CPT Applications - Liquefaction 2

Transcription

1 CPT Applications - Liquefaction 2 Peter K. Robertson CPT in Geotechnical Practice Santiago, Chile July, 2014

2 Definitions of Liquefaction Cyclic (seismic) Liquefaction Zero effective stress (during cyclic loading) Flow (static) Liquefaction Strain softening response

3 Flow (static) Liquefaction Strain softening response in undrained shear Trigger mechanism required Static shear stress greater than minimum (liquefied) undrained shear strength Kinematic mechanism required Uncontained flow Contained deformation

4 Schematic undrained response of saturated, contractive sandy soil After Olson & Stark, 2003

5 Flow chart to evaluate liquefaction After Robertson, 1994

6 Flow (static) liquefaction Sites defined as: steeply sloping ground Steeply sloping (> 5 degrees) or earth embankments (e.g. dams) Sequence to evaluate flow liquefaction: 1. Evaluate susceptibility for strength loss 2. Evaluate stability using post-earthquake shear strengths 3. Evaluate trigger for strength loss If soils are susceptible, and instability possible, it is often prudent to assume trigger will occur

7 Evaluate susceptibility for strength loss Soils must be strain softening in undrained shear Strain softening soils are CONTRACTIVE at large strains Can we identify soils that are either CONTRACTIVE or DILATIVE (at large strains) using CPT?

8 State parameter in coarse-gained soils (sands) (+) Loose CONTRACTIVE Loose (-) Dense DILATIVE y Dense State Parameter after Jefferies and Been, 1985 Relative State Parameter index after Boulanger, 2003

9 State Parameter from CPT Jefferies and Been (2006) summarized ~30yrs of research related to evaluation of liquefaction using a Critical State Soil Mechanics (CSSM) approach Problem is complex & depends mainly on: in-situ stresses, shear stiffness, shear strength, compressibility & plastic hardening Requires combination of in-situ tests (SCPT) and lab testing (reconstituted samples to get CSL) Based on extensive calibration chamber test results, field results (frozen undisturbed samples), lab testing & numerical simulation estimate of state from CPT

10 State parameter & clean sand equivalent Increased resistance to loading Increased resistance to loading DILATIVE CONTRACTIVE Q tn,cs = 70 at y = Based on CSSM theory & CC Based on liq. case histories ~ log Q tn,cs Robertson, 2013

11 Dilative/Contractive boundary CPT-based boundary DILATIVE Jefferies and Been (2006) suggested that when Y > soils are strain softening at large strain CONTRACTIVE Y = Q tn,cs ~ 70 After Robertson, 2010

12 Case histories flow liquefaction Many publications (over past 25 years) that have studied case histories: e.g. Seed (1987), Seed & Harder (1990), Stark & Mesri (1992), Wride et al (1999), Olson & Stark (2002), Jefferies and Been (2006), Olson & Johnson (2008) and Robertson (2010) Some uncertainty with some seismically triggered cases that involved either strength loss (flow liq.) or cyclic liq./modility (lateral spreading)?

13 Case histories flow liquefaction Case histories involve soils that are: Very young Very loose Non- or low-plastic ~78% recent fills ~40% hydraulically placed ~16% tailings ~50% triggered by earthquakes several triggered by very minor disturbance Failures tend to occur without warning, are rapid and can flow large distance

14 Stava, Italy, 1985; 268 deaths, 190,000m 3 No warning Kolontar, Hungary,2010, 700,000m 3 Very fast Travel long distances

15 Case histories - flow-liquefaction DILATIVE Case histories with CPT Nerlerk (sand) 19,20,21 Jamuna (sand) - 34 Fraser River (silty sand) - 27 Sullivan mines (silty tailings) - 35 Northern Canada (silty clay) 36 L. San Fernado Dam (silt) 15 CPT data in critical layers +/- 1 sd. CONTRACTIVE All case histories fall in contractive portion of CPT SBT chart After Robertson, 2010

16 Tennessee Valley, Kingston Coal Ash Failure million m 3 flowed ~1 km December, 2008 Post-slide April, 2008 Pre-slide After AECOM:

17 Kingston failure sequence After AECOM:

18 Kingston failure sequence After AECOM:

19 Kingston failure sequence After AECOM:

20 Kingston failure sequence After AECOM:

21 Cross section through slide After AECOM:

22 Cross section through slide After AECOM:

23 Cross section through slide After AECOM:

24 Laboratory test results (TVA) Most of the Ash and Slimes are contractive and strain softening at large strain

25 Fine-grained - Slimes High water content: 40 to 140% Very high Liquidity Indices: 3 to 7 Sensitivity, St > 30 (i.e. brittleness > 0.98) Low undrained shear strength at large strain When water content > Liquid Limit Sensitivity can be high

26 Post-failure CPT (CPT ) Fly Ash Weakest slimes Slimes Clay Alluvium Data from AECOM: Confirmed with adjacent borehole & samples

27 Post-failure CPT (CPT ) Peak Contractive Dilative Fly Ash Weakest slimes Slimes Clay CPeT-IT software:

28 TVA Fly Ash Failure CPT results (+/- 1 standard deviation) in coarse- grained fly ash and fine-grained (slimes) Weakest slimes

29 Ground Improvement example After Campanella et al., 1983

30 Ground Improvement example Before After 5 7 CPT, qt (MPa) Peak, su/s v CRR (CTX) 0.15 >0.29 SPT, N Silt went from soft normally consolidated (before) to stiff overconsolidated (after ground improvement) Contractive to Dilative

31 Contractive/Dilative boundary Case histories suggest that when Qtn,cs < 70 soils are potentially susceptible to strain softening and hence, flow liquefaction Criteria of Qtn,cs < 70 slightly conservative since mean + one standard deviation of CPT data fall below this criteria if using mean values, Qtn,cs < 50 to 60 maybe more appropriate but very thin weak layers may not be adequately captured by mean values

32 Contractive/Dilative boundary CPT-based SBT chart provides a reasonable estimate of soils that are likely either CONTRACTIVE or DILATIVE (at large strains) Are all CONTRACTIVE soils at risk for flow (static) liquefaction? depends on brittleness (sensitivity), strain to peak strength and static shear stress

33 Generalized soil behavior (undrained shear NC fine-grained soils) Both CONTRACTIVE Low PI High PI Low PI soils tend to have smaller strain to peak strength and tend to be more brittle After Ladd et al, 1977

34 Brittleness versus undrained strength ratio Regardless of fabric and direction of loading High Brittleness τ Su(p) Su(min) γ Low Brittleness Range for NC soils After Yoshimine et al, 1999

35 Liquefied undrained strength ratio Contractive Dilative Most case histories have a calculated liq. strength ratio su/s v ~ 0.1 or slightly smaller After Robertson, 2010

36 Liquefied strength ratio relationships based on normalized CPT tip resistance Robertson, 2010 Clean sand equivalent Low Brittleness Dilative High Brittleness Olsen & Stark (2002) Modified from Olsen & Stark (CGJ, 2002)

37 Generalized CPT Soil Behaviour Type CPT Soil Behaviour A: Drained-dilative A C B D B: Drained-contractive C: Undrained-dilative D: Undrained-contractive

38 Regions of potential liquefaction Cohesionless soils (A1 & A2) - Evaluate potential behavior using CPT-based case-history liquefaction correlations. A1 Cyclic liquefaction possible depending on level and duration of cyclic loading. A2 Cyclic liquefaction and post earthquake strength loss possible depending on loading and ground geometry. Cohesive soils (B & C) Evaluate potential behavior based on in-situ or laboratory test measurements or estimates of monotonic and cyclic undrained shear strengths. B Cyclic softening possible depending on level and duration of cyclic loading. C Cyclic softening and post earthquake strength loss possible depending on soil sensitivity, loading and ground geometry.

39 Summary Strong link between clean sand equivalent cone resistance (Qtn,cs) based on seismic case histories and State Parameter (y) Contractive soil response when Qtn,cs < 70 (based on mean +1 SD) Young, very loose, non- to low plastic soils tend to be more brittle than older, denser, more plastic soils

40 Summary Since case histories show that failures can occur without warning, are rapid and can flow large distances Observation approach may not work If soils are susceptible, and instability possible (i.e. FS << 1.0), it is often prudent to assume that a trigger will occur - unless you can design away all possible triggers