CHAPTER 2THEPROCESSFORSOILANDROCKPROPERTYSELECTION2.1INTRODUCTIONMost geotechnical engineeringtextbooks provide information related to themechanics ofconductingfield and laboratory tests to obtain soil and rock properties.In addition, the American Society forTesting and Materials (ASTM) and American Association of State Highway and TransportationOfficials (AASHTO) standards provide excellent guidance related to the specific procedures forperforming the actual field and laboratory tests.There are, however, few resources that provideguidance tothedesign professional related to arational process for selecting appropriate criticallocations in the geologic deposit and then developing a specific laboratory and field testing programto obtain soil and rock properties appropriate for design.The goal of this chapter is to describe aprocessthathasbeenusedonavarietyof largeandsmall projectsto integratethevariousdecisionsteps necessary to arrive at the final design parameters.In addition to describing this step-by-stepformal process, guidance is provided on the: (1)appropriate use of correlations to aid in engineeringproperty selection, and (2)use of the Observational Method to refine and improve selected soil androckpropertiesused indesign.2.2PROCESSOFSOILANDROCKPROPERTYSELECTIONA rational approach for selecting soil and rock properties for engineering design can be summarizedas a logical twelve-step procedure that encompasses the general activities of site investigation andfield testing, laboratory testing and interpretation, and engineering design.This step-by-step processis presented on the flow chart in figure 1.A brief description of each step of this process ispresented below. More extensive discussion and the methods used to implement these steps areprovided throughout the remainder of this document.SiteInvestigationandFieldTestingReview Available Information: The best place to start the process of material propertyselectionisto reviewany andall informationthat maybeavailable.There are severalsources for this information, many of the sources being in the public domain and readilyavailable at modest expense.Identify Required Material Properties:No investigation should be initiated without specificgoals being established that are related to design and construction issues that must beconsidered (i.e.,performance requirements),engineering properties that are needed, and thetypeof structurethatistobeconstructed.4
4 CHAPTER 2 THE PROCESS FOR SOIL AND ROCK PROPERTY SELECTION 2.1 INTRODUCTION Most geotechnical engineering textbooks provide information related to the mechanics of conducting field and laboratory tests to obtain soil and rock properties. In addition, the American Society for Testing and Materials (ASTM) and American Association of State Highway and Transportation Officials (AASHTO) standards provide excellent guidance related to the specific procedures for performing the actual field and laboratory tests. There are, however, few resources that provide guidance to the design professional related to a rational process for selecting appropriate critical locations in the geologic deposit and then developing a specific laboratory and field testing program to obtain soil and rock properties appropriate for design. The goal of this chapter is to describe a “process” that has been used on a variety of large and small projects to integrate the various decision steps necessary to arrive at the final design parameters. In addition to describing this step-by-step formal process, guidance is provided on the: (1) appropriate use of correlations to aid in engineering property selection; and (2) use of the Observational Method to refine and improve selected soil and rock properties used in design. 2.2 PROCESS OF SOIL AND ROCK PROPERTY SELECTION A rational approach for selecting soil and rock properties for engineering design can be summarized as a logical twelve-step procedure that encompasses the general activities of site investigation and field testing, laboratory testing and interpretation, and engineering design. This step-by-step process is presented on the flow chart in figure 1. A brief description of each step of this process is presented below. More extensive discussion and the methods used to implement these steps are provided throughout the remainder of this document. Site Investigation and Field Testing • Review Available Information: The best place to start the process of material property selection is to review any and all information that may be available. There are several sources for this information, many of the sources being in the public domain and readily available at modest expense. • Identify Required Material Properties: No investigation should be initiated without specific goals being established that are related to design and construction issues that must be considered (i.e., performance requirements), engineering properties that are needed, and the type of structure that is to be constructed
Review available subsurface information and develop preliminary model of subsurface conditionsIdentify material properties required for design and constructability and estimate scope of field programPlan site exploration and field test programConduct field investigations and field testingPerform sample descriptions and laboratory index tests?Summarize basic soiVrock data and develop subsurface profileAreresultsNoconsistentwithpreliminarymodel?YesReview design objectives and inifial resultsPhase II Investigation (if needed)YesAre there additionaldata needsNoSelect representative soi/rock samples and details of laboratory testingConduct laboratory testing Review quality of laboratory test data and summarizeNoAre resultsconsistent and validYesIs a Phase IIYesInvestigationnecessary?NoSelect material properties and finalize subsurface modelaringaPerform design and consider constructability issues<Figure 1. Soil and rock property selection flowchart.5
5 Review available subsurface information and develop preliminary model of subsurface conditions Conduct laboratory testing Site Investigation and Field Testing Laboratory Testing and Test Interpretation Engineering Design Identify material properties required for design and constructability and estimate scope of field program Plan site exploration and field test program Conduct field investigations and field testing Perform sample descriptions and laboratory index tests Summarize basic soil/rock data and develop subsurface profile Are results consistent with preliminary model? Review design objectives and initial results Are there additional data needs Select representative soil/rock samples and details of laboratory testing Yes No Review quality of laboratory test data and summarize Select material properties and finalize subsurface model Are results consistent and valid Is a Phase II Investigation necessary? Yes No Perform design and consider constructability issues Phase II Investigation (if needed) Yes No Yes No Figure 1. Soil and rock property selection flowchart
Plan Site Investigation:Historical information, which will provide anticipated subsurfaceconditions, coupled with knowledge of the specific design will allow an efficient site-specific investigation strategy to be developed.Contingency plans should be consideredbased on anticipated variabilities in subsurface conditions.Sampling intervals should beidentified and an in situ testing program should be developed.Conduct Site Investigation and Field Testing:Once the investigation strategy is developedit is ready to implement.Findings should be communicated to the geotechnical designengineer during the field work and modifications to the number and types of samples andtesting should be determined, as required.Describe Samples:Results from the field investigation program and subsequent laboratoryidentification of samples should be compared to the anticipated conditions based onhistorical information.Selected laboratory samples can be reviewed by the design engineertoobtainfirst-hand observations.These samples shouldbeusedforperforming simplelaboratoryindextests.Develop Subsurface Profile:Using results from the field investigation and the laboratoryindex tests, a detailed subsurface profile should be developed by the geotechnical designengineer.It is helpful at this step to review the initial site investigation objectives andexpectations to be assured that the materials are consistent with expectations.Review Design Objectives:An on-going evaluation offieldand availablelaboratory datarelative to the design objectives should be performed during the implementation of the siteinvestigation.If adjustments are needed or if additional data needs are identified.procedures should be initiated toobtain the necessaryinformation.Laboratory Testing and Test InterpretationSelect Samples for Performance Testing:Prior to initiating the project-specific laboratory-testing program, the design engineer should review the recovered samples and confirm thetesting that needs to be conducted (i.e., type, number, and required test parameters).Ifpossible, selected samples should be extruded in the laboratory and reviewed by the designengineer.ConductLaboratoryTesting:Oncethe sampleshavebeenreviewed andthetestingprogramis confirmed, it is time to continue the index tests and initiate the performance-testingprogram (with index test correlation for quality assurance).Preliminary results should beprovided to the design engineerfor review.ReviewOualityof LaboratoryData:Ifthedata and interpretedlaboratorytestresults arenotconsistent with expectations or if results indicate that the sample was disturbed, it isnecessary to review progress and make adjustments. On some projects, results at this stagecan be used to plan and initiate a more detailed and focused phase of investigation.Aphased investigation approach is particularlyhelpful on large projects and in cases wherethere are many unknowns regarding the subsurface conditions or specific projectrequirements prior to conducting theproposed site investigation program.6
6 • Plan Site Investigation: Historical information, which will provide anticipated subsurface conditions, coupled with knowledge of the specific design will allow an efficient sitespecific investigation strategy to be developed. Contingency plans should be considered based on anticipated variabilities in subsurface conditions. Sampling intervals should be identified and an in situ testing program should be developed. • Conduct Site Investigation and Field Testing: Once the investigation strategy is developed, it is ready to implement. Findings should be communicated to the geotechnical design engineer during the field work and modifications to the number and types of samples and testing should be determined, as required. • Describe Samples: Results from the field investigation program and subsequent laboratory identification of samples should be compared to the anticipated conditions based on historical information. Selected laboratory samples can be reviewed by the design engineer to obtain first-hand observations. These samples should be used for performing simple laboratory index tests. • Develop Subsurface Profile: Using results from the field investigation and the laboratory index tests, a detailed subsurface profile should be developed by the geotechnical design engineer. It is helpful at this step to review the initial site investigation objectives and expectations to be assured that the materials are consistent with expectations. • Review Design Objectives: An on-going evaluation of field and available laboratory data relative to the design objectives should be performed during the implementation of the site investigation. If adjustments are needed or if additional data needs are identified, procedures should be initiated to obtain the necessary information. Laboratory Testing and Test Interpretation • Select Samples for Performance Testing: Prior to initiating the project-specific laboratorytesting program, the design engineer should review the recovered samples and confirm the testing that needs to be conducted (i.e., type, number, and required test parameters). If possible, selected samples should be extruded in the laboratory and reviewed by the design engineer. • Conduct Laboratory Testing: Once the samples have been reviewed and the testing program is confirmed, it is time to continue the index tests and initiate the performance-testing program (with index test correlation for quality assurance). Preliminary results should be provided to the design engineer for review. • Review Quality of Laboratory Data: If the data and interpreted laboratory test results are not consistent with expectations or if results indicate that the sample was disturbed, it is necessary to review progress and make adjustments. On some projects, results at this stage can be used to plan and initiate a more detailed and focused phase of investigation. A phased investigation approach is particularly helpful on large projects and in cases where there are many unknowns regarding the subsurface conditions or specific project requirements prior to conducting the proposed site investigation program
Select Material Properties: The laboratory and field test results should be interpreted andcompared to project expectations and requirements. The role of the design engineer at thisstage is critical as the full integration of field and laboratory test results must be coupledwith the site-specific design.If test results are not completely consistent, the reason(s)should be evaluated, poor data should be eliminated, and similarities and trends in datashould be identified.It may be necessary to return to the laboratory and conduct anadditional reviewof sample extrusion,selection,andtesting.EngineeringDesignPerform Design:At this final stage, the design engineer has the necessary informationrelated to the soil and rock properties to complete the design.Additionally, the designengineer also has first-hand knowledge related to the variability of the deposit and of thematerial properties. Design activities can proceed with knowledge of these properties andvariabilities.As referenced in chapter 1,this process is logical and is generally followed on many projects.Inmany cases, however, old “rules-of-thumb" and “status quo" approaches can result in anunconscious “by-passing"of critical steps.In particular, selection of the correct engineeringproperty tests, their interpretation, and summarization are often poorly performed. Rigorousattentiontothistwelve-stepprocedureisrequiredto assureefficientandthoroughinvestigationandtesting programs, especially since many projects are fragmented in which drilling,testing,anddesign areperformedbydifferentparties.2.3USEOFCORRELATIONSTOASSISTPROPERTYSELECTIONIf time and budget were not an issue, the design engineer could obtain as many samples as necessaryand conduct as many laboratory or in situ tests as desired to obtain a complete assessment ofsubsurface soil and rock conditions.Engineeringproperties could bequantified and anyinconsistentdata could be set aside; additional testing could then be initiated. Unfortunately, time and budgetsaremajor issues and the design engineer must make critical decisions at several steps throughout thedesignto obtain themostreliable and realistic soil and rockpropertyinformation.As describedpreviously, a critical step in obtaining these properties lies in the selection of a specific test and theinterpretation of the test results. For any number of reasons (e.g., cost, sampling difficulties, etc.), itmay be difficult to obtain the specific parameter(s) of interest.Fortunately, the design engineer canoften use well-developed and/or site-specific correlations to obtain thedesired parameter.Also,correlations serve as a quality assurance check on determined test results.Correlations to engineeringproperties comein manyforms,but all havea commontheme;specifically, the desired correlation utilizes a large database of results based on past experience.Inthe best case, the correlation and experience have been developed orcalibrated"using the specificlocal soil; in other cases the correlation may be based on reportedly similar soils.The reliance or useof correlations to obtain soil and rock properties is justified and recommended in the followingcases: (1) specific data are simply not available and are only possible by indirectly comparing toother properties; (2) a limited amount of data for the specific property of interest are available andthe correlation can provide a complement to these limited data; or (3) the validity of certain data is inquestion and a comparison to previous test results allows the accuracy of the selected test to be7
7 • Select Material Properties: The laboratory and field test results should be interpreted and compared to project expectations and requirements. The role of the design engineer at this stage is critical as the full integration of field and laboratory test results must be coupled with the site-specific design. If test results are not completely consistent, the reason(s) should be evaluated, poor data should be eliminated, and similarities and trends in data should be identified. It may be necessary to return to the laboratory and conduct an additional review of sample extrusion, selection, and testing. Engineering Design • Perform Design: At this final stage, the design engineer has the necessary information related to the soil and rock properties to complete the design. Additionally, the design engineer also has first-hand knowledge related to the variability of the deposit and of the material properties. Design activities can proceed with knowledge of these properties and variabilities. As referenced in chapter 1, this process is logical and is generally followed on many projects. In many cases, however, old “rules-of-thumb” and “status quo” approaches can result in an unconscious “by-passing” of critical steps. In particular, selection of the correct engineering property tests, their interpretation, and summarization are often poorly performed. Rigorous attention to this twelve-step procedure is required to assure efficient and thorough investigation and testing programs, especially since many projects are fragmented in which drilling, testing, and design are performed by different parties. 2.3 USE OF CORRELATIONS TO ASSIST PROPERTY SELECTION If time and budget were not an issue, the design engineer could obtain as many samples as necessary and conduct as many laboratory or in situ tests as desired to obtain a complete assessment of subsurface soil and rock conditions. Engineering properties could be quantified and any inconsistent data could be set aside; additional testing could then be initiated. Unfortunately, time and budgets are major issues and the design engineer must make critical decisions at several steps throughout the design to obtain the most reliable and realistic soil and rock property information. As described previously, a critical step in obtaining these properties lies in the selection of a specific test and the interpretation of the test results. For any number of reasons (e.g., cost, sampling difficulties, etc.), it may be difficult to obtain the specific parameter(s) of interest. Fortunately, the design engineer can often use well-developed and/or site-specific correlations to obtain the desired parameter. Also, correlations serve as a quality assurance check on determined test results. Correlations to engineering properties come in many forms, but all have a common theme; specifically, the desired correlation utilizes a large database of results based on past experience. In the best case, the correlation and experience have been developed or “calibrated” using the specific local soil; in other cases the correlation may be based on reportedly similar soils. The reliance or use of correlations to obtain soil and rock properties is justified and recommended in the following cases: (1) specific data are simply not available and are only possible by indirectly comparing to other properties; (2) a limited amount of data for the specific property of interest are available and the correlation can provide a complement to these limited data; or (3) the validity of certain data is in question and a comparison to previous test results allows the accuracy of the selected test to be
assessed. Correlations in general should never be used as a substitute for an adequatesubsurface investigation program, but rather to complement and verify specific project-relatedinformation.Examples ofeachof the threecasesfollows:Specific Data are Unavailable: Several examples of this type exist. Most notable is thestrength of uncemented clean sands.Undisturbed sampling is prohibitivelyexpensiveandcorrelations to Standard Penetration Testing (SPT), Cone Penetration Testing (CPT), andotherin situtests resultshavebeen shown tobequitereliable.As another example,supposethat the strength of a soil in triaxial extension is desired and only triaxial compression dataare available.By reviewingprevious published comparisons of compression and extensiontest results for similar soils, it is possible to approximate the triaxial test extension testresults.Limited Data are Available:For a given application, suppose that only afew, high-qualityconsolidation tests were performed.Compression properties werefound to correlate wellwith Atterberg limits testing results. It is therefore concluded that additional consolidationtest results are not required and that numerous Atterberg limits tests can be used toconfidentlyassess compressionproperties.Assessing Data Validity:Consider that results from tests on two similar soils areinconsistent.By comparing the results to those for similar soils it may be possible toidentify whether thedata are simply inconsistent of if some of thedata are incorrect.There are several sources of correlation data for a range of geotechnical materials and propertiesMany geotechnical textbooks and reference manuals include correlations as part of the text (e.gHoltzandKovacs,1981;NAVFAC,1982).TheElectricPowerResearch Institute(EPRI)(Kulhawyand Mayne, 1990) commissioned the preparation of a very useful document that includes severalcorrelations for laboratory and in situ tests.Regardless of the specific correlation, the following critical components need to be explicitlyrecognized:The selected correlation is only as good as the data used to develop the correlation. Manycorrelations for sands were developed for clean, uncemented, uniform sands primarily forassessing liquefaction potential.Be careful inusing this correlation toassessproperties inawell-graded silty sand deposit.Selectthe appropriate correlation carefullyA correlation provides an “approximate" answer and will undoubtedly exhibit scatter amongthe data points. Assess the data and the scatter by using upper and lower bound (i.e, bestcase/worstcase)scenarios inthedesign calculations.The selected correlation will be most accurate if calibrated"to local soil conditions.Manystate DOTs have developed useful correlations based on specific project experience in theirstate.8
8 assessed. Correlations in general should never be used as a substitute for an adequate subsurface investigation program, but rather to complement and verify specific project-related information. Examples of each of the three cases follows: • Specific Data are Unavailable: Several examples of this type exist. Most notable is the strength of uncemented clean sands. Undisturbed sampling is prohibitively expensive and correlations to Standard Penetration Testing (SPT), Cone Penetration Testing (CPT), and other in situ tests results have been shown to be quite reliable. As another example, suppose that the strength of a soil in triaxial extension is desired and only triaxial compression data are available. By reviewing previous published comparisons of compression and extension test results for similar soils, it is possible to approximate the triaxial test extension test results. • Limited Data are Available: For a given application, suppose that only a few, high-quality consolidation tests were performed. Compression properties were found to correlate well with Atterberg limits testing results. It is therefore concluded that additional consolidation test results are not required and that numerous Atterberg limits tests can be used to confidently assess compression properties. • Assessing Data Validity: Consider that results from tests on two similar soils are inconsistent. By comparing the results to those for similar soils it may be possible to identify whether the data are simply inconsistent of if some of the data are incorrect. There are several sources of correlation data for a range of geotechnical materials and properties. Many geotechnical textbooks and reference manuals include correlations as part of the text (e.g. Holtz and Kovacs, 1981; NAVFAC, 1982). The Electric Power Research Institute (EPRI) (Kulhawy and Mayne, 1990) commissioned the preparation of a very useful document that includes several correlations for laboratory and in situ tests. Regardless of the specific correlation, the following critical components need to be explicitly recognized: • The selected correlation is only as good as the data used to develop the correlation. Many correlations for sands were developed for clean, uncemented, uniform sands primarily for assessing liquefaction potential. Be careful in using this correlation to assess properties in a well-graded silty sand deposit. Select the appropriate correlation carefully. • A correlation provides an “approximate” answer and will undoubtedly exhibit scatter among the data points. Assess the data and the scatter by using upper and lower bound (i.e., best case/worst case) scenarios in the design calculations. • The selected correlation will be most accurate if “calibrated” to local soil conditions. Many state DOTs have developed useful correlations based on specific project experience in their state