Proceedings SF the W4 Zyperateael Co antl PPintlaton Engmecrny, 4 on 20 Mechanes SA, CA Hag s2-1t, (ES. bh L AW. Palleewa Ly Gestey, ag. New developments in field and laboratory testing of soils Nouveaux développements des essais in-situ et de laboratoire M, JAMIOLKOWSKI, Technical University of Turin, Italy cc. LADD, J.T. GERMAINE, Massachussets Institute R. LANCELLOTTA, Technical SYNOPSIS A reliable assessment of the s Ravural soi] deposits by means of any rational geotechnical analysis and design. Eelevant topics for detailed discussion. In Chapter Story, yielding, normalization, anisotropy, infive: Cohesive soils, tine effect in’oedometer testing. advances in the use of in situ techniques in: soil situhorizontal stres Finally Chapter 4 is devoted to some general aspect: crocomputers in geotechnical experinentation. 1 INTRODUCTION OBJECT AND SCOPE the present theme ‘ide spectrum of Of Experimental Soil Engineering. The aim of Branch of Geotechnical Engineering 1s to obtain ~ through experiments performed both in the field ang. the laboratory "the following basic design information 1. Detailed and representative soil profiles, in- Giuding the description of ground water condi~ tions and soil index properties. Assessment of the initial state of stress ‘ating in the ground and the stress history of the sof] prior to any construction activity. 3. Determination of the stress-strain-tine and Strength characteristics of the encountered soil layers. ‘the first two points include the initial physical and geometrical conditions of the foundation Soils, and it i convenient to group them under a Common heading called Initial State Variables (ISV). All the above Gesign information may be obtained by meane of a properly planned and executed explo ration programme, the scope of which should be related to the complexity of the local geology and the structure to be built. Successful exploration prograames are generally Tun in stages, The later and most conclusive sta— Ges are Linked to the soil model which one intends foruse in the design analyses through an appro xi ‘+ Experimental Soil Engineering ‘experimental technique: Within thie context the writers Tn Chapter 3 the discussion focus , ‘Stetss history: deformation 2 Massachussets institute of Technology, Cambridge, Mass., USA of Technology, Cambridge, Mass.. USA University of Turin, Italy train-tine and strength characteristics of fepresents the most challenging ta lected eres: 2, laboratory testing, they incl srofiling and identification, a Su and consolidation characteristics of soil deposits. is of the use of data acquisition systems and mi priate determination of strese-strain-tinc-strength Gharacteristics. 1.2. EXPERIMENTAL SOIL MECHANICS ‘The methods of ESE*uged in geotechnical explora- tion may broadly be grouped as follows: Laboratory testing, including physical models [essa ‘et al. (1977), Saada and Townsend (1981), jchofiela (1980)} 2, tn situ testing [Lada et al. (1977), ‘utehell et al. (1378), Yori (1981), Wroth (1984)). 3. Monitoring the field behaviour of full~ prototypes and/or existing structures by mea Sf Geotechnical Instrumentation which, when pr: pe rly back-analyzed, enables in many Cases the jessnent of the pertinent stress-strain-tine strength characteristics of the soil (Di Bi gio (1975), Dunnicliff (1982), Hanna (1985)]. Obviously, the number of investigation tools and Gpparetus belonging to each of the above thre Groups is almost infinite; a completa review of them is hardly feasible and vill not be attempted in the present paper. The intention of the writers is rather to examine in detail a few topics se- Jected on the basis of the following conside= rations: 1, Priority 1s given to topics in which the need dor research has recently been recognized by the geotechnical community [RNESE (1983), Site Cha Zacterization and Exploration (1978). 2. Recently developed innovative exploration and ‘interpretation methods receive due consideration. Finally, the discussion should be within the » Context of current research activities at the Universities to which the writers belong. a 57Discussion of these topics vill be made after a brief exanination of the following aspect: = levels of soil behaviour investigations, = categories of soil models in relation to expe- Finents, and = laboratory tests versus in situ tests and field measurenents. In the writer's opinion, these aspects are clo- sely interrelated and a’ proper vision of then is necessary for any soil mechanics experiment to be scientifically valid. 1,3, LEVELS OF SOIL BEHAVIOUR INVESTIGATIONS Generally speaking, the behaviour of soils canbe studied at three disferent levels: micro, macro and nega. [bafalias and Lade in RNESE (1963)]> wheres 1, the microwlevel is the level at which the in- teraction between single soil particles and clu sters of particles are investigated and lavs e established which relate the evolution of soil structure and micro-fabric to changes of the imposed stresses and strains in the sot nase [see for example oda (1972), Cungall et al. (1962), Nenat Nasser (1962), ete] 2. The macro-level is the level at which quanti- fies such as stress and strain are usually defi ned and measured by so-called single elenent 1a boratory tests like triaxial, plane strain or simple shear tests. At this level, one also for mulates constitutive equations for the soil, [see for example Roscoe and Burland (1968), Schofield and Wroth (1968), Lade (1977), Prevost. (1980), Dafaiias and Herrmann (198) and many others}. 3. The mega-level refers to an entire geotechni- eal structure, including the surrounding soil mags, the behaviour of which may be analyzed by solving the boundary value problem using macro- constitutive equations. A typical example of Soil mass mega-element study is the beck-analy Sis of a case history, ‘The classification of in situ tests within the scope of these three levels is not straigntfor- ward. As far as methods for in situ test inter- pretation are concerned, they should be assigned $0 the mega-level since they represent boundary value problens. However, the question of the vo- lune of the soil to be investigated falls between 1.4. ANALYTICAL PROCEDURES AND SOIL MODELS The analysis of geotechnical problems usually re quires the choice of an appropriate computational iethod which is strictly linked to the adopted soil constitutive model. The model, in turn, is related to a specific set of soil paraneters to be determined through an appropriate exploration Progranme. This means that there mist be a direct Connection between the method of analysis which the designer intends to use and the type and extent of the proposed soil exploration: Following the RNESE (1983), the presently availa. ble analytical methods may be grouped as shown in abl 1.5. LABORATORY VERSUS FIELD MEASUREMENTS Conventional exploration progrannes make use of both in situ and laboratory measurements for the determination of soil properties, the latter being based on the testing of so-called undistur bed samples obtained from borings and exploration pits. It is quite comon to find that geotechnical engineers declare themselves in favour of one of these basic exploration methods, without conside- Fing that the suitability of one approach or the other is strictly linked to the aims of the explo ration and the method of analysis which is inten: ded to be used in the design. In fact, both me: thods of soil investigation have their inherent merits and disadvantages which are briefly summa- rized here: LABORATORY TESTS 1, Advantag = Well-defined boundary conditions. = Strictly controlled drainage conditions. ~ Preselected and well-defined stress paths are followed during the ests. - In principle, uniform strain fields (this as- sumption is acceptable for small strain levels only and soils vhich do not exhibit strain softening behaviour) are imposed on the speci- mens, which allows the application of continuum mechanics theories to the interpretation of test results. ~ Soil nature and physical features are positive- ly identified. 2. Limttations: = In cohesive soils, the effects of unavoidable sample disturbance in even so-called "high qua- re eceoane ea leaaia sareLee lity" undisturbed samples are sometimes diffi- cult to assess. TABLE I Categories of Analytical Methods for Soil Studie: ‘Category Main Features of Models Determination of Soil Parameters r Very advanced models using nonlinear Only from sophisticated laboratory tests, elastic plastic time dependent laws which | with the exception of the IsV which must’ be Possibly incorporate anisotropic behaviour| obtained fron in situ testa a ‘Advanced models using constitutive incre- | Laboratory tests which are only a little nore Mental elasto-plastic laws and nonlinear | sophisticated than conventional tests. In sit elastic relationships tests are also appropriate, particularly as far as the ISV are concerned —t Er ‘Simple continuum, such as isotropic ela- | Conventional laboratory and in situ teats stic continuum, including layering and empirical models= In cohesionless soils, undisturbed sampling is still an unsolved problen in everyday practice. = The snail volume of laboratory specimens [Rowe (18721 cannot incorporate the frequently pre~ Sent macrofabric and inhomogeneities of natu- Fal soil deposits. This leads to doubts as to Ghat extent the field behaviour of a large soil Rass can be successfully modelled by snall sca Te laboratory tests. = the factore causing the formation of shear pla hes during the testing of laboratory specimens Bre still very poorly understood. Shear planes are frequently associated with such phenomena Qs induced shear stresses, soil volune changes, nonuniformity of laboratory specimens and con Bequent nonuniform strain distributions, boun- Gary and Kinematic constraints, and stress con centrations imposed by the laboratory appar: Sus [RNESE (1963) I must be emphasized that once a shear plane has developed in a laborata fy specimen, deformations are concentrated along this plane and displacements and stres neasured at the specimen boundaries are cof Guently no longer a function of the stress: ferain behaviour of the tested material. - In prineiple, the discontinuous nature of in- formation obtained fron laboratory tests may lead to erroneous modelling of the behaviour of a large soil mass. ~ Ih general terms, soil explorations based on the laboratory testing of soil samples from borings are Likely to be more expensive and time-consuming than explorations which make use of in-situ testing techniques. A special kind of laboratory test is represented by Generifuge models. fed scaling Tiws make ic possible to obtain reasonable simi- Litities between test model and prototype Structure [Rowe (1975), Schofield (1980, 1983), Scott, (1983), Centrifuge models are particularly Sseful in ettdying the Behaviour of complex geo- technical structures where the node of deforna- tion and near-failure kinematics are unknown. ‘Through appropriate paranctric studies, centri- Tuge esting permite the verification of exi- Sting computation nethods and the improvenent of funeficel models where geometry and loading con- Gitions are complex. No'furener attention will be devoted to this in- portant and extremely promising physical nodel~ Bing technique, since the present conference pro vides for a special discussion session on centri Eage testing. IN SITU TESTS 1. Advantages: A larger volune of soil is tested than is usual jy done in most laboratory tests; hence in situ tests should in principle reflect more acura tely the influence of the macrofabric on the jured soil characteristics. - Many devices currently in use produce a conti~ hus record of the soil profile which allows the soil macrofabric and layer boundaries to be determined. = Tn situ tests can be carried out in soil depo- Site in which undisturbed sampling is stil) {ble ot unreliable. Examples include co- hebionless soils, soils with bighly-developed Racrofabrics, intensively layered and/or heterogeneous soils, and highly fissured cle = The sofis are tested in their natural environ- ment which may not be preserved in laboratory ts. For example, the most successful attempts to measure the existing initial total in sity horizontal stress are the recent developnents in situ techniques, e.g., self-boring pressure hecer (SBP), flat Ailatoneter (DMT), Towa Stepped blade (188), spade-like total stress eelis (TSC). = {n general terms, soil exploration by means of in situ techniques is more economical and less time consuming than investigations based on 1a boratory tests. 2. Limitations: = Boundary conditions in terns of stresses and/or Strains are, with the possible exception of the Self-boring pressureneter (SBP), poorly defined, anda rational interpretation of in situ tests is very difficult. = Drainage conditions during the tests are gene- Pally unknown and make it uncertain if the de- Fived soil characteristics reflect undrained, Grained or partially drained behaviour. In this fespect, quasistatic cone penetration tests with pore pressure measurements (CPTU) and SBP Tests (aleo with pore pressure measurenents) » when properly programmed, help to minimize the problen. = The degree of Aisturbance caused by advancing the device in the ground and its influence on the test results is generally (with the possible Gxception of the SBP) large but of unknown ma~ gnitude. = Hodes of deformation and failure imposed on the Surrounding soil are generally different from those of civil engineering structures; further more, they are frequently not well established, as for example in the field vane (FV) test. = The strain fields are nonuniform and strain ra- tes are higher than those applied in laboratory tests or those which are anticipated in the foundation on structure = With the exception of the Standard Penetration Test (SP7), the nature of the tested soil is hot directly identified by in situ tests. ‘The Limitations of in situ tests lead to a situa~ tion in wnich almost all present interpretation techniques are empirical, with ‘the exception of the methods for the SBP test. BACK-ANALYSES OF FULL SCALE PROTOTYPES the rapid grovth of geo technical instrumentation and les widespread ai Plication hae made available a nunber of well-do- Cunented case records of the field behaviour be~ fore and at failure of a wide variety of geo- fechnical construction and full scale prototyp The inportance of the publication of well-docunen ted case records led the University of Missouri~ Rolla in 1988 to organize a special conference in St. Louis on Case Histories In Geotechnical Engi- Reering. An analysis of the more than 150 papers presented to that conference confirms that nea- Bured field performance data are used in the two ways postulated by Lanbe (1973) 1. As controlled boundary value problems, to ve~ Eify a specific constitutive soil model’ and the Eelated computation procedure. In this case, all input parameters are determined fron appropriate laboratory tests which are supposed to represent the stress-strain-tine behaviour and strength characteristics of the soils underneath the structure being monitored. 2. As a set of experimental data such as pore pressures, vertical and horizontal displacenent: 59vertical strains, etc. which make it possible, within the frane of the adopted model of soil behaviour, to assess the relevant soil parame- Of these tvo approaches, only the second one is pertinent to the scope of the present thene lecture. The writers have no doubt that since the publi- cation of the classic works by Terzaghi (1950), Skenpton and Brown (1961), Skempton and La Ro~ chelie (1965), Peck (1969), Bjerrum (1972), Lan- be (1973), this is the procedure which, in’prin- ciple, leads to the most reliable assessment of Geotechnical design paraneters (1f compared only fo what may be obtained fron the results of labo ratory and in situ tests). However, as often-happens in practice, the self- evident may not be sufficient and reliable soil parameters may not be guaranteed by a back-analy Ses of field records alone. In fact, this type Of procedure often requires many simplifying as- sumptions which present the designer with fre quent pitfalls, as has been illustrated by Le- Foueil and Tavenas (1981). Generally involved is one or more of the following 1, Soil constitutive model and related computatio hal procedure. 2, Drainage conditions and boundaries. 3, Initial state variables of the considered pro- ‘biem, including soil stratigraphy, ground water conditions, initial stress state and stress hi- story of the soil deposit, etc. 4, Auxiliary soi characteristics which frequen— tay must be introduced in a back-analysis. This means that the examined problem is controlled by a larger number of parameters than the ana- lysis itself can yield. The sensitivity of back-analyzed soil characters, stics 1s not alvays evaluated with respect to a given set of assumptions.in addition, the desi- Qner must fully understand that a substantial between soil paraneters deri- ved from a back-analysis of field data which re~ Flects the performance of a single point (e.g stress, strain, pore pressure, etc.) and those obtained from the analysis of average foundation Conditions (e.g. load-displacenent curves, etc.). Te must be borne inmind that only the forner type of back-analyses of existing case records us to obtain soil parameters which may be ‘inferred from labo- ratory and in situ tests. ‘The above brief summary of the limitations and ‘Of the three basic methods of BSE ly indicates that their capability for = po ‘aitive and comprehensive identification of the engineering properties of foundation soils is complementary. To support this statement, the current capabilities of the three methods are described in Tables II through IV. 1.6. SUMMARY AND SELECTED TOPICS 1, Only laboratory tests make it possible to in Vestigate such complex features of soil bene viour as initial and evolving anisotropy of both strength and deformability, the effects of streas and/or strain reversal, tine effects, the influence of intermediate principal stre: etc. Therefore, only those tests which examine the soil at a macrolevel allow the determination of parameters which characterize realistic con stitutive soil models. 2, The role of in situ tests in experimental soit hechanice consists mainly in the assessment of; S“Initial state variables and particularly de~ tailed soil profiles, but also the in situ i- Aitial lateral stress, Oyo? in this respect, the linitations of laboratory tests are parti cularly pronounced. = Drained and undrained soil stiffness, especial ly at small and moderate strain levels. The — determination of this paraneter in the labora~ tory ie an almost impossible task for cohesion less soils and often of questionable reliabili ty for cohesive materials due to the well known sensitivity of these parameters to even snall sample disturbance. Plow and consolidation properties, especially 4] inicohestonless deposies’ and cohesive 041s with a well-developed macro-fabric. The unre- ) liability of laboratory tests in these condi- tions is widely recognized. 3. The monitoring of the field behaviour of con- ‘Struction and full-scale prototypes makes it possible, if properly planned, controlled and analyzed to verify constitutive soil models and lytical and numerical methods, and ‘a megalevel, relevant soil cha- racteristics like strength and deformability, by by-passing most of the limitations inherent in laboratory and in situ tests. 4. In all circumstances, the three methods of ESE ‘mentioned above are complementary, and the rela- tive importance of each is strictly Linked to 19 cal geology, the type of construction, and the — menthod of analysis which one intends to adopt, in a specific design. In relation to the above and taking account of the priority research needs in ESE as identified Guring the USA NSF Workshop on Research Needs in Experimental Soil Engineering (1983), the fol- joting topics have been selected for’ coverage jereint Laboratory Testing Preconsolidation pressure, yielding and nor- alization. 3. Strength deformation characteristics under generalized states of stress. 4. Time effects. In Situ Testing ‘The assessment of the initial state variable: with particular reference to the initial in situ lateral stress and soil profiling. ‘The evaluation of soil stiffness with parti- cular reference to deformability parameters under drained conditions. 3. A eritical review of in situ methods used for the evaluation of flow and consolidation characteristics. Instrumentation Due to the already wide spectrum of topics of this theme Lecture and due to the obvious space limita- tions, the writers have decided not to cover in detail topics related to geotechnical instrument: tion, with the exception of data acquisition sy- stems and micromputer applications in geotechni-TABLE IT jes for Soil Modelling Present Laboratory Capabilit. Bold behaviours Tauipeant and/or Procedures Tina Trip Va, OR TE and OF cegea,| {Joes Restricted to cohesive voile jatGet*isth atlovence for ie | \'"p, applicable only to mechanical overconsoligation aetuning wainiy OD eonte Trica Requires good quility cohesive undisturbed samples 31g] ty euteeate bo latinsin aon autora oereonteises Ths fave Hi tere wih Give ‘Gccanaive data for sande yielding conflicting Femults £9 ilosibis ane'tinee Boundaries {S,tepi yeep) values ene wich dugterent.in~ | 2-2. Rather Linited ata for clays ri'gs’and external Pe “ Taveaal aptpouropy Tots Questionante valigity ath respect t2 pd SEbengeh ene sesEine 2 acres 4.2, Lintead te 690" rotation of princts ‘la, proven" for cD and CD teats on sande and CU testa on clays Ca! can explore full range of of ¢ #90" Un theory? ic clays: further development needed £01 ALD, strovent for sande aeel ace, leh te mar anes es Sopa wipe [es baw eee a nian o prinipel Hastie 60 Fare iolatict of” [50. ro cone Ta REEL ata peeeeat i ae an ee eeenie the Lite of plane stain conditions Intensive deta for induced anisotropy in sends available Gojca, Exporinepeel protien due, co eeubrane igaraye, friction on cha" spechacea boundifiee and tempersvare fuuctuarsons citibr tack of experimental data for varying stress conditions feces SSIS SE continue [Teic vow praceicnl Linitaw | TeTvas Tnporeent f0r danse sandy stst® clays and cenested 20108 sieved 200 conventions Tih, mere tava Lusitation 1m We DSC) che applied heer etre efangeh and stile Sasotatory eevapasne lsd ME tteS HE "for clays and laboratory Feconstivated sandey Limited Tr Seaticateein,(geto-38) [Ot TAG, BNC Cente rarer nent agaulue and dncer~ sania aveliunie for wnddscarbed sand esas eed aurcb. in clays the renulte are very seneitive to even emai sample ‘alatisbeace titre, majority of the available date obtained on {s0troptcaily com- soiléated specinens sa, vary luge how about infuence of sntial aactrory en axioun past vertical effective stress cock + ovarconsolidation ratio Ondoneter apparsts x + Teuacial apparatus Darect simple anear apparatus {Teuc = Torsional shear Dollow cylinder apparatus Torsional resonant coluan apparatus bse. = Directional shear cell Plane strain apparatus = Consolidated undrained CConvoliaated drained H+ Coeffictant of earth presrure at rest cal engineering (Chapter 4). The writers believe yavally require at least some design paranetere Chat most readers will realize the relevance of from laboratory ng. shat gone rensereiactions calibration, installa- 1M planning any laboratory Progra. St 202000 oy PEPE see gn eee nse geotechnical anstru- recognized that the in situ undrained and drains mentation. Behaviour of a clay is largely controlled by the Ragnitude of the applied stresses in relation to the "yield envelope” for the soil. Section 2-2. Miscusses the practical aspects of this concept, pecially regarding the influence of stress hi- 2, LABORATORY TESTING Story and clay sensitivity, Section 2.3. deals With procedures for assessing and minimizing Sample disturbance in cohesive soils. The topic eee Of the measurement and influence of stress-strain tn this section the vriters' focus is on equip- strength anisotropy is treated in Section 7-4, the ent and procedures for testing cohesive soil: final section discusses “tine effects” during ment and prt practical problems involving clays consolidation. 61Present In Situ Capabilities for Soil Modelling TABLE 111 Eyuipeent Commence ~ Rear 7 "Great potential Zor" sot profiting nd iaehtiricatione ss Teo" MAE glib, EMARELALLy rigid and extremely veld denied 4 ‘quick Ferponse foe" feliatle ges ecburanenee 1:2.4, Good for agit profiling but lene sensitive to strata changes in gf Elet ion eatio fy/Ge # poor te2se, Potential aay resolution of ge measure fs rellable and Fepeatabie’fy secsurecente 3ot-gfat temaive 261 sagncitier but, since peforned diecontinuc- Aska MYPIY foe sot protiting and saameitications needs further etmacen nondinangiona electricth fornation tater” ich na'Comentation? hay be relevant for Liquefaction steaie th a very 3:2 type identifier in expectatiy 7 i sits a, ence Ky 2.2, Oa (asses 2.3, 0 (anseseas 46) co tahel 2:3, wyranite,teactaring, TEES BEpgMSRERT get cares lass exerisnce Oe Greatest potential anong in situ aachods but titi some pro- SiSt'Sefne cnpltance tnd probe Tenet oredr rtenenreh Bo assees Fatianiingy Tomsnines Bet seizes 2.3.4, Maw device: requires farther intanaive laboratory and in atta ‘tui iae in Other soils untixely "7 *” "Ft $e #64 yer ces 2.422, In st4tt clay overestiantes opt requires correction for wots having 1 aefective yield” Possinie applications limited to relatively honogensovs cohesion inte aepontey “eal japehs in which teste are performed under S71 the influence of plata shape and ciaarnance due to its indeirTstion on’ the locavaceelanent eelationsnip not well uncaeseoet 2. ar teste 45, thane wave valoctey Application Limited to shatiow epthe Proven" sn cohesioniess deposits in veich can datarnine verse riled’ Young stitinens Ey wiesan the onpeh of uathoanee ef the plate jc, I soneaive sole, deapice uncertainty above drainage conditions, “GUTS Mhacaed"to yteia average angreaned Today seittaese Sy 4.14, since £10 obtained from losd-diolacenant nanaurements, an a ‘priott"ateumpeien regarding so1l constitutive model ia neceeeasy LoMOET dbtusats to rater ene © ebtined fron and SAL conte to inves sienenty hence 4,24, crane potential for Girect anscament of shear sodulue Oy in horizontal direction . ali untoading-relonding cycles whose role is to miniaise the oil ‘iatarbance gue to probe insertion 42.4; Trough empirical correlations yielde values of tangent con ‘stihined noauise’in sande and clays” 2g, Teannety atten corzelations hve bean obtain mainly tor inantly gusset sande and aarinn and sllvial clayey taste Sabor Fttory anc field valiaacion in e wider range of sotis hecdede apirical correlations between qy and F of questionable relia~ btitty"ahd' not generally valid excape for ac sand’ 38 any cane applicable only co predoainanely quarts clesn un- Cieticed SthaS*EE Elsi peneeruelon’oteuse andor fully Sreinad coe Sa, “Proven” potential to avaluate ama str Wottuoneally Lapered soil saponin 43d; Te value of @ ts calculated after steumption are aude con seehing aa SacethcelSaSeGt SS "GRS teDeT peta cna tse SST noe 62TABLE IIT CONT. Zan povienrFarnat [ Taipan arora er rane Taree SICRRTIS GAH Gee el reerred Teron Oe ti aan chia hier reteeiy cmpier ssa. taraewoste pumping | Sou, egy reasaie te sie, rerygepnai cass emerge wll i LET MUA tase TOL RULE pole te teu sede aaterin mabe, henee anaes se gona sort dace rer ieeg requarayauttieste interpretation due 80 nme enti, 2 ete ees ET ST a sea geons ot 1retey sens neem gas eid en reeeee rae saa] str toraratnsen ated sential earn ore prnmae 0 80 ful aa bee sslaptth ater nc EStto a Sl seesiiite, mera sc Lmaraugese tat eyateoner sepuetietysoen an etersie in atta fe Tow repo oaae Renin eo haat arewent « neereoe veces ot TABLE IV Possible Current Contributions of Field Instrunentation to Advanced Soil Modelling Foil benaviourPeranetar]| Byuipeent and/or Proceduren ‘Commenta = banat Tr operational serength | 7,3, bateral, soi deformation Nasaeured betov enoenenante. EiTinTsuopess wereieal it-_| "yr, propably the moet reliable vay to assene she average oparttionsl itngretarycubeed acne OUGDLY SP ATopsTand"ChBanheaca'on clays wnich neve fatbed SLOTE'SaEing capacity | SAE &* sense iid Spe enalare sire, Apelor! aamumpetone concerning failure Kineaatica: « constito- LIS eset for the soit te newsee 3 Beforaapansey TiThsswapeamt GF vertical |2,1,n, with at Tease ovo, yalbes of oe, ngumured st ovo elevations 22 Sharactersetic i sMatamtameasaet certict? | ?hctuate'tafees "tne wartioel strain ands conmeguanehy, te average mte,tt SS | SfTeha toll 'becween the two points may be aseeneed ‘Sndar loaded area |211,y, aye distribution of che induced total stregane mont be sateseeds ibs Be AAEE ect Sn regarding she conseitucive Behaviour of a rrey in order to refer the anteseed Eco know drainage, conditions, ida inbtaDTAES BrStatie in the consisered Layer ancula be monitored Lunived research valve gKaetanne an Tae” [3°2°. Gatde am auras within ee dnt of intlonen of te Lote 2g tamarmen of guctge | 22s. vor prcistnof tertet presser ne property “quality” he astensed stiffness, che octet osFes" ana’ Latarel soLl aioplacenants Deiow the Lomded Sten mould be known ana. a prior! sanaption regarding the aoil constitutive model 44 pe TE 18 almoar iapoesible to Link the obtained E to, the stress or Wapfavencatave macro-sienent of $04) “Tae coutTiciene consolidation aay be qutinated for vertical flow 3 Gensohiaatton 3: of, excess pore characteriacies Decay of excnat 2ortse | ievel_ invese ancence of eerticnl arsine oF for horisoncal flow (evn) Sn plesonstere [SrtuetSeetence of vertical araine june & priori a aotl constitutive so- Be afs'tsatecas cyy and evn 3.4, weanurenane of op at 242-0; A8 mentioned under point 2:10 2; Manmurenarsof,24,1, |?ctzhina'sn'the monitored Layer aay be entiaated and, hanes RIE ales Yoaded ares on may be interred a.}rb, Tor a proper interpretation of the atrain evolution vith tint, ae cstSEat BES pEetnune snd lateral soll eforastione beneath the {otaed' tres anoule be sonttored Hori anzunpeton concerning the soit conststative buhaviowt 632.2, PRECONSOLIDATION PRESSURE, YIELDING AND NORMALIZATION 2.2.1, Introduction Casagrande (1936) defined the preconsolidation pressure, oj, as the “largest overburden beneath Which the sil had once been consolidated". He also presented an empirical technique to estimate the value of of from the shape of the void ratio versus logarithm vertical consolidation stress Curve obtained from incremental one-dimensional consolidation (oedoneter) tests performed on un- Giscurbed samples. During the 1960's it becane & vident that the Casagrande of profile for sone Clay deposits was higher than the "maximum past Pressure" (maximum vertical consolidation stress) Believed to have existed during geological histo ry. This discrepancy was attributed to various joi] structure” phenomenon, such as long term ‘ondary compression, aging and cementation ef jects [e.g. Leonards and Altschaeffl (1964), Bjerrum (1567)|. Although special terms were sug- gested to designate the: ieds aifficule fo establish for most clay deposits exactly when they would apply. Hence, while "preconsolidation pressure" may often be a misleading term in the physical sense, it remains the accepted term to Genote the “break” in the oedoneter curve. As large strains accompanied by plastic (irrecove- mensional com pression. The writers will use of = preconsolida- fion pressure in this context irfespective of the physical cau Although the profession can argue over terminolo- Gy, we should all accept the view that an accu- Fate determination of the of profile is generally ‘the single most important step in predicting long term consolidation settlements. The writers be- Lieve that it is equally important for most short term stability problems. Hence we will first review procedures for measurenent of of and the different mechanisms responsible for develo- ping the in situ of profile. This is followed by Sn overview of the’ concept of a yield envelope, ite practical implications and how such enveio, can be affected by the of mechanism. Finally, Will emphasize the usefulness of normalizing un- Grained stress-strain-strength parameters with Stapece tovoge indoing this, the term stceas hi BteEy"ie ceed co" denots the in site effective o- ‘Jerolicden ‘strese, sper and’ preconsolidation ‘prex Serer obs profiles: “Bias cere hes ehe sane’ prom EiSas a8 already discussed for ag, a5 Goes the o- SerconseLidstion ration oon op/Ebo 2.2.2, Preconsolidation Pre: Major variables that can influence the measured op obtained from laboratory testa run on undistur bE samples can be divided into four categori 1, Sample disturbance which will almost always Zeduce the measured value of 03. 2, Te test equipment and procedures used to Obtain the one-dimensional compression curve and related tine effects. 3. The interpretative technique used to estinate the value of 03. 4. Environmental factors auch as pore fluid compo Sition and temperature (a 10°C increase in tempe Fature can cause a significant reduction in a3 for some clays: Ladd et al. 1977, page 4425, 64 Regarding 2., the increnental oedoneter test u- Bing a load increment ratio 6P/P of unity ts moot Sonton, except that At/? ia sonetines reduced te Sbour 0.5 to obtain a better defined curve in the Uicinity of op (eastern Canada uses 0.5 throughout for its sensitive clays). Bach stress increnent is'eypically maintained for one day, and nest practitioners use the 24 h compression curve to Estinate of. Since the end of Prinary usually oc~ curs withik less than 1h, the virgin portion of the 24 ncurve is displaced downward by one oF nore cycles of secondary compression. Others such as £240" (1573) Messi anc hos, (1985) and ASTH (O2435~ $0)" reconnend using end of prinary, BOP, curves as aiacussed in Section 2-5, In any case, convention hal cedoneter tests require considerable time and yield discontinsous compression data, which lea Lo'the development of the controlled gradient (Co) tert [Love et ai, (1965)] and the congtant rate of serain (G85) test. [onith and. Wahis. (1969) ¢Wia Sa ee al, (1971), séLifors (1978), AsTH (D4186~ {5 Both teste’ require measurenent oF control of boundary pore pressures and give sduring primary" Compression curves. since excess pore pressures Scise throughout testing. the co'test also invol~ See's change in strain rate near op, which cen ai gnificanciy affect the conpression®curve, due to” Ganges in compressibility and permeability. Sased on data summarized in Section 2.5-, 1¢ will be concluded hae? 1, The end of primary (EOP) compression curve for incremental loading is independent of sanple (Grainage) thickness (at least for drainage heights equal to or greater than used in conven- tional laboratory oedometer tests) [Neri and choi (1985)]. 2. one day conpré sion curves from standard incre sntal oedoneter tests yield values of of gene- rally 10 + 10% lower than EOP curves due to in- created strains resulting from secondary cos Pression. Controlled gradient (Cc) and constant rate of train (CRS) Consolidation tests run at rates Significantly higher than that corresponding to the end of primary compression in conventional Sedoneter teste can result in values of 03 higher than obtained from oedoneter £0P clrves [eeroueii ee al. (1983) 3 Interpretative o, Techniques Various interpretative techniques have been propo Bed to estinate of from a given compression curve, (ve. the void ratio or strain and oje data), the Rost common still being Casagrande's'<(1936) empi Eical ¢ or ¢ versus log oy, construction, Fig.! (al- ote that the location of the minimum radius point can be influenced by the scale, which should be standardized, and involves judgement with gen- tly rounded curves (disturbed samples, very stiff clays, etc.). Schmertnann's (1955) technique is considered advantageous in estinating of for over consolidates clays wherein sample distufbance is suspected (Peck (1974)]. Reliance on semi-log plots to obtain of has be criticized by Janbu and Senneset (1979) and others Who maintain that arithmetic diagrans of stre Versus strain (or modulus 1/ny, oF coefficient of consolidation, etc.) are better indicators of changes in basic behaviour. The writers agree that high quality samples frequently show an essential ly linear stress-strain relationship during recom Pfession. However, except for highly structured foils that undergo a dramatic increase in compresee Wot 2 bo o z Prosasie g = "(| sn th ma 5 2 s6[—|mnnaonus QL & (CASAGRANDE & 20\—| construction 2al— (a) semi-Loc SCALE 40 100 200 400 1000 LOG SCALE oy, (kPa) Se g Jeverecr| wamece [Ep] E [Tho fame tS itcine [= lle oe wa! VERTICAL STRAM.¢ 08? ol — eo armas seat ri oo 305 a0 505 CONSOLIDATION STRESS. di (Pa? ciget: Compression Curves for Clays Maving SeBEerene Preconsoudaion Pressure Mechs nisms (2) i O08 oo 705 sibiiity beyond 02 (e.g- the James Bay clay in Fig-1(5)), the resbitant range in the probable Pi8sht0n' much greater than can be reasonably Af ofted from che conventional semi-log plot. Preconsolidation Pressure Nechaniens ‘the mechanian(s) responsible for causing the ob- The mechreconsolidation pressure of horizontal seryeGebosita can have several practical implica Gay sepessumarized in Table V. Mechanical one= tions, Shai loading-unloading typically leads tc Sine tor amount of precompression (constant o5- sage) Gnd Ko conditions, although Ko a€ 2 gives Gakehebenas’on whether or not oy. has been i Cfog Seale; (b) Natural Scale. pesiceation due to drying, freeze-thaw cycles, Desiccation saally produce scattered, often diffs. sees te define values of of and the in situ sere: cult te Geviate trom K, conditions, @.g. :s0tron S08 Tey sees, from evaporative drying, ‘The genera pic stressee consolidation pressure due to aging, Sion of fefcinas long term one-dimensional Grai~ defined hetesecondary compression) , is certainly ned creetmented. in the laboratory {Leonards and well docume (i964) and is supported by case Mir Mseiee: (pjerrum {19671}. Je should result in 3 SenstantOgn, but whether Ko remains constant di constaQcondary compression fs in dispute (see ca ene)? Finally, it is now generally acu Section har various physio-chemicai phenonena can ception increase in cn, particularly natural cou ‘carboRates, silica, ion exchange, arate resultant op profile is likely to be v ce rme eoillustrated in Fig.2 for a deposit of saat cay marine clay. Although the in sitt Ko ante chain constant during development of op the yitia stress for horizontal loading vould presuna Zig'd tkease due to cementation. Te should be no" Bly that cementation can be significant in depo- SeG,'UShging fron heavily overconsolidated clay Aten uekowen and Ladd (1982)] to the brittle Guick clays of Canada. B sion We pose two questions: First, what is the correla frbeanuzed, in ine Laboratory tna inthe field? Based on five case hi reat oet{nvelving the highly structured Canadias afertiain clays, Leroueil et al- 1983a,b) and 4o- Champ aT (1983) conclude that the in situ Ein ¢F te low to moderate OCR clays (OCR < 4-5) i fOr ene eeout F108 of that obtained from conventio within aeeter one day curves run on 70 mm piston na) ea. But comprehensive field data on a fairly sampitt geposit in New England suggest @ better, Same Teh fa with oedoneter end of primary (FOP) Corre acts However, given the lack of definitive SBea*on'a variety of clays, no answer to the que- dita ont be made other than the fact that Con stion Gigradiene (cc) and constant rate of strain relied Jee ran at rates significantly faster than (CRS) te esponding BOP rate in oedoneter tests can tne cory bverestimate the in situ op (assuming Regligible sample disturbance) » ‘NeveFtheless, the negligiPie cur with Mesri's (1964) view that it writers Cruslly more correct to use EOP rather is concePiay curves from cedoneter tests, even than sh the difference is generally less than 10 £0 208. WATER CONTENT) VERTICAL EFFECTIVE STRESS."to AND (kPa? a 8 8 ‘00, 200, 0 Deeri.ztm? : Stress History of Janes Bay B-6 Marine lay foata Reploteed from Lefebvre et al. (1983) 65TABLE V Preconsolidation Pressure Mechanisms (For Horizontal Deposits with Geostatic Stresses) a ae — Description Betty] Game] nanarte/Reaferences tee] aa, 5] Sanson mo fscnnwe Tn coves eerexcan | unetaea [es Tce vatea| neve] corsoneTand[osrie eects | eaten eceeaiey eeeraa | eeasesteesllpmmmest somes Et ory Boome esac coneeane | vere ee 2) caanges in gece peonsure | (beep | seiced ve: Giatae’cabSe'sebpage | xcep® | unicaa Sere ae why tn vestocation [11 aeyieg'aue to evavoretion| often,” | can deviate |ucyiog erustx found a8 cope ics, a Seth | Seat eereee eyes and Seno Tae Man 222 T0M noe eor casa lea ceit uct sevactocges 2) Drying due to fi ing capillary within deltaic deposits errs € bratned creep nisore | for nue not | teonards and ateachaetei veninet Sie | Receeenelay | ASSN TSA IT cobtttn: | Somat SPEER" | cores ace P value i fenystemrcoeea | (uieneueestoomenenesentone reoety soterstood and a feet oscee is titees Tocely wedeceteed st ne ce pax los sere tseraases asleseeace | rnttoaeion |eactianreinne, erst aan 2 guner causes of bonding | uniform | znfcrmation | ety" {iSrats Siertin (1973) Ghixotropy, "weathering" Quigley (1960) The second question is: how to deal with scatte- fed cp data from laboratory tests (assumed to be of on® type)? We hat ‘answer, but do Offer the following a parate out effects of @ata in light of the most likely mechanisms re~ sponsible for development of the preconsolidation pressure. Mechanisms A) andC) in Table V should Usually produce relatively linear op profile: whereas 5) and D) are likely to cause true spatial variability. 2.2.3. Yield Envelope: As previously discussed, the real physical signi ficance of the preconsolidation pressure is that At represents the yield stress observed during one-dimensional drained loading. Tt also repre- jenta one point on the yield envelope, defined as the locus of stress states that separates snall strain, "elastic" behaviour from large strain, Plastic (irrecoverable) behaviour for different Grained and undrained stress paths: The concept of a yield envelope (oi Jocus) was originally proposed by re: Cambridge University in their development of the (nodified) Can-Clay soil model to describe the behaviour of isotropically consolidated renolded clays (Roscoe et al. (1958), Rosce and Burland (1968), Schofield and Wroth’ (1968)]. Subsequent research shows that the concept also applies to natural clays [Mitchell (1970), Sangrey (1972), Crooks and Graham (1976), Tavenas. ‘and Lerouesl (1977] although the shape of the yield envelope is significantly dif- ferent from that predicted by the Can-Clay models because of the anisotropy exhibited by natural clays. New soil models used in advanced finite @ Tement analyses now incorporate anisotropic yield surfaces [erg. Prevost and ides (1977), Prevost (1978), Kavvadas and Baligh (1982)]. Gur objective is not to discuss experimental cri feria used to detine yielding (still a conerover= Sial subject) Sew theoretical proper Eles of yield’ surfaces (flow rules, vork harde hing laws, etcr), but rather to present a Brief Overview of the Aature of yield envelopes in order forillustrace their practical significance. The presentation will be restricted for simpiiciey to Ekrese paths wherein the major principal effective Stress acte in the vertical direction, i.e. 0120): ‘Sesulte from various tort les of resedinented Soston Biue clay (wynt0ty Ipr20t) that wore Ky consoliga ted ina slufzy oedebeter to 1:0 kg/on! (with one cycle of secondary compressioni, rebounded to0.25 kg/com?, trimmed from the large oedometer cell and then stored at constant water content. for up te Several months. standard cedonecer eats snowed @ Well defined increase in of with storage, tine due to thixotropy [O'Neill et Bl. (1985)] (op for the ‘shown Fanged from 1.25 to 1.0 kg/en?) .Th tor simplicity to the 93 cor fesponding to the age of the test specimen, although chis involves sone error. Tests (1) through (4) in Pig.) were isotropically consolidated to oj. * 0.25 kg/em2 and then loaded inerementally with drainage to measure the beha- vicar of overconsolidated clay. fest (7) is also Sverconsolidated, but sheared undrained in triaxial compression. Tests (5) and.(6) were Ko consolidated weil into the virgin compression fa Seand sheared undrained in triaxial extension Sna compression, respectively, £o measure the be- Raviour of normally consolidated clay. Yielding was assumed co occur ae stresses causing devia~ Elon"from the initial “linears stress-strain cur Tn the shape of the andrained ef path for fest (7)5 (a + oy ar V/ob = 8/5 conan veers Oy big 3: (a) Normalized Stress-Strain, (b) Effective ELEL3S fatns°and Yield Envelopes. from Drained and rete td Testa on Resedimented Soston Blue clay ees From Bensari (1964) and O'NeiI1 (1988)].- The results for resedimented Boston Blue clay ‘Show the following features: 1. The yield envelope is roughly centered about the Ko Line such that the yield stress for iso Egopig consolidation is significantly less than og. Both of these characteristics typical of soft sedinentary clay deposits; the review by Tavenas and Leroueil (1977). 2, the effective stress path for the Ko consoli- ‘Gated-undrained triaxial extension (CkoUE) test fons on notmally consolidated clay coincide With the lower portion of the yield envelope. 3. ‘The upper portion of the yield envelope at. high values of p'/oy coincides with the initial pare of the effective stress path for the Ko con Colidated-undrained triaxial compression (CKoUC) sat run on normally consolidated clay. At lower values of p'/oe, the envelope falls near, the maximum obliquity Mohr-Coulomb envelope for nor mally consolidated clay. 4, As a reasonable approximation, the yield enve- ‘jope for this clay can be defined from the ef fective stress paths for CKoU compression and Extension tests on normally consolidated clay and the large strain ¢' envelope. This hypothe sis is considered reasonable for overconsolida- fed clays where mechanical preconpression is the principal Rechanism responsible for the pr: consolidation pressure. Results are now presented for the James Bay B-6 Resere clay to illustrate very different behaviour mere naterial wherein cementation is thought to Ee’ he principal mechanism contributing to the Reasured preconsolidation pressure profile shown meeeigc2. this Canadian deposit is a very brittle, highly scructured quick clay. The data are from pegtolidated-undrained triaxial compression and weteRsion tests having the following consolidation Conditions (all using Ke=0.55)! 1, Consolidation to 9% gure the behaviour of fa OCR, which varied from about 1 2. Consolidation to ofc/og between 0.6 and 1.0 at ‘two depths to measure the effect of recompression for intact clay (the volumetric strains were 2¢ or less). 3, Consolidation to oje ranging from 1.3 to3 tines ine in site op to measure the behaviour of "de structured" clay [Leroueil et al. (1979)]- Figure 4(a) plots stress-strain curves fron three typical triaxial compression tests, where yielding Sieeneially occurs at the peak strength (particu~ $S5ty for the highly structured intact clay), The formalized q-0p stresses at the peak strength for Bileriacial compression tests and representative Gffective stress paths are plotted in Fig.4(b) - Approxinate locations of the yield envelopes are Brawn based on the triaxial compression peak Strengths and relevant effective stress paths.The Eesults show the following important behavioral trends: 4, At the in situ OCR, the intact clay is extreme iy'beittle, with approximately Linear effective Skrese paths (As0.2 in compression and A#0.8 in Extension) prior to yielding. Further straining SiSduces a very aranatic decrease in shear resi- stance, ‘Donounced strain softening. Such Behaviour is very different from that shown for y at the in 3 to 3.3. e’nechanically overconsolidated clay such as est (7) in Fig.3. less than ‘in the 3, the upper portion of the yield envelope for in act clay (ive. from samples having dje less than the in situ og) rises well above the large Shain’ Nohr-Coulonb envelope represented by the $'258? Tine, this behaviour is also very diffe fone trom that measured for resedimented Boston Blue clay. 4, Reconsolidation to of, values well beyond the ‘in situ op produces effective stress paths and a Sorrespording yield envelope for destructured Slay that are similar in shape to those measured for Boston Blue clay. ‘the above behaviour for intact versus destructured Zlay is very similar to that reported by Lerouei} Se a1. (1973) for tests on the St. Alban clay, Ghich is representative of the highly structured, “onsitive Champlain sea clays of eastern Canada. The upper portion of the yield envelope form Tian E> soft clays which derive their preconsoli— Gseion pressure principally from cementation bonds (or sintlar effects) lies well above the large Strain Hohr-Coulonb envelope and such materials ite characterized by brittle behaviour prior to fielding during both undrained and drained loa~ Xing. substantial strain softening occurs after Yielding for stress paths well above the Ko lin e7(a) nommaLizeD STRESS-STRAIN DATA FROM CKOUC TESTS 05; T—1——t | 9s] ao aa] ay Tp ww % Nats Ciro = ePa. T 5a 1s H0 40 yo 12 3 «86 78 AXIAL STRAIN, ca(X) ( nomans2to emeorve ernene rari ano VAD WnvtLores ast Sr Pero em O—ai_aF as oa os asaya ap 10 yo 2 tae che mau a ream (O17, Fig.4: (2) Normalized Stress-Strain, (b) Effective Stress Paths and Approxinate Yield Envelo- pes fron Chol Tests on James Bay B-6. Marine Chay‘ [pata fron Lefebvre et al. (1983)]. whereas compressibility aramatically for drained loading near and below the Kp Lin By comparison, mechanically overconsolidited clays have a more compact yield envelope and exhibit a much more ductile behaviour. Mechani- cal overconsolidation also increases the in sity Ko, whereas Ko for cenented deposits may well re~ min near the normally consolidated value (Note: in situ measurenent of Ko in cenented materials 4s difficult since the soil is very brittle). Ap propriate test procedures for minimizing the Adverse effects of sample disturbance (Section 2.3.) also depend on the preconsolidation pres- sure mechanism(s). 2.2.4. Normalization (Normalized Behaviour and Use of Normalized Parameters) It should be evident that the in situ preconsoli dation pressure, 0%, is the single moat impor tant parameter ‘cohtrolling basic behavioral trends for a wide range of problens involving both undrained and drained load applications. It controls the location of the yield envelope, which in turn separates snall strain "elastic" behaviour from large strains involving plastic deformation. However, recent research suggests that the shape of the yield envelope can be strongly affected by the mechanism(s) causing 95 ‘This in turn leads to differences in normalized stress~strain-strength behaviour, as will be {1- lustrated next for a mechanically overconsolida~ ted versus a cenented material. Lada et al. (1977) summarized the effect of over Consolidation ratio on Ky and undrained stress: strain-strength parameters for a variety of clays, Most of those data came fron SHANSEP test programs (see Section 2.3. for discussion of SHANSEP) wherein "undisturbed" samples were Ko consolidated in the laboratory into the virgin compression ran= je and then rebounded to varying OCR. Hence of for all SHANSEP test specimens is caused by mechahical overconsolidation. Data from a SHANSEP testing program performed on the Atlantic Generating Station (AGS) plastic ma- Fine clay (Ipe43:74, Ip0.620.1) are sunmarized as being representative of the normalized beha= Viour of mechanically overconsolidated cla; Koutsoftas and Ladd (1984) describe the site and testing details. Figure 5 plots effective stress Paths from a series of Ckol triaxial compression (fC) and triaxial extension (TE) tests run at no minal OcR's of 1, 2, 4 and OCR, ‘the shape of the effective stress p ges’ from concave to the left to concave to the Fight. Figure 6(a) plots log OCR versus log cy/0y, from the CKoU TC, TE and Geonor [Bjerrum and Landva (1968)) direct simple shear (DSS) tests. ‘The log-log plot gives essentially straight. Lines and hence is closely approximated by the relation ship: ey/ayq "8 (ocr™ where S is the undrained strength ratio for nor- mally consolidated (OcR=1) clay. Figure 6(b) shows the variation in shear strain and Skenpton’ pore pressure paraneter at failure (peak strength). Figure 7 presents sinilar plots from undrained ‘she paths re- spectively). For this test program, all overcon- fe 1 ie an Fw i _ zl ais i ao el Ca 7s tsi NORMALIZED EFFECTIVE STRESS. 6, Fig.5: Normalized Effective Stress Paths from SHANSEP CKoU Tests on AGS CH Marine Clay [Koutsoftas and Ladd (1984)](a) UNORAINED STRENGTH RATIO. 8 Tear ems T=] | 1c T > e 8 & NTS] r 2 & g 8 3 Ki & a 8 UNDRAINED STRENGTH RATIO,Cu/o¥e OVERCONSOLIDATION RATIO, OCR = of /ar4¢, (ib) STRAIN ANO PORE PRESSURE PARAMETER AT FAILURE bu £3 ii 1 1 = « : - tO (a) ononaineo srmenctn Mano eer eo x 6 2 UNDAAINED STRENGTH RATIO,Cu/oye 2 & (>) srnain Ano FONE PRESEUNE PARAMETER AT PARLUNE conorrion | cn | ve | one elel> afelala ‘SHEAR STRAIN AT FAILURE, 940% BGspS PORE PRESSURE PARAMETER, Ay 8 —_ T z «6 8 0 OVERCONSOLIDATION RATIO, OCR 9G3 Ar, Fig.6: (a) Undrained Strength Ratio and (b) Strain and Pore Pressure Paraneter Aat Failure vs. OCR from SHANSEP.CKaU Tests on AGS CH Narine Clay’ (Kousoftas and Ladd (1984)]. solidated data are from samples reconsolidated fo stresses less than the in situ of, i.e. repre senting behaviour of an intact cemented The OCR=1 data on destructured clay had a labo~ Tatory c¢e ranging from 1.3 to 2 times the in si creer PORE PRESSURE. PARAMETER, Ay OVERCONSOLIDATION RATIO, OCR Ag Aje (a) Undrained Strength Ratio and (b) Strain and Pore Pressure Paraneter A at Failure vs. OCR from CKoU Tests Run on Intact and Destructured Janes Bay B-§ Marine Clay [Data from Lefebvre et al. 13831 Fig.7: As previously noted, the yield envelope for in- fact James Bay clay is much higher than for de~ structured clay, where the large strains caused By consolidation beyond the in situ of, presunably Gestroy most of the cementation bonds’ Likewise, jults in Pig.7 show a discontinuity in behi 69viour at OCRe1. For example, the destructured clay has somewhat higher values of Ag and much larger strains at failure for the DSS and TE no- des of failure (note also the same ye in TC and TE for intact clay* compared to the large diffe- rence for the overconsolidated AGS clay). The James Bay intact clay exhibits a linear log cu/ 7Sho V8. 109 OCR felationship, as also typical Of Rechanically overconsolidated clays. But the “normally consolidated" 5 derived from testing intact clay is substantially higher than S measy red on destructured clay (358 larger in TC and about 158 larger in DSS and TE). Geotechnical practice routinely performs labora~ tory consolidation tests to predict long term Settlements of cohesive deposits. The foregoing illustrates that knowledge of stress history is also helpful in evaluating and correlating data from laboratory shear testing programs. This l- 80 applies to in situ strength testing, such as With the field vane. Figure 8 shows field vane data fron several sites plotted in the sane ner as used to correlate laboratory undrained strength ratios versus overconsolidation ratio. ‘These data and results from similar analyses at other sites are summarized in Table VI. The de- Fived values of cy/cjo for normally consolidated clay follow a trend with plasticity index similar to those proposed by Skempton (1948) and Bjerrun (1972). The values Of m are often larger than measured in laboratory CXoU test programs. In pai “ Pew PORE Riven | thee yj oncanic {4 OVERCONSOLIDATION RATIO.OCR 10 4 a we. FIELD VANE STRENGTH .CuCFVI/ 07, sl 8 Lg loomecncur] “LBP ames eay Ze_| vatLeY (7 | MARINE CLAY 4] \VARVED civ 2 «88 OVERCONSOLIDATION RATIO. OCR Blculary a appears tobe Siontticunely greater’ Fig.8: Undrained Strength Ratio vs. OCR from than unity for deposite believed to be fetureily Field Vane fests [Lacasse et al (19)8)]; Senented, although the data base is too snail to (a) Boston Bive clay, 1-95 Sauguss MA Grav definite conclusions. in any case, 4f"S and &) Connecticut Valley Varved Clay, Bare reasonably well defined for a given clay Aaherst, Ws Geposit, field vane testing provides « relatively (o) Organie'clay with Shells, Fore River, inexpensive nethod of interpolating and extrapola ee ting laboratory oedoneter daca to cbeain a more (a) Janes Bay B-2 and B-6 Marine Clays SouBlete: picture of che in situ stress history. Thags eV'at. cigs]: TABLE VI Normalized Pield Vane Test Results wo] site and sos. type | x, cw] 1, --Fieie vane nenarke References s . 135 Saugus a |e y ace ot al * [Roni Bis aay ae 0.165 | 0.96 | Plotted in Fig.6 | tacaase ot al. C978) Saneceicue Valter =—]0-75 = a 2 | Sonnectiew Ts Piotred in Fige | ta ean 1978) 5 Trai hiver organic Wear of weattered Clay with shell, sass | - |o.r4 | 0.02 | Pv piotted tad et al. (1969) Ete. in Pag. aaa eS oH Tange scatter —]Komtsottar an 4 |nazine clay 4927 [o.6s0.1]0.3 | 0.77 | ih ng data Fischer (1976) | BanGROKESF aban SP ESAT GRO ESTE F ean SPER 3 aizzo|o.71,0[0.29 | 0.40 [Mean oF SIGLY | tacasse et al. (1970) Sakata 6 [ommhe 30 fo.sro.sfoze | os? | - Navarro ae Rew Liskenre URSSEESTREY 7 [Cenentes ases | - [o.res] 1.51 | Sn ocn Tacasse et al. (1976) Yazyed oh Bulk : @ | games Bay, 3.523 [2.6z0.3]0.16 | 1.10 | Sean EY Siam warine Cle; profites Y Sproesiee, ig | tad et at. (1983) Janes ay oe pbont ae Hea Bea Marine Clay 5 profiles Plotted in Fig. + The larger DSS ve for intact clay might have in partly due to the use of top and bottom piss. 70Por m near unity and relatively snail changes in OCR, culFV) /af should be approximately constant ic {58 for BOcR= 2 and m= 1.020.2) for a given deposit. Mesri (1975) substantiated this “predic fion’ ia his analyais of Bjerrun's (1972, 1973) field vane and stress history correlations de~ veloped for normally consolidated and slightly Gverconsolidated (OCR < 2) post glacial clays. Soreover, application of Bjerrun's (1972) field Vane corfection factor, u, based on case histo~ ries of embankment failures to the resulting CulEV) /0p versus Ip relationship yielded an es- sentially constant” y cy (FV) /op-cu/ 90.22. Although the above interpretakion did not consi- Ger the scatter about Bjerrun's (1972) u versus {p correlation, Larson's (1980) summary of cu/ 73g values derived from 15 embankment failures fof clays having Ip < 608 yielas ey/op = 0.23 + 0.04 Hence, the in situ cy/og appropriate for analy- ses of embankment stability probably falls ‘Within a fairly narrow range for most soft hentary clays of moderate to low plasticity. Ghis conclusion 1s also supported by results of Taboratory CkoU tests sheared with differing modes of failure to incorporate anisotropy if the peak Strengths are corrected for "strain compatibili- tyre ice, the effects of progressive failure, Gescribed by Koutsoftas and Ladd (1984), (Note: aiso see Section 2.4.). However, there is sone Gvidence indicating that cy/op for highly plastic Groanie claye is higher than Quoted above [e.9. Teak et al; (1980), Holtz and Holm (1979), Lars~ Son (1980)). For example, the ratio for nonfibe- rous peats in very high. ai 2.2.5. Summary and Conclusions Preconsolidetion 1. The of determined fron one-dimensional drained joading? represents a point on the yield envelope where the Korline crosses it. 2. pifferent of ‘wiour as a fuhetion of OCI Genented clays give high FV strength as function of OCR. 3. Many natural soi1 deposits have been subjected ‘to more than one preconsolidation mechanism, acting simultaneously or sequentially and le ding to complex OCR profiles. 4. Te writers recommend to evaluate of on EOP ‘Gurves obtained from the conventional’ increnen- fal loading oedoneters. Reduced load increments (20.3) catios shoula be used when the op is ap~ proached. Continuous loading oedoneter CRS, CHG) Rests rua at fast rates can overpredict in aitu Ge. When using these devices, the strain rate eBzzesponding to the conventional BOP curve may fe inferred tron the formula given in section 25.2. Yield Envelop: 1, The shape of the yield envelor ied mechanically Oc Boston Blue ted sensitive Janes Bay clay are very different. ‘This difference is attributed to the very diffe rent preconsolidation mechanisms experienced by these two soils. The writers emphasize the heed for further research on the relation ind yield mechanisns for a variety of natural clays. 2. Consolidation of the Janes Bay clay beyond of Causes @ significant change in the shape of ith Yield envelope and hence in its mechanical be: Eavsour. This phenomenon, called "destructuration’ fs typical for highly structured clays, €.g. 0- Yerconsolidated cemented eastern Canadian clays. Normalization 1; Since of te the most inporsant parameter, con izolling basic behaviour of soil in both drained and undrained conditions, it is particularly sul fable as the basia for normalizing the relevant strength and stiffne: Sive deposits. 2, Normalized strength and stiffness paraneters ‘do not lead to unique relationships vs OCR be~ cause they depend on the oS mechanism. 3. The existing experimental evidence indicates ‘inat for low OCR inorganic clays of low to moderate plasticity, the cu/op ratio appropriate for en Banknent stability falls within a narrow band of 0.23 0.04, Characteristics of cohe= 2.3. ASSESSMENT OF SAMPLE DISTURBANCE AND PROCE- DURES 70 MINIMIZE ITS EFFECTS 2.3.1. Introduction able VII summarizes sources of disturbance which Zan occur during sampling of cohesive soils from S’Grtil hole, Some of these can be controlled oF Se Teast minimized by using the best available fampling technique for the particular soil conas- {lake aaa by following appropriate handling proce dures. For example, researchers in Eastern Canada developed special techniques [e.g. the Sherbrooke $80 mn diameter block sampler, Lefebvre and Poulin (i379), and the Laval 200 an diameter overcored tube simpler, La Rochelle et al. (1961)] to obtain Gxtrenely high quality samples of their soft Sritele sensitive clays. Their studies show that better sampling can yield significant changes in Stained compressibility and peak triaxial strengens (andreined and drained) conpared to conventional fixed piston samples (e.g. ba Rochelle and Le- febvre’ (1970) +, Raymond ef al. (1971)]. However, ‘special large diameter samplers are either pensive for most geotechnical investigations not feasible for very deep sampling and for “exploration, Noreaver, other sources Like sescclaced with stress relief, and especial Jy the outgassing problen often encountered in deep wlest deposits, axe impossible to minimize without Taking extrene measures: Given che above constraints, most, sampling programs Gust employ procedures that may yield sanples of Tees than deat quality. Hence, practicing engi neers need techniques for assessing sample quality Bnd knowledge of testing techniques that might be Shployed to minimize the adverse effects of sample Sibeuybance, Our treatment of these two subjects Sill be restricted to cohesive soils where design practice relies heavily on laboratory testing. How Biers it should be mentioned that recent progress fas been made in developing procedures to obtain mich better quality samples of granular soils, Tend. Marcuson and Franklin (1979)]. 2.3.2, Sample Quality (Cohesive Soils) ‘Three techniques for assessing sample quality will be discussed: radiography; measurements of effect ‘and evaluating oedoneter compression nTABLE vin Sources of Sample Disturbance in Cohesive Soils Heading Teen Renarke - Stress 1.1, Change in atreases due to + Excessive reduction in cy due to light aril- Relief drilling hole Ling mud causes excessive deformations in extension + Overpressure causes excessive deformations in compression 1.2, Eventual removal of in situ | . Resultant shear strain should usually be shear stres: small 1-3, Eventual reduction (removal) | . Loss of negative u (soil suction) due to pre of confining stre: sence of coarser grained materials + Expansion of gas (bubbles and/or dissolved gas) 2. Sampling 2.1. Sampler geometxy: ‘These variables affect: Technique Diameter /Lengeh Area ratio + Recovery ratio Clearance ratio Accessories ~~ piston, + Adhesion along sample walls ‘coring tube, inner foil, ete. + Thickness of remolded zone along interior wall 2.2. Method of advancing sampler | . Continuous pushing better than hammering 2.3, Method of extraction + To reduce suction effect at bottom of sample, use vacuum breaker 3. Handing 3.1, Transportation + Avoid shocks, changes in temperature, etc. procedures 3.2. Storage + Best to store at in situ temperature to minimize bacteria growth, etc. + Avoid chemical reactions with sampling tube + Opportunity for water migration increases with storage tine 3.3, Extrusion, trimming, ete. + Minimize further straining Radiography ray photons emitted fron a cathode ray tube pe- netrate materials to varying degrees depending upon their density. A radiograph is the photogra- phic record produced by the passage of X-rays through an object onto a white photographic film that darkens in proportion to the intensity of Protons reaching it. Photographs taken of radio~ graphs produce the opposite effect. Hence, a low density object appears as a dark zone on a radio graphy (negative) and as a light zone on a phots graph (positive). Although industrial radiography has been used for many years in geotechnical re- search, primarily in England for measurement of strains Via lead shot embedded in soil samples (arthur, 1972), its systematic use with tube sam ples has only recently become fairly widespread. Based on experience gained at MIT since 1978, ra- Aiography can show the following: 1. Variations in soil types, especially granular Versus cohesive materials. 2. Macrofabric features resulting fron bedding Planes, varves, fissures, shear planes, etc. 3. Presence of “intrusions” such as sand lenses, stones, shells, calcareous nodules, peaty mate- rials, arilling mud, ete. 4. Voids and cracks due to gas pockets. 5. Variations in the degree of sample disturbance, Fanging from barely detectable curvature adja 72 gent to the sample edges to gross disturbance as evidenced by a completely contorted appearance and large voids and cracks (most often occurring at the ends of the tube). tures nay not be readily identi- Inspection of the extruded san~ ples, at Least without trinning or breaking it a: Bare. Hence radiography provides a nondestructive for selecting the most representative and/o} ‘isturbed portions of each tube for enginee Fing tests. Tt also helps in planning the overall festing program based on the anounts of suitable material, Such information can be considered sential for projects having a Limited number of expensive tube samp occurs with offshore exploration. in fact, the Norwegian Geotechnical Institute used onboard radiography in 198) to pro Vide an innediate assessment of sample quality at & deep water (300 a) site in the North Sea (haces, Se et al. (1984)]. The American Society for Testing and Materials 1s currently developing an ASTH Standard for radio~ graphing soil samples. ‘The writers predict that Fadiography facilities will soon become standard fon drilling ships and in geographic areas having @ significant number of organizations engaged if Geotechnical engineering. ‘the expense of radiogra phy is relatively snail (approxinately US $50 Ber tube) compared to the cost of running consol, Gation tests andsophisticated strength tests on @isturbed and/or nonzepresentative 2011 specimens,Eefective Stress After Sampling ba obo Bo Ay = Rll Ei setisrey tBSmechafical disturbance since shear deforma~ other of the Assu- fecti- given pore pre Swelling as water is sucked from: less zones that cannot sustain negative pore pres= sures; (2) more disturbed zones, such as the re nolded (more compressible) layer along the walls of the tube. Internal swelling has been observed ‘to decrease of [Schjetne (1971)] duce the strength of Norwegian Clays CKoU triaxial compression tests reconsolidated to the overburden stress [Bjerrum (1973)]. However, ta Rochelle et al. (1976) conclude that it parently less significant in strongly cemented ela Several researchers have proposed that 0f/op3 Pro vides a quantitative measure of the degree Of sam ple disturbance (e.g. Ladd and Lambe (1963), Ne: gon et al. (1971), Okumura (1971)], and it can serve as a useful guide regarding likely decrea— es in measured UU strengths. But no data are a- vailable to correlate og/o3, with the effects of disturbance on CU or oedonBEer test results. Mo~ Feover, og/op has little relevance in highly mented clays’ since 0; can approach zero without seriously affecting the structure of these soils. Qedoneter Curves Sample disturbance usually affects compression curves fron oedoneter tests run on “ordinary* #@ Einentary clays (soft to stiff consistency, low fo moderate sensitivity) in the following manner: 1, Decreases the void ratio (or increases the strain) at any given consolidation stress, 2. Makes it more difficult to define the point of Eininun radiue, thus obscuring and often lower ing the value of op. 3. Increases the compressibility during recompres sion (always true) and may decrease the com~ Pressibility in the virgin compression region. Figure 9(a) illustrates the above effects via da~ ta’fron oedoneter fests run on specinens from the Same tube of an offshore Holocene clay. Although fest Agave a Feasonable looking curve that indi Fig.9: Effects of Disturbance on Oedoneter Con~ pression Curves: (a) Orinoco Clay [Azzouz et al. (1982)): (b) Champlain Clay, St. Louis [La Rochei~ le and Lefebvre (1971)]. cates an underconsolidated clay, the second t Yielded a significantly higher of which agrees With the stress history developed for the deposit. Jest B also had a much higher virgin compressibi- lity. Figure 9(b) compares compression curves from block and tube samples of a brittle, sensitive clay, Where disturbance increases compressibility during Fecompression but has little effect on op and vir~ gin compressibility. These results appear fairly Eypical of the cemented Champlain Sea clays where jearch [see La Rochelle et al- (1981), for re- ences] “shows that very large dianeter sampling is required to preserve the stress-strain proper~ ties of overconsolidated intact clay (i.e. within the yield envelope shown in Fig.4). Conventional fixed piston samples may or may not affect the lo- cation and shape of the virgin curve depending on the tube diameter and specific deposit. The above discussion and examples denonstrate that no simple criteria exist to evaluate sample quali- ty from conpression curves. Whether or not sample Gisturbance is significant also depends on the pro perties being assessed, e.g. overconsolidated vei Sus virgin behaviour, of, etc. Nevertheless, a Comparative evaluation of compression curves Sithin a given deposit can often indicate diffe- Yent degrees of disturbance fron relative changes In compressibility and shapes of the curves. One Simple, but not precise, criteria is the measured Sereical strain st the effective overburden stress for deposits believed to have a relatively uniform Stress history, as illustrated in Pig.10 for a Ho- locene deposit offshore eastern Venezuela. a Conclusions 1, the required sample quality depends on what pr: perties have to be measured and their required di Gree of accuracy. This can vary from: *Ttarge diameter specially cored samples to mea- sure the intact properties of brittle clays, « More conventional push tubes (preferably fixed Piston) for routine settlement and stability Problems. 2. Quantitative assessment of sample quality 1s ‘alnost impossible if accu- ‘behaviour is needed unless block ‘for comparison are available, But for sample Foutine settlement and stability problems invol~ ving relatively ductile materials, qualitative Besement can be obtained viai s*padiography-ideally suited to help select repre Sentative and best quality materials and to plan overall program based on available suita~* Ble soil. 7312 7 - ° 14] 1 2 u g 2 5 a 5 B ae e g o: a 3 os| O5|- Note: A AND © OENOTE TESTS: NIG sna oll 24 6 8 0 MEASURED €, AT i, (h) Fig.10: Reconpression Strain vs. Preconsolidation Pressure fron Oedoneter Tests in the Soft Orinoco Clay [from Ladd et al. (19800) - « Measurements of sample effective strei for comparison with that of perfect sampling, Opp * Ko or . Shape of and compressibility parameters from Oedoneter compression curve. Conventional tube sampling will generally cause sufficient and variable disturbance such that UU type strength tests will usually not give reli ble of repeatable stress-strain-strength data (also there are problens due to effects of strain rate and anisotropy). The profession is increa~ singly realizing this and hence there now ismore mphasis on using results from samples reconsol Gated in the laboratory. But what procedure should be employed? 2.3.3. Laboratory Recongolidation Procedures to Hininive Errects of sample Disturbance one Dimensional Settlement Even modest degrees of sample disturbance will nerally increase compressibility during recompr Sion by several fold. For ordinary clays, this er For can be minimized by using the recompression Curves from an unload-reload cycle perforned after Sonsolidating the sample beyond the in situ pre~ Consolidation pressure. Por highly structured and 3 14 cenented materials, this technique is Likely to Gverestinate the in situ reconpression curve ang Senay be preferable to unload from a stress lose than 9. (the writers are not sure of this recommendation). The influence cf disturbance on virgin conpressi bility varies considerably with the degree of @isturbance and. soil type. For ordinary clays, Schnertnann's (1995) technique for reconstructing the in situ compression curve is recommended, this typically inereasing CR Co / (1 = €0) by js: St for moderately good samples of nediun to sot clay [Lage (1993), sue with higly seructus Fed clays (high S, and iz) having an extrenely hjgh and variable compressibility Just beyond Gh, such as shown in Fig.9(b) one can only hope that the samples are of sufficient quality to preserve this characteristic. Undrained Stress-Strain-Strencth Properties ‘The in situ soil structure will always be red by the sampling process and hence can_ne\ be exactly duplicated in the laboratory. It is now recognized that unconsolidated-undrained (UU) type testing produces highly unreliable and va~ Fiable results for at least two reasons: (1) va~ zying degrees of disturbance, which often caus. @ substantial reduction in the preshear effecti- Ve stress, 087 Sperfect sampling” signi, ficantly alters ‘strain characteristics since shearing starts from isotropic rather than the in situ Ko stress conditions [Ladd and Lanbe (1964), Skempton and Sowa (1963), Noorany and Seed (1965)]+ Isotropically consolidated~undrai— ned (CIU) testing also has similar drawbacks and hence Ko consolidated-undrained (CKoU) tests are required, Variables to be considered in conducting the con golidation portion of @ CKol test include the preshear values of: (1) the vertical consolid: Eion stress, oye? (2) the consolidation stress Fatio, Ke = che / Ole and (3) the time allowed for Consolidation, tc; and the stress path and conso- Tidation times used to reach the preshear condi ‘on, The following discussion will focus on the first variable, in pareicular the relative ne~ Fits of the so-called Recompression and SHANSEP Feconsolidation techniques which are used in many Jaboratories to minimize the adverse effects of sanple disturbance. ‘These tvo techniques are illustrated in Fig. 11, Which shows hypothetical in situ and laboratory Ko compression curves for a slightly overconso~ Liaated soft clay. Points 1 and 2 designate the in situ condition and the preshear effective stress for UU test respectively (the latter as~ fuming no change in water content during san- pling). In the Reconpression technique, the test Specimen is reconsolidated (ideally along a Ko stress path) to ojer0jo shown by Point 3. Points A through D correspond to typical stresses used for a SHANSEP test progran (to be subsequently described) - Recompression Technique Bjerrum (1973) presents the rationale underlying this technique, which has since been routinely sed by the Norwegian Geotechnical Institute (NGI) to predict in situ behaviour. The following selected quotes attempt to reflect his views (note that Bjerrum was then more concerned with inter~ nal swelling than gross mechanical disturbance) - “provided the swelling which occurred before the testing is so small that it still is of anFi to os ° Ord wv srry conve 3 tetas) and cenented soils. Ae- Straee small stfain behaviour for Canadian type cues’ probably requires very large dianeter san~ C1285 Eithough the increased sample disturbance Eih' smaller samples will most likely lead to wtlanced stiffness and undrained strength, 1:¢- a ree ecstive results, The Reconpression technique $3rafzo recommended for highly overconsoligated Abt: °Ciays and within weathered crusts, where sees ec diseurbance is Likely to.be less of 4 pro- 1 Ditthe other hand, for tube sanples of truly nor oni conscligated deposits, there can be little mattlign that reconsolidation to oj, will usually WeSuce a significant decrease in velune and hence Brgrained strengths that are too hight. (Note: {- foeropie. consolidation to Ojo will cer’ seGnee tne error] «Although one seldom encounters Senet deposits on land (vaste disposal facilities ‘Brine exception) » they are found offshore Tei tbeas near major rivers. There are also many An Sifeery unstructured natural deposits having & reltiSventiy low OCR that Feconsolidation to point PU eigtitaigne lead to overestimates of strength Sain inereasing sample disturbance. Moreover, Yhere is no way of assessing whether Recompres- MNSE* tie'on the safe or unsafe side, except Per- five by testing samples with varying desrees Of eisturbance. Sieee than che unknown potential adverse effects OfMeiaple disturbance, the main uncertainty in Uo Of atthe Reconpression technique lies in selecting in'dppropriave Ke. Ideally it should equal the in an gpproPinien if difficult to estimate for pre~ sare olfdation mechanisns other than mechanical: Gonsgtnensional recompression will not give the in on hee gue to disturbance and the dependence of E's Nfoading versus unloading. Sone laboratories TScetncr use oedoneter tests to measure op and aiSSker and Tretand's (1965) empirical cofrelation Bropksen co estinate Ko (perre (1961)]. NGI also. wien TBotropically consolidates the Sample to che Eke*then increases oje to ho (£0F Ko < 1)**. The Hokieal University Sf Turin (TUT) measures Ko wecans OCR for the particular clay using the SHAN- Gap technique, which also strictly applies only SEP aschanigal overconsolidation. Eastern Canadian Fhe writers believe that overestimates of cu THei“asually be larger than reported by Kizk— fartrik and Khan (1964) from their investiga pation of sampling effects with normally con~ Solidated resedinented kaolin and illite. a» Kirkpatrick and Khan (1984) show little error TAEKPasing ‘this compared to the more time con- cron Gg loading in small increments (but their Sgcults are for resedinented kaolin and illi- te). 8Labora: 0.6 ce: es simply select a Ke of about 0.5 to ‘sidered reasonable for their cemented de~ Problems in selecting Ke, the Recon- echnique is relatively simple to exe- it directly gives stress-strain-strength each depth of testing. Potential errors probably not significant compared to un 8 in the effects of sample disturbance. requires more research of the type carrie! out in eastern Canada. In any casey the writers Fecommend that this technique should alvays Ze accompanied by detailed measrenents of the stress history of the deposit in order to: 1. help estimate Koz 2, interzolate and extrapolate the "point data” via correlations with op and/or OCR; use shis correlation to check the reasonable~ hess of the results compared to normalized stress-strain-strength data on similar clay 3 depesits. SHANSEP Technique SHANSE? is an acronym for Stress History and Nor- Soil Engineering Properties. As described by Ladi and Foot (1974) and Ladd et al. (1977), the SEASSEP design procedure for estimating the in sitt undrained properties of a clay deposit Anvolves the following basic steps (for any given ‘unifers* layer and required mode of failurel: ; Establish the gtress history, i.e. the profi- “Go and op, which determines the range of jes for which data are required. Ob) te Seasure the behaviour of normally consoli- dated slay and also on specinens rebounded to varying OCR to Measure overconsolidated behaviour, 3. Express the results in terms of normalized soil pararecers (NSP) and establish NSP va. OCR rela- Efonstses, €.g. cu/oge va. OCR as in Fig. 6. Use these NSP relationships and the stress hi- Story information to compute profiles of cy,etc. Althour: SHANSEP was originally developed based fon the expirical observation that it yielded rea- sonabie results, the rationale for the reconsoli- @ation sechnique used to minimize the effects of sample turbance was predicated on the as~ ‘that natural clays exhibit normalized "ur. Referring to Fig.11, laboratory com 22 curves typically approach the in situ Virgin compre: 1.5 te: times on. should Save a stFuc 1 ly cozsslidated clay and hence yield reasonably, NSP. tikewise, tests C and D give data on amples having 2 well defined overconsolidation ratio. Foott (1974) state that: "The possibili~ $ that (the above reconsolidation techni— 1 destroy some important aspect of soll "¢ that has developed during and after for the clay deposit". They also note: *. rly true with highly structured ‘quick’ clays and with naturally cemented deposits". But ritics of SHANSEP like Mesri (1975) state 1 natural clays possess a structure that soyed ..." and its recongolidation techni- Rot overcone sample disturbance; on the Af further renolds and disturbs the na- ‘ucture of the clay", This opinion was endorsed by Tavenas and Lerouei! (1977, 1980) who Presented examples of its limitations using test data on Champlain sea clays. The results illustrated Figs.4 and 7 also clearly show that SHANSEP will severely underestimate the brittle nature and high peak strength of very sensitive cenented clays. The writers agree with Ladd and Foott's (1980) views, who in essence state: (1) all natural over consolidated clays develop a structure over thei, geologic history that cannot be precisely duplica ted via mechanical overconsolidation in the Labe: ratory; (2) the Champlain clays represent a sp cial case; (3) with more ordinary clays (e.g. fz, typically less than unity), the practical question is the degree to which the SHANSEP reconsolidation Procedure alters the structure and hence whether or not it will provide a reasonable estimate of in situ properties, especially when compared with o- ther available techniques; (4) quantitative as- sessnent of potential error are difficult to nea sure in the laboratory since it requires testing? natural clay samples that truly retain their in situ structure. Lada and Foot (1980) presented the results in Fig.12, which compares undrained stress-strain ‘triaxial compression data on a block sample of Connecticut Valley varved clay using the Recompres sion and SHANSEP techniques (Ko assumed equal to unity at the in situ Ock of 4). The agreement is considered reasonable for engineering practice. 015} Geel eBOKPA : Ox OKRA 'SHANSEP CKoUC Gigx OKRA; Thx 4cOKPA 030) zg ° ix 240K ; OCR=40 BULK INDEX PROPERTIES 05; Was5I% Wi950% Wp#20% —Tpa20% 68 to AXIAL STRAIN (%> Fig.12: Comparison of Undrained Triaxial Strength Testing Procedures on Block Sample of Connecticut Valley Varved Clay LAfter Sambhandharaska (1977)}.Moreover, SHANSEP tests on tube samples, where Gisturbance produced much lower UUC strengths, fielded cg/ote vs« OCR relationships similar to those obtained with the block sample (though {ese true regarding modulus, which is mich more sensitive to the exact testing procedure) - Unpublished research by the Norwegian Geotechni cal Institute also showed nearly identical Stress-strain curves from Reconpression and SHANSEP CKoU triaxial compression and extension fests run on 20 mm fixed piston samples of Pl Stic Drammen clay (OCR*1.5). Sinilar research is needed on a variety of clay Geposits. However, before reaching general con= Clusions, the research should also include re~ falta’ on’ samples having varying degrees of €5- 2.3.4. Summary and Conclusions The Recompression technique: iy superior for highly structured 6 1, deposits, auch as typical of DENSE Glhadd, ‘where accurate enall strain b haviour may algo require block quality sampl 2. Is preferred whenever block quality samples ‘axe available and for testing weathered and highly overconsolidated deposits where SHANSEP is difficult to apply. As a result of sample disturbance, can unde~ Zestimate cy in overconsolidated cemented depo Sits (well documented for Champlain clays) and fay overestimate cy in normally to lightly ov Consolidated “ordinary” clays (more an opinior than proven fact). Requires an estimate of the in situ Ko, al- though resulting errors are probably not’ signi Eicant compared to potential adverse effects of Gisturbance. 5. Should always be accompanied by measurenents ‘Of the in situ stress history in order to (1) stinate Xo: (2) extrapolate and interpolate the "point® daca; and (3) check the reasonable- ness of the measured cu/o} values. ‘The SHANSEP techniq 1. Requires a more accurate estinate of the in fulstress history and is strictly applicable only to mechanically overconsolidated and tru- {y Rormally consolidated deposits having ideal normalized behaviour. 12. Involves more extensive (and expensive) te- jecially regarding CKoU triaxial (or Plana’ strain) teste which should closely follow Deep’ stress path. Special cells such as descri- by Campanella and Vaid (1972) or Bishop and Wesley (1975) greatly simplify this task. 3, Ie preferred for testing tube samples from re iatively deep deposits of low OCR “ordinary” Gays (again sore an opinion than proven fact). 4, fas the distinct advantage of developing norma, rength parameters than jequent projects involving fhe same deposit (especially if they have yiel> God acceptable results based on evaluated field experience) - Unconsolidated-undrained testing: 1. Cannot yield meaningful stress-strain data be Gause such tests do not start from the correct Gaitial stress state and are seriously affected by sample disturbance (even with block quality samples) - 2, often give unreliable and highly scattered cu data due to varying degrees of sample disturb Ge plus uncontrolled effects from strain rate and anisotropy. Discussion topics and research needs: archers conducted programs, such as jone for the Champlain Sea clays, on 4 variety. Of soil types to well docunent the specific ef- fects of sample disturbance 2. Assuming a negative answer, systenatic studies Gomparing results from Reconpression and SHANSEP €esting on block samples subjected to varying Gegrees of disturbance are sorely needed for a Variety of "ordinary" clays having different a- mounts of overconsolidation. 3, There is also a need for further research in~ to the causes of sample disturbance example, Baligh (1984) has used the fechnigue to evaluate the influence of sampler Geometry on strains induced during sampling. 2.4, STRENGTH-DEFORMATION CHARACTERISTICS UNDER GENERALIZED STATES OF STRESS 2.4.1, Introduction A state of stress is defined by the magnitudes and directions of the three principal stres: (or, 37 and 03) and generalized states of stress Genote“all possible conditions. For most pract: Gal field problena and when comparing various Soratory testing devices, it is usually sufficient fo describe basic differences in the applied state Of stress (applied stress systen) by two varia: Biles! (1) the direction of the applied major prin Cipal stress relative to the vertical (depositic Sal) direction denoted by the 6 angle; and (2) the relative magnitude of the intermediate princi pal stress as defined by b= (02-03) / (04 7 03) Ehanges in the values of 6 and b lead to diffe~ Gene stress-strain responses due to anisotropy and the 07 effect respectively. ‘soil behaviour under aif- yystens, i.e. under genera, {s required to obtain in put parameters for most sophisticated soil nodels Palertainiy for Category I in Table T and often for category II) and to generate basic know {edge needed to develop improved constitutive re- intionships. This latter knovledge also helps valuate Likely errors associated vith results fron tests that do not exactly duplicate in situ stress paths, e.g. use of triaxial test data for a plane strain field problem. Gale section focuses on anisotropy because 1ittle Gata and much confusion exist compared to our un- Geretanding of 2 effects. After describing dif~ ferent types of anisotropy and illustrating their Stfects, the section gives an assessnent of the Current ability (or inability) to measure aniso- {ropy ‘and then concludes with an overview of re~ search needs. 2.4.2. Anisotropy: Initial and Evolving consider identical samples from a horizontal depo Sit wherein the soil was deposited in the verti- Gal (2 axis) direction and underwent only onedi- mensional strains, i.e. Ko consolidation and re bound. Figure 13(a) shows the sample orientation for testing with the directional shear cell (DSC) ny a) SAMPLE ORIENTATION b) IMPOSED STRESSES ; y moss tu sama AY ae ese jo, Sg ee oa s23 ans OWE De LANG, © TERRA EAE a0, %% Fig. 13: Measurenent of Anisotropy with the Directional Shear Cell. that will be used to illustrate anisotropy. This plane strain shear apparatus, described subse— Guentiy, controls the major principal stress di- fection by varying the Rormal stresses, dq anc Ob, and the shear stresses, Tq = ~th, acting on four faces of cubical sample Constrained between two rigid end plattens,.as illustrated in Figure 3b). finen the cubical sample 1s placed in the psc such that shear occurs in the x-z plane [see Figure i3le)], the device measures the influence of ant~ sotropy by varying the 0; direction between the ‘and the x-axis, ise, 6 increasing from 0° fo 90°. Such a test series simulates conditions for elenents within the failure zone beneath a strip footing. When the cubical sample is placed go that shear occurs in the x-y plane (see Pigure 43 (a}], the response of the sample is independent of the’o; direction (the y angle) since soil pro- pert! ‘the same in all radial directions. his stress systen simulates cavity expansion and the self-boring pressuremeter test. The respons will be different fron the above é= 90" test Bheared in the x-2 plane since the soil is cross- anisotropic. This fact illustrates that § and b are not adequate to completely define states of Stress since both are plane strain tests having $=50", but the first is sheared in the x-z plane land the second in the x-y plane. Fig.14(a), which plots results from drained tests run in the DSC on loose Leighton Buzzard sand, shows very dramatic effects of anisotropy on Stress-strain behaviour, Por samples sheared in the x-2 plane (varying 6 angle), anisotropy not only greatly reduced the modulus, but also caused abasic change in the shape of the curves. Shea: fing in the "isotropic" x-y plane produced a re- Sponse about equal to that for ¢ = 45°. Pigure 14(b) shows corresponding data from undrained DSC tests performed on resedimented Boston Blue clay having @ nominal OCR=4. Although the initial mody lus remains constant, anisotropy has a very pro- nounced effect on the yield stress, the modulus at yield and the peak strength. 78 (2) onaineo SHEAR OF LOOSE LEIGHTON BUZZARD BAND 3 BHean wxoz sane | Bee'row rr a a a MAJOR PRINCIPAL STRAIN, €,(%) () UNonamto SHEAR OF OCRa# RESEOMENTED BOSTON 03 (ADO ET AL ; aor | a2] a/oh On —— SHEAR IN x= 2 PLANE | pr) = wean wre ane i ir ance 2a SHEAR STRAIN, P(%) Examples of Inherent Anisotropy for Plane Strain Monotonic Loading in the Directio= nal Shear Cell: (a) Drained Shear of Loose Leighton Buzzard Sand [Arthur et a1, (1981)] and. (b) Undrained Shear of OCied Resedinented Boston Blue Clay (badd et al. (1985)]. Fig.t4: The above data were all obtained from tests ha: ving an initial isotropic state of stress. Hence the observed anisotropy must have resulted enti- rely from a preferred soil structure developed during one-dimensional deposition. This type of anisotropy is commonly called inherent anisotropy. Mitchell (1976) summarizes measurements of pre- ferred interparticle contacts in granular mate rials and of preferred particle orientations (soil fabric) in both sands and clays. Macrosco- pic variations in fabric can also produce inh Fent anisotropy; examples are stiff fissured clays and varved glacial lake deposits containing alter Rating layers of "silt" and “clay” [Ladd et al. (1977)]. In any case, inherent anisotropy will lead to directional changes in basic material pro perties governing elastic behaviour, plastic be~ haviour and failure. Soils can also exhibit directional dependent un- drained strengths whenever shearing starts from an anisotropic initial state of stress. Hansen ‘and Gibson (1943) first predicted the existence of this type of anisotropy which, for lack of @ better term, will be called initial shear stre: anisotropy [section 2.2.2. of Ladd et al. (1977) used the term stress system induced anisotropy to describe the same phenomenon]. The developed theoretical expressions for the variations in undrained strength with mode of failure for nor-maily consolidated clays as a function of Koy the dvorsiev strength paraneters and the equivalent Of Skenpton's pore pressure paraneter AF [see Dun Gan and Seed (1956) "for an updated version]. Their plane strain predictions showed ocr + 1 Clay Eacension Lean Sensitive 0.331 0.193 (k= 0.5) Plastic 0.282 0.282 Insensitive (ky 20-75) where cy=gg*0.5 (01-93)f. Tt should be enphasi- wBSTEnge “ehise strength ratios were computed using {Soeropic material properties, i.e. assuming no [iherene anisotropy: Hence the changes incu re~ galt solely fron the fact that different incre- Sunts of shear stress are required to produce. Ture as oy¢ var: ween the vertical (6*0*) and the hofizontal (6=90") direction. fhe above initial shear stress anisotropy helps Tipiain the undrained behaviour of soils and it SRBL1a ne considered when conducting test pro- Glans or interpreting cy test data. But this ef- Zict’can be theoretically computed from 2 knov- [Sageof material properties, be they isotropic ae °8hsSoerepic. Hence fron a. fundamental view- point, it is not a tmaterial” anisotropy. strain-strengen. behaviour Reasured during monotonic loading (shearing) of TGencical samples vill be called initial aniso- tiepy; Soils that have undergone one-dimensional Soppsition and straining have cross-anisotropic qepetal material properties due to inherent ani— iBthepy. "tn addition, when Xo is not equal to u- seep Punarained behaviour 1s also affected by 1~ nieial shee anisotropy. 2oiie can also attain their initial anisotropy Soe to prestraining. Results from drained tests Sus on Sand in the directional shear cell will be Euga’to illustrate this strain induced anisotropy ESoge sand was deposited along the z-axis and Siacea in the DSC for testing in the isotropic Buy plane (rig. 13(d)]. The effects of strain in- Sued anisotropy were then investigated as fol- Yows (o} always kept constant) 1. Prestrain the initially isotropic test speci~ ‘nen during a first shear to R= 0j/o3» 4.0 with Dy acting at some given direction, Bay Va. 2. Unload the sample and then perform a second ‘anear at a different 0; direction, say vB 3. this produces a rotation in the o1 direction or b= by Phe arthur et al. (1977) firat conducted such tests Arpt geis plots recite fron a similar eri 20g evurr bn the sane sand previously used £0 Tul gtrace inherent anisotropy [shown in Figure {lanl Comparea to isotropic sand, the date TAG) euae "Berean induced anisceropy produces: i3f"a much stiffer response, vith characteristic wr etuolic stress-strain curves, at smell Fotar UGsTengiess and. (2) a much softer response, Short Sikear Rive. cy relationship, at large w zines. ‘but a fundamental difference exists AVERAGE MONOTONIC toaping CURVE DURING FIRST SHEAR io 1 2 3 4 § 6 we Fig.15: Example of Strain Induced Anisotropy for Loose Leighton Buzzard Sand Tested in the Directional Shear Cell [Arthur et al. (19810). between inherent and strain induced anisotropy; petvece Inherent caused a continuous increase in retinas s rotated fron 0 to 90", the strains SSFEG teste with strain induced anisotropy rise 48 PMaxinon at 6 = 70-2 10" and then decrease £2 § Mgt tis important difference also occur= se vien dense sand tArenur et al, (1977) ,(1981)1. Felelal anisotropy 13 now generally recognized £0 H'laporeane nepect of soil behaviour. Advan eosin sbil modeling aze such that the theoreti Cet Sbclity to reasonably predict the in situ re- SSonee of horizontal soil deposits under moncto~ sre tSadings having different oj directions is Reeer thas the experimental sbility to provide betttate input parameters. Moreover, some soil SSGeis havergone one step further by considering moseying anivocropy. [pafalias in RNESE (1963)] ; Wiot’ ie neane by evolving anisotropy? Tt descri- Wot*now the initial anisotropy of & soil chang B8e tov subsequent stressing or steaining. Por . GMinpies the initial cross-anisotropic properties SE iaky consolidated sample will become less prom Sounced if unloaded and then subjected to large Ryarostatic compression, i.e. the initial aniso- Tepe will evolve tovards isotropy. OF if the ERSEY iple is stressed at some 640 angle to Produce substantial plastic strains, then sheared at varying 6 angles, the measured then the will be different fron that of the "vir~ Gin’ sample. Thus the problem of anisotropy not Sit) neails identification of ite initial featy- Only .cevaiss developnent of laws describing its Fee eeton, Moreover, both initial and evolving a~ SiSoeropy ‘can influence ail three components of att meng1ing: elastic behaviour, plastic beha~ Viour and failure. Evolving anisotropy 1s especially important in Ercblens involving large stress reversals; one propeeayaaple is che foundation response of & S78, Pity offshore platform subjected to wave action. Arthur et al, (1960) used the directional shear ceiTEo'neasure the accumulated strains developed Gating drained continuous cyclic shearing. Typie Sith cal results for dense Leighton Buzzard sand after $B cycles wherein R= 0] / oj was kept approximate: iy constant during each test art 799} Rotation R Accumulated ©, (#) 40° 333 0.25 40 13) als a4 70° 3.4 0.8 42 1214 sé 20 (ote: Ry was about 7.5 for initial monotonic Yeading) Evolving anisotropy is also important when at- Eompting to predict the behaviour of axially Toaded friction piles in clay- Pile installation first causes intense shearing of the adjacent soil thereby altering its in situ initial aniso~ tropy, and subsequent reconsolidation produces further changes. The MIT effective stress soil model. developed by Ravvadas and Baligh (1982) Considers evolving anisotropy via a kinematic hardening rule that allows continuous rotation of the yield surfaces. Several other soil models [e.g. Prevost and Heeg (19771) have also incor porated evolving anisotropy in order to predict Behaviour for complex stress paths having cyclic Joading conditions especially. pertinent ¢o off- shore construction. However, little experimental data exist, that can be used to check the validi~ ty of the “lavs of evolution” assuned in these models. 2.4.3. Experimental Capabilities ‘This section as: various laboratory shear ¢e vices regarding their capability to measure strength-defornation characteristics under gen (6 angle) and the intermediate principal stress (b parameter) on constitutive relationships. It then discusses testing procedures that can be em ployed in engineering practice to evaluate ani~ Sotropy for stability problems and/or to obtain input parameters for anisotropic soil models. The focus is on behaviour during monotonic loading, ive. initial rather than evolving anisotropy and effects of cyclic loadings. The writers’ specifications for a shear device ideally suited for basic research would include the following characteristics: 1, Have uniform and well-defined states of stres: and strain, at least prior to formation of rupty Fe surfaces. 2, Provide accurate data over a wide range of ‘Strains and confining stress levels 3. Be capable of performing both stress and strain Gontrolied loading, the latter usually being sential to measure strain softening behaviour. 4, Be capable of performing both drained and un- Grained tests on granular and cohesive soils having a well defined stress history and minimal sample disturbance (the effort required t: Pare test specimens is often a major expense). 5, Be capable of varying the magnitude and di- vary b at constant 6 to measure the effect of b) vary 6 at constant b to measure anisotropy. 6. Be relatively easy and inexpensive to operat 80 Items 1 and 5a or b for any meaningful rei are certainly desirable, although specific require- ments largely depend upon the research cbjecti- ves, e.g. "elastic" behaviour versus plastic behaviour and failure conditions. specific devices, some con- Before discussin rigid end platens (i.e. the conventional triaxial cell) wherein average strains are con- puted from boundary displacenents and overall Volume changes. Stress distributions along the end plattens are seldon measured. Hence little data exist to indicate the possible magnitude of nonuniformities in stress or strain within the samples. In contrast, Arthur and his colleagues at University College London routinely measure Strain distributions within samples by using ta- Aiography (embedded lead shot) or photography (inked lines on a sample boundary). ‘Their data show substantial variations in strain for presu- mably very uniforn sand within the middle third Of triaxial specimens having rigid ends and Within flexible boundary shear devices [Arthur et al. (1981)]. This result suggests that more enphasis should be placed on measuring strain Gistributions in order to check stress-strain uniformity. The three shear devices to be discussed a: the true triaxial apparatus (TTA); the directio- nal shear cell (DSC); and the torsional shear hollow cylinder (TSHC). Saada and Townsend (1981) Present a comprehensive evaluation of these and Other strength testing devices, with emphasis on their rei Figure 16 i1lustrates combinations of b and 6 that can be achieved by the three devices and for comparison, the location of conventional triaxial compression and extension tests. Te sting of samples trimmed at various inclinations fo the vertical depositional direction (z-axis) are excluded because: the specimen cannot be Properly Ko consolidated: and bending monents and shearing forces are generates at the ends when an inclined specimen is tested between ri- gid end platens (Saada (1970), Saada and Bian- Shani (13770) The true triaxial apparatus (TTA) is ideally sus ted to investigate 07 effects, and nunerous u- sually cubical devices exist having all rigid, all flexible or mixed boundary conditions. stire and Desai (1979) and Sada and Townsend (1961) compare their attributes and give references Most have been used to study sands, with consi- stent results for 0 dx = 02 > oy 16 90%) with og < oy = 0 < oy t cavity extension, 69908, with ox'> o;* 02 > oy& INTERMEDIATE PRINCIPAL STRESS. bn (qq 0 M(O,— 95) PRINCIPAL STRESS DIRECTION. 5° DSC = DIRECTIONAL SHEAR CELL(SHOWN FOR ‘SHEAR IN x-z PLANE) ‘TC = TRIAXIAL COMPRESSION TE = TRIAXIAL EXTENSION TSHC = TORSIONAL SHEAR HOLLOW CYLINDER (VARYING 0. y:P}:Po) ‘TTA mw TRUE TRIAXIAL APPARATUS Fig.16: Combination of b and § which can be Achieved by Various Shear Devices {After Germaine (1982)]. ‘The directional shear cell (DSC) vas specifically Geveloped to investigate anisotropy in sands [Ar~ chur e€ a2. (3977)); Figure 17 shows a Giagram of the method used to apply normal and shear stresses to the four faq ces Of a cubical sample, constrained between tvo figid end platens, in order to fully control the 6) direction during plane strain shear. Arthur et al. (1981) describe the operation of its main ses for applying pressure bags for applying the normal stresses da and op; and the rediographie-photographic techniques for neg suring the distribution of strain magnitudes and directions. one of the most attractive features of the DSC is the ability to conduct “proof” tests in order to Check df the device operates as intended. Sample: fre sheared in the isotropic x-y plane with dif- ferent directions of oy to the y axis, i.e. varia tions in the 9 angle in Fig.13(a). Such tests r Guire changes in the relative magnitude of the dhear stress applied via the two sets of pressure bags, 0a # oy, versus that applic sheets, ‘tq = tp (see Pigs. 13 and 17). If the Dse and the experimental techniques work perv fectly, such tests should yield identical strese- strain data for identical samples independent of the ) angle. Drained tests on sand by Arthur et al. (1377, 1981) and undrained test on overconso. fidated clay be Germaine et al. (1985) indicate that the DSC can yield reproducible data having negligible systematic errors. newroncen puseen patno| errs, paessune BAG pefanwme Vanes ee Fig.17: Diagram of Method Used to Apply Nors Bad Shear Seresses: in the Direction Shear Cell [Arthur et al. (1977, 1980 and 1981)]. fe ‘the DSC is the best plane strain device available .reh into various types of anisotro. B aa the adéed advantage of providing RRéailea information regarding strain distribu- Cions (magnitudes and directions) . Hovever, the Gurrent University College London-KIT device has Certain linitations: (1) stress controlled, hen~ Se'cannot easily measure strain softening beh to relatively low stress ‘Limited to 50 kPa) and not shear of sands; (3) the ra~ (2 time consu~ ming and requ pital invest- Beng; and (4) reliable data require careful, expe ‘ators, Sture et al, (1985) describe xP fate DEC device that can Operate at higher stress levels and uses surface GEeplacements rather than radiography-photography. Figure 18 shows the idealized stress conditions: {hit can be achieved in the torsional shear hollow Gylinger (SHC) device, Application of an axial Zaa'w, a torque M, and internal and external Pi and Po generate Oz, T6z, prtgnd o¢ in the wall of the specimen. By control~ ‘of 04, 02 and 03 ean be independently controlled, together with She Srientation of the major principal stress di- Eeceion, 6 angle. Tests performed with identical Inner and cuter cell pressures require that Gos or * Pi Po. Consequently, changes in the angle must be accompanied by changes in the $n— fermadiate principal stress condition as given by qPereelationship b= sin? 6, which is plotted in Fig.16. For example, for tests run at constant 02" Py = Pot b= of anab=0 (triaxial | Increasing tez gives 6= 45° and b*0.5 * pecreasing og gives 6+ 90" and b= 1.0 (triaxial extension) «|) SAMPLE AND REFERENCE AXIS St ©) PRINCIPAL STRESSES ON [AN ELEMENT IN THE WAL STRESSES ON AN ELEMENT iN THe WALL S& Ss Idealized Stress Conditions in the Tor- sional Shear Hollow Cylinder Device [From Hight et al. (1983)) Fig.18: But TSHC tests run with unequal inner and outer cell pressures can, in theory, obtain any conbi- nation of 6 and b. In other words, the device can tudy the influence of changing b at constant é and also the effect of changing é at constant b. For example, tests to measure inherent anisotropy at b=0.5 by maintaining 2 = o,=0.5 (ez * 0) could be performed as follova: + Pp £ Py and og > or lead to & increasing fron 0° to “45° by increasing the applied torque M, increasing Tez. Pi and oz < of lead to 6 increasing from "e890" by decreasing the applied torque 4, decreasing tz. However, in reality, the TSHC device cannot be used for testing at all combinations of b and & due to the nonuniformity of stress (and hence strain) across the specimen wall which arises because of the specimen's curvature, and re- straint at the end platens. Sada and Townsend (1981) discuss the influence of device geometry on nonuniformities for testing with Po - Die Wight et al. (1983) also analyze the added compli cations arising from having unequal inner and outer pressures and summarize the sample dimen- sions, loading capabilities and research objects ves of various hollow cylinder apparatuses. Both references indicate that nonunifornities caused ous problens for many of the devices that have been used to investigate b and/or 6. It should be noted that stress-strain variations oc cur across the wall even when boundary stresses are uniformly applied, even if there is no end restraint due to either the application of torque or different inner and outer pressures. 82 Partial resules from tvo investigations are sun. marized to illustrace the use of the TSHC device to study inherent anisotropy. First, Saada and Bianchini (1975) kept po = pi and varied oz and toz during congolidated-undrained tests run on Ko consolidated clays in a device having nominal @imensions of height = 150 mm; inner and outer dia meters = 50 and 70 mn. ies Figure 19 plots the variation in friction angle and undrained strength ratio versus 6 angle of the four OcR= 1 clays studied, Pertinent connents follow. 1. the data are scattered, but it is difficult to iinow how much should be attributed to; + Non-normalized clay behaviour, i.e. changes due to different consolidation stresses (some trends do exist) . + Specimen variabiisey. 2, The friction angle tends to increase and then Fenain constant at § > 45°, Saada and Bianchini (1975) considered 6" to "vary in a completely random fashion between 26.7" and 54.9" and at~ tributed the large variation to the Hohr-Coulonb failure criterion not being valid for anisotropic materials. Although inherent anisotropy can and does affect $', the vriters believe that change in the intermediate principal stress condition should also be considered since b=sin? é for this test program. The systenatic increase in b could OO OP a er ae OF DIRECTION OF o, DURING SHEAR, 5 Fig.19: Results of CIU Torsional. shear Hollow Glinder Tests on Edgar Plastic Kaolin {bsta Plotted tron Sada. and Bionchint t378)}.explain the general trends in o' if this clay behaved similarly to many sands, €.9. Fig.13 of Ladd et al. (1977). (Note: b vs. ¢' data for Clays are both scarce and contradictory: see fade and Musante (1978). 3, The values of cy/og generally increase between 520° and é +30" and then decrease at greater in Elinations of oyg- But, as with ¢ above, one Eannot separate the relative effects of changes Sn'8 versus b. in sunmary, the writers believe that use of the SHC with equal inner and outer pressures for Sic research into anisotropy is severely hampered because of the unknown influence of changes in the intermediate stress condition. Hight et al. (1983) describe a TSHC device deve- {oped at Imperial College starting in 1975 for in Yestigating anisotropy. Rather extensive theoreti Gal anelyses of stress-strain nonunifornities, Specially with Po # Pi, lead to an apparatus ha~ ving the following feature 1. Sample dimensions: height = 254 mm; inner and outer diameters = 203 and 254 nm. 2. Stress controlled shearing with inside - the ~ Gell measurements of axial load and torque and, for the central portion of the sample, insi Gell measurements of radial, vertical and cir- cunferential displacenents. Ability te conduct either consolidated-drained 3 or undrained tests on saturated sands deposited one Ginensionally. 4, Ability to vary b at constant ‘ae constant b (important for me: py), but wihin certain Limits 6 Btricted to between 0.9 and 1.2 uniformitic (pre-failure stre nably free from error", Symes et al Festriction means that tests cannot be run with Combinations of b and é that fall in the upper Jett and lover right portions of Fig.16. 5 or to vary & suring anisotro~ symes et al. (1984) used this device to measure the inherent anisotropy exhibited by mediun-loose Han river sand during undrained shear. The sand fas deposited through water to form the hollow cy Vinarical specimen, isotropically consolidated to 0 = 200 kPa with backpressure (400 kPa), “pre Sheafed" via drained triaxial compression to ci/os = 2 te check repeatability (considered to fivercaused negligible strain induced anisotropy) and then shear undrained at b= 0.50 while keeping So constant at 600 KPA. Figure 20 plots normali- oe data from three CIU tests sheared at é = 0°, 24.s* and 45°. These results show that inherent anisotropy caused 4. A dramatic change in the “small strain" peak lndrained strength, i.e. cy/og decreasing fron 0.44 to 0.23, due to a combination of higher ind lower obliquities with in- 2. A fairly significant decrease in the rain” failure envelope. ‘These authors also present data wherein samples Were first sheared undrained at 6=0° and sub~ quently at é=45¢, and vice versa, that produ- ‘Fesults considered reasonable by the wri- fers. Hence the device appears to have the im~ portant capability of investigating both initial and evolving anisotropy. Based on the above results, the writers conclude Fe ¥ e os t COCTAHEDRAL SHEAR STRAN, fcr" Fig.20: Results of CIU Torsional Shear Hollow Cylinder Tests on Mediun-Loose Ham Ri River Sandy. (a) Noraaiized Effective Stress faths and (b) Normalized Stress Strain Behaviour [From Symes et al. (1984 that the Imperial College TSHC device offers Gktrenely attractive research capabilities regar~ Ging detasied studies of anisotropy under both Gtained and undrained shear conditions. However, Ghis exciting and greatly needed new testing abi- Lity is very costly, both to construct and to ops fate, and algo restricted to pre-failure stress States. For these reasons, the directional shear Gell is perhaps tore “cost effective" for, studying Snisotropy, especially at large strains, but only Under plane strain conditions and for low strength materials. Evaluating Clay Anisotr’ ‘he directional shear cell and torsional shear hol Tow cylinder test are research devices 111-suited for eagineering practice. On the other hand, major projects often require input parameters for aniso~ Efopic soil models needed for finite element ana~ [yses. Also most stability evaluations should con Sider undrained anisotropy when selecting design ferengths, at least for cases involving stage con- Struction, anusual loading conditions (e.g. ice forces on Aretic gravity structures) and sim Situations where empirical approaches have obvious deficiencies. The following gives the writers’ Views on what test devices and procedures should oF should not be used. cing cannot give input parameters for ‘ROE a reliable estimate of the in si- fu undrained strengeh anisotropy. Ladd et al. (i971) discussed specific limitations of the field vane and self-boring pressureneter tests for such purposes. Laboratory unconsolidated-undrained (U0) in Engineering Practice 83test prograns run on inclined specimens are also unreliable due 1, Variable degrees of sample disturbance, which Usually tends to mask anisotropic behaviour. 2. Neglect of the initial shear stress component Of anisotropy (Section 2.4.2.). 3. Stress and strain nonuniformities generated Within an inclined specimen having rigid end as discussed earlier, Hence estimates of anisotropic behaviour require laboratory consolidated-undrained tests, both to averse effects of sample disturban- eplicate the proper preshear Which may change with time due to consolidation. The discussion is mainly restricted to CKoU te- sting for predicting the response of a horizon= tal clay deposit during monotonic loading. Table VIIr gives an assessment of those shear devices sonably available for consulting oratories can perform CKoU TC snd TE tests. The major drawback is that Ko con- solidation is very tine consuming without Special automatic controls. Hence samples are sonetines first isotropically consolidate to she Xo ce, followed by increasing the axial stress to oje to jelopnents are fe ‘ively rapid, inexpensive and simple procedure for Ko consolidation. In any case, the triaxial cell can shear the soii with Gy¢ Oriented only in the vertical and horizontal alfections. since this also requires that b in- grease from zero to unity, the results reflect the combined influence of anisotropy and diff Fences in the oz condition. In this regard, PSC and PSE tests offer an advantage by naintaining plane strain conditions. But such devices are Jess available and genezally more costly to spe: practice. Most When interpreting TC/TE data, one can: 1, Simply ignore the change in the b condition, Since this should yield conservative results (see below) + 2, Employ an anisotropic soil model that uses TC/ ‘TEdataas input parameters [e.g. Prevost and Beg (1977), Prevost (1978, 1975]] to predict plane strain behaviour. 3. Use past experience to correct the strengths, +g. the following information from Ladd et al. (1977) for normally consolidated clays. ey (Tetaxial) Loading Direction cy (Plane Strain} Compression (50°) 0.92 + 0,08 (6 clays) Extension (6= 50°) 0.82 + 0.02 (4 clays) ‘The amount of the continuous rotation of the 0} direction which occurs in the Geonor direct ple shear (DSS) device is of unknown nagnituse Since the state of str innot be defined from ‘on only one plane, Moreover, the basic configuration Of the ‘DSS causes nonunifors stress-strain condi~ ‘Theoretical analyses have tions within the sampic led sone researchers to conclude that thes unifornities preclude "reliabii lations or absolute failure v |]. However, the writers agree with (3982) who believe that such a view is too pessimistic since yielding of the clay will reduce the stress concentrations. Those authors also present experimental DSS data showing Little effect of height to diameter ratio wnich Supports that conclusion. Nevertheless, there 1s TABLE VILL Assessment of Laboratory Shear Devices Available for Measuring Anisotropic stre: Strain-Strength Parameters of Clays in Engineering Practice (Nainly CKO Testing) apparatus bas Remarks « References to PPE Capabilities ‘Typical Equipment: 1. triaxial Cell [é0° at bo ~ Ko consolidation difficult without (gee te and ves vei Special controis [Campanella and Vaid énsor “Re et (i572) + Sontrolied stress-path testing very advantageous, bur expensive [Bisho and Wesley (1975) ] y 2. Plain Strain [é-0" 4 90° + Wore complex and costly to operate (Pca Pst) | uten'piane ves Yes than triacian strain» <_Vaid and Campanella (1974) 3. Geoner Direct | unknown ~ Unknown state of stress Simple Shear [oea5" e 13° No |! Lease expensive cig test with rapid (038) with plane Ke consolidation ‘strain D = Gin vary thevote 1 Bjerrum andLandva (1966) 4. Torsional shear [5=0" eo 90° ~ Available with controlled etress-path Hollow eylinder| ~ vign vei Yes testing [seada and Townsend (19810; With Piste int 6 + Used extensively for studying anisotra| asic py of clays by Sada et al. bur not Starting #ron kp stressesno generally valid method to determine the values Of 0.8l0y~oy)¢ and ¢ in OSS tests. So why Per form B85 teste? The Norwegian Geotechnical Institute (NGI) and MIT have used the DSs test to simulate the hori- Pontal portion of a failure surface as part of rRStangard" procedure for considering anisotropy, Gither via the "AD?" Reconpression technique Febseeye (03730) Vor oy SHAREED Laas’ and Foose 1304], The writers know of several major geo- technieal fizms, 011 companies, etc. who algo use the DSs teat in’a similar manner. The DSS device was used by Soydeniz (1976) to in- Vestigate stress-strain behaviour at various in- Clinations of the failure plane via consolidation Sith diferent tho/ode ratios and then shearing With ty acting inthe sane or opposite direction Lo thet The weitere’ fudge such use as reaonable for Bigineering practice, but it offers little Savantage over more conventional CK 7C, DSS, TE test. program: Finally, badd’ (1981) considers that the Euoth (max) measured in CKoU O55 tests gives 2 rea Sonable estimate of the average cy appropriate For evaluating foundation stability. such tests are slo much simpler and less expensive than Gkou Te/TE testing. The torsional shear hollow cylinder device used with equal inner and outer pressures enables one Yo'test along the S-shaped combination of b and Poplotted in Fig.16. As such, it should provide move extensive inforsation than can be obtained From a combination of CKaU 7C, DSS and TE tests, especially given the uncertain interpretation of 558 data, Noreover, Saada and his co-workers at Ease inactivate of Technology have used their ful Sy autonated device fo conduct nundreds of tests on Ko consolidated clays, mostiy with renolded Specimens but also with undisturbed samples. How ever, the writers withhold judgement on its u in practice since: (1) the scatter shown in pi blibhed deta is larger than one would Like to See) (2). they have not seen data vherein shear Started fron Ko stresses for comparigon with Zesulte from other devices. Figure 21 summarizes values of cy/ojc measured Tnvckgu Zc, DSS and TE teats runvon's variety of normally consolidated clays. It shows that: 1, Less plastic, andotten nore sensitive, clays ignd’to"have Righer anisotropy than sore plastic elays. a og ps8 — (© camact smarue sean (O80) CunTymen PLASTICITY INDEXIp(%) Fig.21: Undrained Strength Anisotropy from CkoU Tests Performed on Normally Consolidated Clays: (Data from Lefebvre et al. (1983), Vaid and Campanella (1974) and Various Publications by MIT and NGI]. 2, The wide spread practice of using CIUC (or éxguc) tests to obtain undrained strengths for SeSbility analyses will generally yield signi- Eicantly unsafe results for clays of low to no- derate Oc. IE the project cannot afford to run DSS and/or TC And TE tests, estimates of the in situ cy for eva {hating stability during stage construction art better predicted from oedoneter tests to measure the in situ OCR and the relationship ey/M}e = (0.23 + 0.08) (ocR)°® ‘than from CIUC testing. With major projects; CKoU testing having different modes of shearing is con Sidered essential. Koutsoftas and Ladd (1964) de~ Scribe the "strain compatibility" technique deve- joped at MIT to evaluate the influence of aniso- tropy and progressive failure in selecting strength parameters for design. 2.4.4. Summary and Conclusions ‘There are at least two obvious research needs on the influence of anisotropy and oz on the genera- lized stress-strain-strength behaviour of soi. 1, Comprehensive studies of the same soil in dif- ferent shear devices to evaluate this influence. 2, Detailed and intensive studies of anisotropy fora range of a typical soil types e.g., sands at varying relative densities and OCR and clays having varying sensitivity, OCR and of mecha- Aiens, in order to obtain: S‘basic information on initial anisotropy [Gata that can help to develop rules controlling evolving anisotropy uneil such information becones available it is Rot possible to evaluate the errors involved in shear devices such as the TX, ps and DSS. for jaing the stress~ Sefaincgerength relationshipe of d4#ferent nator zal soils. 2.8. TINE EFFECTS 2.5.1. Introduction section refer to: (1) the in Fing consolidation and strength testing: (2) creep, Which reflects continued straining at constant ap plied stress; and (3) relaxation, which refers to Bhanges in stress at constant applied strain. The physical phenomena that cause time effects renain Poorly understood and relatively little data exist Po show the interrelationship among strain rate, creep and relaxation, except perhaps for one~dit Sional consolidation testing. Hence it is not si prising that attempts to develop constitutive re~ Yationships to model all aspects of time dependent stress-strain behaviour have met with limited suc~ ‘The influence of tine effects on various types of Conventional laboratory tests can be divided into the following categories: 1, Tine effects observed prior to testing, ¢.3- variations in sample storage tine. Assuning that Storage occurs ae constant composition and voli- moefea°any change in properties presumably re- Tales from thixotropy (see Mitchell (1976), chapter I). 2, time effects observed during and/or after the ‘and of primary consolidation. This includes v $05.0 efpes of consolidation tests (incremental 85gedoneter, constant rate of strain, etc.) and the consclidation phase prior to CU or cD shear tests. Secondary compression (also called ‘a ging") refers to the specific case of drained Gzeep that follows primary consolidation as ty- Pically measured during one-dimensional consoli Gation testing. Whether the sane physical pheno mena that lead to secondary compression also have a significant effect on what happens du- ring primary consolidation has bec. a highly con~ troversial topic [e.g. Ladd et al. (1977). 3, Time effects observed during drained and un- drained shear testing. These usually involve e1 ther the influence of strain rate on measured behaviour or creep tests vhich measure strain Versus tine at constant applied stress. The writers had originally intended to discus: Several aspects of the above topics, but encoun- tered another tine effect, nanely the lack there Of. Hence this section is restricted to a treatment of tine effects during consolidation, where recent research has clarified sone impor tant controversial issues. It will specifically address three itens: 1, Effect of drainage height on the location of the end-of-prinary (EO?) compression curve and fon the applicability of Tersaghi's theory to predict field rates of consolidation. Laboratory measurements of the preconsolida- tion pressure, 0}. Possible changes in Xo during secondary co Bression. Bore pressures dissipate during primary consoli= Sation, take two extrene cases, Rypotheses A and B. lypothesis A asaunes that creep cccurs only after the end of prinary consoliantion (i after dissipation of excess pore prossures), vith Secondary compression being caused by the eine Physical nechaniana responsible for volune chan- Ses during an increase in effective stress (ergs deformation, slippage and Feorientation of pert ticles; changes in double layer thickness; Gi- stortion of tadsorbed™ water’ filme; eve") [erg Taga. (1373), Neari and Godlewski i977). Ag iti strated in Pigea2, fypothesis A predicts chat Sanple thickness, ive. drainage height. tay and hence the tine reguired for pore pr pation, hi Sf primary (20?) compression curve and hence on fhe'vaiue of op. Furthermore, the strain versus Jog t relationship for a given Load, increment. ie simply displaced in proportion to 1, a0 censonly asauned in practice, Hypothesis 8 assumes that sone sort of “gtructu- zal viscosity" is responsible for creep [e.g. due to the viscosity of adsorbed water fine a ia Bjerrun (1973), that this. phenonenon occurs du- ring pore pressure dissipation, and therefore that the strain at the end of primary consolida- tion increases, The precise effects depend upon the specific rheologic model and. input’ paranecers Those show in Fig.32 are representative Of prew dictions by Garlanger (1972) based on the s0L] mo del described by Bjerrum (1967). It predicts a si Shificane shift in the location of the field ver= Sts laboratory EOP compression curves, with @ cor Fesponding decrease in’ op and incressed consoli tion settlements, 86 ) STRAIN VS STRESS Ar END OF PRIMARY CONSOLIDATION Fon ats sawnue monn soe Cue/Con 004 J cau. samece riicxnesses) AND 9 (THIN SAMPLED VOID RATIO,e —» wvromesis a: £0" ON SaMeue THICKWE! (b) stmain vo TIME FOR OcRe! SAMPLES HAVING EQUAL INITIAL CONoITIONS Ano oy prenomesis 8 YPHOTHESIS log TIME—> Fig.22: Illustration of Hypotheses A and B in Terms of (a) Strain vs. Stress, and (by Strain vs. Th Mesri and Choi's (1985) paper to this conference reviews past attempts to investigate creep ef fects during consolidation. Zt also presents, in the writers’ opinion, conclusive data from an eight year test program on three natural clays of varying plasticity and sensitivity wherein the drainage neight vas varied from conventional 1 Tatory size up to 0.5 m- Figure 23 plots the m fired end of prinary (BOP) conpression curves from three tests run on San Francisco Bay Mud. ‘The tine required for prinary consolidation (defi ned fron pore pressures measured at different. di stances from the drainage e} increased fron days to ‘The other, two 180 gave unique BOP e135 Oe curves az by Hypothesis A. Aithoughmost specimens Were consolidated isotropically via triaxial #p cinens connected in series the results should also apply to Kp consol iaation since both conditions should have the same Cge/Co ratio [Nesri and chot (1984)], where Cae = de/ dlogt and Cc = de/log aye- Mesri and choi (1985) conclude that their results and. "the existing reliable data in the literature support the concept of @ unique EOP e-1oj oje CUE,2 mot TTT 2a 22 ° 9 & 2c z Q Q 3 \ wl slid ! NL 12 4 710 20 40 7 0 CONSOLIDATION STRESS. qi, (PSD Fig.23: End of Primary Compression Curves for San Francisco Bay Mud with Varying Srainate Heights [from Mesri and Choi, Caaf ve for any soft clay clude that laboratory "BOP conso! fa'een be used to predict the field preconsoli- Sation pressure and the amount of consolidation Settienent (for Ko conditions and neglecting Sample disturbance) « The specific probien of how tine effects influen Ge the op measured in laboratory one-dimensional SGnsolidition tests will now be addressed. First Sonsider standaré increnental oedoneter tests, Shere it {a well established that: 1. Since Cye/Ce for a given clay is essentially Constant during both Fecompression and virgin Compression [e.g. Mesri and Godlewski (1977), Nesri and choi (1985)]. 2. then the compression curve for consolidation lanes greater than the end of primary (te > tp) Sill be displaced downwara [similar to Curve K Versus Curve 8 in Fig 22 lal} and give a op le than the BOP curves Similarly, controlled gradient (Cc) and constant Fate of strain (CRS) consolidation tests will un Gerestimate the EOP op if the rate is s0 slow a: ie additional ‘straining due to secondary Compression. Conversely, if the EOP curve is in Gependent of drainage height, as previously con~ Claded, the de/dt should not affect of at strain fates equal to or greater than that required to Prevent secondary compression. However, extensi~ ge data exist fron CG and CRS tests which con~ Yeadict that conclusion. For example, the Lerou~ Gil et al. (1963) summary of results for 11 Clays typically shows a 10% incresse in op per Tog cycle increase in rate of strain. The writers dovnot know why this occurs at the higher strain rates, but do offer two possible explanations (khese being in agreenent with Mesri (1984)]: 1. €G and CRS tests run at "fast" rates have si- gnigicant pore pressure (effective stress) gra~ Bienes scrose the specimen such that resultant Variations in compressibility and permeability may invelidate Terzaghi's theory used to inter- pret the results. 2. at strain rates faster than occur near the end Ge"primary consolidation in conventional incre- mental teste, many clays may indeed exhibit a istructural viscosity". Hesri (1984) gives the following relationship to jelect the strain rate to be used in CRS tests fn order to obtain the same of as obtained from BOP incremental cedometer compression curves: ae K oC, <.— 2%, 2 = were: 280% HE Wy Fe Ce = ~de/alogove KE = initial coefficient of permeability at the in situ void ratio yw = unit weight of water NS = drainage height Cie 2 nde/dlost Ge = de/alogk ‘two empirical correlations are helpful when using this relationship. Mesri and Choi (1985) report Cae/Ce equal t 0.04. 0,01 for "a majority of {RSrganic soft clays” and equal to 0.05 + 0.01 for *highly organic plastic clays". For the per~ Reability chenge index, Tavenas et al. (1983) con Elude that Cy * 0.5 eg 188 proximation for clays having an initia: Pio between 0.8 and 3.0. They also show thi icy varies over a wide range and cannot be e Gofrelated (at least for highly structured Cana- Gian clays). Non-Linear Consolidation ‘the conventional Terzaghi formulation assumes & Tonseant compressibility during consolidation so Ghat the rate of pore pressure dissipation, du/at, ‘at any depth should be directly proportional to Ghanges in void ratio. Hence field cases wherein Settienents continue while the excess pore pri sures renain practically constant have been a\ Ecibated to abnormal behaviour, with creep often Being cited as the possible cause. Mesri (19791 and Hesri and choi (1979) analyzed one such the Gloucester Test Fill reported by Crawford and Burn (1976). They reformulated the Terzayhi theo- Fy to account for the highly non-linear void F: Elo vs. effective stress relationship for this Gverconsolidated clay having an extremely high Compressibility just beyond of, see Fig.24 (a) « They also used a linear e vs."log permeability Telationship controlled by the value of Cy. The Fesults, plotted in Fig.24(b), show: 1, Rapid pore pressure dissipation with little fectlenent at early times, as should be expected Guring consolidation along the flat recompre sion curve. 2. Then when ofc exceeds ofp, there is substantial Seetlenent with relatively little pore pressure Siscipation because of the very steep e-log dic curve during "collapse" of the clay structure. Hence, as emphasized by Mesri (1979) and Mesri and choi (1979), the abnormal pore pressure beha~ Giour observed in highly structured clay deposits Zan be readily explained by nonlinear compressibi 87() mut outa FOR THE THEORETICAL ANALYSES 0508 a7 08 Og + kore? (b) Observes and Predicted Settlement and Pore Pressure Behaviour th PERCENT SETTLEMENT ar [EXCERS PORE PRESSURE, (0/49) «100 Fig.24: Consolidation Analysis of the Gloucester, Canada, Test Fill [Mesri (1979)]- lity rather than some sort of creep phenomenon. Ghanges in Ko During Secondary Compression Schmertmann (1983) showed that no con: exists within the geotechnical profe: ding the question: secondary compression (aging)? practical’ implications when at Renpting to estimate the in situ ojo in deposite which have developed their of Sue ¢9 secondary Compression. In these casen, the Laboratory Ko Ys. OCR relationship. “obtained via mechanics) Qverconsol idation vill obviously overestinate Ko for Faged" deposits if Ro renaine constant oF decreases during secondary compression, Kavazanjian and Witcheli (1984) conclude that, Ko should increase with time for low Gck ciny Proaching a value of 1.0 over geologic tine? Based on Laboratory. triaxial cell dita for two clays (undisturbed san Francisco Day Mud anda Gonpacted kaolinite) and fron a theoretical ana~ Lysis based on the Singh-Hicenell. thi Steep equation [Hitcheli (1976)] and Sonstant rate of secondary compression, Cq. They 88 Also suggest that fo will decrease with tine for overconsolidated clays with fo dreaver than 1-05 Pig.28 shows experimental daca obeained by the weitere, ‘the tests on the undisturbed Panigagiia clay were rin a studio Geovecnico Tealiane of Hslan using a “square” oedoneter (0 "50 mn, Hg 20 mn) witha flush pressure transducer si- nilar to that developed by 8. Ladd (1963). Nei~ ther the OCR = 1 (Ko = 0.39) nor the OCR =10 foe #1.6) sample show afy significant change in to ever about two cycles of secondary compression. Four tests on undisturbed and renolded (after aiz drying) “samples of two organic silty clays ised the MIT Lateral Stress Osdoneter. [develsped by R.T. Martin and A.B 2. Wisea, it mesasren Go Via a vater filled angular fing in'a circular ef" (Osim, Ho = 26 mel}. Ro voraus tine deca Over one to two cycles of secondary compression Were obtained at i to's stregs levels for each feat (Fig.35 plots results for the niseest stress) dusing virgin compression. The results generally shoved a fairiy consistant, incresse in'ip with tine, but the rate was very snails ‘Bto/Eidge~ 0.007 £10,008. Hence even ten evel of°secondary compression would increase Ke’ by Jess than Ov. (Hote! soLl A had an anusueliy ilgh Exiceion angie, which probably explains” sts ow Ko) - The above results were obtained on soils having values of Cq/Ge representative of typical clays. Consequently, the writers believe that the views expressed by'Kavazanjian and Aivehell (1986) either do not apply to all ‘cohesive soils or are Prenature regarding their significance over geoe Togie 103 tine cycles of practical concern, Winer speek gata ton | Mame tear ew ees (© at oenss | arocnae [o mmcroeo reaneotne | ay eat X.tpato| matt seer Tasstare| Tasssate] —olecnmure | Clecmeure 08; oy he 3 2 = 08} fod 8 8 % os ue 2 2 | hs oa) 1h Fig.25: Coefficient of Carth Pressure at Rest vs. time for Undisturbed and Remoulded Clay2.5.3. Summary and Conclusions Physical phenomena causing time effects are poor ly understood and Little data exist to show the snterrelationship among strain rate, creep and relaxation. Among the different time effects, only those acting during consolidation have been discut her The available experimental data are able to clari fy the following items: 4, Data from oedoneter tests on specimens ranging Exon. 76 to 508 mm in height performed by Mesri and choi (1985) support the hypothesis that Creep occurs only after the end of primary conso Ligation, ive. after dissipation of excess pore pressure 2) Field cases wherein settlements continue while the excess pore pressures remain constant (obser~ ved in highly structured clays) can be expleined by nonlinear compressibility, rather than by creep phenomena. 3. Experimental data by the writers show that Ko is practically constant during secondary con pression. Consequently the views expressed by Kavazansian and Mitchell (1984) (i.e, Ko increa ges with time for low OCR clays and decreas for OC clays with Ko greater than 1.0) either do not apply to all cohesive soils or are prena~ ear 3. IN SITU TESTING 3.1, INTRODUCTION ‘The topics of this chapter cover only some se lected aspects of the in situ tests and their in terpretation Readers who are interested in going deeper into the most recent applications of in situ tests are invited to consult the following publications: Mitchell et al. (1978), Schmertmann (1978), Bague, Lin et al. (1978), Marchetti (1960), Baligh et. al. (1978), Lacasse et al. (1978), Marchetti and Grapps (1981), Robertson and Campanella (1964), Mair’ and Wood’ (1985), and Meigh (1985) . ‘This chapter vill deal with the following speci- fic problens: 1, ASSESSMENT OF INITIAL STATE VARIABLES Soil Profiling and Identification The usefulness, the potential, as vell as the 1i~ itaUisas Se aLeteteRe tn ated fechniques for con tinuous or quasi-continuous penetration of the Soil wil be discussed, ‘these techniques make it possible to obtain detailed information on the Goll profile, including the identification of dif ferent layers and the location of drainage bounds Hies. in this respect, the following in situ tests appear to be particularly useful: te Electrical Standard Friction cone Test (cP1] See De muster (1971, 1981, 1982). 2. Phezocone (CPTU) [Yanbu and Senneset (1974), Parez et al. (1976), Baligh et al. (1960), De Rus fer (1981, 1982)] and Pore P: (PrP) [Torstensson (1975) (the Piezocone features essentially a CPT tip which measures the cone resistance, qc, the 10 Gal lateral friction, fg, and the pore pressure, Snax, generated during fenetration. The Pore Pressure Probe neasures Unax only) - 3, Marchetti Flat Dilatomecer (DMT) [Harchetts {1960, 1962), Marcheces and Crapps (1981), Cam- panelia et al. (1985)] . It should be recognized that there are other in dite techniques like the Electrical Resistivity Probe [Kutter et al. (1979), Arulmoli et fi. (1b8i)],, the Acoustic Cone fviaaer ex at. (1981), Tringale and Mitchel? {1982)] and the geisnic Cone [fobertson and Eonparielia (i98si}yaii of which imay have @ poten tial for soil profiling and identification. How- ever, their use in practice is at present ismi~ fed due to a lack of adequate field experience. ‘Thorough laboratory and an situ calibration of these devices 1s therefore recomended. Tt has long been recognized that the measurement df even an estinate of the initial horizontal in Situ stress, Ong OF oo, 48 one of the most chal- Tenging tasks Of BSE. "Despite much effort during the last ten years, che problem still seens far fron being satisfactorily solved. ‘The main Giffi~ culties ave due to the fact that none of the exi= Sting in situ devices can be inserted into the iTaienout disturbance, with the possible excep, Elon of the self-boring pressureneter. Thus, the Values of cho measured by the different inscra~ ments are compared to each other oF to cho (or Ofe) from laboratory teats, without really knowing che! correct result. The following techniques will be examined in this section: 1, Self-Boring Pressuremeter Tests (SBP) [Bague- ‘isn’ana Jezequel (1973), Wroth and Hughes (1973) 2. oat [Marenetts. (19801). 3, Towa Stepped Blade (5B) [Handy et al. (1961, i982}. 4, Spade-1ixe Total sere: inst, ach et als (4381, 19835) 5. Hydraulic Fracture Test, (HPT) (Bjerrum and An- ‘derson (1972),mazr (1974). Because insertion of most devices induces non-ne- Gligible strains, interpretation of the test re: ‘lts must incorporate explicitly or implicitly one empirical Fuies for the correction of the ne sured stress. Assessment of the Stress History of the Soil Deposit At present, there are no in situ techniques which 2llow the direct determination of the maximum past pressure. Baligh et al. (1978) postulated that the simlta~ neous Measurement of Go and Umax during CPTU pene tration may give an indication of the stress hi~ story of cohesive deposits. This possibility will be examined in the Light of experience gained du- Fing the past five years with CPTU tests. XI. DEFORMATION PARAMETERS Given the Limitations of the assessment of soil deformation parameters from laboratory tests, par ticularly when dealing with OC clays and cohesion less soils, their determination by in situ test is considered to be of great practical interest. ‘The following in situ techniques, which are suite ble for the determination of soil stiffness, will be briefly examined: 1, spp [Wroth (1984), Mair_and Wood (1985)], Mé~ hard Pressureneter (PMT) [Baguelin et al. (1976)], and the recently developed Push-In Pressureneter, PrP [Henderson et al. (1979)]. cells (Tc) [Massarsch (1975), Ted and Charles ie2. cpt [Schmertmann (1972, 1978), Schnertmann et 42, (1378), Senneget et a2, (1972), Campanella and Robertéon (1983), Robertson and Campanella (seat. 3, DMT [Marchetti (1980), Boghrat (1982), Robert Son and Canpanella (1983) 4, Plate Loading Tests (PLT) , including Screw Plate Tests (spt) (Janbu and Senneset (1973), Kay and Parry (1982), Marsland (1971) ]- Mainly due to space limitations, the writers did not make an evaluation of the seismic methods, which through measurenents of the propagation ve locity of elastic waves (especially shear waves vg) in supposedly horizontally layered deposits permit the shear modulus Gnax at small shear Strain level (y= 10-58) to be assessed [Hoods (1878). Despite this omission the writers believe that methods such as the measurement of vg by the Govn-hole and especially cross-holes techniques. [Miller et al. (1975), Bodare and Massarsch (1982), Nazarian and Stokoe (1984), Hoar and Stokoe (1984)] are of great relevance to the study of soil deformation moduli in situ, and they provide insight into the soil stress-strain behaviour at. low strain levels. III, FLOW AND CONSOLIDATION CHARACTERISTICS In view of the major role which in situ tests have played in the evaluation of flow and conso~ Aidation characteristics of natural soil deposits, ‘the following testing techniques will be consid reds 1. Borehole permeability tests [ivorslev (1951)]. 2. Pumping tests from wells with bservation pie Zometers [Mansur and Kaufman (1962); Cassan (1980)] 3, Permeability tests fun through piezoneters [wvorsiev (19511, Gibson (1963), Wilkinson 0367, 1368)). 4, Self-Boring Permeaneter Tests [Jézéquel and Mieussens (1975)]- 5, Holding Tests using the sap [Clarke et al. insten] 6. cPTU dissipation tests [baligh and Levadoux (1360), Torstensson (1975)]- 7, Evaluation of consolidation characteristics ‘on the basis of pore pressure and strain ne renents nade under loaded areas [Bishop and Al- Dhahir (1970), Asaoka (1978), Orleach (1983), Janiolkowski and LanceDlotea’(1984)). The possibilities offered by these relatively well-known tests and their limitations will be examined with particular reference to the fol- lowing problens 1, Which type of flow conductivity (k) or consoli- dation (c'} characteristics may be obtained from each of these tests when dealing with an aniso- tropic and/or non-homogeneous soi] deposit? 2. Since both k and e depend on the effective Stress level, what is the significance of the paraneters obtained from the above tests in re- Tation to the existing effective in situ stress system and a given stress history? 90 3.2, ASSESSMENT OF THE INITIAL STATE VARIABLES (SOIL PROFILING AND IDENTIFICATION; INITIAL IN SITU STRESS, STRESS HISTORY) 32 Cone Penetration Test (Electrical Type Spy aneteseton test _tehecesices ee In the late 1960's the CPT was greatly improved by the introduction of the electrical penetrone- ter [De Ruiter (1971)] which permits a continous Measurenent of both the cone resistance qe and local shaft friction f,. Both parameters are sed by two electrical strain gauge load cell: shown in Pig.26. A detailed description of electri ie is found in De Ruiter (1981, 1982), Schaap and Zuidberg (1982) and Robertson and Campanelia (1984), Tes potential has been 1 gely improved due to recent advances in data acqui Sition systems (see Chapter 4.) that allows the = measured data to be obtained and stored in both graphical and digitized forms [De Ruiter (1982), Bruzzi ang Cestari (1983), Robertson and Campanel la (19841). ‘The latter is’extrenely useful for most engineering applications of CPT and CPTU re sults. WATER SEAL SOIL SEAL FRICTION SLEEVE. LOAD CELL Fig.26: Typical Electrical Cone {Sehaap and Zuidberg (1982)]. Because of improvenents in equipment, testing pro cedures, and interpretation, ‘the electrical GPP has becone the major tool for offshore soil inv Stigations [be Ruiter and Richards (19691]. Ie presently offers an alnost exclusive capability For obtaining reliable information on soil depo sits sn deep water. The electrical cone has been tentatively standar~ ized [ISSMPE (1977) and ASTM (1979)) and. at pre jent it haa the folloving relevant. features! S*Cone yith an apex angie 60" and base area of fo en; = friction sleeve located immediately behind the cone having an aves Of 150 ent. ‘through continuous monitoring Of qe and f and their ratio PReqe/fs_(ealled the friction ratio), any attenpts have been made to develop classifi~ Cation charts relating the soil type to the mea Sured gc and FR [pegenann’ (1965), Sehpertmann (oregs When using such charts and generally when inter~ preting cPt and PIU results the following factors ust be Kept in mind: 1, Tt has been reaiized [be Ruiter (1981, 1982); Ganpaneiia and Robertson (1983)] that, when the Glectric cone is subjected to hydrostatic water Pressure, en important shift of the reference Tectronsc signal (sero readings). is observed for both the load cells incorporated in the tip (see Fig.26). Tis phetometion is due to the unequal areas of Both the cone andthe friction sieve on which the water pressure is acting. (Pig.27) - Thuss both ge and fq, as measured during the penetra Elon pocess, ‘do not represent the total resi stance offered by the surrounding soil. They areFig.2! : Unequal End Areas of the Electrical Friction Cone. somewhat lover, depending on the specific con- Seruction of the cone. [n order to convert the measured values of qc and fe into the corrected ‘to be expressed in terms of total stresses qty fi to know the area ratios for the cont for the sleeve b=FL/Fy, respectively, ced in Fig. 27. These values should be obtained through careful Laboratory calibration; The electrical cones prt Of @ between 0.75 and O. hes, these values may be as low as 0.5. The correction of de~ fg into qe~ fe value Scussed herein requires the penetration pore pre: Sure to be measured, and this is possible only when the CPTU is performed. In this case: de Ge te 8) Una sty tk, -d)g, fe max wheres kor ky * correction factors depending on the off- jet between the point where Unax is mea- Sured and the base of the cone and mid~ height of the friction sleeve respective ly [see Lunne et al. (1985)]; gay * Measured penetration pore pressure. a= AN SEE FIG.27 A 20 CONE RESISTANCE a¢(kPa) OF 200 400 600 800 1000 APPLIED WATER PRESSURE u(kPa) ple of Determination of Ay/Ar in ‘the Pressure Chanber ed PBattaglio and Maniscatco (19837). Unfortunately, these corrections are not straight forward since the nay is not constant along the penetrometer shaft a shown in Fig.32. In fact, Pe'depends on the offset between the points in ‘the ground at which the qo and fy and Unax are measured, When Umax 1s monitored only at one loca ion, the correction of both qe and fs may be done by relating the measured pore pressure to the theo Fetical distribution along the penetrometer tip and shaft, as shown in Fig.32. Measurenents of Gpax at the cone base allow a direct correction in the cone (Fig.27) have an is portant influence on the measured gc and fq value Th cohesive soil deposits. This is especially true in soft clays, where high unax and low qc and fg Yead in sone cases to paradoxes like measured unax > de and negative values of £5. Once the importance of the above problem has been Qdmitted, it must also be recognized that all ext Sting correlations of soil type with ac or FR, have been developed for uncorrected qc and fs va~ lues. Therefore in the case of cohesive soil depo sits, these correlations incorporate systematic errors, depending on a combination of specific features of the cone design and characteristics Of the soils used for their formulation. fn view of the development of the CPTU which al- love a correction of both che de and the £5 to be readily made, the preparation of improved soil identification charts based on corrected qt and fe values appears to be warrented. 2, when using the FR for identification purpo: ‘eimust be Rept, in mind that both gp and fs influenced by the inital effective Lateral stress oie existing in the ground [Schmertmann (1972), + SMaraneter which reflects strongly the stress AL story of the penetrated deposit. 91on the basis of the cpr performed in sands in calibration chambers [Schrerenann (1976) Baldi et al. (1983)), it appears that the effect of changes in ofo/is more important on f_ than on ge- Ih’ sensitive clays the severe renoulding and large reduction of the of caused by cone pene- gration lead to situation in waich the messured fg will be close to sero, which nakes the FR Largely questionable Therefore, one Might suspect chat for the sane ily depending on its stress istory and sitivity, classification chares Like the one Shown in'Pig.29 based on qe and FR will be n= Feliable. jased on existing published information iBehmeremann (198]] and on the wrivere™ expe Fience, the following points concerning the use Of CPT’ in stratified soil deposits nay be nad + The thickness of thin stiff layers enbedded in the soft soil ass should exceed approxina tely 70 on in order ve full ge at ite mid height, + Por thin soft layers ina stiff deposit, the minimin thickness to assure that the correct dg 18 measured should probably exceed 20 to + The detection of the presence of thin soft layers enbedded in stiffer sos deposits re- quires digital output of gc and fy values at Teast every 2 cn. Even in this case the de~ tion of soft ienses and layers whose thick ness is less. than 20 en requires mich exper tise and experience. Tt is necessary to recognize that the suc fal use of cP? resuite for soil profiling and identification requires standardization of the 1 bar = 100 kPa = 1.02 kg/cm? ool a a. 8 So 4 g z < a 5 2 = wit o i. 2 3 4 5S 6 FRICTION RATIO FR ‘%) Fig.29: Simplified Classification Chart for Standard Electrical Friction Cone [Robertson and Campanella (1984)]. 92 cone design and of calibration (Schaap and 2uia berg (1982), Robertson and Campanella (1984)] and testing’ procedures [Schmertmann (1978), Ro- bertson and Canpanelia (1984)]. Te 1s especial~ ly inporeane for CPT interpretation to maintain the standard penetration rate +20n/sec because the measured Soil response may be influenced by 8) depending “on the permeability and strain rate sensitivity of the soil deposit. Based on experience with CPTU it may be’ argued that the standard penetration zate assures: + practically undrained penetration conditions in Ronogeneous cohesive deposies) + alnost drained penetration conditions in relat, vely clean sands having a fines content of lest than $08 For this specific point refer to Roy et al. (3962), Bellottl et al. (1983) and. Robertson 1d Catpanelia (1964) and see the following Section. 5. Almost all electrical cones presently in use have load cells (maximum capacity 50 to 80 kN) which allow penetration of soils ranging from very soft clays (qe < 100 kPa) to very dense sands (qc frequent{y"> 30000 kPa)... Thus it ie evident that the gc actually measured in very soft to medium cohesive deposits suffers from low electronic resolution of the load cell and therefore may be barely reliable. this factor is important when any ratio like FR of Umax/de is to be evaluated. To solve this problem, Ridgenet al. (1982) have developed a cone tip with two load cells then is capable of carrying loads corresponding to a maximum qe = 50000 KPa, while the second much nore sensitive one measures gc up to S000 kPa. A specially designed overload mechanism protects the more sensitive cell once its maxi- mum design load has been reached. 3.2.2, Sot} profiling and Identification from cPry st (1974), Schmertmann veneies [Jansu and Senn (1974)]. ‘This vas followed by the development of the Pore Pressure Probe (PPP) [Torstengson (1975), Wisea et al. (1975)] and the Piszocone [Janbu and Senneset (1974), Parez et al. (19761]- Since then, a rapid development of the PPP and CPTU has coeur Fed, which is evidence of the great potential that these devices have in soil exploration. In the present paper attention is devoted only to the piezocone (Fig.30) because this device measu- res simultaneously Gc, fg and the penetration pore Pressure nas. Prom the Interpretation point of View, this fact presents a distinct advantage o- ver tne PPP which allows the measurement of trax only. The most interesting applications of the CPTU in geotechnical engineering are: 1, Soil profiling and identification; this topic Will be briefly discussed in the present section. 2, Tentative assessment of the stress history of Cohesive deposits; see section 3.2.9. 3. Evaluation of the flow and consolidation pro- perties in cohesive deposits; see section 3.4.5. 4. Assessment of ground water conditions [Batt Gio et al- (1581), Robertson and Campanella, tise. 5, Indication concerning the liquefaction suscepti bility of sand deposits [Schnerenann (1976a) » NOE fon, (1983), Rok (1963), Robertaon and Campanelli TSedai}Fig.30: Example of Existing Piezocones. In addition to the above mentioned applications, a'knowledge of Unay rationalizes and Rakes less Snpirical the intefpretation of go in both cohe Sive and cohesionless deposits. Tis makes pos sible: 1, An assesament of the drainage conditions under inten peneeration is performed. ‘The experience Gained in the last five years has showed that $n Clays and uniform silts, cone penetration at the standard rate of ? cp/sec occurs virtually Shundrained conditions [Roy et al. (1962, jSeza), Campanella et al. (1981), Battaglio and Bruget’ (1988)] +_n lean sands (< 108 passing ASIN sieve NO.200), the penetration of CPT at the standard rate is virtually a drained process fants (1982), Belloces etal. (1903)]. This is Jeri“Supported by the Fesuies of CPTU performed Tn the recent alluvial deposits of the Po River Vatley shown in Fig.31. At a depth between 4 and acters below G.l, the soil consists of nediun to coarse dense sand with fron 2 to 108 fines. In this layer the cone penetration occurs vith & fogligible variation of the pore pressure with Fespect to the equilibrium up line. 2, Te interpretation of qc in both cohesive and cohesionless deposits by adopting a more funda al effect ‘approach, [Janbu and Sen Reset (1974), Senneset et al. (1962), Senneset ana danbu (13847). 3. The corzection of qe and f, for the unequal areas effect. Despite the great potential of CPTU, experience accumulated in the last decade with CPTU and PPP Fig.31: Results of CPTU Performed in Recent Miluvial Deposits of the Po River Valley" [Battaghio and Bruzzi (1984)]. strongly suggests that reliable and accurate mea, Surenents of Umax during cone penetration are Still difficult and controlled by a number of factors which have scarcely been investigated and thas are not completely understood. The factors are (i) a comprehensive and generally valid theo- fy for the prediction of the stress and strain Elelds, hence of the Unax around the penetrating Gone, (2) equipment design criteria, and (3) test procedures [see Levadoux and Baligh’ (1980), Rad (i983), May (1983) , Robertson and Campanelia (i364), Baligh (1984), Narayanan et al. (1984)) These factors deserve sone discussion, since they are inherently related to the use of piezocone: in soil profiling and identification, 1, Baligh and co-workers [Baligh et al. (1978, jg80), Baligh and Vivatrat (1978), Levadoux and Baligh (19601) demonstrated both theoretically and experimentally, that Umax depends on the po~ sition of the porous stone on the penetron fip. The theoretical distribution Of the penetra xeess pore pressure predicted by Levadoux GcR=1 Boston Blue clay fe compared to experimental data obtained by aif ferent’ researchers in Pig-32. The theoretical predictions agree quite well with the measure- enti 2. Fig.33 reports some of the values measured at ‘the Pontida site in Italy which is a deposit of neiun stiff silty-clay with a highly developed Gacrorfabric. The measurements have been perfor- med with a PPP having three porous sensors loca~ fed as shown in Fig.33. 93EXCESS PORE PRESSURE FOR 2=0 EXCESS PORE PRESSURE FOR 2+0 2ay-+ 2 3 aL | Jeprepictep FOR | te NC BOSTON BLUE 3-4 CLAY BY LEVADOUX| rat em |AND BALIGH «1980 ny tol if ccm 2 be : | LY 553 FS lemucnerat. | ae 24.3201 7 1 vel.220.2) Soares 00 6| od. 015) cca 21 LAUREN Sa AFT aaszas| “eon! 3 VQAEO2) nov Er AL.ci9en| 2 NC? > sucawana a ! coy te { ol 1 fee 321» ans er aLito0d) CLOT Fig.52: Influence of Filter Location on Pore Pressure Measured During Penetration for 60" Standard Cone Fig.33: Penetration Pore Pressure in a Medium to iff Clay at the Pontida Site. Battagiio and Bruzzi (1984). An analysis of this test and similar data e1 bles the following practical remarks to be made: = The Upa,; Measured at the cone apex is 17 to 260 higher than the unax? measured imnediately behind the cone, and it is 50 to 60¥ higher than the unax3 measured 14 cm above the cone. + The ratios of Unay1/tmax2 and Una; /Unax} seem to be slightly influenced by the*Value of OCR in the sense that they increase as the OCR de 94. 4, the reliability of the monitored pore prt creases (compare Fig.33 to Fig.59). The Unaxy yields the best detail of the soil profife‘and macrofabric. The Usax? gives only a slightly less detailed picture of the pene= trated deposit. Tinally Upay3 measured well behind the cone is toa large extent less Sensitive to the stratigraphical details of the deposit. No single location of the filter stone exists Which provides the "best" pore pressure for all the possible applications Of the CPTU. However, as far as soil profiling and identification are concerned, it appears that the best position for the filter stone is the one located sonewh fon the cone. On the other hand, considering the Still good stratigraphical details obtained with the filter located just above the cone, there are sone reasonable arguments [Robertson and Campane] ta (1984), Tavenas et al. (1982)] for this posi- tion also; if the main aim of CPTU is soil profi- Ling and identification. The arguments are: = The porous stone is much less subject to damage and abrasion. ~The measurements are less influenced by dynamic pore pressure due to the deformability of the Eip and of the porous stone. This point is especially important in sands. = This position is very appropriate for measuring the pore pressure needed to correct the qc for the unequal area effect (and to estimate it for fs). depends largely on the rigidity of the measuring systen. For properly designed piezocones this is mainly controlled by the quality of de-airing.campanella et al. (1981) and Battaglio etal. (1981) have Presented experimental data indicating the impor- tance of appropriate de-airing if reliable infor- mation both during the penetration and dissipa- tion stages of CPTU is to be obtained. Experience (Baligh et al. (1981), Acar (1981), Franklin et (1981), Lacasse and Lunne (198da) ,Rad (1983), ‘taglio and Maniscalco (1983)] indicates that the de-airing and assembly of piezocones should preferably be performed in the laboratory under high vacuum in specially designed chambers with thoroughly de-aired saturation fluids, water, glycerin, and silicone oil can be used. After the completion of the de-airing operation, the tip is enclosed in cial plastic container filled with the saturation fluid. Tt can chen be transferred to the testing site without losing jaturation. Many research organizations check dynamic time Fesponse of CPTU tips after their saturation and after the test, see Fig.87. Very Little is known about the criteria concer- ‘hing the selection of the filter element to be U- sed in CPTU. The effects of the filter type, di- mension, permeability, and compressibility are scarcely understood. Based on the writers’ expe- rience and that of Smits (1962), Rad (1983), Bat~ taglio and Bruzzi (1984), the following points may be made about filters: = In cohesive deposits a high entry value fine graded filter should be used. According to Smits (1982), a 2 y) pore diameter porous stone, saturated with silicone oil, leads to a correct pore pressure response under a cyclic loading of 100 Hz. This kind of porous nsor can possibly penetrate unsaturated soils without losing its saturation.