Role of Binders in Pavement Performance

Transcription

1 Role of Binders in Pavement Performance Presented by H. Bahia Research conducted by The University of Wisconsin-Asphalt Group The Pavement Performance Prediction Symposium Western Research Institute, Laramie, WY, June 21-23, 2006

2 Outline Binder rheology modeling Role of binders in mixture rheology modeling A universal phenomenological model Damage resistance modeling

3 Why discuss binder effects on pavement performance? Pavements are built with Asphalt Mixtures, not binders! But, we select binders before mixture is even designed. This requires understanding the range of binder effects on specific mixture behaviors. Effective binder specification should probably be based on a scale of mixture performance.

4 Rheology of Asphalt Binders Before and after Field Aging Thermal Cracking / Fatigue / Rutting

5 Effect of Modification on Asphalt Rheology

6 Effect of Fillers

7 Binder Modeling Efforts Linear Visco-Elastic Models 1950 s: van der Poel 1960 s: Jongepier and Kuilman, Dobson 1970 s: Dickinson and Witt (1974) 1980 s: de Bats et al., Maccarrone 1990 s: Christensen, Anderson, Marasteanu Recent attempts to include nonlinearity / modified binders / Aging 2000 s: Zeng et al. (Strain dependency) 2000 s: Witczak et al. ( Aging effects ) 2000 s: Damage Resistance Characterization

8 Reality: A Very Complex Material Asphalt + Rocks + Air Voids Very, very complex composite

9 Advanced Technology Combined Digital Image/ X-Ray New Tools

10 We Can determine Film Thickness Distribution Can be very Precise

11 Binder to Mixture Stiffness Relations- Not a new Topic -Empirical Hukelom and Klomp Relations 50s -60s: 2.5 C v where n = f(s b ) S m = S b 1 + n 1 C Bonnaure s relations 70s: v If S b is between 5.0 MPa and 1.0 GPa log β + β ( Sm ) = ( log( Sb ) 8) + log( Sb ) 8 + β 2 2 β β If S b is between 1.0 GPa and 3.0 GPa 2 ( S ) β + β ( β β β )( log( S ) 9) log m = 2 b

12 Binder to Mixture Stiffness Fundamental (Micro-mechanical) Paul s Equation and the Law of Mixtures Hashin and Shtrikman s Arbitrary Phase Geometry Model Hashin s Composite Sphere Model Christensen and Lo s Solution for Generalized Self Consistency Scheme * c G c G G G c G c + + ( ) ( ) * G K G c G K G G c G G L = ( ) σ η 1 * 1 1 y c G G m L + = ( ) ( ) ε η 1 * 1 1 y c G G m U + = 0 2 * 2 * = + + C G G B G G A m m

13 Most Recent Models AASHTO 2002 and after Empirical-Witczak et al. ( ) Last 25 years. The latest version of this model is based on 1,429 mixture testing data on 149 asphalt mixtures. Included in The 2002 AASHTO Design guide Micromechanical- D.W. Christensen, Pellinen, and Bonaquist (2003)

14 Using same formulation for binder and mixtures Illustration of G* master-master curve 1E+07 G* g R 1E+06 1E+05 G* (kpa) 1E+04 G* e R' G*g me 1E+03 m e G*e 1 G*e = 0 G*e > 0 1E+02 1E-10 1E-06 f' c 1E-02 1E+02 f c 1E+06 1E+10 Frequency (Hz)

15 Rheology (G*) Formulation Fitted model ( Phenomenological) G* g G* e G* = G* e + [7] me / k [ ( ) k ] 1+ fc f G* e = G* (f 0), equilibrium complex modulus, G* g = G* (f ), glassy complex modulus, f c = location parameter, (temperature effects) f = reduced frequency, (speed effects) k, m e = shape parameters, dimensionless. (Aging effects)

16 Phase Angle Formulation δ m = phase angle constant, the value at f = f d, f d = location parameter, R d, m d = shape parameters, I = 0 for mixtures; I = 1 when f > f d and I = 0 when f < f d for binders. 2 / 2 ) log( 1 ) (90 90 d d d m m R f f I I = + δ δ [11]

17 Mixture SST FSCH-- G* data (PG76-22 binder, Limestone agg.) 1E+08 Temp Strain 1E+06 Complex Modulus (kpa) 1E+04 1E+02 1E+00 1E-04 1E-02 1E+00 1E+02 1E+04 1E+06 1E+08 Frequency (Hz)

18 Temperature and Strain Shift Factors log a a T T ( T ) ( T 0) c 1(T-T 0 ) = [14] c2 + (T-T 0 ) T 0 = reference temperature (52C), c 1, c 2 = constants. log a a γ γ ( γ ) ( γ 0) d1( γ-γ 0 ) = [15] d 2 + ( γ-γ 0 ) γ 0 = reference strain (0%), d 1, d 2 = constants.

19 Binder and mixture G* master-master curves 1E+08 Binder fit Mix fit Complex Modulus (kpa) G*, MPa 1E+06 1E+04 1E+02 Mixture Binder 1E+00 1E-04 1E-02 1E+00 1E+02 1E+04 1E+06 1E+08 Reduced (Hz) Reduced Frequency, Hz

20 Binder and mixture Phase angle - δ -master curves 90 Binder fit 75 Mix fit Phase Angle (deg) Binder 15 Mixture 0 1E-04 1E-02 1E+00 1E+02 1E+04 1E+06 1E+08 Reduced Frequency (Hz)

21 Binder and mixture temperature shift factors Binder fit Mix fit 4.0 log a T Temperature (C)

22 Binder and mixture strain shift factors Binder fit Mix fit log aγ / / / / /0.15 Binder/Mixture Strain (%)

23 Evaluation of Errors Items Max Error R 2 Binder Mixture Binder Mixture log G* δ (degree) log a T log a γ

24 If we have binder rheology model, can we predict mixture rheology? Best direct approach is Finite Element Method

25 From Image to FEM MATLAB Binary Image Finite Element Mesh

26 Experimental vs. FEM Estimated Mixture Stiffness B13 G* (Pa) Mix (FEM) Binder Mix I (Experiment) Mix II (Experiment) Frequency (Hz), Hz

27 Aggregate Interlock

28 Aggregate Interlock (cont d) Measured G* (Pa) Mix (FEM) w/ interlock Mix I (Experimental) Mix II (Experimental) Binder (RTFO) (Hz) Frequency, Hz

29 Transition Zone

30 Transition Zone (cont d) B G* (Pa) Mix (FEM) Binder Mix I (Experiment) Mix II (Experiment) (Hz) Frequency, Hz

31 How is modeling currently used in prediction performance? Empirical Models are the Most Common

32 AASHTO 2002 Dynamic Modulus /E*/Phase Angle, φ σ o sinωt φ/ω σ, ε σ 0 ε 0 ε o sin(ωt-φ) Time, t E* = σ 0 ε 0 φ = ωti Slide from Dr. Ray Bonaquist

33 AASHTO 2002 (Witczak et al.) Model for Plastic Strain (ε p ) All Data log ε p = c 1 + c 2 * log N + c 3 * log No4 + c 4 * log No200 + c 5 * log temp + C 6 * log th_ei + c 7 * log Va + c 8 * log vfa + c 9 * log th_fr + c 10 * log Dev_st + c 11 * log visc_t + c 12 * (log N) 2 +c 13 * (log No4) 2 + c 14 * (log thei) 2 + c 15 * (log Va) 2 + c 16 * (log temp) 2 c 1 = , c 2 = , c 3 = , c 4 = , c 5 =44.16, c 6 = , c 7 =0.4515, c 8 =0.6739, c 9 = , c 10 = 0.7, c 11 = , c 12 = , c 13 = , c 14 = 5.69, c 15 =31.219, c 16 = R 2 = 71.8% Se/Sy= N= 4995

34 How important is effect of Binder in typical mixture? 1. Rheology 2. Damage Resistance

35 Sensitivity to temperature: Ratio of G* at 46C to G* at 52C (f=0.1hz) 5 4 f (Hz) = 0.1 Strain (%) = 0.01 PG82 (SBSr) PG82 (PE) PG82 (ss) PG82 (SBR)/(SBSl) PG76 (ET) PG76 (oxd) PG58 (SBSl) PG58 (SB) PG58 (oxd) G* 46C /G* 52C Binder Change per 6C= 2.3 Mix Change per 6 C= AI Coarse AI Fine NCAT Coarse NCAT Fine Binder Aggregate Type

36 Sensitivity to Traffic Speed : Ratios of G* at 10 Hz to G* at 0.1 Hz (T=52C) Temp (C) = 52 Strain (%) = 0.01 PG82 (SBSr) PG82 (PE) PG82 (ss) PG82 (SBR)/(SBSl) PG76 (ET) PG76 (oxd) PG58 (SBSl) PG58 (SB) PG58 (oxd) G* 10Hz /G* 0.1Hz Binder Change per 100Hz= Mix Change per 100 Hz= 5 0 Binder Aggregate Type AI AI Coarse AI AI Fine NCAT Coarse NCAT Fine

37 Effect of Traffic Speed: Ratios of G* at 10 Hz to G* at 0.1 Hz (T=IT) 15 Temp (C) = IT Strain (%) = 0.01 PG82 (SBSr) PG82 (PE) PG82 (ss) PG82 (SBR)/(SBSl) PG76 (ET) PG76 (oxd) PG58 (SBSl) PG58 (SB) PG58 (oxd) 10 G* 10Hz /G* 0.1Hz 6 Binder Change per 100Hz = Mix Change per 100 Hz = Binder AI Coarse AI Fine NCAT Coarse NCAT Fine Aggregate Type

38 Sensitivities to Temperature and Traffic Speed Mixture sensitivity to temperature is lower than binder. Changing grade from PG-58 to PG-64 results in: Binder G* increase by 230% Mix G* increase by only 40 % Mixture sensitivity to traffic speed is also lower than binder. Increasing speed from 0.5 mph to 50 mph results in: Binder G* increase by 1800% (52 C), 700% (25 C) Mix G* increase by only 500 % (52C), 300% (25C)

39 Mixture vs. Binder (G*) Mixture G* at 10Hz and 0.1Hz at 52C and IT (0.1%) (KPa) B Mixture G* (Kpa) = A*(Binder G* (Pa)) Mixture A B R 2 Binder 10% - Mixture 0.1% AI NCAT ALL Crushed Limestone Gravel ALL Binder G* at 10Hz and 0.1Hz at 52C and IT (10%) (Pa) AI NCAT AI NCAT

40 Simplified Binder to Mixture Relationships Limestone Coarse - Low Strain Mixture G* (KPa) = A*[Binder G* Binder G* (1% strain) - Mixture G* (0.01% Modifiers A B R 2 PG 82 SBSr PG 82 PEs PG 82 Oxd PG 82 SBR PG 76 Eter PG 76 Oxd PG 58 SBSl PG58 SB PG 58 Oxd ALL B

41 The Best Simplified Relationship Analyses of data (9 binders and 36 mixtures) indicate that the best model is: Mix G* = 720 (G* of Binder) 0.50 This model could be used for estimating binder effects in pavement analysis and design of layer thickness. It should allow revision of binder grading to effective grades.

42 How do binders affect mix damage? 1.Rutting 2.Fatigue

43 Flexible Pavement Failures. 1. RUTTING

44 Mixture Rutting (log-log plot) Permanent Strain Number of Cycles

45 Damage Testing for Rutting (Repeated Creep Test for Binders) 14.0 Creep Tests at 70C, 300 Pa shear stress (Loading 1s Recovery 9s) 100 cycles 12.0 Accumulated Strain Normal Scale PG 82- Oxidized PG82- PEs PG-82- SBSr Time (s)

46 Evaluation of Binder Effect on Mixture Rutting Average of All Aggregates Mixture rutting slope S Mix = Binder + C 2 R = Binder rutting slope, (1/Kpa)

47 Longitudinal Cracking In the Wheel Path Alligator cracking Fatigue

48 Mixture Fatigue Response Stiffness (Mpa) AI Coarse Average Fit AI Fine Average Fit NCAT Coarse Average Fit NCAT Fine Average Fit 0 1E+00 1E+01 1E+02 1E+03 1E+04 1E+05 1E+06 Number of Cycles of Cycles

49 Binder only Fatigue Test Results (DSR) 1.6E+07 G*, Pa 1.4E E E+07 G* (Pa) 8.0E E E E E Cycles Number of Cycles in the DSR

50 Correlation between Binder & Mixture Fatigue Lives Mix = Binder +C R2 = No. of Cycles at 50% G* (Mixture) Average

51 Findings from Damage Testing Rutting Resistance On average, the rule of 5 to 1 applies: 100% change in binder resistance could results in only 20 % change in mixture rutting resistance. Fatigue Resistance Surprisingly, the same rule of 5 to 1 can be applied: 100% change in fatigue life of binders results in 20 % change in fatigue life of mixtures. Aggregate type and gradation can have significant influence on rutting and fatigue.

52 In summary Binder rheology is important, but we need better methods to estimate effect on performance Effective PG specification should be based on a scale of mixture behavior.

53 Moisture Damage!

54 Adhesion Measurement Pull-Stub Metal Blocks Aggregate or Glass Surface Asphalt Binder Schematic of Piston Specimen Preparation

55 Binder Cohesion Testing Concept of Tack Test The relationship between the stress (f) and the time (t) when the plates move from d 0 to d 1 f 2 = 3ηa t d 2 0 d mm/s f = stress applied t = duration η = viscosity of adhesive a = radius of specimen d 0 = initial thickness of adhesive layer d 1 = thickness after time interval t Adhesive Probe Temperature Chamber Solid Surface

56 Hamburg Wheel Tracking Test Asphalt: B1: Original PG (RTFO) B2: PG Chemically Treated (RTFO) B3: PG w/ Elvaloy (RTFO) B4: PG w/ Elvaloy (RTFO) B5: PG w/ Elvaloy (RTFO) Aggregate: Limestone and Granite Condition at 50C *Testing Results and Pictures Provided by Mr. Reinke of the MTE Services, Inc. Test at 50C

57 HWT Test Results Cycles to Failure Base Limestone Granite Acid Polymer 0 B1 B2 B3 B4 B5 Asphalt

58 Linear Model Predicting the HWT Test results from Adhesion and Cohesion of Binder Cycles to Stripping Failure Limestone Granite R 2 = 0.87 Study will be Published in2005 trb f(adhesion, Cohesion)