Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
50 6.1 The Mechanistic-Empirical Pavement Design Guide The MEPDG (AASHTO 2015) is the current recommended method for the structural design of heavily trafficked pavements in the United States. The MEPDG methodology is implemented in the AASHTOWare Pavement ME Design software. This software predicts distresses and ride quality in various types of pavements (flexible, rigid, semi-rigid/composite) as functions of traf- fic, climate, material properties, and other design inputs. It evaluates flexible pavement perfor- mance based on rutting, fatigue cracking, thermal cracking, and the International Roughness Index (IRI). Three levels of inputs are in the Pavement ME Design software, corresponding to different levels of accuracy. Level 1 data, which provide the highest level of accuracy, typically are project- specific values measured in the field or in the lab. Level 2 data typically are based on correlations and require less measured data from the field or lab. Level 3 data, the least accurate level, provide typical default values for inputs. Material properties for flexible pavements are categorized into three groups: asphalt materials, chemically stabilized materials, and unbound materials. The principal material characteristics are the thickness and stiffness of each layer. Asphalt material stiffness is defined by the dynamic modulus, which takes into account the time-temperature sensitivity of the material. Stabilized and unbound material stiffness levels are specified by the materialsâ elastic and resilient moduli, respectively. Additional inputs for rehabilitation designs include the pavement condition at the time of rehabilitation. Rutting in each layer, percentage of fatigue cracking, and milled thickness are the principal inputs. The damaged modulus as measured from nondestructive testing (NDT) also can be input. The principal outputs from the Pavement ME Design software are the predicted distresses, which are then compared to the design criteria. The primary pavement distress considered for the cold-recycled pavement rehabilitation scenarios is asphalt rutting. Although CIR/CCPR lay- ers could be candidates for bottom-up fatigue cracking, little in the literature suggests this as an important distress mode for the types of pavements considered in this study. The exceptions cited in the literature are primarily from South Africa, where the pavements have high stress-to- strength ratios because of the thin surfacing over the CIR coupled with high traffic/load levels. In the United States, only very lightly trafficked roads are likely to have thin surfacing over the CIR/ CCPR layer. Most other pavementsâand specifically the types of higher traffic volume pave- ments that would be designed using the MEPDG, the focus of this studyâwill have moderately thick HMA surface/wearing courses that will suppress stress ratios below the threshold at which fatigue cracking develops. C H A P T E R 6 Performance Evaluation
Performance Evaluation 51 6.2 Initial Comparisons 6.2.1 Analysis Scenarios Two rehabilitation scenarios having equivalent structural capacity were designed to evaluate HMA versus cold-recycled material performance. The two pavement structures are shown in Figure 40. The first structure is a recycled pavement with a cold-recycled inlay (RP-CIR). It con- sists of, from bottom to top, an A-7-5 subgrade with an input resilient modulus of 5,000 psi, 12 in. of A-1-a granular base with an input resilient modulus of 25,000 psi, 2 in. of existing HMA, 5.5 in. of cold-recycled material, and an HMA wearing course of variable thickness (1.5 in., 2 in., 3 in., and 4 in.). The second structure is an HMA pavement (RP-HMA), but the 5.5 in. cold- recycled layer in the first section is replaced with a 4-in. HMA layer. This difference in overlay thickness is consistent with the typical ratios of structural layer coefficients for these materials in AASHTOâs 1993 Guide for Design of Pavement Structures (i.e., 0.32 for cold-recycled versus 0.40 for a base HMA [Khosravifar, Schwartz, and Goulias 2015]). The RP-HMA structure is the standard against which the cold-recycled inlay in the RP-CIR structure is compared. Level 1 dynamic modulus (|E*|) and RLPD characteristics developed in the 2015 study by Khosravifar, Schwartz, and Goulias were used for the cold-recycled inlay; similar properties for the HMA wearing course and intermediate layers were taken from previous studies. Typical Level 3 prop- erties were used for the existing asphalt. Varying annual average daily traffic (AADT) values consistent with typical traffic volumes were applied. 6.2.2 Analysis Inputs 220.127.116.11 Dynamic Modulus Each pavement section had three different asphaltic material layers: the HMA wearing course, the HMA or CIR inlay, and the existing underlying HMA layer. Level 1 dynamic modulus data for the wearing course and inlay in both sections were used as inputs in the HMA wearing course (variable thickness) HMA wearing course (variable thickness) 5.5 in. cold-recycled mixture 4 in. HMA mixture 2 in. existing asphalt 2 in. existing asphalt 12 in. granular base 12 in. granular base RP-CIR RP-HMA Figure 40. Pavement sections: RP-CIR and RP-HMA.
52 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete Pavement ME Design software (Version 2.0). The HMA wearing surface properties for both sections were taken from a typical Maryland State Highway Administration (Maryland SHA) 9.5 mm NMAS surface mix. Several 9.5 mm NMAS mixes were tested in the lab with a fairly narrow range of dynamic modulus master curves, as shown in Figure 41. The HMA inlay properties for the RP-HMA structure correspond to a typical Maryland SHA 19 mm NMAS mix designated H151B19. The dynamic modulus master curves for this mix and for a range of other 19 mm mixtures are illustrated in Figure 42. For the RP-CR layer, Level 1 dynamic modulus properties from recycled projects within this study were used. The recycled materials had a wider range of dynamic modulus values when com- pared to the HMA mixtures tested, as shown in Figure 43. The three cold-recycled inlay materials selected for evaluation in this comparison were from Delaware, Maine, and San Jose, California. The Delaware project (14-1025), a CIR material using emulsified asphalt as the recycling agent, Upper and lower bounds for typical Maryland SHA 19 mm NMAS mixtures also shown. Figure 42. Master curve for 19 mm NMAS HMA mid-layer mixture (H151B19). Figure 41. Master curve for 9.5 mm NMAS HMA surface wearing course mixture. Upper and lower bounds for typical Maryland SHA 9.5 mm and 12.5 mm NMAS mixtures also shown.
Performance Evaluation 53 exhibited higher laboratory-measured permanent strains in comparison to the other materials. The Maine project (15-1003), a CCPR material also using emulsified asphalt as the recycling agent, had moderate measured permanent strains. The San Jose project (13-1124), a CIR material using foamed asphalt as the recycling agent, had the smallest measured permanent strains. The temperatures and loading rates specified for dynamic modulus testing in AASHTO TP 79 do not conform to the temperatures and loading rates required for input to the Pavement ME Design software. This introduces an intermediate step as a minor complexity. A master curve must be developed externally from the measured dynamic modulus values collected using AASHTO TP 79. Then, the dynamic modulus values at temperatures and loading rates required by Pavement ME Design must be computed for input. The master curve optimization/fitting algorithm embedded in Pavement ME Design then regenerates a new master curve for internal use within the program. However, the Pavement ME Design master curve optimization algorithm did not converge for some of the cold-recycled materials. Ranges of suggested dynamic modulus values for input into Pavement ME Design are included in Tables 14 through 19. For Level 1 mixture |E*| inputs, the Pavement ME Design software also requires a measured binder G* master curve. This is used only by the global aging model when Level 1 mixture |E*| data have been entered. Past studies have shown that the Pavement ME Design distress predictions are insensitive to the binder G* master curve when Level 1 mixture |E*| data have been input. It is sufficient to have a binder G* master curve that is reasonably close to the actual effective binder behavior for the cold-recycled material. 18.104.22.168 RLPD Input Rutting was the main distress measure evaluated in these comparisons. Therefore, the Level 1 rutting coefficient inputs were required for the Pavement ME Design software. The capability of specifying layer-specific rutting coefficients was a new feature added to Version 2.0 of the Pave- ment ME Design software. These coefficients are not input directly as layer properties, but rather as layer-specific calibration values. Upper and lower bounds for the cold-recycled materials in this study also shown. Figure 43. Master curves for cold-recycled overlay materials (Delaware/CIR, Maine/CCPR, San Jose/CIR).
54 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete Temperature Dynamic Modulus (psi) 25 Hz 10 Hz 5 Hz 1 Hz 0.5 Hz 0.1 Hz 14Â°F 811,952 753,611 709,019 605,319 561,177 461,749 40Â°F 912,834 856,773 813,384 710,483 665,744 562,612 70Â°F 545,097 488,271 446,547 355,300 318,943 242,589 100Â°F 278,091 236,651 207,977 150,492 129,610 89,589 130Â°F 196,731 163,746 141,519 98,585 83,582 55,849 Table 14. Suggested dynamic modulus values for CIR projectsâupper range. Temperature Dynamic Modulus (psi) 25 Hz 10 Hz 5 Hz 1 Hz 0.5 Hz 0.1 Hz 14Â°F 1,440,057 1,274,516 1,151,136 878,230 769,393 544,112 40Â°F 1,666,833 1,501,676 1,376,078 1,089,337 970,929 716,479 70Â°F 731,008 601,272 512,384 339,123 278,911 170,307 100Â°F 268,680 203,720 163,365 94,584 73,819 40,654 130Â°F 117,325 85,120 66,253 36,334 27,913 15,119 Table 15. Suggested dynamic modulus values for CIR projectsâlower range.
Performance Evaluation 55 Temperature Dynamic Modulus (psi) 25 Hz 10 Hz 5 Hz 1 Hz 0.5 Hz 0.1 Hz 14Â°F 926,778 862,526 812,585 693,861 642,325 524,408 40Â°F 1,032,195 972,539 925,525 811,263 760,450 640,975 70Â°F 623,417 556,082 506,190 396,141 352,112 259,894 100Â°F 308,060 257,624 222,928 154,384 130,026 84,692 130Â°F 205,122 166,312 140,578 92,298 76,050 47,279 Table 16. Suggested dynamic modulus values for CCPR Projectsâupper range. Temperature Dynamic Modulus (psi) 25 Hz 10 Hz 5 Hz 1 Hz 0.5 Hz 0.1 Hz 14Â°F 553,849 492,244 446,812 347,297 307,773 225,594 40Â°F 625,372 562,692 515,819 410,920 368,278 277,420 70Â°F 293,825 246,546 213,915 149,123 125,967 82,624 100Â°F 123,317 97,360 80,712 50,844 41,242 24,887 130Â°F 60,230 45,910 37,140 22,304 17,805 10,483 Table 17. Suggested dynamic modulus values for CCPR projectsâlower range.
56 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete Temperature Dynamic Modulus (psi) 25 Hz 10 Hz 5 Hz 1 Hz 0.5 Hz 0.1 Hz 14Â°F 1,071,198 1,037,336 1,010,164 941,923 910,367 832,404 40Â°F 1,122,287 1,092,240 1,067,996 1,006,561 977,875 906,204 70Â°F 898,442 854,246 819,495 735,019 697,311 607,839 100Â°F 683,188 632,276 593,431 503,380 465,194 379,587 130Â°F 545,072 494,033 455,994 370,944 336,258 261,676 Table 18. Suggested dynamic modulus values for FDR projectsâupper range. Temperature Dynamic Modulus (psi) 25 Hz 10 Hz 5 Hz 1 Hz 0.5 Hz 0.1 Hz 14Â°F 692,192 647,721 612,802 528,551 491,434 405,248 40Â°F 733,752 691,449 657,969 576,150 539,595 453,347 70Â°F 477,733 428,573 391,774 309,450 276,058 205,246 100Â°F 273,960 232,529 203,434 144,313 122,701 81,466 130Â°F 162,615 132,148 111,827 73,463 60,490 37,474 Table 19. Suggested dynamic modulus values for FDR projectsâlower range.
Performance Evaluation 57 To derive the rutting calibration coefficients, the MEPDG rutting model was fit to the measured laboratory RLPD test results. The MEPDG rutting model is as follows: 10 (6)1 1 2 2 3 3k T N p r z r k k kr r Îµ Îµ = Î² Î² Î² in which Îµp = the measured permanent strain, Îµr = the resilient strain, T = temperature (Â°F), N = number of load repetitions, k1, k2, and k3 = RLPD properties for the secondary portion of the response, and Î²r1, Î²r2, and Î²r3 = field calibration coefficients. The kz term is a depth correction function given as ( )( )= + 0.328196 (7)1 2k C C zz z with = â + â0.1030 2.4828 17.342 (8 )1 2C H H aa a = â â0.0172 0.7331 27.428 (8 )2 2C H H ba a in which z = depth from the surface, and Ha = total thickness of the asphalt layers. The depth function kz was set to 1, as it is not relevant for interpreting laboratory test data hav- ing homogeneous stress conditions. The Î²r1, Î²r2, and Î²r3 field calibration coefficients also were set to 1. The resilient strain Îµp required for the strain ratio dependent variable Îµp/Îµr is not measured or recorded by the AMPT used in this study and thus was estimated using the deviator stress and measured unconfined dynamic modulus |E*| at the RLPD test temperature and frequency (10 Hz). Best estimates for the k1, k2, and k3 material coefficients were determined through least squares multivariate linear regression analysis in a transformed log-log coordinate space. It has been assumed here that the HMA rutting model in the MEPDG also applies to bitu- minously stabilized cold-recycled materials. Given that the laboratory RLPD behavior of cold- recycled materials is similar to that of HMA (i.e., both materials follow the standard power-law relationship between permanent strains and number of load cycles), this assumption seems rea- sonable. It could be disproved if there were large discrepancies between predicted and measured rutting of cold-recycled pavements under in-service conditions, but these data do not exist at present. Although not evaluated for this research, it is also assumed that the IRI prediction mod- els apply to cold-recycled materials. This assumption also seems reasonable, as the IRI prediction models are a function of total rutting only and do not distinguish among the layer sources. One problem in the analysis is that the RLPD tests on the cold-recycled materials were per- formed at a single temperature. The MEPDG rutting model (Equation 6) is dependent on temperature, so plastic strain data for at least two other temperatures are needed. The tech- nique developed by Khosravifar et al. (2015) was used to predict plastic deformations at other temperatures. The process is similar to fitting a master curve; after a reference temperature is picked, the temperature shift function determined from the dynamic modulus testing is used to shift the permanent strain data to the desired temperature, as show conceptually in Figure 44.
58 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete The process uses the concept of reduced load cycles and reduced intercept, similar to the way dynamic modulus testing uses the concept of reduced frequencies: [ ]( ) ( )( ) = âlog log log (9 )N N a T aR [ ]( ) ( ) ( )â² = +log log log (9 )A A B a T b in which N = number of load cycles, NR = reduced number of load cycles, A = intercept of the secondary zone of the RLPD response, Aâ² = reduced intercept value, and B = slope of the secondary zone of the RLPD response. A typical fitted RLPD model for a cold-recycled material is shown in Figure 45. The HMA materials were tested at three temperatures, so there was no need to predict the permanent deformations at other temperatures for these materials. During the fitting procedure for some mixes, negative values were obtained for the temperature coefficient, which suggests that resilient strains are more sensitive to temperature than are perma- nent strains. Figure 46 shows the rates of change for the resilient strains and the plastic strains at the 10,000th cycle as a function of temperature. Each strain type is normalized by its respective strain at 20Â°C. For the Delaware CIR material, the plastic and resilient strains vary nearly identically with temperature, implying that their strain ratio (Îµ/Îµ at 20Â°C) is insensitive to temperature (i.e., that k2 is nearly zero). On the other hand, the San Jose, California CIR material shows resilient strains increasing faster than plastic strains with temperature, implying that the ratio of plastic to resilient strains decreases with temperature (i.e., that k2 is negative). These negative coefficients were hypothesized to be the consequence of using unconfined dynamic modulus values to estimate the resilient strains. Dynamic modulus is sensitive to confinement at high temperatures. Because resilient strains were not measured in the AMPT, they were estimated from the applied stresses and the dynamic modulus of the material. However, the RLPD test is performed Figure 44. Time-temperature shift factor to form a RLPD master curve using method from Khosravifar et al. (2015).
Performance Evaluation 59 Strain ratio = Îµ Îµ at 20Â°C . Delaware CIR Mixture (Positive temperature coeï¬cient kr2) San Jose, California CIR Mixture (Negative temperature coeï¬cient) Figure 46. Strain ratios for two mixes, one with a positive temperature coefficient and one with a negative temperature coefficient. Figure 45. Typical RLPD fitted model for cold-recycled mixture 15-1003.
60 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete under confined conditions while the dynamic modulus test is performed under unconfined conditions. Previous researchers have shown that there can be significant differences between confined and unconfined dynamic modulus values, especially at the high temperatures of the RLPD test (Seo et al. 2007; Yun et al. 2010; Zhao et al. 2013). Confined dynamic modulus values are required to estimate more realistically the resilient strains in the RLPD test. Zhao et al. (2013) proposed a model to derive confined dynamic modu- lus values at different confining pressures: ( )( ) = Î´ + Î± + + â +( ) ( )Î²+Î³ Ï â â + Ï ln * 1 1 (10) ln 5 ln 6 0 6 3 4 E e C e e e C P C P C Cr r in which P0 = test confining pressure, P = desired confining pressure, Ïr = reduced loading frequency, Î´, Î±, Î³, Î² = master curve fitting parameters (similar to Equation 2, which is discussed in the section on âDynamic Modulus Testingâ in Chapter 3), and C3, C4, C5, C6 = fitting parameters. The model must be calibrated with confined dynamic modulus data. Zhao et al. (2013) per- formed the calibration for 19 mm and 25 mm NMAS Superpave HMA mixtures. Zhao et al.âs calibrated coefficients for the 19 mm NMAS mixture were assumed to be representative for the cold-recycled materials tested in this study. The coefficients are summarized in Table 20. These calibrated coefficients were therefore used to estimate the confined modulus values for the HMA and cold-recycled materials in this study at the 10 psi confining pressure in the RLPD test, and thus to estimate the resilient strains for the confined conditions. Figure 47 shows C3 C4 C5 C6 1.632 0.421 4.031 3.259 Table 20. Calibration coefficients for confined master curve. Figure 47. Confined versus unconfined dynamic modulus master curves for Project 13-1114.
Performance Evaluation 61 typical confined versus unconfined dynamic modulus master curves for a cold-recycled material as determined using this procedure. A recommendation for future research is to either measure confined dynamic modulus directly for the cold-recycled materials or have resilient strain reported directly during RLPD testing using the AMPT. Summaries of the MEPDG RLPD material coefficients attained from applying these proce- dures to the HMA and cold-recycled materials are provided in Table 21 and Table 22, respec- tively. For completeness, all HMA and cold-recycled materials evaluated in this study are included in the tables. 22.214.171.124 Traffic Input Differing traffic loads were applied for the various HMA wearing course thicknesses. Appropri- ate AADT values were determined based on the 1993 AASHTO flexible pavement design stan- dard. The 1993 AASHTO procedure predicted 10 million, 15 million, 27 million, and 46 million equivalent single axle loads (ESALs) over a 20-year design life for pavement structures having HMA wearing courses with thicknesses of 1.5 in., 2 in., 3 in., and 4 in., respectively. For inputs to the Pavement ME Design software, the vehicle mix was set as a 100% distribution of Class 5 vehicles. Class 5 vehicles include 2-axle vehicles with dual rear tires (e.g., single-unit trucks, mini school buses, and camping vehicles). To simplify the traffic loading, the load for all rear axles was set at 18 kips (i.e., one ESAL), and the load for all front axles was set at zero. The traffic distribu- tion was assumed to be constant over all months with zero growth rate. 126.96.36.199 Climate Input All analyses were conducted for temperate climates. The Baltimore, Maryland, weather station data were used as input to the MEDPG. 6.2.3 Comparisons of Predicted Performance The predicted asphalt rutting results for the three HMA surface course thicknesses are shown in Figure 48. The predicted rut depths include the contributions from all bituminous materialsâ the HMA surface course, the HMA/cold-recycled structural inlay, and any additional rutting in Mixture k1 k2 k3 H160A09 1.55E-01 0.821 0.163 H151B19 6.64E-01 0.407 0.228 H135A19 6.45E+00 0.101 0.117 H077A09 5.22E-01 0.610 0.130 H083A12 6.06E-02 1.018 0.138 H127A12 5.09E+00 0.100 0.144 H135A12 1.66E-01 0.787 0.124 H168A09 8.25E+00 0.0002 0.116 Table 21. MEPDG RLPD material coefficients for HMA mixtures.
62 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete Project (Mixture) k1 k2 k3 13-1093 3.72E-07 3.3054 0.340 13-1111 1.61E-09 4.5055 0.530 13-1112 4.47E-03 1.5668 0.366 13-1113 5.62E-02 0.9332 0.304 13-1114 1.88E-01 0.7226 0.314 13-1115 1.59E+01 0.0687 0.027 13-1116 2.97E+00 0.2029 0.071 13-1117 6.06E-03 1.4279 0.126 13-1124 3.15E+00 0.1545 0.149 13-1127 2.11E+00 0.0968 0.155 14-1001 3.01E-04 2.1380 0.362 14-1002 1.73E-16 7.3551 0.705 14-1003 3.87E-03 1.5091 0.279 14-1011 2.92E-03 1.3187 0.183 14-1025 2.73E-02 1.1658 0.346 14-1026 3.03E-04 2.1724 0.413 14-1027 1.06E+00 0.5540 0.181 14-1028 3.17E-01 0.6648 0.118 14-1055 1.99E-08 4.0716 0.470 14-1057 2.39E-02 1.2063 0.140 14-1058 3.88E+00 0.2544 0.156 14-1062 5.58E-01 0.5587 0.168 15-1002 7.48E-05 2.3050 0.383 15-1003 4.47E+00 0.1373 0.159 Table 22. MEPDG RLPD material coefficients for cold-recycled mixtures.
Performance Evaluation 63 the underlying existing HMA material. From Figure 48, the predicted rutting performance for the CCPR materials from Maine and the CIR materials from San Jose, California, were similar or better to that predicted for the HMA inlay scenario. However, the predicted rutting performance for the CIR materials from Delaware was much worse than that predicted for the HMA inlay at HMA wearing course thicknesses of 1.5 in. and 3 in. The predicted rutting for the Delaware CIR with a 4-in. HMA wearing course was similar to that predicted for the HMA inlay condition with a 4-in. HMA wearing course. The two recycled materials that showed similar or better predicted rutting performance than HMA also included cement as a chemical additive. The recycled material that showed worse predicted rutting performance than HMA did not include any chemical additive. The higher rutting susceptibility of the Delaware CIR material can be observed in the labora- tory RLPD curves where the Delaware CIR material is at upper range of plastic strains at all tem- peratures. In the future, the interactions between pavement structure and cold-recycled material properties could be examined more comprehensively to develop guidelines for the appropriate inlay and surface wearing course thicknesses. 6.3 Rutting Performance Evaluation of All Cold-recycled Materials In addition to the initial three projects selected to represent the range of material quality, all recycled projects assessed in this study were similarly analyzed for rutting performance. Dif- ferent climatic conditions were also added to the study to observe the effect of temperature on rutting performance. Three locations having different climatic characteristics were selected: Maryland (temperate), Arizona (hot), and Minnesota (cold). The pavement sections considered were the same as those used in prior analyses (see Figure 40). Four structural sections were analyzed (1.5-in., 2-in., 3-in., and 4-in. HMA surface wearing course thicknesses). For comparison purposes, approximated FDR âinlaysâ were included in the analyses along with CIR and CCPR inlays. (The pavement sections do not accurately represent how FDR * HMA surface wearing course thickness. Figure 48. Asphalt rutting for selected cold-recycled inlays in comparison to HMA inlay.
64 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete materials would be used in a real pavement structure, but the approximation permits a better comparison of the FDR performance to the other cold-recycled materials.) The predicted asphalt rutting results are summarized in Figures 49 through 51; the asphalt rut- ting includes the contributions from all of the bituminous layersâsurface HMA, cold-recycled/ HMA overlay, and additional rutting from underlying existing HMA layer. Projects 13-1093, 13-1115, 13-1117, 13-1127, 14-1001, 14-1002, 14-1057, and 15-1002 are not included in in ch es FDR - Emulsion FDR - Foam CIR - Emulsion CIR - Foam CCPR Figure 49. Predicted rutting performance for all cold-recycled projects: Arizona weather (hot). CCPRCIR - Foam CIR - Emulsion FDR - Foam FDR - Emulsion in ch es Figure 50. Predicted rutting performances for all cold-recycled projects: Maryland weather (temperate).
Performance Evaluation 65 Figures 49 through 51. The program algorithms of the Pavement ME Design software were unable to fit master curves to the input laboratory-measured dynamic modulus data for these projects; consequently, these projects could not be analyzed. A more robust dynamic modulus master curve algorithm may be warranted for the Pavement ME Design software. In addition, rutting could not be predicted for the I-81 cold in-place recycling lane because no RLPD testing was performed for the CIR material. As can be observed in Figures 49 through 51, five of the 17 analyzed projects had very high rut- ting values. These large rutting values are a consequence of the input RLPD material coefficients. The coefficients on temperature and traffic load strongly influence predicted rutting. The five projects with the highest predicted rutting (13-1111, 13-1112, 14-1025, 14-1026, and 14-1055) also exhibited high permanent strains in the laboratory RLPD tests as compared to the other cold-recycled materials. The ranges of predicted rutting for all of the cold-recycled projects analyzed as compared with the predicted rutting of conventional HMA inlays are summarized in Figures 52 through 54. In these box-and-whisker plots, the vertical lines delineate the minimum and maximum limits of the data, the boxes delineate the first and third quartile values, and the horizontal lines in the middle of each box represent the median of the data. Each dot represents a single project, and project types (CIR/FDR/CCPR) are differentiated by color. The blue lines in the graphs represent the predicted rutting for the HMA inlay sections at each wearing course thickness. The CIR projects exhibited a greater range of values in part because there were more of them. The Pavement ME Design software could be used to analyze only two CCPR and three FDR projects. The trends seen in Figures 52 through 54 clearly show that wearing course thickness is an important factor for predicted rutting. The cold-recycled inlay sections having 3 in. and 4 in. wearing courses had predicted rutting values in a narrow range with a mean value very close to their HMA inlay counterparts for all weather conditions. Rutting decreased as the wearing course thickness increased despite increases in input traffic volume (noted as ESAL) with increased FDR - Emulsion FDR - Foam CIR - Emulsion CIR - Foam CCPR in ch es Figure 51. Predicted rutting performance for all cold-recycled projects: Minnesota weather (cold).
66 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete Figure 52. Predicted rutting: Arizona weather. Figure 53. Predicted rutting: Maryland weather.
Performance Evaluation 67 wearing course thickness. It can be concluded that, as long as HMA wearing course thickness is above some thresholdâapproximately 2 in. to 3 in.âthe cold-recycled inlay sections exhibit predicted rutting performance comparable to that of conventional HMA inlay sections. Figures 52 through 54 also show that the range and mean values of predicted rutting for the cold-recycled inlay sections decreases with decreasing temperature, as was expected. The mean predicted rutting under all three weather conditions is acceptable, except perhaps for the thinnest wearing course (1.5 in.). The average values of predicted asphalt concrete rutting for the temperate Maryland weather conditions for the 1.5 in., 2 in., 3 in., and 4 in. wearing course thicknesses were 0.44 in., 0.31 in., 0.16 in., and 0.084 in., respectively. The last two values are well below this studyâs assumed default design limit of 0.25 in. For the thinner (1.5 in. and 2 in.) wearing courses, the rutting performance can still be considered reasonably good, considering that the traffic applied to these sections was quite high. Although cold-recycled rehabilitation had historically been most commonly used on low volume roads, the results from the present analyses suggest that with a wearing course thickness of more than 2 in., these cold-recycled materials can be used successfully in higher traffic roads. As discussed previously, the five projects with poor rutting performance also had substandard laboratory RLPD behavior. Good quality cold-recycled materials that exhibit satisfactory laboratory RLPD behavior should exhibit satisfactory predicted rutting performance similar to that of conventional HMA mixes. The five projects with the highest predicted rutting along with selected performance criteria are summarized in Table 23. The arrows in the table indicate whether the respective property fell at the high end or low end of the range of values as compared to the other cold-recycled materials. The results suggest that a higher slope of the RLPD permanent strain versus load cycles is most Figure 54. Predicted rutting: Minnesota weather.
68 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete correlated with higher predicted rutting, followed by the RLPD intercept and, in two cases, by the E* lower shelf. Project 13-1111 (a CIR project with foamed asphalt as the recycling agent and exhibiting the highest rutting) had a very high k3 value (exponent on N) at 0.53, double the average value of 0.27 for all cold-recycled materials in this study (see Table 22). This project also had a very high k2 value (exponent on T) at 4.51, nearly three times the average value of 1.59. (Project 14-1002, although also having high k3 and k2 values, had an exceptionally low k1 value.) Among all of the cold-recycled mixtures that could be analyzed using the Pavement ME Design software, the five materials listed in Table 23 were among the top six with respect to the highest k3 values and among the top nine with respect to the highest k2 values. Interestingly, four of the five projects listed in Table 23 included no chemical additive. Project Type Stabilizer E* Lower Shelf RLPD Intercept RLPD Slope 13-1111 CIR Foam 14-1026 CIR Emulsion 14-1055 CIR Emulsion 13-1112 CIR Emulsion 14-1025 CIR Foam Table 23. Projects with highest predicted rutting.