Geotechnical properties of evaporite soils of the Dead Sea area

Engineering Geology 101 (2008) 236–244

Contents lists available at ScienceDirect

Engineering Geology

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

Geotechnical properties of evaporite soils of the Dead Sea area

Sam Frydman a,⁎, Josef Charrach b, Ian Goretsky c

a Faculty of Civil and Environmental Engineering, Technion—Israel Institute of Technology, Haifa 32000, Israelb Dead Sea Works Ltd., Beer-Sheba, Israelc Building and Infrastructure Testing Laboratory, Haifa, Israel

⁎ Corresponding author. Department of StructuralManagement, Faculty of Civil and Environmental engineeof Technology, Haifa 32000, Israel. Tel.: +972 4 8293320

E-mail address: [email protected] (S. Frydman).

0013-7952/$ – see front matter © 2008 Elsevier B.V. Aldoi:10.1016/j.enggeo.2008.06.003

A B S T R A C T

Article history:

Keywords:Evaporite soilsRock saltDead SeaLisan marlSaline soils

west point on earth and is tectonically subsiding. During the Holocene Period theclimate became much drier with increasing evaporation whereby initially lacustrine sediments weredeposited from the non-marine brines, giving a multi-layered stratigraphy of lime carbonate and halitesediments. The lime carbonate sediments are comprised of laminated, clay to silt sized, clastic sediments(calcite) and authigenic aragonite and gypsum. The halite commonly appears as rock salt. Chemicalindustries, based on harvesting the salts from the Dead Sea, have developed on both the Israeli and theJordanian sides of the basin. The lime carbonate soils are used for dike construction, and these soils, togetherwith significant salt layers, are encountered in the foundations of structures, dikes, and tailings dams,requiring definition of their geotechnical properties. Use of standard soil mechanics definitions and testingapproaches for the lime carbonates have been found inapplicable, particularly in view of their exceptionallyhigh saline content, and it has been necessary to develop new concepts. The rock salt is encountered atshallow depths, with unit weights considerably lower than those usually discussed in the literature, and withcorrespondingly different mechanical properties. The geotechnical properties of these soils, and approachesused to define them, are discussed in the paper.

© 2008 Elsevier B.V. All rights reserved.

1. Introduction

TheDead Sea Basin (Fig.1) is locatedwithin theDead Sea Transformthat separates the Arabian and Sinai plates. It is the lowest point onearth and is tectonically subsiding. The basin is part of an active plateboundary that connects the Red Sea toTurkey. It formed during the lateCenozoic period by the break up of the once continuous Arabian–African continent (Garfunkel and Ben Avraham, 2001). The basin isbordered by steep normal/strike slip fault scarps and by longitudinalintra-basinal faults. These faults delimit the interior graben from theless subsidingmarginal blocks. The stratigraphy of the graben fill, up to13 km (Ginsburg and Ben Avraham, 1997), consists of severalkilometers of clastic sediments, mainly quartzose sandstones of theMiocene age. During the late Miocene–lower Pliocene, a 2 km thickevaporate sequence was deposited which later formed diapirs and theMt. Sdom salt wall (Charrach, 2006a). The upper 5 km of sedimentsconsists of continental fluviatile and lacustrine sediments. During thelate Pleistocene (70,000–11,000 yr BP), lime carbonate and gypsumsediments were deposited from the terminal Lake Lisan, a precursor

Engineering and Constructionring, Technion—Israel Institute; fax: +972 4 8293323.

l rights reserved.

lake to the present day Dead Sea. During the Holocene period therewere major climatic changes ranging from arid to subtropical (Enzelet al., 2003; Klinger et al., 2003), with added tectonic subsidence.During arid periods the Dead Sea basin was divided into a deepnorthern basin and a shallow southern basin. Lacustrine sedimentswere deposited ranging from lime carbonate sediments to halite bedsand alluvial fan detrital sediments at the deltas of wadi channels. Thelime carbonate sediments are typically comprised of laminated, clay tosilt size clastic sediments (predominantly calcite), and authigenicaragonite and gypsum. The clastic detrital material was brought intothe basin by seasonal floods. The aragonite laminae may representdrier years when the upper water mass was diminished due to lowerwater input and enhanced evaporation (Migowski et al., 2004). Withcontinued evaporation, halite (NaCl) precipitated out of the brine. Thehalite crystallizes at the brine/air interface and gravity settles to thebase. The halite sediments range frommassive rock to friable sand likesediments. The halite is in general recrystallized in the form of cleareuhedral cubic crystals. This will occur due to the influx of undersaturated brine with a climatic change or a flooding event, into theshallowsouthernbasin of theDead Sea, dissolving the halite sedimentsand then crystallizing out in equilibrium under a cover of overlyinglime carbonate sediments. The halite sediments in the upper 30 m areporous with voids and channel ways. Large euhedral crystals areobserved within these voids. Displacement hopper crystals are alsolocated within the lime carbonate sediments. A further phase of haliteprecipitation has been observed, in particular within the artificial

Fig. 1. Map of the Dead Sea area.

Fig. 2. Change in calculated water content with drying time at 60 °C.

237S. Frydman et al. / Engineering Geology 101 (2008) 236–244

halite pans which are located on the natural Holocene sediments;whereby halite precipitates out by the common ion effect when lighterNaCl brine percolates into heavier oversaturated brine. This mechan-ism for halite precipitation was first described by Raup (1970). Thechemically precipitated halite crystallizes in the interstitial voids andcements the evaporated salt crystals forming very hard competent, lowporosity salt sediments (Charrach, 1986).

Core drillingwithin the southern basin of the Dead Sea has enabledthe construction of a stratigraphic column for the Holocene periodshowing multiple lime carbonate/halite interbedding suggestingmultiple changes in the climate during this period. The lime carbonateand halite sediments are statigraphically correlatable throughout thebasin. During this period no carnallite sediments have beendiscovered and no desiccation textures have been observed, whichsuggests that the basin never dried out and a continuous sedimentaryrecord of greater than 100 m is preserved. This illustrates the intensityof tectonic activity during this period. The lime carbonate sedimentsrepresent periods of relative brine dilution, a fluviatile climate, whilehalite sediments represent periods of relative salinity increase, an aridclimate (Charrach, 2006b).

The Holocene sedimentation of the southern basin of the Dead Seais comparable with quaternary non-marine, closed basins like: Bristol

Dry Lake and Cadiz Lake Basins California, Great Salt Lake Utah,Qarhan Playa, China, Lake Uyani, Bolivia, Danakil Depression, Ethiopia,and the Ushtagan Lake in Kazakhstan (Casa and Lowenstein, 1989;Hardie, 1990). On burial these deposits lose their porosity and formlithified rock salt and shales which on deeper burial will creep flowunder low stress conditions to form salt walls and diapirs. The nowvertically bedded Mt. Sdom salt wall of the Upper Miocene age, is a2 km sequence of more than 40 evaporite cycles (dolomitic shales tocarnallitites), which still exhibit their original syndepositionaldiagenetic sedimentary textures (Charrach, 2006a).

The present Dead Sea is a terminal lake of Mg, Ca, Na, K chloridebrine composition. Until recently, the lake inundated both thesouthern and northern basins of the Dead Sea. Due to pumping ofthe fresh headwaters from the Sea of Galilee, the river Jordan and itstributaries, the Dead Sea has receded and the southern basin is nowdry except for the artificial evaporation pans of the Dead Sea Works inIsrael and the Arab Potash Company in Jordan. The chloride brines arenow saturated with respect to halite, except in the areas of the alluvialfan deltas and the areas close to the fault escarpments.

Large chemical industries, based on harvesting the minerals fromthe Dead Sea, have developed on both the Israeli and the Jordaniansides of the basin. Use of the lime carbonate soils for dike construction,and their presence, together with significant salt layers, in thefoundations of structures such as, dikes, tailings dams, chemical plantsand power stations, requires definition of the geotechnical propertiesof both these materials. This paper describes some of the uniquefeatures of the geotechnical properties of these Dead Sea deposits, andsuggests some original approaches for dealing with them.

2. The lime carbonates

2.1. Effect of brine solution on water content

The chloride brines result in exceptionally high salinity of the limecarbonate soils, rendering the use of standard soilmechanics definitionsand test procedures inapplicable. A prime example is the determinationof the water content of these sediments. According to the conventionalAmerican (ASTM, 2005) andBritish (BSI,1990) testmethods, samples forwater content measurement are oven dried to 110 °C until constantweight is achieved. In the present case, this method is the source ofmajor errors due to the precipitation of halite (NaCl), carnallite(KCl·MgCl2 MgCl2 ·6H2O), and bischofite (MgCl2 ·6H2O) during thedrying process. Not only is the dry sample weight increased due to thesalt precipitation, but water is also taken up as water of crystallization,adding further inaccuracy to the measurements. Drying to constantweight is further impaired since a mineral crust crystallizes out on thesurface of the sample, inhibiting the release of all the moisture. Clearly,evaluationof all soil characteristicsdependentonmeasurementofwater

Fig. 3. Elementary cube showing phases of the Dead Sea soil.

238 S. Frydman et al. / Engineering Geology 101 (2008) 236–244

content (e.g. void ratio, dry density and their derivatives—e.g. compres-sion index) will also suffer from the problematics of this measurement.

2.2. Water content determination

The standard procedure for evaluation of water content of soils(e.g. ASTM, 2005; BSI, 1990) involves weighing the soil at its naturalwater content, and then drying it to constant mass at about 110 °C—ASTM requires a temperature of 110±5 °C, while BSI specifies atemperature between 105 °C and 110 °C. The British standard addsthat certain soils contain gypsum which loses its water of crystal-lization on heating, so affecting the evaluated water content by about0.2% for each 1% of gypsum. Consequently, the standard recommendsthat soils containing gypsum should be heated at less than 80 °C. TheAmerican standard also refers to materials containing gypsum,suggesting that they should be dried at a temperature below 60 °C,and further makes mention of soils containing water with substantialamounts of salt. In these cases, themass of solids obtained after dryingincludes the previously soluble salts, and this should be accounted forin calculating the water content.

The Dead Sea lime carbonate sediment (often referred to as “LisanMarl”) is an extreme example of the relevance of the commentsincluded in the above standards. Thematerial is made up of about 95%calcite, with the other 5% being aragonite, quartz and gypsum.

The brine in the soil is saturated with respect to halite, and has aspecific gravity of about 1.25 —i.e. it contains approximately 25% byweight of dissolved salt. The problem of water content determinationin such soils is not only one of procedure, but also one of definition; dowe, for example, mean the weight of fresh water divided by the totaldry weight of solids (i.e. soil particles + salt)? For a long period, thisdefinition was adopted, and water content was determined by thetraditional procedure set out in the standards—weighing wet,

Fig. 4. Fluid content versus water content.

weighing dry (albeit after drying at 60 °C), and then calculatingwater content as the ratio between the difference between the twoweights and the dry weight. Even this definition proved problematic,as illustrated in Fig. 2. This figure shows results of tests on 32 differentspecimens which, after being weighed at in-situ water conditions,were dried at 60 °C for up to 8 days, with intermediate weighingsbeing made. The water content values, defined as above, are shownrelative to the values after drying for 3 days. It is seen that even after8 days of drying, an equilibrium condition was not usually achieved.

In addition to the above procedural problems, the classicaldefinition of water content used does not appear to be physicallymeaningful, as it includes the precipitated salts as part of the solidweight, and ignores their part in the fluid weight. It would appear tobe more meaningful to define a ratio between the weight of brine tothe weight of solid soil particles. Noorany (1984), discussing marinesoils, emphasized this distinction, and suggested two definitions—water content, defining the former ratio, and fluid content definingthe latter. He developed phase relations, including a relationship forcalculation of fluid content, from known water content and fluidsalinity. Unfortunately, in the present case, the difficulties involved inmeasuring water content make the use of such relationshipsirrelevant. Consequently, an alternative approach has been sought.In the Dead Sea area, ground water table is commonly close to groundsurface, so that the lime carbonate deposits under consideration arebelow water, and therefore saturated. For the case of these saturatedlime carbonates, an alternative approach has been developed forevaluation of fluid or water content.

Fig. 3 shows the elementary cube, indicating the phases of anelement of fluid saturated soil. The indices sw and sa indicate brineand salt, respectively; all other symbols are as commonly used ingeomechanics. As the soil solids consist, mainly, of calcite (specificgravity 2.71), with a small quantity of aragonite (specific gravity 2.93)and some other minerals, the specific gravity, Gs, of the soil solids may

Fig. 5. Standard liquid limits of washed versus unwashed soil.

Fig. 6. Standard liquid limit versus value based on calculated fluid content. Fig. 8. Liquid limit of washed soil versus value based on Eq. (3).

239S. Frydman et al. / Engineering Geology 101 (2008) 236–244

be approximately taken as 2.75. The brine has a specific gravity, Gsw,of 1.24–1.26 and a value 1.25 may reasonably be adopted.

Referring to Fig. 3, the following phase relations can be developed:

Total unit weight;γt : γt ¼Gs⋅Vs⋅γw þ Vsw⋅Gsw⋅γw

Vsþ Vswð1Þ

Resulting in : Vsw ¼ Vs⋅Gs⋅γw−γt

γt−Gsw⋅γwð2Þ

Fluid content;wf : wf ¼WswWs

¼ Vsw⋅Gsw⋅γw

Vs⋅Gs⋅γw¼ Gsw

Gs ⋅Gs⋅γw−γtð Þγt−Gsw⋅γwð Þ

ð3Þ

In order to evaluate fluid content of saturated specimens of the DeadSea soils, Eq. (3) has been applied, using total unit weight measurementsmade on undisturbed soil samples, and adopting the specific gravityvalues specified above; γw, is the unit weight of cleanwater. Fig. 4 showsthe relationship between the water content, w, obtained after drying at60 °C for 8 days, and the fluid content, wf, obtained from Eq. (3). Thedifference between the two is seen to be extremely large, illustrating theproblematicsofwater contentdefinition andevaluation in these soils. It isfelt by the authors that “fluid content”, as defined above, and evaluatedhere, is the physically relevant measure of water content for engineeringpurposes. In order to use this definition, it is necessary to have un-disturbed samples for measurement of total unit weight. Fig. 4 shows,however, that there appears to be a relationship between the two values,allowing at least an approximate evaluation of fluid content on the basisof standard water content measurements.

The proposed procedure has been adapted for field estimation offluid content, by measuring core samples and weighing them on ananalytic balance immediately after they are obtained from a borehole. Amajor advantage of this procedure is that by performing the measure-ments at field conditions of temperature and pressure, precipitation ofNaCl due to climatic changes is avoided.

Fig. 7. Liquid limit based on Eq. (3) versus value from correlation in Fig. 4.

2.3. Consistency limits

Beingmeasures ofwater content at particular conditions of plasticityor liquidity, determination of the consistency limits of theDead Sea soilsobviously suffers from the same problems as does determination ofwater content. However, an additional problem now arises—what ismeant by, for example, liquid limit? Do we mean the percent of brinewhich will bring the natural soil to a “liquid” state? Or do we mean theamount of fresh water which will bring the clean, saltless soil (i.e. afterwashing it clean of salt) to the “liquid” state? If the purpose is to definean in-situ fluid content at which the soil becomes “liquid” withincreasing brine solution, the former definition would be the relevantone. On the other hand, if the liquid limit is being evaluated forclassification purposes of the soil, the value determined should beindependent of the particular fluid present at a particular site, andshould depend only on the properties of the soil particles. In this case,the latter definitionwould bemore relevant. It appears, then, that thereis no unique answer as to how to define the consistency limits of thesesoils, and probably both values should be determined.

Liquid limit values of Dead Sea soil have been determined in fourways, as follows:

(a) According to the standard procedure, using brine as the fluid,and drying at 60 °C.

(b) As (a), but using the relationship in Fig. 4 to estimate the fluidcontent at the liquid limit condition, and defining the liquidlimit as this fluid content.

(c) Undisturbed samples were used, and initial fluid content wasestimated from Eq. (3). The samples were then remolded, andliquid limit tests performed keeping record of the amountof brineadded for each test point. Thefinal fluid content of each test pointcould then be calculated, and used for evaluation of liquid limit.

Fig. 9. Compression curves obtained by two procedures.

Fig. 10. Comparison of Cc values resulting from different procedures.

240 S. Frydman et al. / Engineering Geology 101 (2008) 236–244

Note that method (a) is based on determination of a water content,while methods (b) and (c) define liquid limit as a fluid content. Allthree methods are based on the in-situ soil and fluid.

(d) The soil was washed free of all soluble salts, and liquid limit wasdetermined by the standard method, using distilled water.

Figs. 5–8 show relationships between various values obtained fromthese tests. Fig. 5 shows the relationship between the standard liquidlimits obtained from tests performed on natural material with brine(method a), and onwashedmaterialwith distilledwater (methodd). It isseen that the washed soil yields much higher liquid limits than theunwashed soil; in fact, while the unwashed soil always has a liquid limitbelow 50, and would, therefore, always be classified as CL according tothe Unified Classification System, the washed soil usually has a liquidlimit greater than 50, and so would usually be classified as CH.

Fig. 6 shows a comparison of results obtained by methods a and c.Again, a large difference is noted between the results; on the basis ofthe standard tests, the soil would be classified as CL, while the liquidlimit defined on the basis of fluid content, as calculated from Eq. (3), isgenerally greater than 50, and so the soil would be classified as CH.

Fig. 7 provides an opportunity to check the usefulness of thecorrelation shown in Fig. 4. The liquid limit obtained on the basis of fluidcontent and Eq. (3) is compared to the value obtained from the standardtest, corrected by the correlation of Fig. 4. Of thenine comparisonsmade,seven are seen to be reasonable, suggesting that the correlation may bepotentially useful for providing anapproximate estimate offluid contentand liquid limit without the need for undisturbed samples.

A comparison between the liquid limit obtained on the basis offluid (brine) content, using Eq. (3), and that of washed soil is shown inFig. 8. It is seen that the former is generally higher.

Summarizing the above observations, the liquid limit obtainedusing the standard testing method, with brine solution, appears to beextremely low, partly because the water content values measured aretoo low (the soil has not dried sufficiently), partly because the salts

Fig. 11. Relationship between compression index, Cc and liq

influence the surface activity of the particles, and partly because thefluid content, rather thanwater content should be used as the basis forcalculation. It is recommended that method (c), based on Eq. (3),provides a more physically relevant definition for in-situ liquid limit,indicating the consistency property of the soil under field conditions.The standard approach, corrected by a correlation such as that shownin Fig. 4, may be used to obtain an approximate estimate of this value.On the other hand, for classification of the soil particles, divorced fromtheir geoenvironment, the liquid limit should be obtained from awashed sample, using distilled water.

Evaluation of plastic limit of these soils suffers from the sameproblems as above. A procedure has been developed for estimation offluid content at the plastic limit by measuring the weight anddimensions of anundisturbed sample, and then using Eq. (3) to estimatefluid content. A portion of theundisturbed sample is then separated, andused for plastic limit determination, air drying as necessary and keepingtrack of weight (i.e. moisture) changes. During drying, some mineralsmay precipitate, and this is corrected for in calculation of the solid andfluid weights. For classification of the soil, washed samples and distilledwater should be used. No clear relationship was seen between plasticlimit values obtained by the standard method and by the new methoddescribed, although a trend was observed between plastic limits ofwashed samples and values obtained by the suggested method.

2.4. Compressibility

Since the void ratio used for establishment of compression curves(e versus σ′) is simply an expression of the moisture/fluid content ofthe soil, the resulting relationship between void ratio, e, and effectivecompression stress, σ′, clearly depends on the procedure forestimating e. Fig. 9 shows the result of a typical consolidation teston a brine saturated lime carbonate sample. Two compression curveshave been drawn—one based on calculation of void ratio from watercontent obtained from standard oven drying, and the second based onvoid ratio calculated from the fluid content (Eq. (3)). The significantdifference between the curves is apparent. Since the compressionindex, Cc, of saturated soil is a function of change in void ratio or ofmoisture/fluid content, different estimates of void ratio result indifferent estimates of Cc. Fig. 10 shows the relationship between the Ccvalues obtained from the two approaches; a quadratic relationshipappears to be reasonable between the two. Clearly, use of Cc obtainedby the standard procedure would result in unconservative (i.e. toolow) estimates of settlement.

Terzaghi and Peck (1967) suggested the following empiricalrelationship between compression index, Cc, and liquid limit, wL:

Cc ¼ 0:009 WL−10ð Þ F30k errorð Þ ð4Þ

Fig. 11(a) and (b) shows the relationships between Cc and wL asobtained from tests on Dead Sea lime carbonate soil. In Fig. 11(a), both

uid limit, LL; (a) standard procedure; (b) use of Eq. (3).

Fig. 14. Failure stresses in triaxial and Hoek tests on rock salt cores of bulk unit weights14 kN/m3 and 18 kN/m3.

Fig. 12. Failure stresses in triaxial tests on saturated, undisturbed lime carbonate.

241S. Frydman et al. / Engineering Geology 101 (2008) 236–244

liquid limit and Cc have been evaluated from standard test procedures—oven drying at 60 °C for 8 days. Fig. 11(b) shows values of bothwL and Ccbased on use of Eq. (3). The figures show that while the standardprocedure results in large deviation from Terzaghi and Peck's relation-ship, use of Eq. (3) results in reasonable, agreement, with Terzaghi andPeck's relation giving a slightly conservative estimate.

2.5. Shear strength

Numerous triaxial tests have been carried out on undisturbedsamples of the lime carbonate soils. The tests have included bothisotropically and anisotropically consolidated, undrained (ICU andACU) tests, on brine saturated samples, with dry unit weights (basedon fluid contents calculated from Eq. (3)) ranging from 7–15 kN/m3.No systematic relation between strength parameters and dry unitweight was observed; a summary of effective stress values at failure isshown in Fig. 12. It is seen that one strength envelope can reasonablybe fitted to all the test data, yielding an effective cohesion of zero, andan effective friction angle of 34°. These parameters are not typical ofclay soils, and are significantly different from those representative ofother Israeli clays (Frydman, 2000), which have an average effectivecohesion of about 10 kPa, and friction angle of about 28°. Thisdifference may be explained by the very different mineralogicalcomposition of the Dead Sea material, composed predominantly ofcalcite and not clay minerals, whereas the Israeli clays are comprisedlargely of montmorillonite (spectite) clay minerals.

3. Rock salt

3.1. Porosity and permeability

Testing has been carried out on cores of rock salt extracted fromdrill holes in the upper 35 m of the Holocene section. The porosity has

Fig. 13. Rock salt permeability as a function of porosity.

been found to vary between 5% and 30%, with no clear trend withdepth. Sonnenfeld (1984) discussed porosity of freshly precipitatedevaporites, which he described as being initially very porous, loose,watery slush that gradually firms up, with a porosity of 10%–50%. Hisspecific inclusion of halites of the Dead Sea area in this group issubstantiated by the presently measured porosity values. Consider-able diagenic changes are required to convert this material into amassive, compact deposit. After burial and considerable compaction,the porosity of halite drops to less than 5%, and even to less than 0.5%.Hofrichter (1976) suggested that salt becomes tight (i.e. effectivelyzero porosity and permeability) under an overburden of about 100 m;Baar (1977) suggested that this occurs at depths exceeding 300 m.

Fig.13 shows the relationshipbetweenpermeabilityandporosity. Thetestswere carried out in triaxial equipment, under a confining pressure ofbetween 120–170 kPa, and a pressure head of 100–150 kPa. There is seento be a general trend of increasing permeabilitywith increasing porosity;the rangeof permeability,10−8–10−5m/s is consistentwith that quotedbySonnenfeld for other freshly precipitated evaporites (e.g. the sebha ElMelah, Tunisia, which has a porosity of 25–54%).

3.2. Shear strength

Triaxial shear tests performed on samples of the rock salt, haveincluded both conventional drained tests, and consolidated undrainedtestswith porewater pressuremeasurement, and tests performed in theHoek cell (Hoek and Franklin,1968). Fig.14 shows effective failure stressconditions obtained from these tests, on the assumption that the Hoekcell tests are essentially drained, with no significant pore pressuredevelopment. Results are shown from tests on samples with bulk unit

Fig. 15. Unconfined compression strength of rock salt cores.

Fig. 16. Mohr failure stress circles for (a) triaxial; (b) unconfined compression.

Fig. 17. Effective cohesion of rock salt (ϕ′ assumed 55°).

242 S. Frydman et al. / Engineering Geology 101 (2008) 236–244

weights of about14 kN/m3 and 18 kN/m3. It is seen that triaxial andHoektests give consistent results,with common failure envelopes. The frictionangle, ϕ′, appears to be independent of unit weight, and equal to about55°, consistent with the order of values (55°–65°) presented byKlayvimut (2003), quoting Hansen et al. (1984), for a range of rocksalts from ten different locations in theUnited States. On the other hand,cohesion, c′, appears to be a function of bulk unit weight. An estimate ofthis function has been made from a combined analysis of the results ofthese triaxial tests, and results obtained from a large number ofunconfined compression tests performed on samples of different bulkunit weights. The relationship between unconfined compressionstrength and bulk unit weight, as obtained from these tests, is shownin Fig. 15. Unconfined compression strength is seen to be highlydependent on bulk unit weight, varying from the order of 2 MPa to10MPawhen unit weight varies from 14 kN/m3 to 20 kN/m3. This rangeof strengths is significantly lower than the values quoted by Klayvimut(2003) based on strengths reported by Hansen et al. (1984) (13.3 MPa–33.6MPa); Fuenkajorn andDaemen (1988) (18.4MPa); Ratigan andVogt(1993) (15MPa–30MPa); and Boontongloan (2000) (18.5MPa–26MPa).Unfortunately, the data quoted by Klayvimut does not include bulk unitweights, and so it is difficult to evaluate the relevance of a comparison ofhis data to that for the Dead Sea deposits. However, it is suspected thatthe bulk unit weights were higher than those of the Dead Sea samples.

The estimation of cohesion was made assuming that both the Hoekcell and unconfined tests are essentially drained, with no significantpore pressure development, and that the friction angle of the rock salt is55°, independent of bulk unit weight. Fig. 16(a) and (b) shows Mohrcircles of effective stress for triaxial and unconfined samples, respec-tively, at failure. From these figures, it can be shown that for any sampleat failure, cohesion, c′, can be calculated from the following expressions:

Triaxial : c 0¼ σ1−σ3ð Þ− σ1 þ σ3ð Þ sin�0n o

=2 cos�0 ð5aÞ

Unconfined : c 0 ¼ su 1− sin�0� �

= cos�0 ð5bÞ

where su=unconfined shear strength, equal to half the unconfinedcompression strength.

The effective cohesionwas calculated fromEqs. (5a) and (5b) for eachsample tested in triaxial, Hoek cell and unconfined compression test,and the results are shownplotted against bulk unit weight in Fig.17. Thesets of data appear to followa consistent trend, cohesion increasingwithbulk unit weight. Again, the range of cohesion values obtained is lowerthan those commonly published in the literature. For example, Hansenet al's (1984) results, as quoted byKlayvimut (2003) are in the range 2.0–4.7 MPa, compared to the values of up to 1.8 MPa shown in Fig. 17.

3.3. Elastic modulus

Baar (1977) claimed that elasticity parameters of rock salt, derivedfrom uniaxial compression tests, are insignificant, since under in-situ

conditions, rock salt exhibits plastic behavior as soon as the low creeplimits are exceeded. He suggested that such parameters should never beused in calculation of pillar strengths and stress accumulations nearunderground openings. Baar's comments were directed to problems ofunderground openings, where long-term behavior, under decreasingconfining stress conditions, are of interest. The purpose of the presentinvestigations was to provide data for foundation problems, wherestress levels are usually increasing. Although the creep problem is stillrelevant, the creep limit is higher under compression stresses thanunder tension, and creepmaybe less significant at thehigh safety factorsassociated with working loads. Secant modulus values were estimatedfrom the stress-strain curves obtained in the unconfined compressiontests, as a function of strain, following correction of the initial portion ofthese curves for seating and contact effects. Considerable variabilitywasobserved in themodulus—strain functions, and theyare shown, for threebulk unit weight ranges, in Fig. 18(a)–(c). The three figures have beendrawn to the same scale in order to facilitate comparison between them.Theelasticmodulus is seen tobe significantlydensitydependent; at 0.5%strain, E decreases from the order of 1000 MPa at a bulk unit weight of18–20 kN/m3 to the order of 300 MPa at a bulk unit weight less than16.5 kN/m3. Small strain values of E were also estimated, based oncompression wave velocities measured on cores, using Pundit equip-ment (Elvery, 1972). Fig. 19 shows compression wave velocity, vp, as afunction of bulk unit weight. The velocity values were translated to Evalues, applying the following elasticity expressions:

G ¼ v2p ρ 1−2�ð Þ= 2−2�ð Þ ð6aÞ

E ¼ 2G 1þ �ð Þ ð6bÞwhere ρ=mass density, ν=Poisson's ratio, taken as 0.4, within therange 0.35–0.42 quoted by Klayvimut (2003) from data of Boon-tongloan (2000) for Thai rock salt.

Fig. 18. Elastic modulus of rock salt cores.

243S. Frydman et al. / Engineering Geology 101 (2008) 236–244

The small strain moduli are seen to be an order higher than thosemeasured in the unconfined compression tests at strain levels of theorder of 0.1%–0.5%; this is consistent with what is generally found forgeomaterials (e.g. Kulhawy and Mayne, 1990). It should be notedthat these E values are extremely sensitive to the assumed valueof Poisson's ratio, slightly higher assumed ν values resulting insignificantly lower E values.

4. Concluding remarks

The standard procedure for evaluating water content is notapplicable to soils containing highly saline water. An alternativeprocedure has been suggested here for use with the highly saline, limecarbonate soils, based on testing of undisturbed samples, resulting insignificantly higher values. Use of these higher values influences othersoil parameters; for example, compression index, is found to be

Fig. 19. Compression wave velocity, vp, of rock salt cores.

higher, and more conservative, indicating that settlements would belarger than those estimated on the basis of conventional procedures.

Due to its shallow burial depth, the rock salt of the Dead Sea area isless dense thanmost rock salts reported in the literature, and as a result,use of mechanical properties reported in the literature may bedangerously unconservative. The data obtained in the present investiga-tion support this caution, and provide a basis for analysis and design ofengineering structures constructed on this and similar materials.

Acknowledgments

The authors thank the Dead Sea Works Ltd. and Eng. S. Kogan, forpermission to publish the material presented in this paper.

References

ASTM, 2005. Standard D2216, Standard test method for laboratory determination ofwater (moisture) content of soil and rock mass. Annual Book of Standards, Section4-Construction.

Baar, C.A., 1977. Applied Salt-Rock Mechanics 1: The In-situ Behavior of Salt Rocks.Elsevier. 294 pp.

Boontongloan, C. (2000). Engineering properties of the evaporitic and clastic rocks ofMaha Sarakam formation, Sakon Nakhon evaporite basin. M.Sc. thesis, AsianInstitute of Technology, Thailand.

BSI, 1990. BS 1377: British Standard Methods of Test for Soils for Civil EngineeringPurposes; Part 2: Classification tests.

Casa, E., Lowenstein, T.K., 1989. Diagenesis of saline pan halite: comparison ofpetrographic features of modern, Quaternary and Permian Halites. J. Sediment.Petrol. 59 (5), 724–739.

Charrach, J., 1986. Genesis of Halite Islands in the Southern Basin of the Dead Sea. Geol.Soc. of Israel Annual Meeting, Ma'alot, Abstract, pp. 26–27.

Charrach, J., 2006a. A re-evaluation of theMount Sdom evaporite sequence. Geol. Soc. ofIsrael Annual Meeting, Beit Shean, Abstract, pp. 19–20.

Charrach, J., 2006b. Holocene stratigraphy of the Southern Dead Sea Basin—paleoclimatic and tectonic implications. Geol. Soc. of Israel Annual Meeting, BeitShean, Abstract, p. 21.

Elvery, R.H., 1972. Ultrasonic Testing of Concrete—The Use of The PUNDIT. CNS Farnell/Fulton Group, Borehamwood, UK.

Enzel, Y., Bookman (Ken Tor), R., Sharon, D., Gvirtzman, H., Dayan, U., Liv, B., Stein, M.,2003. Late Holocene climates of the Near East deduced from Dead Sea levelvariations and modern regional winter rainfall. Quat. Res. 60, 263–273.

Frydman, S., 2000. The shear strength of Israeli soils. Isr. J. Earth-Sci. 49, 55–64.

244 S. Frydman et al. / Engineering Geology 101 (2008) 236–244

Fuenkajorn, K., Daemen, J.J.K., 1988. Boreholes Closure in Salt. Technical Report Preparedfor The U.S. Nuclear Regulatory Commission. Report No. NUREG/CR-5243 RW.University of Arizona.

Garfunkel, Z., Ben Avraham, Z., 2001. Basins along the Dead Sea Transform. In: Ziegler,P.A., Cavazza, W., Robertson, A.H.F., Crasquin-Soleau, S. (Eds.), Peri-Tethis Memoir6: Peri-Tethian Rift/Wrench Basins and Passive Margins. Men. Natn. Hist. nat.,vol. 186, pp. 607–627. Paris.

Ginsburg, A., Ben Avraham, Z., 1997. The deep structure of the north basin of the DeadSea, Israel. Geophys. Res. Lett. 24, 2063–2066.

Hansen, F.D., Mellegard, K.D., Senseny, P.E., 1984. Elasticity and strength of ten naturalrock salts. Proc. First Conference on Mechanical Behavior of Salt, Clausthal-Zellerfeld, Federal Republic of Germany: Trans Tech Publications, pp. 71–83.

Hardie, L.A., 1990. The role of rifting and hydrothermal CaCl2 brines in the origin ofpotash evaporites: an hypothesis. Am. J. Sci. 290, 43–106.

Hoek, E., Franklin, J.A., 1968. A simple triaxial cell for field and laboratory testing of rock.Trans. Inst. Min. Metall. 77, A22–A26.

Hofrichter, E., 1976. Zur frage der porositat und permeabilitat von saltzgesteiner. ErdolErdgas 93 (3), 77–80.

Klayvimut, K. (2003). Mechanical performance of underground excavation in rock saltformation for nuclear waste repository in Northeastern Thailand. MSc thesis,Suranaree University of Technology, Thailand.

Klinger, Y., Avouac, J.P., Bourles, D., Tisnerat, N., 2003. Alluvial deposition and lake-levelfluctuations forced by late quaternary climate change: the Dead Sea case example.Sediment. Geol. 162, 119–139.

Kulhawy, F.H., Mayne, P.W., 1990. Manual on estimating soil properties for foundationdesign. Cornell University Final Report EL6800, for the Electric Power ResearchInstitute.

Migowski, C., Agnon, A., Bookman, R., Negendank, J.F.W., Stein, M., 2004. Recurrencepattern of Holocene earthquakes along the Dead Sea transform revealed by varve-counting and radiocarbon dating of lacustrine sediments. Earth Planet. Sci. Lett.222, 301–314.

Noorany, I., 1984. Phase relations inmarine soils. J. Geotech. Eng., ASCE 110 (4), 539–542.Ratigan, J.L., Vogt, T.J., 1993. LPG storage at Mont Belvieu, Texas: a case history. SPE Adv.

Technol. Ser. 1 (1), 204–211.Raup, O., 1970. Brine mixing: an additional mechanism for formation of basin

evaporites. Bull. Am. Assoc. Pet. Geol. 54 (12), 2246–2259.Sonnenfeld, P., 1984. Brines and Evaporites. Academic Press Inc. 613 pp.Terzaghi, K., Peck, R.B., 1967. Soil Mechanics in Engineering Practice. JohnWiley & Sons,

Inc., New York.