Performance-Based Mix Design Y. Richard Kim North Carolina State University Presented to the Asphalt Mixture ETG Fall River, MA September 14, 216
Integration between PBMD and PRS Same test methods and same underlying principles and models used in PBMD and PRS Index properties can be used in PBMD whereas full models are used in PRS. Integration necessary to apply incentive/disincentive to contractors PBMD index properties allow go/no-go decisions during construction Allows changes in mix production during construction
https://en.wikipedia.org/wiki/disability-adjusted_life_year Predict Pavement Performance How much life was lost? Gained? Make it simpler PRS Binder Content Hit the target. Walk away. Calibrated to performance data. PBMD a number a number
PBMD Framework
Performance-Based Mix Design Volumetric Final optimum Cracking Resistance Rutting Resistance Minimum Required % AC Minimum Required Candidate Performance Optimum
Performance-Based Mix Design Volumetric optimum Cracking Resistance? Predictive Equations Rutting Resistance % AC Candidate Performance Optimum
PBMD Framework Step 1: Perform Superpave volumetric mix design to determine the volumetric optimum. Step 2: Conduct performance tests on the volumetric optimum using AMPT. Step 3: Check against the minimum performance criteria. Step 4: If okay, the volumetric optimum becomes the final optimum. Step 5: If not okay, adjust the asphalt content using predictive equations. Step 6: Conduct performance tests on the adjusted optimum. Step 7: Check against the minimum performance criteria. Step 8: If okay, the adjusted optimum becomes the final optimum. Step 9: If not okay, use different aggregate gradation and repeat the above steps.
Possible Scenarios for PBMD Pavement structure unknown Pre-approval of mix design Use index properties to determine pass/fail Or run LVECD program on critical pavement designs with measured mixture properties to check against the minimum required pavement performance Pavement structure known Run LVECD program on known pavement design with measured mixture properties to check against the minimum required pavement performance.
Test Methods and Models for PBMD and PRS
Asphalt Mixture Performance Tester
38 mm Cores for AMPT Cyclic Fatigue Testing Only 1 gyratory specimen needed for PBMD fatigue testing
Horizontal Cores from Field Core 15 mm 38 mm 11 mm AC Base Subgrade 15 mm
S-VECD Material Functions Time-Temperature Shift Factor These characteristic relationships remain the same under different modes of loading, different E* Mastercurve temperatures, different stress/strain amplitudes, and different loading histories. Damage Characteristic Curve Energy-Based Failure Criterion
S-TSS for Rutting Test Test Method S-TSS TSS Reference - 1 (T H ) Temp. Pulse Time (s) 2 (T H and T L ) 3 (T H, T I, and T L ).4.4 Rest Period (s) 3.6 (T H ) 1.6 (T L ) 1 (T H ) 1.6 (T I, T L ) Deviator Stress (psi) 1, 7, and 13 (T H ) 7, 1, and 13 (T L ) 7, 1, and 13 Number of Samples 4 8 Testing Time (days) 1.5 3 Permanent strains determined from machine displacements. No on-specimen LVDTs necessary.
Shift Model as the Rutting Model 2.% (a) S-TSS 2.% (b) Reference Curve Permanent Strain 1.5% 1.%.5% T H 1 psi 7 psi 13 psi Permanent Strain 1.5% 1.%.5% Predicted Reference ε = vp -3 a = p log( ξ ) + p 1.E-5 1.E-4 1.E-3 1.E-2 1.E-1 1.E+ 1.E+1 a = d1log( σ / P ) + d2 ξ p ε N red 1 T H β Total ξp σv + red ( N N ) I 1 p 2 T L 7 psi 1 psi 13 psi.% 2 4 6 Number of Cycles 2 (c) Total SF (1 psi) Accounts for the effects of Total SF (a total ) -1 loading time on rutting -2 T L 7 psi 1 psi 13 psi Reduced Load Time (Sec).% 1 2 3 4 5 6 Number of Cycles stress level, temperature, and a = a + a σ v v a Load Time SF (a ξp ) 2 1-1 -2-3 (d) Reduced load time SF Stress SF (a σd ) 2 1-1 (e) Vertical Stress SF -4 1.E-5 1.E-4 1.E-3 1.E-2 1.E-1 1.E+ 1.E+1 Reduced Load Time (Sec) -2 7 9 11 13 15 Vertical Stress (=dev+con)
LVECD for Pavement Model 3-dimensional viscoelastic analysis under moving load and changing temperature
Damage after 2 Years Loading Damage Factor (N/N f ) Distribution - @ September 1, 221 1 Damage Factor (N/N f ) Distribution - @ September 1, 221 1 1.9 1.9 2.8 2.8 3.7 3.7 Z (cm) 4 5 Control.6.5 Z (cm) 4 5 Advera.6.5 6.4 6.4 7.3 7.3 8.2 8.2 9.1 9.1 1-1.5-1 -.5.5 1 1.5 X (m) 1-1.5-1 -.5.5 1 1.5 X (m) Damage Factor (N/N f ) Distribution - @ September 1, 221 1 Damage Factor (N/N f ) Distribution - @ September 1, 221 1 1.9 1.9 2.8 2.8 3.7 3.7 Z (cm) 4 5 Sasobit.6.5 Z (cm) 4 5 Evotherm.6.5 6.4 6.4 7.3 7.3 8.2 8.2 9.1 9.1 1-1.5-1 -.5.5 1 1.5 X (m) 1-1.5-1 -.5.5 1 1.5 X (m)
Rut Depth Prediction in LVECD Time History
Required Testing Time Property Operation Time Modulus a Dynamic Modulus Test 1 day Cracking AMPT Cyclic Fatigue Test (TP-17) 1 day Rutting S-TSS Test (TP-116 Option B b ) 1.5 days Pavement Performance a LVECD Program 4 min. Total Time for PBMD Performance Testing For Index Properties For Pavement Performance Note: a Only needed when the pavement performance analysis is desired. b AASHTO specification being developed. 2.5 days 3.5 days
Validation Using Field Data
Laboratory-to-Field Correlation FHWA-ALF (1 mm Pavement) LVECD Analysis Field Cracking Data Control SBS Crumb Rubber (Terminal Blend) Terpolymer 1 mm Crushed Stone Aggregate Base
Fatigue O Field Prediction NCAT Test Track FW AW C RW R Before Calibration O FW O FW After Calibration AW R RW, C AW R RW, C
Fatigue Field Prediction MIT-WMA S C A E Before Calibration S A After Calibration S C E A C E
Rutting Performance Prediction Field Measured LVECD Predicted MIT RAP MIT WMA
Rutting Performance Prediction Field Measured LVECD Predicted FHWA ALF NCAT
Index Property for Pass/Fail
S@C avg as Cracking Index Property S@C avg is cumulative effective dissipated pseudo strain energy Use the temperature recommended in TP 17 as the reference temperature. S@C avg = 8, is the preliminary minimum required value.
S@C avg for ALF Mixtures C 1.8.6.4.2 ALF-Control ALF-CR ALF-TR ALF-SBS Cracking Area (%) 1 8 6 4 2 1.E+ 1.E+5 2.E+5 3.E+5 4.E+5 5.E+5 S 3 1 2 3 S @ C avg Cracking Area (%) 8 6 4 2 S @ C avg 2 1 Control CR TR SBS Control CR TR SBS
Factors Affecting S@C avg Binder Content Binder Grade 2 12 16 1 S @ C avg 12 8 S @ C avg 8 6 4 4 2 NH644-opt NH644opt NH644+opt NH584-opt NH644-opt
MSR for Rutting Index Property TP-116 by Azari and Mohseni MSR = a(t P) b = 1-7 TP b where T= temperature ( C) and P=deviatoric stress (MPa) FHWA ALF Mixtures 5 4 L1-Control b=5.4 5 4 L5-4%RAP b=5.2 5 4 L6-2%RAP b=5.1 MSR 3 2 MSR 3 2 MSR 3 2 1 1 1 1 2 3 4 5 6 TP ( C MPa) 1 2 3 4 5 6 TP ( C MPa) 1 2 3 4 5 6 TP ( C MPa) MSR 5 4 3 2 L7-2%RAS-SB b=4.98 MSR 5 4 3 2 L8-4%RAP-SB b=5.3 MSR 5 4 3 2 ALF All L6-2%RAP L1-Control L8-4%RAP-SB L5-4%RAP L7-2%RAS-SB 1 1 1 1 2 3 4 5 6 TP ( C MPa) 1 2 3 4 5 6 TP ( C MPa) 1 2 3 4 5 6 TP ( C MPa)
Classification of Mixtures 3 MSR at.6 (MPa) and T H @ 5% ( C) 2 1 Mixture Type RS9.5B RI19B RB25B Criteria Design ESALs (million) TP-116 1-3 < 1 3-1 6.12 4.96 NCDOT 4.59 QMS.34. - 3 4.71 < 3 3.67< 3 2.76 13.96 21.14 4.8 L1-C L5-4R L6-2R L7-2SS L8-4RS MIT-5RSB MIT-E2 RS9.5B RI19B RB25B Traffic Level TP-116 Criteria Design ESALs (million) Maximum MSR Value Light < 1 24 Standard > 1 to 3 17 Heavy > 3 to 1 1 Very Heavy > 1 to 3 3 Extreme > 3 1 NCDOT 216 QMS Manual Mixture Design ESALs (million) S9.5B >.3 to 3 I19B < 3 B25B < 3
Predictive Equations
Materials and Mix Designs Accelerated Loading Facility (ALF) Lane 6 Superpave 12.5 mm HMA mixture 23% RAP PG 64-22 binder Volumetric Design Target Design VMA: 13, 14, 15% Design AV: 3, 4, 5% In-Place AV: 5, 7, 9% Total of 21 Mix Designs AMPT Cyclic Fatigue and TSS Testing Completed
Predictive Equations for Damage Characteristic Curve C = exp (a S b ) b = 2.388+.1464xV a -.1235xVMA -.1452xV beff +.1241xVMAxV beff a =.1826+.46641xV a -.21855xVMA
Prediction Results for Mix B 35 % Damage (Measured Coefs) 3 25 2 15 1 5 y =.989x R² =.9966 5 1 15 2 25 3 35 % Damage (Predicted Coefs)
Prediction Results for Mix O 35 % Damage (Measured Coefs) 3 25 2 15 1 5 y =.8611x R² =.954 5 1 15 2 25 3 35 % Damage (Predicted Coefs)
LVECD Prediction for 21 Pavements 12 Months Loading 24 Months Loading 35 35 % Damage (Measured Coefs) 3 25 2 15 1 5 y =.99x R² =.6947 % Damage (Measured Coefs) 3 25 2 15 1 5 y = 1.145x R² =.7919 5 1 15 2 25 3 35 % Damage (Predicted Coefs) 5 1 15 2 25 3 35 % Damage (Predicted Coefs)
Current PBMD Database Relative not absolute distress Applicable for a particular structure and traffic BUT we can generate a catalog with LVECD Anchor point is standard Superpave Minimum VMA for NMAS 4% Design Air Voids 7% Air Voids In-Place Density
Generalization to Any Mixture Other mixes will be different The pattern should be the same We need to verify with other mixes, incl. WesTrack
Summary of PBMD Starts with Superpave volumetric mix design AMPT cyclic fatigue and S-TSS tests as the performance tests LVECD program for pavement performance analysis Either index properties or pavement performance as the pass/fail criteria Predictive equations to adjust the mix design
Additional Remarks PBMD is a necessity in adequately implementing PRS. PBMD and PRS must be based on the same test methods and engineering properties. PBMD and PRS models have been successfully validated using the field data. Excel programs to be available for determination of material properties Predictive equations are being developed by testing additional mixtures at different volumetrics.
Thank you! Questions?