teat TT96-11. The solid vertical lines show the radial stress for test TT95-I 2. The solid vertical lines show the
aatimated onset point The broken lines indicate thesharPneas estimated onset point. The broken lines indicate the sharpness
of the onset. of the onset. The drop in activity at 10200s is due to an intended
reduction in measurement sensitivity.

----
oi. !~ S2Co Esoo
I
b 0,1
7W0
01
5scrJ 58W
I I
&3Co
I
SSoo
! 0.01
X-30a

Time [s] GRON019


Time [s] GRON023

Fi&J.6, AE rate and the fitted model together with the axial and Fig. 8. AE rate and the fitted model together with the axial and
radial stress for test GRON023. The solid vertical lines show the radial atress for test GRONO19. The solid vertical lines show the
aatImated onset point. The broken lines Indicate the sharpness estimated onset point. The broken lines indicate the sharpness
of the onset. of the onset.

29
Ii%

8 T.E. UNANDER, R.M. HOLT SPE 47201

Iwo

I
so

~
mzo- -
k
z

10

5. -

0.
5420 5470 5520 5570
llme [s] GRONOI 5

Fig. 9. AE rata and the fitted model together with the axial and radial atresa for test GRONOI 5. The solid
vertfosl lines ahow the estimated onset point. The broken lines indicate the sharpness of the onset.

45

40 78 GRONO14 I
GRONW3
GRONOO1
I
/
/
/
?$ /
/
/
36 /
/
/
,/
/
so GRON023 o & ~ x Forming stress
/
/
~ ,/

~ /
/
~ 25 /
T-T96-12 .O /
/
o /
g /
/
/
20 /

f!
/
GRON019 ,? /
/
, w
15
,~; GRON015

/ b
/
10 / /
/
/
/
/
/
5 /
/
/
/
/

0 I
o 5 10 15 20 2s 20 35 40

Radial stress, u, [MPa]

Fig. 10.The virgin Holcomb surface of Gronstone plotted in stress apace. The solid marks indicate estimated
onset points Square points are estimated using the Step model, and triangular points are estimated using the
Emerge model. The open circles give an indication of the transition region of the onset.

30
. . .

--..- -...-..
SPE 47201 VIRGIN DAMAGE SURFACE OF A SYNTHETIC ANALOGUE TO A RESERVOIR SANDSTONE V

4-

?
3.5- -
1
I
!
t GRON015
1
t, q 3- -
\\
1! aI
$1
tl
!t 2.5- 1
!1 o
t
1,

TrS6-11 ;, ,,+)
~ 2/ - c1
,(
GRONO19 GRON022 f,
:; { ,/
.
m, t ,/
GRON023 1,1. +-
.
\ ! f? GRON014 I
\ Q Al
,, (
GRoNcKr3 I
\ \ -, ! /, 1
\ GRONCKN
. \l
q o , ib ~d
\
\ \ #c)
\ , ; t 4

. I
Q
I
-2 -1.5 -1 -0.5 0 0.5 1

Normal strain, ~t [1 O?

Fig. 11.The virgin Holcomb surface of Gronstone plotted in strain space. The solid marks indicate estimated
onset points. Square points are estimated using the Step model, and the triangular points are estimated using
the Emerge model. The open circles give an indication of the transition region of the onset.

a ----. . ... .. . ,
CL 8.5-
~

J 8- - - - -- ~*- -
+v/ ~
g- 7.5- ----
~ g 19.5
!l/~ i i
% 7- , , --- -G
Shear ~ K
. .- tJW
% ,, stain ,
~ 6.5- -
a 7i Volumetric -
u) //~ [
ii strain
6- 1 18.5-1 1
-0.5 0 0.5 1 1.5 -0.1 0 0.1 0.2

Strain [1 O-T -rW3-11


Strain [107 ~w.12

Fig. 1Z-stress veraus strain curves for the small load cycles Fig. 13-Stress versus strain cuwes for the small load cycles
performed prior to main Iosding in test lT96-11. The arrowa performed prior to main loading in test TT96-12. The arrowa
indicate direction of the first load cycle. T~. = Iu.-u,II2 indicate direction of the first load cycle. C&I= (cr. + 2u,)13

31
10 T.E. UNANDER, R.M. HOLT SPE 47201

r.

I
s

1-=1
[
3150
A A


A
A
A 31(XI

Hi rate

L
3050
Shear stress
Axiil P-wave
---=. A Radml P-wsve
\
2950

i 2850

o Zaca
S3cm 55C0 570U 5900 6100 8WJ 8500
Thna [s]

Fig 14.Time history plot of the shear loading cycle in the test TT96-11, The shear stress is
maximum shear stress, TN = Iu.-u,II2. The mean stress was kept constant during these cycles.

II I
I
2. loading I
1

lo~
7 8 9 10 11

. Shear stress tm.x ~Pa]

1s. Acoustic emission activity during the first shear Iosding and the second main shear
Fig.
loading in test TT96-I 1.

32

You might also like