A Perspective on Geotechnical Testing: The Details Matter

1. A Perspective on Geotechnical Testing: The Details Matter John T. Germaine Massachusetts Institute of Technology Department of Civil and Environmental Engineering

2. The Question How well are we doing as a profession with regards to the characterization of soils?

3. Outline • Overview of soil testing industry • Establishing quality control • Some example industry data • Specific gravity • Shrinkage limit • Compaction • Hydraulic conductivity • Conclusions and recommendations

4. Laboratory Testing Goals • Diversity in test type • Broad range of materials • Accurate results • Timely delivery • Profitability

5. Testing Considerations • Test methods • Index Tests • Engineering Tests • No correct answer • Extreme variability of natural materials • Huge range in results • Quality control concerns

6. Testing Organizations • Commercial companies • About 1200 • Commercial laboratories • In-house engineering consultants • Small independent laboratories • Government organizations • About 110 • Academic research laboratories • About 180

7. Distribution of Tests • Very informal poll • Three large commercial • One in-house engineering • Test numbers, not revenue

8. Distribution Minus Index • Significantly different distributions • Large number of strength tests • In-house QC type testing

9. Quality Control Tools • ISO Certification • Management, documentation and training • ASTM D3740 • Guidance for technical, documentation and training requirements • NICET • Certifies technician capabilties • AMRL laboratory assessment • Certifies conformance to standard • AMRL proficiency sample testing • Sends out uniform subsamples • Evaluates collective test results

10. Documented Protocols • Facilitate communication • Product uniformity • Solidify professional practice • Expand domain of expertise • Improve product quality • Formal Standards • ASTM • AASHTO • BS • In-house procedures

11. Quality of a Test Method • Precision and Bias • Bias: deviation relative to true value • Precision: variation for given test method • D18 standards have no Bias! • Quantities generally do not have a “correct” result • Use standard caveat statement in all standards

12. Quantifying Precision • ASTM Standard E691 • Round Robin or Interlaboratory • Ruggedness testing • Impact of allowable variables • > 6 laboratories • Triplicate testing in each lab • Acceptable range • 2.8 x standard deviation • Repeatability for single operator • Reproducibility for between labs • Limited to independent observations

13. l: Classification and Index • Simple equipment • Considerable labor • Technical skill and finesse • Difficult to check results • Rely on consistency and correlations

14. Example: Specific Gravity Test • AMRL proficiency program • Method: ASTM D854 • 542 Laboratories • Samples 157 and 158 • Distributed uniform dry powder • One test on each sample

15. Sample 157 <200 67 % < 2m 29 % Gs 2.644 LL 29 PI 13 USCS CL AMRL Sample Specifics • Sample 158 • <200 62 % • < 2m 27 % • Gs 2.645 • LL 28 • PI 13 • USCS CL 2008 Proficiency Testing Program

16. Specific Gravity Results • Huge range in results • Within laboratory correlation • Systematic error in procedure • 1995 study same variability Specific Gravity of Sample 157, (gm/cm3) Specific Gravity of Sample 158, (gm/cm3)

17. Specific Gravity Results • Eliminate outliers • Wide distribution • Bias towards low values Number of Observations • Useful range 0.01 • ASTM • Repeatability • 0.02 • Reproducibility • 0.06 Specific Gravity, (gm/cm3)

18. Example: Shrinkage Limit Test • Comparison of Wax and Hg Method • AMRL proficiency program • Method: ASTM D4943 & D427 (old) • About 50 Laboratories • Samples 159 & 160 and 161 & 162 • Distributed uniform dry powder • One test on each sample

19. Sample 159 / 160 <200 89 / 83 % < 2m 39 / 37 % Gs 2.704 / 2.699 LL 43.0 / 43.2 PI 20.8 / 20.9 USCS CL AMRL Sample Specifics • Sample 161 / 162 • <200 65 / 46 % • < 2m 24 / 20 % • Gs 2.733 /2.694 • LL 24.8 / 23.7 • PI 10.2 / 10.1 • USCS CL 2009 & 2010 Proficiency Testing Program

20. Shrinkage Limit: Wax Method • Huge range in results • Within laboratory correlation • Systematic error in procedure

21. Shrinkage Limit: Wax Method • Wide distribution • Second year improvement • Distribution skewed to higher values

22. Shrinkage Limit: Hg Method • About the same range as Wax method • Within laboratory correlation • Systematic error in procedure

23. Shrinkage Limit: Hg Method • Clear difference between each year • Most labs in narrow range • Serious outliers

24. Shrinkage Limit: Summary • Wax gives lower values • Wax method has more scatter • Average values capture subtle differences

25. ll: Laboratory Compaction • Simple equipment • Calibration of automatic hammers • Energy transfer • Material processing very important • Technical skill • Interpretation of results

26. Example: Standard Proctor • AMRL proficiency program • Method: ASTM D698 • Samples 157 and 158 • 963 Laboratories • Report only wopt and gmax

27. Compaction Results • Water Content • Weak correlation • Processing issues • 157 higher • Serious outliers • Unit Weight • Better correlation • Technique differences • 157 lower 158 Opt. Water Content, % 157 Opt. Water Content, % 158 Max. Dry Unit Weight, lbf/ft3 157 Max. Dry Unit Weight, lbf/ft3

28. Compaction Results • Outliers Removed • Water Content • Broad distribution • Subtle difference • Unit Weight • Narrow center band • Clear shift in average • Symmetrical tails Number of Observations Opt. Water Content, % Number of Observations Max. Dry Unit Weight, lbf/ft3

29. Compaction Results • Considerable scatter • Clear outliers • No trend Dry Unit Weight, lbf/ft3 • Unlikely results • Impossible results Water Content, %

30. Compaction Results • wopt =10.7 % • gmax =122.6 lbf/ft3 • Field specification • +/- 2 % wc • 92 % R.C. Dry Unit Weight, lbf/ft3 • Field specification • Including 2 Std. Dev. Water Content, % AMRL Proficiency Sample 158

31. lll: Hydraulic Conductivity • Widest range of any parameter • Extreme equipment demands • Little automation • Expertise more than finesse • Attention to detail • QC equipment

32. Example: Establishing Precision • ASTM D5080 • Craig Benson conducted study • ISR ML, CL, and CH material • Provided compacted test specimens • 12 laboratories • 3 tests per laboratory

33. ISR Sample Specifics • ML Sample • <200 99 % • < 2m 8 % • LL 27 • PI 4 • USCS ML • Vicksburg silt • CL Sample • <200 89 % • < 2m 31 % • LL 33 • PI 14 • USCS CL • Annapolis clay • CH Sample • <200 96 % • < 2m 46 % • LL 60 • PI 39 • USCS CH • Vicksburg clay ASTM ISR managed 15,000 lbs of each soil NSF, FHWA, and private sponsorship Started 1993 7 Precision statements

34. Hydraulic Conductivity Results • Variable Scatter with in labs • Two outlier labs • Some labs very consistent • Log std. dev. fairly good Hydraulic Conductivity, (cm/s) (10-6) Laboratory Number

35. Hydraulic Conductivity Results • ML (x10-6) • naturallog • 1.21.1 • 0.8-1.60.8-1.5 • CL (x10-8) • 3.83.7 • 3.2-4.43.2-4.4 • CH (x10-9) • 3.62.6 • <0-8.21.3-5.2 Avg. S. D. Hydraulic Conductivity, cm/s Laboratory Number

36. Hydraulic Conductivity Results • Log provides better representation • Equip. tuned to 10-7 • < one sign. digit • Real problems for low permeability Hydraulic Conductivity, (cm/s) Laboratory Number

37. lV: Consolidation and Shear • Significant advances in equipment • Extensive automation • Technical expertise • Sample quality and handling • Testing decisions based on soil behavior • Essentially no precision data

38. Conclusions • QC tools are available • Equipment adequate • Too much scatter • Causes of scatter are not obvious • No data for consolidation or strength • Substantial room for improvement

39. Recommendations • Formal protocols for every test • Technician training • Consistency evaluation of results • Reference material testing • In-house databases • Participation in ASTM

40. Acknowledgements • Friends associated with ASTM • Ron Holsinger; AMRL • Craig Benson; U of Wisconsin