You are on page 1of 8

Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 191 (2017) 394 – 401

Symposium of the International Society for Rock Mechanics

An Assessment of the Post-Peak Strain Behavior of Laboratory


Intact Rock Specimens Based on Different Dilation Models
I. Pérez-Reya, L.R. Alejanoa,*, E. Alonsoa, J. Arzúab, M. Araújoa
a
Department of Natural Resources and Environmental Engineering, University of Vigo, Vigo 36800, Spain
b
Universidad Católica del Norte, Antofagasta 0610, Chile

Abstract

The complete stress-strain response of a number of intact rocks —including igneous, metamorphic and sedimentary ones— has
been studied based on more than 300 compressive uniaxial and triaxial strength tests. Obtained results have been typically
interpreted to register peak and residual strength as well as other post-peak parameters. Particular attention was given to the study
of dilatancy. Dilation angle of these intact rocks was fitted to two pioneering but still recent dilatancy models
(Alejano & Alonso 2005; Zhao & Cai 2010), having the feature of being plastic-shear-strain and confinement-stress dependent.
The aim of this paper is to identify various dilation trends according to the type of rock tested and to check the ability of
the models to accommodate actual results. This can be the base to further extend these models to the rock mass scale. To do that,
the dilation angle database produced in our laboratory and some results from other tests taken from literature have been compiled
in order to explain the main behavioral trends of this parameter according to the addressed models. These approaches are
intended to insight rock behavior models emphasizing the post-failure stage as well as to contribute to more realistic but still
practical dilatancy models.
©©2017
2017The Authors.
TheAuthors. Published
Published by Elsevier
by Elsevier Ltd. is an open access article under the CC BY-NC-ND license
Ltd. This
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of EUROCK 2017.
Peer-review under responsibility of the organizing committee of EUROCK 2017
Keywords: post-peak; dilation angle;strain;strain-softening

* Corresponding author. Tel.: +34-986-812-374.


E-mail address: alejano@uvigo.es

1877-7058 © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of EUROCK 2017
doi:10.1016/j.proeng.2017.05.196
I. Pérez-Rey et al. / Procedia Engineering 191 (2017) 394 – 401 395

1. Introduction

For several years, the stability assessment of rock engineering projects was mainly based on avoiding failure,
hence the elastic and peak mechanical behavior of rocks have been extensively studied since they may represent,
a priori, the most determining features to control stability.
Nevertheless, the post-peak behavior of rocks and rock masses plays a relevant role in important fields of rock
engineering where failure take place. For instance mining caving methods, petroleum engineering or the estimate of
plastic zones around underground excavation largely depend on post failure behavior. It is therefore interesting to
have available improved post-failure behavior models to be able to study these phenomena both analytically and
numerically, as highlighted by various authors [1, 2].
The study of rock behavior on the stage beyond the peak of the stress-strain curve started to attract the interest of
the rock mechanics community during late 1960s and early 1970s, when various authors [3, 4] were able to carry out
compressive strength tests with closed-loop servo-controlled testing machines. Techniques to obtain reliable
volumetric strain values after peak were also enhanced by authors like Crouch [5], who improved testing capabilities
and opened new ways to better characterize the post-peak behavior of a rock.
Given that some rocks and rock masses still withstand certain load once overpassed the peak strength, obtaining
a complete characterization of the stress-strain curve is of paramount relevance to understand and control their
mechanical behavior. This permits estimating elastic parameters, peak, residual and evolving failure criteria as well
as post-peak deformability parameters. The elastic part of the curve up to peak strength is standardly obtained in
laboratories worldwide. However, post-peak strain behavior is often disregarded or not computed; due to lack of
interest and equipment to register this part of the tests.
Our rock mechanics laboratory disposes of a compression frame equipped with a servo-controlled confining-
pressure system (Fig. 1) that allows not only to carry out triaxial tests under constant confining pressures, but also to
measure the amount of fluid that is injected / withdrawn from the Hoek’s cell to keep the already mentioned pressure
at a constant value. Taking advantage of this fact, post-peak volumetric strains can be registered and data can thus be
utilized to describe post-yield dilatational properties of rocks.

Fig. 1. General view of the loading frame and servo-controlled confining pressure system at the rock mechanics laboratory
of the University of Vigo.

The aim of this work is to provide an overview and assessment of dilation angle trends according to
recently-developed dilatancy models [7–10]. This study was carried out by analyzing experimental results found in
literature and from more than 300 compressive strength tests performed in different lithology samples at the rock
mechanics laboratory of the University of Vigo.
396 I. Pérez-Rey et al. / Procedia Engineering 191 (2017) 394 – 401

2. Recently-developed models for rock dilation

Dilatancy can be defined as the volume change of a material that is subjected to a shear stress and it is
numerically represented by the dilation angle, ȥ. The value of this parameter can be estimated from experimental
p
laboratory plastic strains, i.e: major principal plastic strain H 1 , equivalent to axial plastic strain and volumetric
p
plastic strain H V through Equation 1 [6]:

HVp
sin \
2H1p  HVp (1)

Two pioneering but still representative models were selected for the assessment of dilatancy, which are
painstakingly described in [7, 8]. Despite new models were recently proposed [9, 10] it was decided to implement
only those shown in this section; firstly, not only for the sake of briefness, but also due to the fact that these models
were already analyzed by the authors only for particular lithologies (Moura coal and Blanco Mera granite) but not
studied for the presented rocks.

2.1. The Alejano & Alonso (2005) dilatancy model

In order to address a historical poor treatment given to rock dilation, the authors proposed a dilatancy model [7]
that depends on the accumulated plasticity of the rock as a consequence of progressive failure and on the confining
stress to which the material is subjected.
Actually, the peak-dilation-angle model is confining-stress dependent and can be estimated by applying
the equations proposed in [7]. To estimate the peak dilation angle, obtaining peak friction angles (‫׋‬peak) for each
confining stage is first required (obtainable by the expression shown in [7]: page 494, Equation 28), for instance
starting from the slope of the Hoek-Brown failure criterion. The decaying trend of dilation angle with respect to
plastic shear strain is also addressed by an exponential expression, (see [7]: page 494, Equation 29) and omitted in
this work for the sake of briefness.

2.2. The Zhao & Cai (2010) dilatancy model

Zhao & Cai [8] proposed a new confining-stress and plasticity dependent dilatancy model based on a nine-
parameters equation (see [8]: page 374, Equation 25), that has shown a reasonable ability to capture dilation trends
of many different rock types. Nevertheless, it presents some inconveniences related to the involved coefficients,
which have a non-clear physical meaning and affect altogether the shape of the dilation trend of the model, that is,
different sets of parameters would yield practically equal results.

3. Computing dilation from UCS and triaxial strength tests

The development of uniaxial compressive (UCS) and triaxial strength tests is a very common practice to obtain
parameters intended to correctly characterize the behavior of rocks and rock masses. In our laboratory, we have
available a good number of triaxial test results performed to insight post-failure behavior [11, 12] besides helping on
the development of several engineering projects that required elastic and post-peak information to obtain reliable
rock characterization and representative numerical models.
Equation 1 allows to compute dilation angle from strain values directly obtained from a complete stress-strain
curve. These strains can be analytically and graphically obtained. For the development of this work, the second
method was utilized. It requires depicting the so-called irrecoverable volumetric strain locus, which is the envelope
curve of the minimum strain points in the axial strain-volumetric strain graph. By considering the slope of first
loading cycles, it is possible to relate the elastic and plastic components of each pair of axial strain–volumetric strain
point (Fig. 2).
I. Pérez-Rey et al. / Procedia Engineering 191 (2017) 394 – 401 397

Fig. 2. Graphical estimate of axial and volumetric plastic shear strains from a stress-strain curve obtained for a triaxial strength test on Blanco
Mera granite at ı3 = 4 MPa (modified from [12]).

Plastic shear strain (Ȗ p) was the selected plastic parameter and it is defined as the difference between the major
and minor principal plastic strains (Eq. 2). This is based on a practical approach that considers the plastic parameter
as a function of internal variables, for planar deformations (H2 = H2 p = 0).

Jp H 1p  H 3p (2)

According to this definition of the plastic shear strain, the foregoing presented dilatancy models can be defined
all along the post-failure stage of every test curve by means of the following general form:

\ \ V 3 , J p
(3)

Resorting to Eqs. 1 and 3 and considering that the confining stress applied to the sample, for homogenous
deformation conditions, H2 = H3 and so HV = H1 + 2H3 it is possible to produce point clouds for each confining pressure
consisting of i pairs of values (ȥi, Ȗ pi). The mentioned dilatancy models will be thus applied to analyze their ability
to capture the observed trends of the obtained dilation angles for each plasticity level at different confining levels.

4. Dilation results from different types of rocks

Recently developed dilatancy models highlighted a dependence of the evolving dilation angle with the rock
nature, particularly related to grain size and uniaxial compressive strength [8]. A relevant range of different rocks
was tried to be covered. So, various igneous, metamorphic and sedimentary rocks were analyzed. Three granites
(two moderately weathered ones, like Amarelo País and Vilachán and a crystalline one, Blanco Mera granite)
398 I. Pérez-Rey et al. / Procedia Engineering 191 (2017) 394 – 401

represent igneous rocks. An orthogneiss (Noia gneiss) and a garnet amphibolite (Touro amphibolite) represent
foliated metamorphic rocks together with a very homogeneous marble (Carrara marble) as a crystalline
metamorphic one. Finally, a coal (Moura coal, described in [13]) and a limestone (Indiana limestone) represent
sedimentary materials.
Dilation angle results are represented in Fig. 3 with respect to plastic shear strain for these rocks. Three datasets,
corresponding to the highest, medium and lowest values of the available confining pressures for each testing series
were considered. The Alejano & Alonso model [7] was first fit to the available data series with the help of MatLab.
Despite capturing the decaying trend of dilation angle reasonably well, it seemed to be not very appropriate for
the estimation of peak dilation angle, a fact that had been previously reported [9, 11]. Only the sedimentary and
partially metamorphic rocks, in particular those presenting foliation, yielded reasonable fitting curves for
the complete stages. Peak dilation angle was not well captured by the current model for the grained rocks (granites
and marble). The model fit for this rocks gives very mildly decaying curves, which do not follow observed trends.
In the same way as previously explained, dilation angle results are represented in Fig. 4 with respect to the plastic
shear strain. Now, the Zhao & Cai model [8] was fitted to the available datasets again with the help of MatLab,
yielding reasonably accurate results while capturing peak dilation angle and confining-stress dependency.
The decaying trends were also well displayed.

Fig. 3. Adjusted fits for the Alejano & Alonso (2005) model [7] for metamorphic rocks (gneiss and amphibolite) and for sedimentary rocks (coal
and limestone). Dilation angle, ȥ [°] is represented against the plastic shear strain, Ȗ p [mstrain] for the mentioned lithologies. The selected
confining pressures were those corresponding to the highest, medium and lowest available ones.
I. Pérez-Rey et al. / Procedia Engineering 191 (2017) 394 – 401 399

Fig. 4. Zhao & Cai (2010) model [8] fits for eight rocks, including igneous (Amarelo País, Blanco Mera and Vilachán granites), metamorphic
(Carrara marble, Noia gneiss and Touro amphibolite) and sedimentary ones (Moura coal and Indiana limestone). Dilation angle, ȥ [°] is
represented against the plastic shear strain, Ȗ p [mstrain] for the mentioned lithologies.

5. An assessment of dilation models for the studied rocks

A general assessment of the ability to capture dilation trends by the proposed dilatancy models was carried out.
Data shown in Table 1 gives an idea of the fitting goodness of both models representing dilatant behavior of
different rocks. This is set out through the coefficients of determination, R2, representative of the two models
400 I. Pérez-Rey et al. / Procedia Engineering 191 (2017) 394 – 401

defining equations. Since dilation angles are obtained via laboratory-testing strain data, dispersion is quite visible
although it is considered to be reasonably representative. Since the original dilation results come from various tests
for every confinement level (except for the case of coal), not very high regression coefficient values are expectable.
In this way, and also accounting for the variability in the response of rock samples, values of R2 within the range 0.5
to 0.9 can be considered reasonably accurate and within the range 0.1 to 0.5 acceptable, in rock mechanics terms. In
any manner, they are in the range of those observed in already published dilatancy studies [7, 8]. Remark that,
dilation angle tended to be considered in practical engineering numerical models, typically equal to the friction
angle (an associated flow rule) or 0º. So, the presented models, even though not too accurate, they largely improve
traditional practices.

Table 1. Coefficients of determination, R2, for every rock dataset at each confining pressure level for the two studied models (R12 = Alejano &
Alonso (2005); R22 = Zhao & Cai (2010)). (und. = undetermined coefficient).
Confining pressure (MPa)
Rock 0.2 1 2 4 6 10 12
Amarelo País granite R12 = und. R12 = und. R12 = und.
— — — —
R22 = 0.71 R22 = 0.77 R22 = 0.84
2 2
Blanco Mera granite R1 = und. R1 = und. R12 = und.
— — 2
— 2

R2 = 0.75 R2 = 0.82 R22 = 0.77
Vilachán granite R12 = und. R12 = und. R12 = und.
— — 2
— 2

R2 = 0.52 R2 = 0.69 R22 = 0.79
Carrara marble R12 = und. R12 = und. R12 = und.
— — 2
— 2

R2 = 0.80 R2 = 0.71 R22 = 0.89
2 2 2
Noia gneiss R1 = 0.39 R = 0.29 R1 = 0.51
— — 2 2
— —
R2 = 0.53 R = 0.33 R22 = 0.18
Touro amphibolite R12 = 0.30 R12 = 0.23 R12 = 0.21
— — 2
— 2

R2 = 0.61 R2 = 0.42 R22 = 0.68
Moura coal R12 = 0.39 R12 = 0.66 R2 = 0.65
2 2
— — — —
R2 = 0.99 R2 = 0.97 R2 = 0.91
Indiana limestone R12 = 0.11 R12 = 0.11 R12 = 0.08
— — 2
— 2

R2 = 0.51 R2 = 0.10 R22 = 0.31

It has been observed that the Alejano & Alonso model [7] displays, not too badly, peak and post-peak dilatant
behavior of sedimentary rocks —as previously observed by the authors—, but the proposed formulation is not able
to capture the dilatant behavior of another rock types (in particular, the marble and the three granites), a fact that had
also been observed for other crystalline rocks when studying this model [9]. Nevertheless, the fit is again acceptable
for two of the metamorphic rocks analyzed, those presenting lower dilation in terms of peak dilatancy. For
the observed results, the model present confining-stress and plasticity dependence.
The more sophisticated Zhao & Cai model [8], keeps the ability of more or less representing the complete variety
of studied rocks. Peak dilation angles and decaying trends are reasonably well captured in the case of granites,
marble and amphibolite. Best fits were particularly encountered for the Moura coal, a fact that might be attributed to
the relatively-small size of the analyzed datasets.

6. Conclusions

Taking advantage of the available information on elastic and plastic parameters recovered from laboratory
compression tests, a study on the performance of two dilation models related to this experimental data were
I. Pérez-Rey et al. / Procedia Engineering 191 (2017) 394 – 401 401

carried out. The Alejano & Alonso [7] and Zhao & Cai [8] models were addressed since they are pioneering
formulations, even if extending these study to recently publish models [9, 10] is envisaged.
The Alejano & Alonso [7] model was found to display good results for sedimentary rocks, both for peak and
post-peak dilation angles. Two of the studied metamorphic rocks, the amphibolite and the gneiss, yielded acceptable
results for both peak and post-peak states. This result is attributed to a good correlation between the model
estimation of peak dilation angles and the observed ones. Lower dilation and failure mechanisms through
preexisting weakness planes, specially encountered in foliated rocks are thought to be the main reason behind this
fact, as previously observed in [14]. Results for crystalline igneous and metamorphic rocks (granites and marble),
were not correctly captured by the model [7], generating meaningless coefficients according to the observed natural
decaying data trends. In particular, peak dilatancy in hard crystalline rocks is not at all well reflected by this
formulation [7].
The Zhao & Cai model displayed reasonable results for all the selected rock types. Peak dilation angles and
decaying trends are well captured. In the case of crystalline rocks, higher peak dilation angles were observed, being
quite lower for the case of foliated-metamorphic and sedimentary rocks. This fact may be attributed to different
failure mechanisms: through grains breaking, in the case of crystalline rocks and by following preexisting weakness
planes in metamorphic foliated rocks. When sufficient information is available, this formulation is interesting. Its
main back-draw, from the authors’ scope, is related to the fact that various sets of parameters would yield practically
equal results.
This study may be of help regarding to further study and develop a new reasonably accurate but still
non-excessively-complex dilation model, complementing and enhancing two recently proposed models [9, 10]. This
may involve to improve laboratory equipment to obtain more precise results, especially in the range of low
confinements.

Acknowledgments

The authors thank the Spanish Ministry of Economy and Competitiveness for funding this research, awarded
under Contract Reference No. BIA2014-53368P, partially financed by means of ERDF funds from the EU.

References

[1] X.G. Zhao, M. Cai, Influence of plastic shear strain and confining-dependent rock dilation on rock failure and displacement near
an excavation boundary, Int. J. Rock Mech & Min. Sci. 47(5) (2010) 723–738.
[2] G. Walton, M. Diederichs, A. Punkkinen, J. Whitmore, Back analysis of a pillar monitoring experiment at 2.4 km depth in the Sudbury Basin,
Canada, Int. J. Rock Mech & Min. Sci. 85 (2016) 33–51.
[3] F. Rummel, C. Fairhurst, Determination of the post-failure behavior of brittle rocks using a servo-controlled testing machine, Rock Mech.
2 (4) (1970) 189–204.
[4] W.R. Wawersik, W.F. Brace, Post-failure behavior of a granite and diabase, Rock Mech. 3 (2) (1971) 61–85.
[5] S.L Crouch, Experimental determination of volumetric strains in failed rock, Int. J. Rock Mech. & Min. Sci. 7 (6) (1970) 589–603.
[6] P.A. Vermeer, R. de Borst, Non associated plasticity for soils, concrete and rock, Heron 29 (3) (1983) 1–64.
[7] L.R. Alejano, E. Alonso, Considerations of the dilatancy angle in rocks and rock masses, Int. J. Rock. Mech. Min. Sci. 42 (2005) 481–507.
[8] X.G. Zhao, M. Cai, A mobilized dilation angle model for rocks, Int. J. Rock. Mech. Min. Sci. 47 (2010) 368–384.
[9] G. Walton, M.S. Diederichs, A new model for the dilation of brittle rocks based on laboratory compression test data with separate treatment
of dilatancy mobilization and decay, Geotech. Geol. Eng. 33 (2015) 661–779.
[10] M. Rahjoo, E. Eberhardt, A simplified model for modeling the inelastic behavior of rock, ARMA 50th US Rock Mechanics/Geomechanics
Symposium 760 (2016).
[11] J. Arzúa, L.R. Alejano, Dilation in granite during servo-controlled triaxial tests, Int. J. Rock. Mech. Min. Sci. 31 (2013) 43–56.
[12] J. Arzúa, Study of the mechanical behavior of intact and jointed rocks in laboratory with particular emphasis on dilatancy, PhD Thesis.
University of Vigo,Spain, 2016.
[13] T.P. Medhurst, E.T. Brown, A study of the mechanical behavior of coal for pillar design, Int. J. Rock. Mech. Min. Sci. 35 (1998) 1087–105.
[14] I. Pérez-Rey, J. Arzúa, L.R. Alejano, J. Barbiero, G. Walton. A lab-testing based geomechanical characterization of metamorphic rocks
focusing on post-failure behavior, in: L.R. Alejano, A. Perucho, C. Olalla, R. Jiménez (Eds.), Proceedings of the EUROCK 2014. Rock
Engineering and Rock Mechanics: Structures in and on Rock Masses (2014) pp. 82.

You might also like