Separating Load from Moisture Effects in Wet Hamburg Wheel-Track Test

1 Separating Load from Moisture Effects in Wet Hamburg Wheel-Track Test Quan (Mike) Lv, Ph.D. Hussain Bahia, Ph.D. March 6, 2019 Fort Worth, Texas School of Transportation Engineering Tongji University

2 Acknowledgments The financial supports: WisDOT: Wisconsin Highway Research Program 17-06. Project Collaborator: Dr. Preeda Chaturabong

3 Outline Background Materials & Testing Methods Identification of Confounding Effect Proposal of a Novel Analysis Method Validation of the Proposed Method Findings & Conclusions

4 Background Part 01 Background

5 Background The wet HWTD test is widely used to identify asphalt mixes that are prone to rutting and moisture damage (Aschenbrener et al., 1993). Confounding effects of loading and moisture (Lu, 2005; Mohammad et al. 2015, NCHRP-W219; Tsai et al., 2016; Swiertz et al., 2017). Limited specifics are provided in AASHTO T324-17 for the analysis of results (Mohammad et al., 2017).

Background There is a need to separate Loading effects from Moisture effects Option 1: Conducting the HWTD test under both dry and wet. The moisture sensitivity related performance can be determined by subtracting the rutting response curve of a dry HWTD test from that of a wet HWTD test. Lu, Q. Investigation of conditions for moisture damage in asphalt concrete and appropriate laboratory test methods. University of California Transportation Center, 2005. 6

7 Background Option 2 Separating Load from Moisture Effects in Wet HWT test. (Yin et al., NCHRP Project 9-49, 2014)

8 Materials & Testing Methods Part 02 Materials & Testing Methods

Materials & Testing Methods Experimental Plan Eight different mixtures BBS AASHTO T 361-mastic test Dry HWT test Wet HWT test Identification of Confounding Effect (initial consolidation, confound effect, existing solution) Proposal of a Modified Analysis Method Wet HWT test Validation of the Proposed Method 9 Validation of the proposed parameters Validation of the normalization procedure

10 Materials Eight mixture types: 2 aggregate types, 2 traffic levels and 2 binders Mixture ID Aggregate Type Traffic Binder Type Mix Level PG 58 C-MT-S28 MT S-28 C-MT-V28 Cisler MT V-28 C-HT-S28 (Granite) HT S-28 C-HT-V28 HT V-28 W-MT-S28 MT S-28 W-MT-V28 Waukesha MT V-28 W-HT-S28 (Limestone) HT S-28 W-HT-V28 HT V-28

11 Testing Methods HWT test: AASHTO T324-17, 50 ± 1 C (a) PMW Hamburg Single Wheel Tracker (b) Set up for the dry condition test.

Slope of Curve (1st Derivative) Rut Depth (mm) Testing Methods HWT test: Iowa DOT analysis method Creep Slope: CS Stripping Inflection Point: SIP Strip Slope: SS 0-5 -10-15 -20-25 0.0E+00 HWTD test rutting curve Number of Wheel Passes 0 1000 2000 3000 4000 5000 6000 7000 8000 Creep range First Derivative Number of Wheel Passes Measured rut depth Fitted rut depth (6th polynomial) SIP 0 1000 2000 3000 4000 5000 6000 7000 8000-5.0E-04-1.0E-03-1.5E-03-2.0E-03-2.5E-03-3.0E-03-3.5E-03-4.0E-03 Creep pass Strip range 12-4.5E-03-5.0E-03 Strip pass

Testing Methods Binder Bonding Strength (BBS) test: Based on AASHTO T361 Loss of POTS = POTS dry POTS wet POTS dry 100 13 (a) BBS test device (b) the equipment to control the temperature.

14 Identification of Confounding Effect Part 03 Identification of Confounding Effect

Rutting depth at first 1,000 passes (mm) Number of Passes to Failure (pass) Identification of Confounding Effect Confounding effect of initial consolidation (First 1000 Cycles) Rutting depth at first 1,000 passes NPF 6.0 18000 Rutting depths after first 1000 5.0 R² = 0.3525 5.2 mm (vs. 12.5mm) 16000 14000 wheel passes are highly correlated to the AV contents. 4.0 12000 3.0 2.0 1.0 1.4 mm Rutting depth at first 1,000 passes R² = 0.9162 10000 8000 6000 4000 2000 There are strong confounding effects of specimen air void and post-compaction consolidation 0.0 6.0 6.5 7.0 7.5 8.0 0 15 Air Void (%)

Rut depth (mm) Identification of Confounding Effect Effects of water conditioning on the creep stage 5 0 Rut-Difference=Rut(Wet)-Rut(Dry) water conditioning enhancement Number of wheel passes 0 2000 4000 6000 8000 10000 12000-5 -10-15 Rut-Wet(Measured) Surpassing point water conditioning enhancement will affect the CS, SS, SIP. -20-25 Rut-Dry (Measured) 16

Rut depth (mm) Identification of Confounding Effect Effects of water conditioning on the creep stage C-HT-V28 C-MT-V28 W-HT-V28 W-MT-V28 4 C-HT-S28 C-MT-S28 W-HT-S28 W-MT-S28 2 0-2 -4-6 Number of wheel passes 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 water conditioning enhancement is observed in all eight mixtures. -8-10 -12-14 -16 17

Strain (ɛ) 18 Identification of Confounding Effect Existing method to solve the confounding effect (Texas method, NCHRP Project 9-49 ) 0.00-0.05-0.10-0.15 Inflection point (LC SN ) 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 Negative Curvature viscoplastic strain increment ( ε_10,000^vp) Positive Curvature Number of wheel passes ε^vp=ε_ ^vp exp[ (α/lc) ^λ ] Assumption: the inflection point of the curve is an indicator of the onset of the stripping. -0.20 Stripping life (LC ST ) -0.25-0.30-0.35-0.40 Measured wet HWT data Projected viscoplatic strain ε^st=( 12.5 ε^vp)/ (Specimen Thickness)

Slope of curve (1st Derivative) Identification of Confounding Effect Existing method to solve the confounding effect (Texas method, NCHRP Project 9-49 ) Number of wheel passes 0 2000 4000 6000 8000 10000 0.0000-0.0005-0.0010-0.0015-0.0020-0.0025 Inflection point secondary zone Wet 1st Derivative Dry 1st Derivative Two models to fit 2 parts of the trend; before and after inflection point -0.0030-0.0035-0.0040-0.0045-0.0050 19

Slope rate at 10,000 passes in Dry HWT (mm/pass) Identification of Confounding Effect Using the proposed method to solve the confounding effect (Texas method, NCHRP Project 9-49 ) applied to our data. -0.001-0.0009-0.0008 y = 151.99x - 9E-05 R² = 0.53-0.0007-0.0006-0.0005-0.0004-0.0003-0.0002 Wet HWT (Texas method) vs. Dry HWT Correlation - R 2 is not high enough. Need discount the post-compaction. -0.0001 20 0 0-0.000001-0.000002-0.000003-0.000004-0.000005 Predicted viscoplastic strain increment by Texas method (ɛ) -0.000006

21 Identification of Confounding Effect Part 04 Proposal of a Novel Analysis Method

Proposal of a novel analysis method for wet HWT test Assumptions The total rutting depth = the contribution from visco-plastic deformation + the moisture-induced damage. But we need to discount the contribution from the post-compaction phase. The inflection point of the curve (when the curvature changes from negative to positive. ) is where the water starts to affect. Need an easier model and fit method. 22

Second derivative of rutting curve Rutting depth (mm) Proposal of a novel analysis method for wet HWT 0 2000 4000 6000 8000 10000 12000 14000 0.0 Step test 1:Fitting of the raw data Fit curve with a sixth-degree polynomial function. (Eq.1) The inflection point where the second derivative of the polynomial first reaches zero after first 1,000 passes. (Eq.2) -2.0-4.0-6.0-8.0-10.0-12.0-14.0-16.0-18.0-20.0 3.50E-06 Negative Curvature Inflection point Inflection point Raw data Number of wheel passes Positive Curvature Eq. 1: RD N = P 1 N 6 + P 2 N 5 + P 3 N 4 +P 4 N 3 + P 5 N 2 + P 6 N + P 7 3.00E-06 2.50E-06 2.00E-06 Eq.2 : 2 RD N = 30 P N 2 1 N 4 + 20 P 2 N 3 + 12 P 3 N 2 + 6 P 4 N + 2 P 5 = 0 23 1.50E-06 1.00E-06 5.00E-07 0.00E+00-5.00E-07-1.00E-06-1.50E-06 Inflection point Number of wheel passes 0 2000 4000 6000 8000 10000 12000 14000

Rutting depth (mm) Normalized rutting depth, RD(N^ ) ^ Proposal of a novel analysis method for wet HWT test Step 2: Normalization of the fitted data The fitted rutting depth at first 1000 passes should be subtracted from the fitted rutting curve to normalize the data. (Eq.3) 0.0-2.0-4.0-6.0 Number of wheel passes 0 2000 4000 6000 8000 10000 12000 14000 Normalized loading passes, N^ 24 Eq. 3: RD N = RD N + 1000 RD 1000-8.0-10.0-12.0-14.0-16.0-18.0-20.0 Inflection point Normalized Data Inflection point Raw data RD N is the normalized rutting depth, N is the normalized number of loading passes,rd 1000 is the fitted rutting depth at 1,000 passes.

Normalized rutting depth (mm) Proposal of a novel analysis method for wet HWT Step 3: Calculation of the performance-related parameters Overall testperformance evaluation: Number of Passes to Failure (12.5mm, NPF) or maximum Rutting Depth (Rut max ). Rutting resistance evaluation: Visco-plastic Ratio (VR) 0.0-2.0 NPF Normalized loading passes 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 RD _final^vp RD vp N = a (N ) VR -4.0-6.0 Moisture resistance evaluation: Moisture Ratio (MR) -8.0-10.0-12.0 12.5 mm RD _final^m MR = RD m final m RD final vp 100 + RD final -14.0-16.0-18.0-20.0 Data in negative curvature Inflection point Normalized rutting depth "RD" (N^ ) ^ Projected visco-plastic deformation RD^vp (N^ ) ^ 25

26 Validation of the Proposed Method Part 05 Validation of the Proposed Method

Rutting depth (mm) 27 Normalized rutting depth (mm) Validation of the Proposed Method Validation of the normalization procedure Number of wheel passes 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 0 0-2 -5-4 -6-10 -8 Negative curvature Normalized loading passes 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 Creep stage -15 Original rutting curve in dry HWT test -10-12 Negative curvature part of normalized wet rutting curve -20-25 Original rutting curve in wet HWT test Modeled viscoplatic rutting-texas method -14-16 -18-20 Normalized rutting curve in the wet HWT test Normalized rutting curve in the dry HWT test Modeled visco-plastic deformation-new method (a) Current method no normalization (b) New method after normalization

Rut depth (mm) 28 Validation of the Proposed Method Validation of the normalization procedure 2 0-2 C-HT-V28 C-MT-V28 W-HT-V28 W-MT-V28 C-HT-S28 C-MT-S28 W-HT-S28 W-MT-S28 Number of wheel passes 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000-4 -6-8 -10 water conditioning enhancement is reduced. -12-14 -16-18

Slope of curve (1st Derivative) Validation of the Proposed Method Evaluating the Post Compaction Phase Limits Number of wheel passes 0.0000-0.0005-0.0010-0.0015-0.0020 0 1000 2000 3000 4000 5000 6000 50% Initial slope Inflection point Assumption: the post-compaction phase is the first 1,000 passes in the wet HWT test. -0.0025-0.0030-0.0035-0.0040-0.0045-0.0050 80% Initial slope Initial slope Wet 1st Derivative The point at where the slope is decreased to 80% or 50% of its initial value. 29

Validation of the Proposed Method Evaluating the Post Compaction Phase Limits Mixture type Inflection point passes to 80% passes to 50% (pass) initial slope (pass) initial slope (pass) C-HT-S28 2,100 250 700 C-MT-S28 1,850 200 600 C-HT-V28 4,300 450 1350 C-MT-V28 4,300 450 1450 W-HT-S28 2,000 300 1050 W-HT-V28 4,760 3,000 Not reached W-MT-S28 1,800 250 850 W-MT-V28 2,600 300 850 Average 2,964 650 979 30 Reasonable to define the post-compaction range as the first 1,000 passes. There are enough data between 1,000 passes to the inflection point that can be used to build the visco-plastic rutting model.

Validation of the Proposed Method Validation of the proposed parameters vp RD final Mixtures NPF-new m RD final VR MR NPF-Iowa (pass) (mm) (mm) (log mm/ log pass) (%) (pass) C-HT-S28 6320-8.7-3.8 0.89 30.4 (rutting sensitive) 6377 C-MT-S28 4300-8.1-4.4 0.95 34.9 4073 C-MT-V28 13400-6.9-5.6 0.75 45.0 11892 C-HT-V28 16300-4.0-8.5 0.74 67.6 16143 W-HT-S28 5300-5.3-7.2 0.92 57.6 5395 W-HT-V28 13000-3.9-8.6 0.79 69.0 13172 W-MT-S28 4560-5.8-6.7 0.94 53.9 4774 W-MT-V28 9700-3.2-9.3 0.83 74.6 (moisture sensitive) 10093 31 This different ranking confirms that the initial consolidation affects the calculated parameters in the current analysis method and thus should be discounted. MR parameter can be used to evaluate the sensitivity of mixtures.

Measured rutting depth in dry HWT test (mm) 32 VR determined from the dry HWT test (log mm/ log pass) Validation of the Proposed Method Validation of the proposed parameters -12 0.95-10 -8 y = 1.2173x + 1.1041 R² = 0.87 0.90 0.85 y = 0.7874x + 0.1634 R² = 0.96-6 0.80-4 -2 0.75 0 0-2 RD _final^vp -4-6 Predicted from the wet HWT test (mm) -8-10 0.70 0.70 0.75 0.80 0.85 0.90 0.95 1.00 VR determined from the wet HWT test (log mm/ log pass)

33 Validation of the Proposed Method Validation of the new parameters: Moisture Effects are related to Adhesion Groups Current parameters BBS test Loss of Adhesion SS (mm/pass) SIP (pass) SS/CS POTS (%) Value Rank Value Rank Value Rank Value Rank C-HT-S28-0.0052 B 6189 A 4.33 C 16.85 A C-MT-S28-0.0059 C 4549 C 2.68 A 30.09 B W-HT-S28-0.0045 A 4685 B 3.46 B 33.73 C Texas method New method Groups LC SN (pass) LC ST (pass) MR (%) Value Rank Value Rank Value Rank C-HT-S28 1800 A 6300 A 30.4 A C-MT-S28 1400 B 4700 C 34.9 B W-HT-S28 1400 B 5500 B 57.6 C Loss of Adhesion Can Explain The MR

Findings & Conclusions Part 06 Findings & Conclusions 34

35 Findings and Conclusions The rutting depths at the first 1,000 wheel passes are very sensitive to the AV contents of the specimen. This initial consolidation should be discounted if the interest is in shear deformation rutting. After eliminating the post-compaction stage, fitting of a simple power-law model allows effective procedure for separating the visco-plastic response due to loading from the moisture effects. Visco-plastic Ratio (VR), the power factor in the rutting modeling, is proposed to characterize the mixture s rutting resistance under dry conditions; Moisture Ratio (MR), the percentage of the moisture-induced deformation in the final rutting depth, is recommended as a moisture resistance parameter and can be used to indicate the damage sensitivity of the mixture. More work is needed to verify this method, especially the comparison with the field performance.

Effects of Modifiers and Binder Properties on the Performance of Asphalt Mixtures in the Hamburg Question Wheel-Tracking or Comments? Device Test Quan Lv (Mike) Ph.D. Candidate 1991Lvquan@tongji.edu.cn March 19, 2018 Jacksonville, FL Hussain Bahia, Ph.D. bahia@engr.wisc.edu March 3-6, 2019 Fort Worth, Texas AAPT 94 rd Annual Meeting School of Transportation Engineering Tongji University