Layer Coefficient Calibration of Fiber Reinforced Asphalt Concrete Based On Mechanistic Empirical Pavement Design Guide

Transcription

1 Layer Coefficient Calibration of Fiber Reinforced Asphalt Concrete Based On Mechanistic Empirical Pavement Design Guide Prepared by B. Shane Underwood, Ph.D. Assistant Professor Waleed Zeiada, Ph.D. Postdoctoral Scholar School of Sustainable Engineering and the Built Environment Department of Civil, Environmental, and Sustainable Engineering Arizona State University, Tempe, AZ Submitted to FORTA Corporation 100 Forta Drive Grove City, PA March 10 th, 2015

2 EXECUTIVE SUMMARY The primary objective of this study is to estimate the potential thickness reduction possible by using FORTA fiber-reinforced asphalt concrete (FRAC) on an overlay project in the country of Peru. Technical data for conditions in the United States are used to develop these estimates because experimental data for Peru is not available. However, the data used does include material characteristics from more than 12 laboratory studies and FRAC performance predicted from the mechanistic-empirical pavement design guide for a range of climatic conditions. To carry out the study, Mechanistic Empirical Pavement Design Guide (MEPDG) simulations were conducted to predict the pavement performance. Comparative assessment was performed based on conventional AC and FRAC. The MEPGD material inputs of the fiber-reinforced asphalt concert mixture using 0.5 kg/tonne fiber dosage were based on the average results from 12 laboratory studies. The conventional AC case was first simulated so that the total rutting was approximately 19 mm after a service life of 20 years. MEPDG runs were then performed for 0.5 kg/tonne cases under decreasing AC thicknesses until either same rutting or the same bottom up fatigue cracking of the control case was achieved. The reduction of the AC thickness was then used to calibrate a FRAC-based layer coefficient Based on this analysis, the FRAC layer coefficient values vary depending on the climate and the subgrade strength conditions. The most conservative estimate for the layer coefficient of FRAC is 0.52 (when the thickness is in the unit of inches). In the Peruvian project, the overlay mixture design is 5 cm, which is determined using a layer coefficient for asphalt as 0.44 (when the thickness is in the unit of inches). In this scenario a FRAC mixture could be theoretically placed at a thickness of approximately 4.2 cm (a 16% reduction). If the FRAC were used in both the MAC-2 (3 cm replacement of milled material) and MAC-3 mixtures, the total thickness could be reduced to 6.8 cm. 1

3 INTRODUCTION Although many highway agencies are exploring the use of new mechanistic empirical pavement design (MEPDG) method, many currently still use a version of so-called empirical design method released in either 1972, 1987, or One of the key inputs to this method is the layer coefficient for the hot mix asphalt (HMA) layers. This HMA layer coefficient has not been updated since then despite the advances and improvements in mix design methods, new materials and modifiers technologies, quality control, and construction of HMA. For HMA, AASHTO recommends using a layer coefficient of 0.44 (if the units of thickness are inches) (1). This value was taken as a weighted average of values obtained from the Road Test, which ranged from 0.33 to 0.83 (2). This range of values is surprisingly broad, and the AASHTO Design Guide recommends that each state determine or recalibrate coefficients individually if possible. Only limited studies have been conducted to recalibrate the AC layer coefficient either based on actual pavement performance (3) or pavement performance form accelerated pavement facilities (4). Results from these studies have showed that the AC layer coefficient under the current practice is 0.54 (4). The use of synthetic fibers to reinforced HMA pavements has shown major benefits with respect to pavement performance. FORTA Corporation introduces FORTA-FI, a next-generation synthetic fiber blend for reinforcing asphalt designed to reduce rutting and cracking. The technological capabilities of these fibers have been demonstrated in multiple laboratory and field studies over the years. Using the empirical design method to design FRAC pavements is cumbersome as the AC layer coefficient depends only on the stiffness at 68 F, where FRAC show only an average of 13% increase compared to conventional AC. Assuming an AC layer coefficient of 0.44, this increase in FRAC modulus results in only a 6% higher layer coefficient (0.47) or, equivalently 6% lower AC layer thickness. The primary problem with this established relationship between modulus and layer coefficient is that only one relationship, which was fitted to a wide range of limited data, is used (2). This means that the effect of the stiffness changes on the AC layer coefficient might be more than this relationship suggests if other factors (structural, traffic, and environmental conditions) are considered. For this reason, this study investigates the sensitivity of the AC layer coefficient to FRAC stiffness change considering different structural, traffic, and climate conditions. 2

4 PREDICTION OF PAVEMENT PERFORMANCE USING MEPDG The pavement performance of the control and the 0.5 kg/tonne reinforced mixture was predicted using the Mechanistic Empirical Pavement Design Guide (MEPDG). For this analysis MEPDG version 1.1 has been utilized. Two main distresses were considered; total rutting and bottom-up fatigue cracking (alligator cracking). The MEPDG simulations were conducted with respect to national calibration. For the FRAC pavement, the alligator cracking distress prediction model was calibrated to increase the FRAC pavement resistance against alligator cracking by three times compared to the control pavement (concluded based on the average laboratory performance of 9 fiber studies). The dynamic modulus E* values used for the FRAC pavement simulations were also increased by 13% which is the average increase in the modulus due the use of FORTA fibers based on the results obtained from 11 FORTA studies. The AC rutting distress prediction model for the FRAC pavement was kept the same as the control pavement as there is not enough data to calibrate the model for fiber-reinforced mixture. Table 1 shows the conventional and the fiber-reinforced E* values used in the MEPDG simulations where the E* values of a typical 19-mm conventional dense graded mixture was used and then was increased by 13% for the fiber-reinforced mixture. Table 2 shows the MEPDG inputs for both control and FRAC pavements. Two climate conditions (warm and moderate to cold) and three subgrade stiffness (soft, moderate, and stiff) were considered. Only one pavement structure was used for the control section for different climate and subgrade conditions. It should be noted that since the MEPDG is currently only capable of performing analysis in the US customary units, all quantities given henceforth will be listed in US customary system since all inputs were made as round numbers in this system. Preliminary MEPDG simulations were conducted on control pavement sections so that a rutting failure is attained after 20 years from the initial construction. The total rutting failure criteria was 0.75 inch (approximately 19 mm). To accumulate 0.75 inch rutting with different conditions, traffic level was adjusted accordingly via trial simulations. The MEPDG simulations were then conducted on the FRAC pavement at different AC thickness until the FRAC pavement structure that gives the same total rutting compared to the control section is obtained. The future traffic for each simulation period was predicted using 4.0% compound traffic growth rate. Figure 1 and Figure 2 show a comparison of predicted total rutting and alligator cracking of FRAC pavement as a function of AC depth for Phoenix and Raleigh climate conditions respectively. In both figures, threshold value of the control pavement at AC thickness of 5 inches (12.7 mm) were plotted for each combination to determine the FRAC thickness that is corresponding to similar rutting or alligator cracking performance as for the control pavement depending on which distress was more critical. It is observed that the rutting distress was always more critical than alligator cracking distress for all the considered cases. Figure 3 shows an example of determining the equivalent FRAC thickness. Equivalent FRAC thickness was determined as the maximum thickness required to yield the same rutting or fatigue that the conventional pavement exhibited with 5 inches AC thickness. Figure 3 showed that the FRAC thicknesses required to show the same rutting and alligator cracking as of control were 4.15 and 3.75 inches (10.5 and 9.5 mm) respectively. In this case, the equivalent FRAC thickness was chosen as 4.15 inches which is corresponding to the more critical distress (rutting), where greater FRAC thickness is needed. 3

5 Table 1 E* Values for Conventional and Fiber-Reinforced Mixtures Dynamic Modulus, ksi - MPa (Test Values) Temp. o F ( o C) 14 (-10) 40 (4.4) 70 (21.1) 100 (37.8) 130 (54.4) Freq. Hz Fiber-Reinforced Conventional Modular Ratio 25 6,847 47,206 6,059 41, ,313 43,528 5,587 38, ,215 42,850 5,500 37, ,631 38,822 4,983 34, ,397 37,206 4,776 32, ,760 32,812 4,212 29, ,736 32,654 4,191 28, ,551 31,378 4,027 27, ,286 29,548 3,793 26, ,621 24,961 3,204 22, ,322 22,905 2,940 20, ,663 18,359 2,357 16, ,552 17,590 2,258 15, ,223 15,326 1,967 13, ,989 13,715 1,760 12, ,454 10,023 1,287 8, ,252 8,630 1,108 7, , , ,141 7,865 1,010 6, , , , , , , , , , , , , , , , , , , , , ,

6 Table 2 Inputs of MEPDG Simulations Traffic Data Initial two-way Annual Average Daily Truck Traffic (AADTT) 3,000-3,500-4,500-7,500-8,500-18,000 Number of lanes in design direction 2 Percent of trucks in design direction (%) 50 Percent of trucks in design lane (%) 80 Operational speed, kph (mph) 60 (96.6) Traffic Growth Factor Comp. 4% Climate Conditions Weather Station Phoenix Airport, PHX Raleigh, NC Latitude (degrees.minutes) Longitude (degrees.minutes) Elevation, ft (m) 1106 (337) 403 (131) Depth of water table, ft (m) 20 (6.1) 20 (6.1) Mean annual air temperature, F (ºC) (22.92) (14.93) Pavement Section Data Material type Asphalt concrete Layer 1 Layer thickness, in. (cm) 5 (12.5) Reference temperature, F (ºC) 70 (21.1) Unbound Material Crushed Stones Thickness, in. (cm) 14 (35) Layer 2 Modulus, psi (kpa) 30,000 (206,843) Plasticity Index, PI 1 Liquid Limit, LL 6 Unbound Material A2-4, A4, A7-6 Layer 3 11,500-16,500-21,500 (79, ,763- Modulus, psi (kpa) 148,237) 5

7 Total Rutting (in) Total Rutting (in) Total Rutting (in) (a) (b) (c) Figure 1 Total rutting and alligator cracking results of FRAC pavement for Phoenix climate (a) soft subgrade, (b) moderate subgrade, and (c) stiff subgrade. 6

8 Total Rutting (in) Total Rutting (in) Total Rutting (in) (a) 0.71 (b) (c) Figure 2 Total rutting and alligator cracking results of FRAC pavement for Raleigh climate (a) soft subgrade, (b) moderate subgrade, and (c) stiff subgrade

9 Total Rutting (in) (a) 4.15 in in. (b) Figure 3 Determination of equivalent FRAC thickness for Raleigh with soft subgrade condition (a) with respect to total rutting, and (b) with respect to alligator cracking. Based on this analysis, it can be observed that, the FRAC pavement requires less AC thickness for all the cases compared to the control pavement to yield the same total rutting performance as of control pavement (Table 3). It is observed that the enhancement in rutting performance for the FRAC pavement compared to the control pavement increases for warmer climates and stiffer subgrade soils. The only exception was for the low subgrade condition under Raleigh climate condition, where the FRAC low subgrade condition showed lower total rutting compare to the medium subgrade condition. One possible reason for this observation is that the subgrade strength increases substantially during the freezing season, which decreases the subgrade rutting considerably. Knowing the AC thickness for the control and FRAC pavement, the layer coefficient of the FRAC layer can be calculated through Equation (1), where control and FRAC 8

10 pavement have the same structural number (SN) assuming the layer coefficient of the control AC layer is 0.44 as used by most of the DOTS in the US. SN acontol DControl afrac DFRAC (1) where: SN = Structural number of the AC layer, a control = Layer coefficient of the control AC layer, D control = Thickness of the control AC layer, A FRAC = Layer coefficient of the FRAC layer, and = Thickness of the FRAC AC layer. D FRAC Table 3 includes a summary of the FRAC layer coefficient and percent change for different climate and subgrade condition. It can be observed that the FRAC based layer coefficient ranjges from 0.52 to 0.62 with a percent change ranging from of to 40.85% compared to the control. Thus the used of the FORTA fibers can reduce the thickness of the AC layer/increase the AC layer coefficient by 19% to 41%. Table 3 Summary FRAC Layer Coefficients and Percent Changes for Different Climate and Subgrade Conditions. Control Layer FRAC Layer Control AC FRAC Percent Subgrade Climate Thickness Thickness Layer Layer Change Strength (inch) (inch) Coefficient Coefficient (%) Phoenix, AZ Raleigh, NC Low Medium High Low Medium High REFERENCES 1. AASHTO (1993). AASHTO Guide for Design of Pavement Structures. American Association of State Highway and Transportation Officials, Washington, D.C. 2. HRB (1962). The AASHO Road Test. Special Reports 61A, 61C, 61E. Highway Research Board. 3. Michael Pologruto. (2006) Study of In situ Pavement material properties determined from FWD testing. Journal of Transportation Engineering, Vol. 132(9), pp Davis K.P. and D.H. Timm. (2009) Recalibration of the Asphalt Layer Coefficient, NCAT Report 09-03, Auburn University. 9