U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590

Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

This report is an archived publication and may contain dated technical, contact, and link information
Publication Number: FHWA-HRT-06-121
Date: November 2006

Long-Term Pavement Performance (LTPP) Data Analysis Support: National Pooled Fund Study Tpf-5(013)

Appendix A. Literature Review

This appendix details relevant literature reviewed for this study to gain an understanding of pavement response in various freezing conditions. Although the available literature directly related to the study was relatively limited, it provided insight that was used to formulate the activities conducted. The majority of the literature regarding frost effects dealt with quantifying the change in material properties and performance characterization on particular projects. In addition, some limited information was found on using LTPP data to model pavement performance in frost areas.


In this literature example, Zhang and MacDonald(31) reported frost studies using the Danish road testing machine in their paper "The Effects of Freeze-Thaw Periods on a Test Pavement in the Danish Road Testing Machine" presented at the Ninth International Conference on Asphalt Pavements, International Society for Asphalt.

Tests performed in the testing machine were subjected to freeze-thaw cycles to study the effects of frost on pavement performance. Initially, Road Test Machine 2 (RTM2) was loaded with more than 150,000 repetitions of a 60,000 newton (N) (13,488 lbf) wheel load. A freeze-thaw cycle was then simulated and the pavement structure was loaded with 1,800 additional repetitions during the thawing period. This process was then repeated; however, the number of load repetitions during thawing was modified to 3,000. Rut levelup and overlay layers were then placed on the RTM2 pavement structure. The resulting structure was labeled as RTM3. It was subjected to one freeze-thaw cycle and loaded with 2,800 repetitions during the thawing period, followed by 50,000 repetitions after the completion of the thawing process. Soil suction, material response, profile, and temperature data were collected throughout the testing periods. The profile data were obtained using a profiler specially built for profile measurements in t e RTM. To simulate freezing conditions, the chamber was maintained at -15 °Celsius (C) (5 °Fahrenheit (F)). During thawing, the temperature of the chamber was gradually raised to 10 °C (50 °F) and then to 25 °C (77 °F). The water table level in the RTM was maintained at 0.8 m (2.62 feet (ft)) below the level of the pavement surface.

As a result of the testing, a new subgrade permanent strain model was developed as well as roughness models in terms of slope variance, international roughness index (IRI), and rut depth. Equation 11was developed for permanent strain in the subgrade.

View Alternate Text(11)
ePZ = plastic strain at depth Z in microstrain
N = number of load repetitions
sZ = vertical stress at depth Z
P = reference stress (atmoshperic pressure 0.1 MegaPascha l)
eZ = vertical resilient( elastic) strain at depth Z in microstrain
A,a,ß = constants

Plastic strain in the subgrade material was determined to increase 60 to 75 percent during the thaw-loading period, while the transient resilient response increased 40 percent. The increase in strain was attributed to movement, reorientation, and resettlement of the soil particle displaced by previous frost heave. This study provided good numerical information on the change in material properties from frost effects.


Doré, Konrad, and Roy(32) presented a paper, "A Deterioration Model for Pavements in Frost Conditions," analyzing thermal cracking. Using data from test sections in Quebec, Finland, and Minnesota (the Minnesota Road Research Project (Mn/ROAD) as well as two LTPP SMP sites, Doré, Konrad, and Roy developed two empirical models. The first model provides curves for delimiting different levels of cracking extents based on the solicitation index. The second model estimates the equivalent annual evolution of cracking, which is defined as the total cracking developed over 1 year divided by the age (in years) of the pavement.

A sensitivity analysis was performed by altering all contributing factors used in the model to predict the life span of the pavement. The results showed that the frost depth and width of snow removal contribute the most to pavement performance. The model is based on a number of assumptions, and it is valid only for pavements with substantial frost penetration. In addition, the data used in generating the model came from a small set of locations. Construction variability was not taken into account and would contribute significantly to crack development in areas of local weak spots.


Doré and Savard(33) reported additional studies of thermal cracking which in their paper "Analysis of Seasonal Pavement Deterioration." Two test sites in Quebec, Canada, were monitored for 3 years to evaluate seasonal accumulation rates of distress. One site was constructed in 1994 specifically for this project. It consisted of five test sections (three of which were insulated). The other site was part of C-SHRP and consisted of four noninsulated test sections. Each of these sections had the same structural design and differed only in the grade of asphalt used. Noninsulated sections were used to compare the accumulation of different distress types in each season. Similar analysis was performed on the insulated sections, thus evaluating the effectiveness of insulation in preventing frost-related distress.

The noninsulated sections were found to experience 65 percent of transverse cracking damage in the winter followed by 25.5 percent in the spring. Transverse cracking is caused by the accumulation of tensile stress in the pavement due to thermal shrinkage. The relatively large amount of damage incurred during the spring is likely to be the propagation of existing transverse cracks originally developed in the winter. Similarly, 55 percent of longitudinal cracking damage occurred in the winter compared with 23 percent in the spring and 22 percent in the summer. The research team believed that differential heaving during winter was the main mechanism in longitudinal cracking; however, the data does not support this belief. Other factors may also contribute to the longitudinal cracking. Approximately one-half (49 percent) of the fatigue damage was accumulated during the winter, while 42 percent occurred in the spring. The large fatigue damage in the winter is believed to be due to partial thawing of the pave ent base that occurred during the winter season. The ratio of winter-to-fall roughness was approximately 1.7 to 1. A residual roughness was found to remain after the spring thaw. The increase in roughness is believed to be caused by frost heave. Based on the data collected, seasonal effects are not significant in terms of rutting damage. Last, the comparison of distress accumulated on insulated and noninsulated sections reveale that insulation is effective in reducing overall distress by 40 percent.


More specific to this project was a report by Senn, et al.(34) on "Development of Performance Prediction Models for Dry No Freeze and Dry Freeze Zones Using LTPP Data." Using data from approximately 50 GPS sites in the LTPP Western Region, performance trends were developed based on roughness, fatigue cracking, transverse cracking, and rut depth. For purposes of analysis, the sites were categorized by climatic region and structural number. Only two climates were considered in this study (dry freeze and dry no-freeze). Similarly, two structural number levels were used to group the sites using a structural number of four as the breakpoint between the two levels. This was done to distinguish between the relatively thick and thin pavement sections. Regression analysis was performed using ESALs as the independent variable for all parameters with the exception of transverse cracking, which incorporated pavement age as the independent variable.

From the analysis, fatigue cracking and transverse cracking both increased with traffic or time:

Although the preceding conclusions could be drawn from the data, the correlations were very poor, and many factors made the analysis difficult. In some cases, monitoring terminated before the failures in the pavement were exhibited. Without data on pavement structures near the end of their service life, developing a prediction model is very difficult. Adding to the difficulties was the variability in the field measurements. In particular, the difference between longitudinal cracking in the wheelpath and fatigue cracking is very subjective, resulting in drastically different amounts of fatigue cracking on the same section. Last, there is a large difference in the traffic levels experienced in the Western Region’s dry freeze and dry no-freeze zones.


A study(35) was undertaken to evaluate the spring thaw period on a forest highway in northwestern Montana. By developing a methodology to accurately and easily determine the thaw conditions of the highway at various locations, responsible entities could place load restrictions at optimum times throughout the year. Air and subsurface soil temperature thermistors were installed at 17 sites equating to a density of one site every 5.63 km (3.5 mi). The locations of the instrumentation were selected based on divisions of the highway that could be easily closed without affecting the remaining segments. Data were collected throughout the year with increased frequency during the spring season.

The research attempted to predict the thaw condition based solely on soil temperature. Because of the presence of dissolved minerals in soil moisture, the soil will freeze at temperatures below 0 °C (32 °F). Therefore, laboratory testing was performed on eight soil samples typical of the region to determine the actually freezing temperature. An average value of -0.17 °C (31.7 °F) was determined from this testing and used in the analysis. In addition, pavement strength was measured using Benkelman beam deflection testing.

Through the combination of temperature and strength testing the following results were obtained. The thawing period did not begin until a temperature of -0.17 °C (31.7 °F) was achieved at the base of the asphalt. It is interesting to note that this condition was not met until the average soil temperature rose to approximately 1.11 °C (30 °F). Pavement strength data were used to determine the pavement strain, and thus, damage factors throughout the year. The end of the critical thaw weakening period was determined by identifying the time when the damage factor returned to unity, indicating the pavement strength returned to prefreeze levels. From this analysis, a thaw depth of approximately 1.22 m (4 ft) was indicated as the end of the critical thaw period. These results are based on limited data from a very specific region of Montana.


The Modified Berggren equation (MBE) was used to estimate the maximum frost penetration experienced at 40 Mn/ROAD test sections over three winters.(36) These predictions were compared with the measured maximum frost penetration values obtained from electrical resistivity probes installed at the site. Further, sensitivity analysis was conducted to study the contribution of many factors on estimated frost depth. The factors considered in the analysis were pavement material properties, moisture content, density, layer thickness, and mean soil temperature, as well as factors dealing with thermal properties of materials.

The results from the comparison were inconsistent between the three winters. For the 1993–1994 winter, the difference between calculated and measured frost depth was within ±10 percent for most of the sections. Generally, the calculated values underestimated the actual frost depths measured. Similar results were obtained for the 1995–1996 winter. Conversely, only approximately 50 percent of the data from the 1994–1995 winter were within ±20 percent. The rest of the data exhibited greater differences. Using the MBE resulted in an overestimation of frost depth in the majority of the test sections for that winter. For all three winters, test sections with granular subgrade experienced measured frost depths far less than the estimated values. In this case, it is believed that errors in the frost penetration measurement caused the discrepancy.

Several conclusions were drawn from the sensitivity analysis. The effect of small variations in layer thickness on estimated frost depth can be considered negligible. Sizeable deviation in moisture content will result in a change in calculated frost depth of less than 10 percent. One of the variables in the MBE is the n-factor, which is used to convert air freezing index to surface freezing index. The most appropriate n-factors were found to be 0.90 for flexible pavements and 0.95 for rigid pavements based on the conditions at Mn/ROAD. Thermal conductivity values were estimated using equations developed by Kersten. Better results of frost penetration were observed when the thermal conductivity estimates were increased by 25 percent. Last, a reduction of the estimated mean annual soil temperature from 11.1 °C to 9.4 °C (51.98 to 48.92 °F) resulted in better agreement of frost penetration data. If the changes in n-factors, thermal conductivity, and mean annual soil temperature are implement d, the majority of the calculated depths fall within ±13.3 (excluding granular subgrade data).


In an effort to evaluate the effect of thaw weakening on pavement structures, eight test sections were monitored in 1993, 1995, and multiple times in 1996.(37) FWD, distress, rut, roughness, and traffic data were collected and used in the analysis. To monitor thaw propagation, hourly subsurface temperatures were collected to a depth of 1.93 m (6.33 ft) below the surface. Regression analysis was performed on the temperature data to develop thaw propagation models, which were used to determine the limits of the thaw period in the spring of 1996. For this study, the end of the thaw period was set as the time at which the thaw depth reached 1 m (3.28 ft).

Results from distress data revealed that rutting and water bleeding were the most common forms of distress accumulated during the thaw period. Rutting occurred very rapidly, and it first occurred when the thaw depth was at approximately 0.3 m (0.98 ft). The southern lane, which carries 70 percent less EALs, was found to exhibit more damage than the northern lane. Dynamic effects of lighter weight traffic are believed to be the cause of this phenomenon. Rut depth and roughness were both found to be larger in the early spring than in the late spring. The improvement in remaining life based on fatigue damage obtained when load restrictions were enforced was determined using FWD.


A study(38) was undertaken to evaluate the factors contributing to both good and poor performance of PCC pavements. To conduct this analysis, definitions of good and poor performance needed to be determined in terms of specific distress types. This was achieved through the collaboration of a panel of experts who set limits on performance considering roughness, faulting, cracking, and localized failures such as a function of pavement age. Data from the LTPP database were the sole source of information used for this study.

Based on the analysis, jointed plain concrete pavements (JPCP) were found to exhibit increased roughness values when subjected to multiple freeze-thaw cycles. In addition, sections in colder climates were rougher, as were sections in wetter climates. Seventy-one percent of the poor performing JPCP (in terms of roughness) sections had fine-grained subgrade. Similarly, all of the poor performing JRCP sections (roughness) were on finegrained subgrade. Increased subdrainage was also found to decrease roughness in three surface types: JPCP, JRCP, and continuously reinforced concrete pavements.

Faulting of nondoweled JPCP was found to be higher in wet climates. This trend did not hold true for doweled JPCP or JRCP. Fine-grained soils contributed to increased faulting of both JPCP and JRCP. Adequate subdrainage resulted in lower amounts of faulting for both types of jointed concrete but particularly for nondoweled JPCP.

Transverse cracking of JPCP was found to occur more often in the western part of North America. It is believed that the increase in solar radiation in the west contributes to a larger thermal gradient through the slab leading to increased warping and curling.

Cold and wet climates increased the amount of localized failures experienced in continually reinforced concrete pavement (CRCP); However, more information is required to fully understand all factors contributing to these failures.


In this study(39) data obtained from LTPP SMP sites were used to determine the presence and extent of frost penetration. Three types of electrical resistance measurements (2-point resistance, 4-point resistivity, and voltage drop) were used in conjunction with subsurface temperature profile data to estimate frost penetration. A computer program (FROST) was developed to assist in evaluating the temperature and electrical resistance data.

Overall, temperature profile data produced reasonable and expected trends. Based on the concept of latent heat of fusion, two conditions were defined to indicate a phase change in subsurface material. The temperature must be less than or equal to 0 °C (32 °F), and it must remain constant for 2 days. Reasonable frost predictions were obtained using these criteria and the temperature data; however, some limitations exist using the established method. The 2-day minimum time period for constant temperature prevents short freezethaw cycles from being captured. On the other hand, temperature data from lower in the pavement structure usually exhibit less than 1 °C (33.8 °F) daily variation throughout the year. Therefore, the 2-day time period erroneously identifies phase changes in some cases. Because of these limitations, temperature data alone cannot be used in predicting phase changes.

There is a large increase in resistivity when frozen conditions exist, thereby allowing for the identification of phase changes; however, many factors influence the bulk resistivity of soils: the type of soil, moisture content, dissolved salt concentration, and temperature. Identifying an increase in resistance due to frozen soil is very difficult. When comparing the three resistance measurement techniques it was discovered that agreement between all three methods occurred 60 percent of the time. Coupling this with the fact that each method had unique advantages and disadvantages, the use of all three techniques simultaneously would produce the most reliable results. Concurrently using all resistance data still produced large nonwinter, seasonal, and diurnal variability that could not predict phase changes with accuracy.

As a result of the analyses of temperature and electrical resistance data, algorithms were developed to incorporate data from all three resistance measurement techniques, temperature data and time of year to identify the extent of frost penetration. These algorithms were packaged into the FROST program. The FROST program will normalize all three resistance measurements. The user must then select an appropriate threshold based on these normalized values. All data points below the threshold are considered to be in the nonfreeze state. Data points above the threshold are analyzed further based on the average temperature. If the temperature is greater than zero, the data point is considered to be in the nonfreeze state. Conversely, temperatures below zero yield a data point classified in the freeze state. Graphs are then produced that visually depict the extent of frost throughout the monitoring period.

The output from FROST was compare with historical data and found to be in reasonable agreement. Time domain reflectometry (TDR) data, also collected at LTPP SMP sites, was compared with the FROST output. TDR data provide information about the moisture content of soil; however, this information does not reflect frozen water. Therefore, a sudden drop in moisture content recorded by the TDR is expected during freeze periods. In almost all cases, prediction of frost by the program corresponded to a drop in moisture content recorded by TDR. The overall reliability of the freeze-state determination is 93 percent. These results have been added to the LTPP IMS and can be found in two tables: SMP_FREEZE_STATE and SMP_FROST_PENETRATION.


An empirical rutting progression model was developed using data from the AASHO Road Test.(40) One of the major differences between this and other models is the inclusion of a thawing index to capture the effects of the environment on rutting. Considerable accumulation of permanent deformation was experienced during spring thaw periods; therefore, including a thawing index seems logical. Furthermore, it is well accepted that the strength of unbound layers reduces considerably with excess moisture. The largest increase in moisture content of subsurface material is experienced during thaw periods.

To determine a thawing index, some measure of freezing is necessary. An accumulated freeze index was utilized to quantify freezing and was based on the mean minimum temperature of each two week period. This index was combined with the mean maximum temperature over each two week period to determine the thawing index. The thawing index was then incorporated in the permanent deformation model by increasing the rate of rutting as a function of thawing index.

Overall, the model was found to estimate rut depth with a standard error of regression of 3.3 mm (0.13 inch). The predicted rut depth values compare very well to the observed rut depth measurements for sections used in the model development. In addition, actual rutting measurements from other sections, not used in the model development, were compared with predicted values and found to be in agreement, further confirming the model. Note that this model is specific to the conditions of the AASHO road test.


A study(41) was undertaken to develop a new pavement performance equation using duration models based on data from the AASHO road test. Duration models incorporate the variable nature of pavement failure as well as accounting for censoring and truncation biases, which are introduced into models when failure events are not observed due to limited duration data collection. In some instances, pavement failure is reached before the monitoring period begins, while other failures occur after the completion of the monitoring period. Although these data points were not observed, they need to be accounted for in model development. An extension of the Weibull model can be used to eliminate this bias.

From the Weibull model, the rate at which pavement failure will occur after a given time can be estimated using the hazard rate function. In addition, the survival function can be used to estimate the probability that a pavement will last longer than a given time period. In turn, the probability distribution can be determined using the previous two functions. Of the three models discussed above, only the hazard rate function was used in this study.

For comparison, a model using the hazard function with the same variables as the original AASHO equation was developed to predict pavement life. Results from this duration model were compared with the predicted values of the original model using data from the AASHO road test. Overall, the life prediction from the new duration model estimated the life of the test section with more accuracy than the original model. The standard error of the new model was 0.42 compared with 0.65 from the original model. However, the duration model overestimated the life of pavement structures that failed relatively early in the testing period. Conversely, the model underestimated the pavement life of pavement structures failing relatively late. By the end of the testing phase, 237 test sections reached failure. The new equation predicted 253 failures with an error of 6.8 percent, while the original equation predicted 215, which equates to an error of -9.3 percent.


Test sections in the Frost Effects Research Facility at the Cold Regions Research and Engineering Laboratory were loaded with a 133-kN wheel load during a simulated thawing period.(42) Using in situ measurement equipment, base and subgrade responses such as stress, strain, resilient modulus, and permanent deformation were monitored before and throughout one freeze-thaw cycle to quantify the changes induced by this environmental process. In addition, the timing of these changes was of interest to researchers to define critical periods in the cycle. Results from this research were used to evaluate the validity of the U.S. Army Corps of Engineers Modulus Reduction Factors for Frost-Susceptible Soils. This table is used for design to adjust modulus values obtained during frost-free conditions to reflect values experienced during thawing periods.

From this testing, the base was found to experience a 67 percent maximum reduction in resilient modulus during the thaw-weakening period, which was approximately 2 weeks. Similarly, the subgrade experienced a reduction of 56 percent over a 3-week period. The vertical strain in the base reached a maximum level 15 days into the thawing phase, which was approximately 530 percent of the prefreeze value. The strain never fully recovered, remaining at 160 percent on completion of the cycle. The subgrade behaved in the same manner, reaching a maximum strain 1,100 percent higher than the prefreeze strains 15 days into the thaw process and recovering to only 241 percent. The stress values followed similar patterns. The minimum stress did occur at the same time as the maximum strain in the subgrade but not in the base layer. Permanent deformation was found to increase rapidly within the first 10 days. Based on these findings, the reduction factors used in the design process significantly overestimate the stiffnes of both the base and subgrade material during the thawing period.


The cyclic increase of pavement roughness in roadways subjected to frost penetration prompted a study of the change in IRI on a section of the Doto Expressway in east Hokkaido, Japan.(16) The section of road was divided into 31 consecutive sections each with a length of 1 km (0.62 mi). The IRI values were obtained from longitudinal profile data collected using an inertial profiler. Profiles were collected once in August 1999 and once in November 1999 to establish prefrost values. Between February and April 2000, longitudinal profile data were collected once a week to monitor the change in IRI during the propagation of frost penetration and subsequent thaw period. Weather data were also gathered from a weather station adjacent to the test sections. Using a modified Berggren’s formula with this data and knowledge of the subsurface material, the depth of frost penetration was estimated to reach a maximum in late March 2000. In addition, frost penetration was estimated to reach the subgrade layer on February 11, 2000.

The IRI in the summer and fall were found to be similar and considerably less than the winter IRI values. The winter IRI values increased as the temperature decreased, and they reached a maximum value in early to mid-March just before the maximum depth of frost penetration was achieved. The spring IRI values appeared to have returned the level of the prefreeze values of the summer and fall. For further analysis, the test sections were divided into 100-m (328.1-ft) segments and categorized into following three groups based on the nature of material under the pavement structure: cut, embankment, and bridge. The average IRI values in each category were compared and the cut sections were found to exhibit the greatest increase in roughness during the frost penetration. Additional evaluation of the six sections demonstrating the largest increase in IRI established a linear relationship between IRI and freezing index.


Multiple models have been developed to forecast the propagation of frost and resultant heave of roadways both in Quebec and France. In order to validate these models, in this study(43) four test sections were constructed and monitored for 3 years. The test sections were equipped with temperature sensors, TDRs, frost tubes, piezometers, and heaving sensors to monitor frost depth as well as the amount of heaving in the subsurface layers of the pavement structure. Weather data were obtained from a weather station in the nearby vicinity. Two pavement structures commonly constructed in France were selected for the monitoring sections. Two test sections conforming to each typical pavement structure were constructed; however, only one test section of each structure was insulated with extruded polystyrene. This allowed for the evaluation of the effect of frost on pavement performance while keeping all other variables (i.e., traffic, natural subgrade, climate, and pavement structure) constant. Furthermore, the efficiency of the insulation could also be assessed.

Frost depth and soil heave estimates from the SSR, GEL1D, and CESAR-GELS models were compared with actual values obtained for the onsite monitoring equipment. The SSR model is based on thermal equilibrium at the frost front, and it estimates soil heave using water migration as a function of thermal gradient. The GEL1D and CESAR-GELS models are similar; they use finite element analysis as the foundation for frost-depth estimation. These two models do not incorporate heave estimates into their frost-depth calculations.

All three models were found to estimate frost penetration within 10 percent of the values recorded onsite. Because the GEL1D and CESAR-GELS models are dependent on initial conditions and do not incorporate frost heave, substantial differences between the models were observed. The SSR model was found to predict frost heave within the same order of magnitude with the maximum difference between predicted and measured values found to be 7 mm (0.276 inch). The ability of the insulation layer to protect subsurface materials from frost penetration was confirmed.


Survival analysis has been performed four times over the past 15 years on the majority of the freeway system in Illinois. This analysis(44) provides the probability of failure as a function of age or cumulative ESALs. Original pavement structures were categorized by pavement type and thickness. JRCP and CRCP were further separated with the presence of durability cracking. All pavement types were also classified by geographical location. Receiving rehabilitation treatment was defined as the failure criteria. Due to the length of the study, some pavement structures experienced failure multiple times, with some sections receiving three overlays; therefore, overlays were also included in the analysis.

Durability cracking greatly reduced the life expectancy of both JRCP and CRCP sections in the southern region of Illinois. JRCP sections without durability cracking are expected to carry 30 percent more ESALs than sections with durability cracking before reaching the 50th percentile life expectancy. Similar results were found in CRCP sections with a reduction in cumulative ESALs ranging between 32 and 63 percent. Thick asphalt cement overlays placed over both JRCP and CRCP experienced a larger reduction in life expectancy due to durability cracking compared with thin overlays.

One major downfall of this study was the failure criteria selected. Due to budgetary and other issues, rehabilitation activities are not performed on roads with consistent conditions. One road may receive treatment 1 year after the pavement has reached a poor condition, while other roads may be overlaid within 1 month of deteriorating to a similar condition. Further, some sections receive rehabilitation before failure occurs, introducing censoring bias.


Using data available from the LTPP database, a study was conducted to examine the effects of climate and subgrade on pavements subjected to limited loading.(45) All SPS-8 projects, monitored under LTPP, were considered in this analysis. Each site was categorized based on climate (precipitation and temperature), subgrade type, pavement structure, and age. Distress and roughness data were the only parameters evaluated in the study.

The research effort revealed that sections constructed on active subgrade soil in wetfreeze climates exhibited significantly more nonwheelpath longitudinal cracking. Statistical analysis of pavement roughness showed that subgrade type was the most influential factor for flexible pavements, while precipitation was most important in rigid pavements. It should be noted that these statements were not statistically significant at the 95 percent confidence interval. Sections constructed on active subgrade (in any climate) had the highest average IRI.


LTPP data from the SPS-1, SPS-2, and SPS-8 experiments were used to study the relative influence of structural design factors as well as site conditions on pavement performance.(46) The SPS-1 projects were used to investigate HMA layer thickness, base type, base thickness, and drainage. The SPS-2 experiments were evaluated to determine the effects of PCC slab thickness, flexural strength, base type, drainage, and slab width. The SPS-8 experiments were used to look into environmental effects without the contribution of heavy traffic. Performance comparisons were made for various surface distress types as well as roughness, rutting (HMA), and transverse joint faulting (PCC). The results from this study can be used to improve and implement design procedures that make better use of design options. Also, the contribution of environment and site conditions can be evaluated. Following are a few of the findings from the final report:

Previous | Contents | Next

Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000
Turner-Fairbank Highway Research Center | 6300 Georgetown Pike | McLean, VA | 22101