You are on page 1of 9
SURVEY OF TRIAXIAL AND DIRECT SHEAR TESTING METHODS Robin Fell, School of Civil Engineering, University of NSW Maureen Ng Ove Arup & Partners, Hong Kong 1. Introduction Methods for carrying out triaxial and direct shear testing. of clays to determine peak effective strength are not covered by Australian Standards. Most laboratories follow procedures which are detailed in books or technical papers and when carried out correctly, result in good quality results, However, in reviewing a number of landslide and dam projects over the last eight years, the first author has observed that not all laboratories rigorously follow these procedures, and errors in estimates of strength can result With a view to ascertaining how widespread these problems are, and to raise awareness of the issues, the authors surveyed Australian laboratories to determine the details of their test equipment, test procedures, and reporting of results, This paper summarises the results of the survey, and discusses some of the common errors and deficiencies in testing being carried out. The emphasis is on test procedures, particularly where these can lead to errors. 2. The Survey Method and General Comments All Australian laboratories likely to be carrying out triaxial and direct shear testing were identified through their NATA registration, and from personal knowledge. Organisations were included in the survey even when it was considered that it was unlikely that they did such testing. In all there were 70 questionnaires sent out, resulting in 21 detailed replies. Another 17 organisations replied indicating that they did not do such testing. Few of those who failed to respond would do much, if any, \iaxial or shear strength testing, so the survey probably represents about 70% to 80% of Australian commercial testing (including government departments). The survey document consisted of an 11 page questionnaire, with 42 questions relating to equipment, test ‘methods and reporting. Respondents were also asked to attach examples of detailed testing and reporting. ‘The data was assembled and assessed by the second author for a BE(Civil) thesis (Ng, 1993). To maintain confidentiality, laboratories which responded have been identified only by a letter of the alphabet. The first author hhas the "key", and each laboratory has been advised which letter has been assigned to them. Some more detailed comments have been given to individual laboratories where this was considered likely to be useful. Table 1 details the number of tests carried out by each laboratory in a typical year. In can be seen that there are only a relatively small number of tests carried out each year, with only about 5 laboratories carrying out more than 50 tests in a typical year. What is not apparent from Table 1, is the relatively large number of organisations who indicated that they have stopped doing tests because of changes in the structure of government. Table 1: Number of tests performed for projects in ‘a typical year Laboratory Triaxial Tests Direct Shear Tests A 3010 40 60 t0 80 B None 20 c 10 None D 8 2 E 10.15 None F 30 20 G 15 50 H 6 None 1 10 None 1 None recently None recently K 10 t0 20 None L 2 days IS N ef D FC Eguil GEIS Y by F Fc? Equil? Is N 7 G H Equil GEIS Y cy J FC? Overnight GEAV N E M Fc? Overnight GEIS N E Q Fe? Equil GE N E R Is Equil? GEAV Y 2 s FC 133 days GEISAV Y a LEGEND FC First consolidation stage GE Selected by geotechnical engineer HH Hanger or nominal small load AV Arbitrary value IS In-situ stress c, Based on coefficient of consolidation Equil To equilibrium in swell or consolidation E Based on experience NOTES (1) Use 0.003 to 0.005 mm/minute (2) Shear boxes are usually 60 mm x 60 mm or 75 mm x 75 mm 68 Australian Geomechanics ~ October 1994 SHEAR STAESS - kPa SHEAR STRESS - kPa SHEAR STRESS - kPa 80 60 40 20. 5 0 On 400KPa IN 200KPa On 100KP2 0 2 4 6 & w 21 3t CUMULATIVE STRAIN ~ % NORMAL STRESS Gn = 400kPa 4143 454749 ot 60 62 64 66 68 70 72 82 84 86 88 90 92 CUMULATIVE STRAIN ~ % NORMAL STRESS Gn = 200kPa foo «102 104 308 108 110 12 Fig CUMULATIVE STRAIN ~ 4% NORMAL STRESS Gn = 100kPa LEGEND © CORRECT RESIDUAL STRESS x RESIDUAL STRESS INTERPRETED BY LABORATORY 4 -REVERSALS jure I: Residual strength testing from laboratory $ - shear stress vs strain plots Australian Geomechanics ~ October 1994 69 10. 3 sg LABORATORY INTERPRETATION SHEAR STAESS - KPa ‘CORRECT INTERPRETATION Las ° 100 200 300 400 NORMAL STRESS ~ kPa LEGEND @ CORRECT RESIDUAL STRESS X RESIDUAL STRESS INTERPRETED BY LABORATORY CORRECT INTERPRETATION: c’ = KPa os LABORATORY INTERPRETATION: c' = OkPa Figure 2: Residual strength testing from laboratory $ - shear stress vs normal stress plots (a) Failure criteria adopted With direct shear testing, there is only the shear stress vs strain (or displacement) plot from which to establish failure. Unfortunately, the curves are often irregular, with “false” peaks, or peaks followed by a slight reduction, then gradual increase duc 10 interference between the (wo halves of the box, or smearing of soil onto the facing surfaces of the box resulting in an increased (not decreased) area as the sample shears, This calls for careful interpretation in the data provided. Head (1986) Vol 3, pp. 761 and 974, gives definitions of the failure criteria, (e) Residual strength determination ‘The residual strength of a soil is the minimum shear strength (c’g, ’g) achieved at large displacements. In a direct shear test, itis achieved by shearing the sample to the limit of travel of the shearbox, then returning the sample to the starting position, rehearsing and repeating the process until a constant (minimum) shear strength is obtained. As pointed out by Skempton (1985) this may involve many cycles of shearing of the shearbox, and total displacements of up to 500 mm may be required. It is unlikely that residual strength will be achieved in the first cycle of shearing, and in any case one needs to check if it has been reached by a second (and preferably third) cycle subsequent to the minimum being reached. As part of the survey, laboratories were asked to give definitions of residual strength. These are reproduced in Table 4, with the authors’ remarks as to their correctness. The Inboratories were also asked to detail how they would obtain the residual strength in the direct shear and this has bbeen used in preparing the remarks in Table 4 It can be seen that laboratory A has an incorrect definition, since it uses the lowest shear stress in the first cycle of shearing only. Laboratory S has the correct definition, but there was an error in interpretation of one set of data. This is shown in Figure 1. The laboratory has used the strength at limit of travel of a cycle (which is influenced by interference between the halves of the box), rather than the minimum strength. This results in an overestimation of the residual strength by 2° (9° instead of 7°) as shown in Figure 2. “The first author has seen the same error in results from other laboratories. ‘The problem seems to be that the laboratory personnel do not understand what is meant by residual strength and the geotechnical personnel involved either don't know or don't check. Several other laboratories had imprecise definitions, and they would be wise to clarify definitions and procedures 70 Australian Geomechanics ~ October 1994 Table 4: Definition of residual strength Laboratory Definition of Residual Strength Remarks A Lowest shear stress value after the peak shear stress Incorrect; shearing ince it implies only one stage of Ata given normal load, the irreducible strength reached at large shear strains applied slowly ‘enough to effectively ensure drained condition on a failure surface with a form that simulates that which would occur in field at large strains (ie. reasonably planar over the scale of the shear box sample Correct Strength measured after large displacements on a shear plane (up to I m may be required for some clays to reduce their strength to residual) Correct (essentially) but not defined in laboratory testing terms Achievement of “constant” stress condition following extended shear displacement (min 80 mm) Correct (essentially) ‘When shear stress is constant after multi- reverse testing Correct (essentially) attained. This is the ‘steady state’ or ultimate value which is maintained as displacement continues. I Value at which load levels off, ie, as sample | Potentially correct, but imprecise, since it shears, shear stress neither increase or decrease | could be applied to only one stage Residual strength is the strength of the First half not really correct, second M materials after the peak value and dilated of the | potentially correct but not precise lowest strength of the material Q Strength after a large displacement Correct (essentially), but imprecise, and not defined in laboratory testing terms R ‘The constant value after peak shear strength | Correct (essentially) but imprecise and not has been achieved defined ‘The shear resistance that a sample of soil can | Correct, maintain when subjected to large shear s displacement after peak shear strength has been Table 5 details some of the methods used (0 hasten the achievement of residual stength. It will be seen that half the laboratories leave the current normal load in place on rewinding, and use the slow rate of shearing to rewind, “The first author's experience is that it is better to unload the sample (as laboratory G does) because there is less “wearing” or squeezing of the soil from the shear surface, and hence less likelihood of interference between the halves of the box. The first author also favours quick rewinding because of the time saved and there does not appear to be a loss of quality in the results. The use of precutting and polishing of the shear surface (e.g. on a lass plate) is acceptable but not usually necessary if one adopts fast rewinding. 5. Use of Staged Testing ‘Staged testing, where the one sample is used for say three consolidation stresses (triaxial) or three normal stresses, (direct sheat) is quite widely used, The advantage is that only one sample has to be cut to size, and saturated, so. there is less technician time and less equipment time. ‘The disadvantage is that if the soil is strain weakening, or if the staging affects the pore pressure response, incorrect results can be obtained. In particular, higher effective ‘cohesion (c’) (and lower effective friction angle ¢’) may result than the true properties. Australian Geomechanics ~ October 1994 n Table 5: Direct shear testing - some details of procedures to obtain residual strength Laboratory Normal Load Rate of Special During Rewinding Rewinding Techniques A GE,CNL or SL 'S (sometimes Q) FW B CNL s cs D CNL s None F CNL s MR G SL. Q FW, CS J CNL Q FW M CNL Ss None Q CNL s cs, P R GE Q None 5 CNL Q FW LEGEND GE As determined by geotechnical engineer FW Fast winding back and forth CNL Current normal load CS Precutting the shear surface SL Small load P Polishing the shear surface s Slowly (usually at same rate as forward shearing) Q Quickly MR — Multiple reversal ‘There is some evidence that this is a problem in some of the testing forwarded as part of the survey. For example the following results of triaxial testing were obtained: Table 6: Results of staged triaxial testing Laboratory Soil cle E ‘ML sandy clayey silt | 18 | 27 CH silty clay | 27 CH silty clay 7 | 26 1 2 30 | 30 2 26 | 27 ? 27| 30 P ‘ML sandy sitt 12 | 23 Q CL sandy clay 19 | 25 CH silty clay 12] 21 CL sandy clay 2 [27 It can be see than in a number of cases, particularly laboratories I and Q quite high effective cohesions were obtained. As pointed out by Lade (1986) and others, effective cohesion should be quite small, provided one allows for curvature of the envelope. rR In the cases shown in Table 6, there was no particular evidence of curvature, and strain rates and saturation seemed reasonable, so the high c’ may be due to staged testing thas been the first author's practise to use staged testing but, where high care obtained, to check with separate samples tested at higher consolidation stresses without staging, ‘The same problems can occur for direct shear testing as for triaxial 6. Data Provided to the Client Table 8 details the data which is supplied in the report by the various laboratories, It will be seen that some laboratories are failing to provide sufficient details of the testing procedures to allow critical review of the results If one is presented with a result, say c’ = 32 kPa, 6 = 25° for a clay of high plasticity, one will be sceptical of the result. If all data 1 to 11 is provided, one can check through the results and be satisfied either thatthe result is acceptable, or that, for example, strain rates were too high, the sample not properly saturated, staged testing used etc. At least it is usually possible to salvage some useful data from well reported data, even if the overall, result is wrong. Those laboratories which do not provide all the data are encouraged to do so in the future. ‘Australian Geomechanics — October 1994 ‘Table 7; Triaxial and direct shear testing - information supplied to client Laboratory Triaxial Direct Shear A 1,2,3,4,5.6,10,11 1,12,13,10,11 B NIA 1,12,13,10 c 123,456 NIA D 123,4,5.6,7,8,9,10,11 1,12,13,8,10,11 E 123,4,5.6.7,11 NIA F 123,4,5.6.10 42 G 789 1,12,13,10 H NA I 8,9,10 N/A I 1,12,10 K 8,9,10,11 NIA L 3,4,5.6,7,8.9,10,11 NIA M 1,23,69,10 1,12,13,10,11 N 124,5.6.7,9,11 NIA ° NA P 8,9,10,11 NIA Q 42 R 1 s 12,4,5,6,10,11 1,12,13,10,11 T 1,2,6,10 NA u 1,2,4,5,6,7 NIA LEGEND c.g Mohs Circle plot Stress path plot Pore pressure vs strain Volume change vs strain Deviator stress vs strain ‘Major principal stress vs strain 7. Acknowledgements The authors would like to thank all those laboratories who responded to the questionnaire References Bishop, A.W. and Henkel, DJ. (1957, 1971). The measurement of soil properties in the tiaxial test - Edward Amolé. Fell, R. and Jeffery, RP. (1987). Determination of drained shear strength for slope stability analysis, in Soil Instability and Stabilisation (eds. BF. Walker and R. Fell), Balkema, Rotterdam. 8. Coefficient of consolidation 9, Degree of saturation (B value) 10. Strain of displacement rate 11. Failure criterion adopted 12. Shear stress vs normal stress 13, Shear stress vs strain or displacement Head, KH. (1980, 81, 85). Manual of Soil Laboratory Testing, Vols. 1, 2 and 3. Pentech Press. Lade, P.V. (1986). Advanced triaxial testing of soils. University of Sydney, School of Civil and Mining Engineering. Ng, M. (1993). Triaxial and direct shear testing - some common errors, School of Civil Engineering, UNSW, BE Thesis. Australian Geomechanics — October 1994 3

You might also like