Models for Predicting Reflection Cracking of Hot-Mix Asphalt Overlays (2010)

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.

48 Introduction This project developed a HMA overlay design program which is based on calibrating mechanistically predicted reflec- tion cracking lives to observed overlay reflection cracking distress. The program is written to be a subprogram to the MEPDG software for new pavements. The program uses the same input and type of input that the MEPDG software requires and additional data that is appropriate to the design of an over- lay. This includes the condition of the pavement at the time of the overlay construction and the properties of the layers of the old pavement, which may be assumed or measured by nondestructive testing means. The design subprogram per- mits the user to make exploratory runs with trial mix designs, thicknesses, and reinforcing or strain-absorbing interlayers. As with the MEPDG, the overlay design subprogram requires the traffic to be input as an axle load spectrum. Over 150 sep- arate weather databases were assembled in the course of this project and are incorporated into the files available to the user of this subprogram in addition to others that may be generated using the Enhanced Integrated Climatic Model (EICM). All data may be input at one of three levels of detail selected by the user (Levels 1, 2, or 3). The project has also developed a model calibration program which may be used to create a set of calibration coefficients, for use in the design model, to design overlays calibrated to specific local or regional conditions and practices. The design program is arranged as a single mechanistic crack growth model with different sets of calibration coefficients for overlays in different climatic zones or with a different pavement structure, such as overlays over cracked asphalt surfaces or over jointed concrete pavements. A total of 11 separate types of pave- ments were identified with a sufficient amount of quality data to develop sets of calibration coefficients. The final set of cali- bration coefficients for each type of overlay was selected to reproduce, as closely as possible, the actual field observations of the growth of reflection cracking distress in both its extent and severity. The process by which these sets of model calibration coefficients were developed is described in Chapter 2 (more details are provided in Appendices L, M, and N). Twenty two sets of calibration coefficients were developed, two sets for each combination of overlay and climatic zone (for which data were available). The Model Development Process The application of mechanics to the prediction of reflection cracking through HMA concrete overlays involves a number of steps including the use of finite element analysis of crack growth and the modeling of those results with an ANN algorithm to speed up the computational time. The computational task of determining the material properties of the overlay under a vari- ety of loading conditions and temperatures, including traffic and thermal stresses, must be done rapidly to make the design model a practical tool for design. A third part is to generate accurate weather characteristics that can be used to provide realistic material properties and stresses throughout each day and over the observed service life of an overlay. A fourth part of the assembly of this model is to develop a consistent means of describing the distress observed in the field. The fifth part was to devise a means to relate the predicted distress to the observed distress in a simple way, and produce predicted distress that matched well with what was actually observed in the field. A discussion of these five steps follows. Mechanistic Prediction of Crack Growth The model was selected for the Design Program based on several factors. One of the most important factors was the speed with which daily crack growth could be computed to facilitate consideration of several material, thickness, and re- inforcing options in the overlay. This led to the decision to use the ANN algorithms to compute both the changing modulus of the overlay mix and the growth of the cracks up from the C H A P T E R 3 Interpretations, Appraisal, and Applications

49 cracks in the existing pavement and to develop mechanistic data with which to “train” the ANN algorithms for crack growth. This was done by calculating a large set of stress intensity factor data for a variety of overlay and pavement structures using a two-dimensional finite element approach with the transverse third dimension being represented by a series solution. When the calculated results were compared to the correct answers generated by a full three-dimensional set of computational results, the errors were acceptably small (discussion is provided in Appendix Q). The ANN models were found to fit all 18 sets of computed databases very well (as described in Appendix F). Neither the two- nor the three-dimensional finite element analysis was used within the Design Program because of the long computational time that each requires. HMA Overlay Material Properties The ANN algorithms were also used for generating the material properties of a HMA overlay material as it responds to traffic and thermal stresses. It was also necessary to use a well constituted and widely available database of HMA material properties to represent these properties. The database assem- bled and modeled by Witczak in 1999 and 2006 (2, 3) satisfied these criteria. Representing both models by an ANN algorithm provided two models that were computationally fast yet have a better coefficient of determination (R2) than Witczak’s regres- sion models. By inputting binder properties, some aggregate gradation, volumetric composition of the mix, and frequency of loading and temperature into these ANN algorithms, HMA mixture properties over a wide range of loading times and tem- peratures can be generated quickly. The binder properties used as reference properties within the program were binder properties extracted from field cores and reported earlier (4). Although the user may input other binder properties at the detailed Level 1 input, binder data from actual constructed pavements may be used for Level 2 or 3. The PG of the binder and the internal reference data may be used for Level 2 to generate the remainder of the required data. The fracture properties of an asphalt mixture depend on simpler and more fundamental properties of that mixture as shown by Schapery (36, 37) and confirmed in other studies (4), which calibrated to field fatigue cracking data. The calibration coefficients that were developed in these studies were used in this project without any alteration, even though the type of distress was different. Weather Data and Temperature Prediction Accurate temperature prediction is key to making accurate predictions of thermal crack growth, especially in an overlay. Comparisons, between the temperature predictions and actual temperatures measured in the field, demonstrated the need for a higher degree of accuracy in calculating the temperature within the overlay. Such a temperature model was incorpo- rated into the Design Program (characteristics of this model are described in detail in Appendix B). The temperatures calcu- lated by this program and those measured in the field rarely dif- fered by more than 2° C. A complete set of weather data were assembled from databases available to the public for about 150 different locations within the United States and Canada for use in the Design Program. Consistent Description of Reflection Cracking Distress The S-shaped curve for the accumulating extent of reflec- tion cracking adopted in this project matches the pattern that is observed in the progressive development of many kinds of distress. There are three curves that represent the extent and severity of the reflection cracks as they are observed in the field. The curves show the one growth of the high severity reflection cracks; the sum of the percentages of the high and medium severity cracks; and the sum of percentages of the high, medium, and low severity cracks. Each curve is plotted against the percent of the original length of transverse cracks in the old pavement surface. The difference between the curves represents the percentages of the individual levels of distress severity. This S-shaped curve is defined by two parameters: ρ, the scale parameter, and β, the shape param- eter. The scale parameter is the number of days required for the percentage of reflected cracks to reach 36.8 percent (i.e., 1/e), of the original length of the transverse cracks or joints in the old pavement surface. The shape parameter deter- mines how steep the growth of the curve is as it reaches the 36.8 percent mark. This method allows a simple, consistent, and comprehensive description of the distress history of an overlay. It also makes the task of calibrating the calculated reflection cracking lives, due to traffic and thermal stresses to the field observations, possible. Calibration of Calculated Overlay Life to the Observed Distress The two-step method used to develop calibration coeffi- cients first involved linear regression analysis and observation of the patterns of the predicted versus the observed values of both ρ and β. If the coefficient of determination (R2) was acceptable and the scatter of the data was clustered around the line of equality, the calibration coefficients were considered acceptable. Otherwise, a second step was undertaken if there was a non-random pattern of scatter around the line of equal- ity or R2 was unacceptable. This approach had to be taken because only about 150 of the sections were unique, the rest were similar in pavement features (structure, materials), traffic, and weather, such that

50 Figure 43. LMH Regression results of  and  for AC over AC pavement and Wet-No Freeze climate zone. Figure 44. MH Regression results of  and  for AC over AC pavement and Dry-No Freeze climate zone. there were not enough sections to separate the sections into two categories (calibration sections and validation sections). The approach taken raises concern about how well the cal- ibrated predictions match the field data. The next section dis- cusses that concern. Calibrated Results Compared with Observed Field Data The details of each set of calibration coefficients are pro- vided in Appendices L, M, and N. The scale and shape param- eters ρ and β that were fit to the field observations are provided in Appendix M, and the calibration coefficients and plots of the predicted versus the observed values of ρ and β are provided in Appendix N. Calibration Coefficients by Regression Analysis Figure 43 shows the predicted versus observed values of ρ and β for HMA overlays over a cracked asphalt surface in a Wet-No Freeze zone. Figure 44 shows the predicted versus the observed values of ρ and β for a HMA overlay placed over a cracked asphalt pavement surface in the Dry-No Freeze zone. The details of each set of model calibration coefficients are given in Appendix N. Figure 45 shows the predicted versus observed values ρ and β for HMA overlays over a cracked asphalt pavement surface in a Wet-Freeze Zone. Figure 46 shows the predicted versus the observed values of ρ and β for HMA overlays over jointed concrete pavements in a Wet-Freeze Zone. Figure 47 shows the predicted versus the observed values of ρ and β for HMA overlays with reinforcing interlayers over cracked asphalt pavement surfaces in a Dry-Freeze Zone. Figure 48 shows the predicted versus observed values of ρ and β for HMA overlays with reinforcing interlays over jointed concrete pavements in a Wet-No Freeze Zone. Some of the difference between the predicted and the observed distress in these figures is due to the difference in levels of construction quality control. Predictions of Overlay Reflection Cracking Eleven sets of calibration coefficients were developed, one set for each combination of pavement structure and climatic zone for which sufficient data were available. Each set of model calibration coefficients have a maximum of three pairs

Figure 45. MH Regression results of  and  for AC over AC pavement and Wet-Freeze climate zone. Figure 46. H Regression results of  and  for AC over JPC/JRC pavement and Wet-Freeze climate zone. Figure 47. LMH Regression results of  and  for AC with reinforcing over AC pavement and Dry-Freeze climate zone. Figure 48. LMH Regression results of  and  for AC with Reinforcing over PCC pavement and Wet-No Freeze climate zone. 51

52 of ρ and β values corresponding to the three levels of distress severity. In some cases, there were no observed high or medium severity distress levels. Thus data were available for 24 out of a total of 33 possible sets of model calibration coefficients. The 24 sets of calibration coefficients are tabulated and graphs of the observed versus the predicted values of ρ and β are presented in Appendix N. Figures 49 through 59 present 11 sample sets of calculated distress curves, one for each of the pavement structure and climatic zone combinations, as listed in Table 25. Figure 49 shows the predicted development of transverse reflection cracking (at the L-M-H, M-H, and H levels of sever- ity) for an HMA overlay over a cracked asphalt pavement sur- face in a Wet-Freeze Zone (Lincoln, Maine). The H level of severity begins to appear at around 100 days of service life. Figure 50 shows the predicted development of transverse reflection cracking and severity for an HMA overlay over a jointed reinforced concrete pavement in a Wet-Freeze Zone (Beaver, Pennsylvania). The High level of severity remains low for a long time before beginning its sharp rise. The difference between the rates of distress development shown in Figures 49 and 50 is due mainly to the difference in thermal stresses. Figure 51 shows the predicted development of transverse reflection cracking and severity of an HMA overlay over an open graded friction course which was used as a strain reliev- ing interlayer over a cracked asphalt pavement surface in a Wet-Freeze Zone (Frederick, Maryland). Figure 52 shows the predicted development of transverse reflection cracking and severity of an HMA overlay over a continuously reinforced concrete pavement surface in a Wet- Freeze Zone (Minnesota, Washington). The figure indicates no observed high severity reflection cracks. Figure 53 shows the predicted development of transverse reflection cracking and severity of an HMA overlay reinforced Figure Number Overlaid Pavement Type Climate Zone Distress Severity Levels 49 Asphalt WF L, M. H 50 Jointed Reinforced Concrete WF L, M, H 51 Friction Course over Asphalt WF L, M, H 52 Continuously Reinforced Concrete WF L, M 53 Reinforcing Geosynthetic over Jointed Concrete W-NF L 54 Reinforcing Geosynthetic Over Asphalt DF L 55 Asphalt W-NF L, M 56 Friction Course Over Asphalt W-NF L 57 Asphalt DF L, M, H 58 Asphalt D-NF L, M, H 59 Reinforcing Geosynthetic Over Jointed Concrete WF L, M Table 25. Figures of calculated reflection cracking distress curves. 0 10 20 30 40 50 60 70 80 0 200 400 600 800 % T ot al L en gt h of C ra ck s No. of Days L+M+H M+H H 100/e Figure 49. Development of transverse reflection cracking distress extent and severity for HMA overlay over asphalt surface in wet-freeze zone (Lincoln, Maine). 0 10 20 30 40 50 60 70 80 90 0 1000 2000 3000 4000 5000 % T ot al L en gt h of C ra ck s No. of Days L+M+H M+H H 100/e Figure 50. Development of transverse reflection cracking distress extent and severity for HMA overlay over jointed reinforced concrete in wet-freeze zone (Beaver, Pennsylvania).

010 20 30 40 50 60 70 80 90 0 1000 2000 3000 4000 5000 % T ot al L en gt h of C ra ck s No. of Days L+M+H M+H H 100/e Figure 51. Development of transverse reflection cracking distress extent and severity for HMA overlay over friction course over asphalt surface in wet-freeze zone (Frederick, Maryland). 0 10 20 30 40 50 60 70 0 2000 4000 6000 8000 10000 % T ot al L en gt h of C ra ck s No. of Days L+M+H M+H H (Not calibrated) 100/e Figure 52. Development of transverse reflection cracking distress extent and severity for HMA overlay over continuously reinforced concrete pavement in wet-freeze zone (Minnesota, Washington). 0 10 20 30 40 50 60 70 80 90 0 1000 2000 3000 % T ot al L en gt h of C ra ck s No. of Days L+M+H M+H (Not calibrated) H (Not calibrated) 100/e Figure 53. Development of transverse reflection cracking distress extent and severity for HMA overlay with reinforcing geosynthetic over jointed concrete in wet-no freeze zone (Waco, Texas). 0 10 20 30 40 50 60 70 80 90 100 0 1000 2000 3000 % T ot al L en gt h of C ra ck s No. of Days L+M+H M+H (Not calibrated) H (Not calibrated) 100/e Figure 54. Development of transverse reflection cracking distress extent and severity for HMA overlay with reinforcing geosynthetic over asphalt surface in dry-freeze zone (Amarillo, Texas). 0 20 40 60 80 100 120 0 500 1000 1500 2000 % T ot al L en gt h of C ra ck s No. of Days L+M+H M+H H (Not calibrated) 100/e Figure 55. Development of transverse reflection cracking distress extent and severity for HMA overlay over asphalt surface in wet-no freeze zone (Pittsylvania, Virginia). with a geosynthetic material and placed over a jointed con- crete pavement in a Wet-No Freeze Zone (Waco, Texas). Because medium or high levels of severity were not observed during the monitoring period, only the low level of severity could be modeled. Figure 54 shows the predicted development of transverse reflection cracking and severity of an HMA overlay reinforced with a geosynthetic material and placed on a cracked asphalt pavement surface in a Dry-Freeze Zone (Amarillo, Texas). No medium or high level severity distress was observed on any of the test sections during the monitoring period. Figure 55 shows the predicted development of transverse reflection cracking and severity of an HMA overlay over a cracked asphalt pavement surface in a Wet-No Freeze Zone (Pittsylvania, Virginia). The low severity distress appeared after about 1800 days and medium level severity began to 53

54 appear after about six years; no high level severity distress was observed. Figure 56 shows the predicted development of transverse reflection cracking and severity of an HMA overlay over an open graded friction course which was used as a strain relieving interlayer over a cracked asphalt pavement surface in a Wet-No Freeze Zone (Yazoo, Mississippi). Only the low level severity of distress was observed. Figure 57 shows the predicted development of transverse reflection cracking and severity of an HMA overlay over a cracked asphalt pavement surface in a Dry-Freeze Zone (Deaf Smith County, Texas). The high, medium, and low levels of distress severity appeared within the first year. Figure 58 shows the predicted development of transverse reflection cracking and severity of an HMA overlay over a cracked asphalt pavement surface in a Dry-No Freeze Zone (Pinal, Arizona). The medium level severity of distress appeared and began its sharp rise after about 12 years. Figure 59 shows the predicted development of transverse reflection cracking and severity of an HMA overlay with geo- synthetic reinforcing over a jointed concrete pavement surface in a Wet-Freeze Zone (New York City). No high level severity distress was observed. Figures 49 through 59 illustrate the predictions for each of the sets of reflection cracking model calibration coefficients. Each of the four major climatic zones are represented, but not all of the pavement structure-overlay types are. Although 11 sets of calibration coefficients were developed, a total of 40 combinations are possible (four climatic zones and 10 pave- ment structure-overlay types). All of these additional sets of 0 20 40 60 80 100 120 0 1000 2000 3000 4000 % T ot al L en gt h of C ra ck s No. of Days L+M+H M+H (Not calibrated) H (Not calibrated) 100/e Figure 56. Development of transverse reflection cracking distress extent and severity for HMA overlay over friction course over asphalt surface in wet-no freeze zone (Yazoo, Mississippi). 0 20 40 60 80 100 120 0 100 200 300 400 500 600 % T ot al L en gt h of C ra ck s No. of Days L+M+H M+H H 100/e Figure 57. Development of transverse reflection cracking extent and severity for HMA overlay over asphalt pavement surface in dry-freeze zone (Deaf Smith County, Texas). 0 10 20 30 40 50 60 70 80 90 100 0 2000 4000 6000 % T ot al L en gt h of C ra ck s No. of Days L+M+H M+H H 100/e Figure 58. Development of transverse reflection cracking extent and severity for HMA overlay over asphalt pavement surface in dry-no freeze zone (Pinal, Arizona). 0 10 20 30 40 50 60 70 80 90 0 1000 2000 3000 4000 % T ot al L en gt h of C ra ck s No. of Days L+M+H M+H H (Not calibrated) 100/e Figure 59. Development of transverse reflection cracking distress extent and severity for HMA overlay with reinforcing geosynthetic over jointed concrete pavement in wet-freeze zone (New York, New York).

calibration coefficients can be entered into this design program and can be used in the design of HMA overlays to represent each unique combination of climatic zone and pavement structure-overlay type. The remaining sets of calibration co- efficients may be developed when a sufficient collection of field distress, pavement structure and materials, and traffic and weather data are assembled to repeat the process described in Chapter 2. In order to assist in such calibration efforts, a Cali- bration Program has been assembled and the User’s Guide is provided in Appendix P. It should be noted that if unrealistic input values are used, an unrealistic result can be expected from the predicted curves of reflection cracking extent and severity levels. Numerical tri- als in this project with the stress intensity factor and complex modulus ANN models have shown that they provide accurate extrapolations well beyond the values used to train them. These ANN algorithms are an integral part of both the calibration and design processes. However, they cannot be expected to produce realistic results if the input values are too far outside the range of their training values. Calibration of the Computational Model to Field Data The process of developing calibration coefficients is de- scribed in Chapter 2 and details of the process are contained in Appendices I through N. The Calibration Program, presented in Appendix P, generates five computed pavement reflection cracking lives that are required of each overlay in the original field data set. At this point, the calibration process is completed by using linear regression analysis to produce the calibration coefficients for the overlay. The form of the model for the time scale value ρ is seen in Equations 40 through 43 and the same form of equation is used to model the shape factor β. Equations 40 through 42 are the form used if the principal cracking mecha- nisms are assumed to be bending stresses up to Position I and thermal stresses are the principal mechanism from Position I up to the surface of the overlay. Equation 43 is an example of the form of the equation if bending stresses are assumed to be the principal cracking mechanism up to Position I and shear- ing stresses are assumed to be the principal cracking mecha- nism from Position I up to the surface of the overlay. All of these forms of equation are tried and calibration co- efficients are found for each form of equation using linear regression analysis. The form of equation that produces the highest R2 value and its corresponding set of calibration co- efficients are selected as the best set for that type of overlay. Experience has shown that bending stresses are the principal cracking mechanism up to Position I and thermal stresses are the principal cracking mechanism from Position I up to the surface of the overlay, except for overlays with reinforcing interlayers. From this overlay, bending is the principal mecha- nism up to Position I and shearing is the principal mechanism from there up to the surface of the overlay. The reinforcing interlayer reduced the influence of thermal stress on the rate of crack growth in this type of overlay. As a general rule, at least 10 overlay sections which are distinctively different from one another in their traffic, layer thicknesses, and overlay mix designs are needed to form a set of calibration coefficients. Validation of the developed sets of model coefficients involves the prediction of the distress of overlays that were not used in developing the original set of calibration coefficients. Such validation is desirable if sufficient additional overlay sections are available for this purpose. The shape of the extent-severity-life curves was chosen to limit the percentage of cracking to be no more than 100 per- cent. The rate of rise of these curves is also constrained by the magnitude of the shape factor β. Generally, the scale factor ρ will be larger for the higher level of severity, although occasion- ally the H and the M+H curves will cross because of inverted values of β, as observed in Figures 49, 50, and 55. The logical structure in the design program does not prevent two severity curves from intersecting and this will occur on occasion. 55