Tài liệu Airfield and highway pavements 2017 design, construction, evaluation, and management of pavements

d from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all right Airfield and Highway Pavements 2017 Design, Construction, Evaluation, and Management of Pavements Selected Papers from the Proceedings of the International Conference on Highway Pavements and Airfield Technology 2017 Edited by Imad L. Al-Qadi, Ph.D., P.E. Hasan Ozer, Ph.D. Eileen M. Vélez-Vega, P.E. Scott Murrell, P.E. Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. AIRFIELD AND HIGHWAY PAVEMENTS 2017 DESIGN, CONSTRUCTION, EVALUATION, AND MANAGEMENT OF PAVEMENTS PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON HIGHWAY PAVEMENTS AND AIRFIELD TECHNOLOGY 2017 August 27–30, 2017 Philadelphia, Pennsylvania SPONSORED BY The Transportation & Development Institute of the American Society of Civil Engineers EDITED BY Imad L. Al-Qadi, Ph.D., P.E. Hasan Ozer, Ph.D. Eileen M. Vélez-Vega, P.E. Scott Murrell, P.E. Published by the American Society of Civil Engineers Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. Published by American Society of Civil Engineers 1801 Alexander Bell Drive Reston, Virginia, 20191-4382 www.asce.org/publications | ascelibrary.org Any statements expressed in these materials are those of the individual authors and do not necessarily represent the views of ASCE, which takes no responsibility for any statement made herein. No reference made in this publication to any specific method, product, process, or service constitutes or implies an endorsement, recommendation, or warranty thereof by ASCE. The materials are for general information only and do not represent a standard of ASCE, nor are they intended as a reference in purchase specifications, contracts, regulations, statutes, or any other legal document. ASCE makes no representation or warranty of any kind, whether express or implied, concerning the accuracy, completeness, suitability, or utility of any information, apparatus, product, or process discussed in this publication, and assumes no liability therefor. The information contained in these materials should not be used without first securing competent advice with respect to its suitability for any general or specific application. Anyone utilizing such information assumes all liability arising from such use, including but not limited to infringement of any patent or patents. ASCE and American Society of Civil Engineers—Registered in U.S. Patent and Trademark Office. Photocopies and permissions. Permission to photocopy or reproduce material from ASCE publications can be requested by sending an e-mail to [email protected] or by locating a title in ASCE's Civil Engineering Database (http://cedb.asce.org) or ASCE Library (http://ascelibrary.org) and using the “Permissions” link. Errata: Errata, if any, can be found at https://doi.org/10.1061/9780784480922 Copyright © 2017 by the American Society of Civil Engineers. All Rights Reserved. ISBN 978-0-7844-8092-2 (PDF) Manufactured in the United States of America. Airfield and Highway Pavements 2017 iii Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. Preface An ever-growing number of highway and airport agencies, companies, organizations, institutes, and governing bodies are embracing principles of sustainability in managing their activities and conducting business. Overarching goals emphasize key environmental, social, economic, and safety factors in the decision-making process for every pavement project. Therefore, the theme of the conference was chosen as “Sustainable Pavements and Safe Airports.” It is dedicated to the state-ofthe-art and state-of-practice areas durability, cost-effective, and sustainable airfield and highway pavements. In addition, recent advancements and technologies to ensure safe and efficient airport operations are included. This international conference provides a chance to interact and exchange information with worldwide leaders in the fields of highway and airport pavements, as well as airport safety technologies. This conference brought together researchers in transportation and airport safety technologies, designers, project/construction managers, academics, and contractors from around the world to discuss design, implementation, construction, rehabilitation alternatives, and instrumentation and sensing. The proceedings of 2017 International Conference on Highway Pavements and Airfield Technology have been organized in four (4) publications as follows: Airfield and Highway Pavements 2017: Design, Construction, Evaluation, and Management of Pavements This volume includes papers in the areas of mechanistic-empirical design methods and advanced modeling techniques for design of conventional and permeable pavements, construction specifications and quality, accelerated pavement testing, pavement condition evaluation, and network level management of pavements. Airfield and Highway Pavements 2017: Testing and Characterization of Bound and Unbound Pavement Materials This volume includes papers in the areas of laboratory and field characterization of asphalt binders, asphalt mixtures, base/subgrade materials, and recent advances in concrete pavement technology. This volume also features papers for the use of recycled materials, in-place recycling techniques and unbound layer stabilization methods. Airfield and Highway Pavements 2017: Pavement Innovation and Sustainability This volume is dedicated to the papers featuring most recent technologies used for structural health monitoring of highway pavements, intelligent compaction, and innovative technologies used in the design and construction of highway pavements. The volume also includes papers in the area of sustainability assessment using life-cycle assessment of highway and airfield pavements and climate change impacts and preparation for pavement infrastructure. © ASCE Airfield and Highway Pavements 2017 Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. Airfield and Highway Pavements 2017: Airfield Pavement Technology and Safety This volume is dedicated to recent advances in the area of airfield pavement design technology and specifications, modeling of airfield pavements, use of accelerated loading systems for airfield pavements, and airfield pavement condition evaluation and asset management. The papers in these proceedings are the result of peer reviews by a scientific committee of more than 90 international pavement and airport technology experts, with three to five reviewers per paper. Recent research was presented in the technical podium and poster sessions including the results from current Federal Aviation Administration (FAA) airport design, specifications, and safety technologies; design and construction of highway pavements; pavement materials characterization and modeling; pavement management systems; and innovative technologies and sustainability. The plenary sessions featured the Francis Turner Lecture by Dr. Robert Lytton and the Carl Monismith Lecture by Dr. David Anderson. In addition, two technical tours were offered: Philadelphia International Airport and the Center for Research and Education in Advanced Transportation Engineering Systems (CREATEs) Lab of the Henry M. Rowan College of Engineering at Rowan University. Three workshops were presented prior to the conference: hands-on FAA’s FAARFIELD software, design and construction of permeable pavements, and environmental product declarations. The editors would like to thank the members of the scientific committee who volunteered their time to review the submitted papers and offered constructive critiques to the authors. We are also grateful for the work of the steering committee members in planning and organizing the conference: Katie Chou, Jeffrey Gagnon, John Harvey, Brian McKeehan, Shiraz Tayabji, and Geoffrey Rowe; as well as the local organizing committee chaired by Geoffrey Rowe and members including James A. McKelvey, Timothy Ward, Ahmed Faheem, and Yusuf Mehta for their help with the technical tours. Finally, we would like to especially thank the ASCE T&DI staff who helped put the conference together: Muhammad Amer, Mark Gable, Drew Caracciolo, and Deborah Denney. Imad L. Al-Qadi, Ph.D., P.E., Dist. M.ASCE, University of Illinois at Urbana-Champaign Hasan Ozer, Ph.D., M.ASCE, University of Illinois at Urbana-Champaign Eileen M. Vélez-Vega, P.E., M.ASCE, Kimley-Horn Puerto Rico, LLC Scott D. Murrell, P.E., M.ASCE, Applied Research Associates © ASCE iv Airfield and Highway Pavements 2017 v Contents Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. Mechanistic-Empirical Design Method Implementation and Regional Calibration Comparing Methods for Determining In Situ Asphalt Stiffness Using Pavement ME .............................................................................................................. 1 N. D. Bech, J. M. Vandenbossche, A. Mateos, and J. T. Harvey Analysis of Site-Specific MEPDG Traffic Inputs Parameters for the State of Tennessee in Comparison to National Inputs .......................................... 15 A. Ziedan, M. Onyango, S. Udeh, W. Wu, J. Owino, and I. Fomunung Effects of Concrete Stiffness on Mechanistic-Empirical Performance of Unbonded Jointed Plain Concrete Overlay ............................................................ 25 Gauhar Sabih and Rafiqul A. Tarefder Stability Control of the Unbound Aggregate Base in Multi-Layer Pavement Structures ................................................................................................ 35 Mojtaba Asadi and Reza S. Ashtiani Development of Traffic Inputs Library in Pennsylvania for the Use in AASHTOWare Pavement ME Design Software .................................................... 45 Biplab B. Bhattacharya, Olga Selezneva, and Lydia Peddicord Recalibration of the Flexible Pavement Rutting Model in Utah .......................... 58 Biplab B. Bhattacharya, Michael I. Darter, Leslie Titus-Glover, and Steven Anderson Advanced Modeling and Analysis of Pavements A Molecular Dynamics Simulation Approach to Predict Release of Polycyclic Aromatic Hydrocarbons from Asphalt Concrete Pavements ............. 70 M. I. Hossain, J. P. S. Yadavalli, H. M. Azam, and J. Pan Laboratory Simulation of Extreme Cooling Effects on the Propagation of Reflection Cracks Using Customized Texas Overlay Tester ................................ 81 T. Mandal, H. Yin, R. Ji, and R. Rutter Extended Finite Element Modeling of Crack Propagation in Asphalt Concrete Pavements Due to Thermal Fatigue Load ............................................. 94 M. I. Hossain, A. Adelkarim, M. H. Azam, R. Mehta, M. R. Islam, and R. A. Tarefder © ASCE Airfield and Highway Pavements 2017 vi Load Format Comparison with StratCalc: A 3D Finite Element Method Pavement Analysis Model ...................................................................................... 107 Geoffrey Rowe and Sérgio Raposo Pavement Response to Full Scale and APT Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. Effect of Loading Conditions on the Magnitude and Variation of Pavement Responses in Accelerated Loading Testing .......................................................... 119 Cory Zimmerman Performance Based Specifications Effect of Sample Size and Methods on Percent within Limits for Quality Control and Assurance ........................................................................................... 134 Syed Waqar Haider, Gopikrishna Musunuru, and Karim Chatti Developing Performance-Related Specifications for Preservation Treatments—Micro-Surfacing .............................................................................. 145 Syed Waqar Haider, Gopikrishna Musunuru, and Karim Chatti Pavement Monitoring, Evaluation, and Nondestructive Testing Modeling a Hybrid Pavement Conditions Performance Framework for Botswana District Road Transportation Networks ............................................. 156 Adewole S. Oladele Deep Learning for Asphalt Pavement Cracking Recognition Using Convolutional Neural Network ............................................................................. 166 Kelvin C. P. Wang, Allen Zhang, Joshua Qiang Li, Yue Fei, Cheng Chen, and Baoxian Li Experimental Study on Macrotexture of Asphalt Pavement .............................. 178 Zhi Li, Wenliang Wu, Zhixiong Qiu, and Zhixian Tu Network Level Performance Indicators Lessons Learned from the Canadian Agency Implementation of Transportation Asset Management Systems ........................................................ 191 David K. Hein Use of Multiple Non-Destructive Evaluation Approaches in Connecticut to Establish Accurate Joint Repair and Replacement Estimates for Composite Pavement Rehabilitation ..................................................................... 201 Tamim U. Khan, Steven T. Norton, Katherine Keegan, Jonathan S. Gould, and Christopher D. Jacques © ASCE Airfield and Highway Pavements 2017 vii A Framework for Maintenance Management of Pavement Networks under Performance-Based Multi-Objective Optimization............................................. 209 Sakthivelan Ramachandran, C. Rajendran, A. Veeraragavan, and R. Ramya NDT for Pavement Condition Assessment Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. Potential Applicability of Slab Impulse Response (SIR) in Geophysical Investigation of Pavement Structures ................................................................... 222 Masrur Mahedi, M. D. Sahadat Hossain, Ahmed N. Ahsan, Asif Ahmed, Mohammad Sadik Khan, and Kelli Greenwood Field Investigation of Dowel Misalignment at LTPP Sections ........................... 232 Shreenath Rao and Laxmikanth Premkumar Towards Improved Temperature Correction for NDT Data Analyses ............. 244 M. Broutin and A. Duprey Pavement Surface Characteristics Prediction of International Roughness Index of Flexible Pavements from Climate and Traffic Data Using Artificial Neural Network Modeling .............. 256 M. I. Hossain, L. S. P. Gopisetti, and M. S. Miah Certification of Inertial Profilers .......................................................................... 268 Rohan W. Perera © ASCE Airfield and Highway Pavements 2017 Comparing Methods for Determining In Situ Asphalt Stiffness Using Pavement ME Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. N. D. Bech1; J. M. Vandenbossche2, Ph.D., P.E.; A. Mateos3; and J. T. Harvey, Ph.D., P.E.4 1 Dept. of Civil and Environmental Engineering, Univ. of Pittsburgh, 713 Benedum Hall, 3700 O’Hara St., Pittsburgh, PA 15261. E-mail: [email protected] 2 Dept. of Civil and Environmental Engineering, Univ. of Pittsburgh, 705 Benedum Hall, 3700 O’Hara St., Pittsburgh, PA 15261. E-mail: [email protected] 3 Pavement Research Center, 1353 South 46th St., Building 480, Richmond, CA 94804. E-mail: [email protected] 4 Dept. of Civil and Environmental Engineering, Univ. of California, Davis, 3153 Ghausi Hall, One Shields Ave., Davis, CA 95616. E-mail: [email protected] Abstract The in-situ asphalt stiffness master curve is a critical input for flexible pavement design. The master curve can be established directly by performing dynamic modulus testing on field cores, predicted using the volumetrics of the mixture, binder grade and aggregate gradation established using field cores, or backcalculated using falling weight deflectometer testing. There can be significant differences between the master curves established using these three different methods. The effects of these differences on design were compared using distress predicted by the asphalt pavement design module in Pavement ME. It was found that the method used to determine the in-situ asphalt stiffness can have a significant effect on the predicted distress. Introduction There are three primary methods for determining the in-situ stiffness of the asphalt for an in-service pavement. A master curve that describes the dynamic modulus of the asphalt concrete as a function of load frequency and temperature can be developed using data from dynamic modulus testing performed on cores pulled from the pavement (AASHTO T 342). A dynamic modulus master curve can also be estimated based on the volumetrics, gradation, and binder grade of the asphalt mixture using a predictive equation, such as the Witczak, Hirsch, or Al-Khateeb equations (Andrei, et al., 1999, Bari, 2005, Christensen, et al., 2003, Al-Khateeb, et al, 2006). Artificial neural networks have also been used to predict dynamic modulus from asphalt mixture parameters (Ceylan, et al., 2007, Kim, et al., 2015). Finally, the modulus can be backcalculated using falling weight deflectometer (FWD) data. © ASCE 1 Airfield and Highway Pavements 2017 Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. Differences between the three methods for determining in-situ asphalt stiffness were evaluated using data collected at the Minnesota Road Research Facility (MnROAD). Asphalt concrete cores were pulled and dynamic modulus tests were performed to establish master curves. Additional cores were taken and their volumetric parameters were used to estimate dynamic modulus master curves using the Witczak equation. Finally, moduli backcalculated from FWD testing were used to estimate dynamic modulus master curves. The effect of the difference in the estimated moduli on predicted performance was then evaluated using Pavement ME. Background Each of the three approaches for establishing the in-situ asphalt stiffness has advantages and disadvantages. Performing dynamic modulus testing on cores pulled from the in-service pavement (AASHTO T 342) is the best means for measuring the “true” asphalt stiffness over a range of temperatures and frequencies. However, it is very expensive and time consuming to perform. Also, dynamic modulus test specimens are only 151 mm thick, so the entire thickness of the asphalt layer is not tested if it is greater than 151 mm thick. The use of predictive equations, such as the Witczak or Hirsch equations, is a more cost effective method for establishing the dynamic modulus. The Witczak equation is based on two volumetric parameters of the asphalt mixture (percent air voids (Va), and volumetric effective binder content (Vbe)), the characteristics of the aggregate gradation (P38, P34, P4, and P200), and the binder grade (Andrei et al., 1999). These parameters can be determined from destructive testing of field cores or historic mixture design information. Predictive models, however, do not provide an exact measure of the dynamic modulus. The Witczak equation was developed using labmixed, lab-compacted, undamaged asphalt specimens. Additionally, many of the dynamic modulus tests used to fit the Witczak model were performed diametrically, whereas the current dynamic modulus test (AASHTO T342) uses axial loading (Andrei, et al., 1999). All of these factors can contribute to a difference between the dynamic modulus predicted using the Witczak equation and the dynamic modulus measured in the lab. It is worth noting that even if the use of the Witczak equation does not provide a “true measure” of the dynamic modulus, it is possible that this method provides the most realistic predicted pavement thicknesses when using Pavement ME for design. This is because the calibration of the performance prediction curves was performed using dynamic moduli established primarily using the Witczak equation (ARA, 2004). The stiffness of the asphalt can also be backcalulated from FWD data. This stiffness corresponds to the temperature of the asphalt at the time the FWD testing is performed and to the load frequency of the FWD. Moduli backcalculated from deflection data collected when the asphalt pavement is at different temperatures can be used to “calibrate” a master curve developed using the Witczak equation. First, the Witczak equation and the measured volumetric and aggregate properties of the asphalt mix are used to construct a dynamic modulus master curve that describes the asphalt stiffness as a function of load frequency and temperature. This master curve is then used to determine the dynamic modulus at the temperature of the asphalt layer when the FWD testing was performed. Finally, the load frequency can be established © ASCE 2 Airfield and Highway Pavements 2017 by determining the frequency that provides a shift in the master curve estimated using the Witczak moduli, such that this master curve is equal to the moduli backcalculated from FWD testing. Backcalculation can be performed quickly and cost effectively, but requires accurate layer thickness measurements and appropriate seed moduli. Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. Several studies have evaluated the relationship between the different methods for establishing in-situ asphalt stiffness (Clyne, et al., 2004, Loulizi, et al., 2007, Mateos, et al., 2012). Master curves obtained from performing dynamic modulus testing on field cores and through backcalculation using Evercalc 5.0 were compared for three pavement sections at MnROAD (Clyne et al., 2004). Testing was performed on Cells 33, 34 and 35 of the Low Volume Road. Each cell had a 100-mm thick asphalt concrete layer over a 300-mm thick granular base. A load frequency of 17.9 Hz was assumed for the Dynatest FWD based on measurements from embedded strain gauges. The backcalculated modulus was found to be lower than the dynamic modulus measured in the lab for asphalt temperatures between 0oC and 40oC. Cells 34 and 35 showed that variability in the backcalculated moduli within cells was largest at temperatures less than 10oC. In another study, master curves established using data from dynamic modulus testing performed on cores were compared to master curves predicted using the Witczak equation and to moduli backcalculated using ELMOD (Loulizi, et al., 2007). Cores were taken from nine different pavements in Virginia. A load frequency of 5.3 Hz was assumed based on a FWD load pulse duration of 0.03 seconds. This is over three times less than the frequency assumed in the previous study. The type of FWD used in this study was not provided, but it can be inferred from the load duration that it was a Dynatest. The dynamic moduli predicted using the Witzcak equation were greater than lab-measured dynamic moduli for frequencies greater than 0.0001 Hz and at a reference temperature of 21.1oC. Additionally, the backcalculated moduli, averaged over each pavement section, were lower than dynamic moduli predicted using the Witczak equation. In a study performed by Mateos et al., the dynamic moduli measured on field cores and the moduli backcalculated using EVERCALC 5.0 were compared for four new pavement sections on the CEDEX test track in Madrid (Mateos, et al., 2012). The four sections had asphalt layers between 120 mm and 150 mm thick and granular bases between 500 mm and 1000 mm thick. A load frequency of 15 Hz was assumed for the KUAB FWD used in this study. The average backcalculated moduli were roughly equivalent to the dynamic moduli measured from the field cores. There was moderate variability among the backcalculated moduli within each section. For instance, two FWD tests performed on the same newly-placed section at 17oC resulted in backcalculated moduli of 4 GPa and 5 GPa. This study will further evaluate the difference between the dynamic moduli estimated for in-service pavements. The effect these differences have on the performance predicted using Pavement ME is also investigated. Data Sources Data from MnROAD Cells 15, 16, and 21 were used to establish stiffness vs. temperature relationships using each of the methods described above (MnDOT, © ASCE 3 Airfield and Highway Pavements 2017 4 2016). These cells experienced traffic diverted from Interstate 94 and travelling at highway speeds. All cells were constructed in 1993 and removed in 2008 (Johnson, et al., 2009). The pavement structure for each cell is detailed in Table 1. Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. Table 1: Pavement structure of MnROAD cells Ave. Asphalt Thickness (mm) Base Thickness (mm) Cell 15 280 0 Base Material (MnDOT, 2016) N/A Subgrade AASHTO Classification Depth to Rigid Layer (m) Cell 16 200 710 Class 3. Sp. Granular Base Cell 21 200 580 Class 5 Sp. Granular Base A-6 A-6 A-6 3.9 4.0 3.5 The dynamic modulus was measured for four cores pulled from Cell 21. A master curve was based on the average of these four cores. Additional cores were taken from each section and in-situ volumetrics and aggregate gradations for each asphalt mix were measured using destructive testing. The binder grade for each asphalt mix was determined from original mixture designs. Additionally, representative complex modulus and phase angle data for each binder grade were determined by testing extracted binder. This information is detailed in Table 2. All asphalt mixtures are dense-graded. FWD testing was performed and the asphalt temperature was measured using thermocouples. Backcalculation of the FWD data was performed using Evercalc 5.0. A stiff layer was assumed, and the depth to the stiff layer was obtained using the “line of influence” method (Rhode and Scullion, 1990). Seed values were based on typical seasonal values for MnROAD materials (Ovik, et al. 2000). Moduli were backcalculated at two stations within each cell. Table 2: Asphalt mix parameters of MnROAD cells Cell Asphalt Lift Thickness (mm) Binder Grade 1 33 15 2 108 3 143 AC-20 (PG 64-22) 1 40 16 2 83 3 79 AC-20 (PG 64-22) 21 1 2 3 36 81 79 Pen 120/150 (PG 58-28) Air Voids (Va) 8.3 9.9 7.9 5.4 8.1 9.6 4.2 5.6 6.8 (%) Effective Binder 10.5 10.4 10.9 10.3 9.9 10.0 11.8 11.7 11.4 Content (Vbe) (%) P38 (%) 100 100 99.9 P34 (%) 85.6 84.3 84.3 P4 (%) 69.5 68.2 68.6 P200 (%) 4.40 4.53 4.76 Note: Gradation information represents entire asphalt core. Gradation information for individual lifts was not available. © ASCE Airfield and Highway Pavements 2017 5 Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. In order to exclude the effects of damage and compaction due to traffic on the asphalt stiffness, only FWD test data collected at mid-lane was used. The assumption that there was minimal damage at the mid-lane locations was confirmed through distress surveys (MnDOT, 2016). Very little fatigue cracking was observed in any of the cells, but significant longitudinal cracking in the wheelpath and transverse cracking was observed in all cells. Furthermore, the effect of binder aging on asphalt concrete stiffness was accounted for by using FWD data collected between 1996 and 2005. All cells were constructed in 1993, and it was assumed that any significant increase in asphalt concrete stiffness due to binder aging would have occurred by 1996 when FWD testing was started. The load frequency of the Dynatest FWD used at MnROAD was assumed to be 30 Hz, since this is the recommended value provided by the developers of Pavement ME (Rao and Von Quintus, 2016). It should be noted that this value is substantially higher than what was used in the previous studies presented above. Direct Comparison of Methods Plots comparing the three methods for determining in-situ asphalt stiffness are shown in Figures 1, 2, and 3 for Cells 15, 16, and 21, respectively. The modulus was backcalculated at two different locations in each cell. It was also predicted using the Witczak equation based on the volumetrics and aggregate gradations obtained from testing field cores and binder grades from original mixture designs. Prediction intervals for the stiffness estimated using the Witczak equations were generated as well (Andrei, et al., 1999). For Cell 21, cores were pulled at mid-lane in 2001 so that dynamic modulus testing could be performed. Dynamic moduli measured in the lab and predicted using the Witczak equations were used to establish master curves. The 30-Hz FWD load frequency was used with these master curves to generate the stiffness vs. temperature curves shown in Figures 1 through 3. Witczak Predicted Stiffness using Volumetrics from Cores (2001) Witczak Stiffness 50% Prediction Interval 16 14 Modulus (GPa) 12 10 Witczak Stiffness 95% Prediction Interval 8 6 Mid-lane Backcalculated Stiffness - Station 119620 (1996-2005) 4 2 0 -10 0 10 20 30 o Temperature ( C) 40 50 Mid-lane Backcalculated Stiffness - Station 119770 (1996-2005) Figure 1: Methods for determining in-situ asphalt stiffness, MnROAD Cell 15 © ASCE Airfield and Highway Pavements 2017 6 16 Witczak Predicted Stiffness using Volumetrics from Cores (2001) Witczak Stiffness 50% Prediction Interval 14 12 Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. Modulus (GPa) 10 Witczak Stiffness 95% Prediction Interval 8 6 Mid-lane Backcalculated Stiffness - Station 120415 (1996-2005) 4 2 0 -10 0 10 20 30 o Temperature ( C) 40 50 Mid-lane Backcalculated Stiffness - Station 120465 (1996-2005) Figure 2: Methods for determining in-situ asphalt stiffness, MnROAD Cell 16 Stiffness from Lab Testing of Field Cores (2002) 16 14 Witczak Predicted Stiffness using Volumetrics from Cores (2001) Witczak Stiffness 50% Prediction Interval Modulus (GPa) 12 10 8 Witczak Stiffness 95% Prediction Interval 6 4 Mid-lane Backcalculated Stiffness - Station 122995 (1996-2005) 2 0 -10 0 10 20 30 40 50 Mid-lane Backcalculated Stiffness - Station 123345 (1996-2005) Temperature (oC) Figure 3: Methods for determining in-situ asphalt stiffness, MnROAD Cell 21 The backcalculated moduli are similar to those established using the Witczak equation for Cell 15, but there is a difference observed between the moduli obtained using these two methods in Cells 16 and 21. Here, the backcalculated moduli are lower than the moduli established using the Witczak equation at all asphalt temperatures. This trend is consistent with a study performed by Clyne et al. (2004) for cells on the Low Volume Road at MnROAD. The backcalculated moduli are similar to the moduli obtained from dynamic modulus testing performed on cores for © ASCE Airfield and Highway Pavements 2017 Cell 21. This trend agrees with observations made at the CEDEX test track (Mateos et al., 2012), but not with observations by Clyne et al. (2004). Finally, the Witczak moduli are greater than the moduli from laboratory testing for Cell 21. This is consistent with observations made at sections in Virginia (Loulizi et al., 2007). Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. Comparison of Methods Using Pavement ME The Pavement ME flexible pavement design module was used to quantify the effect of the differences between methods for determining asphalt stiffness on the predicted performance. Each cell was evaluated in Pavement ME using the pavement structures summarized in Table 1. For a Level 1 analysis, Pavement ME uses statistical regression to construct the master curve using the measured dynamic modulus of the asphalt concrete and complex modulus and phase angle of the binder. For a Level 3 analysis, Pavement ME constructs the master curve using dynamic moduli estimated using the Witczak equation (ARA, 2004). For the dynamic moduli measured in the laboratory, the moduli were entered directly into the Pavement ME Level 1 stiffness input field. The complex modulus and phase angle data measured for the extracted binder were used as Level 1 binder inputs. When using the Witczak equation to establish the dynamic modulus, the mixture volumetrics and gradations were entered in the Level 3 stiffness input fields and the binder grade was selected using the drop-down menus. Volumetrics and aggregate gradation data were entered separately for each sub-layer in the asphalt. “Calibration” was required to convert the backcalculated moduli to Level 1 inputs for Pavement ME. First, a master curve representative of the in-situ asphalt was created using the Witczak equation and the volumetrics, aggregate gradation, and binder grade for each asphalt lift. Next, the backcalculated moduli and dynamic moduli predicted by the representative master curve were plotted on the same set of stiffness vs. temperature axes. Finally, the frequency at which the dynamic moduli were predicted was changed until the dynamic moduli were approximately equal to backcalculated moduli at equal temperatures. This “matching frequency” was found to be 1.08 Hz for Cell 21, Station 122995, as shown in Figure 4. © ASCE 7 Airfield and Highway Pavements 2017 8 16 Witczak Predicted Stiffness at 30 Hz 14 Modulus (GPa) Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. 12 10 Witczak Predicted Stiffness at 1.08 Hz 8 6 Midlane Backcaclulated Stiffness - Station 122995 (1996-2005) 4 2 0 -10 0 10 20 30 Temperature (oC) 40 50 Figure 4: Adjusting frequency so Witczak predicted moduli match backcalculated moduli. MnROAD Cell 21, Station 122995 The representative master curve was then used to create a matrix of dynamic moduli at various temperatures and frequencies that could be used as Level 1 Pavement ME inputs. The frequencies for this matrix were shifted an amount equal to the logarithmic difference between the frequency used to predict dynamic modulus with the Witczak equation (30 Hz in Figure 4) and the “matching frequency.” This master curve calibration procedure was performed for each station where moduli were backcalculated and separate Pavement ME analyses were then conducted for each station. The matching frequencies for all the stations analyzed are shown in Table 3. The average matching frequency for Cell 15 is significantly higher than for Cells 16 and 21, which have roughly equal matching frequencies. This trend correlates with asphalt concrete layer thickness. Cell 15 has 284 mm of asphalt, which is more than Cells 16 or 21, which have 202 mm and 196 mm of asphalt, respectively. Table 3: “Matching frequencies” for stations used in Pavement ME analysis Cell 15 15 16 16 21 21 © ASCE Station 119620 119770 120415 120465 122995 123345 Matching Frequency (Hz) 14.6 15.4 0.305 0.257 1.08 0.365 Airfield and Highway Pavements 2017 9 The Pavement ME inputs for Cell 21 for the dynamic modulus testing on field cores, Witczak equation, and backcalculation methods are summarized in Tables 4 through 7. Pavement ME inputs for Cells 15 and 16 followed the same format. Table 4: Dynamic modulus testing on field cores method Level 1 dynamic modulus (GPa) inputs, MnROAD Cell 21 Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. Frequency (Hz) Temperature (oC) -10 4 20 40 54.5 1 5 10 25 10.43 3.17 0.68 0.12 0.03 13.38 5.43 1.77 0.25 0.12 16.42 9.70 2.28 0.39 0.20 17.40 10.62 2.92 0.90 0.49 Table 5: Witczak equation method Level 3 volumetric inputs, MnROAD Cell 21 Asphalt Lift 2 81 Pen 120-150 5.6 Property 1 3 Thickness (mm) 36 79 Binder Grade Pen 120-150 Pen 120-150 Air Voids (Va) (%) 4.2 6.8 Effective Binder Content 11.8 11.7 11.4 (Vbe) (%) P38 (%) 99.9 P34 (%) 84.3 P4 (%) 68.6 P200 (%) 4.76 Note: Gradation information represents entire asphalt core. Gradation information for individual lifts was not available. Table 6: Backcalculation method Level 1 dynamic modulus (GPa) inputs, MnROAD Cell 21, Station 122995. Matching frequency = 1.08 Hz o Temperature ( C) -10 4.4 21.1 37.8 54.4 © ASCE 1.39 12.55 5.96 1.31 0.32 0.11 Frequency (Hz) 2.78 27.75 13.72 17.66 6.82 10.13 1.60 2.94 0.40 0.80 0.14 0.27 138.76 20.36 12.74 4.30 1.28 0.43 Airfield and Highway Pavements 2017 10 Table 7: Lab testing and backcalculation method Level 1 binder inputs, MnROAD Cell 21 Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. Temperature (oC) 35 45 60 Binder Complex Modulus (Pa) 126100 24720 2281 Phase Angle (degrees) 63.87 70.16 80.13 A design period of 15 years was used for all Pavement ME analyses because all cells were constructed in 1993 and removed in 2008. Traffic data from MnROAD were used to establish traffic inputs that represented the actual traffic for Cells 15, 16, and 21 between 1993 and 2008. The initial average annual daily truck traffic (AADTT) was set at 1391 in the design lane. Vehicle class distributions, growth rates by vehicle class, and monthly adjustment factors were also established using MnROAD traffic data. The same traffic inputs were used for all Pavement ME analyses. The climate inputs were based on weather data from the Crystal Airport northwest of Minneapolis, Minnesota. The default threshold values for predicted distress were used along with a 90% reliability. Table 8 shows the predicted distress and the observed surface distress for Cells 15, 16, and 21. If predicted distress exceeded the threshold criterion, the approximate pavement age at which this occurred is listed in parentheses. All observed distress measurements are based on distress surveys performed in April, 2008, at the end of the pavement life (MnDOT, 2016). The method used to estimate the in-situ asphalt concrete stiffness has a different effect on each predicted distress. First, there is no apparent relationship between the method of estimating asphalt stiffness and predicted thermal cracking. Cell 15 fails by thermal cracking over two times faster (5.5 yr) when the Witczak equation is used than when backcalculation is used (12.5 yr), even though there is little difference in moduli between the Witczak and backcalculation methods (Figure 1). In contrast, predicted thermal cracking in Cell 16 is almost identical between methods of stiffness estimation even though these methods have very different moduli (Figure 2). The lack of correlation between method of estimating asphalt stiffness and thermal cracking is not surprising because predicted thermal cracking is determined primarily by low-temperature properties of the binder and not by dynamic modulus. Furthermore, measured thermal cracking is two to three times greater than predicted thermal cracking across all cells. This may be because typical low temperature binder performance inputs, based on the binder grade, were used for the Pavement ME models rather than measured values. The method used to predict asphalt concrete stiffness has a significant effect on predicted rutting. A large difference in moduli between methods of estimating the insitu asphalt stiffness correlates with a large difference in predicted rutting. In Cells 16 and 21, the moduli estimated using backcalculation are lower than the moduli estimated using the Witczak equation (Figures 2 and 3) and the AC rutting predicted using the backcalculated moduli is between 62% and 115 % greater than that predicted using the Witczak moduli. For Cell 15, the moduli estimated using © ASCE Airfield and Highway Pavements 2017 11 Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. backcalculation and the Witczak equation are similar (Figure 1) and the difference in predicted AC rutting between these methods is less than 10%. For Cell 21, the moduli estimated using the dynamic modulus testing are in between the moduli estimated using the other two methods (Figure 3). The dynamic modulus testing inputs predict 16 mm AC rutting, which is in between 11 mm predicted using Witczak moduli and an average of 20 mm predicted using backcalculated moduli. Predicted total rutting is at least two times greater than measured rutting across all methods of estimating asphalt stiffness and across all cells. Predicted AC rutting, however, is within 20% of measured rutting across all cells when the Witczak equation is used to estimate asphalt stiffness. Assuming that all rutting at MnROAD was in the asphalt, using the Witczak equation to estimate the asphalt stiffness gives a reasonable prediction of measured rutting. Table 8: Predicted and measured distress for MnROAD Cells 15, 16, and 21 Distress Cell 15 Cell 16 Cell 21 © ASCE Method of Asphalt Stiffness Estimation Total Rutting (mm) Threshold 19 Witczak Equation 23 (7.5 yr) Backcalculation Station 119620 22 (8 yr) Backcalculation Station 119770 22 (8 yr) Measured Distress 7 Witczak Equation 25 (5 yr) Backcalculation Station 120415 38 (2 yr) Backcalculation Station 120465 39 (2 yr) Measured Distress 10 Witczak Equation 26 (4.5 yr) AC Rutting (mm) 6 8 (Not reported) 7 (Not reported) 7 (Not reported) Not measured 10 (Not reported) 21 (Not reported) 21 (Not reported) Not measured 11 (Not reported) AC Thermal Cracking (m/km) 189 AC Bottomup cracking (%) 25 AC Topdown Cracking (m/km) 379 375 (5.5 yr) 5 69 254 (12.5 yr) 9 73 254 (12.5 yr) 10 74 884 1 1078 466 (5.5 yr) 25 (15 yr) 289 433 (6 yr) 31 (8.5 yr) 1219 (1 yr) 433 (6 yr) 31 (8 yr) 1260 (1 yr) 1132 0 570 106 12 187 Downloaded from ascelibrary.org by RMIT UNIVERSITY LIBRARY on 01/04/19. Copyright ASCE. For personal use only; all rights reserved. Airfield and Highway Pavements 2017 12 Backcalculation Station 122995 35 (2.5 yr) Backcalculation Station 123345 40 (2 yr) Dynamic Modulus Test 33 (3 yr) Measured Distress 13 17 (Not reported) 22 (Not reported) 16 (Not reported) Not measured 345 (9.5 yr) 19 497 (8 yr) 345 (9.5 yr) 22 559 (5.5 yr) 345 (9.5 yr) 17 457 (10 yr) 394 0 570 The method used to predict asphalt concrete stiffness also has a large effect on predicted bottom-up and top-down cracking, and the magnitude of this effect correlates to the difference in moduli between methods of estimating asphalt stiffness. For example, in Cell 16, the moduli estimated using backcalculation are much lower than the moduli estimated using the Witczak equation (Figure 2). Cell 16 fails by bottom-up cracking in 15 years when the Witczak equation is used to estimate the asphalt stiffness, but in only 8 years when backcalculation is used. Similarly, the predicted top-down cracking in Cell 16 is more than four times greater when backcalculation is used than when the Witczak equation is used. For Cell 15, the moduli estimated using backcalculation and the Witczak equation are similar, and the difference in predicted top-down cracking between these methods is less than 7%. The predicted bottom-up cracking in Cell 15, however, is different between the backcalculation and Witczak equation methods, with an average of 9% predicted when backcalculated moduli are used and 5% predicted when moduli are estimated using the Witczak equation. This lack of correlation may be due to the non-linear transfer function that relates asphalt damage to bottom-up fatigue cracking in Pavement ME. For certain levels of predicted damage, a small change in asphalt stiffness has a large effect on predicted cracking (ARA, 2004). There is no consistent relationship between measured bottom-up and top-down cracking and predicted bottom-up and top-down cracking. There was an average of 0% bottom-up cracking measured for all cells, and between 5% and 31% predicted. For Cell 15, measured top-down cracking is over ten times greater than top-down cracking predicted using either Witczak moduli or backcalculated moduli. For Cell 21, on the other hand, measured top-down cracking is approximately equal to that predicted using either backcalculated moduli or dynamic moduli measured in the lab and three times greater than that predicted using Witczak moduli. Thus, it is not possible to conclude that any single method of estimating the asphalt concrete stiffness more accurately predicts fatigue cracking. Finally, it is important to note that, within each cell, the moduli backcalculated at two different stations give very similar predictions across all distresses. Rutting and fatigue cracking predicted at Cell 21 are an exception to this, however, as these predicted distresses are between © ASCE