Engineering Geology 122 (2011) 160–168

Contents lists available at SciVerse ScienceDirect

Engineering Geology

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

Influence of carbonate content on the stress–strength behaviour of neogene marlsfrom the betic cordillera (Spain) in cu triaxial tests using a quasilinear elastic(hyperbolic) model

F. Lamas a, C. Oteo b, J. Chacón a,⁎a Department of Civil Engineering, University of Granada, ETSICCP, Avda. Fuentenueva, s/n, 18071, Granada, Spainb Department of Construction Technology, University of A Coruña, ETSICCP, Spain

⁎ Corresponding author. Tel.: +34 958246136; fax: +E-mail address: jchacon@ugr.es (J. Chacón).

0013-7952/$ – see front matter © 2011 Elsevier B.V. Adoi:10.1016/j.enggeo.2011.05.013

a b s t r a c t

a r t i c l e i n f o

Article history:Received 22 April 2010Received in revised form 30 April 2011Accepted 3 May 2011Available online 6 August 2011

Keywords:Marly soilsHyperbolic modelTriaxial CUCarbonate content

The stress–strain behaviour of 135 samples of calcareous silt and clay soil, known as marl, from three quarriesis presented and discussed in connection with its adjustment to a hyperbolic non-elastic constitutive model.Themarl samples were classified and identified, following conventional soil-mechanics procedures and testedunder triaxial CU (consolidated undrained) stresses. Also the carbonate, sulphate, and soluble salts contentswere analysed by chemical and X-ray methods. Considering the obtained strength parameters and chemicalcompositions, a high correlation between the carbonate content and the parameters of the hyperbolic model,was observed, indicating a parallel increase in the brittle behaviour, and carbonate content, with a slightincrease in bearing capacity. Nevertheless, when all the samples were considered, the parameters of thehyperbolic model seemed not to be clearly related to the carbonate content, this being interpreted asexpressing a hyperbolic model influenced not only by the carbonate content but also by several othervariables such as the percentage of fine particles or the dilatationing behaviour of the soil particles expressedby the dry density. Given the extensive use of marly soils in the construction of earth dam cores and slopebanks, the main interest of the results is to present new insights into its mechanical behaviour not usuallyconsidered in common engineering practice.

34 958246138.

ll rights reserved.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

The study of the mechanical properties of the calcareous silt andclay (marl) cropping out in the Neogene of the Betic Cordillera(southern Spain) and Tetouen (northern Morocco), on both sides ofthe Mediterranean, leads to different contributions concerning: theeffect of expansive clays and the activity of the fine soil fraction; theinfluence of the carbonate content on the plasticity behaviour ofthe marls and their strength properties; the problems arising withthe engineering classification of marls; and, finally, the proposal of anappropriate method for the chemical analysis of the carbonatecontent. These studies were based on the sampling of marly soils inthe Neogene basins of Granada (Spain), Tetouen (Morocco) and alsoin three quarrying sites in the Guadix-Baza Neogene basins (Spain)where this marl was extracted for the construction of three earthdams. (El Amrani and Chacon, 1996; El Amrani et al., 1998, 2000;Lamas et al., 2002, 2005).

The geology and engineering geology of the marls in both studyareas was described (El Amrani and Chacon, 1996; El Amrani et al.,1998). An inverse correlation was shown between the carbonate

content and dispersion index and swelling behaviour, while a directcorrelation between carbonate content and shear strength wasestablished from the study of 171 samples of cohesive calcareoussoils. Compaction increases the shear strength and the clay fractionactivity while the sulphate content augments susceptibility toweathering (El Amrani et al., 1998).

These samples were taken in the field following the instructionsdetailed in ASTM technical norm E-4 (2004). For the preparation ofundisturbed samples in laboratory, the necessary number of sampleswere previously cut up, not only for triaxial but also for confinedcompression tests, splitting apart enough amounts for other tests. Thesampling zones were selected by direct application to quarry fronts ofsomedam-project technical requirements concerning theconstructionofimpervious embankment cores (PG-3, 2004), disregarding those quarryfronts not fulfilling these requirements. Therefore, the selected quarrieshave similar lithologies and the soil extracted for the construction of coreof earth dams shows compositions in between the acceptablemaximumand minimum values of different physical parameters.

Also, an inverse correlation between the clay content and theresidual friction angle was found. The carbonate fraction of themarlsoil had a significant influence on the residual friction angle, andthere was a threshold of 25% of carbonate contents, above which thevalue of ϕ′r increased until the clay fraction and the percentage of

161F. Lamas et al. / Engineering Geology 122 (2011) 160–168

smectite within that fraction tended to increase the residual cohesionvalue. It was proved that the presence of smectitic clay as dominantconstituent produces a low value of residual shear strength,associated with high fragility in unconsolidated undrained tests, asreported by Lamas et al. (2002), although this parameter was notcorrelated because of the similarity in minor smectitic clay contentamong the available samples. Also it has been proved that the degreeof compaction of the soil and, therefore, the initial porosity has adirect influence on the effective residual friction angle ϕ′r (El Amraniet al., 2000; Lamas et al., 2002).

The existing engineering classifications fail to provide an adequateresponse to clay minerals and, especially, carbonate content of themarls and, thus, none of these systems satisfactorily explains thedifferent behaviour of the marls studied (Lamas et al., 2002).

Despite the large number of experimental methods available todetermine the carbonate content in soils, a comparative analysis ofthe three most commonly used methods (flame photometry, theBernard calcimeter, and EDTA complexometry) indicates that theaccuracy of the method is a function of the carbonate concentration inthe sample to be analysed and the Bernard calcimeter has beendemonstrated to be the most suitable method for analysingcarbonates in soils (Lamas et al., 2005).

This paper provides a more detailed study of the relationshipsbetween the strength properties of marls tested in CU triaxialcompression, and the carbonate content. For that purpose, samplesfrom the above-mentioned three quarrying sites for earth dams in the

Fig. 1. Setting of the

Guadix-Baza basins were selected (Granada province, Spain; Figure 1)with increasing carbonate content.

In order to analyse changes in parameter values in these soils ofknown type and density, the results concerning the stress–strainbehaviour are fitted to a quasilinear elastic or hyperbolic model whichis considered appropriate to address problems involving stress andstrain prior to failure (Kondner and Zelasko, 1963;Wong and Duncan,1974; Duncan, 1980; Duncan and Chang, 1980).

2. The nonlinear elastic hyperbolic constitutive model

The nonlinear elastic constitutive models simulate stress–strain soilbehaviour by considering the elastic parameters as depending on thestress state and the accumulated deformation. The hyperbolic model(Kondner and Zelasko, 1963), modified by Duncan and Chang (1980) tofit it to the c and ϕ soil-strength parameters of the Mohr–Coulombequation and to the finite-element analysis (Domaschuk andValliappan,1975), depends in the model of Duncan and Chang (1980) only on thestress state andmodel's parameters, suchas thebulkmoduleK, the shearmodule Et and also the Poisson ratio ν. The coefficients necessary todefine the model depend on the failure criteria. The main advantage ofthe model of Duncan and Chang (1980) is usually a best fit to thelaboratory data because it permits the variation of the Poisson ratioduring the analysis, attaining critical values in the stress deviation andvolume change throughout the different stages of the analysis, as it isusually observed in real materials. Nevertheless, there are a number of

sampling area.

162 F. Lamas et al. / Engineering Geology 122 (2011) 160–168

limitations concerning the nonlinear elasticity related to its inability toconsider the dependence between the stress history and pathways, fromwhich the strain increments dependon thestress increase rather thanonthestress state inwhichagiven increment is applied. Thedilatancyeffectof the soil during the increase in compressive load violates principles ofthermodynamics because if the shearmodule depends only on the stressor strain state it is not possible a simulation of changes during thestresses release.

Thedevelopmentof thehyperbolicmodels for sandandcohesive soilsis found inKondner andZelasko (1963), and furtherwork byDuncan andChang (1980), Wong and Duncan (1974), and Duncan (1980). Dapena(1993a,b,c) applied themodel to the behavioural analysis of clay cores inearth dams. Kondner and Zelasko (1963) demonstrated the stress–strainbehaviour (σ, ε) of sand and cohesive soils adjustment to a nonlinearelastic behaviour, represented by a hyperbole, defined by two constantparameters, a and b, in the following equation:

σ1−σ3 = ε1.a + bε1

ð1Þ

where σ1 and σ3 are the highest and lowest stresses, ε1 is the verticalor main higher strain, until a and b are experimentally determinedsample constants. Thus, for small strain values the constant, a, is thereverse of the initial shear module Ei, until b is the reverse of theasymptotic value of the stress deviation in the critical stage, i.e. atlarge strain values.

For a given constant value of σ3, the shear module Et (3) iscalculated from Eq. (1) by differentiating the stress deviation againstthe vertical stress:

Et =∂ σ1−σ3ð Þ

�∂ε1 = a

a + bε1� ��

ð2Þ

With value ε1 in Eq. (1) Kondner and Zelasko (1963) arrived at:

Et = 1= að Þ 1−b σ1−σ3ð Þ½ �2: ð3Þ

Duncan and Chang (1980) proposed the following equation toestimate the initial shear module Ei, based on experimental resultsfrom Jambu (1963):

Ei = Kpaσ3

Pa

� �n

ð4Þ

where K and n are parameters depending on the variation of Ei withregard to σ3, both being experimentally determined for each samplein triaxial tests, and Pa the atmospheric pressure. For that determi-nation the Ei with regard to σ3 values are plotted in doublelogarithmic scale, and the experimental results are adjusted to astraight line with the slope indicating the value of n and the ordinateat the origin the value of K.

Also, Duncan and Chang (1980) proposed the failure ratio Rfbetween the failure stress deviation (σ1−σ3)f and the asymptoticvalue of the stress deviation at large strains or critical state (σ1−σ3)u.

σ1−σ3ð Þf = Rf σ1−σ3ð Þu ð5Þ

If the failure is reached at a constant σ3 confining pressure,following the Mohr–Coulomb failure law, the relationship betweenthe stress deviator at failure, the strength parameters, c cohesion andϕ friction angle, and the σ3 confining pressure, is expressed by:

σ1−σ3ð Þf = 2c cosϕ + 2σ3senϕð Þ= 1−senϕð Þ: ð6Þ

From the b parameter value, and combining Eqs. (4), (5), and (6),Duncan and Chang (1980) got:

Et = Kpaσ3

Pa

� �n

1−Rf 1−senφð Þ σ1−σ3ð Þ2c cosφ + 2σ3senφ

� �2: ð7Þ

Following Wong and Duncan (1974), the parameters n and k aredetermined from the Eqs. (8) and (9):

n =Δ log

E1pa

� �

Δ logσ3

pa

� � ð8Þ

k = lim σ3pa

=1� � E1

pa

� �: ð9Þ

From CU triaxial tests, the transformation of the different stress–strain plots was calculated. According to Wong and Duncan (1974)the transformed curves follow a straight line with a slope dependingon the cell pressure and compaction conditions. However, JiménezSalas (1993) suggested that under large strain the transform equationshows a nonlinear pattern with an asymptotic trend.

Duncan and Chang (1980) calculated the variability of the earlyelastic module from charge–discharge–recharge tests in sand, usingthe equation:

Eur = KurPa σ3 =Pað Þn ð10Þ

where Eur is the elastic charge–discharge module and Kur is thecorresponding module number.

3. Sampling, laboratory tests and results

3.1. Sampling and samples

A total of 118 samples of cohesive marl soils extracted from thequarries supplyingmaterials for the clay cores of the S. Clemente, Portilloand Francisco Abellán earth dams were tested (Figure 1). The samplingprocedures follow the E-4 (ASTM, 2004) technical prescriptions.

3.2. Sample description: phase relationships, particle size, compaction,and composition

The determination of the phase relationships, particle size, consisten-cy, compaction, and specific weight were determined following technicalnormsand recommendations byASTMD422, D854,D4318, andUNE107(UNE, 2001).

Table 1 shows data concerning to natural moisture, maximum sizeof particle, percentage of particle sizes below 0.08 mm, optimummoisture in the Proctor normal test, Atterberg's consistency param-eters (liquid limit LL and plasticity index PI) and activity of the finesoil fraction (Skempton and Bishop, 1954). It is observed that theplasticity index, the activity, and the percentage of fine particles werelower in samples from “Francisco Abellán” earth dam and higher insamples from the “S. Clemente” earth dam, with intermediate valuesin samples from the “EL Portillo” earth dam. A similar distribution wasfound in the carbonate content (Figure 3), which increased with soilplasticity and activity of the fine fraction. The particle-size distributionshowed an increasing dispersion of the percentage of soil fractionb0.008 mm with regard to its standard deviation. Thus, the hetero-geneity increased with a parallel reduction in the percentage of finefraction, from S. Clemente earth dam (90.76%±6.4%, dispersion 7.1%;n° of samples 24), to El Portillo earth dam (82.32%±11.1%, dispersion13.5%; n° of samples 24) and F. Abellán earth dam (61.1%±7.7%,dispersion 12.6%; n° samples 24).

Table 1Standard test results.

S. Clementesamples

Hn Tm Fl Dm Ho LL Ip AI

24

Maximum 22.50 20.00 98.89 1.816 22.21 57.60 34.51 0.730Mean 15.82 10.10 90.76 1.708 18.30 43.82 25.03 0.590Minimum 10.87 2.00 71.02 1.621 15.20 31.00 15.20 0.300Desv. Typical 3.07 7.10 6.400 0.048 1.58 5.01 3.85 0.100

PortilloSamples 24Maximum 18.10 40.00 93.30 2.10 17.20 40.30 24.00 0.842Mean 14.48 17.01 82.32 1.85 14.40 31.94 14.33 0.574Minimum 5.54 2.50 20.76 1.76 10.70 27.00 9.70 0.447Desv. Typical 2.87 10.48 11.11 0.05 1.17 3.31 3.23 0.075

Fco AbellánSamples 24Maximum 20.40 40.00 79.56 1.92 18.70 36.40 17.50 1.210Mean 14.27 17.21 61.10 1.81 14.98 31.40 12.96 0.543Minimum 7.72 5.00 41.38 1.70 11.82 27.80 9.00 0.339Desv. typical 2.671 7.134 7.70 0.045 1.438 2.090 1.67 0.180

Legend: Hn=natural moisture (%); Tm=maximum size (mm); Fl=%b0.08 mm;Dm=max dry density (Tn/m3); Ho=optimum moisture (%); LL=limit liquid (%);Ip=plasticity index (%); AI=activity.

Fig. 3. Carbonate content; histogram.

Table 2Odometer and specific weight of solid particles.

SanClementesamples

σ′P (kPa) cv (cm2/s) γS (Tn/m3)

24

163F. Lamas et al. / Engineering Geology 122 (2011) 160–168

Fig. 2 shows a plot of LL and PI values of the samples in aCasagrande abacus. Except for a sample of the CH type, all samplescorrespond to inorganic low-plasticity clays of the CL type in theUnified Soil-Classification System (ASTM, 2004).

The expansive behaviour of the sample was determined by theprocedure recommended by the “Technical Committee on ExpansiveSoils”, TC6 de ISSMFE (1987) and contributions by El Amrani (1996).Table 2 shows the results from 24 free swelling and oedometer testson disturbed samples following procedures described in ASTMD2435-04 and UNE 74.527 (UNE, 2001) and the pre-consolidationpressure, consolidation coefficient, and the specific weight of theparticle. A parallel reduction was observed in pre-consolidationpressure, PI, and carbonate content but not for the other parameters.

Table 3 shows results related to the soil-phase distribution andmineralogical composition. The clayminerals predominant in the studiedsamples, determined following procedures described by Warshaw andRoy (1961), were weatheredmuscovite in transition to illite, with minoramounts of montmorillonite and caolinite, and traces of chlorite.

The carbonate content was determined following the methoddescribed by Lamas (2003), using a Bernard calcimeter in order toavoid errors in the ends of sample series. It was found that as thesamples increased their carbonate content, there was a decrease inthe quartz content and the percentage of sand fraction as well as anincrease in the percentage of the fine fraction (silt and clay particles).Thus, in 37 samples of the F. Abellán earth dam the carbonate content

Fig. 2. Plasticity chart showing tests result.

was 20.20%±8.48%, with a dispersion of 42%, in 46 samples of ElPortillo earth dam there is a carbonate content of 32.67%±6.30%, witha dispersion of a 19.3% and in 35 samples of S. Clemente earth dam thecarbonate content was 53.5%±10.55% with a dispersion of a 19.7%.The percentages of the quartz content and sandy-soil fractiondiminished by an average of 56.77% to 21.76% and 39% to 10.36%,respectively, while the fine-soil fraction increased an average of61.01% to 89.64%. Also the homogeneity of the sample compositionincreased correlatively to the decrease in sandy-soil fraction andcarbonate and quartz contents. The association between carbonatecontent and fine-soil fraction is clear from these data, where thecarbonate particles may be represented in silt and coarse-clay sizes.

The compaction tests were conductedwith reference to four stagesof dry density, following the technical suggestion for embankments inthese materials (PG-3, 2004). Table 4 offers the density data forsamples collected in the three sampling zones.

3.3. Triaxial CU tests

Given the low permeability of the tested soil samples, and theirbehaviour as cohesiveundrainedsoils, evenunder themostunfavourablegeotechnical conditions, an undrained triaxial CU test with previousconsolidation and water pressure measuring was considered the bestapproach to the in situ stress–strainbehaviour in order to determine theireffective parameters.

Maximum 205 8.70*10−4 2.75Mean 124 3.10*10−4 2.68Minimum 60 1.20*10−5 2.52Desv. typical 0.351 1.80*10−4 0.052

PortilloSamples 24Maximum 271 3.10*10−3 2.73Mean 139 4.45*10−4 2.68Minimum 75 1.04*10−4 2.63Desv. typical 0.507 4.83*10−4 0.023

Fco AbellánSamples 24Maximum 250 1.04*10−3 2.78Mean 156 4.30*10−4 2.71Minimum 47 1.79*10−4 2.64Desv. typical 0.473 2.16*10−4 0.036

Legend: σ′p=preconsolidation pressure; cv=coefficient of vertical consolidation,γS=specific weight of solid particles.

Table 3Texture and mineralogical composition.

Quarry % Sand % Silt % Clay %Carb. %Quartz %Feld.

Samples Fco Abellán 37 37 37 37 37 37Maximum 58.62 41.38 42.40 47.47 83.00 5.00Mean 39.00 35.83 25.18 20.20 56.77 2.35Minimum 20.44 33.39 9.20 9.32 45.00 1.00Desv. typical 7.70 1.69 8.22 8.48 11.25 1.89Samples Portillo 46 46 46 46 46 46Maximum 79.24 72.00 39.00 40.50 60.00 6.00Mean 17.68 58.64 23.68 32.67 48.93 2.35Minimum 6.70 20.76 12.00 10.40 30.00 1.00Desv. typical 11.11 6.43 9.11 6.30 8.01 2.00Samples S. Clemente 35 35 35 35 35 35Maximum 34.00 60.00 58.00 77.13 45.00 15.00Mean 10.36 48.80 40.84 53.50 21.76 1.82Minimum 1.11 35.50 29.00 31.36 5.00 1.00Desv. typical 7.73 5.56 6.52 10.55 9.86 3.40

Table 5Cohesion and angle of internal friction from CU triaxial tests.

San Clemente Portillo Fco Abellán

Carbonate content (%) 70 63 37 27 22 15Initial moisture (%) 17.5 17.7 12.4 15.7 17.05 15.48Final moisture (%) 18.6 18.9 13.5 16.6 17.6 16.2γd (Tn/m3) 1.725 1.706 1.780 1.821 1.770 1.799Cohesion c′ (kPa) 21 12 20 17 27 10ϕ′ (degree) 23.3 22.5 31.0 27.0 35.0 26.0

164 F. Lamas et al. / Engineering Geology 122 (2011) 160–168

Table 5 shows values of cohesion c and friction angle ϕ determinedin 135 triaxial tests arranged in groups of carbonate content in orderto analyse the relationships. As c values are fully independent, ϕincreases with decreasing carbonate content.

3.4. Parameters of the hyperbolic model

Following the suggestions of Boscardin et al. (1990) and Al-Shayeaet al. (2003), CU triaxial tests were performed on 72 samples selectedas representative of the carbonate-content variability (Table 6). UsingEqs. (5) to (9), parameters K, n and Rf of the hyperbolic model(Kondner and Zelasko, 1963) modified by Duncan and Chang (1980)were determined. The intervals in carbonate content of samplesselected for the calculations are indicated. The results are expressed inratios εa: εa/(σ1−σ3) and are shown in Figs. 4 to 6 ordered incarbonate content, from more that 60% (Figure 4b), between 50 and60% (Figure 4a), 30 to 50% (Figure 5b), 25–30% (Figure 6b), and 0 to15% (Figure 6a).

Some minor differences between the mechanical parameters ofsamples with carbonate content below 15% and above a 50%, areattributed to the different geological sedimentary conditions. This isthe reason for the way in which the samples were ordered for thepurposes of this research.

From the above data the relationships between the failure moduleand the carbonate content were estimated (Figure 7). A higherincrease in the failure module was found with the soil dry density insamples having more than 50% carbonate content, while a lowerincrease was noted in samples with less than 50% carbonate content.This may be interpreted as an expression of the contribution of thecarbonate particles to the dilatancy effect on soil strength. Therelationships between the initial elastic module E1 and the confinedpressure σ3 (Figure 8) or the soil dry density (Figure 9) with thesample carbonate content, also indicate a trend of increasing elastic

Table 4Values of density for different degrees of compaction.

P.N. % γd (Tn/m3) γs (Tn/m3) γω (Tn/m3)

Fco Abellán 95 1.65 2.70 1.89100 1.70 2.70 1.94102 1.81 2.70 2.03105 1.919 2.70 2.11

Portillo 95 1.74 2.70 2.00100 1.82 2.70 2.05102 1.866 2.70 2.08105 2.102 2.70 2.12

S. Clemente 95 1.59 2.67 1.82100 1.62 2.67 2.01102 1.7 2.67 2.04105 1.82 2.67 2.07

behaviour with the increase in carbonate, and a direct correlationwith dry density (Figure 7).

4. Discussion of the results

4.1. Samples description: phase relationships, particle size, compactionand composition

The samples are low-plasticity clays of the group CL in the UnifiedSoil-Classification System for Engineering Purposes (ASTM, 2004,Figure 2).

The distribution of soil-particle-size fractions is significant for theanalysis of the mechanical behaviour of soils. The homogeneity of thesamples increases inversely with the sandy-soil, carbonate and quartzcontents. The results clearly show the association between carbonatecontent and fine soil fraction (Tables 1, 2 and 3), where the carbonateparticles may be represented in silt and coarse clay sizes. This is aconsequence of the sedimentology of the marly soils. The Neogenedeposits of the Guadix-Baza basin (Betic Cordillera, Spain) resultedfrom a deltaic sedimentation in a shallow marine environmentdescribed by Fernández and Guerra-Merchán (1996), where gravita-tional flows and avalanches were more fluid in distal parts, giving riseto sediments with a predominant fine fraction and coarser in proximalparts with sandy marls and related sediments.

Also Fukue et al. (1996) established that the greater the waterdepth and the larger the distance from the shoreline, the higher thecalcium carbonate content becomes and that, at a constant precipi-tation rate of calcium carbonate, the accumulation rate of sediments ishigher near shorelines. The reason for this is that more sedimentaccumulates near and around river mouths and shorelines. A detaileddescription of palaeo-environmental changes around the Miocene–Pliocene boundary along a West–East transect across the Mediterra-nean and the changes in the carbonate content in relation to distal orproximal facies is due to Pierre et al. (2006).

Given the complexity of water-flow mechanism through theimpervious core and, taking in account the proximity of the dams tothe river sources, except for San Clemente dam because there the riveris highly concentrated in carbonates after crossing previous karsticareas, the waters have low mineralization and, therefore, areaggressive to soil by dissolution. Therefore, in the river watersupplying the dam reservoirs the chemistry was tested in order toestablish whether the mineralization was weak, and thus the waterwas aggressive or high and therefore the carbonate depositpredominated over the dissolution mechanisms.

The results obtained by “At marble test” (Rodier, 1978), whichquantifies the dissolution of calcium carbonate in water, indicatedthat incrustation processes were widespread.

In view of the above, it becomes clear that in the study zone theproximaldeposits are representedby thecoarser samples inSanClementequarry and the distal deposits by the finer samples in F. Abellán. Also thehigher variability or dispersion in the proximal deposits in San Clement isconsistent with the changeable environmental conditions in the coastaldeltaic alluvial processes and also the overall homogeneity of the particle-size distributions is related to the sedimentology of the sandy- to silty-marl units selected for building up the cores of the three earth dams. Themineralogy is also consistent with the type and origin of the sediments.

Table 6Calculated means of E1 and Rf, in CU triaxial tests for different carbonate contents.

Carbonate content (%) interval Sample Nº W % γd (Tn/m3) σ3+σu (kPa) a E1=1/a (σ1−σ3)p (kPa) (σ1−σ3)f (kPa) b Rf

Fco Abellán 0bCO3Cab15 I 15.4 1.8 50 0.0133 75.19 212.9 180 0.409 0.845II 15.0 1.82 150 0.0072 142.86 292 264 0.310 0.904III 14.9 1.84 300 0.0061 166.67 407.6 353.4 0.218 0.821

Fco Abellán 15bCO3Cab25 I 14.8 1.852 50II 14.6 1.860 150 0.006 166.67 310.3 289 0.297 0.931III 15.1 1.851 300 0.003 333.33 363.8 336.3 0.267 0.924

Portillo 25bCO3Cab30 I 15.7 1.82 50 0.034 29.41 121 104.1 0.672 0.860II 16.0 1.81 150 0.0076 131.58 186 163.2 0.523 0.877III 15.5 1.816 300 0.003 333.33 352 334 0.268 0.948

Portillo 30bCO3Cab40 I 13.8 1.759 50 0.0148 67.57 184 143 0.479 0.777II 14.8 1.781 150 0.009 111.11 298 249 0.286 0.835III 14.8 1.804 300 0.008 125.00 410 327 0.196 0.798

S. Clemente 50bCO3Cab70 I 18.3 1.74 50 0.006 166.67 184.2 183.1 0.308 0.995II 17.9 1.74 150 0.004 250.00 267.6 266.2 0.352 0.994III 18.4 1.75 300 0.002 500.00 355.9 354.3 0.270 0.995

S. Clemente CO3CaN70 I 19.6 1.718 50 0.0104 96.15 160 159 0.570 0.993II 19.8 1.697 150 0.0025 400.00 188 170 0.529 0.904III 19.8 1.72 300 0.0025 400.00 311 300 0.311 0.965

165F. Lamas et al. / Engineering Geology 122 (2011) 160–168

The association of the carbonate content and the fine fraction isrelated to environmental factors determining its deposition, and it isexpressed by positive correlation to the fine-soil size fraction andnegative correlation to the sand-size fraction.

The variability of the internal or intra-particle void ratio mayexplain a significant part of the reduction of the total volume inbiogenic sediments as the sampled soils, in which skeleton and fossil

Fig. 4. Transformed of the stress–strain curves, for carbonate “bigger than 50%”. Iσ3=100 kPa; II σ3=150 kPa; III σ3=300 kPa.

fragments include a major fraction of the total void ratio (Golightlyand Hide, 1988).

In addition, internal erosion and piping are two different processesto be considered in these impervious cores, in order to avoid changesin its stability and water tightness.

Although beyond of the scope of this research, some SAR index andsoil aggregate self-stability studies were also made in order toquantify the internal erosion and piping hazards of the tested soilsamples, following recommendations published by Sherard (1982).

Fig. 5. Transformed of the stress–strain curves, for “Between 50% and 25%”. Iσ3=100 kPa; II σ3=150 kPa; III σ3=300 kPa.

Fig. 6. Transformed of the stress–strain curves, for “Smaller than 25%”. I σ3=100 kPa; IIσ3=150 kPa; III σ3=300 kPa.

Fig. 8. The parameters k and n, according to the content in carbonates.

166 F. Lamas et al. / Engineering Geology 122 (2011) 160–168

4.2. Stress–strain relationships

4.2.1. Behaviour in confined compressionThe results were compared using empirical relationships between

the compression index and Atterberg's LL (Valent et al., 1982),following the equations:

cv = 0:009� LL−10ð Þ ð11Þ

cv = 0:0054 2:6w−35ð Þ ð12Þ

cv = 0:54 e0−0:35ð Þ ð13Þ

Fig. 7. Relationship between the breakmodule and the dry density for the different cellspressures.

The previously presented results were clearly lower thanexpected, in all the three empirical correlations, although the valuesfrom Eqs. (11), (12), and (13) were above the acceptable level. Thesesignificant differences suggest an influence of the high carbonate

Fig. 9. Relationship between the module of initial elasticity and the dry density in thedam cores.

Table 7k, n, Kur and Poisson coefficient, as a function of the carbonate content.

% [(CO3)=] γd (Tn/m3) n k G F d Kur

Fco Abellán 15 1.79 0.45 104.52 0.41 0.14 2.5 209.0425 1.77 0.99 111.70 0.41 0.14 2.5 223.40

Portillo 30 1.82 1.36 86.02 0.41 0.14 2.5 172.0440 1.78 0.34 81.59 0.41 0.14 2.5 163.18

S. Clemente 60 1.71 0.61 212.15 0.41 0.14 2.5 424.3070 1.72 0.79 225.05 0.41 0.14 2.5 450.10

167F. Lamas et al. / Engineering Geology 122 (2011) 160–168

content on the marl behaviour under compression which differenti-ates this particular type of soil from the normal or general trend and,therefore, there is a need for special empirical relationships for marls(Bjerrum, 1967; Valent, et al., 1982).

The distribution of the coefficient of vertical consolidation, cv(Table 2), indicates lower compressibility of these marls with thecarbonate content of the fine-soil fraction.

4.2.2. Behaviour under triaxial compressionThe assessment of the results from the CU triaxial tests was based on

Eq. (7), which expresses the value of the tangent module from Eqs. (4),(5), and (6) (Duncan andChang, 1980). Theparameters of these sampleswere studied in order to determine the influence of the carbonatecontent. Each sampling series was considered to represent a single soilunit with a given carbonate content range. The samples tested areaggregate of mineral particles of feldspar, calcite, quartz, clay minerals,occasional minor amounts of gypsum, and some rock fragments.

The results indicate that the friction angle depends on theabundance of coarse particles, diminishing with the increase incarbonate content associated to the fine-soil fraction. Nevertheless,the cohesion did not change correlatively with the carbonate contentin the fine-soil fraction, although it clearly decreased when associatedwith the coarse particles. This relative independence of the cohesivestrength from the presence of carbonate in the fine-soil fraction isrelated to the low plasticity shown by most samples and theassociation of carbonate to the coarse-soil fraction where particlesof calcite are abundant.

The samples tested are mixtures of feldspar, calcite, quartz, clayminerals and gypsum in variable proportions. The correlationsbetween calcite content and the uniaxial compressive strength,imply that when the amount of carbonate is concentrated not in thefine matrix but in fragments of grain minerals, the contribution to thecohesive strength is very limited’. The results indicate an increase infrictional strength, expressed by the internal peak friction angle, withthe proportion of the coarse fraction.

4.2.3. Parameters for the hyperbolic modelFig. 8 shows the variation of initial elastic module, E1, with dry

density, γd, and carbonate content, for the experimental range ofconfining pressures of σ3=0.5; σ3=1.5 and σ3=3.0 Kp/cm2.

As reflected in Fig. 7, the results for the variation of the failurecoefficient Rf, with the dry density, support the use of Rf as a constantparameter in the hyperbolic model. For concentrations of carbonateabove 50% the Rf values varied between 0.9 and 1.05, with an averageof 0.92; and for concentrations below 50% the range is between 0.95and 0.78, with an average of 0.88. Taking in account results publishedby Dapena (1993a,b,c), Oteo (1995) and Wong and Duncan (1974) arange between 0.62 and 1.02 and an averaged value of 0.9 may bereasonably selected for the coefficient Rf, in the hyperbolic model.

A value of K=130 for the Poisson coefficient, which is maintainedpractically constant in the hyperbolic model, was obtained from theresults ofWong andDuncan (1974) concerning a low-plasticity silty clay.

The value of the Kur parameter was determined from dischargecurves in oedometer test, assuming that the exponential module ismaintained as in load curves. For Wong and Duncan (1974) the ratioKur:K varied from 1.2 in hard soils as dense sands, to 3.0 for soft soils asloose sands. Dapena (1993a,b,c) confirmed these results for clayeymarls with a value of Kur:K=2, i.e. Kur=2 K. Taking in account thisratio the values of the Kur parameter shown in Table 7were calculated.

5. Conclusions

The results, concerning three different series of samples (in S.Clemente , F. Abellán, and El Portillo earth dam quarries), show highcorrelations between the carbonate content and the parameters of thehyperbolic model, according to which the fragility of the marls, and its

bearing capacity (though slightly), increases with the carbonatecontent. Nevertheless, an overall view of the mechanical behaviour ofall the samples, by analysing these three series of samples together,shows no clear overall correlation between carbonate content and thehyperbolic model's parameter. These results are evidencing thecomplexity of the dependence relationships, which are influencednot only by the carbonate content but by several other variables.

Because of the limited presence of coarse-grain particles in the soilaggregate, the carbonate content concentrates in the fine-size soilfraction, disturbing the expected mechanical behaviour of the clay-type soils following the CAM-CLAY model, which is useful for typicalclays (Roscoe and Burland, 1968; Burland, 1990; Wood, 1994).Nevertheless, the hyperbolic model permits an adequate adjustmentof the results and the correlation of the carbonate content with otherparameters in each of the three series of samples studied instead ofthe percentage of fine particles or the dilatancy effect expressed bythe dry density (Chien and Oh, 1998).

The behaviour of clayey marly soils offers peculiarities which arenot being taken into account sufficiently, even when these types ofsoils are frequently applied for construction purposes in embank-ments and earth-dam cores.

As was mentioned above, in the context of engineering construc-tion of dams, problems arising from dissolution, piping, and internalerosion may be controlled by imposing requirements defined by theproject in its section on quality control of material, which includes the“At marble” test, SAR index calculation and the recommendations bySherard (1982) about the soil aggregate self stability of the dam core,not only in preliminary studies for the selection of appropriatequarries but also later during the dam-construction stages.

References

Al-Shayea, N., Abduljauwad, S., Bashir, R., Al-Ghamedy, H., Asi, I., 2003. Determinationof parameters for a hyperbolic model of soils. Proceeding of the Institution of CivilEngineers; Geotechnical Engineering 156 (GE2), 105–117 April.

ASTM, 2004. Annual book of ASTM standard. Section 4 construction, Vol. 04.08: Soil andRock, building Stones ASTM Philadelphia, PA, 19103, p. 1078.

Bjerrum, L., 1967. Engineering geology of Norwegian normally — consolidated marineclays, as related to settlement of building. Geotechnique 17 (4), 81–118.

Boscardin, M.D., Selig, E.T., Lin, Ruey-Syan, Gwo, 1990. Hyperbolic parameters forcompacted soils. Journal of Geotechnical Engineering ASCE 116 (1), 88–104.

Burland, J.B., 1990. On the compressibility and shear strength of natural clay.Geotechnique 40 (3), 329–378.

Chien, L.K., Oh, Y.N., 1998. Hyperbolic model parameters for mechanical behaviour ofreclamation soils. Geotechnical Engineering Journal 29 (1), 45–69 June, Publishedby the Southeast Asian Geotechnical Society; ISSN 0046–5828.

Dapena, E., 1993a. Influencia de la rigidez del material del núcleo de arcilla sobre elcomportamiento de una presa de materiales sueltos; Geotecnia de Presas demateriales Sueltos, Zaragoza, 1993. Sociedad Española deMecánica del Suelo y Cim.217–226.

Dapena, E., 1993b. Comportamiento tenso — deformacional de un suelo arcilloso paranúcleo de presas.; Geotecnia de Presas de Materiales Sueltos Zaragoza 1993.Sociedad Española de Mecánica del Suelo y Cim. 88–97.

Dapena, E., 1993c. Comportamiento tenso— deformacional de zahorras naturales comomaterial para espaldones de presas.; Geotecnia de Presas de Materiales SueltosZaragoza 1993. Sociedad Española de Mecánica del Suelo y cim 77–87.

Domaschuk, L., Valliappan, P., 1975. Nonlinear settlement analysis by finite elements.Journal of Geotechnical Division ASCE 101 (7), 601–614.

Duncan, J.M., 1980. Hyperbolic stress relationships. In: Yong, R.K., Ko, H.-Y. (Eds.),Proceedings Workshop on Limiting Equilibrium, Plasticity and Generalized Stress–Strain in Geotechnical Engineering: ASCE, pp. 443–460.

168 F. Lamas et al. / Engineering Geology 122 (2011) 160–168

Duncan, J.M., Chang, C.Y., 1980. Review of shearing strength for cohesive soils. Int.Symp. On Earth Fills ICE London.

El Amrani N.; (1996); “Caractérisation géotechnique des sols cohésifs du bord sudorientel de la depresión de Grenade”; Thése de 3eme cycle. Université AbdelmalekEsaadi Tetouan Maroc; 312.

El Amrani, N., Chacon, J., 1996. Altérabilité, gonflement et propriétés d'indices desmarnes néogenés de la depresion de Grenade, Espagne. Bulletin of the InternationalAssociation of Engineering Geology 54, 77–89 Paris.

El Amrani, N., Lamas, F., Irigaray, C., Chacón, J., 1998. Engineering geologicalcharacterization of Neogene marls in the Southeastern Granada Basin, Spain.Engineering Geology 50, 165–175.

El Amrani, N., Lamas, F., Irigaray, C., Chacón, J., Oteo, C., 2000. The residual shearstrength of neogene marly soils in the Granada and Guadix basins, southeasternSpain. Bulletin of Engineering Geology Environement. Springer-Verlag, pp. 99–105.

Fernández, J., Guerra-Merchán, A., 1996. A coarsening-upward mega sequencegenerated by a Gilbert-type fan-delta in a tectonically controlled context (UpperMiocene, Guadix-Baza Basin, Betic Cordillera, southern Spain). SedimentaryGeology, 105. Elsevier, pp. 191–202.

Fukue, M., Nakamura, T., Kato, Y., Naoe, K., 1996. Correlations among carbonate content,accumulation rate and topography of seabed. Soils and Foundations, 36, 1. JapaneseSoc of Soil Mechanics & Foundation Engineering, Tokyo, Japan, pp. 51–60.

Golightly, C.B., Hide, A.F.L., 1988. Some fundamental properties of carbonate soils. Proc.Int. Conf. on Calcareous Sediments, Vol 1, pp. 69–78. Perth, Australia.

ISSMFE, 1987. Evaluation of swelling pressure of expansive soils in laboratory. Draft standardof Int. Soc. Soil Mec. And Found. Eng. Technical Comitee of Expansive Soil TC6, pp. 1–5.

Jambu, N., 1963. Soil compressibility as determined by oedometer and triaxial test.European Conf. On soil Mechanics SMFE, 1, pp. 85–88. Salisbury.

Jiménez Salas, J., 1993. Núcleos, espaldones y escolleras: estudio de préstamos ycaracterísticas de los materiales. Geotecnia de Presas de Materiales SueltosZaragoza 1993 Sociedad Española de Mecánica del Suelo y cim, pp. 11–44.

Kondner, R.L., Zelasko, J.A., 1963. A hyperbolic stress strain formulation for soils. 2nd Int.Pan-American Conf. Of Soil Mechanics and Found. Eng. Brazil, Vol. 1, pp. 289–324.

Lamas F.; (2003); “Influencia del contenido de carbonatos en las característicasresistentes, en materiales margosos usados en la construcción de núcleosimpermeables de presas”; Tesis Doctoral; Universidad de Granada; 452 p. Granada(Spain).

Lamas, F., Irigaray, C., Chacón, J., 2002. Geotechnical characterization of carbonate marlsfor the construction of impermeable dam cores. Engineering Geology, 66. Elsevier,pp. 283–294.

Lamas, F., Irigaray, C., Oteo, C., Chacón, J., 2005. Selection of the appropriate method todetermine the carbonate content for engineering purposes with particular regardto marls. Engineering Geology, 81. Elsevier, pp. 32–41.

Oteo, C., 1995. Propiedades geotécnicas de los materiales margosos del Sur de España ysu aplicación a problemas de excavaciones. 2º Coloquio Internacional sobre elEnlace Fijo Europa — África, pp. 1–15. Sevilla (Spain).

PG-3, 2004. Pliego de condiciones técnicas generales para obras de carreteras y puentes.Centro de Publicaciones. Ministerio de Fomento, Dirección General de Carreteras.;Madrid. ISBN: 978-84-498-0730-5.

Pierre, P., Caruso, A., Blanc-Valleron, M.M., Rouchy, J.M., Orzsag-Sperber, F., 2006.Reconstruction of the paleoenvironmental changes around the Miocene–Plioceneboundary along a West–East transect across the Mediterranean. SedimentaryGeology, 188–189. Elsevier, pp. 319–340.

Rodier, J., 1978. L'Analyse de L'Eau, Eaux Naturelles, Eaux Résiduaires et Eau deMer. Ed.Dunod, París, p. 578.

Roscoe, K.H., Burland, J.B., 1968. On the Generalised Stress Strain Behaviour on WetClay. In: Heyman, Leckie (Eds.), Eng. Plasticity Cambridge Univ. Pres, pp. 535–609.

Sherard, J.L., 1982. Filter test on fine-grained clayey soils using high water pressures.Manuscripts S.C.S. Soil Mechanics Laboratory Lincoln, Nebraska.

Skempton, A., Bishop, A., 1954. Soils. Building materials, their elasticity and inelasticity.North Holland Public Company, Amsterdam. Chap. X.

UNE, 2001. Normas españolas de ensayo. AENOR.Valent, P.J., Altschaeffl, A.G., Lee, H.L., 1982. Geotechnical properties of two calcareous

oozes. In: Demars, Chaney (Eds.), Geotechnical Properties Behaviour andPerformance of calcareous Soils, ASTM STP 777: ASTM, pp. 79–95.

Warshaw, C.H., Roy, R., 1961. Classification and scheme for the identification of layersilicates. Bulletin Geological Society of America 72, 1455–1492.

Wong, K., Duncan, J.M., 1974. Hyperbolic stress–strain parameters for non linear finiteelement analyses of stresses anmovements in soilmasses. Report No TE-74-3: 74-3.University of California, Berkeley; USA.

Wood D. M.; (1994); “Soil behaviour and critical state soil mechanics”; CambridgeUniversity Press 1994; 462 p. Press Syndicate of the University of Cambridge U. K.