Engineering Geology 166 (2013) 81–89

Contents lists available at ScienceDirect

Engineering Geology

j ourna l homepage: www.e lsev ie r .com/ locate /enggeo

Determination of mechanical properties of flysch using laboratory methods

G.P. Tziallas a, H. Saroglou b,⁎, G. Tsiambaos b

a Lahmeyer International GmbH, Germanyb Dept. of Geotechnical Engineering, School of Civil Engineering, National Technical Univ. of Athens, Greece

⁎ Corresponding author. Tel.: +30 210 7722440; fax: +E-mail address: saroglou@central.ntua.gr (H. Saroglou

0013-7952/$ – see front matter © 2013 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.enggeo.2013.09.002

a b s t r a c t

a r t i c l e i n f o

Article history:Received 18 June 2013Received in revised form 22 August 2013Accepted 2 September 2013Available online 8 September 2013

Keywords:Heterogeneous rock massComposite materialFlyschSiltstone contentUniaxial compressive strengthTriaxial compressionGSI

The paper deals with the determination of the mechanical properties of heterogeneous rocks in the laboratoryand proposes a methodology for predicting the rock mass strength of flysch formations consisting of siltstone–sandstone alternations in different proportions. In order to simulate such formations, composite specimens com-prising superimposed disks of intact sandstone and siltstone material with different thickness ratios were pre-pared. Wave velocity (P and S), uniaxial and triaxial compressive strength and Young's modulus of thecomposite sampleswere determined in the laboratory. According to the laboratory results, the uniaxial compres-sive strength and Young's modulus decrease with the increase of the siltstone percentage in the specimens, andyield the strength of siltstone when the siltstone percentage in the rock is equal to 37%. Additionally, the triaxialcompression tests on composite samples revealed a significant decrease of the mi parameter for a siltstone per-centage of only 17%. A comparison is made between the results of the present study with similar laboratory andempirical references from literature. Finally, it was possible to suggest how the relation between uniaxial com-pressive strength, σci, and siltstone percentage, sl, may differ according to the ratio of the compressive strengthof the two main rocks, weak and strong, which comprise the heterogeneous rock.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Heterogeneous or composite rock masses can be characterized asformations, which consist of at least two lithological units with differentengineering behaviors and properties. Heterogeneous rock masses ex-hibit a structure of alternating layers of hard and weaker rocks withvarying thickness and they are usually of sedimentary origin.

Flysch and molasse are typical examples of such rocks. They arecharacterized by rhythmic alternations of sandstone and pelitic (finegrained) rocks such as siltstones, marls, shales and clay shales. Con-glomerates and limestones can also be present.

A number of engineering projects like tunnels, foundations and rockslopes are constructed in heterogeneous rock masses in countries (e.g.Greece) where such rock masses are common. Due to the highly aniso-tropic behavior that they exhibit, the engineering uncertainties and riskincrease disproportionately, often leading to highly conservative engi-neering judgment and overpriced solutions. Determining the mechani-cal properties of heterogeneous formations both in uniaxial and triaxialconditions is vital for the selection of characteristic values for the designand construction of such works.

In the present study, an attempt is made to estimate the intact rockproperties of heterogeneous flysch rock (alternations of sandstone andsiltstone) using composite rock specimens for laboratory tests.

30 210 7723428.).

ghts reserved.

2. Literature review

Although the determination of properties of heterogeneous rockmasses constitutes a challenging engineering problem, the available lit-erature is relatively limited. Only in the last decade, owing to the in-creasing underground infrastructure projects, there has been interestin the research on such rock masses.

Duffault (1981) published a study on modeling and simulating het-erogeneous rock masses introducing the term “sandwich (single ormultiple) rock mass”.

Greco et al. (1992) and Greco (1994) conducted a study for deter-mining the strength and failure mechanism of composite rocks in thecontext of the stability analysis of the columns and masonry walls of aCathedral, built with stones of different rock types.

Goodman (1993) had emphasized that any combination of morethan one lithological types of rock exhibiting different properties im-poses a complex geotechnical engineering problem.

Marinos and Hoek (2001) and Hoek et al. (2005) suggested empiri-cal charts for estimating GSI, σci and mi for heterogeneous rock masses,based on the lithology, structure and discontinuity surface conditions.

Vlasov andMerzlyakov (2004) studied the deformability of the strat-ified rock mass considering it as a transversally isotropic medium usingthe asymptotic method of averaging, whereas Lydzba et al. (2003) stud-ied the conditions at failure of a layered sandstone/claystone rock massformation.

Liang et al. (2007) conducted laboratory experiments on natural lay-ered specimens consisting of rock salt and anhydrate concluding that

82 G.P. Tziallas et al. / Engineering Geology 166 (2013) 81–89

the overall strength is determined by the behavior of the weaker com-ponent. Zainab et al. (2007) conducted tests on composite specimensconsisting of highly weathered sandstone and siltstone from KualaLumpur, Malaysia. The laboratory method that they have proposed issimilar with the one that is used in the present study.

Saroglou and Steiakakis (2010) investigated the variation of uniaxialcompressive strength of flysch intact rock samples, which presentedlayering due to existence of thin siltstone and sandstone alternations.

Budetta and Nappi (2011) have assessed the rock strength of a het-erogeneous formation, comprised of sandstones with argillaceousmarls, using the sandstone/pelite ratio (S/P) and the classification ofGSI proposed by Marinos and Hoek (2001).

3. Laboratory testing

3.1. General

The laboratory method comprises the preparation and testing ofrock specimens that were made up of superimposed disks of intactsandstone and siltstone material. The disks were formed in differentthickness ratios of siltstone–sandstone (sl–st)material so as to simulatedifferent siltstone and sandstone proportions in the flysch formation aspresented in Fig. 1 (Tziallas, 2010).

The proportion of siltstone to sandstone in heterogeneous rock for-mations in the field is not directly analogous to the proportion of silt-stone to sandstone in the laboratory, due to the lack of continuity ofthe alternations of siltstone to sandstone in the specimens. Thus, het-erogeneous formations in the field with a specific proportion (definedby thickness of alternating layers) of siltstone and sandstone, willhave a lower proportion in the laboratory specimens, since the alter-ations of layers in the specimen result in more sandstone than siltstonelayers. For example, a heterogeneous rock mass with equal proportionsof siltstone to sandstone (sl:st = 1:1)will have a percentage of siltstoneequal to 30% in the laboratory specimen, as shown in Specimen B inFig. 1. The siltstone percentage is indicated as sl and sandstone percent-age as st (Tsiambaos, 2010).

The heterogeneous rock mass types with different proportions ofsiltstone to sandstone (equal to 1:5, 1:1, 3:1, 5:1) most commonly

Fig. 1. Heterogeneous flysch with different proportions o

encountered in nature are shown in Fig. 1. These rock mass typeswere modeled in the laboratory and the proportion of siltstone to sand-stone is shown in Fig. 1.

The composite specimens with varying siltstone–sandstone ratiowere subjected to wave propagation tests, uniaxial compression testswith determination of elasticitymodulus aswell as triaxial compressiontests.

All the laboratory tests were determined according to the ISRM(2007) procedures.

3.2. Description of materials

The samples that were used for the laboratory tests were collectedfrom Kalydona tunnel of the Ionia Odos Highway, in the area ofEvinochori village in central Greece, where the Ionian flysch of theGavrovo–Tripoli geotectonic unit is encountered. The samples were ei-ther prepared from intact sandstone and siltstone blocks or from bore-hole cores. More specifically: a) Light gray to gray, medium to coarsegrained, fresh, sandstone with no signs of anisotropy. Siltstone interca-lations in the form of lenses were encountered in the sandstone but didnot exceed 1–2% of the specimen volume. The mean unit weight of thesandstone specimens was γst = 25.4 Mg/m3.

The siltstone used was gray-green to dark gray, homogeneous andfresh, sensitive to water, and displayed macroscopically swelling andshrinkage and loosening finally leading to slaking. Siltstone sampleswith thin sandstone intercalations were rejected during the selectionof disks for the preparation of the composite specimens. The meanunit weight of the siltstone specimens was γsl = 24.4 Mg/m3.

3.3. Preparation of the specimens

Generally, the greater difficulty on testing heterogeneous rockmasses is to obtain “intact” core samples for determining the uniaxialor triaxial compressive strength in the laboratory. Moreover, laboratorytests carried out on natural heterogeneous samples often result in alower strength value due to sampling and preparation disturbance(Saroglou and Steiakakis, 2010). Marinos and Hoek (2001) suggestthat using the results of such tests in the Hoek–Brown criterion “will

f siltstone and sandstone in the field and laboratory.

83G.P. Tziallas et al. / Engineering Geology 166 (2013) 81–89

impose a double penalty on the strength (in addition to that imposed byGSI) andwill give unrealistically low values for the rockmass strength”.

For this reason, in the present study it was decided to form artificialcomposite specimens consisting of superimposed disks of intact rockmaterial. Each separate disk was cored from intact blocks and then cutin the respective thickness using a water-cooled diamond disk. Thesamples were extracted always perpendicular to bedding planes. Fur-thermore, since inherent anisotropy of sedimentary rocks due to bed-ding influences the mechanical properties, in the present study all thelaboratory tests were conducted perpendicular to bedding. The surfacesof the disks were then flattened and polished with a diamond wheel.Since the siltstone specimens were very sensitive when exposed towater and the vibrations of the trimming equipment, they were cutand polished by using a PVC coating. Following this procedure it hasbeen possible to prepare thin undisturbed siltstone disks (theminimumthickness was 1.13 cm) for simulating the specimenswith low siltstonepercentage.

The final cylindrical specimens with the superimposed disks wereprepared with length to diameter ratio between 2.0 and 2.5. Twenty-one composite specimens were prepared in total, comprising twodisks of sandstone and an intermediate disk of siltstone except of spec-imen E15 that consisted of 3 disks of sandstone and 2 intermediate silt-stone disks. Five sandstone and four siltstone specimens were alsoprepared in order to assess the properties of the original materials.The siltstone percentage of each specimen has been calculatedvolumetrically.

It should be noted that no adhesive material was used in order toconnect the separate disks in the composite sample. Initially, plasterwas used for this purpose, as proposed by Zainab et al. (2007), but plas-ter failed at a lower load during the uniaxial compression tests, causingthe disks to displace and introducing non-uniform stress distributionthroughout the disk interfaces. Considering the pattern of the fracturesurfaces and the fact that no displacements of the disks were observedeven in higher compressive loads, the use of a bonding material wasnot considered necessary.

A high-speed camera (250 fps) was used in order to capture thefracture of the composite rocks under uniaxial compression and facili-tate the understanding of the failure mechanism. An example of a cap-tured frame at failure is shown in Fig. 2.

3.4. P and S wave velocities

The ultrasonic pulse method was used to determine the velocity ofP- and S-waves of the intact and composite heterogeneous rock sam-ples. Thirty-two wave propagation tests were carried out in total, four-teen (14) of which were performed on sandstone, five (5) on siltstone

Fig. 2. Fracture of composite sample (specimen A of Figure 1) in compression test(captured by high speed camera).

and thirteen (13) on composite specimens. The correlation of the longi-tudinal and shear wave velocity for all samples is described by the fol-lowing equation:

Vp ¼ 1:48·Vs R2 ¼ 0:85� �

:

The correlation of the wave velocity with the siltstone percentage isshown in Fig. 3 and the results are summarized in Table 1.

The average wave velocities of the two original rock materials areVP st = 4290 m/s, VS st = 2867 m/s and VPsl = 2950 m/s, VSsl =2067 m/s for sandstone and siltstone respectively.

As expected, the wave velocities in the composite specimen de-crease with the increase of the siltstone percentage (sl). The correlationis linear and is described by the following equations:

VP ¼ −13:28 slþ 4250 R2 ¼ 0:87� �

ð1Þ

VS ¼ −8:16 slþ 8246 R2 ¼ 0:74� �

: ð2Þ

The linear decrease of the velocity of the heterogeneous samples(0 b sl b 100%) and the fact that their value is not lower than the veloc-ity of the siltstone samples (sl = 100%), indicate that the interfaces ofthe superimposed disks are in contact during the test, thus the inter-faces have no effect on the wave propagation. This fact demonstratesthat the composite specimens do not present any deficiencies thatcould lead to significant divergence in thedeterminationof themechan-ical properties.

3.5. Uniaxial compressive strength and Young's modulus

The uniaxial compressive strength was determined according to theISRM (2007) procedure. The tests were carried out using a servo-controlled loading frame with a 1.5 MN capacity. The stress rate waskept constant in a range of 0.17–0.35 kN/s and failure occurred within8–10 min from the start of loading.

The axial strain (εα) during uniaxial compression tests was mea-sured using an LVDT mounted at the mid-height of the composite rockspecimen. The samples were cylindrical with a height to diameterratio of about 2.0 to 2.5. Prior to testing, the samples were air-dried atroom temperature.

The results of the uniaxial compression tests are presented in Table 2.

3.5.1. Uniaxial compressive strengthThe intact sandstone has a mean compressive strength σci st =

95.3 MPa while the siltstone σci sl = 43.0 MPa.

Fig. 3. Correlation of the longitudinal and transverse wave velocity with siltstone percent-age (sl).

Table 1Results of the wave propagation laboratory tests.

Specimen type Specimen no. sl (%) Vp (m/s) Vs (m/s)

Sandstone St1 0.00 4290⁎ 2867⁎

St2 0.00St3 0.00St4 0.00St5 0.00

Composite specimens E14 7.0 4110 2588E7 9.8E9 10.3 4390 2792E8 10.4 3899 2670E5 16.4 3762 2585E11 18.0 3886 2802E15 20.6 3620 2358E12 23.2 3891 3056E13 24.4E4 26.9 3884 2678E3 33.3Ε1 34.2E2 34.6 3892 2483E6 36.8 3918 2618E10 46.8 3542 2387Ε16 65.0 3237 2201Ε17 73.7 2898 2137

Siltstone Sl1 100.0 2950⁎⁎ 2067⁎⁎

Sl2 100.0Sl3 100.0Sl4 100.0

⁎ Mean value of 14 tests.⁎⁎ Mean value of 5 tests.

Fig. 4. Compressive strength, σci, reduction curve relevant to siltstone percentage (sl).

84 G.P. Tziallas et al. / Engineering Geology 166 (2013) 81–89

Based on the results of the tests, it is suggested that the increase insiltstone percentage results in a decrease of the strength, while thestrength values of the composite specimens range between the strengthof the two original rock materials (Figure 4). This applies up to a silt-stone percentage equal to sl = 35–37%, while for greater percentageof siltstone the composite specimens display the average strength of in-tact siltstone i.e. σci = 43 MPa.

Two different sections can bedistinguished in the strength reductioncurve shown in Fig. 4, a section of the low siltstone percentage(0% ≤ sl ≤ 37%) and a section where the specimens exhibit a siltstone

Table 2Uniaxial compression and Young's modulus (Ei) test results.

Specimen type Specimen no. sl (%) σci (MPa) Εi (MPa)

Sandstone St1 0.00 85.4St2 0.00 103.8St3 0.00 93.6 39,250St4 0.00 97.2 42,810St5 0.00 93.5

Composite specimens E14 7.0 70.2 29,340E7 9.8 67.2 24,780E9 10.3 94.1 39,240E8 10.4 80.5 29,430E5 16.4 77.4E11 18.0 58.5 19,050E15 20.6 61.6 25,010E12 23.2 62.3 16,070E13 24.4 57.0 16,400E4 26.9 55.5E3 33.3 40.6Ε1 34.2 56.5E2 34.6 36.8E6 36.8 42.1 19,620E10 46.8 51.4Ε16 65.0 28.0 19,010Ε17 73.7 42.4 12,360

Siltstone Sl1 100.0 53.7Sl2 100.0 48.7Sl3 100.0 34.6Sl4 100.0 37.4

behavior (37% b sl ≤ 100%). In thefirst section, the decrease of strengthof intact sandstone is intense and is described by an exponential de-scending curve:

σci ¼ 95:3 e−0:022 sl: ð3Þ

The uniaxial compressive strength of the intact sandstone materialexhibits a strength reduction equal to 20% even when a thin intercala-tion of siltstone is present (sl equal to 10%).

In the second section, the strength of the composite heterogeneousmaterial remains constant and equal to the strength of intact siltstone.Thus, if a percentage of siltstone equal or greater to 37% is present,which corresponds to a siltstone–sandstone proportion of 1:1 in thefield (as shown in Figure 1), theweakmember determines the heteroge-neous rock behavior and strength. For this range of siltstone percentage(greater than 37%) the presence of the strongermember (sandstone) haspractically no effect on the overall strength of the heterogeneous rock.

It is noted that specimen E15, plotted in Fig. 3 with an empty trian-gle, comprised five superimposed disks and matches perfectly on thepredicted exponential curve leading to the conclusion that the factorthat mainly affects the compressive strength of the flysch rock mass isthe percentage of siltstone and sandstone materials rather than thenumber of layers which are present.

3.5.2. Fracture pattern of composite specimensThe fracture pattern in the composite specimens changes according

to the variation of the siltstone percentage sl. For specimens with silt-stone percentage between 0% and 22%, the fracture is brittle and clearand the fracture surfaces propagate through the entire specimen(Figure 5). For a range of 22% b sl ≤ 37% the fracture pattern is moreductile and the fracture surfaces propagate from the siltstone into thesandstone disks in the form of cracks. The failure surface is formedalong part of the specimen and thus this is not separated during failure(Figure 5b). As sl increases (N37%) and the specimen acquires thestrength of the siltstone material, the fracture surfaces appear mainlyin the siltstone disk and they propagate in the form of faint cracks or su-perficial spalling (flaking) in the sandstone disks (Figure 5c).

This difference in the fracture patterns can be explained by the stressdistribution throughout the rock specimen. Assuming the compatibilityof deformations of the interfaces of the separate disks and consideringthe fact that siltstone and sandstone radial strains differ due to the var-iation of Poisson ratio, the common uniaxial strain introduces the stresspattern that is presented in Fig. 6.

Consequently, the sandstone disks introduce confinement stresseson the siltstone's upper and lower end preventing it from displayingits original low strength. The opposite effect is exerted by the siltstonedisk to the sandstone disks, thus the thinner the siltstone disk is(lower sl %), the greater the confinement and the higher the specimen's

Fig. 5. a. Specimen E8with sl = 10.4%, b. Specimen E13with sl = 24.4%, c. Specimen E10with sl = 46.8%.

Fig. 6. Radial stress distribution on the disk interfaces of the composite sample.

Fig. 7. Correlation of Ei and σci of the composite samples.

85G.P. Tziallas et al. / Engineering Geology 166 (2013) 81–89

overall compressive strength. When the siltstone disk has a thicknesswhich results to the threshold siltstone percentage (sl = 37%), the ben-eficial influence of sandstone ceases to exist and the specimen'sstrength drops to the level of intact siltstone.

3.5.3. Young's modulusTheYoung'smoduluswas determined at 50% of the uniaxial compres-

sive strength. The Young'smodulus for sandstone and siltstonewas equalto Ei–st = 45.0 GPa and Ei–sl = 17.0 GPa respectively, as presented inTable 2. Fig. 7 presents the correlation of uniaxial compressive strengthwith the modulus of elasticity. In homogeneous materials the regressioncurve of such a correlation is linear. The MR value (Ei / σci) for sandstone

and siltstone ranges between 200–350 and 350–400 respectively,according to Hoek andDiederichs (2006). Based on the laboratory resultson heterogeneous composite samples an average MR value equal to 390was determined, which agrees well with those presented in Hoek andDiederichs (2006).

Due to theheterogeneity of the samples, an exponential curve can beapplied described by Eq. (4):

E ¼ 8620 e0:015σci 4

where E and σci in MPa.

Table 3The modified and the original siltstone percentage with the respective Ei values.

Specimen type Specimen no. sl (%) Modified siltstonepercentage slc (%)

Ε (MPa)

Sandstone St3 0.0 0.0 39,200St4 0.0 0.0 42,800

Compositespecimens

E14 7.0 26.3 29,300E7 9.8 25.1 24,700E9 10.3 23.8 39,200E8 10.4 22.5 29,400E11 18.0 58.4 19,000E15 20.6 43.3 25,000E12 23.2 84.1 16,000E13 24.4 85.5 16,400E6 36.8 92.3 19,600Ε16 65.0 100.0 19,000Ε17 73.7 100.0 12,300

86 G.P. Tziallas et al. / Engineering Geology 166 (2013) 81–89

An interesting correlation is the one of the Young's modulus versusthe siltstone percentage sl. For that relationship, a discussion arises onthe laboratory procedure of measuring the axial deformations. TheLVDT device that was used for the laboratory tests is able to measurethe strains of a fixed part of 50 mm of the mid-height of the specimen.However, using composite samples, the composition of the central partof the specimens is constantly changing due to the variation of the silt-stone disk thickness (Figure 8). As a result, if the LVDT ismounted solelyon the siltstone intercalation, it would be mistaken to assume that theelastic modulus represents the whole composite specimen. To over-come this issue, the “modified siltstone percentage”, slc, is introducedthat represents the siltstone percentage that refers to the part of thespecimen where the LVDT device measures (50 mm distance of mea-suring points). The modified siltstone percentage, slc, compared to theoriginal sl togetherwith themeasured Ei values are presented in Table 3.

Fig. 9 presents the decrease of the Young's modulus with the modi-fied siltstone percentage. The equation is an exponentialfit described bythe equation:

E ¼ 41:79e0:01Slc; R2 ¼ 0:83: ½5�

The decrease of the elasticity modulus is more gradual compared tothat of the compressive strength (Figure 4). This is anticipated, as theelasticity modulus depends on the overall deformability pattern duringloading and not only by the moment of fracturing at failure.

3.5.4. Comparison with the elastic theoryThe above laboratory data are in agreement with the elastic defor-

mation theory of heterogeneous materials. With the assumption thatthe stress–strain relationship is linear elastic, a rock specimen, thatcarries a horizontal discontinuity of lower elasticmodulus than the orig-inal intact rock, is considered. The specimenwith total height “H”, intactrock height “h” and discontinuity thickness “tf” is under the compres-sive stress “σ” (Figure 10). The elasticity modulus of the intact rock is“Ei” and that of the discontinuity “ED”.

The contribution of the intact rock to the overall deformation equalsto δi ¼ σ �h

Ei:

The contribution of the discontinuity to the overall deformationequals to δD ¼ σ �≤ t f

ED:

According to the above:

δtotal ¼ δi þ δD ¼ σ � hEi

þ σ � t fED

⇔εδtotalH

¼ σEi

� hHþ σED

� t fH⇔

ffiffiffi2

p

Etotal ¼σε¼ 1

1Ei

� hH

� �þ 1

ED� t fH

� � :ð6Þ

Fig. 8. Difficulties in the measurement of You

If the intact rock is sandstone and instead of a discontinuity, a silt-stone disk is assumed, Eq. (6) transforms to:

Etotal ¼1

1Esandstonei

� EsandstoneH

!þ 1

Esiltstonei

� hsiltstoneH

! or

Etotal ¼1

Ssandstone=100Esandstonei

!þ Ssiltstone=100

Esiltstonei

! ð7Þ

where Ei, ED, Etotal in GPa,Ssiltstone = sl (%) andSsandstone = 100-sl (%).

Based on Eq. (5) the Young's modulus for modified siltstone per-centage was determined and presented in Table 3. Using Eq. (7) withEi st = 41.8 GPa and Ei sl = 15.3 GPa, as determined from the laborato-ry tests, the values of Etotal were calculated and presented in Table 4.Both the laboratory results curve and that predicted from the elasticitytheory are presented in Table 4 and in Fig. 11. It is obvious that the dif-ferences are marginal which can lead to the conclusion that elastic the-ory as described in Eq. (7), can give a very realistic approach concerningthe elasticity modulus of heterogeneous rock masses.

3.6. Triaxial compressive strength

Triaxial compressive tests were carried out using a 70 MPa capacitytriaxial cell placed in a 1.5 MNcapacity-loading frame. The triaxial cell is

ng's modulus on composite specimens.

Fig. 9. Correlation of Young's modulus Ei and the corrected siltstone percentage slc.

Table 4Comparison of the measured Ei values to the predicted by the elastic theory ones.

Specimen Siltstonepercentage sl (%)

Elasticity theoryΕtotal (MPa)

Laboratory valuesElab (MPa)

Deviation(%)

Sandstone 0.0 45,340 41,790 7.84Compositespecimens

10.0 38,860 37,810 2.7120.0 34,000 34,210 −0.6130.0 30,220 30,960 −2.4240.0 27,200 28,010 −2.9750.0 24,730 25,340 −2.4960.0 22,670 22,930 −1.1770.0 20,920 20,750 0.8380.0 19,430 18,780 3.3690.0 18,130 16,990 6.31

Siltstone 100.0 17,000 15,370 9.58

87G.P. Tziallas et al. / Engineering Geology 166 (2013) 81–89

a Hoek–Franklin cell for specimens of 54 mm diameter (NX size). Fourdifferent confining pressures applied during the triaxial tests andthese were 5.0, 10.0, 20.0 and 30.0 MPa.

Since the scope of thiswork is to comment on the parametersσci andmi of the Hoek–Brown criterion for use in heterogeneous materials, therange of confining pressures of 0 b σ3 b 0.5σci proposed by Hoek andBrown (1980) were used. The specimens were first subjected to the re-quired confining pressure and then the axial load was applied until thespecimen failed.

The parameters of the heterogeneous specimen were derived byfitting the Hoek–Brown failure criterion on the data presented inTable 5. The uniaxial compressive strength was found equal to σci =71 MPa andmi = 11.8 (thus 12). The failure envelope is given in Fig. 12.

It was assumed that the relationship between themi parameter andthe siltstone content would be similar to the one between uniaxial com-pressive strength (σci) and siltstone percentage (sl), as presented inFig. 4. Consequently, substituting σci with mi in Eq. (3) and using mi

maximum value as 17 (proposed for sandstone by Hoek and Brown,1980), the resulting equation for the prediction of mi is as follows:

mi ¼ 17 e−0:022sl: ½8�

Considering a siltstone percentage equal to 17% (equal to that of thespecimens onwhich triaxial testswere performed), the value ofmi fromEq. (8) is found equal to 11.6, which is in good agreement with the pa-rameter derived byfitting theHoek–Brown failure criterion to the triax-ial laboratory test results (mi = 11.8).

It must be noted that due to the difficulty of the triaxial tests on com-posite specimens, only a set of tests was performed. The above observa-tions should be confirmed by more triaxial tests on different siltstonepercentages (e.g. sl = 20% or 25%).

Fig. 10. Geometrical features and elasticitymodulus of a rock specimenwith a discontinuity.

4. Discussion

4.1. Comparison with similar laboratory results

Zainab et al. (2007) carried out a laboratory study to determine thestrength of tropically weathered sandstone and shale in Malaysia. Thecomposite samples were prepared into three different thickness ratiosof shale to the total height of specimens, H, each having 0.1H, 0.2Hand 0.3H of shale. These composite samples were prepared to simulatethe possible geometry profile of the constituents in the interbeddedsedimentary formation. The laboratory method that was used was thesame to the present study, with the only difference that plaster wasused between the interfaces of the rock disks as a bonding material.

Fig. 13 presents the comparison between the results of Zainab et al.(2007) and the present study. Because of the difference in the compres-sive strength of the original materials and in order that the laboratoryresults are compatible, the comparison was made using the ratio ofthe uniaxial compressive strength of each sample to the strength ofthe corresponding intact sandstone, σci/σci

sandstone.Zainab et al. (2007) has concluded that only a low percentage (i.e.

10%) of shale reduces the strength of the composite rock specimen byalmost 70%, which is close to the strength of the weaker material. Onthe contrary, the decrease of the strength in the present study is withlower gradient and the composite specimen strength falls to the silt-stone strength only when the siltsone percentage in the compositespecimen is 37% or greater.

An explanation for the above differences can be given by the ratio ofthe compressive strength of the original materials.

In Zainab et al. (2007), σ cisandstone

σ cishale ¼ 33:6MPa

7:6MPa ¼ 4:4:

While in the present study, σ cisandstone

σ cisiltstone ¼ 95:3MPa

43:0MPa ¼ 2:2:

Fig. 11. Comparison of the measured Ei values with those predicted by the elasticitytheory, Etotal.

Table 5Triaxial laboratory test results for composite specimens with sl = 16–17%.

Siltstone percentagesl (%)

Confining pressureσ3 (MPa)

Compressive strengthσ1 (MPa)

16.4 0.0 77.416.9 5.0 95.716.0 10.0 120.716.8 20.0 161.416.4 30.0 212.7

Fig. 13. Comparison of laboratory results of the present article to Zainab et al. (2007).

88 G.P. Tziallas et al. / Engineering Geology 166 (2013) 81–89

As the ratio of the compressive strength of the two original rocks in-creases, the decrease in the compressive strength ration is greater withincreasing percentage of the low strength material. This can be under-stood by comparing two extreme cases of composite rock materials:a) a sandstone with a soft marly intercalation (σci hard/σci weak N 8)and b) a specimen comprising of 3 disks of sandstone (the weak mate-rial considered the same as the harder one thus σci hard/σci weak = 1). Inthe first case, even the thinnest marl disk will fail under the initial load-ing steps and the overall compressive strength of the sample will reachthe lower limit at once. In the second case the “composite” specimenwill respond as an intact sandstone sample, so even by altering thethickness of the disks, the strength will remain constant.

This conclusion is represented graphically in Fig. 14. It must bepointed out that the curves presented in this figure, except those byZainab et al. (2007) and the present study, constitute hypothetical sche-matic presentation of the above theory and have not been verified bylaboratory testing.

4.2. Comparison with empirical methods

Marinos and Hoek (2001) proposed a GSI chart for heterogeneousrock masses, which constitutes an empirical method for rock mass de-scription and mechanical properties estimation. Marinos (2010) sug-gested an updated GSI classification chart for heterogeneous rockmasses.

Regarding the selection of the “intact” rock properties σci and mi offlysch composite rocks, Marinos (2010) suggested proportions of values(“weighted average”) for each rockmass type of the proposedmodified

Fig. 12. Analysis of triaxial laboratory test r

charts, reducing values of σci and mi of competent rocks (sandstones)up to 40%, when the layers of these rocks are separated from eachother by weaker layers of siltstones or shales.

Therefore, having defined the parameters σci, mi and GSI, the me-chanical properties of the heterogeneous rock mass are estimated byusing the Hoek–Brown failure criterion (Hoek and Brown, 1997).

Here, a comparison ismade between the propositions of the newGSIchart and the laboratory results of the present paper. The comparisonwas made for flysch Type IV, which comprises of siltstone and sand-stone layers of equal thickness (sl–st = 1:1). This rock type corre-sponds to a composite specimen with a siltstone percentage equal to30% (specimen B), as shown in Fig. 1.

Using the compressive strength values of the intact materials asdetermined through the laboratory tests i.e. σci–sl = 43.0 MPa andσci–st = 95.3 MPa for the siltstone and the sandstone material re-spectively, the suggestion by Marinos (2010) results in a compres-sive strength of the “intact” flysch rock equal to 64.4 MPa (for thinbeds) and 69.2 (for thick beds). According to the laboratory tests ofthe present paper (presented in Figure 4 and based on Eq. (3)) the

esults (Roclab, Rocscience Inc., 1998).

Fig. 14. Anticipated behavior of heterogeneous materials with σci hard/σci weak ratios.

89G.P. Tziallas et al. / Engineering Geology 166 (2013) 81–89

compressive strength of the corresponding composite specimen(thus flysch rock type IV), is equal to 50.0 MPa.

Thus, the compressive strength of the flysch “intact” rock, estimatedby using themethod proposed in GSI classification, appears to be higherthan the one determined in the present study.

5. Conclusions

The present paper deals with the determination of mechanical prop-erties of heterogeneous rock, such as flysch based on laboratory testingon artificial composite samples of flysch type rocks, consisting of silt-stone and sandstone in different proportions. Based on these laboratorytests, the variation of different properties, such as the ultrasonicwave ve-locity, uniaxial compressive strength, Young'smodulus and triaxial com-pressive strength in relation to the percentage of the weak rock material(siltstone) present in the composite rock samples was investigated.

This investigation can further assist the effect of siltstone proportionon the compressive strength and modulus of deformation of flyschrockmasses, comprising siltstone and sandstone alternations in differentproportions.

The uniaxial compressive strength of the flysch composite rock wasfound to decrease exponentially with increasing siltstone percentageup to the 37%, thereafter the composite specimen displays the strengthof the siltstone material. The Young's modulus also decreased expo-nentially with the increase of siltstone percentage in the specimens(expressed by the modified siltstone percentage).

The Young modulus predicted by the proposed equation was com-pared to that determined by the elastic deformation theory and wasfound to correlate very well.

The Hoek–Brown criterion parameter mi was determined by tri-axial testing on composite samples with siltstone percentage equalto 17%. Prediction of mi values for composite flysch rocks with differ-ent siltstone percentages can be achieved using a similar equation tothe one proposed for uniaxial compressive strength (given here asmi = 17 e−0.022sl). Based on the equation proposed, the value pre-dicted for mi coincided with the one determined by fitting Hoek–Brown failure criterion to triaxial data.

Comparing the results with previous studies (Zainab et al., 2007), itwas possible to suggest how the relation between uniaxial compressivestrength, σci, and siltstone percentage, sl, may differ according to theratio of the compressive strength of the twomain rockswhich comprise

the heterogeneous rockmass. It can be concluded that as the ratio of thecompressive strength of the two participating rocks (σci

hard/σcisoft) in-

creases, the decrease in the compressive strength ratio (σcicomposite/σci

hard)is greater with increasing percentage of the low strength material.

Finally, regarding the selection of the “intact” rock properties, name-ly σci, of flysch composite rocks, a comparison was made for a certainflysch rock type, consisting of equal proportion of siltstone and sand-stone, between the compressive strength proposed by the GSI classifica-tion methodology and the present laboratory test results on compositesamples. It appears that the latter is lower.

Concerning the selection of the parameter mi of flysch rocks for usein derivation of rockmass parameters, it was concluded that the pro-posed equation between mi and siltstone percentage can yield very re-alistic values.

The present study presented the complexity and the challenge of es-timating the strength and deformation parameters of heterogeneousrocks, such as flysch. By testing composite rocks, with different mate-rials and various proportions of each one, it can further assist in the es-timation of the properties of heterogeneous rocks and thus rockmasses.

References

Budetta, P., Nappi, M., 2011. Heterogeneous rock mass classification by means of the geo-logical strength index: the San Mauro formation (Cilento, Italy). Bull. Eng. Geol. Envi-ron. 70, 585–593.

Duffault, P., 1981. Structural weaknesses in rocks and rockmasses. Tentative classificationand behaviour. Weak rock: soft, fractured and weathered rock. Proc. Symposium,Tokyo, September 1981, volume 1, pp. 93–97.

Goodman, R.E., 1993. Engineering Geology—Rock in Engineering Construction. In: Smith,M.R., Collis, L. (Eds.), John Wiley & Sons, Inc. (412 pp.).

Greco, O.D., 1994. Behaviour of composite rock specimens under uniaxial compressivetests. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 32 (2), A76.

Greco Del, O., Ferrero, A., Peila, D., 1992. Behaviour of laboratory specimens composed ofdifferent rocks. Proc. of ISRM Int. Congress on Rock Mechanics, Aachen, pp. 241–245.

Hoek, E., Brown, E.T., 1980. Underground Excavations in Rock. Inst Min Metall, London.Hoek, E., Brown, E.T., 1997. Practical estimates of rock mass strength. Int. J. Rock Mech.

Min. Sci. Geomech. Abstr. 34 (8), 1165–1186.Hoek, E., Diederichs, M.S., 2006. Empirical estimation of rock mass modulus. Int. J. Rock

Mech. Min. Sci. 43, 203–215.Hoek, E., Marinos, P., Marinos, V., 2005. Characterization and engineering properties of

tectonically undisturbed but lithologically varied sedimentary rock masses underpublication. Int. J. Rock Mech. Min. Sci. 42 (2), 277–285.

ISRM, 2007. The complete ISRM suggested methods for rock characterization, testingand monitoring: 1974–2006. In: Ulusay, R., Hudson, J.A. (Eds.), Suggested MethodsPrepared by the Commission on Testing Methods, International Society for Rock Me-chanics, Compilation Arranged by the ISRM Turkish National Group, Kozan Ofset,Ankara, Turkey (628 pp.).

Liang, W., Yang, C., Zhao, Y., Dusseault, M.B., Liu, J., 2007. Experimental investigation ofmechanical properties of bedded salt rock. Int. J. Rock Mech. Min. Sci. Geomech. 44,400–411.

Lydzba, D., Pietruszczak, S., Shao, J.F., 2003. On anisotropy of stratified rocks: homogeni-zation and fabric tensor approach. Comput. Geotech. 30 (4), 289–302.

Marinos, V., 2010. New proposed GSI classification charts for weak or complex rockmasses. Bulletin of the Geological Society of Greece, 2010, Proc. of the 12th Int. Con-gress, Patras, May, 2010.

Marinos, P., Hoek, E., 2001. Estimating the geotechnical properties of heterogeneous rockmasses such as flysch. Bull. Eng. Geol. Environ. vol. 6, 85–92.

Rocscience Inc., 1998. Roclab Users manual.Saroglou, H., Steiakakis, C., 2010. Prediction of strength of anisotropic and layered flysch-

type rocks. Proc. of 6th Hellenic Conference on, Geotechnical Engineering, vol. 2,pp. 243–249.

Tsiambaos, G., 2010. Engineering geological behaviour of heterogeneous and chaotic rockmasses. Bulletin of the Geological Society of Greece, 2010. Proc. of the 12th Int. Con-gress, Patras, vol. XLIII, pp. 183–195.

Tziallas, G., 2010. Estimation of strength and deformability characteristics of compositerocks. MSc Thesis Geotechnical Engineering Department, NTUA (in Greek).

Vlasov, A.N., Merzlyakov, V.P., 2004. Analysis of test results of composite specimensmodeling rock. J. Soil Mech. Found. Eng. 41 (6), 191–199.

Zainab, M., Kamaruzaman, M., Cho, Gye C., 2007. Uniaxial compressive strength of com-posite rock material with respect to shale thickness ratio and moisture content. Elec-tron. J. Geotech. Eng. 12.