Impact of Design Features on Pavement Response and Performance in Rehabilitated Flexible and Rigid Pavements

Chapter 8. Study Findings, Conclusions, and Recommendations

This chapter provides the findings and conclusions from the analysis of preventive maintenance treatments and performance of different pavement rehabilitation alternatives. Recommendations for future research are provided at the end of the chapter.

Preventive Maintenance Treatments

The findings presented in this section are based on the analysis of 81 SPS-3 flexible pavement sites and 34 SPS-4 rigid pavement sections subjected to different preventive maintenance treatments. Most of the flexible pavement sites were monitored for at least 4 years, and about 22 percent of the sites were monitored for 10 years or more. Most of the rigid pavement sites were monitored for at least 4 years.

Preventive Maintenance Effectiveness for Flexible Pavements

From the analysis of SPS-3 sites, the following effects of preventive maintenance treatments on pavement performance were observed:

IRI:

• Of all of the SPS-3 treatments, only thin overlay was effective in mitigating and delaying the progression of roughness; however, this treatment was effective only for pavements in freeze zones, high traffic, or poor condition.

Rutting:

• Thin overlay mitigated and slowed the progression of rutting under all circumstances.
• Chip seal was more effective than slurry seal in wet freeze zones but was only marginally more effective in dry freeze zones.
• There were no significant differences between slurry seal, crack seal, and the no treatment control scenario with respect to rutting.

Fatigue cracking:

• Thin overlays and chips seals were more effective than slurry seal and crack seal treatments in mitigating fatigue cracking.
• With respect to fatigue cracking, thin overlays performed better than most other treatments if the pavement was in a freeze zone, in a wet climatic region, subject to high traffic, or initially in poor condition.

Preventive Maintenance Effectiveness for Rigid Pavements

The study of the SPS-4 sites showed that the performance of the joint/crack sealed sections and undersealed sections was not significantly different from the performance of control sections. Additionally, no meaningful difference between the two treatments was found. The analysis was weakened by the small number of sites and only 4 years of performance history that included recorded surveys with undersealing treatment. While 34 sites included the survey measurements for joint/crack sealed sections, only 10 sites had data for undersealed sections.

Rehabilitated Flexible Pavements

The findings presented in this section are based on the analysis of 18 SPS-5 rehabilitated flexible pavement experimental sites with 162 core test sections. Most of the sections were monitored for at least 9 years.

Evaluation of Rehabilitation Strategies with Respect to Performance

To analyze data from the SPS-5 experiment, a gradual statistical analysis was used  in which the data from each site were analyzed first, followed by a consolidated analysis of all sites simultaneously in search for general trends and broader conclusions about pavement performance and its dependency on design features and site conditions. The results obtained in the consolidated analysis mostly agree with the results found in the individual site analyses. A summary of the analysis findings with respect to major pavement performance indicators is provided below.

IRI:

• Rehabilitation strategies with milling prior to overlay provided better roughness performance (i.e., smoother) for all site conditions.
• Strategies with thick overlays provided smoother pavements for all site conditions.
• Strategies with virgin or RAP mixes had equivalent performance when used under wet conditions.
• Strategies with RAP mixes provided smoother pavements when used in dry conditions.
• Traffic level and freeze conditions did not impact roughness performance ranking.

Rutting:

• Rehabilitation strategies with thin overlays performed better than thick overlays in the short term. In the long term, the ranking of best strategies was more evenly distributed for both thick and thin overlays.
• The ranking of best strategies was evenly distributed among the two mix types (virgin and RAP) in the short term. In the long term, rehabilitation strategies with virgin mixes were in the top ranking of performance more frequently (lowest rutting), with the exception of fair pavement surface under freeze conditions, which corresponded to 33 percent of all sites.
• Strategies with milling did not improve rutting performance more than alternatives without milling.
• Surprisingly, the level of traffic did not affect rutting performance of the selected rehabilitation strategies.

Fatigue cracking:

• Short-term fatigue cracking performance was not affected significantly by any design features under any site conditions, which makes sense because overlays were designed to minimize fatigue cracking in the short term.
• Rehabilitation strategies with thick overlays provided better performance for fatigue cracking under all site conditions evaluated.
• Strategies with milling prior to overlay performed better to mitigate development and propagation of fatigue cracking in all site conditions.
• In regions with a dry climate, alternatives without milling performed as well as solutions with milling.
• Strategies with RAP mixes were better ranked for sites with low traffic.

Transverse cracking:

• There were no differences identified in short-term performance among the rehabilitation strategies in freezing zones.
• Among the sites located in no-freeze zones, the remaining site conditions did not have any impact on short-term performance.
• Strategies with virgin mixes and thick overlays ranked best for long-term transverse cracking performance.
• Strategies with RAP mixes performed better than virgin mixes when the site had low traffic and when the site had high traffic and a dry climate.
• Milling prior to overlay did not improve performance more than alternatives without milling.

Longitudinal cracking (in wheel paths):

• Rehabilitation strategies with milling prior to overlay were consistently better for improving performance than alternatives without milling.
• Strategies with virgin mixes were consistently better than alternatives with RAP in sites located in wet climates.
• There was no difference in ranking between strategies with virgin and RAP mixes in dry conditions.
• Overlay thickness was not a significant factor affecting performance associated with longitudinal cracking.

In terms of the effect of design features or construction practices, the following conclusions were made:

Overlay thickness:

• Overlay thickness was the most influential design feature. Thick overlays consistently performed better than thin overlays, as expected.
• The impact of thickness on performance was more evident in the long term (more than 5 years) for most of the distresses. The exception was rutting, which had no evidence suggesting that either thin or thick overlays provided less rutting.

Milling:

• The analysis of milling prior to overlay suggested that replacing the distressed portion of the surface layer improved the performance for the majority of distresses commonly observed in flexible pavements.

RAP mixes:

• The majority of sites did not show significant differences in performance between sections overlaid with virgin and RAP mixes. However, when differences existed, they were mostly in favor of virgin mixes.

Evaluation of Rehabilitation Strategies with Respect to Structural Responses

For evaluation of structural responses, a maximum FWD deflection measured under the center of the load was used as a structural response indicator. The study concentrated on evaluating FWD maximum deflections against the average pavement performance during the service life of SPS-5 sites. As with the analysis of pavement performance presented above, a gradual statistical analysis was used beginning with the analysis of individual sites, followed by a consolidated analysis of all sites simultaneously. The results from the consolidated analysis supported the findings from the analysis of individual sites. A summary of the analysis findings with respect to structural response is as follows:

• Rehabilitation strategies with thick overlays provided the lowest structural response independent of site conditions.
• Strategies with RAP mix overlays had the lowest structural response in freeze regions, while strategies with virgin mixes presented lower deflections under
  no-freeze conditions.
• Milling prior to overlay did not affect the structural response. In fact, in no-freeze zones, strategies without milling presented lower deflections.
• When comparing wet and dry climates, pavement surface condition, and traffic level, none had a significant impact on structural responses associated with each rehabilitation alternative.

In terms of the effect of design features or construction practices, the following conclusions were made:

Overlay thickness:

• Rehabilitation strategies with thick overlays had lower maximum deflection values compared to alternatives with thin overlays, as expected.

RAP mixes:

• There were no differences in pavement response between strategies using virgin and RAP mix overlays.

Milling:

• Strategies with milling prior to overlay did not impact the structural response more than alternatives without milling.

Evaluation of Structural Responses Immediately After Rehabilitation and Future Performance

The objective of this evaluation was to identify trends in the relationship between response measured after the rehabilitation and the observed performance in subsequent years of the pavement's service life. Only long-term performance was used for this analysis, which included performance data of 5 years or more. The following summarizes the analysis findings:

IRI:

• The trend between roughness and maximum deflection suggests that roughness, as measured by IRI values, was positively related to the deflection values measured after the rehabilitation of the pavement structure. The higher the deflection after rehabilitation, the higher the IRI over the long term.

Rutting:

• The center load FWD deflection after the pavement's rehabilitation did not provide adequate qualitative information about the rutting performance predictions. The instantaneous FWD deflections could not be associated with the material's behavior impacting performance for rutting.

Fatigue cracking:

• No statistically significant trend was found between fatigue cracking performance and center load FWD deflection.

Transverse cracking:

• Higher values of transverse cracking were found when the pavement had higher center load FWD deflections.

Longitudinal cracking:

• No significant correlation was found between longitudinal cracking and center load deflection.

Rehabilitated Rigid Pavements

Findings presented in this section are based on the analysis of 14 SPS-6 rehabilitated rigid pavement sites, specifically 8 JPCP and 6 JRCP. Most of the sections were monitored for at least 6 years. The findings from the analysis are described separately for JPCP and JRCP sites.

Evaluation of JPCP Rehabilitation Strategies with Respect to Performance

The results from the statistical analysis led to the following conclusions with respect to major pavement performance indicators, total cracking and IRI:

Total cracking:

• Rehabilitation strategies without overlays were the most effective to mitigate cracking development and propagation. Specifically, HMA overlays over jointed concrete pavements exhibited more total cracking than when the pavement was not overlaid.
• Saw and seal (when counted as an existing crack) showed more total cracking than other alternatives, but the control of reflection cracks (through sawing and sealing) was the design goal for this alternative. The smoothness of sawed and sealed overlays was similar to other overlays of similar thickness.
• Crack/break and seat of JPCP had no significant effect in reducing the amount of cracking, since it performed similarly to the 4-inch (102-mm) overlay over noncracked JPCP (with both minimum and maximum restorations).
• The three alternatives without overlays, no treatment control scenario, minimum restoration, and maximum restoration, were found to be the best choices (i.e., reduced total cracking) for both short-term and long-term performance.
• Crack/break and seat with 4-inch (102-mm) overlays was the best alternative among those which involved overlays to reduce cracking.
• The sawed and sealed joints did not deteriorate significantly on these sections, and they became an effective control of reflection cracking. If they were counted for total cracking, the sawed and sealed sections would have shown similar performance to other HMA overlays.

IRI:

• Rehabilitation strategies with HMA overlay were significantly smoother than treatments without overlay for both the short term and long term.
• The best alternative to improve roughness performance was the thicker overlay alternative crack/break and seat with 8-inch (203-mm) overlays. This alternative also has the highest cost.
• Crack/break and seat with 8-inch (203-mm) overlays and minimum restoration with 4-inch (102-mm) overlays (without crack/break and seat) were statistically equivalent and were found to be the best alternatives for most of the scenarios evaluated when both short-term and long-term roughness performance were considered.
• Crack/break and seat with 4-inch (102-mm) overlay was among the worst alternatives to improve roughness performance.
• Saw and seal alternative provided similar smoothness to other 4-inch (102-mm) overlays.

It should be noted that the best performance alternative may not be the lowest cost alternative. Selection of a rehabilitation alternative must also consider the cost and long-term maintenance.

The analysis of impact of site conditions led to the conclusion that different climate regions and surface conditions did not have a significant impact on roughness and total cracking performance for the rehabilitation strategies included in the SPS-6 JPCP experiment.

Effect of PCC Restoration Prior to Overlay

The impact of PCC restoration preoverlay treatments on performance of overlaid sections was investigated in the JPCP sites of the SPS-6 experiment. Transverse cracking was the only distress for which statistical differences were found between the four treatments. The conclusions from this study were as follows:

• The best alternative to limit the development and propagation of transverse cracking among all options with 4-inch (102-mm) overlays was crack/break and seat.
• Minimum and maximum restorations had an equivalent impact on short-term transverse cracking performance.

The analysis of impact of site conditions led to the following additional conclusions:

• Statistical differences in performance of overlaid sections were observed only for transverse cracking and short-term roughness when individual site characteristics were considered.
• The rankings of best treatments prior to overlay remained the same regardless of site characteristics (i.e., surface condition or climate region).
• There was no impact on performance due to variations in surface condition prior to rehabilitation or the climatic region where the LTPP site was located.
• Minimum, maximum, and saw and seal restorations provided the best short-term roughness performance, but there was no difference between the rehabilitation alternatives for long-term performance. Specifically, for the long term, these three restorations showed the same roughness.
• Crack/break and seat and minimum restoration were the best alternatives to mitigate the development and propagation of transverse cracking in the long term.

Effect of PCC Restoration Without Overlay

Three sections in each SPS-6 site did not receive overlays as part of their rehabilitation strategies. These sections were used to evaluate the impact of PCC restoration on performance. The small number of sections available for this study significantly reduced the power of the analysis and the chances of finding statistical differences among the treatment alternatives.
No statistical differences in performance were found for short-term performance. The only performance indicator that showed statistical differences between the treatments was long-term roughness. From the analysis of long-term roughness, the findings supported by the statistical analysis were as follows:

• The maximum restoration treatment produced the smoothest pavement over the long term.
• The minimum restoration treatment produced the second smoothest pavement over the long term.
• The no treatment control section, which received no rehabilitation, was the roughest pavement over the long term, as expected.

An attempt was made to evaluate the impact of site conditions on performance; however, the results were not statistically significant.

Evaluation of JRCP Rehabilitation Strategies with Respect to Performance

Similar to JPCP findings, the results of the JRCP analyses suggested that rehabilitation strategies with HMA overlays improved roughness performance, while strategies without overlays were better at improving total cracking development and propagation. The main conclusions were
as follows.

Total cracking:

• Rehabilitation strategies without overlays were the best when considering total cracking.
• Saw and seal, when counted as cracking, had the highest total cracking among all options evaluated. However, the sawed and sealed joints remained in reasonably good condition over time.
• Crack/break and seat of JRCP had no significant effect on reducing the amount of cracking since it performed similarly to the 4-inch (102-mm) overlay over noncracked JRCP (with minimum and maximum restoration).

IRI:

• Rehabilitation strategies with overlay had significantly better roughness performance (i.e., were smoother) than treatments without overlay.
• Minimum and maximum restorations with overlays were the best strategies to improve short-term performance for roughness.
• For long-term performance, the best alternative was the thick overlay alternative crack/break and seat with 8-inch (203-mm) overlays.

The sawed and sealed joints did not deteriorate significantly on these sections, and they became an effective control of reflection cracking. If they were removed from total cracking, the sawed and sealed sections would have shown similar performance to other HMA overlays.

Effect of PCC Restoration Prior to Overlay

The impact of PCC restoration treatments on the performance of overlaid sections was investigated in JRCP sites of the SPS-6 experiment. Transverse cracking was the only distress for which statistical differences were found between the four treatments. The conclusions from this study were as follows:

• Crack/break and seat was the best alternative for short-term performance.
• Minimum and maximum restorations had an equivalent impact on short-term performance.
• The best alternatives to limit the development and propagation of transverse cracking in the long term were crack/break and seat and minimum restoration.
• Saw and seal prior to overlay when counted as cracks had the highest total cracking among all options evaluated in the SPS-6 experiment.

The sawed and sealed joints did not deteriorate significantly on these sections, and they became an effective control of reflection cracking. If they were removed from total cracking, the saw and sealed sections would have shown similar performance to other HMA overlays.

Effect of PCC Restoration Without Overlay

Three sections in each SPS-6 site did not receive an overlay as part of their rehabilitation strategies. These sections were used to evaluate the impact of PCC restoration on performance. Distresses common to rigid pavements were used as performance measures. The only performance indicator that showed statistical differences between the treatments was short-term transverse reflection slab cracking. The findings from this statistical analysis were as follows:

• The maximum restoration treatment and the control section had statistically equivalent performances for short-term transverse slab cracking.
• The minimum restoration treatment provided the worst transverse slab cracking performance.
• The small number of sites limited the statistical findings in this study.

Evaluation of Rehabilitation Strategies with Respect to Structural Responses

FWD deflections were used as the response measure of the pavement structure. Deflections at the center of the slab and at the transfer joints were used in this study. JPCP and JRCP structures were evaluated independently. Sections that received an HMA overlay were monitored like flexible pavements, and deflections at the center of the lane were used.

There were limitations due to the amount of data available, especially after the data were grouped by pavement structure type and surface. Because of the small sample size (eight sites), the statistical power of the analysis was low, and no statistical differences were found in the pavement response of JPCP structures.

The only analysis that provided some statistically meaningful results was the evaluation of maximum deflection at the center lane of overlaid JRCP structures. The results suggested that crack/break and seat significantly increased the overall deflections measured on the pavement surface. The remaining treatments interchangeably provided equivalent maximum deflection magnitudes. These results were expected since crack/break and seat was an alternative in which the concrete slab was reduced to smaller pieces resulting in lower stiffness, and this increased the maximum deflection at the center of the slab.

Evaluation of Structural Responses Immediately After Rehabilitation and Future Performance

The objective of this study was to identify trends in the relationship between response measured immediately after the rehabilitation and the observed performance in the subsequent years of the pavement's service life. LTE between slabs and maximum deflection at the center of the slab were used when the surface remained concrete slabs after rehabilitation. Maximum deflection at the center of the lane was used when the surface changed to HMA after rehabilitation.

LTE Versus Performance in JPCP

Only transverse slab cracking exhibited a clear trend with LTE values in no overlaid JPCP sections, indicating that as the efficiency of the load transfer increased, the amount of transverse slab cracking decreased. This trend suggested that good load transfer joint restoration was important to mitigate the development and propagation of slab cracking.

Maximum Deflection at Center of Slab Versus JPCP Performance

The trend between performance based on roughness and deflection measured at the center of the slab suggested that the higher the deflection, the smoother the JPCP over time. This trend was not what would normally be expected. The level of slab cracking also showed an inverse trend with maximum deflection measured at the center of the slab. The trend suggested that slabs with higher deflections under FWD loading were less likely to develop cracking. A possible explanation was that stiffer subgrades resulted in higher slab curling and warping stresses, which led to increased slab cracking. This same result was found in MEPDG.⁽¹⁾ While stiffer foundations reduced axle load stresses, they increased curling and warping stresses, which often tended to dominate cracking.

Faulting was also investigated, and the observed trend suggested that faulting was inversely proportional to deflection measured at the center of the slab. High deflection values yielded low faulting, although the trend was weak and depended on only one or two points. There was no logical explanation for this result.

Maximum Deflection at Center of Lane Versus Performance of Overlaid JPCP

For sections that received an overlay as part of the rehabilitation strategy, there was a clear indication that overlaid JPCP with high deflections were more likely to become rougher pavements in the long term compared to sections with low deflection values.

Overlaid JPCP sections with high center lane deflections were more likely to experience higher rutting than sections with low deflection values. Since all rutting occurred in the HMA layer, the cause for this result was not explainable unless the HMA was so soft that it contributed significantly to the total deflection. Normally, nearly all deflection was in the foundation for JPCP.

The fatigue cracking trend suggested that high fatigue cracking was expected when deflection values were high. High longitudinal cracking values were observed when maximum deflections at the center of the lane were low, which indicated that the pavement structure was less deformable and more susceptible to surface tensile stresses, which was an important contributor to the development and propagation of longitudinal cracking.

Response Versus Performance in JRCP

The investigation of possible trends between response and performance in JRCP structures did not result in any meaningful conclusions. Different performance measures were analyzed against LTE, maximum deflection at the center of the slab, and maximum deflection at the center of the lane; however, no relevant conclusion was determined.

Findings from MEPDG Analyses

MEPDG analysis was used to compare MEPDG-predicted performance of rehabilitated pavement sections with field-measured data and to verify current calibration against predictions of rehabilitated pavement structures. The following summarizes the findings from the MEPDG analysis.

The findings for the roughness model are as follows:

• The roughness models for flexible pavements and rigid pavements provided good overall predictions of rehabilitated sections with and without HMA overlay.
• There was some bias in the predictions, which could be addressed with local or revised general calibration. The model had a tendency to underpredict roughness for rigid pavement sections with IRI values above  9.50 ft/mi (1.8 m/km). This bias was more characteristic of sections located in dry and freeze regions.

The findings for the rutting model are as follows:

• This model needs further enhancement to more accurately predict permanent deformation in HMA overlay over flexible and rigid pavements before it can be used with overlaid pavements.
• The model overpredicts performance of HMA overlays over crack/break and seat restored rigid pavements
• The model underpredicts performance of HMA overlays for saw and seal and minimum and maximum restorations prior to overlays.
• MEPDG considers the cracked/broken PCC layer as new granular base layer. Permanent deformation is predicted for the new layer and even the subgrade, which is normally the cause of the overprediction of total rutting identified in this study.

The finds for cracking models for HMA overlays are as follows:

• The cracking models for HMA overlays (particularly the empirical reflection cracking) need further enhancement to provide more accurate predictions.
• The models for fatigue cracking (new and reflective) and longitudinal cracking were not capable of predicting consistent and comparable performance with measured values.
• MEPDG did not predict transverse cracking in any of the SPS-5 or SPS-6 sections; however, some transverse cracking was measured during surveys.

Recommendations for Future Research

The following list provides suggestions for future research to further build on the knowledge gained from this study.

Researchers could monitor or create a new LTPP experiment to examine new and rehabilitated sections, including the following focus areas:

• Monitor rutting for flexible pavements, focusing on trenching and forensic studies to measure the contribution of individual layers to total rutting.
• Develop a means to assess and gather data on reflective cracking to obtain data to calibrate improved models to incorporate in MEPDG.
• Monitor response and performance data to develop or improve MEPDG models.
• Add new rigid pavement sections to the experiment to monitor undersealing of rigid pavements. The sample size used in this analysis was too small to obtain any meaningful conclusions.
• Review LTPP procedures to measure fatigue cracking of rehabilitated sections. There is evidence that cracking for such conditions was mostly associated with the propagation of cracks from the existing pavement prior to the overlay.

Another possible research plan includes future improvements of MEPDG models, including the following:

• Reassess the impact of existing conditions prior to overlay on MEPDG models for IRI. In general, MEPDG underpredicts IRI levels for poor surface conditions and thin overlays, particularly with RAP mixes.
• Use LTPP data to reduce bias on IRI estimates for aged pavements. In general, MEPDG overpredicts IRI for the control sections.
• Revise MEPDG models. MEPDG analysis results for IRI of SPS-6 sections with no overlays presented some bias. In general, the MEPDG models underpredict the IRI level, particularly for sections with maximum restoration. It is recommended to revise the MEPDG models for such conditions to remove bias, particularly for pavements with restoration procedures and pavements located in freeze zones. In general, MEPDG overpredicts rutting for sections rehabilitated with crack/break and seat alternatives. MEPDG models can be improved if the bias is removed with revised models or with improved calibration based on LTPP data for such sections. Except for crack/break and seat, MEPDG estimates for rutting on sections with overlays are underpredicted. The MEPDG models for such conditions should be revised or recalibrated to remove existing bias.
• Improve the MEPDG empirical model for reflective cracking through calibration efforts based on results obtained in this study. Two issues need to be addressed in the future. The first is how fatigue, longitudinal, and transverse cracking may be masked by reflective cracking. It was difficult to separate reflective cracking from other types of cracking during the survey measurements. The attempt to incorporate every type of cracking in the analysis conducted in this study was not very successful, particularly due to the possible confusion when measuring individual types of cracking and due to the poor performance of reflective cracking models in MEPDG. Revised models and procedures to identify reflective cracking during the surveys can help improve MEPDG estimates for cracking of rehabilitated sections of both flexible and rigid pavements.

The recommendations for technology transfer are as follows:

• Develop publications based on findings of this study that can be distributed to the industry to help engineers find the best maintenance treatment or rehabilitation alternatives for their specific conditions.
• Promote use of SPS-5 and SPS-6 data and findings to improve local and general calibration for MEPDG analysis of rehabilitated pavement sections.
• Make MEPDG input data from this project available to users for local calibration of MEPDG models.