PORTABLE FWD (PRIMA 100) FOR IN-SITU SUBGRADE EVALUATION FINAL REPORT by K.P. George Conducted by the DEPARTMENT OF CIVIL ENGINEERING UNIVERSITY OF MISSISSIPPI In cooperation with THE MISSISSIPPI DEPARTMENT OF TRANSPORTATION And U.S. DEPARTMENT OF TRANSPORTATION FEDERAL HIGHWAY ADMINISTRATION The University of Mississippi University, Mississippi June, 2006 i Technical Report Documentation Page Form DOT F 1700.7 (8-72) Reproduction of completed page authorized 1.Report No. FHWA/MS-DOT-RD-06-179 2. Government Accession No. 3. Recipients Catalog No. 5. Report Date June 2006 4. Title and Subtitle PORTABLE FWD (PRIMA 100) FOR IN-SITU SUBGRADE EVALUATION 6. Performing Organization Code 7. Author(s) K.P. George 8. Performing Organization Report No. MS-DOT-RD-06-179 10. Work Unit No. (TRAIS) 9. Performing Organization Name and Address University of Mississippi Department of Civil Engineering University, MS 38677 11. Contract or Grant No. State Study 179 13. Type Report and Period Covered Final Report 12. Sponsoring Agency Name and Address Mississippi Department of Transportation Research Division Jackson, MS 39215-1850 14. Sponsoring Agency Code 15. Supplementary Notes 16. Abstract Subgradesoilcharacterizationmeasuredintermsofresilientmodulus,MR,hasbeenaprerequisiteforpavementdesign.Fornew pavementdesign,MRisobtainedbyconductingrepeatedloadtriaxialtestsonreconstituted/undisturbedcylindricalspecimens, employingAASHTOT-307testprotocol.Because ofthecomplexitiesencounteredwiththetest,in-situtestswouldbedesirable,if reliable correlation can be established.Subgrade characterization for rehabilitation selection, however, in-situ tests are the norm than the exception.The focus of this study is to investigate the viability of Prima 100, a Portable Falling Weight Deflectometer (PFWD), fordirecttestingofsubgradewiththeobjectiveofestimatingresilientmodulus,viaacorrelationbetweenMRandPFWDmodulus (EPFWD).Thirteenas-builtsubgradesectionsreflectingtypicalsubgradesoilmaterialsinMississippi,wereselectedandtestedfor elasticmodulusemployingaFallingWeightDeflectometer(FWD),followedbyPFWD.In-situunitweightandmoisturewere measured using a nuclear device.Soil samplescollected from those sections were subjected to repeated load triaxial test(AASHTO T307) and to other routine laboratory tests for classification purposes. ThefirststepinanalyzingthedatawastoauthenticatethePrimaelasticmodulus,whichwasaccomplishedbyestablishingan acceptable relation between Prima modulus and FWD modulus (EFWD).A statistically significant relation between MR and EPFWD and was derived, though three other explanatoryvariables emerged in the model equation.Since moisture and density of in-situ material rarely match those prescribed in the repeated load test sample, those two attributes were included in the model.A third variable, which was soil-related (namely, PI/P200), emerged to account for the range of soil types, and intentionally retained in the model equation.A similar, but abbreviated version of the model was also derived, deleting the soil-related variable.An investigation of the significance of unit weight and moisture on the Prima modulus resulted in a correlation equation between EPFWD and those two variables. An exclusive program, PFWDSUBGRADE was developed to analyze Prima modulus and calculate resilient modulus.The program, inadditiontocalculatingstation-by-stationresilientmodulus,relyingonwhatisknownascumulativedifferencetechnique, delineateshomogeneoussubsectionsoftheproject,outputtingmeanandstandarddeviationoftheresilientmodulusforeach homogeneous section.A graphical plot of resilient modulus of each station is another output of the program. 17. Key Words Elastic Modulus, In-situ Test, Portable FWD, Resilient Modulus, Subgrade,18. Distribution Statement Unclassified 19. Security Classif. (of this report) Unclassified 20. Security Classif. (of this page) Unclassified 21. No. of Pages 125 22. Price ii ACKNOWLEDGMENT ThisreportincludestheresultsofastudytitledPortableFWD(Prima100)forIn-situ SubgradeEvaluation,conductedbytheDepartmentofCivilEngineering,TheUniversityof Mississippi, in cooperation with the Mississippi Department of Transportation (MDOT), and the U.S. Department of Transportation, Federal Highway Administration (FHWA).Funding of this project by MDOT and FHWA is gratefully acknowledged. The author wishes to thank Bill Barstis with MDOTs Research Division for his efforts in coordinatingtheoverallworkplanoftheproject.M.HowardofMDOTcoordinatedthe fieldwork,includingFWDtests;BurnsCooleyDennisofJacksonconductedtherepeatedload triaxial tests. BiplabBhattacharyawasthekeypersonnelfromtheUniversityconductinglaboratory work and providing support in the field.The service of Sherra Jones in preparing this report is gratefully acknowledged. iii ABSTRACT Subgradesoilcharacterizationmeasuredintermsofresilientmodulus,MR,hasbeena prerequisiteforpavementdesign.Fornewpavementdesign,MRisobtainedbyconducting repeatedloadtriaxialtestsonreconstituted/undisturbedcylindricalspecimens,employing AASHTOT-307testprotocol.Becauseofthecomplexitiesencounteredwiththetest,in-situ tests would be desirable, if reliable correlation can be established.Subgrade characterization for rehabilitation selection, however, in-situ tests are the norm than the exception.The focus of this studyistoinvestigatetheviabilityofPrima100,aPortableFallingWeightDeflectometer (PFWD), for direct testing of subgrade with theobjective ofestimating resilient modulus, via a correlationbetweenMRandPFWDmodulus(EPFWD).Thirteenas-builtsubgradesections reflectingtypicalsubgradesoilmaterialsinMississippi,wereselectedandtestedforelastic modulusemployingaFallingWeightDeflectometer(FWD),followedbyPFWD.In-situunit weightandmoistureweremeasuredusinganucleardevice.Soilsamplescollectedfromthose sectionsweresubjectedtorepeatedloadtriaxialtest(AASHTOT-307)andtootherroutine laboratory tests for classification purposes. The first step in analyzing the data was to authenticate the Prima elastic modulus, which wasaccomplishedbyestablishinganacceptablerelationbetweenPrimamodulusandFWD modulus (EFWD).A statistically significant relation between MR and EPFWD was derived, though three other explanatory variables emerged in the model equation.Since moisture and density of in-situmaterialrarelymatchthoseprescribedintherepeatedloadtestsample,thosetwo attributes were included in the model.A third variable, which was soil-related (namely, PI/P200), emerged to account for the range of soil types, and intentionally retained in the model equation.Asimilar,butabbreviatedversionofthemodelwasalsoderived,deletingthesoil-related iv variable.An investigation of the significance of unit weight and moisture on the Prima modulus resulted in a correlation equation between EPFWD and those two variables. Anexclusiveprogram,PFWDSUBGRADEwasdevelopedtoanalyzePrimamodulus andcalculateresilientmodulus.Theprogram,inadditiontocalculatingstation-by-station resilientmodulus,relyingonwhatisknownascumulativedifferencetechnique,delineates homogeneoussubsectionsoftheproject,outputtingmeanandstandarddeviationofthe resilient modulus for each homogeneous section. Agraphical plot ofresilient modulus of each station is another output of the program. v TABLE OF CONTENTS 1.INTRODUCTION.1 1.1SUBGRADE CHARACTERIZATION IN MECHANISTIC EMPERICAL PAVEMENT DESIGN GUIDE...1 1.2CRITIQUE OF RESILIENT MODULUS TEST(AASHTO, T-307)...........3 1.3OBJECTIVE AND SCOPE.5 1.4CHAPTER SUMMARY..6 2.LITERATURE REVIEW ....7 2.1OVERVIEW7 2.2RESILIENT MODULUS OF UNBOUND MATERIAL7 2.2.1Resilient Modulus Determination7 2.2.2Resilient Modulus in M-EPDG9 2.2.3Estimation of MR from Correlation Equations...10 2.2.3.1 Resilient Modulus Estimation Software11 2.3FACTORS AFFECTING RESILIENT RESPONSE OF SOILS..11 2.3.1Factors Related to the State of Stress12 2.3.2Factors Related to Soil Physical State...13 2.3.3Factors Related to the Structure/Type of Material14 2.4MODELING RESILIENT MODULUS14 2.5NON-DESTRUCTIVE TEST DEVICES..16 2.5.1Non-destructive Impulse Test Devices for Stiffness Modulus..16 2.5.1.1 Falling Weight Deflectometer (FWD)...17 2.5.1.2 Prima 100...18 2.6RELATION BETWEEN RESILIENT MODULUS (MR) AND IN-SITU STIFFNESS MODULUS (E).18 2.6.1Relation Between MR and FWD Modulus, E back or E..19 2.6.2Relation Between MR and EFWD or EPFWD: A Critique..21 2.6.3Relation Between MR and Portable FWD Modulus, EPFWD..23 2.6.4 Relation Between EFWD and EPFWD...24 2.6.5 Relation Between In-situ Resilient Modulus andIn-situ FWD Modulus25 2.7CHAPTER SUMMARY25 3.EXPERIMENTAL WORK - FIELD AND LABORATORY..27 3.1OVERVIEW..27 3.2FIELD TESTS...27 3.2.1FWD Test On Prepared Subgrade and Modulus Calculation27 3.2.1.1 Modulus from FWD Deflection Data.29 3.2.2Measuring In-Situ Modulus Employing PRIMA 10030 3.2.2.1 Description and Operation of Prima 100..30 3.2.2.2 Spectral Analysis of Time History of Load and Deflection38 3.2.2.3 Prima 100 Modulus Data...39 3.2.2.4 Factors Affecting PRIMA 100 Tests..39 vi 3.2.3In-Place Density and Moisture...41 3.2.4Soil Sampling and Tests.44 3.2.4.1 Routine Laboratory Tests on Bag Samples44 3.2.4.2 Resilient Modulus Tests on Reconstituted Samples...44 3.2.4.3 Representative Resilient Modulus From AASHTO T-307 Tests45 3.3CHAPTER SUMMARY46

4.ANALYSIS AND DISCUSSION OF RESULTS..50 4.1INTRODUCTION.50 4.2PRIMA 100 TEST RESULTS...50 4.2.1Unreliable EPFWD Measurement Owing to Uneven Surface...50 4.2.2Outliers of Prima Modulus, EPFWD.50 4.3PRIMA MODULUS RELATED TO FWD MODULUS..51 4.4PROJECT DATABASE....53 4.4.1Resilient Modulus Test Results55 4.5PREDICTION OF MR AT 95% COMPACTION (MR95) FROM PRIMA MODULUS (EPFWD)..56 4.5.1Development of Statistical Model to Predict MR9556 4.5.2Sensitivity of the Model.67 4.6ABBREVIATED PREDICTION MODEL...68 4.7IN-SITU TESTS INFLUENCED BY SEASONAL VARIATION .70 4.7.1Timing of Prima Test.........70 4.7.2Prima Modulus Influenced by In-situ Moisture and Unit Weight.71 4.8DATA ANALYSIS SOFTWARE.78 4.9CHAPTER SUMMARY79 5.PLANNING PRIMA 100 TEST AND CALCULATION OF DESIGN RESILIENT MODULUS...80 5.1OVERVIEW..80 5.2PLANNING PRIMA 100 TEST IN THE FIELD..80 5.2.1Equipment Selection..80 5.2.1.1 Test Procedure...81 5.2.2When and Where To Test?82 5.2.3Additional Data Required..83 5.3SELECTION OF DESIGN UNIT.85 5.4COMPUTER PROGRAM, PFWDSUBGRADE, TO CALCULATE DESIGN MODULUS............................................88 5.5CHAPTER SUMMARY89 6.SUMMARY AND CONCLUSIONS..91 6.1SUMMARY...91 6.2CONCLUSIONS...91 6.3RECOMMENDATIONS FOR FURTHER RESEARCH.92 6.4IMPLEMENTATION93 6.5BENEFITS.94 vii REFERENCES.....95 APPENDIX A - RESILIENT MODULUS OF SAMPLES AS A FUNCTION OF STRESS STATE103 APPENDIX B - OPTIONAL PREDICTION MODELS108 APPENDIX C - DETAILED FLOW CHARTS OF SOFTWARE PFWDSUBGRADE111 viii LIST OF TABLES 2.1Test Sequence and Stress Levels in Harmonized Repeated Load Test (2)..9 2.2Input Levels for Mechanistic-Empirical Pavement Design...10 2.3Test Device Specification (Adapted from Reference 14)..17 2.4AASHTO Modulus Correction Values from Long-Term Pavement Performance Sections (43).(Backcalculated Value Shall be Multiplied by Correction Factor to get Resilient Modulus)20 3.1Summary of Section Locations and Test Performed.28 3.2Falling Weight Deflectometer and Portable Falling Weight Deflectometer Test Results.31 3.3Soil Properties of Bag Samples and Comparison of Nuclear Dry Unit Weight and Moisture to Maximum Dry Unit Weight and Optimum Moisture.42 3.4Resilient Modulus Calculated at Different Stress States Employing Regression Constants k1, k2 and k3 (see equation 3.2).47 3.5Calculated Stress State in Subgrade under Different Loads Including Overburden..49 4.1List of Soils and Their Properties Employed in Regression Analysis...54 4.2Dependent and Independent Variables Considered and Their Ranges in Developing Prediction Model58 4.3Correlation Matrix of Basic Variables Considered in Developing Prediction Model.......58 4.4Correlation Matrix of Transformed-Dependent and-Independent Variables Considered in Developing Prediction Model59 4.5Summary Statistics of Prediction Models..63 4.6Dependent and Independent Variables Considered and Their Range in Developing Correction Equation...74 4.7Correlation Matrix of Basic Variables Considered in Developing Correction Equation..74 4.8Correlation Matrix of Transformed-Dependent and-Independent Variables of Correction Equation.75 4.9Summary Statistics of Correction Equation...77 ix 5.1Comparison of AASHTO T-99 Optimum Moisture and Maximum Dry Unit Weight with Those Calculated from Empirical Equations.86 B.1Summary Statistics of Prediction Models110x LIST OF FIGURES 3.1Prima 100, Portable Falling Weight Deflectometer (PFWD), with Laptop Computer.30 3.2Data Collection Screen (49)...35 3.3Load Pulse and Corresponding Deflection Bowls from Prima 100 Software, 1 lb = 4.44 kN, 1 in. = 2.54 cm36 3.4Effect of Plate Rigidity on Deflection, r = Radial Distance and a = Plate Radius (Adaptedfrom Reference 48).37 3.5Repeatability of Prima 100 Modulus (EPFWD) Test, 1 psi = 6.89 kPa40 4.1FWD Modulus (EFWD) Compared to Prima 100 Modulus (EPFWD), 1 psi = 6.89 kPa 52 4.2Scatter Plot of Modulus Ratio (EPFWD/MR95) Versus Density Ratio (D(f/95)).60 4.3Scatter Plot of Modulus Ratio (EPFWD/MR95) Versus Moisture Ratio (M(f/o))60 4.4Scatter Plot of Modulus Ratio (EPFWD/MR95) Versus PI/P200.61 4.5Modulus Ratio (Measured) Versus Modulus Ratio (Predicted) [Comprehensive Model] 64 4.6 Residuals Plotted Against Predicted Modulus Ratio, EFWD/MR95......65 4.7Residuals Plotted Against Independent Variable D(f/95).65 4.8Residuals Plotted Against Independent Variable M(f/o).....66 4.9Residuals Plotted Against Independent Variable (PI/P200)66 4.10Residuals Plotted Against Independent Variable (PI/P200)69 4.11Modulus Ratio (Measured) Versus Modulus Ratio (Predicted) [Abbreviated Model]..70 4.12Residuals Plotted Against Predicted Modulus Ratio [Abbreviated Model]..73 4.13Scatter Plot of Prima 100 Modulus (EPFWD) Versus Dry Unit Weight (d), 1 psi = 6.89 kPa, 1 pcf = 0.157 kn/m3...73 4.14Scatter Plot of Prima 100 Modulus (EPFWD) Versus Moisture Content (w), 1 psi = 6.89 kPa..76 xi 4.15Scatter Plot of Prima 100 Modulus (EPFWD) Versus Density Ratio (D(f/95), 1 psi = 6.89 kPa..77 5.1 Photograph of Imprint Showing Loose Coarse Particles Congregating Around the first Sensor Tip...83 5.2FN (40) Results Versus Distance Along Project (Adapted from Reference 39)...........87 5.3Delineating Analysis Units by Cumulative Difference Approach (Adapted from Reference 39)..88 5.4Flow Chart of Program PFWDSUBGRADE.90 APPENDIX A - RESILIENT MODULUS OF SAMPLES AS A FUNCTION OF STRESS STATE.103 FIGURE A1:Resilient Modulus versus Deviator Stress of Soil 5(4), 1 psi = 6.89 kPa104 FIGURE A2:Resilient Modulus versus Deviator Stress of Soil 9(2), 1 psi = 6.89 kPa104 FIGURE A3:Resilient Modulus versus Deviator Stress of Soil 14(5), 1 psi = 6.89 kPa..105 FIGURE A 4:Resilient Modulus versus Deviator Stress of Soil 13(1), 1 psi = 6.89 kPa..105 FIGURE A5:Resilient Modulus versus Bulk Stress of Soil 5(4), 1 psi = 6.89 kPa..106 FIGURE A6:Resilient Modulus versus Bulk Stress of Soil 9(2), 1 psi = 6.89 kPa..106 FIGURE A7:Resilient Modulus versus Bulk Stress of Soil 13(1), 1 psi = 6.89 kPa107 FIGURE A8:Resilient Modulus versus Bulk Stress of Soil 14(5), 1 psi = 6.89 kPa107 APPENDIX B - OPTIONAL PREDICTION MODELS.................108 APPENDIX C - DETAILED FLOW CHARTS OF SOFTWARE PFWDSUBGRADE...111 FIGURE C1:Flow Chart of First Phase of Program Calculating Resilient Modulus from ElasticModulus ..............................................................................................112 FIGURE C2:Flow Chart of Second Phase of Program Delineating Homogeneous Sections...113 1 CHAPTER 1 INTRODUCTION 1.1SUBGRADE CHARACTERIZATION IN MECHANISTIC EMPIRICAL PAVEMENT DESIGN GUIDE Subgradesoilstiffnessisanimportantparameterinpavementdesign.Theresilient modulus(MR)hasbecomethestandardparametertocharacterizeunboundpavementmaterials becausealargeamountofevidencehasshownthattheelastic(resilient)pavementdeflection possessesabettercorrelationtofieldperformancethanthetotalpavementdeflection(1).Resilientmodulusisdefinedastheratioofdeviatorstress,d,to the recoverable strain, r, MR = d/r(1.1) NowthatMDOThasembarkedonaprogramofimplementingMechanistic-Empirical PavementDesignGuide(M-EPDG),laboratoryresilientmodulus,MR,ofsubgradesoilisa requisiteinputintothedesignsoftware.TheDepartmenthasalreadyinitiatedastudyto determineMRoftypicalMississippisoils,eventuallydevelopingamaterialslibraryofMR values.Withthis materialslibrarycompleted,aMRvaluebasedonsoil classificationcouldbe obtainedforLevel2pavementdesign.ForLevel1design,however,laboratoryresilient modulusisaprerequisite,withAASHTOadoptingtheharmonizedtestprotocolinNCHRP1-28A(2).Meanwhile,thecomplexityofthelaboratorytestprocedureshaspromptedhighway agenciestoexploreothertestmethods,especiallyinsitufieldtests.Deflectionmeasurements withtheFallingWeightDeflectometer(FWD)and,inturn,moduluscalculationthrough backcalculation have been routinely employed in evaluating pavement layers, and the underlying subgrade.Elasticstiffnessmodulus(abbreviatedasstiffnessmodulus)ofsubgrade,however, could be determined employing forward calculation of the surface deflection induced by devices 2 similartoFWD.Forroutineuse,itisimperativethatthedevicebereliable,fast,andcost effective.TheMississippiDepartmentofTransportation(MDOT)hasfundedthisstudyto investigate the use of a Portable FWD (PFWD) for subgrade characterization. The AASHTO Guide allows the use of both laboratory and in situ backcalculated moduli, butrecognizesthatthemodulideterminedbybothproceduresarenotequal.TheGuide, therefore,suggeststhatthestiffnessmodulusdeterminedfromdeflectionmeasurementsonthe pavementsurface(Eback)needstobeadjustedbyafactorof0.33.However,otherratioshave been documented.Ali and Khosla (3) compared the subgrade soil resilient modulus determined in the laboratory and backcalculated values from three pavement sections in North Carolina.The ratio of laboratory- measured modulus values to the corresponding backcalculated values varied from 0.18 to 2.44.Newcomb (4) reported results of similar tests in Washington State, suggesting a ratio in the range of 0.8 to 1.3.Von Quintus and Killingsworth (5) reported ratios in the range of0.1to3.5inastudybasedondataobtainedfromtheLongTermPavementPerformance (LTPP) database.In the same reference, different average ratios were reported based on the type oflayersatopthesubgradelayer.Laboratoryvalueswereconsistentlyhigher(nearlydouble) than the backcalculated values, according to Chen et al. (6).Note that the previous studies relied onbackcalculatedmodulifromdeflectionmeasurementsonthetopofthepavementstructure.Manyfactorsmayhavecontributedtothedisagreementbetweenthelaboratorymeasuredand backcalculated moduli.One issue is the difficultyof obtaining representative samples from the field because of the inherent variability of the subgrade layer itself.A detailed discussion of the differencesbetweenlaboratorymeasuredMR(lab)andbackcalculatedmodulicanbefound elsewhere (7). 3 WhilenumerousstudieshaveattemptedFWDmeasurementsonthepavementsurface, only a few have targeted FWD tests conducted directly on the subgrade surface.In their study of theMinnesotaResearchRoadProject(Mn/ROAD),VanDeusenetal.(8)reportedthatthe laboratoryresilientmodulustestsconductedonthethin-wallsamplesyieldedvaluesthat comparedwellwiththebackcalculatedvaluesfromthedeeplayerofthesubgrade.Resilient modulusvs.elasticstiffnessmodulus,E,relationwasexploredinarecentstudytitledThe VirginiaSmartRoadProject(9).Theone-to-onerelationshipsought,however,waslessthan satisfactory.A recent investigation, conducted for the Mississippi Department of Transportation (MDOT),showedthatthebackcalculatedmoduli(Eback)obtainedfromtestingdirectlyonthe subgrade were in satisfactory agreement with the laboratory resilient modulus (10).On average, Ebackwas3%largerthantheresilientmodulus.Ina2003study(11),aconcertedeffortwas made to correlate stiffness modulus from the light weight package of FWD to resilient modulus of undisturbed (Shelby tube) samples.The relation turned out to be of dual nature involving the first sensor and the offset sensors stiffness moduli.Note that stiffness modulus is directly related toresilientmodulusincontrasttoapplyingacorrectionfactortobackcalculatedmodulusas customary when testing a pavement system. 1.2CRITIQUE OF RESILIENT MODULUS TEST (AASHTO T-307) SinceAASHTOrecommendsusingalaboratoryresilientmodulustestinarelatively smallsoilsampleonethatisundisturbedorreconstituteditisworthwhiletoexaminehow realisticthistestis.Despiteseveralimprovementsmadeovertheyears,researchershavecited several uncertainties as well as limitations associated with this laboratory test procedure (12): 1.The laboratory resilient modulus sample is not completely representative of in situ 4 conditionsbecauseofsampledisturbanceanddifferencesinaggregate orientation,moisturecontent,in-situsuctionandlevelofcompaction(or recompaction). 2Laboratory specimens represent the properties of a small quantity of material, and not necessarily the average mass of the material that responds to a typical vehicle axle load. 3.Questionableaccuracyindeflectionmeasurementevenemployinginternal LVDTs.. 4.Lackofuniformequipmentcalibrationandverificationproceduresleadto differences between labs. 5.Thetime,expenseandpotentialimpactassociatedwithastatisticallyadequate sampling plan as well as testing add up to large expenditure. Overall,theseissueshavekepttheresilientmodulustestfromachievinggeneral acceptance by the pavement and materials testing community, whereas a nondestructive test such astheFWDdeflectiontestiscreditedwithprovidingin-situmodulus,andisalsocapableof identifyinginherentspatialvariation.Somerecentstudies(13,14,15,16)suggestthata PortableFWD(PFWD)couldaswellaccomplishthesameobjectiveasthatrealizedbya conventional FWD, at a fraction of the cost.Though the direct use of in-situ stiffness modulus is desirable for pavement design, its application should await until the M-EPDG model is calibrated withthisinput.Thisresearch,therefore,exploredtheviabilityofPFWDinestimatingelastic stiffnessmodulus,alikelysurrogatemeasureforresilientmodulusinpavementdesign.Many differentversionsofPFWDhavebeenintroducedinrecentyears(13).Abriefreviewofthe 5 prominent PFWDs can be seen in chapter 2.Prima 100, manufactured by Carl Bros, Denmark, was selected as the most promising technology for this investigation. 1.3OBJECTIVE AND SCOPE The first objective of this study was to investigate the feasibility of employing Prima 100 toestimatein-situstiffnessmodulusofconstructedsubgradeand/orembankments.In accomplishingthisobjective,arelationwassoughtbetweenthein-situstiffnessmodulus measured by Prima 100 and resilient modulus determined in accordance with the harmonized test procedure.Should a viable correlation exist between the two, PFWD tests could be advanced for subgrade characterization for Level 1 M-EPDG design.These objectives were accomplished by: (i)Selecting 14 test sections covering a wide range of soils employed in subgradeconstruction in Mississippi. (ii)Conducting in-situ tests with PFWD and conventional FWD at predeterminedlocations, characterizing those 14 subgrade soils. (iii)Conducting resilient modulus tests as well as index tests (for soil classification)on bag samples collected from the 13 test sections (Note only 13 section-datawere employed in the model development). (iv)PerformingcorrelationanalysisbetweenPFWDstiffnessmodulus(EPFWD)and conventional FWD modulus (EFWD) in order to authenticate the former test. (v)Developing a prediction model, between PFWD stiffness modulus and laboratory resilientmodulus,facilitatingthetransformationofPFWDmodulustoresilient modulus. (vi)Developingofacomputerprogramtodetectspatialvariationofestimated resilient modulus along the road and thus facilitate subdividing the road way into 6 uniformsections,assigningrepresentativedesignresilientmodulusforeach uniform section. 1.4CHAPTER SUMMARY Thischapterdiscusseshowresilientmodulusevolvedtobecometheprimarysubgrade characterizingparametersinM-EPDG.DespitetheoriginalrecommendationofAASHTOto use laboratory resilient modulus, the current trend is to rely more on in-situ tests for assessing the subgrade design modulus. This report comprises six chapters and threeappendices.Chapter 2 presents a literature reviewofvariousin-situdevicesforelasticstiffnessmodulusdeterminationandcorrelations betweenthestiffnessmodulifromdifferentimpacttestingdevices.Resultsofin-situtests (Prima100,FWDandothersupportingtests)on13testsections,fivestationsineachtest section,arepresentedinChapter3.Acomprehensivedataanalysis,culminatinginarelation between elastic stiffness modulus and resilient modulus comprises chapter 4.A methodology for Prima 100 test is described in the first part of Chapter 5.Presented in the latter part is an outline of a computer program designated PFWDSUBGRADE for analyzing Prima 100 data, arriving atadesignresilientmodulusmeanandstandarddeviationofso-calleduniformsection.A summaryandobservationsregardingthefindingsofthestudyconstituteChapter6.Typical resilientmodulustestresultsarepresentedinAppendixA,andoptionalpredictionmodelsin Appendix B.Detailed flow chart of the program, PFWDSUBGRADE, can be seen in Appendix C. 7 CHAPTER 2 LITERATURE REVIEW 2.1OVERVIEW Thischapterpresentsabriefdiscussionofthesignificanceofresilientmodulusin pavement design, its laboratory determination including the factors affecting it.An overview of devicesforin-situmodulusdeterminationofpavementlayersisalsopresentedleadingto selection of Prima 100 in this investigation.In addition, this chapter provides information on the existing correlations between the in-situ devices and resilient modulus. 2.2RESILIENT MODULUS OF UNBOUND MATERIAL The concept of a resilient modulus of a material was originally introduced by Seed et al. (17) in 1962.Seed et al. defined resilient modulus as the ratio of applied replicated deviatoric stress to the resilient or recoverable strain under a transient dynamic load.The resilient modulus has become the standard parameter to characterize unbound pavement materials because a large amount of evidence has shown that the elastic (resilient) pavement deflection possesses a better correlation to field performance than the total pavement deflection. In the last several decades, theresilientmodulushasbecomeawellrecognizedmodeofmaterialcharacterizationforall pavement material layers (subgrade, subbase, and base). 2.2.1Resilient Modulus Determination The resilient modulus of soils can be determined from repeated load triaxial (RLT) tests inthelaboratoryorbackcalculatedfromnondestructive(deflection)tests(NDT)inthefield usingmethodssuchasfallingweightdeflectometer,RoadRaterorDynaflect.The1986 AASHTO Guide, however, has stipulated and the 2002 Guide reaffirmed, that laboratory MR be theparameterforcharacterizingthesubgrade.Respondingtotheneed,AASHTOT278-82 8 laboratory test was proposed to describe the behavior of pavement materials subjected to moving traffic.In 1991, AASHTO modified the T278-82 test procedure in terms of sample conditioning, loadmagnitude,andloadapplication,withtherevisedtestdesignationchangedtoTP292-92I, and subsequently to TP46-94.In conjunction with M-EPDG, a harmonized MR test protocol was proposed in the NCHRP 1-28A study (18).For undisturbed test samples, Shelby tube sampling is relied upon.Reconstituted samples are molded in the laboratory to obtain desired density and moisturecontentrepresentativeofthefield.Thetestmethodrecommendskneading,impactor vibratory methods (depending on soil type) for sample preparation. Inaccordancewiththeharmonizedtestprotocol,afinesoilsampleissubjectedtoa combinationoffourconfiningstressesandfourdeviatorstresses,thusyielding16resilient modulusvaluesforeachsample.Thestressfactorialincludingthetestsequenceislistedin Table2.1.Now,aconstitutivemodelcomprisingMR-stressrelationischosen,describingthe resilient property of the material.The model employed in this study can be seen in equation 2.2.This model is then fitted to the data of each sample by regressing, and the resulting equation can be used for calculating MR at any desired stress level. Generally, the RLT test requires well-trained personnel, expensive laboratory equipment andistime-consuming,tosaytheleast.Theresilientmodulusbackcalculatedfromthefield NDTdeflectiondatacanproduceinconsistentbackcalculatedmodulusresultswhendifferent backcalculationprogramsarechosen.Manyfactorscontributetouncertainoutcomesin backcalculatedmoduli,suchastheuseofelastic-layertheory,thestaticloadassumption, variable and unknown depths of stiff layers at the bottom of subgrade in a pavement structure, to name a few. 9 Table 2.1 Test Sequence and Stress Levels in Harmonized Repeated Load Test (2) Confining Pressure Contact Stress/Seating Stress Cyclic (Deviator) Stress Sequence kPapsikPapsikPapsi Number of Load Applications 0*27.64.05.50.848.37.0 1000 155.28.011.01.627.64.0100 241.46.08.31.227.64.0100 327.64.05.50.827.64.0100 413.82.02.80.427.64.0100 555.28.011.01.648.37.0100 641.46.08.31.248.37.0100 727.64.05.50.848.37.0100 813.82.02.80.448.37.0100 955.28.011.01.669.010.0100 1041.46.08.31.269.010.0100 1127.64.05.50.869.010.0100 1213.82.02.80.469.010.0100 1355.28.011.01.696.614.0100 1441.46.08.31.296.614.0100 1527.64.05.50.896.614.01001613.82.02.80.496.614.0100(*Conditioning) 2.2.2Resilient Modulus in M-EPDG The general approach for selecting design inputs for materials and subgrade soils in 2002 Design Guide is a hierarchical system. In its simplest and most practical form, the hierarchical approachisbasedonthephilosophythatthelevelofengineeringeffortinpavementdesign processshouldbeconsistentwiththerelativeimportance,sizeandcostofthedesignproject 10 (18). In keeping with the hierarchical approach,material characterization is comprised of three inputlevels.Level1representsadesignapproachphilosophyofthehighestpractically achievable reliability, Levels 2 and 3 have successfully lower reliability.A general tabulation of resilient modulus characterization methods is given in the Table 2.2. Table 2.2Input Levels for Mechanistic-Empirical Pavement Design MaterialInput Level 1Input Level 2Input Level 3 Granular MaterialsMeasured resilient modulus in laboratory Estimated resilient modulus from correlations Default resilient modulus Cohesive MaterialsMeasured resilient modulus in laboratory Estimated resilient modulus from correlations Default resilient modulus InselectinganappropriateMRvalueforlevel1design,itisimperativethatthestress statecorrespondingtotypicalrollingloadbegivenconsideration.M-EPDGsuggestsusinga deviatorstress,d=6psi(41kPa)andaconfiningstress,c=2psi(14kPa)inthestress dependent constitutive equation. 2.2.3Estimation of MR from Correlation Equations Various empirical correlations have been proposed to determine the resilient modulus in thelastthreedecades;mostofthemsuitthelevel2requirements.VanTiletal.(19)related resilientmodulusofsubgradesoilstothesoilsupportvalueemployedintheearlierAASHTO designequation.HealsomadeacorrelationchartinwhichthevaluesofMRcouldbe determined from R-value, CBR, and Texas triaxial classification value.Many other correlations between MR, CBR, R-value and soil support values were also developed. A comparative study ofthepresentMRpredictionequationshadbeencompletedforTheMississippiDepartmentof Transportation(20),recommendingLTPPequationsforestimatingitforlevel2design.The 11 writer has completed two other studies on relating Dynamic Cone Penetrometer Index (DCPI) to resilientmodulus(21),andanotherestablishingarelationbetweenFWDelasticmodulusand resilient modulus (11). Note that FWD tests were conducted directly on the subgrade and with modulus calculation employing forward technique. 2.2.3.1 Resilient Modulus Estimation Software:With numerous correlation models available for MR prediction, a software program was recentlydeveloped byHan et al.(22), incorporatingall ofthe30availablemodels.Theprogramwasdevelopedusinganexpertsystemapproach.In this system (program), the information (soil properties) entered by the user is first examined for reasonablenessandaccuracy.Then,datasearchingprocessesareinitiated,andoverthirty estimationmodelscanbeinvoked,dependingontheavailabilityofinputdata.Allresultsare evaluated, based on certainty rules such as how well the data is meeting the limitations existing during the models original development environment.The user is given four alternate methods on which to base the choice of resilient modulus that is most appropriate for the site:one based oncertaintyrulesandothersstatisticalaveragewithdifferentconfidenceintervals.Also included is a provision to estimate an average MR, based on all of the 30 models.This option is employed to estimate MR95 of 18 soils, reported in Table 4.1. 2.3FACTORS AFFECTING RESILIENT RESPONSE OF SOILS Beginning with the 1986 AASHTO Guide (including the current M-EPDG), require that soilcharacterizationtoincorporatechangesinmaterialpropertiesasafunctionofthestateof stress(stressdependency),environmentalconditions(moistureandtemperature),agingand continualdeteriorationundertrafficloading.Acomprehensivediscussionofthefactors affecting resilient modulus is presented in M-EPDG report, Appendix DD1 (23).Those factors 12 arelistedunderthefollowingthreecategories.Thesignificanceofthosefactorsinthe formulation of MR-prediction models can be seen in section 4.5.1. 2.3.1Factors Related to the State of Stress Themostoftenusedstressparametersincludebulkstress,octahedralsheerstressand porepressure.Forlaboratorytestconditionssuchastriaxial,volumetric(bulk)stressis determined from: = 1 + 2 + 3 OctahedralShearStressisdeterminedfrom:( ) ( ) ( ) .3123 223 122 1 + + + =oct

Unbound materials used in pavement design are generally in a partly saturated state, especially if theyfallabovethegroundwatertable.Thestateofstressinunsaturatedmaterialscanbe characterized by the following parameters (Fredlund et al. (24)): ( )au 3= net confining pressure (also called net normal stress); ( )3 1 = deviator stress, d; and( )w au u = matric suction, m where: 3 = total confining pressure; 1 = total major principal stress; ua= pore air pressure; and uw = pore water pressure. Matricsuctiongreatlyaffectsthestateofstressandconsequentlythemodulus(24,25,26,27, 28).Aresilientmodulusmodelincorporatingsoilsuctionisproposedinarecentstudy(29).The equation is of the form: ( )21km d Rk M + = (2.1) where: m= matrix section; k1, k2 = regression constants; and 13 = a parameter thought to be a function of degree of saturation ( = 0 fordry soils, = 1 for saturated soils) Notethatmatrixsuctioncomplementsthedeviatoricstressincreasingtherigidityofthesoil skeleton, and in turn, the resilient modulus. 2.3.2Factors Related to Soil Physical State MoistureContent:Allotherconditionsbeingequal,thehigherthemoisturecontentthe lowerthemodulus;however,moisturehastwoseparateeffects:first,itcanaffectthestateof stressthroughsuctionorporewaterpressurebecausesuctionandwatercontentarecorrelated throughthesoil-watercharacteristiccurve.Second,itcanaffectthestructureofthesoil, through destruction of the cementation between soil particles. DryDensity:Atlowmoisturecontents,alowerdensitywillgivealowerMR.The relationshipisreversedforhighmoisturecontents,asreportedinreference23.Anychangein volume is reflected in a change in dry density; therefore, void ratio (e) can be used instead of dry density. DegreeofSaturation:Athirdparameter,uniquelydefinedbymoisturecontent,dry density (or void ratio) and specific gravity of solids (Gs) is the degree of saturation (S).There is a unique relationship between the three physical state parameters depending on which parameter is used as a measure of volume changes (i.e. dry density or void ratio): Whatitmeansisthatknowinganytwoofthethreeparameters,w,Sanddry,thethird maybefound,providedGsisknownorcanbeestimated.Theuseofallthreeparametersas predictorsinamodelisthereforeincorrect,duetoredundancy.Forcaseswherevariationsin moisture contentare accompanied by volume changes,any two of the three parameters need to be used to correctly predict the change in modulus, together with known Gs. 14 Temperature:It becomes the most important factor in predicting the resilient modulus of frozen materials while for thawed materials it has little or no significant influence. 2.3.3Factors Related to the Structure/Type of Material Compaction Method:Roller compaction in addition to pre-stressing the material, results in microcracks which could affect resilient response of the material. ParticleSize(GrainSizeDistribution):Nodoubtawellgradedsoilcouldresultin improved rigidity of the soil skeleton, and in turn, larger resilient modulus. Particle Shape:Frictional resistance, and in turn, stiffness of the soil, could be enhanced by irregular-shaped particles rather than spherical particles. CohesiveStrength:Rightquantityoffines(-#200material)wouldenhancethebond between particles, and in turn, increase the resilient modulus. 2.4MODELING RESILIENT MODULUS Summarizing,theeffectofvariousfactorsonresilientmodulus,itisimportantto realize/recognize the three broad categories of factors in MR prediction model.If the model for a givenunboundmaterial(UBM),atconstantmoistureanddensityisdesired,bulkstressand octahedral shear stress could be the predictor variables.The M-EPDG model (Equation 2.2) is a prime example of this approach: 3 211kaoctkaa Rp pp k M+= (2.2) where: pa = atmospheric pressure Shouldresilientmodulusbedesiredforanunboundmaterialsubjectedtovarying environmentalconditions(suchas,moisturefluctuation),theexplanatoryvariablesinthe constitutive model shall include a moisture-related factor.Equation 2.1, a deviator stress-matrix suctionmodel,thoughsimple,notonlytakesintoaccountthestateofstressbutalsosuction-15 generated internal stress resulting from moisture changes.Alternately, postulating that the state of stress and physical state factors are uncoupled, the moisture/density effect may be solved as an independent problem, and the latter used in tandem with equation 2.2.The MR saturation level formulation proposed in M-EPDG is represented by Equation 2.3: ( )opt sRoptRS S kMM = log (2.3) where: MR = resilient modulus at saturation level S (%); MR(opt) = resilient modulus at maximum dry density and optimum moisture content; Sopt = degree of saturation at maximum dry density and optimum moisture content, (%); and ks = gradient of log resilient modulus ratio (log (MR/MRopt)) with respect tovariation in degree of saturation (S Sopt) expressed in percent; ks is amaterial constant and can be obtained by regression in the semi-logspace. ThespecificrelationdevelopedbasedonEquation2.3,happenstobeasigmoidal equation, which can be seen in reference (23).Factors related to structure/type of material could beincludedinthemodelbyintroducingadditionalindexpropertiesofUBM(forexample, material passing #200 sieve, PI etc.).General models, empirical though, encompassing all of the threecategories of variables have been proposed in the past (30, 31, 32).The validity of those three and four other equations have been investigated by the author (20), and the LTPP equation (30) is found to be suitable for purposes of predicting resilient modulus of Mississippi subgrade soils.Specifically,intheLTPPequation,physicalstatefactorsincludemoisturecontentand density,stateofstressfactorsareand,andfinally,factorsrelatedtomaterialtypeinclude 16 material passing 3/8" sieve, #4 sieve, percent silt and percent clay, and liquid limit and plasticity index. 2.5NON-DESTRUCTIVE TEST DEVICES Non-destructivetestingofpavements,especiallydeflectiontesting,hasbeenavitalpart inevaluatingthestructuralcapacityofpavement.Adetailedreviewofdeflectionmeasuring methodsandanalysistechniquestoderivematerialpropertyofthelayeredsystemcanbeseen elsewhere (33). TheBenkelmanBeam,theLaCroixDeflectograph,andtheCurviameterapplystaticor slow moving loads.Vibratory loads are applied by the Dynaflect, the Road Rater, the Corps of Engineers16-kip(71-kN)VibratorandtheFederalHighwayAdministrationsCoxVan.Geogauge is a portable device which againemploys a vibratory load.Near field impulse loads areappliedbytheDynatest,KUABandPhoenixfallingweightdeflectometers.Small-scale impulsetestdevicesincludeLoadman(34),GermanDynamicPlateBearingTest(GDP)(35), TRLFoundationTester(TFT)(36)andPrima100(37).Farfieldimpulseloadsareagain appliedbytheimpactdeviceswhoseprimaryuseisinSpectralAnalysisofsurfacewave technique.WavepropagationisusedbytheShellVibrator,whichloadsthepavement harmonicallyandsetsupstandingsurfacewaves,thepeaksandnodesofwhicharefoundby using moveable sensors. 2.5.1Non-destructive Impulse Test Devices for Stiffness Modulus Foradescriptionofprincipalimpulsetestdevicesandothers,includingGeogaugeand Dynamic Cone Penetration Test, the reader may consult references 15, 16 and 38.Impulse test devices described here include the Falling Weight Deflectometer (trailer mounted), and dynamic plate test devices such as GDP, TFT and Prima 100.All those devices mimic the moving vehicle 17 loading by measuring the response of a transient load pulse of 20 to 40 milliseconds and the load appliedthroughabearingplateofdiameterbetween300mmor450mmatacontactstressof about100kPato200kPa.FlexibilityintheloadingisfacilitatedintheFWDandPrima100.Theportabledevicesmeasuredeflectionviaacentralgeophone(oraccelerometer)exceptthe Prima100,whichhasanoptiontoaccommodatetwomoregeophones.Table2.3presents pertinent features for easy comparison.Whereas a detailed description of all of the devices can be seen in reference 15, FWD and Prima 100 employed in this investigation will be summarized for ready reference: 2.5.1.1 FallingWeightDeflectometer(FWD):FWDhasbeenafavoredpavementevaluation deviceoverthelast 25 years.Itistrailer-mounted andcomprisesaweightthatisraisedanddroppedmechanicallyontothe300mmdiametersteelbearingplateviaasetofrubber buffers by in-vehicle computer control. Thedropheight, weight and plate size can be varied to Table 2.3Test Device Specification (Adapted from reference 14) MassDeflection Transducer DevicePlate Diameter (mm) Falling Weight (kg) Bearing Plate (kg) Total Load Pulse (ms) TypeOn Plate or On Ground Stress Rangea (kPa) GDP300101718 2AccelerometerPlate100 TFT300, 200102015-25VelocityGround