APPLICATIONS to GEOTECHNICAL DESIGN - Marchetti DMT

THE FLAT DILATOMETER (DMT).APPLICATIONS to GEOTECHNICAL DESIGNMarchetti, Monaco, Totani, CalabreseUniversity of L'Aquila, ItalyTorino 20 Nov 2001Documentation: Report TC16 ISSMGE (2001)updated comprehensive ref.• Derivation of geotechnical parameters(Design via parameters)• Literature summary on:SettlementsCoeff. of Consolidation and PermeabilityVerify if a slope contains slip surfacesLaterally loaded pilesLiquefiability of sandsCompaction controlPavement subgrade modulusSubgrade Kh for diaphragm wallsDMT for FEM input parametersTorino 20 Nov 2001 1

STANDARDSKEY REFERENCESEurocode 7 (1997). Geotechnical design - Part 3: Designassisted by field testing, Section 9: Flat dilatometertest (DMT).ASTM D6635 (2001). "Standard Test Method forPerforming the Flat Plate Dilatometer ". In corso dipubblicazione.MANUALSMarchetti, S. & Crapps, D.K. (1981). "Flat DilatometerManual". Internal Report of G.P.E.Schmertmann, J.H. (1988). Rept. No. FHWA-PA-87-022+84-24 to PennDOT, Office of Research andSpecial Studies, Harrisburg, PA, in 4 volumes.è US DOT - Briaud, J.L. & Miran, J. (1992). "The FlatDilatometer Test". Dpt. Transp.- Fed. HighwayAdm.., Washington, D.C., FHWA-SA-91-044, 102 pp.ISSMGE TC16 (2001): DMT in Soil Investigationsè Short Course NOTES on Test Execution (Bali, 2001)DMT on the INTERNETIn Report: Link to download key papers.Torino 20 Nov 2001 2

GENERAL LAYOUT of DMTTorino 20 Nov 2001 3

INSERTION of the BLADEDMT USING APENETROMETERDMT USING ADRILL RIGTorino 20 Nov 2001 4

SOILS that can be TESTED by DMT• SAND, SILT, CLAY But can cross throughGRAVEL layers » 0.5 m• Due to balance of zero (null method) : highresolution even in NEARLY LIQUID SOILS• Clays : Cu = 2- 4 kPa to Cu= 10 bar(MARLS)• Moduli : up to 400 MPa• Not just soft soils. LIMIT is push capacity(blade 25 tons). Trucks 20 ton DMT fast &easily in hard soils/ soft rocks.Torino 20 Nov 2001 5

Only Mechanical Parts. Fixed dimens.(–0.01mm). Cannot be regulated.Torino 20 Nov 2001 6

REPRODUCIBILITY of DMTCestari (SGI), Lacasse (NGI), Lunne (NGI), Marchetti(Aq)NC clay Onsoy, NorwayTorino 20 Nov 2001 7

BASIC (1980) REDUCTION FORMULAEFrom A, B (p o , p 1 ) :Torino 20 Nov 2001 8

PRESENTATION of DMT RESULTSHOW TO USE DMT RESULTS• M and Cu : common, usual way• Id : soil type (sand, silt, clay)• Kd similar shape OCR (useful to understandhistory of deposit). NOTE : Kd=2 fi OCR » 1ACCURACY of Parameters by DMT(using 1980 published correlations)Mostly comparisons on (National) Res. SitesTorino 20 Nov 2001 9

CORRELATION OCR vs KdKamei T. and Iwasaki K.(1995). "Evaluation ofUndrained shear strenght of cohesive soils using a flatdilatometer"COMMENTPractically no appreciable difference vs 1980 correlationTorino 20 Nov 2001 10

Torino 20 Nov 2001 11

DMT in NC sitesCu in Tokyo Bay ClayGeotechnical Research CenterKiso-Jiban Consultants Co., TokyoIwasaki, K Tsuchiya H., Sakai Y., Yamamoto Y. (1991) "Applicability ofthe Marchetti Dilatometer Test to Soft Ground in Japan", GEOCOAST'91, Sept. 1991, Yokohama 1/6Torino 20 Nov 2001 12

Cu at Skeena Ontario CanadaVs Field VaneMekechuk J. "DMT Use on C.N. Rail LineBritish Columbia", First Int.Conf. on the FlatDilatometer, Edmonton, Canada, Feb 83, 50Torino 20 Nov 2001 13

Cu at NATIONAL SITE BOTHKENNAR UK0KPa5Z (m)101520Nash et al., Géotechnique, June 1995, p. 173Torino 20 Nov 2001 14

Cu at NATIONAL SITE FUCINOKPaZ (m)Note the various Nc for CPT(U)A.G.I., 10 th ECSMFE Firenze 1991 Vol. 1, p. 37Torino 20 Nov 2001 15

Cu in 2 Malaysian ClaysVs Field VaneWong, J.T.F. & Dobie, M.J.D. 1990. Marchetti Dilatometer: Interpretation in MalaysianAlluvial Clays. Seminar on Geotechn. Aspects of the North-South Expressway, 5-6th Nov,pp. 87-96.Torino 20 Nov 2001 16

Cu in Recife Clay – BrazilUniv. of Pernambuco Research Site 1Coutinho et al., Atlanta ISC 1999Torino 20 Nov 2001 17

CONCLUSIONS on Accuracy Cu,DMT• Comparisons (+ many others) indicateCu,DMT fits right in middle of other Cu.• Results : many geograph areas, » Ok foreveryday practice, w/o local correlations• True, Cu not unique (TRX compression - ext.,simple shear, pl. strain, FV..). But normaldesign rarely use various Cu – Ok good ave.• Cu,dmt ++fast than Lab (disturbance ?).Half fast as CPT (but must choose Nc!).Torino 20 Nov 2001 18

M in ONSOY Clay (NORWAY)Norwegian Geotechnical Institute (1986)."In Situ Site Investigation Techniquesand interpretation for offshore practice"Report 40019-28 by S. Lacasse, Fig. 16a, 8Sept 86Torino 20 Nov 2001 19

M in Tokyo Bay ClayGeotechnical Research CenterKiso-Jiban Consultants Co., TokyoIwasaki, K Tsuchiya H., Sakai Y., Yamamoto Y. (1991) "Applicability ofthe Marchetti Dilatometer Test to Soft Ground in Japan", GEOCOAST'91, Sept. 1991, Yokohama 1/6Torino 20 Nov 2001 20

M in Sites in Virginia, U.S.A.Failmezger, 1999Torino 20 Nov 2001 21

M in Bangkok ClaySeah and Rasheed – unpublished resultsSIMILAR CONCLUSIONS for MGenerally satisfactory for everyday practice,w/o local correlations (clay)Torino 20 Nov 2001 22

APPLICATION N° 1 SETTLEMENTSby BoussinesqS= ∑ ∆ σvM⋅∆ Zor 3-D with E » 0.8 M -⇒ Similar Predictions(Poulos : important is Modulus, not Formula!)Torino 20 Nov 2001 23

DMT-calculated vs observedSETTLEMENTSSCHMERTMANN, 1986 - 16 CASE-HISTORYProc. In Situ '86 ASCE Spec. Conf. VIP, Blacksburg, p.303.No Location Structure Compressi Settlement (mm) Ratioble soilDMT/DMT ** meas meas.1 Tampa Bridge pier HOC Clay *25 b,d 15 1.672 Jacksonville Power Plant Compactedsand*15 b,o 14 1.07(ave.3)3 Lynn Haven Factory Peaty sd. 188 a 185 1.024 BritishColumbiaTestembankmentPeatorg. sd.2030 a 2850 0.715abc6abFredricton""Ontario"Surcharge3' platebuildingRoadembankmentbuildingSandSandQuick cl.SiltPeatPeat*11*22*78*300*262aaaa,oa,o1528352752700.730.792.231.090.977 Miami 4' plate Peat 93 b 71 1.318abPeterborough"Apt. bldgFactorySd. & si. *58*20a, oa, o48171.211.189 " Water tank Si. clay *30 b,o 31 0.9710a Linkoping 2x3 m plate Si. sand *9 a,o 6.7 1.34b "1.1x1.3m Si. sand *4 a,o 3 1.33plate11 Sunne House Silt &sand*10 b,o 8 1.25DMT-CALCULATED vsOBSERVED. Ave : 1.18-30%+50%Torino 20 Nov 2001 24

DMT-calculated vs observedSETTLEMENTSHAYES, 1990 (°)HAYES J.A. (1990) "The Marchetti Dilatometer and Compressibility"Paper to Southern Ontario Section of Canad. Geot. Soc.Seminar on "In Situ Testing and Monitoring". Sept.Torino 20 Nov 2001 25

Agreement by a large No. of case-historiesMayne P. (2001) "Settlements predicted by SPT and DMT vs settlementmeasured of a 13-story Dormitory Building for Georgia State University inAtlanta", [Prof. Mayne compares settlements predicted by SPT and DMT vsmeasured. "The measured settlement was 9.8 inches. SPT had predicted 1inch (in this case 1 order of magnitude lower). DMT + theory ofelasticity gave essentially the correct answer.Failmezger R., Rom D., Ziegler S.B.(1999) "BehavioralCharacteristics of Residual Soils. SPT?- A Better Approach to SiteCharacterization of Residual Soils using other In-Situ Tests", ASCE Geot.Special Pub. No. 92, Edelen, Bill, ed., ASCE, Reston, Virginia, [5 casehistoriescomparing settlements predictions by SPT and DMT. AtBlacksburg Bypass SPT predicted 100 mmsettlements, while DMT predicted 27 mm leading tochange in design and large savings. Generally SPToverpredicted settlements (in one case by a factor 10)Pelnik, Fromme, Gibbons, Failmezger (1999)."FoundationDesign Applications of CPTU and DMT Tests in Atlantic Coastal PlainVirginia",KCI Technologies, Md, Usa (2000): By DMT a more costeffective design can result compared to using theSPT alone (producing) savings in construction costTorino 20 Nov 2001 26

Sawada S. and Sugawara N. (1995) "Evaluation of densification ofloose sand by SBP and DMT"Tice & Knott (2000) "Cape Hatteras Light Station Relocation" - ASCEOutstanding Civil Engineering achievement for 2000, Geo-Strata Oct.Good agreement was observed between DMTpredictedand measured settlements at the sandy siteunder Cape Hatteras Light StationSteiner W. (1994) "Settlement Behaviour of an Avalanche ProtectionGallery Founded on Loose Sandy Silt", Settlement '94 ASCE Conf. at TexasA&M, [An earthfill on a loose sandy-silt produced settlementssubstantially higher than anticipated based on conventional soil borings. DMTwere then performed. The DMT-predicted settlements agreedwell with observed settlementsWoodward, McIntosh (1993) "Case history : Shallow FoundationSettlement Prediction Using the Marchetti Dilatometer", ASCE AnnualFlorida Sec. Meeting - Sandy site. Use of modulus fromDMT permitted considerable savings vs using datafrom SPT. SPT, for this project, underpredicted themodulusGeopac (1992) "Comparisons of settlements predicted by PMT and DMTin a silty-sandy soil in QuebecLacasse & Lunne (1986) "Dilatometer Tests in Sand". Proc. In Situ '86ASCE Spec. Conf. Virginia Tech, Blacksburg,. Report very goodagreement between DMT-predicted and measuredsettlements under a silos at a sandy site• Real Settlements in Real Jobs• Many of Above Multiple Case-Histories.Global Experience of Investigation Firms/Organizations / Practicing Engns.Torino 20 Nov 2001 27

Accuracy of Settlement predictionsOCR?CPTDMTRange CPT's E/Qc:• CC: Jamiolkowski 2.5 to 25. f(OCR?).• From real settlements Schmertmann initially 1.5 or 2.5,Fahey in Perth stiff sand 13, Fonseca (Asce Oct 01) 4 etc.Compare band amplitude E/Qc vs Hayes:Settlem. prediction bandwidth (max/min) for DMT 3narrower than CPTFailmezger Asce Sep 01 analysed St Dev Settlem. predictionsSDspt= 86% SDdmt= 29% (x3!) "poor practice predictsettlements by SPT rather DMT (Soils testable – no Gravel)"Possible reasons higher accuracy DMT :1. Availability of Stress History parameter Kd2. Wedges deform soil

CLAYc u = undrained strengthg T = unit weightI R = rigidity indexN' = friction angleOCR = overconsolidationK 0 = lateral stress statee o = void ratioV s = shear waveE' = Young's modulusC c = compression indexq b = pile end bearingf s = pile skin frictionk = permeabilityq a = bearing stressIs One NumberEnough???(Mayne 2001)NsptSANDD R = relative densityg T = unit weightLI = liquefaction indexN' = friction anglec' = cohesion intercepte o = void ratioq a = bearing capacitys p ' = preconsolidationV s = shear waveE' = Young's modulusY = dilatancy angleq b = pile end bearingf s = pile skin frictionTorino 20 Nov 2001 29

DMT is a 2-PARAMETER TESTDMT 2 parameters. Permits elementary formof combination M DMT = R M (I D , K D ) E DOne of param (Kd) reflects Stress History!Jamiolkowski (Isopt-1, '88, Vol. 1) : without StressHistory, impossible to select reliable E (or M)from QcLeonards (Asce 88) : …calculating settlementson granular soils based on correlations Penetr.Resistance – Soil Modulus will seriouslyoverestimate settlements if deposit has beenprestressed [Schmertmann 70,Terzaghi 67]…Torino 20 Nov 2001 30

DISTORTIONS due to INSERTIONCONEOK rupture. Butdeformation?Strength „StiffnessWEDGESPT !!!Photographs of distortions in clay.Baligh & Scott (1975) ASCETorino 20 Nov 2001 31

Effect of shape of probe (Cylindr/Flat)on : Arching & Sensitivity to shHughes & Robertson (Canad. 1985) analysed stresses aroundcircular probes in sand :• Behind tip, enormous stress reduction• Ring high residual stresses• "Parasitic" RING obstacle to sh (blinding effect)limits/ makes undetermined sh,sleeveHuang (sand), Canad. 1994 ...sh,sleeve insensitive to sh,insitu & variable.Frost (2001) …"underuse" of fs is related to commonsentiment that fs is unreliable…Flat shape (B:L = 1:7) much less arching.Torino 20 Nov 2001 32

DMTbefore-after for Compaction ControlReasonant vibrocompaction techniqueMDMT (bar)Van Impe, De Cock, Massarsch, MengéNew Delhi (1994)Torino 20 Nov 2001 33

DMT more sensitive to COMPACTIONFrom before-after CPT/DMTs to monitorcompaction∆Mdmt/ Mdmt ≈∆Qc/Qcoften found 2Schmertmann (1986) DYNAMIC COMPACTION ofsand site. MDMT % increase » twice % increase in qc.Jendeby (1992) monitored DEEP COMPACTION in asand fill by VIBROWING. MDMT increase » twiceincrease in qc.Pasqualini & Rosi (1993) VIBROFLOTATION job :"DMT clearly detected improvement even in layerswhere benefits were undetected by CPT".Ghent group (1993) before-after CPTs DMTs to evaluateeffects (–Dsh , Dr) by PILE (Atlas) INSTALLATION"DMTs before-after installation demonstrate more clearly[than CPT] beneficial effects of Atlas installation".Torino 20 Nov 2001 34

FLAT SHAPE MORE REACTIVE TOSTRESS HISTORYJendeby 92 measured Qc & MDMT before andafter compaction of a loose sandfillBEFORE AFTERMDMTQcMDMTQcTorino 20 Nov 2001 35

COMPACTION CONTROLSensitivity of DMT esp. advantageous.In fact, if Qc is used to control compaction:• Since Qc scarcely sensitive to sh• Qc reflects PART of the benefit (Dsh mostlymissed !)• Settlement predicted from Qc aftercompaction too big : +200-300%(Massarsch'94). Consequence : compactmore than necessary waste $If Mdmt is used to control compaction:• Since Mdmt incorporates sh,compaction• Settlement predictions incorporate benefit ofincreased sh. Waste avoided.Scope of compaction: limit settlements.More rational specs on Modulus than Dr(from SPT/Qc) - Schmertmann, 1986Torino 20 Nov 2001 36

a= Mdmt/Qc increases with OCR.Hence a » OCR indicatorFigure: If a loose sandfill is compacted(increase OCR), Mdmt increases faster thanQc (i.e. Mdmt + sensitive to SH)COMPACTIONa = Mdmt/Qc usable as » indicator of OCR :a = 5-10 means NC sanda = 12-24 means OC sandReverse true: wetting a loess (i.e. reduce SH,destructure): a from about 20 to 8 (Devincenzi 01)Torino 20 Nov 2001 37

CC results confirm Kd ++ sensitive toPrestraining than Penetr. ResistanceJamiolkowski & Lo Presti. Atlanta ISC '98CC on Ticino SANDEffect of prestraining : Kd increase was 3 to 7higher than Penetration ResistanceTorino 20 Nov 2001 38

INFLUENCE OF PRESTRAININGON DMT RESULTSCC TEST N. 216 IN TICINO SANDJamiolkowski & Lo Presti. Atlanta ISC '98Torino 20 Nov 2001 39

INFLUENCE OF PRESTRAININGON DMT RESULTSCC TEST N. 241 IN TICINO SANDJamiolkowski & Lo Presti. Atlanta ISC '98Torino 20 Nov 2001 40

DMT FOR LIQUEFACTION• Factors (SH) making DMT + sensitive tocompaction and + accurate settlements :are known to affect "liquefiability"Jamiolkowski's res. group (S.F. 1985):"reliable predictions [of liquefiability] incomplex stress-history deposits require thedevelopment of some new in situ device [otherthan CPT or SPT] more sensitive to the effectsof past stress and strain histories".• Less disruptive insertion in loose sand1. Enter Kd – get CRR.Then Seed method.RECOMMENDEDCURVE2. or Use Kd with Table:Clean sand is safe againstliquefaction (M=7.5 earthquakes) if Kd > :KDSeismicity of the area a max /g Kd,minNonseismic / 1.7Low seismicity 0.15 4.2Average seismicity 0.25 5.0High seismicity 0.35 5.5Torino 20 Nov 2001 41

Evaluating STABILITY of a SLOPEVerify if a Clay SLOPE contains ACTIVE(or old quiescent) SLIP surfacesTorino 20 Nov 2001 42

Examples of Kd » 2 in SLIP SURFACESLANDSLIDE "FILIPPONE" (CHIETI)DOCUMENTEDSLIP SURFACELANDSLIDE "CAVE VECCHIE" (S.Barbara -AR)DOCUMENTEDSLIP SURFACETorino 20 Nov 2001 43

Coefficient of CONSOLIDATION /PERMEABILITY from TflexSTOP Penetration and Monitor sh DecayCh≅7cm2Tflexk=⋅γ CM wTorino 20 Nov 2001 44

DMT for FEM Input PARAMETERSŒSimplest : use MDMT (E » 0.8 MDMT ) +linear elasticity (eg. Hamza, Cairo Metro).Use DMT to calibrate Model Parameters.Must predict correct settlement under simpleload i.e. use settlement by DMT as a virtualload test – w/o doing it.Then use FEM to analyse complicated casesŽ ….Torino 20 Nov 2001 45

Many FEM need G-g DEGRADATION.DMT may help define G-ga) When G-g needed, often Go by SWV (CH, DH)b) Sites have different degrad. shapes.\ Go (1 point) not sufficient. ⇒ 2 nd pointMaugeri S. Louis 1995c) DMT can help : but must define g-range Mdmt.d) Analysed case-histories where DMT predictedwell settlement: in zone most of settlement.Found g = 0.05 to 0.1% ⇒ 2 nd point.Torino 20 Nov 2001 46

g - RANGE of MODULUS by DMTMayne – Insitu 2001, BaliIshihara – Insitu 2001, BaliTorino 20 Nov 2001 47

If forced to guess complete G-g from 1 point1 way: If Go available, downgrade by 2 or 5 toguess operative modulus @ g =0.05-0.1%2 way: Colleague of Mayne (Atlanta) upgradeEd to guess Go (x• 2 or 5). Reason:Errors less important where g (settlements)small. Operative important.Aim not Go! But curve for PC if Go missingGo elegant (unique). But often far from range.Torino 20 Nov 2001 48

Often most of settlement in g >0.05% rangeApprox calculation Lambe & Whitman tankLoad 100 kPa, D=40 to 50 m, Medium sandEssential:correct modulus in operative rangeTorino 20 Nov 2001 49

Ratio Go / Ed (mostly sand)Sully & Campanella 1989Baldi et al. 1989Tanaka et al 1994Hrycyw 1990Lunne Lacasse Rad 1992Tanaka and Tanaka 1998Cavallaro, Lo Presti, Maugeri, Pallara 1999………………………..Baldi Bellotti Ghionna Jamiolkwski Asce Jan 1991Various sands, NC and OC, CC and in situTorino 20 Nov 2001 50

Is MDMT an "operative modulus"?MDMT is very similar to "operativemoduli" recommended by textbooksp. 567 Foundation EngngHandbookWinterkorn & Fang, VanNostrandTYPE OF SOIL E (Kg/cm 2 )Very soft clay 3.5-30Soft clay 20-50Medium clay 40-80Hard clay 70-180Sandy clay 300-400Silty sand 70-200Loose sand 100-250Dense sand 500-800Dense sand & gravel 1000-2000(E ≈ 0.8 M• Operative moduli » moduli that, input in theoryelasticity, ⇒ reasonable Settlem. Predictions• Moduli by Manuals : normalized case-histories.Predicted well for decades real settlem. real cases.• DMT moduli : same range as above (operative)(Oglio loose sand 100 bar, Nile Dense sand 1000)Torino 20 Nov 2001 51

SDMT Test SetupMayne & Martin (1998). "Seismic flat dilatometerin Piedmont residual soils", ISC '98, Atlanta .Torino 20 Nov 2001 52

Why Difficult to get Operative Modulusfrom Qc or Go-Fahey (2001) : when doing CPT, take also Go-Reason : Qc unsatisfact. for settlem. (jump)-But even Go to Operative : jump-Mostly needed : moduli in operative rangeGIf have Go :divide by 2-5? If have Qc :multiply by 2-20?Operative range0.05 to 0.1 %g%Modulus from TRXson NC & OC specimenE(OC)E(NC) 1 (5) 1v.low strainhigh strainYamashita,Jamiolkowski,Lo PrestiAsce Oct 2000-OCR does not affect G at v.low/high g (ratio=1)-OCR mostly affects modulus at operative strain-Help explain why Go, Qc do not reflect OCR-Confirms notion OCR v. important for settlem. –cannot predict by parameters scarcerly f(OCR).Torino 20 Nov 2001 53

Yamashita, Jamiolkowski, Lo Presti Jnl. Asce Oct 2000Torino 20 Nov 2001 54

When use S-DMT : 2 points in G-g?From Go or Qc is a big jump to get operativeBut if have operative (mechan. DMT) whybother to do SDMT to get complete G-g?Moreover: if DMT predicts settlem. ± 30% (or± 50%) need/possible to improve?Two reasons:(1) Complex geometry/ complex loading:require FEM, requires complete G-g(2) Today's schoolboys even for simplegeometry/loading will push button on a PC.May want routinely G-gHowever for everyday practice: Majority(90%?) of problems are simple geometry/loading.Then: Get Mdmt and(a) Use elastic theory + Mdmt(b) Use FEM and G-g with Go=2-5 • Ed.Torino 20 Nov 2001 55

DMT BEST APPLICATIONS• M and Cu profiles• Estimating settlements, deformation• Monitoring soil improvement• Recognize soil type• Verify if a clay slope contains active/old slipsurfacesUseful information also on:• OCR and Ko in clay• Coefficient of consolidation / permeability• P-y curves for laterally loaded piles• Sand liquefiability• Friction angle in sand• (Some info OCR and Ko in sand)Torino 20 Nov 2001 56

FINAL REMARKSIf purpose is to profile hard/soft layers :SPT/CPT more convenient (cost half).Also depends on needed accuracy: not rarelysufficient to know if settlement = 1 cm or 10 or100. Then use SPT/CPT.Use DMT when(a) Settlements important, e.g. strict specs (eg.by railways) or need decide: Piles or Shallow ?(b) If need numbers (at least for deformationparam.) rather than a penetration resistance.Torino 20 Nov 2001 57