Geo-characterisation and properties of natural soils by in situ tests

Transcription

1 Poceedings of the 16th Intenational Confeence on Soil Mechanics and Geotechnical Engineeing Milless Science Publishes/IOS Pess. Published with Oen Access unde the Ceative Commons BY-NC Licence by IOS Pess. doi: / Geo-chaacteisation and oeties of natual soils by in situ tests Geo-caactéisation et oiétés des sols natuels as essais in-situ Fenando Schnaid Fedeal Univesity of Rio Gande do Sul, Bazil ABSTRACT Site chaacteisation and in situ test inteetation have been evolving fom basic emiical ecommendations to a sohisticated aea demanding a thoough knowledge of mateial behaviou and numeical modelling. With the advent of moden testing techniques and moe igoous methods of analysis, site chaacteisation in natual soils is gaining momentum. This Reot esents a citical aaisal in the undestanding and assessment of the stess-stain-time and stength chaacteistics of natual soil conditions and exloes new inteetation methods caable of measuing soil oeties shaed by effects of micostuctue, stiffness non-lineaity, small and lage stain anisotoy, weatheing and destuctuation, consolidation chaacteistics and ate deendency. Inteetation methods in diffeent soil fomations such as clay, sand, silt and bonded geomateials ae exloed using diffeent testing techniques. Since the in situ behaviou of natual soils is comlex, a single geneal ecommendation is to coss-coelate measuements fom diffeent tests. When data ae combined thee is moe scoe fo ational inteetation and, fo this eason, emhasis has been laced on coelations with mechanical oeties that ae based on the combination of indeendent measuements. RÉSUMÉ La caactéisation des sols et l inteétation des essais in situ a ofondément évolué, assant de ecommandations simles et emiiques à des méthodes lus sohistiquées nécessitant une connaissance aofondie du comotement des matéiaux et des méthodes de modélisation numéique. Avec le déveloement de nouvelles techniques d essais et de méthodes d analyses de lus en lus igoueuses, la caactéisation des sols natuels end de lus en lus d amleu. Ce aot ésente une évaluation citique de note coméhension et des méthodes de détemination des elations containtes-defomations-tems et des caactéistiques de ésistance des sols natuels en lace. Il exloe de nouvelles méthodes d inteétation caable de mesue les oiétés du sol diectement liées a sa micostuctue, tels que le module d élasticité non linéaie, les etites et lages défomations en anisotoie, la déstuctuation, les caactéistiques de consolidation et la déendance a la vitesse de chagement. Les méthodes d inteétation dans des sols difféents tels que agile, sable, limon et autes geomatéiaux fomes a déôt sont analysées, à ati de difféentes tyes d essais. Du fait de la comlexité du comotement des sols en lace, une emièe et simle ecommandation est de coise les ésultats de difféents essais. Losque les ésultats sont combinés, il y a une lus gande ou une inteétation ationnelle, et ou cette aison, l attention a été lacée su les coélations ente oiété mécaniques basées su des combinaisons d essais indéendants. 1 INTRODUCTION Because the Eath is a comlex thee-dimensional object with numeous stuctues and hydo-geomohologies that ae difficult to imagine and to chaacteise, it is imotant to develo concets and techniques that will suot the identification and descition of its satial aangements in maniulative two-dimensional eesentations. Although this consists on a challenge to be aoached by a hybid of discilines, it has omted the develoment of the scientific field of Soil Mechanics and, in aticula, the aea of site chaacteisation. The basic objective of site chaacteisation is to acquie toogahical, hydo-geological, geotechnical and geoenvionmental infomation that is elevant to the equiements of a oject. The wok of geo-enginees has exanded enomously ove the ast decades and theefoe ofessionals must become awae and acknowledgeable in many aeas to fully undetake the issues coveed by the lanning and oganization of site chaacteisation ogammes. Seveal ublications ovide suveys of many of the technical details of the field and give a good oveview of the immediate uoses associated to site chaacteisation (e.g. Clayton et al, 1995; Rowe, 2001). Placing the focus on geotechnical engineeing, the immediate uoses associated to an investigation ogamme is to (a) detemine the geneal natue and sequence of the subsuface stata, (b) locate the wate table and goundwate conditions and (c) measue o assess secific oeties of the gound. Any oject o elated activity would theefoe equie a fundamental undestanding of its envionmental constains and, fo that eason, a ofessional should comehend the intelay of ocesses that leads to site chaacteisation. Figue 1 illustates the stages associated to chaacteisation in a flowchat that identifies mechanical testing as an inheent at of site investigation, a at that detemines the basic soil classification, suots the concetual model adoted in design and establishes the eesentative stength and stiffness aametes equied fo engineeing design calculations. Undestanding laboatoy and in situ tests and the constitutive elationshis that link mateial behaviou is theefoe consideed essential to otimise engineeing geotechnical design. Imotant inteelationshis exist between the field and laboatoy which, given the numeous existing techniques available, equie stategies fo chaacteising soils to be established. A geneal ecommendation is to examine the chaacteistics of the soil fom a maco to a mico esective. Suface wave methods ovide a satial 3-D o 2-D subsuface eesentation of lage aeas. Geohysical methods give a qualitative ictue of the site which does not substitute the need fo diect measuements attained by in situ tests. SPT, CPT, DMT and SBPM ae designed to eveal 1-D infomation of the gound that, unde some simlified assumtions, can be inteeted to assess aveage oeties of soil ofiles. Finally laboatoy test deals with a close examination of an elemental mateial oety. Since each testing technique esonds to diffeent hysical oeties and gives infomation of diffeent 3

2 natue, a successful, cost-effective site chaacteisation ogamme should consist of an aoiate combination of field tests (3,2 and 1-D eesentations) and laboatoy tests, so that the elevant infomation can be synthesized and undestood with confidence. The aoach of this eot is to give a geneal ictue of the field of geo-chaacteisation with emhasis on in situ testing soil mechanics, to discuss the theoetical backgound that suots inteetation of testing data, lus add examles of the use of exeimental techniques. An attemt is made to highlight new otential methods of inteetation of in situ tests, by emhasising what is taken to be tendencies in the field of site investigation. Rathe than tying to exhaustively discuss established methods of inteetation, I would like to focus on giving enginees a close examination of ecent and cuent develoments in the following: a) inteetation of in situ tests in clay, whee the needs to account fo stess histoy, soil stuctue, anisotoy, consolidation and viscous effects ae consideed; b) analysis of the stess-stain and stength behaviou of ganula soils, with emhasis on the effects of fabic, aging and cementation and thei imact on coelations based on both lage calibation chambe tests and centifuge tests; c) a citical eview on the behaviou of bonded geomateials and the develoment of inteetative models caable of extending the existing theoetical and emiical aoaches established on the basis of the exeience on standad clays and sands; d) evaluation of atial dainage effects on esults of in situ tests caied out in intemediate emeability silt soils. Figue 1. Site chaacteisation flowchat Each of these toics is a majo subject on itself. Only fundamental asects ae selected fo discussion based on the authos exeience and inteests and the need not to deviate fom fundamental Soil Mechanics. 2 BACKGROUND INFORMATION In this Reot I stive to oduce a boad view of technical infomation and theoetical backgound that ae necessay to eae ofessionals to oduce sounding engineeing judgement in evey stage of the geo-chaacteisation of natual soils. This comises undestanding of the ways the key exeimental and theoetical elements ae coelated in the domain of geotechnical engineeing eseach elating to site chaacteisation, acknowledging otential and limitations of diffeent tests and measuements and ecognising hyotheses and assumtions associated to mechanical models develoed to descibe the behaviou of natual soils. Since no mathematical theoy can comlete descibe the comlex bounday conditions of a field test, the ole of centifuge and lage laboatoy chambe testing is emhasised in ode to imove ou undestanding of the behaviou of ototyes unde aametic studies. We stat by ecalling that fundamental undestanding of soil behaviou is necessaily develoed on the basis of laboatoy in tests. The esults of tiaxial tests have evealed the decisive elements to incooate in many constitutive models of soil behaviou, in aticula on the develoment of Citical State Soil Mechanics and the family of Cam-Clay models (e.g. Schofield &Woth, 1968; Roscoe & Buland, 1968). The ecent innovations in laboatoy testing fo exeimentally detemining the stess-stain-stength and time deendent oeties of geomateials comises (a) gowth of a new geneation of devices such as hollow cylinde, esonant column and tosional shea aaatus, (b) extensive use of stess ath comute based systems and (c) develoment of new techniques fo moe accuate measuements of local stains and imosed loads (e.g. Shibuya et al, 1996; Stoke et al, 1995; Tatsuoka et al, 1997; Lo Pesti et al, 1999). Inceasing sohistication, accuacy and caability of laboatoy measuements has omted a bette undestanding of the behaviou of geomateials to assist in the solution of a vaiety of geotechnical oblems. Recent eseach has ovided the backgound necessay fo the assessment of asects of soil behaviou in moe comlex envionments, including the effects of micostuctue (fabic and bonding), small stain stiffness and stiffness non-lineaity, small and lage stain anisotoy, weatheing and destuctuation, atial satuation and viscosity. These imotant featues of natual (and man-made) gound behaviou ae now ecognised and ae addessed on the basis of a famewok that has been established fom a comehensive chaacteisation of laboatoy tests on econstituted soils and a numbe of well known natual clays and sands (e.g. Almeida & Maques, 2003; Hight et al, 2003; Dias-Rodiguez, 2003; Lo Pesti et al, 2003; Coo & Aiey, 2003; Jamiolkowski & Lo Pesti, 2003). The consideable body of knowledge accumulated fom laboatoy tests is hee e-viewed with the uose of (a) oviding a famewok fo descibing soil behaviou that suots the inteetation of in situ tests and (b) oducing a database against which esults of in situ tests ae calibated. An attemt is made to extend this existing backgound knowledge to the field of in situ testing, and a necessay ste in this diection is to develo a new geneation of inteetation methods and constitutive models that caitalizes on existing exeience. As ecently ointed out by Schnaid et al (2004), the challenge in the field of in situ tests is theefold: to evaluate the alicability of existing theoetical and emiical aoaches in ode to extend the exeience of standad clays and sands to othe geomateials, to develo inteetative methods that incooate new constitutive models wheneve equied, and to 4

3 gathe exeimental data that justifies the alicability of oosed inteetation methods to engineeing alications. A imaily ste in this diection is to identify the alicability and otential of existing techniques, a task that necessaily contains a citical aaisal on how esults can be comiled to obtain a gound model and aoiate geotechnical aametes. A vaiety of in situ tests is now available to meet the needs of geotechnical enginees (e.g. Mayne, 2001; Van Ime et al, 2001). Existing field techniques can be boadly divided into two main gous: (a) non-destuctive o semi-destuctive tests that ae caied out with minimal oveall distubance of soil stuctue and little modification of the initial mean effective stess duing the installation ocess. The non-destuctive gou comises seismic techniques, essuemete obes and late loading tests, a set of tools that is geneally suitable fo igoous inteetation of test data unde a numbe of simlified assumtions; (b) invasive, destuctive tests wee inheent distubance is imated by the enetation o installation of the obe into the gound. Invasive-destuctive techniques comise SPT, CPT and dilatomete. These enetation tools ae obust, easy to use and elatively inexensive, but the mechanism associated to the installation ocess is often faily comlex and theefoe a igoous inteetation is only ossible in few cases. Fo examle, CHT and DHT ae defined in geohysics as intusive methods since they ae geneally efomed within boeholes. Howeve shea waves oagate in a soil mass that has not been distubed by installation, which in geomechanics suggests a non-destuctive tye of test in attemting to distinguish fom invasive enetation techniques in which inteetation is faily sensitive to the shea zone ceated aound a enetating obe. Since the in situ behaviou of geomateials is comlex, cuent eseach effots ae lacing emhasis on coelations with mechanical oeties that ae based on the combination of diffeent sensos in a single test device, usually combining a non-destuctive to an invasive technique such as the seismic cone and cone essuemete. A summay of the key infomation egading the commonly used in situ tests is given in Table 1, in which measuements of each testing technique ae descibed and common alications ae identified. The main chaacteistics of the field test techniques descibed in Table 1 ae summaised in the following sections. Table 1: Commecial in situ testing techniques (modified fom Schnaid et al, 2004) Categoy Test Designation Measuements Common Alications Nondestuctive o semidestuctive tests Geohysical tests: Seismic efaction Suface waves Cosshole test Downhole test Pessuemete test Pe-boed Self-boing SR SASW CHT DHT PMT SBPM P-waves fom suface R-waves fom suface P & S waves in boeholes P & S waves with deth G, ( x ) cuve G, ( x ) cuve Gound chaacteisation Small stain stiffness, G o Shea modulus, G Shea stength In situ hoizontal stess Consolidation oeties Plate loading test PLT (L x ) cuve Stiffness and stength Invasive enetation tests Cone enetation test Electic Piezocone Standad Penetation Test (enegy contol) CPT CPTU SPT q c, f s q c, f s, u Penetation (N value) Soil ofiling Shea stength Relative density Consolidation oeties Soil ofiling Intenal fiction angle, Flat dilatomete test DMT o, 1 Stiffness Shea stength Vane shea test VST Toque Undained shea stength, s u Combined tests (Invasive + Nondestuctive) Cone essuemete CPMT q c, f s, (+u), G, ( x ) Seismic cone SCPT q c, f s, V, V s, (+u) Resistivity cone RCPT q c, f s, Soil ofiling Shea modulus, G Shea stength Consolidation oeties Soil ofiling Shea stength Small stain stiffness, G o Consolidation oeties Soil ofiling Shea stength Soil oosity Seismic dilatomete o, 1, V, V s Stiffness (G and G o ) Shea stength 5

4 2.1 Seismic tests A geohysical suvey is egaded as a oweful technique fo subsuface exloation. Tests ae geneally non-destuctive in natue and can be efomed fom the gound suface. Desite due ecognition its isks and limitations, thee has been a steady incease in the eceived value of geohysics in eesenting comlicated subsuface conditions involving lage satial vaiability and statified soils. In addition, coss and downhole methods have been extensively used in geotechnical engineeing, including the adatation of sensos in the seismic cone. The theoetical bases uon which seismic and othe geohysical measuements ae found ae not within the scoe of this Reot. Fo that uose thee is a numbe of efeence textbooks that extensively coves this subject aea such as Richad et al (1970), Shama (1997) and Santamaina et al (2001). Fo us it is imotant to ecall that geohysical methods ely on a significant contast in hysical oeties of mateials unde investigation. Intinsic oeties such as density, esistivity o electical conductivity, magnetic suscetibility and velocity of shock waves of the subsuface mateials should be consideed when evaluating the suitability of a given technique. Fequently used geohysical techniques ae seismic efaction, high esolution suface wave eflection, vibation, down-hole and coss-hole, electical esistivity, magnetic and gavity tests. (e.g. Stokoe & Santamaina, 2000; Stokoe et al, 2004; Becke, 2001). The imay alications in the use of geohysical methods in geotechnical engineeing ae (Becke, 2001): to ma statigahy, detemine thickness of stata, deth of bedock and define majo anomalies such as channels and cavities; to locate deosits of aggegates and othe constuction mateials; and to detemine engineeing oeties of stata and thei satial vaiation. Geo-envionmental ojects comlement the list of alications. It is always necessay to bea in mind that geohysical techniques ae intended to sulement gound investigation methods. To enhance its consistency, a site investigation camaign should always encomass a combination of geohysical suveys with a mesh of boeholes and/o enetation tests. In this eot attention is given to measuement of shea wave velocities fom which it is ossible to obtain the small-stain stiffness of the soil at induced stain levels of less than %: G 2 o V s (1) whee G o is the shea modulus, the mass density and V s the velocity of shea waves fo a linea, elastic, isotoic medium. The CHT and DHT enable the velocity of hoizontally oagating, vetically olaized (S hv ), vetically oagating, hoizontally olaized (S vh ) and hoizontally oagating, hoizontally olaized (S hh ) shea waves to be measued. 2.2 Piezocone enetation test (CPTU) The CPT, with the ossible inclusion of oe wate essue, shea wave velocity and esistivity measuements is now ecognized woldwide as an established, outine and costeffective tool fo site chaacteisation and statigahic ofiling, and a means by which the mechanical oeties of the subsuface stata may by assessed. CPTs wee aticulaly oula in sands and in maine and lacustine sediments in costal egions, but ae now also commonly used in eats, silt, esidual soils, a vaiety of had mateials (chalk, cemented sands) and eclaimed land fomed by hydaulic fills, dedging and mine tailings. Fo a geneal eview on the subject the eade is encouaged to efe to Lunne et al. (1997) CPT in Geotechnical Pactice, and the Poceedings of the Symosia on Penetation Testing (1981, 1988, 1995, 1998, 2004). Figue 2 shows a diagam of a tyical cone. Routine enetometes have emloyed eithe one midface element fo oe wate essue measuement (designated as u 1 ) o an element ositioned just behind the cone ti (shoulde, u 2 ). The ability to measue oe essue duing enetation geatly enhances the ofiling caability of the CPTU, allowing thin lenses of mateial to be detected. Geotechnical site chaacteisation can be futhe imoved by indeendent seismic measuements, adding the downhole shea wave velocity (V s ) to the measued ti cone esistance (q t ), sleeve fiction (f s ) and oe wate essue (u). The combination of diffeent measuements into a single sounding ovides a aticula oweful means of assessing the chaacteistics of natual mateials. The seismic CPT is becoming a outine site investigation tool in many counties giving the facility of adding acceleomete and/o geohones to measue comession (P) and shea (S) wave velocities (e.g. Camanella et al, 1986). The additional cost and time equied fo a seismic measuement is modest and the inut ovided by the small stain shea modulus is essential fo soil chaacteisation and ediction of gound-suface settlements unde dynamic loading. Figue 2. Design featues of a iezocone 2.3 Pessuemete Pessuemetes ae cylindical devices designed to aly unifom essue to the wall of a boehole by means of a flexible membane. Both essue and defomation at the cavity wall ae ecoded and inteetation is ovided by cavity exansion theoies unde the assumtion that the obe is exanded in a linea, isotoic, elastic, efectly lastic soil. Unde this assumtion the soil suounding the obe is subjected to ue shea only. Acknowledging that the geatest otential of the essuemete lies in the measuement of modulus, it is a common actice to cay out a few unloadingeloading cycles duing the test. If the soil is efectly elastic in unloading, then the unloading-eloading cycle will have a gadient of 2G u, whee G u is the unload-eload shea modulus. Numeous aes have been ublished on this theme and thee ae imotant textbooks such as Baguelin et al (1978), Mai & Wood (1987), Biaud (1992), Clake (1995) and Yu (2000). Pessuemetes ae geneally classified in thee gous accoding to the method of installation into the gound. Peboed essuemete, self-boing essuemete and ush-in essuemete ae the thee boad categoies. The Menad essuemete is the most well known examle of a e-boed obe in which the device is loweed into a e-fomed hole. In a self-boing obe the device boes its own way into the gound with minimal distubance (see Figue 3), wheeas in a ush-in device the essuemete is ushed into the gound attached to a 6

5 cone ti. The method of inteetation should take account of the installation ocess. Theoetical inteetation methods develoed fo essuemetes involve axially symmetic exansion and contaction of an infinitely long cylindical cavity. Unde this fundamental assumtion the cavity-exansion/contaction cuve can be analytically modelled to obtain soil oeties. The symmety of the well defined bounday conditions of a essuemete is the main advantage of this technique ove othe in situ tests. In ode to investigate the aoiateness of this basic assumtion, seveal numeical studies have been caied out to investigate the effects of the finite length of a essuemete in clay (Yeung & Cate, 1990; Houlsby & Cate, 1993; Chales et al, 1999). These studies geneally suggest that ignoing length to diamete effects will significantly oveestimate the undained shea stength. In sand, the ossible effects of geomety ae quantified by both numeical methods (Yu, 1993) and calibation chambe tests (Schnaid & Houlsby, 1992; Ajalloeian & Yu, 1993). Results suggest that the finite length of a essuemete yields a stiffe loading esonse which imacts the edicted values of fiction angle. This effect is less onounced on the unloading otion of the test. Coneessuemete tests caied out in a laboatoy chambe study ae shown in Figue 4 to illustate the influence of the length to diamete atio on the measued essuemete cuve. effects of installation have to be accounted fo and lage stain analysis is equied. This technique is eceived as having a geat otential that has not yet been fully ecognized in actice. Analysis of the test in clay is achieved by a simle geometic constuction of the cuve to detemine the undained shea stength, the shea modulus and the in situ hoizontal stess (Houlsby and Withes, 1988). Analysis in sand is, howeve, significantly moe comlex and inteetation is lagely based on calibation chambe tests (Schnaid & Houlsby, 1992; Nutt & Houlsby, 1992). Reseach in the ast 10 yeas has ovided basic inteetation ocedues to allow the use of test esults to detemine the engineeing oeties of soils, including assessment to shea stength, elative density, state aamete, fiction angle and in situ stess state. Figue 3. Self-boing essuemete Figue 5. Cone-essuemete (afte Withes et al, 1989) Figue 4. Laboatoy esults of finite essuemete length effects (Schnaid & Houlsby, 1992). 2.4 Cone essuemete The CPMT is an in situ testing device that combines the 15 cm 2 cone with essuemete module mounted behind the cone ti, as illustated in Figue 5 (afte Withes et al, 1989). Since the essuemete test is not caied out in undistubed gound, the 2.5 Dilatomete The flat dilatomete test (DMT) was develoed in Italy (Machetti, 1980) and has become a outine site investigation tool in moe than 40 counties. A geneal oveview of the dilatomete, guidelines fo oe execution and basic inteetation methods ae given by Machetti et al (2001) in a eot issued unde the ausices of ISSMGE Technical Committee TC 16. The dilatomete consists in a stainless steel blade having a flat, cicula steel membane mounted flush on one side (Figue 6). The blade is diven into the soil using ushing igs nomally adoted fo the CPT. Afte enetation, the membane is inflated and a sequence of essue eadings ae made at escibed dislacements, coesonding to the essue at which the membane stats to exand ( lift-off ) and the essue equied to move the cente of the membane by 1.1mm against the soil. Inteetation methods ae essentially based on coelations obtained by calibating DMT essue eadings against high quality aametes (e.g. Luttenegge, 1988; Lunne et al, 1989; Machetti, 1997). These coelations ae essentially emiical 7

6 based and ae suoted by a limited numbe of numeical studies (e.g. Baligh & Scott, 1975; Finno, 1993; Yu et al, 1993; Smith & Houlsby, 1995; Yu, 2004). * The maximum otential enegy, PE, deliveed to the soil should theefoe be exessed as a function of the nominal otential enegy E*, and an additional enegy elated to the samle enetation and the weight of both hamme and ods. * PE E M h M ) g (3) ( whee: M h = hamme weight; M = od weigh; g = gavity acceleation; = Samle enetation unde one blow; E* = nominal otential enegy = 0.76m 63.5kg 9.801m/s 2 = 474 J Figue 6. Schematic eesentation of a dilatomete 2.6 Standad Penetation Test (SPT) The SPT is the most widely used in situ testing technique, imaily because of its simlicity, obustness and its ability to coe with difficult gound conditions in addition to oviding distubed soil samles. A comehensive eview of ocedues and alications of the SPT is given by Decout et al. (1988) and Clayton (1995). Thee is a ange of tyes of SPT aaatus in use aound the wold (fo examle those emloying manual and automatic ti hammes) and, consequently, vaiable enegy losses cannot be avoided. Vaiability due to unknown values of enegy deliveed to the SPT od system can now be oely accounted fo by standadizing the measued N value to a efeence value of 60% of the otential enegy of the SPT hamme (N 60 ), as suggested by Skemton (1986). In many counties, howeve, this ecommendation has not been incooated into engineeing actice. Moeove, even an SPT N value nomalized to a given efeence enegy is not standad because of the esently contentious issue of the influence of the length of the od sting. The enegy tansfeed to the comosition of SPT ods was ecently investigated by the autho (Odebecht, 2003; Odebecht et al, 2004a; 2004b; Schnaid et al, 2004). This study has omted a numbe of ecommendations outlined to inteet the test in a moe ational way on the basis of wave oagation theoy. Recommendations ae summaized as follows: a) the enegy tansfeed to the od and to the samle due a hamme imact should be obtained though integation of equation 2, calculated by the F-V method, and known as the Enthu enegy: 0 E F( t) V ( t) dt (2) with an ue limit of integation equal to infinity (actical is 1/10s but may equie longe time intevals of integation (1/5s) in soft soils o long comosition of ods). b) the samle enegy can be conveniently exessed as a function of nominal otential enegy E*, samle final enetation and weight of both hamme and ods. Influence of od length oduce two oosite effects: wave enegy losses incease with inceasing od length and in a long comosition of ods the gain in otential enegy fom od weight is significant and may atially comensate measued enegy losses. c) efficiency is accounted fo by thee coefficients 1, 2 and 3 that should be obtained fom calibation. The hamme efficiency 1 is obtained fom measuements at the to of the od stem. Efficiency facto 2 can be assumed as unit. The enegy efficiency 3 is negatively coelated to the length of ods. The nominal otential enegy E*= 474 J (ASTM, 1986) eesents a at of the hamme otential enegy to be tansmitted to the soil. An additional hamme otential enegy is given by M h g. The othe at is tansmitted by the od otential enegy M g which cannot be disegaded fo tests caied out at geat deths in soft soils, i.e. conditions in which and M ae significant. Fo convenience, equation 3 can be witten in two ats whee the fist eesents the hamme otential enegy (nominal + additional) and the second the od otential enegy: PE * (0.76 ) M g M g (4) h Equation 4 deals with an ideal condition, whee enegy losses duing the enegy tansfeence ocess ae not taken into account. Howeve, it is well known in engineeing actice that these losses occu, they should not be disegaded and should be consideed by the efficiency coefficients eviously intoduced. Equation 5 becomes: PEh 3[ 1(0.76 ) M h g 2 M whee 1 g] (5) F ( t) V ( t) 0 hamme efficiency (0.76 ) M enegy efficiency It follows fom the foegoing that nomalization to a efeence value of N 60 is no longe sufficient to fully exlain the mechanism of enegy tansfe to the soil and it is oosed that the system enegy should be calculated using equation (5). Futhemoe, it inteesting to ecall that the maximum otential enegy can be tansfomed into wok by the non- consevative foces (W nc ) acting on the samle duing enetation, and since the wok is ootional to the measued emanent enetation of the samle, it is ossible to calculate the dynamic foce tansmitted to the soil duing diving: PE h W nc F d o (6) F d PE h / The dynamic foce F d can be consideed as a fundamental measuement fo the ediction of soil aametes fom SPT esults. dt h g 8

7 3 INTERPRETATION METHODS The fonties of a scientific field ae defined as much by the tools available fo obsevation as by the beakthoughs in theoetical develoments. Seveal imotant concets have been intoduced o adated to Soil Mechanics in ast decades binging new esectives fo the chaacteisation of geomateials. Inteetation of in situ tests became a highly secialised subject that evolves using fomal analytical o numeical solutions in addition to eviously acceted semiemiical aoaches. Guidance is hee ovided on how to link theoy to actice by establishing fou classes in which inteetation of in situ tests can be boadly goued: Class I: Analytical solutions caable of idealising the field test into an equivalent ealistic fom. Ability to extact a lausible solution deends on the accuacy of constitutive models in eesenting soil behaviou and on the coectness of the imosed bounday conditions. This is achieved in a limited numbe of cases. Exact close fom solutions ae only alicable in tests that contain sufficient geometic symmeties to educe the oblem to a simle fom, such as the exansion of sheical and infinite long cylindical cavities in a semi-infinite elasticlastic continuum. Class II: Fo cases in which close fom solutions cannot be obtained, adequacy of numeical solutions lagely deend uon the constitutive model adoted to eesent soil behaviou. Thee ae solutions that offe a vey close aoximation of the hysical mechanism of a test; these solutions ae egaded as igoous and ae defined as Class II, as fo examle the family of numeical analyses alied to descibe the enetation mechanism in high lastic clays of a cone o flow enetomete (e.g. Yu, 2004; Randolh, 2004). Class III: Aoximate analytical solutions develoed on the basis of simlified assumtions imosed to educe the constaints equied fo a moe igoous aoach. Fo enetation tools, the concets of beaing caacity (e.g. Dugunuglu & Mitchell, 1975; de Mello, 1971) and cavity exansion (e.g. Vesic, 1972; Salgado et al, 1997) fom the backgound fo this categoy of analysis. Since beaing caacity theoies caied out using limit analysis ae unable to account fo soil stiffness and volume change, they can just be egaded as aoximation to model enetation oblems. Calibation chambe tests and centifuge tests ae ecommended to validate Class III tye of coelations. Simle solutions such as Class III should be efeably achieved fom a combination of measuements fom indeendent tests. Since in the inteetation of in situ tests the numbe of contolling vaiables (soil aametes) well exceeds the numbe of measued vaiables, the combination of indeendent measuements educes the degee of uncetainty. The autho foesees inceasing use of inteetation methods unde this thid categoy with a gowing tend towads the combination of vaious sensos incooated in a single enetation obe. This eot exloes extensively the atio of the elastic stiffness to ultimate stength (G o /q c, G o /N 60 ), the atio of cone esistance and essuemete limit essue (q c /) and the association of stength and enegy measuements (N 60 and enegy). Class IV: Emiical analysis based on diect comaisons with stuctue efomance and coelations to laboatoy test esults. Comlexities in the inteetation of in situ tests in unusual geomateials and ooly defined gound conditions still omts the use of emiical aoaches in geotechnical engineeing actice (e.g. Schnaid et al, 2004). Both the limitations associated to the inteetation of in situ tests in diffeent geomateials and the difficulty of isolating the indeendent factos that contol the test mechanism have omted the develoment and use of laboatoy hysical modelling. Two techniques ae now ecognised as efeence: centifuge and lage calibation chambe tests. Well-lanned seies of laboatoy hysical modelling tests can be used to establish coelations fo field tests that, although emiical, ae set against contolled vaiables (such as density, stess histoy, hoizontal and vetical stesses) and eoduce some of the envionmental equiements fo calibation against field data. A calibation chambe consists of a lage cylindical homogeneous samle of known density. Chambe-walls can eithe suot flexible ubbe membanes fom which unifom stesses ae alied o can be igid in the lateal diection when imosing zeo lateal stain conditions (K o ) on the samle. Stess-contolled bounday conditions ae moe geneally adoted, in ode to investigate the indeendent effects of vetical and hoizontal stess on enetation tests (e.g. Pakin & Lunne, 1982; Schnaid and Houslby, 1992). Fo tests caied out in sand, whee a igoous theoetical analysis is difficult, this technique eesents the best way of establishing well defined calibations of in situ testing devices. Note that this is in contast with clays, fo which calibation chambe tests ae nealy unfeasible due to time constaints, and calibation of analytical ocedues at well-documented test sites is a moe sounding aoach. In sands laboatoy calibation chambe tests have ovided databases that suot emiical inteetation ocedues between ti cone esistance q c and elative density (D ) o fiction angle ( ). Tyical calibation ogammes involve the state of stess of the samle and eflect the combined effects of sand stiffness, dilatancy and minealogy. Thee has been enough exeience on enetation tests to demonstate that thee is essentially no coelation between ti cone esistance q c and the vetical stess fo tests caied out at diffeent combinations of vo and ho (fo K o values between 0.4 and 1 consideed to be the ossible limits fo sand). Howeve, thee is a sounding coelation between q c and the hoizontal stess which suggests the need to take into account the hoizontal stess, o at least the mean stess, in field coelations (e.g. Houlsby & Hitchman, 1988; Schnaid & Houlsby, 1992). Centifuge tests have been gaining inceasing oulaity in the ast 20 yeas as a hysical modelling technique in all tye of soils. The technique has been successfully alied to model classical oblems such as foundation and sloes, envionmental asects of contaminant tansot and seismic events. Results have been documented in confeences oganised unde the ausices of TC2. The fact that scale models can be eaed with escibed soil oety ofiles in clay, sand o intemediate soils, and shaken in the simulated gavity envionment unde contolled inut motion (e.g. Schofield& Steedman, 1988; Taylo, 1995) makes this aoach aticulaly attactive as a means to oduce contolled sets of in situ test data. The main disadvantage of laboatoy hysical modelling is that models ae of limited size and assessment of chambe size effects and boundaies conditions in the calibation of in situ tests is always necessay (Baldi et al, 1982; Pakin & Lunne, 1982; Been et al, 1987; Mayne & Kulhawy, 1991; Schnaid and Houslby, 1992). Exeimental and numeical data demonstate that fo all sand densities the chambe size can affect the esults, with the effect being moe sevee fo dense sands due to samle dilation (even fo chambe to obe diamete atios of 40). It is vital that such effects should be oely quantified if accuate calibations of in situ testing devices ae to be obtained. Fo lage laboatoy calibation chambes, a sounding altenative to substantially educe bounday effects is the use of simulato caable of contolling hoizontal stess duing enetation (Foay, 1991; Ghionna & Jamiolkowski, 1991; Huang & Hsu, 2004). Using a sevo-contolled mechanism connected to a seies of ings, that elace the single ubbe membane, Huang & Hsu (2004) allow the lateal stess to vay 9

8 as the cone is being ushed into the samle. The incease in stess is calculated fom the adial stains measued at the cavity wall (Foay, 1991) and, by doing so, the authos simulate a semi-infinite half-sace at the soil-membane inteface. Figue 7 shows CPT tests in dense samles (D=84%) fo stess state that coesonds to vo =56kPa and ho = 22 kpa, in which faily simila q c vesus deth ofiles have been obtained fo chambe to obe diamete atios D/B of 22 and 44 indicating the eduction in size effects. As fo the centifuge, a diect imlication is the ecognition that a vaiation in hoizontal stesses at the boundaies of the stongbox needs to be accounted fo, i.e. the stain field dislaced by diving a cone into the samle may eventually each the igid bounday of the box and will esult in an incease in hoizontal stess with esect to the initial ho value. Load cells embedded in the centifuge secimen fo monitoing ossible vaiations on h duing cone enetation ae the only eliable way of assessing bounday effects. An examle is given in Figue 8, in which the vaiation with time of the measued hoizontal stess at fou given deths (0.06m, 0.12m, 0.18m and 0.24m) fo fou CPT tests caied out at 30g, 20g and 10g is esented. The numbe inside the squae indicates the distance in centimetes between the cone enetomete and the load cell whee the hoizontal stess has been ecoded. Cone enetation oduced an incease in hoizontal stesses h to a distance u to 80 cm. This distance coesonds to 66 times the diamete of the obe and is significantly highe than those ecognised by a numbe of authos, usually esticted to the ange of about 10 to 20 times the diamete of the obe (e.g. Gui et al, 1998; Bolton et al 1999). As exected, the magnitude of h educes with inceasing distance between the essue cell and the location of the test. In addition, the stongbox is ectangula and unlike the axi-symmetic conditions imosed in a cylindical laboatoy calibation chambe, the benefits of symmety aound the cone ae no longe valid. Any coection due to inceasing bode stesses is theefoe emiical in natue and should be viewed with caution. Use of cicula containes is stongly ecommended when continuous in-flight stength ofile of the secimens is coelated to soil oeties. A fist test can be conducted at the cente of the secimen and subsequently the enetomete can be moved to the quate oints of the containe. This bief eview highlights the fact that geo-chaacteisation and oeties of natual soils can be made using diffeent testing techniques and can ely on vaious methods of inteetation (Class I, II, II and IV). These aoaches ae extensively coveed in this Reot and ae alied to geomateials such as clay, sand, intemediate emeability silt and bonded soils. Deth, mm q c, MPa D/B 1600 Figue 7. Reduction in chambe size effects (afte Huang & Hsu, 2004). B D D = 84% Hoizontal stess h (kpa) test test 45 z=0.24 m z=0.18 m z=0.12 m z=0.06 m 0 30:00 35:00 40:00 45:00 50:00 55:00 60:00 65:00 Time (min/s) Figue 8. Centifuge bounday effects (modified fom Gaudin, Schnaid & Ganie, 2005). 4 CLAY PROPERTIES test Backgound eseach Assessment to the oeties of clay is on itself an extensive subject that is much beyond the scoe of this eot. Inevitably a shot discussion cannot meet the comlexity of the subject and is bound to suffe fom being too selective. A comehensive oveview of the toic comising the inteetation of both laboatoy and in situ tests in clay has been given by Ladd et al (1997); Jamiolkowski et al (1985); Buland (1990), Leoueil & Hight (2003), Yu (2004), among othes. This Reot aims only at detemining the factos which affect the inteetation of in situ tests, and fo that uose a famewok fo descibing the behaviou of clays is outlined. Imotant contibutions ove the last 15 yeas have attemted to integate all geomateials in a consistent and unified famewok by demonstating that soil stuctue is a common featue in a wide ange of soils, fom natual deosits to manmade eath-fills. To distinguish featues of behaviou emeging fom stuctue fom those elated to changes in state, a widesead aoach has been to comae the esonse of the natual soil to that of the coesonding econstituted mateial (e.g. Buland, 1990). Whilst in econstituted soils the shea behaviou is contolled solely by a combination of deviato stess, mean effective stess and secific volume (and as a consequence the shea stiffness at any stain is exessed as a function of thei cuent state), natual clay exhibit a stuctual behaviou that does not confom with the famewok develoed fo econstituted mateials (Buland, 1990; Leoueil & Vaughan, 1990). Tyical behaviou of stuctued clay in isotoic comession is illustated in Figue 9 fo tests caied out in Paadai clay comessed to stesses well beyond yield stess (Cotecchia & Chandle, 1997). Because of micostuctue, natual clay samles can each a domain of the (e-log v ) sace that is not emitted fo the econstituted clay leading to a highe econsolidation essue. Beyond yield stess, lastic stain incements become substantially lage as a esult of stuctue degadation. As fo stess-stain cuves and stess aths fom shea comession and extension tests, the chaacteistics to note ae: a) due to stuctue and oveconsolidation, natual clays show a stong esistance in shea until a eak is eached at elatively small stains. Lage stains develoed beyond eak ae associated to stain softening and stain softening effects educe with inceasing OCR. b) stiffness inceases with soil stuctue and educes with inceasing OCR and shea stain amlitudes. Fom yield locus of natual clays detemined fom laboatoy tests, it is ecognized that multile kinematic yield sufaces have to be consideed since soil does not behave as a simle elasticlastic mateial. A simlified scheme oosed by Jadine test

9 (1992) is illustated in Figue 10 and is adoted as efeence in this Reot. Soils dislay non-linea stess stain behaviou that can be boadly chaacteised by the linea theshold stain, Y1 (oint A in Fig. 10), the stain making the limit to ecoveable behaviou Y2 (oint B) and the stain denoting the onset of lage scale yielding, Y3 (oint C) (e.g. Jadine, 1985; Tatsuoka et al., 1997). At vey small stains, within the limit state cuve defined by Y1, soils ae believed to behave as elastic mateials eesented by the initial elastic stiffness G o. The magnitude of G o is measued in the laboatoy using bende elements o esonant column tests (e.g. Jadine et al, 1984; Tatsuoka et al, 1995; Ishihaa, 1996; Clayton & Heymann, 2001; Lo Pesti et al, 2001) and in the field by seismic techniques (Shama, 1997; Santamaina et al, 2001). Comehensive eviews of this toic ae eoted both at the Intenational Confeences of Pe-Failue Defomation Behaviou of Geomateials in 1995, 1997, 1999 and 2003 and by numbe of individual aes (e.g. Hadin, 1978; Stokoe et al, 1994; Jamiolkowski et al, 1995; Lo Pesti et al, 1999; Tatsuoka et al, 1997; Ramello & Viggiani, 2001). the anisotoy of fabic develoed duing deosition and one-dimensional consolidation of natual soft clay deosits. With the incease in isotoic consolidation, the LSC exands and changes shae, the stuctue gadually beaks and becomes symmetical with esect to hydostatic axis. These featues ae shown in Figue 11, in which the LSC cuve of Saaui clay is nomalised with esect to the econsolidation essue (Almeida & Maques, 2003). 1.3 Reconstituted clay Void atio e Peconsolidation P Yield Natual clay v (kpa) Figue 9. One-dimensional comession behaviou of the natual and econstituted Paadai Clay (Cotecchia & Chandle, 1997). Figue 10. Scheme of multile yield sufaces and soil esonse (Jadine et al, 1991). c) fo a given clay ofile, the limit state cuve (LSC) obtained at diffeent deths have simila shaes and can theefoe be nomalised by the econsolidation essue (Shanse method, afte Ladd et al, 1977; Gaham et al, 1983). The geneal shae of the LSC cuves of natual clays is known to deend mainly on the fiction angle (Dias Rodiguez et al, 1992) and is aoximately cented on the K o (NC) line in a ( a - )/2 vesus ( a + )/2 diagam, which is attibuted to soil anisotoy (Tatsuoka & Shibuya, 1991; Belloti et al., 1996; Hight et al., 1997; Tatsuoka et al, 1997). This eflects Figue 11. Limit State Line fo Saaui Clay (Almeida & Maques, 2003). d) in addition, exeimental obsevations in econstituted soft clay samles show that stess anisotoy should also be taken into account and in this case the effects of anisotoy may not be eased even afte a comlete destuctuation of the soil (e.g. Tatsuoka et al, 1997; Koskien et al, 2002). e) at lage stains the soil tends towads an ultimate condition at which thee is no change in void atio and effective stess. This ultimate state is defined as citical state fom whee the basic elastic-lastic concets that goven the esonse of clay have been integated in a single famewok by Citical State Soil Mechanics, given ise to comlete Cam-Clay models (e.g. Schofield & Woth, 1968; Roscoe & Buland, 1968). The model aametes ae the shea modulus, the sloe of the citical state line in a q: diagam M, the sloe of the vigin consolidation and of the swelling line in the (e ln ) diagam and the citical void atio e c. The M aamete is not the same in comession and in extension, the ultimate state in tiaxial comession exceeds that in tiaxial extension. The yield function illustated in Figue 12a can be conveniently exessed in tems of mean effective stess and deviato stess q (Roscoe & Buland, 1968): m 2 M 2 2 M 0 (7) whee m define the size of the yield suface. The model conceived on the basis of tests efomed on econstituted and isotoically consolidated clays has been systematically extended to othe geomateials, moe geneal stess states and consolidation histoies in attemting to eflect some of the imotant featues of soil behaviou eviously descibed. Gens & Nova (1993) intoduced the concets 11

10 necessay to incooate bonding and destuctuation effects within elastic-lastic constitutive models (see Figue 12b). In addition to the yield suface of a natual stuctued geomateial, an intinsic yield suface is adoted to eesent the size of the yield suface of an equivalent unbonded mateial that is assumed to have the same shae as the one of the natual soil. Concets intoduced to descibe soil anisotoy ae eesented in Figue 12c, following the wok by Whittle & Kavvadas (1994), Pestana & Whittle (1999), Whelle et al (2003), among othes. Whelle et al (2003) has ecently demonstated that fo simlified axi-symmetic conditions of a tiaxial test on a cossanisotoic samle, with the hoizontal lane in the tiaxial samle coinciding with the lane of isotoy of the samle, the yield cuve can be exessed as: ( q ) ( M )( ) 0 (8) whee defines the oientation of the yield cuve. The model incooates two hadening laws: the fist to descibe changes of size of the yield suface and the second to descibe the change in oientation of the yield suface with lastic staining so that when =0 soil behaviou becomes isotoic. Finally thee is a ecognition that stength and stiffness incease with inceasing sheaing ate (e.g. Shibuya et al, 1997; Stoke et al, 1995; Tatsuoka et al, 1995, 1997). Since the loading ate of most common geotechnical in situ tests (SPT, CPT, vane) is significantly highe than that of constuction activities subjected to monotonic loads, tyically u to 6 to 8 odes of magnitude, ate effects ae of significance when deiving oeties fom test data. Investigation fom laboatoy tests has demonstated that natual clays will exhibit highe econsolidation essue, undained shea stength and stiffness than the same soil in the same state, but unstuctued. The mechanical esonse eflects stiffness non-lineaity, small and lage stains anisotoy and stain ate effects. This Reot discusses the inteetation of in situ tests in the light of the chaacteistic behaviou of clay, intoduces the assumtions that ae made as at of the inteetation of test data and highlights the consequences of these assumtions on the edicted soil oeties. 4.2 Stiffness Soil stiffness deends uon inteactions of stuctue (bonding, fabic, degee of cementation), stain level (and effects of destuctuation), stess histoy and stess ath, time deendent effects (aging and cee) and tye of loading (monotonic o dynamic). Desite these comlex inteactions, the chaacteistic esonse of clay with esect to small stain stiffness, small stain anisotoy and stiffness non-lineaity can be diectly assessed fom in situ tests. Hadin & Black (1968) and Hadin (1978) identified vaious factos govening the G value and oosed a geneal exession: G = f ( v, e o, OCR, S, C, K, T) (9) whee v is the effective ovebuden stess, e o initial void atio, OCR oveconsolidation atio, S degee of satuation, C gain chaacteistics, K soil stuctue and T temeatue. Laboatoy esonant column tests caied out to chaacteise G o have shown that in a econstituted clay stiffness is a function of mean stess and that the effect of OCR is atially embedded into the effect of void atio. An emiical coelation to descibe G o can then be witten as: G o v h m n SF( e)( ) (10) (12 n) a whee S and n ae exeimental constants and F(e) is the void atio function. F(e) can take the fom of (1/e), (1+e) o (2.17- e) 2 /(1+e), leading to the following eesentative coelations: G 1 ( e ) 0.5 k o 625 ( o ) OCR 2 k=pi 0.72/50 <0.5 Hadin (1978) (11) G o 480( e) 1.43 ( ) v 0.22 ( ) h 0.22 ( ) 12(0.22) a Jamiolkowski et al (1995) (12) whee o = mean effective stess, a = atmosheic essue and PI= lasticity index. Coefficients in the above equations eesent aveage ecommended values and units ae in kpa. The need to exess G o as a function of the mean stess is ecognized, the effects of vetical and hoizontal stesses being investigated as seaate vaiables by Jamiolkowski et al (1995). Since ho is not usually known accuately, Shybuya et al (1997) consideed moe acticable to exess G o as a function of vo with the due ecognition that its validity is esticted to nomally consolidated clays. G o (1 e ) ( ) ( kpa) (13) o Fo eliminay ojects, vaious authos oosed to estimate G o diectly fom ti cone esistance q c o q t (e.g Rix & Stoke, 1992; Mayne & Rix, 1993; Tanaka et al, 1994; Simonini & Cola, 2000; Powell & Butche, 2004; Watabe et al, 2004). Mayne & Rix (1993) suggested a coelation that exlicitly consides the deendency of G o uon void atio: G o v q e ( kpa) (14) c o Tanaka et al (1994) develoed the following elationshi between the small stiffness measued fom the seismic cone and net cone esistance: G o 50( q )( kpa) (15) t vo Cae must always be taken when using equations 14 and 15, since stictly seaking a small stain value cannot be deived fom an ultimate stength measuement and theefoe these coelations should be seen just as an indication of stiffness that do not elace the need fo diect measuements of shea wave velocities. Infomation of the anisotoy of small stain stiffness (inheent and stess induced) fom seismic measuements is becoming moe eadily available and is a geneally ecommended technique. The G vh value fom a down-hole suvey whee the shea wave oagates in the vetical diection with the aticle motion in a hoizontal diection diffes fom the comaable G hh fom a coss-hole suvey. Leoueil & Hight (2003) esented a comilation of data on the anisotoy of small stain stiffness in natual and econstituted soft and stiff clays shown in Figue 13, exessed in tems of the atio G hv /G hh o G vh /G hh vesus consolidation stess atio v / h. In Bothkenna Clay, the fabic has endeed a stiffness atio anging fom 0.8 to 1.0 in an unconfined state. Jamiolkowski et al (1994) found G vh /G hh atios unde isotoic stess conditions of 0.7 and about 0.65 fo the natual Pisa and Panigalia clays esectively. In situ data fo the stiff and vey stiff London and Gault Clays ae vey simila, although unde isotoic stess conditions the Gault shows a lowe atio fo G vh /G hh. 12

11 Figue 12. Yield sufaces fo clay: (a) modified Cam-Clay, (b) stuctued mateial and (c) stuctue and anisotoy Figue 13. Small stain stiffness anisotoy in natual and econstituted clays (Leoueil & Hight, 2003). Since due to anisotoy G o is not indeendent on the diection of oagation and olaization of shea waves, coelations with enetation esistance should efeably be exessed as a function of G vh, G hv o G hh. Powell & Butche (2004) ecently suggested that thee is no single coelation between q t and G vh fo all clays as often advocated fom DHT. The authos have found a stong coelation between G hh and q t, which is atially exlained by the stong deendency of q t on the hoizontal stess. Figue 14 shows a set of esults in a logxlog lot that, given the scatte, should also be viewed as an indication of G o only. Wheeas the initial shea modulus G o is consideed to be a fundamental soil oety, a knowledge of the non-linea and inelastic stess-stain esonse of geomateials is now fully ecognized as being citical to the ediction of gound movements. The eve-inceasing oulaity of the shea wave velocity and hence G o measuement is encouaging, but, in its own ight, is of limited value because govening stain levels in the vicinity of geotechnical stuctues ae small to intemediate and fa geate than the vey small stain etaining to seismic measuements Reduction of shea stiffness with shea stain amlitude can be obtained fom a combination of seismic measuements and laboatoy o in situ tests. Fom the existing in situ tests (iezocone, essuemete, dilatome, late loading), the essuemete offes the most ealistic ossibility of assessing the non-linea esonse of clay. In a essuemete test soil stiffness can be assessed fom the comlete essue-exansion cuve o fom inteetation of unload-eload loos, which is aticula attactive since it ovides a measuing of the elastic shea stiffness in a ocedue that avoids many of the oblems of distubance that ae associated with laboatoy testing (e.g. Woth, 1984; Fahey, 1998; Whittle, 1999). The oblem is that the test does not, howeve, measue an element stiffness and inteetation of (non-linea) elemental stiffness chaacteistics elies on an aoiate numeical backanalysis method couled with a ealistic soil constitutive model (e.g. Fahey & Cate, 1993; Bolton & Whittle, 1999). Figue 14. Comaison between G hh with q t (Powell & Butche, 2004). The stains undegone by elements of soil at diffeent distances fom the essuemete ae invesely ootional to the squae of the adius in an undained test and theefoe a efeence shea stain has to be abitaily selected as eesentative of an unload-eload shea modulus, G u. This efeence value is often taken as the stain alied at the essuemete suface. Houlsby (1998) justifies the choice by demonstating that the measued G u is vey much dominated by the stiffness of the soil close to the essuemete. Mui Wood (1990) exloes the vaiation of G u with shea stain amlitude in clays. In a essuemete it is ossible to define a secant modulus G s ) / 2 and ( ho a tangent modulus G d t 1. This is simila to 2 d laboatoy tests in which it is also ossible to define a secant Gs / and a tangent modulus G t d / d, whee is the shea stess. Following these definitions it is staightfowad to demonstate that: dgs G G (16) G t t and G s s G s d dgs 2 2d (17) G t G s dgs 2 2d (18) Thus in clay, accoding to Mui Wood (1990) the tangent modulus measued fom the essuemete cuve is equal to the secant modulus fom a conventional laboatoy test. This enables the esults of essuemete tests to be oely 13

12 elated to those of othe tests by means of degadation models designed to exess the vaiation of G with. Simle to faily elaboated fomulations have been oosed to exess G vesus elationshis in the fom of a logaithm o a hyebola. Regadless the model adoted in the analysis, Fahey (1998) stongly ecommends that combining seismic G o measuements and SBP tests incooating multile unload-eload loos is cuently the only accuate method of obtaining non-linea stiffness aametes fom in situ tests. 4.3 Undained shea stength The undained shea stength of clays, in common with all soils, deends on the mode of failue, ate of sheaing, soil anisotoy and stess histoy. Fo a clay that confoms with CSSM, the atio S u / anges fom 0.25 to 0.30 deending on sheaing mode, whee is the econsolidation essue (Ladd et al, 1977; Jamiolkowski et al, 1985; Ladd, 1991). The efeence test adoted to measue S u, against which an in situ test is calibated, should be identified and S u should be defined as S u (TX-CIU), S u (TX-CKoU) o S u (FVT). Vane tests ae used imaily to detemine the undained stength of clays. S u values inteeted fom vane tests ae influenced by distubance, otation ate, delay between insetion and testing, as well as by the shea stess distibution aound the blades. Assumtions egading these effects ae well known and systematically eoted (e.g. Aas, 1965; Woth, 1984; Chandle, 1988). Although the conventional analysis assumes that the shea stength is isotoic, the vane offes the simlest ossible way of investigating the lage stain anisotoy of clays. The undained shea stength can be detemined fom the measued toque T by the following geneal exession: S uv 3 D H D 2T 1 S ( n 3) S uh uv (19) whee H/D is the asect atio, S uv / S uh is the stength atio descibing anisotoy and n is the owe law descibing the shea stength distibution on the to and bottom lanes (Woth, 1984). The vane toque on the vetical and hoizontal lanes can be seaated ovided a seies of vane test data with diffeent H/D is available, allowing assessment to soil anisotoy. In this case, the angula otation ate should be adjusted to avoid the influence of the eiheal velocity on measuements of undained shea stength (e.g. Biscontin & Pestana, 2001). This imlies that vanes of diffeent sizes and same asect atio should yield the same undained shea stength, wheeas diffeent asect atios would give assessment to soil anisotoy. An inceasing tend fo the use of iezocone in the chaacteisation of soft clay deosits is being obseved woldwide, gadually elacing the standadized actice of efoming vane tests. This tendency is justified by the ecognition of the CPTU as a sueio ofiling tool, as well as by its caability of deiving soil aametes fo geotechnical design. When caying out cone enetation tests unde undained conditions, cone ti esistance q c is elated to S u as follows: q N S (20) c c u 0 whee N c is a theoetical cone facto and o is the in situ total stess (eithe vetical o mean total stess). The theoetical solutions available fo detemining N c can be goued as (e.g. Yu and Mitchell, 1998; Yu, 2004): a) beaing caacity theoy (BCT) b) cavity exansion theoy (CET) c) stain ath methods (SPM) d) finite element methods (FEM) As ointed out by Yu (2004), while each theoy may be used alone fo cone enetation analysis, bette edictions of cone enetation mechanisms may be achieved if some of the methods ae used in combination (Teh and Houlsby, 1991; Abu-Fasakh et al, 2003). Table 2 summaises the most significant theoetical solutions fo N c. Solutions geneally assume isotoy of stength and stiffness and adially symmetic initial stesses. A single solution can account fo effects of anisotoy that esult fom both stuctue and otation of incial stesses aound an advancing cone (Su & Liao, 2002). The authos oosed cone factos exessed as a function of the stength anisotoy atio A =(S u )tc/(s u )te, with the undained shea stength S u tc and S u te measued fom CK o U comession and CK o U extension tiaxial tests esectively. The effect of anisotoy is not always significant, maximum diffeences in cone factos of u to 20% obtained fo nomally to slightly oveconsolidated clay with modeate to high stength anisotoy. Wheeas theoetical solutions have been contibuting in the undestanding of the fundamental mechanics of cone enetation, emiical coelations ae still widely used in actice to estimate S u fom cone esistance. The most widely used coelation is: q t N S (21) kt u v0 whee q t is the cone esistence (coected fo oe essue), N kt is an emiical cone facto and vo is the total in situ vetical stess. Values of N kt ange fom 10 to 20 and ae influenced by soil lasticity, oveconsolidation atio, samle distubance, stain ate and scale effects, as well as the efeence test fom which S u has been established (Aas et al, 1986; Mesi, 1989; 2001; Lunne et al, 1997). Stess histoy affects stength, stiffness, stain softening and stain to failue which in tun changes N kt. Comaisons between theoetical and emiical edicted N kt factos ae shown in Figue 15, fo I anging fom 50 to 500 and stength anisotoy atio of 0.5 to 0.9. In actice the choice of aveage I values is abitay and N kt values fom 12 to 15, often adoted in engineeing design oblems, ae geneally highe than theoetical edicted values fo igidity indexes between 50 and 200. Figue 15. Theoetical edictions of N KT factos fo clay Recent develoments using full flow obes, including T- ba, sheical ball and late enetometes (e.g. Randolh, 2004), have been used with the uose of ovecoming the uncetainties in detemining the undained shea stength in soft clay, due to the usual coections necessay with the CPTU and the comlicate mechanism aound the cone that 14

13 eflects on N kt. In a flow ba enetomete lowe and ue bound solutions can be found, as the oblem of consideing an intusive volume intoduced into the soil duing cone enetation is avoided. The beaing caacity facto can be conveniently exessed as the atio of an aveage enetation esistance q u to the shea stength S u, so that N=q u /S u. A theoetical beaing caacity solution of 10.5 fo the T-ba N facto was fist esented by Randolh & Houlsby (1984). Recently, Randolh (2004) intoduced a new solution fo this enetation mechanism which is shown in Figue 16 fo a lot that elates the N facto to the inteface fiction atio at the T-ba-soil inteface. In this figue the theoetical solution fo T-ba is comaed to cone and ball enetometes and is shown to be insensitive to I (ue and lowe bounds ae essentially the same). Long & Phoon (2004) summaized ecent ublished data using the T-ba and demonstated that on aveage the exeimental values ae close to the theoetical beaing caacity numbe of 10.5 fom Randolh & Houlsby (1984) and within the ange oosed by Randolh (2004). The above discussion is concentated on enetation oblems in nomally and slightly oveconsolidated clay whee modified Cam-Clay eoduces chaacteistic featues of behaviou with easonable accuacy. Heavily oveconsolidated clay cannot be successfully modeled and theefoe theoetical values of N kt oduce unealistic edictions of undained shea stength. Enginees should elay solely on emiical coelations which yield N kt values in the ange of 15 to 30 (e.g. Lunnet et al, 1997). Without ealistic theoetical solutions, and given the fact that enetation in OC geomateials may be difficult, the SPT still offes consideable aealing fo qualitative and quantitative evaluation of soil oeties in situ. A simle aoach based on limit equilibium and wave oagation analysis has been oosed by Schnaid et al (2004) and can be useful in assessing the undained shea stength in OC clays. Fist an idealization of the enetation mechanism is equied. In OC mateials the lug of soil which has been fomed within the hollow section of the samle is sufficiently stong not to be emoved by enetation foces, and consequently beaing caacity has to be edicted on the basis of both shaft and base esistance. Fo a uely cohesive soil, beaing caacity equations would then take the following fom: F e whee: N A S LA ) ( S A ) (22) ( c b u b u l F e = Static enetation esistance N c = beaing caacity facto = bulk unit weight L = deth of the samle = adhesion facto A b, A l = aea of base and shaft of samle Combination of equations 4, 5 and 22, leads to the detemination of undained shea stength which equies a calibation fom enegy measuements to detemine 1, 2 and 3 (section 2), estimation of (Flaate, 1968; Tomlinson, 1969) and a ediction of N c (o N k,spt ). N k,spt should be obtained by calibation against undained shea stength values measued fom a efeence test in site-secific coelations and will eflect the dynamic natue of the test. Alication of equation 22 to Guabiotuba clay (Schnaid et al, 2005) is shown in Figue 17, in which consideable scatteed is obseved in a lot of undained shea stength vesus deth. Pedicted values of S u fom SPT fall within the egion defined by othe tests in this OC clay. Table 2: Theoetical cone facto solutions N c 7.0 to lnI 12+lnI I 1.67 c s ln I Penetation model BCT CET CET SPM+ FEM Refeence Caquot & Keisel (1956); de Bee (1977) Vesic (1975) Baligh (1975) Teh & Houlsby (1991) ln I FEM Yu et al (2000) 1 A 1 2A ln I 1 A 3 R1 1 A 1 2A 0.52A 1/ 8 (1 A ) 2 CET+ FEM Su & Liao (2002) lnI -2.1 G S Note: I ; u (1 K ) vo o ; = cone oughness; A =stength anisotoy atio. 2S u CET+ FEM Abu-Fasakh et al (2003) 15

14 Figue 16. (a) Beaing caacity fom T-ba: (a) Long & Phoon (2004) and (b) Randolh (2004) Jeffeies (1988) extended the elasto-lastic solutions to include unloading by exessing the essuemete unloading cuve in tems of small stains as: P P max G a max a 2S u 1 ln 2Su ln (24) Su a amax Figue 17. Undained shea stength of stiff Guabiotuba Bazilian Clay (Schnaid et al, 2005). The essuemete is a well-suited altenative to enetation, since theoetical inteetation methods that educe the exansion ocess to one dimension by idealizing the geomety of a essuemete obe as an infinitely long, cylindical cavity ae well established. A total stess loading analysis of cylindical exansion in cohesive mateial was fist examined by Gibson & Andeson (1961). The authos deived a theoetical exession fo the total essuemete essue P at the stage of lastic loading: G V P ho Su 1 ln Su S ln (23) u V whee V/V is the volumetic stain and ho the total in situ hoizontal stess. By lotting essuemete esults in tems of cavity essue against the logaithm of the volumetic stain, as eesented in Figue 18a, the sloe of the elastic otion is linea and is equal to the undained shea stength, S u. Figue 18. (a) Inteetation of (a) essuemete (Gibson & Andeson, 1961) and (b) cone-essuemete (Houlsby & Whithes, 1988). whee P max and a max ae esectively the cavity essue and cavity adius at the end of the loading stage and a denotes cavity adius at any stage of essuemete unloading. Pessuemete unloading esults should be esented as essuemete essue vesus ln(a max /a-a/a max ); the sloe of the lastic unloading otion in this aticula lot is linea and is equal to twice the undained shea stength. Installation oblems associated to the self-boing technique ae atially ovecome by lage stain unloading analysis which eventually led to the develoment of the cone essuemete (Withes et al, 1989). A solution to deive the undained shea stength and shea modulus fom CPMT data in clay has been develoed by Houlsby and Withes (1988). Penetation is modelled theoetically as the exansion of a cylindical cavity within the soil. The exansion of the essuemete test is taken as a continued exansion of the same cylindical cavity and the contaction hase as a cylindical contaction. As fo the selfboing essuemete, the analysis gives ise to a simle constuction of the exeimental essuemete cuve, whee the sloe of the unloading lastic data in a lot of P vesus ln[( c ) max - c ] is equal to 2S u, as eesented in Figue 18b. An altenative aoach to these well established methods is to oduce an image cuve of the comlete essuemete loading and unloading data (Jeffeies, 1988) using a softwae develoed fo esonal comutes to efom a cuve-fitting analysis fom which soil aametes and the in situ hoizontal stess can be assessed. The ogam executes the calculations fo loading and unloading equations and lots the data. The oeato comaes the quality of the match and modifies the inut soil aametes until a cuve fitting is eached. The aametes that oduce an analytical cuve that satisfactoily fits the exeimental esults ae, in theoy, eesentative of the soil behaviou. The dawback is that an equally good fit to the data may be achieved by diffeent combinations of aamete values, so that engineeing judgment is always equied to avoid an unlikely set of aametes to be adoted. Cuve fitting analysis has been extensively used by the autho and is late exloed in the inteetation of essuemete test esults in cohesive-fictional soils, following ecommendations made by Mantaas & Schnaid (2002) and Schnaid & Mantaas (2003). In addition to isotoic mateials, coss-anisotoy mateials can also be model by cavity exansion (e.g. Gaham & Houlsby, 1983; Wu et al, 1991). Analytical solutions that can find alicability to essuemete tests conside the axial diection as the axis of symmety and the adial (,) lane as the isotoic lane. Solutions ae alicable to small stain 16

15 anisotoy and have elative little alication to soft clay whee lastic defomations ae likely to goven geotechnical design. As fo lage stain anisotoy, the mode of sheaing in a essuemete test is edominantly hoizontal and gives an aveage hoizontal stength that is neve measued in outine laboatoy tests excet fo a geotechnical testing ogamme that comises hollow cylinde tests. Unfotunately measued values of essuemete undained stength ae often subjected to inaccuacies that emege fom geomety, installation distubance, atial dainage and stain ate effects, which can oveshadow the influence of anisotoy. A steady gowth in the use of DMT esults in geotechnical actice has led to the develoment of seveal coelations to estimate S u (Machetti, 1980; Lacasse & Lunne, 1988; Powell & Uglow, 1988). The oiginal coelation as oosed by Machetti (1980) is: S 0. K u 0.22 vo 5 D (25) Numeical analysis of the installation of flat dilatometes eoted by Huang (1989), Finno (1993), Whittle & Aubeny (1993) and Yu (2004) have ovided useful insights of the dilatomete test. These studies geneally suot the emiical coelation fom equation 25 (e.g. Machetti, 1980; Machetti et al, 2001). In addition, the geotechnical liteatue offes a wide numbe of examles in soft clay deosits, in which emiical coelations have been used successfully. A numbe of eoted cases is comlied in Figue 19. whee aamete should be obtained fom site secific coelations but can be adoted as a fist estimate as 0.48 FV 22( I ), whee I is the lasticity index. Pediction of OCR fom iezocone data can be made fom both theoetical and emiical coelations (Sennesset et al, 1982; Woth, 1984; Mayne, 1991; Tavenas & Leouiel, 1987; Konad, 1987). Mayne (1991) suggested an aoach based on cavity exansion theoy and citical state theoy, exessed as: 1 q 2 c u OCR 1.95M 1 vo 2 (27) In addition, emiical estimates of the econsolidation stess can be obtained as (Chen & Mayne, 1996): 0.305( qt ) vo o (28) ( qt vo )( I ) Othe coelations in the fom of k( u 2 u o ) o k( qt u2 ) can also be adoted beaing in mind that they ae all site-secific coelations that should be validated locally. A similaity between the dilatomete K D and the OCR ofiles was fist ointed out by Machetti (1980), and latte confimed by seveal authos (e.g. Jamiolkowski et al, 1988; Powell & Uglow, 1988; Kamey & Iwasaki, 1995). Fo uncemented clays OCR can be simly edicted as: 1.56 OCR ( 0.5K D ) (29) Figue 19. Undained shea stength fom dilatomete (Powell & Uglow, 1988). 4.4 Stess histoy Stess histoy is exessed by the oveconsolidation atio OCR, defined as the atio of the maximum ast effective mean stess and the cuently alied stess. In actice the cuent stess is taken as the esent effective ovebuden stess which chaacteises a mechanically oveconsolidated soil. The geotechnical liteatue offes a wide vaiety of methods designed to estimate OCR in clay based mainly on FVT, CPT and DMT, which ae biefly summaized hee. The field vane measues the undained shea stength and as an additional use of the device esults may ovide assessment of OCR fom nomalised undained stength to ovebuden atio (Mayne & Mitchell, 1988): OCR S ufv FV (26) vo This coelation is simly built-in the evidence of K D 2 fo OCR=1 and its usefulness has been extensively demonstated exeimentally (Jamiolkowski et al, 1988; Finno, 1993; Kamey & Iwasaki, 1995). Numeical studies caied out by Yu (2004), assuming that the installation of a flat dilatomete can be simulated by a flat cavity exansion ocess, enabled a theoetical elationshi between K D and OCR to be obtained. In Figue 20, a numeical estimative of OCR fo thee diffeent clays ae comaed to edictions obtained diectly fom equation 29. With the excetion of heavily oveconsolidated clays (OCR>8), equation 29 can be used with easonable confidence. Figue 20. Theoetical coelation between K D and OCR (Yu, 2004). 17

16 4.5 Consolidation coefficients Coefficients of consolidation can be assessed in situ fom obsevations of settlements unde embankments o diectly fom in situ test esults, efeably fom iezocone dissiation tests and SBPM holding tests. Analytical and numeical ocedues have been develoed to ovide an estimate of the coefficient of consolidation C h fom iezocone dissiation tests in which the decay of excess oe essue with time is monitoed. Methods ely eithe on onedimensional cavity exansion (Tostensson, 1977; Randolh and Woth, 1979) o two-dimensional stain ath method (Levadoux & Baligh, 1986; Baligh & Levadoux, 1986; Teh & Houlsby, 1991; Buns & Mayne, 1998). The analysis of dissiation tests ae mostly edicted on the basis of uncouled consolidation theoy and equies two distinct stages. The fist consides the enetation of a cone into an elastic-efectly lastic isotoic, homogeneous mateial viewed as a steady flow of a soil ast a static cone, leading to the ediction of total stesses and oe wate essues in the soil aound the obe. The second stage of the analysis takes these oe essues as the initial values fo a Tezagui uncouled consolidation ocess, and calculates the subsequent dissiation aound a stationay obe. Although the ecise values of the oe essues as a function of time deend on the secific initial oe essue distibution, Teh & Houlsby (1991) showed that this fom could be genealized by the definition of a suitable dimensionless time facto fo the consolidation ocess T*, exessed as: T c t * h 2 (30) R I whee R is a obe adius in a soil of igidity index I and t is the dissiation time. Thus, fo a given T* and t values, the calculated C h values is diectly ootional to the squae oot of the I value. A constant I is used in the solution although in fact the value of the shea modulus will deend on the shea stain amlitude, which is shown by stain ath calculations to vay in a comlex manne aound a 60 o enetomete. Thee indeendent sets of data fom both field and laboatoy testing ogammes eoted by Schnaid et al (1997) in which the oe essues wee measued at fou diffeent locations on the iezocone suggest that equation 30 should be alied to oe essues measued at the shoulde immediately above the cone face. The theoetical cuves ovide a less good match with the exeimental dissiation ecods measued at othe locations (u 1, u 3 and u 4 ). Based on the stain ath method, Robetson et al (1992) oduced the chat illustated in Figue 21 that can be diectly adoted to estimated C h fom the actual time that takes fo 50% consolidation t 50. The success of the analysis in edicting C h deends on the use of aoiate I values and in tun deends on the shea modulus G. Fo natual clay deosits, I anges fom less than 50 to moe than 600; it is known to decease with inceasing OCR and, fo the same OCR, it inceases with deceasing PI. Recommendations ae made to use the Houlsby and Teh theoetical solution with an I value calculated fom tiaxial tests, in which the shea stiffness G is taken at a level of 50% of the yield stess (Schnaid et al, 1997). Tiaxial CIU, CK o U and UU tests oduce the following eesentative values: I of 70 fo Oxfod calibation chambe tests (afte Schnaid et al, 1997), 44 fo the Saaui site in Bazil (Danzige et al, 1996) and 135 fo Ceasa site in Bazil (Schnaid et al, 1997), also illustated in Figue 21. Figue 21. Pediction of C h fom iezocone tests (Robetson et al, 1992). Stength, stiffness and hoizontal stess can be assessed fom SBPM tests. A futhe alication to this device is to obtain an indiect estimate of hoizontal coefficient of consolidation and emeability fom a dissiation-tye holding test. Take the case in which the essuemete cavity is held constant and the measued oe essues aound the obe dissiates. The consolidation coefficient is estimated fom Figue 22, in which a dimensional time facto T 50 = C v t 50 /a 2 is lotted against the nomalised maximum excess oe essue, U max /S u (Clake et al, 1979). Since the solution does not take into account the viscous behaviou of the clay, the stess elaxation that may occu duing stain holding tests cannot be eoduced. Flow chaacteistics of clays can also be assessed fom dilatomete tests (e.g. Machetti et al, 2001). Available methodologies consists in stoing the DMT blade at a given deth fo monitoing some fom of decay with time fom which C h is infeed. Machetti and Totani (1989) suggests monitoing the decay of the total contact hoizontal stess wheeas Schmetmann (1988) and Robetson et al (1988) ecommends the decay in oe essue in the soil facing the membane. Although a diect analogy to essuemete holding tests and CPT dissiation tests can be made, the coesonding theoy fo DMT has not yet been develoed; C h values edicted fom DMT should theefoe be viewed as aoximation. Figue 22. Solution fo holding tests (Clake et al, 1979) 18

17 Much has been witten with egad to the vaiations on the measued values of coefficients of consolidation. Figue 23 shows a tyical case study at the Chamlain Sea Site (Leoueil et al, 1995), whee measuements on C h ange fom 10-8 to 10-5 m 2 /s. Disceancies between field and laboatoy tests ae due to one o moe of the of the following main easons (Simon, 1975; Hamouche et al, 1995; Leoueil et al, 1995; 2002; Leoueil & Hight, 2003): a) hydaulic conductivity measued in small laboatoy secimens may undeestimate the hydaulic conductivity of the soil mass, aticulaly in heteogeneous statified deosits; b) inaccuacy of constitutive models and bounday conditions adoted in the inteetation in situ tests; c) the two o thee-dimensional asect of field conditions and of emeability anisotoy neglected in the inteetation of field oblems. can be exessed in the fom of a logaithm o an invese of a hyebolic sine function (e.g. Randolh, 2004). Its most widesead fom is:. S u S u, ef log. ef whee (31) S u = undained shea stength at a given loading ate S u,ef = undained shea stength at a efeence ate = ate law coefficient (tyically anging fom 0.10 to 0.15). = stain ate. = efeence stain ate ef An examle of the influence of ate effects on cone enetation is illustated in Figue 24, in which both the measued cone atio (q cnet /q cnet,undained ) and the excess oe essue atio is lotted as a function of the nomalised enetation atio, V (Randolh and Hoe, 2004). The nomalised enetation atio V is exessed as: vd V (32) C v Figue 23. Range of vaiability of C v fo the Chamlain Site (Leoueil et al, 1995). 4.6 Viscous effects Stain ates may account, at least in at, fo the disceancy between field and laboatoy measuements. Tyical ates of laboatoy testing ae of 1%/hou (10-3 %/s) wheeas field tests ae associated to stain ates of 10 3 %/s. In comaison, nominal stain ates associated to geotechnical constuction may ange fom 10-5 %/s to 10-7 %/s. The influence of sheaing ate on undained behaviou unde monotonic load is fully ecognised (Tavenas & Leoueil, 1977; Vaid et al, 1979; Kulhawy & Mayne, 1990; Sheahan et al, 1996; Biscontin & Pestana, 2001; Einav & Randolh, 2005) but, desite its majo imact on both stength and stiffness, attemts to incooate this effect into the inteetation of in situ testing ae faily ecent (e.g. Randolh, 2004). The effect of stain ate on the measued o deduced shea stength of clays whee v = enetation velocity d = diamete of the enetomete C v = coefficient of consolidation Ti esistance and oe wate essue incease due to viscous effects as the enetation ate inceases. Penetation data follows a easonably linea elationshi in a semi-logaithm sace, defining a ate law coefficient = Fo a dimensionless velocity lowe than 10 the cone atio also inceases, this time as a function of consolidation effects. Section 7 of this Reot esents a detailed discussion of the effects of atial dainage and oduces ecommendation fo test ocedues in intemediate emeability soils. Figue 24. Stain ate effects duing cone enetation (afte Randolh, 2004) 19

18 4.7 Closing emaks Available theoetical and emiical aoaches ae mainly elated to high lasticity clays in which simle ateindeendent, efectly-lastic isotoic soil models ae geneally adoted in the inteetation of in situ test data. Wheneve the exeience of soft clays is extended to stiff oveconsolidated clays, inteetation of test data should follow secific ecommendations. Given the uncetainties that emege fom stuctue, stiffness non-lineaity, anisotoy, stess histoy, stain ate, among othe effects, site secific coelations obtained fom calibation against laboatoy tests ae geneally ecommended. Desite the due ecognition of limitations in the inteetation of in situ tests, it is in the difficult conditions of soft clay that both laboatoy and in situ tests offe the best and most eliable combination of test ocedues and inteetation methods available fo assessing constitutive aametes fo engineeing design oblems. 5 SAND PROPERTIES Engineeing design in ganula soils is about finding means of deiving constitutive aametes without ecalling to the benefits of efoming elemental laboatoy tests. Without the ossibility of etieving undistubed samles, academics develoed inteetative methods fom in situ tests that enginees often use without being awae of the simlifications embedded in the solutions. In this Reot advantages and limitations of cuently adoted inteetative techniques and analyses ae highlighted. Since the in situ behaviou of ganula soils is comlex, and the chaacteistic dilatant esonse is not always accuately catued by existing soil models, a single geneal ecommendation is to coss-coelate measuements fom indeendent tests. When data ae combined thee is moe scoe fo ational inteetation and, fo this eason, emhasis has been laced on coelations with mechanical oeties that ae based on the combination of measuements fom indeendent tests such as the atio of the elastic stiffness to ultimate stength (G o /q c, G o /N 60 ), the atio of cone esistance and essuemete limit essue (q c /) and the association of stength and enegy measuements (N 60 and enegy). 5.1 Backgound eseach Fundamental undestanding on the behaviou on ganula soils has been obtained fom laboatoy tests in econstituted samles. Fom the macomechanics oint of view, a numbe of chaacteistic featues ae now ecognized as imotant in ganula geomateials: a) dained esonse evails unde monotonic load due to the elatively high hydaulic conductivity of sands and gavels. Unde cyclic load hydaulic conductivity may not be sufficient to event the develoment of excess oe wate essue; b) basic linea and non-linea stess-stain concets alied fo clay can be extended to othe geomateials by ecognizing the dominant influence of void atio and mean stess on soil stiffness. The vaiation of G o values as a function of void atio e fo a vaiety of soils is shown in Figue 25, in which it is demonstated that the sloe that eflects the void atio function F(e) does not seem to be significantly influenced by the tye of soil (e.g. Lo Pesti, 1989; Jamiolkowski et al, 1991); c) fo clean sands, the initial stiffness is unaffected by the ate of sheaing and stess histoy; fo cabonate and cushable sands, the effects of ove-consolidation ae moe onounced, as comehensively discussed by Tatsuoka et al. (1997); Figue 25. Nomalised maximum shea modulus vesus void atio (Jamiolkowski et al., 1991). d) a fundamental issue in ganula geomateials is to descibe soil dilatancy, defined as the atio of lastic volumetic stain incement to lastic deviatoic stain incement. An ealy, widely used and well tested stess-dilatancy equation is the one oosed by Rowe (1962) that exesses dilatancy as a function of the stess atio and the intenal inteaticle fiction angle: 1 3 tan cv d. 1 d v 1 (33) whee 1 and 3 ae incial stesses, v is the volumetic stain and 1 is the majo incial stain. Rowe s wok was followed by seveal eseach effots that attemted to integate the behaviou of ganula soils ove a ange of densities and stess levels (e.g. Nova & Wood, 1979; Bolton, 1986; Wan & Guo, 1998; Li & Dafalias, 2000); this can be achieved by the state aamete; e) the behaviou of ganula soils io to the achievement of the citical state is lagely contolled by the state aamete. A concet intoduced by Woth & Basset (1965) and develoed by Been & Jeffeies (1985), the state aamete is defined as the diffeence between cuent void atio e and citical state void atio e c, at the same mean stess (Figue 26a). It can be conveniently exessed as: e ln 1 (34) whee 1 is the efeence mean effective stess taken as 1 kpa. In a manne simila to sedimented clay which is well chaacteised by the oveconsolidation atio, sands can be chaacteised by the state aamete following the ecognition that it combines the effects of both elative density and stess level. Been & Jeffeies (1985) evocate this deendency to establish a sounding elationshi between and the intenal fiction angle fo sands. Hid & Hassona (1986) obseved that sands can be chaacteised by the atio of cs / o, when o is the cuent mean effective stess and cs is the mean effective stess on the citical line at the same void atio. This atio is diectly elated to. f) although the state aamete aoach offes a simle and yet fundamental contibution to the chaacteisation of ganula geomateials, by imlicitly combining void atio and comessibility, imotant limitations associated to have been ecognised fo some time. Thee is now an inceasing consensus that, unlike fo clays, the citical state line on sands is non-linea in the -ln sace (e.g. Rieme 20

19 et al, 1994; Konad, 1998; Jeffeies & Been, 2000), exhibiting a steee sloe at high stesses due to changes in gadation and gain shae induced by gain cushing, as shown in Figue 26b. The amount of aticle cushing that occus in an element of soil unde stess deends on aticle size distibution, aticle shae, mean effective stess, effective stess ath, void atio and aticle bonding (e.g. Hadin, 1978). Tiaxial tests in samles of equal elative density have distinctively shown that the dilatant comonent educes with inceasing mean effective stess, since at high stesses, cushing eliminates dilation (e.g. Bolton, 1986). g) most natual soils and soft ocks ae micostuctued so that, at a given void atio, they can sustain stesses highe than could the same mateial non-micostuctued (e.g. Buland, 1990; Leoueil & Vaughan, 1990; Schmetmann, 1991; Clayton & Seatice, 1993; Leoueil & Hight, 2003). Although this is evident fo had soils (and will be detailed in section 6) thee is also evidence of micostuctue in sands (Mitchell & Solyma, 1984; Schmetmann, 1991; Fenandez & Santamaina, 2001). Given the difficulty of samling ganula soils, evidences of bonding have to be obtained fom field tests, allowing not only soil behaviou to be chaacteised but the influence of micostuctue on the small stain shea modulus to be quantified (e.g. Eslaamizaad & Robetson, 1997; Schnaid, 1997; Schnaid et al, 2004). Wheeas the stength-dilation theoy and citical state concets geneally aly, in sand uncetainty aises fom the shea stength enveloe non-lineaity which inceases with inceasing elative density and gain cushability. Since cementation, aging and cushability ae henomena ecognised as imotant in the behaviou of ganula mateials, a eseach challenge is to incooate these effects into inteetation methods of in situ tests. We deal almost exclusively with noncemented, cohesionless soils, because ou exeience has been built on calibation against a database fom lage laboatoy calibation chambe tests on fesh sands. 5.2 Chaacteisation of aging and cementation The eceding discussion in clay has aleady emhasized the imotance of the igidity index to soil chaacteisation, fequently used fo numeical analysis in geomechanics and aticulaly elevant to the inteetation of in situ tests. In incile, a mateial that is stiffe in defomation may be stonge in stength, yielding the following emiical elationshi that aeas to be valid fo vaious geomateials anging fom soft soils to had ocks (e.g. Tatsuoka & Shybuya, 1992; Shibuya et al, 2004): E q (35) max / max This tye of coelation aims at evaluating the satial vaiability of stiffness fom measued stength obtained in conventional laboatoy testing. Since ganula mateials ae difficult to samle and the chaacteistic stuctue of natual sands cannot be eoduced by econstituted samles in the laboatoy, it is necessay to develo a methodology caable of identifying the existence of distinctive behaviou emeging fom aging o cementation fom field tests. Following ecent ecommendations that soil chaacteisation and mechanical oeties should be efeably based on the combination of measuements fom indeendent tests (Yu et al, 1996; Schnaid, 1997; Odebecht et al, 2004; Schnaid et al, 2004), an aoach has been develoed on the basis of the atio of the elastic stiffness to ultimate stength (G o /q c, G o /N 60 ). This is achieved by combining cavity exansion theoy and citical state concets to the vaiables that contol the small stain stiffness of sand (Schnaid & Yu, 2005). Theoetical solution fo cone enetation and backgound eseach of small stain stiffness that fom the basis of the esent analytical study ae biefly summaised as follows. The stiffness of sand has been long ecognized to be contolled by the confining stess and voids atio (section 4, e.g. Hadin & Black, 1968; Hadin, 1978; Lo Pesti, 1989). This has omted the establishment of many useful coelations fo edicting G o adoting slightly diffeent F(e) functions (e.g. Iwasaki and Tatsuoka, 1997; Lo Pesti et al, 1997; Shybuya et al; 1997). Lo Pesti et al (1997) suggest the following exession: G n ( ) ( ) ( kpa) x 1n o Ce a (36) having C=710, n=0.5 and x=1.3 as aveage values fo the mateial aametes. Fo solutions of cone enetation in sand, the significant volume change that occus in shea has to be catued. Give its comlexity, the existing methods of inteetation of the enetation mechanism in sand can only be egaded as aoximation and solutions have to be calibated against exeimental data fom calibation chambe tests. Using a state aamete based, citical state soil model, Collins et al (1992) esented a solution fo the exansion of a sheical cavity that can be conveniently exessed as: C2C3 (1eo ) C ) ex C (1 e ls 1( o 4 o ) (37) Figue 26. Definition of the state aamete (afte Been & Jeffeies, 1985) and (b) non-linea chaacteistic of the citical state line (afte Konad, 1998). 21

20 whee ls is the effective sheical cavity limit essue. Citical state aametes and constants C 1, C 2, C 3 and C 4 have been calibated against laboatoy chambe testing data in six efeence sands. The effective cone esistance q c can be estimated fom diectly fom ls (Ladanyi & Johnston, 1974): q c ls (1 3 tan s (38) whee s is the lain stain eak fiction angle of the soil, as cone enetation is assumed to occu unde lane stain conditions. In equation 38 effective stesses and a ough soilcone inteface ae assumed to be valid. Theoetical values of q c calculated fom equation 38 can be elated to the magnitude of G o calculated fom equation 36, ovided that the initial void atio and in situ stess state ae known o estimated. Fom this coelation it is ossible to exess both G o / q c and q c1 fom a set of given aametes: e, M,, and. The aoach makes use of a nomalized dimensionless aamete q c1, defined as: q c q c a a 1 (39) v whee a is the atmosheic essue. The theoetical coelation between G o /q c and q c1 is shown in Figue 27 in a elationshi eesentative of unaged uncemented soils. Results wee calculated in a stess inteval between 50 kpa to 500 kpa which should cove the ange of alications encounteed in geotechnical engineeing actice. The comuted values ae shown to be insensitive to both the initial stess state and soil comessibility which fully justifies its use fo soil chaacteisation. Fo comaison with natual sands, the ue and lowe boundaies emiically established by Schnaid et al (2004) ae esented togethe with the theoetical deived database. The boundaies that eesent the vaiation of G o with q c fo fesh unaged uncemented sand deosits ae exessed as (Schnaid et al, 2004): G 3 o qa v a (40) whee is a aamete that anges fom 110 to 280 (units in.eq. 40 ae in kpa). Note that these boundaies have oughly the same sloe as the one oduced by the analytical solution. Figue 27. The atio of the elastic stiffness to cone esistance edicted fom citical state aoach (Schnaid & Yu, 2004). Fou indeendent sets of data have been used to validate this theoetical aoach (Figue 28): centifuge tests caied out on Fontaineblue sand (Gaudin et al, 2004), comilation of a database fo natual sands by Eslaamizaad & Robetson (1997), a vaiety of sand tyes fom Westen Austalia (Schnaid et al, 2004) and data eoted fo the Canadian Liquefaction exeimental site, Canlex (e.g. Wide et al, 2000). Gaudin et al (2004) esented a set of centifuge tests in which both G o and q c have been measued. By taking account of chambe size effects, the measued data lotted in Figue 28a confims the theoetically deived line and demonstates that G o /q c atio is elatively insensitive to stess level and elative density. The atio of G o /q c obseved in the centifuge is shown to be geneally highe fo loose sand than that fo dense secimens, imaily because the latte exhibit highe q c1 values. Comaison with natual sands is ovided by Figues 28b and 28c. Fesh uncemented sand chaacteises a egion in the G o /q c vesus q c1 sace defined by equation 40. Values of q c and G o ofiles that fall outside and above this egion suggest ossible effects of stess histoy, degee of cementation and ageing, as demonstated by Eslaamizaad & Robetson (1997). A vaiety of sand tyes in Austalia shows that the statigahy includes a ecent deosited siliceous sand fill laced hydaulically fo eclamation woks and a dune sand fom Seawood which was laid down in the late Pleistocene as a limestone but was subsequently leached of vitually all its calcium cabonate content, both of which show no taces of a defined stuctue. In contast, Safety Bay Sand, which was deosited unde littoal and aeolian conditions in the mid- Holocene, contains many shells and has a calcium cabonate content in excess of 50% and Lowe Guildfod siliceous sand, which was laid down by steams duing the ealy Pleistocene, is shaed by ageing effects. As indicated by Schnaid et al (2004), G o /q c atios in the lowe Guildfod sand ae tyically about five times highe than those ecoded in the hydaulic fill and esults of these cemented aged sands fall above the egion defined fo uncemented mateials. Futhe evidence can be obtained fom field tests using the database eoted fo the Canadian Liquefaction exeimental site, Canlex (e.g. Wide et al, 2000). The Canlex oject has oduced detailed investigation of thee majo locations containing hydaulic laced loose sand deosits, comising both field and laboatoy tests whee undistubed samles have been etieved fom gound feezing techniques. Two locations ae aticulaly useful fo the esent discussion and will be extensively analysed: the 12 yeas old laced, 27 to 37 m dee Milded Lake deosit and the 2 months aged at time of test, 3 to 7 m dee J-it deosit. Reoted ti cone esistance and small stain stiffness have been lotted in Figue 28d and aea to indicate no cementation in both sites and no aging effects at the Milded Lake 12 yeas old deosit. Significant scatte obseved in natual sand deosits is not a function of soil comessibility as suggested in ealy studies, but obably esults fom soil anisotoy, as well as some degee of cementation and aging. Due to anisotoy, G o is not indeendent on the diection of oagation and olaization of shea waves and coelations with enetation esistance should efeably be exessed as a function of G vh, G hv o G hh. Scatte can also be atially attibuted to the influence of the hoizontal stess on both initial stiffness and ti cone esistance. (e.g. Schnaid & Houlsby, 1992). Equation 40 should ideally be efeed to hoizontal stess o mean in situ stess athe than to vetical stess. The efeence fo vo is justified by the imossibility of detemining with easonable accuacy the value of the hoizontal stess in most natual deosits, because they have undegone comlex stess histoy, cementation and aging effects that ae difficult to econstuct. The geneal good ageement between the theoetical coelation, centifuge data and field data entiely justify the use of G o /q c fo the chaacteisation of natual sand deosits. The G o /q c atio is useful because it ovides a measue of the atio of the elastic stiffness to ultimate stength that may be exected to incease with sand age and cementation, imaily because the 22

21 a) Fountanebleau Sand centifuge tests (Gaudin et al., 2004) b) Uncemented unaged sands (Eslaamizaad & Robetson, 1997) c) Westen Austalia sands (Schnaid et al., 2004) d) Calibation chambe data (Schnaid & Yu, 2005) Figue 28. The atio of elastic stiffness to cone esistance fo centifuge tests and natual sands effect of these on G o is stonge than on q c. The atio is not sensitive to changes in mean stess, elative density and sand comessibility. G0 110 q c a 3 v : lowe bound, uncemented (42) 5.3 Stiffness Assessment of eliable stess-stain elationshis of soils equies a coect evaluation of both the small stain stiffness and the shae of the degadation cuve. In the eceding section, the vaiation of G o with q c obseved fo natual sands was summaised in Figue 28. Equation 40 can match the ange of ecoded G o values and desite the fact that this equation has oiginally been oosed to distinguish cemented and uncemented soils, it is likely that actitiones may be temted to emloy it to estimate G o. Fo eliminay design, in the absence of diect measuements of shea wave velocities, the oosed lowe bounds ae ecommended fo an evaluation of the small stain stiffness fom q c (Schnaid et al, 2004): G0 280 q c a 3 v : lowe bound, cemented (41) These equations edict values of G o that ae not fa fom eviously ublished elationshis develoed fo sands (Bellotti et al., 1989; Rix & Stoke, 1991; Jamiolkowski et al., 1995b). Howeve the effect of natual cementation and ageing is quantified hee and is shown to oduce a maked incease in both G o /q c atio. Given the consideable scatte obseved fo diffeent soils, these coelations ae only aoximate indicatos of G o and do not elace the need fo in situ shea wave velocity measuements. il stess-stain behaviou is non-linea and this intoduces uncetainties on the selection of design aametes fo outine design calculations, as outlined in Section 4. Seveal aoaches have been develoed fo the descition of the non-linea stessstain behaviou of soils, adoting mathematical models exessed, fo examle, by logaithm o hyebolic cuves. Fo sands, in the absence of laboatoy tests fom undistubed samles, coss-efeence between small stain stiffness fom seismic tests and intemediate stiffness fom essuemete o 23

22 late loading tests should be sufficient to eesent deviatoy non-linea stess-stain behaviou inside the small stain egion. The oosed method of descibing the modulus degadation cuve is to cuve-fitting a essuemete test that includes unload-eload loos using a non-linea soil model. Assuming that G o is known, the fitting ocess may be sufficiently accuate to detemine the model aametes that eesent the soil non-linea esonse. It is hee woth to ecall that a tyical essuemete cycle is highly hysteetic and as a consequence the unloading and eloading sections do not coincide so that the actual shae of the cycle eesents the nonlinea esonse of the soil, as demonstated in Figue 29. This figue shows tyical essuemete unload-eload cycles and the theoetical fitting cuves using a numeical analysis couled to the hyebolic elationshi oosed by Fahey and Cate (1993): G G o 1 f max g (43) whee f contols the stain to eak stength ( max ) and g detemines the shae of the degadation cuve as a function of mobilised stess level (). Seveal simila aoaches have been oosed in the liteatue giving secific ecommendations to bette chaacteise the non-linea esonse of sand fom in situ tests (Bellotti et al, 1989; Mui Wood, 1990; Feeia & Robetson, 1992, Fahey and Cate, 1993). Uncetainty aises when soil stuctue and destuctuation becomes imotant as ecently ointed out by Schnaid et al (2004) and outlined in Section 6 (Bonded geomateials). Figue 29. Tyical unload-eload cycle in a essuemete test 5.4 Shea stength and state aamete In cohesionless soils, the stength aamete of majo inteest is the intenal fiction angle and its assessment has to be necessaily made fom in situ tests given the difficulties in etieving samles fo laboatoy tests. The significant volume change that occus in sand duing shea can be catued in the analysis of essuemete tests. Due to the comlicated stain field oduced duing CPT/SPT/DMT enetation in fictional dilative soils, in situ enetation tests emain imaily based on emiical evidence suoted against calibation in lage laboatoy chambes. Some established methods fo inteetation of shea stength fom in situ tests ae biefly discussed and new develoments ae addessed. Analysis of essuemete tests in sand assume fully dained conditions so that no excess oe wate essues will be develoed thoughout the test. A small stain cavity exansion solution develoed to deive the angles of fiction and dilation was fist deduced by Hughes et al (1977) and is widely accet in actice. The sand is assumed to behave as an elastic-lastic Moh-Coulomb mateial and the elastic defomation is disegaded in the lastic zone. An analytical solution fo the cavity exansion cuve in the lastic egion can be exessed as: ln( P ) s ln c A (44) whee P is the effective cavity essue, c is the cavity stain, s=(1+sin)sin/(1+sin), A is a constant and is the angle of soil dilation. A diect imlication is that in a essuemete esult lotted as P vesus c in a logaithmic scale, the staight line sloe of the lastic otion is equal to s, which in tun is a function of the fiction and dilation angles. Fom Rowe s stess-dilation (equation 33), it is ossible to exess and as a function of s and cv : sin sin s 1 ( s 1)sin (45) cv s ( s 1) sincv (46) Seveal othe solutions have been late oosed to the analysis of both the loading (Manasseo, 1989; Sousa Coutinho, 1990) and unloading (Houlsby et al, 1986) of a essuemete in sand. Textbooks such as Clake (1993) and Yu (2000) ae excellent efeences to eview these methods. As fo inteetation of the CPT in sand, thee ae two ossible aoaches: (a) analysis based on beaing caacity and cavity exansion theoies which, given the comlexities of modelling enetation in sand, can only be egaded as aoximate (Vesic, 1972; Dugunoglu & Mitchell, 1975; Salgado et al., 1997) and (b) methods based on esults fom lage laboatoy calibation chambe tests (e.g. Bellotti et al., 1996, Jamiolkowski et al., 1985). A ecent tend has been to inteet cone enetation esults in tems of the state aamete by combining ti esistance with seismic measuements o essuemete limit essue (Yu, Schnaid & Collins, 1996; Schnaid & Yu, 2005). Analytical solutions have been useful as means of undestanding the enetation mechanism, as well as identifying the influence of diffeent constitutive aametes on the enetation ocess. Dugunoglu & Mitchell (1975) esented a well-known beaing caacity solution fo dee cone enetation. A simle exession is obtained fom a lane stain failue mechanism couled to an emiical shae facto to account fo the axisymmetic geomety of the cone, fo cases in which the soil-cone inteface fiction angle is half of the soil fiction angle: q 0.194ex(7.63 tan ) c N q (47) vo whee N q is the cone facto in sand. Altenatively, Salgado et al (1997) used the analogy between cavity and cone enetation to deive a simlified solution exessed as: 1l [(1 C) (1 l) C 1] qc 2ex( tan ) P 2 lc (48) C l(1 l) whee P lc denotes the effective cylindical cavity limit essue, l is detemined numeically and C is linked to the soil dilation 24

23 angle by C 3 ex[( / 2) tan )]. Calibation against a lage numbe of chambe tests suggests that measued cone esistance can be edicted to within 30%. A wide-sead aoach in engineeing actice is to estimate elative density D fom cone ti esistance adoting emiical coelations (e.g. Jamiolkowski et al., 1985; 2003; Houlsby, 1998). Jamiolkowski et al, (2003) suggested a coelation between cone esistance, elative density and mean effective stess in the fom: 1 C 2 qc ln C0 C1 D (49) Accoding to the authos, vetical stess can substitute mean stess in equation 49 ovided that the soil is nomally consolidated having a K o constant with deth. In this case aametes C o, C 1 and C 2 can be taken as 17.68, 0.50 and 3.10 esectively. Calibation chambe data in sand have clealy shown that, fo a given density, cone esistance deends imaily on the in situ hoizontal stess and theefoe ho must be accounted fo in a ational inteetation of field tests (Houlsby & Hitchman, 1988); Schnaid & Houlsby, 1992). The deendency of elative density (D ) on q c and ho is illustated in Figue 30 and can be exessed in ecentage as (Schnaid and Houlsby, 1992; Nutt & Houlsby, 1992): efectly-lastic Moh-Coulomb theoy. A comaison between emiical and numeical edictions fo a case consideing K o =1 is esented in Figue 31, in which the K D /K o atio is lotted against the igidity index. Although the numeical solution is still unable of oducing ealistic edictions of fiction angle, it does identify the need to take the igidity index into account. It aeas that on a stiff soil the dislacement that oiginates K D is sufficiently small to be egaded as elastic but in a moe soft soil lastic stains may occu. Limitations to these emiical aoaches ae that (a) the database uon which the coelations have been established is edominantly based on tests caied out on unaged, clean fine to medium, unifom silica sands and (b) most available coelations ae efeed to effective ovebuden stess instead of mean stess and ae theefoe alicable only in nomally consolidated deosits. Given these limitations, the autho advocates the use of coelations with mechanical oeties that ae based on the combination of measuements fom indeendent tests using the cone-essuemete, the seismic cone and the SPT with enegy measuements. D 1 3 qc h0 h 0 10 (50) It is clea fom Figue 30 that even well-contolled calibation chambe tests lead to a consideable scatte fo diffeent soils and soil conditions. Given the scatte obseved in test data a logaithm equation was consideed moe aoiate (Houlsby, 1988): log q c 1.23 D ho ho (51) Values of D estimated fom the above equations can be combined with oeational stess levels to oduce an estimate of eak fiction angles (e.g. Bolton, 1986): m D Q ln R (52) cv whee is the mean effective stess (in kpa), R is an emiical facto found equal 1 as a fist aoximation and Q is a logaithm function of gain comessive stength, known to ange fom about 10 fo silica sand to 7 fo calcaeous sand (e.g. Randolh et al, 2004). An altenative to essuemete and cone tests, the dilatomete is egaded as a useful mean of edicting the intenal fiction angle (e.g. Machetti, 1997). The basic emiical aoach elies on the measued K D fom the DMT and equies an indeendent ough evaluation of K o following the inciles advocated by Dugunoglu & Mitchell (1975). A ecent numeical analysis esented by Yu (2004) ovided an instuctive divegence in the way a DMT should be inteeted in sand by demonstating that although K D (o a nomalised K D /K o dilatomete hoizontal index) incease with soil fiction angle the influence of the igidity index (G/ o ) is also vey significant. In the numeical analysis the DMT has been modelled as a flat cavity exansion ocess using linea-elastic, Figue 30. Relative density vesus nomalised cone esistance fom calibation chambe tests (afte Houlsby, 1988). fiction angle Machetti (Ko=1) I=100 I= KD/Ko Figue 31. Pediction of fiction angle fom dilatomete tests fo K o=1 (numeical edictions afte Yu, 2004). 25

24 a) Cone essuemete Analysis of the cone essuemete test in clay is achieved by a simle geometic constuction of the cuve to detemine the undained shea stength, the shea modulus and the in situ hoizontal stess (Houlsby and Withes, 1988). Analysis in sand is, howeve, significantly moe comlex (Schnaid & Houlsby, 1992; Nutt & Houlsby, 1992). Exeimental esults of calibation chambe testing of the Fugo cone essuemete evealed that the atio of cone ti esistance (q c ) and to the essuemete limit essue ( L ) coelates well with many soil oeties such as elative density and fiction angle. It is woth noting that both cone esistance and essuemete limit essue ae deendent on the size of the calibation chambe used (Schnaid & Houlsby, 1991), but the atio of these two quantities is elatively unaffected by chambe size, and theefoe coelations established in the laboatoy may be alied diectly to field conditions. Aoximate emiical exessions fo elative density, D (Schnaid and Houlsby, 1992; Nutt & Houlsby, 1992), exessed as a ecentage, ae: D qc h 9.6 L 0 h (53) s 14.7 q ln I s c L 22.7 (56) Although futhe veification of the oosed inteetation method fom field tests is still needed to enhance the confidence of these coelations in engineeing actice, esults shown in Figue 32 fom Ghionna et al (1995) and Robetson et al (2000) suot its validity. The basis of these coelations is that both the cone esistance and the limit essue deend on the combined effects of hoizontal stess and elative density. The combination of equation (50) and (53) give ho as a function of q c and L, exessed as the oot of a quadatic equation. A theoetically sound coelation based on CPTM data has been oosed by Yu, Schnaid & Collins (1996). In the theoetical develoment, the authos have assumed that both the cone esistance q c and the essuemete limit essue L ae stongly elated to the limit essue of sheical (P ls ) and cylindical (P lc ) cavities esectively. The atio of these two quantities can be theefoe estimated by the following equation: ls C2 C3 (1 e ) C1 ( ) ex 4 (1 o o C e lc o ) (54) whee C 1, C 2, C 3, C 4 ae constants calibated against six efeence sands. Solutions fo cavity exansion in an elastic efectly lastic Moh-Coulomb soil have been used to coelate the atio of q c / L to the eak fiction angle of the soil. In addition, the limit essue solutions fo cavity exansion in a stain hadening/softening soil using a state-aamete-based soil model ae used to coelate q c / L to the in situ sand state aamete. This ecognizes the idea that io to the achievement of the citical state, the behaviou of ganula mateials is lagely contolled by the state aamete and that can be diectly coelated with tiaxial fiction angles (Been & Jeffeies, 1985). Figue 32 demonstates that the atio of q c / L is mainly deendent on the initial state aamete of the soil and that it can be conveniently exessed as: q c (55) ln The (lane stain) fiction angle (in degees) is (Yu & Houlsby, 1991): L Figue 32. Theoetical atio of cone esistance to essuemete limit essue and in situ state aamete (afte Yu et al., 1996). b) Seismic cone Since G o and q c ae contolled by void atio, mean stess, comessibility and stuctue and ae theefoe diffeent functions of the same vaiables, it s ossible to anticiate that as a atio these two measuements can be useful in ediction soil oeties. In the eceding discussion, a set of citical state aametes combined to initial state conditions have been used to calculated both G o and q c. Acknowledging that citical state aametes and initial soil state ae in the oot of the so called state aamete,, an obvious aoach is to coelate the G o /q c atio and (afte Schnaid & Yu, 2005). It is imotant to ecall that the state aamete has been successfully coelated with tiaxial fiction angles tc using an emiical elationshi of exonential tye (Been & Jeffeies, 1985): A ex( ) 1 tc cv (57) whee A is a cuve fitting aamete anging fom 0.6 to 0.95 deending on the tye of sand. A theoetical elationshi between G o /q c with can be obtained fom equations 34, 54, 55 and 57. Results ae lotted in Figue 33 fo the six efeence sands and fo a stess inteval between 50 and 500 kpa. Fo a given mean stess, the atio of G o /q c deceases with deceasing (i.e. G o /q c deceases with inceasing elative density). Although the coelation is not vey sensitive to vaiations in sand stength and stiffness, fo a given the G o /q c atio educes with inceasing mean stess. The theoetical database geneated by this aoach can be eesented by the following exession: 26

25 ln a Go ln qc (58) whee =-0.520; =-0.07 and =0.180 ae aveage coefficients obtained fom calibation chambe data. In actice it is ecommended to obtain these coefficients fom site-secific coelations. Pedictions of the state aamete fom the measued G o /q c atio would theefoe equie an indeendent assessment to the in situ hoizontal stess, fo calculating mean stesses in equation 58, which fo a nomally consolidated sand can ely on Jacky s equation. Figue 34. Pedictions of the state aamete fom centifuge tests (modified fom Gaudin, 2005). Figue 33. Theoetical atio of elastic stiffness to cone esistance and in situ state aamete (Schnaid & Yu, 2005). Validation of the oosed method fom a numbe of case studies is necessay befoe this aoach is used with con fidence. The database fom both Fountainebleau centifuge tests and Canlex Site can ovide eliminay assessment to the usefulness of equation 58. The measued centifuge data fo elative densities coesonding to 45.1%, 68.8% and 84.2% ae shown in Figue 34, in which G o /q c is lotted against fo mean stesses of anging fom 30 to 100 kpa. Exeimental data clealy show that G o /q c deceases with deceasing and inceasing mean stess as indicated fom equation 58. Pedicted tends fo = 30kPa and 100 kpa ae also shown in the figue and aea to oduce a easonable aoximation of measued data. Penetation and seismic testing data fom the Canlex site (e.g. Wide et al, 2000) ae vey useful to evaluate the oosed coelation since thee is no othe aea in which the feezing technique has been adoted so extensively to etieve undistubed samles in sand fom which the state aamete was assessed. Results ae esented in Figue 35, in which data fom both Milded Lake and J-it sites ae summaised. Substantial scatte is obseved in this lotted which eflects the actual scatteed data eoted by the authos. The two sets of measued data fall in athe distinct egions in the G o /q c vesus sace, with the J it data falling consistently above the data eoted at Milded Lake as a esult of the diffeent mean in situ stesses at the two locations. The J-it data follow a line having a sloe simila to that edicted fom equation 58 and the state aamete can be edicted with easonable accuacy. Data fom Milded Lake is much moe scatteed and does not show a clea tend of educing G o /q c with educing. Figue 35. Pediction of the state aamete at Canlex Site (Schnaid & Yu, 2005). c) SPT in association to enegy measuements In a seies of ecent ublications, the autho has advocated that SPT esults should be inteeted as a dynamic foce F d fom which soil oeties can be assessed (Odebecht et al, 2004; Schnaid et al, 2004; 2005). One examle to illustate ossible alications is given in Figue 36, in which the F d is diectly elated to the fiction angle by assuming a igid-lastic stessstain elationshi and Vesic s beaing caacity factos (Vesic, 1972). The coelation esults in a simle set of elationshis whee the combined values of N 60 and vo ae diectly elated to values. The oosed aoach deicts the tends obtained fom the database of the United States Bueau of Reclamation (Gibbs & Holtz, 1957), eoduces the coelation oosed by de Mello (1971) and incooates into the analysis the effect of the igidity index. In exloing the ossible alications of SPT test data, it will be imotant to evaluate asects of samle enetation that may contol the actual enetation mechanism and theefoe contol the theoetical assumtions adoted in modelling the 27

26 test. Effects of gain size, soil cushing and samle lugging may have to be bette undestood, but since they violate the continuum mechanics assumtion they may have to be mathematically teated as a system of discete aticles. The Distinct Element Method (DEM) descibed by Cundall & Stack (1979) ovides the theoy fo aticle and discontinuous mechanics. Pimay limitations to the aoach ae its extensive comutational demands, the imossibility to conside the effect of intestitial fluid and the qualitative natue of esults. The method can ovide useful insights in descibing enetation mechanism though, as ecently demonstated by Huang & Ma (1994) and Daniel et al (2004), among othes. Nst = 0,30 / de Mello (1971) = 45 o I = 80 = 45 o I = 40 = 40 o I = 20 = 35 o I = 20 = 25 o I = 10 = 40 o I = 60 = 35 o I = 60 = 30 o I = 40 = 30 o I = 10 = 25 o I = 40 = 45 o = 40 o = 35 o = 30 o = 25 o Tensão Vetical vetical effective efetiva stess, - v vo (kpa) Figue 36. Pediction of intenal fiction angle fom SPT (Schnaid et al, 2004). Figue 37 shows an examle of DEM samle enetation simulations fo diffeent gain size to samle diamete (Daniel et al, 2004). In this examle a two-dimensional, lane stain DEM code demonstates the enetation of a 50.8m (2 ) and 127mm (5 ) samles in a ganula soil. The sueimosed black lines ae ootional in thickness to the magnitude of the inteaticle foces. The analysis demonstates that the equied enetation enegy deceases with laten sacing, desite the fact that the laten dimensions wee ket constant. This eduction in enegy is thought to be elated to the fomation of stess aches in the soil ahead to the samle oening, the effectiveness of which inceases as the laten sace deceases. In addition, the comlex henomenon of samle lugging is edicted wheneve conditions ae aoiate. DEM modelling of dee enetation in sand is equied not only fo studying the mechanism aound the SPT samle, but also aound othe enetometes ushed into the gound such as the cone and dilatomete. 5.5 Concluding emaks The otential use of in situ tests fo engineeing site chaacteisation of ganula deosits has been widely ecognised and may comise identification of aging, cementation and cushability. In many design oblems whee the focus is imaily on defomations athe than stength, soil stiffness can be measued fom seismic techniques and stiffness non-lineaity can be conveniently assessed by measuing the stess-stain esonse of soils by essuemete o late loading tests. Convesely, inteetation of enetation tests is not always staightfowad and may equie a combination of measuements fom indeendent tests such as the atio of the elastic stiffness to ultimate stength (G o /q c, G o /N 60 ), the atio of cone esistance and essuemete limit essue (q c /) and the association of stength and enegy measuements (N 60 and enegy). Scoe fo futue eseach comises a bette undestanding of soil cushability and the effects of soil stuctue (which is eoted in Section 6). 6. BONDED GEOMATERIALS 6.1 Backgound eseach Buland (1990) defined the tem stuctue of a natual soil as consisting of two ats: the fabic that eesents the satial aangement of soil aticles and inte-aticle contacts, and bonding between aticles, which can be ogessively destoyed duing lastic staining (given lace fo the tem destuctuation ). Although most geomateials ae ecognized as being stuctued afte the concetual famewok oosed by Leoueil & Vaughan (1990), the natual stuctue of bonded soils has a dominant effect on thei mechanical esonse since the cohesion/cementation comonent can dominate soil shea stength at engineeing alications involving low stess levels. In fact, oblems such as sloe stability, excavations, shallow foundations and oad avements cannot be addessed without accounting fo a cohesion/cementation comonent in the maintenance of long-tem shea stength in oveconsolidated soils, esidual soil ofiles, weak ocks and soil gound imovement. Backgound investigation on the effects of bonded stuctue is based on laboatoy tests caied out on natual secimens etieved fom the field (e.g. Buland, 1989 and 1990; Leoueil Figue 37. DEM samle enetation simulations (Daniel et al, 2004) 28

27 & Vaughan, 1990; Aiey & Fahey, 1991; Smith et al., 1992; Clayton et al., 1992; Aiey 1993; Petley et al., 1993; Cuccovillo & Coo, 1997 and 1999; Kavvadas et al., 1993; Lagioia & Nova, 1995; Consoli et al., 1998) as well as in atificially cemented secimens made u though the addition to the soil of a cementitious agent (e.g. Duas & Pecke, 1979; Clough et al., 1981; Coo & Atkinson, 1993; Cuccovillo & Coo, 1993; Huang & Aiey, 1993 and 1998; Zhu et al., 1995; Pietto, 1996; Consoli et al., 2000 and 2001; Schnaid et al., 2001; Rotta et al, 2003). Difficulties in testing natual soils ae twofold: distubance to the stuctue that can occu duing the samling ocess and satial vaiability inheent to natual deosits emeging fom both the degee of cementing and the natue of the aticles (e.g. Clayton et al., 1992; Stokoe & Santamaina, 2000). To ovecome these shotcomings, atificially cemented secimens ae fequently used desite the fact that they may not eoduce the deosition ocess and the distinctive stuctue of natual soils. To distinguish featues of behaviou emeging fom bonded stuctue fom those elated to changes in state, constitutive laws conceived fo the unbonded mateial ae modified accodingly to intoduce the bond comonent (e.g. Leoueil & Vaughan, 1990, Gens & Nova,1993; Lagioia & Nova, 1995; Kavvadas & Amoosi, 1998; Rouainia & Mui Wood, 2000). These featues, descibed unde both consolidation and shea, ae outlined in this section. Laboatoy tests caied out by Rotta et al (2003) simulate, in the laboatoy, the fomation of a cemented sedimentay deosit in which cemented bonding occus afte buial and unde geostatic stesses. Isotoic comession tests caied out on atificially cemented secimens ae esented in Figue 38 and ae used to illustate concets associated to bonding: a) in stongly bonded mateials, the zone of elastic behaviou is enlaged in esect to an unbonded soil (e.g. Tatsuoka et al., 1997; Matthews et al., 2000; Cuccovillo & Coo, 1997) and as a consequence the value of G o becomes aticulaly imotant as a bench-mak fo engineeing alications; b) the bonded mateial can be at a highe void atio fo a fixed mean stess as comaed to the unbonded one, as clealy shown in Figue 38 and ecognized by Vaughan (1985) and Leoueil & Vaughan (1990); c) afte imay yield, a consolidation cuve follows a ostyield comession line that tends to convege towads the intinsic comession line of the uncemented soil (Cuccovillo & Coo, 1997; Rotta et al, 2003). The zone between the two descibed yield lines (uncemented and 1% cemented) identifies the stuctue emitted sace defined by Leoueil & Vaughan (1990). It is woth noticing that the onset of lastic yielding is exected to be vey maked in bonded geomateials and theefoe the yield locus denoting the onset of lage scale yielding, Y3 can usually be detemined exeimentally faily ecisely (e.g. Leoueil & Vaughan, 1990; Cuccovillo & Coo, 1997). This is unfotunately not always the case, and thee ae many examles eoting the difficulties in detemining the yield oint (e.g. Baksdale and Blight, 1997; Kavvadas et al, 1993; Cecconi et al, 1998). Take once again the examle of isotoic yielding in an atificially cemented soil cued unde stess illustated in Figue 38 (Rotta et al, 2003). The secimens wee initially much stiffe than the soil in its destuctued state, then becoming gadually softe as the isotoic stess inceases. Thee is no singula oint that can be undisuted egaded as Y3 which may be a esult of a gadual onset of the beakage of the cement bonds Void Ratio : kn/m² Figue 38. Isotoic comession esonse fo the 1% cement content secimens (Rotta et al., 2003) f 2. c f. tan. tan d 3 v d 1 Pimay Yield Point ISO(0) ISO(1) ISO(1) ISO(1) ISO(1) Stess-stain cuves and failue enveloes obtained in dained and undained tiaxial shea tests exhibit the chaacteistic behaviou of natual bonded soils. Results fom a gneiss saolitic soil shown in Figue 39 ae used to illustate these attens of behaviou (Futai et al, 2004): a) bonding imats tensile stength and eal cohesion to geomateials. The cohesive-fictional natue of these mateials should then be chaacteized by a shea fiction angle and a cohesive intecet c ; b) secimens tested unde dained shea show a tansition fom a bittle/dilatant behaviou to a ductile/comessive esonse, as confining stess inceases. Initial stiffness and deviato stess at yield may decease at high confining stesses. Most stess stain cuves fom undained tests dislay eak deviato stesses and tests at low confining stesses dislay negative oe essues. c) factos contolling the shae of the limit state cuve in bonded soils has not yet been fully identified, due to the lack of exeimental data in natual samles. Wheeas eseach suggests that esidual soils and soft ocks exhibit isotoic behaviou with the yield cuve cented to the hydostatic axis (Uiel & Seano, 1973; Sandoni & Macainni, 1981; Leoueil & Vaughan, 1990), data fom Figue 39 clealy indicate highly anisotoy of the gneiss investigated by Futai et al (2005); d) effects of bonding on the stess-stain-volumetic esonse of natual and atificially cemented geomateials evealed that dilation of the intact soil is inhibited by the esence of the cement comonent (e.g. Coo and Atkinson, 1993, Cuccovillo and Coo, 1999; 2000, Schnaid et al, 2001, Mantaas & Schnaid, 2002). In tems of enegy it is suggested that the total wok done by the stesses at the bounday of an element is atly dissiated in fiction and atly in disuting the stuctue of the soil. Alying the incemental enegy atio concet, Rowe (1963) demonstates that: (59) whee 1 is the majo incial stain, v is the volumetic stain (often exessed as dilatancy ate 1-(d v /d 1 )=D) and 1 / 3 =K atio of the incial stesses. Fo a uely fictional soil equation 59 is simly exessed as equation

28 Figue 39. Tiaxial esonse of a natual gneiss (Futai et al., 2004) Figue 40 shows the stess-dilatancy elationshis fo dained tiaxial tests caied out in the atificially cemented samles, in which esults ae lotted in tems of stess atio R vesus dilation comonent D, as well as the atio of R/D lotted against axial stains a. Pio to eak the dilatancy exeienced by the cemented samles at a given stess atio is smalle than that of the econstituted samles. The ate of dilation inceases with inceasing shea stain amlitude in a continuous atten that goes u to eak stesses. Peak states ae accomanied by dilation and lastic stains which develoed afte the soil had yielded and the bonds stated to degade. This ocess stats at vey small stains. 30

29 Stain measuements afte eak ae vey uneliable due to stain localisation, but it aeas that an ultimate state has been eached when the exeimental data cuves down towads the econstituted line and each a value of dilatancy D aoximately equal to 1 in all tests. Mántaas (2000) extended the above discussion to the eview of a lage database of tiaxial and lane stain tests eoted by Lambe & Withman (1979), Confoth (1961), Baden et al (1969), Oda et al (1978), Fonseca (1996) and Cecconi et al (1998). Geneal tends lead to the same basic conclusion that it is necessay to exess soil dilation as a function of both fiction and cohesion in such a way that the intenal enegy absobed by soil aticles is a minimum. e) the eceding discussion also demonstates that as shea ogesses, the amount of bonding will necessaily decease with the develoment of ievesible lastic stains (soil degadation). In ecent yeas, a numbe of constitutive models incooating bonding and destuctuation has been ublished, examles given by Lagioia & Nova (1995), Kavvadas & Amoosi (2000); Rouainia & Mui Wood (2000) and Whelle et al (2003) among othes. Diffeent assumtions ae made in esect to small stain and/o lage stain anisotoy, stess induced anisotoy, associated and non-associated flow ules. Desite the fact that this Reot does not aim at discussing constitutive modeling, it seems aoiate to ecognize featues of behaviou that ae essential to constitutive models in the light of the assumtions illustated eviously in Figue 12, following the famewok intoduced by Gens & Nova (1993). A tyical yield suface on the, q sace is eesented in Figue 41. As the amount of bonding inceases the yield suface must gow towads the ight to account fo the fact that highe mean stesses can be alied to the mateial without causing it to yield. In addition, bonding also imats the samle with cohesion and tensile stength that is eflected in the fact that the yield sufaces ae enlaged also towads the left of the stess diagam. Imlementing models with all these featues of behaviou in numeical finite element solutions would oduce sounding inteetation of in situ tests data (in aticula fo cavity exansion analysis), but would distact fom the simlicity of decoding the essuemete cuve to obtain the constitutive aametes of the soil. Mantaas & Schnaid (2002) and Schnaid & Mantaas (2003) intoduced a simlified altenative model caable of eoducing bonding and destucutation unde a numbe of simlified conditions. Lets fist conside esults fom dained tiaxial tests esented in Figue 42 in ode to illustate that inceasing sheaing stain [ = 2 / 3 ( 1-3 )] oduces a continuous and substantial eduction in cohesion fo tests caied out in an atificially cemented soils (Schnaid et al, 2001). Results ae esented in tems of both nomalized stesses and inteaticle cohesion intecet lotted against shea stain amlitude. These obseved exeimental esults suggest that, as a fist aoximation, degadation is attibuted to the eduction in inteaticle cohesion only, and that this eduction could be exessed simly as a hyebole descibed as a function of shea stains and asymtotic to zeo at lage stains, so that: c f = f [(1+) -n ] (60) Figue 41. Yield sufaces in bonded soils (afte Gens & Nova, 1993). In the Moh-Coulomb citeion, the vaiation of shea stength can then be attibuted only to the eduction in the cohesion intecet duing the shea ocess. Figue 43 illustates this idealized concet fo shea degadation on a shea stess vesus nomal stess lot, in which the eduction in cohesion is eesented by a tanslation of the lastic failue enveloes. In a simle stength eduction idealization no degadation is consideed fo shea stesses lowe than the eak shea stength and fo values geate than the eak shea stain,, the obseved degadation atten is asymtotic to zeo and can theefoe be exessed by a simle geneic equation: R K eak 3 = 20 kpa K = 3,39 K eak 3 = 60 kpa R/D K eak 3 = 100 kpa = 20 kpa 3 = 60 kpa 3 = 100 kpa K = 3, D a Figue 40. Stess dilatancy elationshi fo silty sand mixed with 1% Potland cement (K=3.39 coesonds to uely fictional mateial). 31

30 c c o 1 n (61) being: c o = Moh-Coulomb tiaxial eak cohesion intecet; n = degadation index; = eak shea stain calculated using Hooke`s law. In conclusion, the bonded stuctue of soils influences the esults of in situ tests. Since these tests have assumed inceasing imotance in the chaacteisation of bonded geomateials, in aticula in esidual soils, an attemt has been made to develo inteetative methods caable of chaacteising and even measuing some of the oeties of these geomateials. 1 / Botucatu aenite+ 1% of Potland cement 3 = 20 kpa 3 = 60 kpa 3 = 100 kpa shea stain inteaticle cohesion [kpa] Figue 42. Vaiations of cohesion intecet with shea stain in atificially cemented sand (Schnaid & Mantaas, 2003) shea stain soil-ock inteface fo a global geological chaacteisation of weatheing attens. To enhance consistency, ecommendations ae made to encomass geohysical suveys. In this highly vaiable envionment, both laboatoy and in situ test still assist in chaacteising stess-stain and stength oeties on esidual soils. It follows fom the foegoing on ganula mateials that a bonded/cemented stuctue oduces G o /q c and G o /N 60 atios that ae systematically highe than those measued in cohesionless soils. These atios theefoe ovide a useful means of assisting site chaacteisation. Tyical esults fom esidual ofiles ae esented in Figue 44, in which the G o /q c atios ae lotted against nomalized aamete q c1 fo CPT data (afte Schnaid, 1999; Schnaid et al, 2004). The bond stuctue geneates nomalized stiffness values that ae consideably highe than those fo uncemented soils and as a esult the dataoints fo esidual soils fall outside and above the band oosed fo sands by Eslaamizaad & Robetson (1997) and theoetically detemined by Schnaid & Yu (2004). As fo the CPT, SPT N values can also be combined with seismic measuements of G o to assist in the assessment of the esence of a deosit s bonding stuctue and its vaiation with deth. Such a combination is ovided on Figue 45, which lots G o /N 60 vs (N 1 ) 60 in esidual soils, whee (N 1 ) 60 = N 60 ( a / vo ) 0.5 and is analogous to q c1 on Figues 44. Database comise soils fom Bazil and Potugal (Baos, 1997; Schnaid, 1997; Vianna da Fonseca, 2003; Lemos et al, 2004). The bond stuctue is seen to have a maked effect on the behaviou of esidual soils, oducing values of nomalised stiffness (G o /N 60 ) that ae consideably highe than those obseved in fesh cohesionless mateials. A guideline fomulation to comute G o fom SPT tests is given by the following equations: G N 0 a a 60 N 60 vo o 60 G0 a N 1 N 60 (62) whee is a dimensionless numbe that deends on the level of cementation and age as well as the soil comessibility and suction. The small stain stiffness to stength atio embodied within the G o /N 60 tem is seen on Figue 45, at a given (N 1 ) 60 (o elative density), to be geneally aeciably highe fo lateitic soils than that of the saolites, imaily because the latte geneally exhibit highe N 60 (o stength) values. Figue 43. Shea stength degadation (Schnaid & Mantaas, 2003). 6.2 Chaacteisation of bonded geomateials Chaacteisation of bonded geomateials will be mainly efeed to the exeience ecently accumulated in the inteetation of in situ tests in esidual soils. A gound investigation in the so called esidual soils often eveals weatheed ofiles exhibiting high heteogeneity on both vetical and hoizontal diections, comlex stuctual aangements, exectancy of onounced meta-stability due to decomosition and lixiviation ocesses, esence of ock block, bouldes, among othes (e.g. Novais Feeia, 1985; Vagas, 1985). The ocess of in situ weatheing of aent ocks (which ceates esidual soils) gives ise to a ofile containing mateial anging fom intact ocks to comletely weatheed soils. Rock degadation geneally ogesses fom the suface and theefoe thee is nomally a gadation of oeties with no sha boundaies within the ofile. Site investigation camaigns in esidual soils ae geneally imlemented fom a mesh of boeholes associated to eithe SPT o CPT, to deths defined by the caacity of the enetation tool, and followed by continuous otational coing below the Figue 44. Relationshi between G o and q c fo esidual soils (Schnaid et al., 2004). 32

31 Figue 46. Relation between G o and o fo esidual soils (modified fom Gomes Coeia et al., 2004). Figue 45 Coelation between G o and N 60 fo esidual soils (afte Schnaid, 1997). 6.3 Soil stiffness The magnitude of the small stain stiffness in bonded soils is bette undestood in comaison to values detemined fom natual sands. A efeence equation adoted in the comaison is: G o F ( e) S n (63) with units in MPa and a void atio function exessed as: (2.17 e) F( e) 1 e 2 (64) Values of aamete S and n ae given in Table 3 and a diect comaison is shown in Figue 46, having the data fo alluvial sands fom Ishihaa (1982) as efeence. Values of G o divege significantly fom those established fo tansoted soils when they exhibit the same ganulomety but ae uncemented. Paamete S is much highe than the value adoted fo cohesionless soils, wheeas n vaies significantly as a esult of local weatheing conditions as demonstated fom the Caxingui Bazilian Subway database (Baos, 1997). Given the vaiations in both S and n, the need fo site-secific coelations becomes evident. Table 3: Paametes fo esidual soil stiffness Soil S n Refeence Alluvial sands 7.9 to Ishihaa (1982) Viana da Poto saolite 65 to to 0.07 Fonseca ganite (1996) Guada saolite 35 to to 0.35 Rodigues & ganite Caxingui gneiss saolite 60 to ( <100kPa) Lemos (2004) Baos (1997) Consideing the vaiation obseved in natual bonded soils it is efeable to exess coelations in tems of lowe and ue boundaies designed to mach the ange of ecoded G o values, as extensively shown thoughout this Reot. The vaiation of G o with q c can be exessed as (Schnaid et al, 2004): G G G q c q q c c v v v a a a ue bound : cemented lowe bound : cemented ue bound : uncemented lowe bound uncemented (65) The vaiation of G o with N can also be exessed by ue and lowe boundaies, similaly to the cone enetation data (Schnaid et al, 2004): G G G N v a N v a N v a ue bound: cemented lowe bound : cemented ue bound : uncemented lowe bound uncemented (66) Once again it is emhasized that given the consideable scatte obseved fo diffeent soils, coelations such as given in equations (65) and (66) ae only aoximate indicatos of G o and do not elace the need fo in situ shea wave velocity measuements. The eduction in the atio of G/G o with shea stess and shea stain is known to be sensitive to degadation of cementation and stuctue, among seveal othe factos (Tatsuoka et al., 1997). The moduli degadation can be measued in the laboatoy with high esolution sensos ovided that high-quality undistubed samles can be obtained. In contast, Schnaid et al (2004) have demonstated the level of uncetainty associated in back-figuing the degadation cuve fom in situ tests in bonded soils. Take the examle given in Figue 47 intended to illustate the inteetation of non-linea stiffness soil oeties fom a dained Plate Load Test, PLT (afte Schnaid et al, 2004). Tyical (simlified) vaiations of the secant Youngs modulus (E sec ) of a cemented sand is seen to educe fom a high initial value (E o ) of 350 MPa, which evails until a esumed yield stess ( vy ) of 100 kpa is exceeded, to a stiffness comaable to that of an uncemented sand at axial stains in excess of 0.4%. Fom a numeical analysis, these two athe diffeent mateials oduced the non-linea alied stess 33

32 (q a )-settlement (s) esonse of a 300mm diamete (D) late shown in the same figue. It is aaent that, desite the significant diffeences in stiffness, the cuves ae almost linea without evealing the significant softening that may be exected on insection of Figue 47a when the alied stesses exceeded vy. Indeendent attemts have been made to back-analyse the load-settlement cuves in ode to edict the inut degadation cuves of both cemented and uncemented mateials. Deating fom a given G o, enginees wee able to match the behaviou of the uncemented soils but wee unable to identify the cement comonent of the bonded soil (as seen in the dashed aea in Figue 47a). 6.4 Shea stength Since the natual stuctue of bonded soils has a dominant effect on thei mechanical esonse and the bonded stength is ecognized as a net sign of this stuctue, identification of the two comonents of stength is cucial in geotechnical design oblems. Use of tiaxial tests on high quality samles to quantify the cohesive-fictional aametes of bonded soil is always advisable, but given the acknowledged difficulties of maintaining the elic stuctue duing samling, which masqueades bonding effects on stiffness and stength, in situ tests emain as a viable otion in engineeing actice. Penetation tests such as SPT, CPT and DMT, extensively used in tansoted soils, ae also commonly adoted in the investigation of stuctued cemented deosits, with emiical coelations fo assessing soil oeties being locally adated to meet standads that eflect egional engineeing actice. A limited gound investigation based on these enetation tools will not oduce the necessay database fo any ational assessment of soil oeties, fo the simle eason that two stength aametes cannot be deived (, c ) fom a single measuement (q c o N 60 ). Limited investigations ae, howeve, often the efeed otion. In such cases involving cohesive fictional soils, enginees tend to (consevatively) ignoe the c comonent of stength and coelate the in situ test aametes with the intenal fiction angle. Aveage c values may be late assessed fom evious exeience and back-analyses of field efomance. Since the actice of assessing aametes fom limit gound investigation and evious exeience aeas to be widesead, the autho ecommended the essuemete as a means of insecting the accuateness of a given set of design aametes (Schnaid et al, 2004). All the theoies fo the inteetation of the essuemete in bonded soils make use of the in situ hoizontal stess, soil stiffness and stength aametes: angle of intenal fiction, angle of dilation and cohesion intecet (which educes with destuctuation at high shea stains). The essue exansion cuve eesents theefoe a combination of all these aametes that cannot be assessed indeendently. Given this comlexity, instead of attemting to deive a set of aametes fom a single test in esidual soils the essuemete should be viewed as a tial bounday value oblem against which a theoetical essue-exansion cuve edicted using a set of indeendently measued aametes is comaed to field essuemete tests. A good comaison between the measued and edicted cuves gives eassuance to the ocess of selecting design aametes, wheeas a oo comaison indicates that one o moe of the constitutive aametes ae unealistic. Adequacy of the method lagely deends uon the constitutive model adoted to eesent soil behaviou which, fo a cohesive-fictional mateial, is comlicated by a numbe of factos such as the influence of bonding on the stess-dilatancy esonse of soils and the effects of destuctuation. Ideally the c and should be couled to stiffness, dilatancy and mean stess level and fo that eason the cylindical cavity exansion analysis develoed by Mantaas & Schnaid (2002) and Schnaid & Mantaas (2003) is ecommended. The concet intoduced by Rowe (1963) that lastic dilatancy is inhibited by the esence of soil bonding, discussed eviously in this section, was investigated and used to descibe the lastic comonents of the tangential and adial incements in an exanding cavity. A new solution is fomulated within the famewok of non-associated lasticity in which the Eule Method is alied to solve simultaneously two difeential equations that leads to the continuous vaiations of stains, stesses and volume changes oduced by cavity exansion. Fo the loading hase of cavity exansion the nonassociated flow ule can be exessed as whee whee and ae the lastic comonents of the tangential and adial stain incements. The geneal definition oosed by Rowe s law based on the hyothesis of minimum absolute enegy incement duing shea was eviously exessed in equation 59. With the aim at genealising Rowe s law, not only by consideing both the cohesive and fictional comonents but also allowing fo degadation of cohesion duing shea, it is ossible to combine equations 59 and 61: 1400 Alied stess (kpa) Cemented sand Uncemented sand Plate load test (300m m late) vi =20 kpa, hi =10 kpa s/d (%) Figue 47. Analysis and back-analysis of late loading tests fo cemented and uncemented sands (modified fom Schnaid et al, 2004) 34