Evaluation of Pavement Rehabilitation Strategies on Route 165 and Prediction Performance

Many roads in Rhode Island are coming to their intended design life and are now considered in poor condition. Significant number of roads with severe deterioration is being rehabilitated through full depth reclamations (FDR) with various additives. FDR can rejuvenate subbase and pavement structures. Route 165 in Exeter, Rhode Island, was selected as a test road with four different treatments and a control. The road had severe pavement distresses such as alligator cracking, pot holes, shoving and raveling and was not a candidate for a resurfacing. The road had a FDR in 2013 which included a control section, three test sections with additives which consisted of calcium chloride, asphalt emulsion, and Portland cement, and a geo-grid section. Triaxial testing was performed on the subbase materials and subgrade soils before and after the FDR treatments to determine the resilient modulus. The results of the material testing were used to predict the performance of each of the test sections by using AASHTOWare Pavement ME Design software. The 200 mm (8.0 in.) rehabilitated base/subbase layer was covered with 62.5 mm (2.5 in.) Hot Mix Asphalt (HMA) with Warm Mix Asphalt (WMA) additive base and 50 mm (2 in.) surface. The maximum sizes of base and surface aggregate were 19 mm (3/4 in.) and 12.5 mm (1/2 in.), respectively. Properties of HMA mixtures with WMA additives including dynamic modulus were determined as input parameters and for further analysis Of the five test sections, it was predicted that the pavement with the FDR layer stabilized with Portland cement would perform the best overall. Next is calcium chloride followed by the control (no additive), geo-grid and asphalt emulsion. A condition survey was conducted in 2015 winter, but there was no surface distresses found on any of the five test sections. A plan for long term performance evaluation has been developed, and an optimal strategy has been recommended, i.e., predicting performances before rehabilitating any broken roads.


INTRODUCTION
It has been estimated that the amount of miles of truck traffic on our highways will be increasing and surpassing all other modes of freight shipments in the near future. Tractor trailers and heavy vehicles account for a majority of the damage done to highways (Lee and Peckham et al. 1990). The states, especially Rhode Island, are having a hard time keeping up with and paying for maintenance and rehabilitation (M&R). This means there will be more wear done to our highways than ever before, and the states will have to do more M&R with less funding. To meet upcoming highway demand, the Rhode Island Department of Transportation (RIDOT) has been testing alternative subbase material strategies such as full depth reclamations (FDR) and has been expanding their use.
RIDOT wants to build a better road, to have less physical maintenance and to control costs. The final product according to RIDOT is a high performance road. Reclaiming a roadway can fulfill RIDOTs wants by increasing the stiffness of the subbase and increase the pavement life.
Achieving higher performance at a 30% to 50% cost savings can be realized with a full reclamation according to pavement recycling systems. Reclaimed materials are retained and reused on site, consequently reducing trucking costs for new materials. In the 1980, RIDOT had a program to reclaim pavements throughout the state. In 2013, Route 165 was slated to re-reclaimed. An idea was formed to use four different strategies and a control (Figure 1.1). The objectives of this project are: to test the existing subbase materials before and after full depth reclamation, to predict the performance of different subbase strategies and to evaluate the short-term and long-term performance over time.

Objective of Study
Route 165 is a unique candidate for research because it is seven miles long without many intersections, the subgrade layer has a high water table and there is severe frost action in winter. The objectives of the present research project are as follows:

7.
Provide an optimal design and strategies for future RIDOT rehabilitation projects.
* Assist future testing and evaluation after the 2016 summer.
The outcome of this research project will provide a guideline for future maintenance and rehabilitation (M&R) projects. It may be noted that, Route 165 was originally built on soft deposits (swamp). Depending on the nature of the soft deposit, "construction" dealt with this in one of two ways: one was by removal of the unsuitable material and the other was by "floating the embankment on the soft soil, often with considerable settlement" (Nacci et al., 1987).
Eleven test borings were completed for the reconstruction, which found embankments consisting of sand, some gravel, silt, fibrous organic deposits (peat), and organic silt. Other test borings indicated that Route 165 was built on glacial till and stratified kame deposits. There were pockets in the granite bedrock near the surface, which contributed to a high water table. An exploration and analysis found an additional seven areas of swamp deposits.   There are several WMA properties that are needed to be inputted into the Pavement ME program to perform a Level 1 design. The RIDOT performs fifteen AASHTO and ASTM testing on its materials and warm mix asphalt as shown in Table 3-1 for all its construction projects. These tests are the MED inputs that are needed to predict longitudinal cracking, alligator cracking, transverse cracking, rutting or other permanent deformation, IRI, and reflection cracking over a selected design life.

Resilient Modulus of Subgrade Soils
As discussed in Chapter 2, subbase and subgrade soil samples were collected during construction for testing. According to the results of a sieve analysis on the material, the subbase consisted of gravelly sand or A-1-b AASHTO classification which is consistent with the material shown for that location in the 1981 Soil Survey of Rhode Island; this soil was also found under a previous URI study (Lee et al. 2003). The URI study reported resilient modulus (Mr), which is deviator stress over recoverable strain, values for Rhode Island subgrade soils ranged from 7,506 psi to 9,304 psi (Lee et al. 2003) and an Idaho study for comparison shows the same types of gravel material ranged from 8,000 psi to 19,000 psi (Hardcastle et al. 1993).

Resilient Modulus of Existing Subbase.
Before the Full Depth Reclamation (FDR), four inches of asphalt pavement were removed from the roadbed for ten test sections located throughout the length of the road.
Approximately twelve inches of existing subbase layer including five inches of previously recycled material were collected. It should be noted, the collected samples were mix with seven inches of the existing gravel borrow and the five inches of previously reclaimed material was not tested separately. Resilient moduli of the ten subbase test sections were determined by using triaxial chamber apparatus according to AASHTO T 307-99 procedure.
Resilient moduli values are presented in Table 3.2.
The laboratory resilient moduli values varied from 17, 000 psi to 74,000 psi.
Subsequent material testing of the ten samples for the percent RAP content using AASHTO T 308, are shown in Table 2.5. For the 1980 old recycled mixture, the subbase samples have an asphalt content between fifteen and forty percent with most of the samples having a pavement content of twenty-five percent.
A grain size distribution, AASHTO T 27, Figure 3.2 sieve analysis shows the samples as relatively uniform . The samples show that the contractor performed a good job of reclaiming.

Resilient Modulus of New Full Depth Reclamation (FDR) of New Base/Subbase
In construction, four inches of old asphalt surface and base layers were reclaimed into four inches of previously reclaimed subbase, and a new eight-inch homogeneous FDR base/subbase layer was formed. Samples were taken, before the new construction FDR base/subbase layer were mixed with the three different strategies, to URI for testing. Before triaxial testing, four samples were mixed with additives in the lab according to RIDOT specifications for Route 165. Out of the six samples two control FDR samples were tested without additives, one sample was mixed with CaCl2, one sample was mixed with asphalt emulsion, and two samples were mixed with Portland cement. For the Portland cement samples, one was cured for 4 hours and the other 7 days before testing.
The resilient moduli of FDR base/subbase layer were determined by using AASHTO T307-99, and results are shown in Table 3.3 .
Six samples for the percent asphalt (RAP) content, maximum dry unit weight, optimum moisture content and Dry unit weight shown in Table 3.4 and a grain size distribution, Figure 3.3, show the samples as relatively uniform .

Determination of Physical Properties of Asphalt Base and Surface Layers
The Route 165 project used two and one half inches of Class 19 WMA for the base layer and two inches of Class 12.5 WMA for the surface layer. Mechanical properties of WMA including Dynamic modulus, (EHMA, E *) for the surface and base layers were acquired from Villanova University and Cardi Corporation's WMA testing, and are shown in Appendix B. An example of the physical Properties of Class 19 WMA base layer is shown in Table 3.5, and are used as input parameters for MED software.
Creep compliance was acquired from a URI study, and used as an input parameter for the MED software (Lee et al., 2014 The control test section on Route 165 was reclaimed to a depth of eight inches and did not receive any additives. After the FDR, as shown in Figure 2.2, a one inch of old 1980 recycled blend with the old CaCl2 subbase material was assumed left and is represented in the MED.

Prediction of Performance for the Calcium Chloride Section
The calcium chloride section was full depth recycled and mixed with CaCl2 to a depth of eight inches on Route 165. After the FDR, as shown in Figure 2.3, a one inch of old recycled blend with CaCl2 is assumed left and is represented in the MED.

Prediction of Performance for Portland Cement Section
The Portland cement section was full depth recycled to a depth of eight inches with the cement mixed throughout. A one inch of old recycled blend mixed with CaCl2 is assumed left and is represented in the MED, Figure 2.4.
There were two Portland cement samples tested for this project. Sample 4a was mixed with Portland cement (PC) and tested after four hours, while sample 4b was mixed with PC and tested after 7 days. The Portland cement section on Route 165 was micro cracked after four hours and traffic was allowed on the newly compacted surface. Since micro cracking prevents the PC to gain any more stiffness, sample 4a is used in this study.
The MED was run with the layer thicknesses and Mr as shown below in Table   3.11. There are no predicted output failures. The target and predicted values are from the MED Portland Cement URI generated report page 1. Shown in Appendix A.

Prediction of Performance for Asphalt Emulsion Section
The asphalt emulsion section was full depth recycled to a depth of eight inches with only the first three inches mixed with emulsion as shown in Figure 2.5.
The MED was run with the layer thicknesses and Mr as shown below in Table 3.10.
There is one predicted design output failure for AC top-down fatigue cracking (longitudinal cracking).

Prediction of Performance for Geo-Grid Section
The Tensar geo-grid section used full depth recycled material to a depth of ten inches and did not receive any additives. To install the geo-grid, sixteen inches of subbase were removed from the road after FDR and stockpiled. The geo-grid was installed on top of the subgrade and six inches of filter stone were placed on the geo-grid. Another geo grid layer was placed over the filter stone and ten inches of FDR were placed and compacted, Figure   2.6. For this test, the control material for sample 7b Mr mean values for confining stress of 35 kPa were used from Table 3.3.
The MED was run with the layer thicknesses and Mr as shown below in Table 3.12.
There is one predicted design output failure for AC top-down fatigue cracking (longitudinal cracking).

Summary
The comparison between the control and the other four test sections are shown below in The test sections in order of best performance are: Portland cement, CaCl2, control, geo-grid and asphalt emulsion with the smallest amount of cracking and highest predicted threshold distresses in years. All the test sections predict that there will not be any permanent deformation in the subbase or AC layer, or AC bottom up fatigue cracking (alligator cracking). The higher resilient moduli, the better the results for less distresses.    Thus, based on the aforementioned, we know that it will take a number of years before the threshold distress can be predicted for the five test sections i.e control, CaCl2, asphalt emulsion, Portland cement, and geo-grid. We do know that year number eight is the "unofficial" time where pavements start showing distress and will most likely need to receive some form of maintenance treatment e.g crack sealing; chip seal. That said, the asphalt emulsion and geo-grid test sections are predicted to need treatment in four and five years in this study, with predicted AC top down cracking of 3,502 ft/mile and 3,367 ft/mile.
It was surprising to see how high the AC top down fatigue cracking number in the asphalt emulsion test section because the material has been successfully used as an FDR additive in many states.

Control Test Section
According to Table 4.4, the control section was predicted to reach a threshold distress in AC top down cracking in eight years. If the pavement thickness had been increased by one inch, all of the predicted values would have been below the target thresholds for twenty years. The down side to this would have been the approximately $526,000 in additional costs for 8,552 tons of asphalt pavement. The yearly cost for eight years is $332,676 and $159,368 for twenty years. The total cost with the extra asphalt pavement, the control test section ranks two out of five.

Calcium Chloride Section
The CaCl2 section is predicted to reach a threshold of 2,585 linear feet of AC top down fatigue cracking (longitudinal cracking) in ten years see Table 4.5. If the pavement thickness had been increased by one inch, all of the predicted values would have been below the target thresholds for twenty years. The down side to this would have been the approximately $526,000 in additional costs for 8,552 tons of asphalt pavement. The yearly cost for ten years is $364,715 and $208,655 for twenty years. The total cost with the extra pavement, the calcium chloride section ranks four out of five.

Portland Cement Section
In Table 4.6, the Portland cement section, which had the highest Mr of any of the other sections, should not show any distress until roughly year thirty and it is in the terminal IRI (in/mile). In spite of this, the cement was not allowed to fully cure for seven days and thus micro-cracked after four hours. From Table 3

Asphalt Emulsion Section
According to Table 4.7, the asphalt emulsion section was predicted to reach a threshold distress in AC top down cracking in four years. If the pavement thickness had been increased by one and a half inch, all of the predicted values would have been below the target thresholds. The down side to this would be the increased cost of 12,828 tons of asphalt needed to reach the additional thickness. The $788,000 is a high price to pay for fifteen additional service years. But the cost for four years would be $742,351 and $187,916 for twenty years. The total cost with the extra pavement, the calcium chloride section ranks three out of five.

Geogrid Section
According to Table 4.8, the geo-grid section was predicted to reach a threshold distress in AC top down cracking in five years. If the pavement thickness had been increased by one and a half inch, all of the predicted values would be below the target thresholds and the pavement would not see any predicted cracking for twenty years. The down side would be the increased cost of $788,000.00 for 12,828 tons of asphalt pavement. The up side, however, would be an increase in fifteen additional service years. The yearly cost for ten years is $1,326,543 and $357,933 for twenty years. The total cost with the extra pavement, the geo-grid section ranks five out of five.

Forecasting Future Performance through Tie-ins with Pavement Structural
Health Index (PSHI) MED can fit within the states' preservation system by using performance indicators that dTIMS does not. For instance, dTIMS catalogues IRI, rutting, cracking and deformation through yearly field surveys, while MED uses AADTT, resilient modulus, pavement layer make-up, HMA properties, and climate to predict the same pavement distresses over time. DTIMS surveys the surface course and MED predicts the subsurface conditions. Thus, MED predictions can be adjusted accordingly.

Optimal Strategies for Rehabilitation
One finding of the MED, is determining the right combination of subbase Mr to pavement thickness. MED makes it very easy to run multiple models for worst and best case scenarios. Resurfacings and reclamations should have Mr values checked before construction to find if an additive would benefit the subbase stiffness. Monitoring pavement PSHI can catch a road before it deteriorates too far, but the subbase Mr should be evaluated before treatment is determined. For example, too many times RIDOT has milled two inches and put back two inches of pavement on a road only to have the pavement break up in a short amount of time. The MED reported output shows how longitudinal cracking can be a sign of too thin of a pavement thickness.

Guidelines for Long-Term Evaluation and Optimal Rehabilitation Design Strategies
A material database consisting of resilient moduli, pavement core data and sieve analysis needs to be created for easy reference for Design Engineers. The RIDOT has years of collected data but unfortunately no "on-line" database. URI, on the other hand, has already done extensive resilient moduli testing with seasonal variations on subbase and subgrade materials and needs to incorporate these results into the state's database. The results of the testing should be included in one main database along with any new testing done. (Lee et al., 2001) LTPP currently has a Microsoft Excel Program that uses falling weight deflectometer (FWD) deflections to predict resilient moduli of the asphalt layers, subbase and subgrade materials. The program, however, requires pavement and subbase thicknesses as input parameters which a GPR can provide. FWD testing is already being performed on state highways and this information should be appropriately documented and compiled into a database.         There were several findings while investigating the performance of the five test sections.

Pavement Distress and Cost
Of the five test sections, the section with Portland cement performed the best overall in having the least amount of distresses and the longest predicted service life; next is calcium chloride, followed by cold recycled (control), geo-grid, and asphalt emulsion. Table 4.8 shows the comparison of years of predicted threshold distress and cost for AC top down fatigue cracking (ACTDFC) in feet per mile. The AC top-down fatigue cracking and predicted years were obtained from the AASHTOWare Pavement ME design software using the lab tested resilient moduli. According to the prediction models, in Table 4.8, the Portland cement section was the only section that doesn't fail in the ACTDFC. In addition, is predicted to last over twenty-five years. The other test sections are predicted to reach the ACTDFC between five and ten years.
The costs of the pavement structure for all the sections were between 2.6 million to 6.6 million dollars. The asphalt emulsion and geo-grid sections are not cost effective since they are predicted to reach threshold distress in a very short amount of time.

Permanent Deformations, AC Bottom Up, Fatigue Cracking and Thermal Cracking
None of the test sections are predicted to have permanent deformation, AC bottom up fatigue, or thermal AC cracking for over a twenty year period as shown in Table 4.3. Terminal IRI is predicted to be sixteen percent below the target goal for twenty years. Other distresses such as permanent deformation in the asphalt layer and AC bottom up fatigue cracking are predicted not to be present for twenty years. As discussed, AC top down fatigue cracking fails in all of the test sections except for Portland cement.

Portland Cement Section
It appears that Portland cement is an excellent additive, but the curing time can be a problem on narrow roads like Route 165 where detours are not possible. Detours drive up the costs for the project because of the additional traffic control and the delays to the traveling public.
Portland cement should be considered for future projects only where a detour is feasible. Route 150 heavy trucks per day, and would benefit greatly from a more durable pavement like Portland cement. However this roadway could not support a detour therefore used alternating traffic was used for construction.

Resilient Modulus
The various test sections had different lab calculated resilient moduli.

Condition Survey
Condition surveys from the dTIMS database can be used to verify the MED predictions of the five test sections. For this project, a windshield survey was completed after construction in 2015 and an automated survey for download into dTIMS. RIDOT will track the pavement performance for years to determine the best performance section.   Table 4.9 for costs associated with the ideal pavement thickness and the order changes slightly as follows: Portland cement, control, calcium chloride, asphalt emulsion and geo-grid.

Recommendations
There are several recommendations that will be presented to RIDOT on how to increase pavement performance. They are: 1. Perform triaxial tests and FWD testing on subbase material on future FDR projects in the planning stage of design. Any subbase material which has less than 25,000 psi of resilient modulus should be modified with an additive to increase its stiffness. The number of triaxial tests and FWDs can be determined by how distressed the pavement is. The poorer the pavement condition, the more testing should be done to determine the cause of failure.
2. MED and PHSI have great potential to predict and monitor future performance of FDR roads.
Thus, RIDOT may consider using both tools for the design of new reclaimated pavement structures.
3. The Hogg Method should be investigated to see if its results can be used as MED input data.
The Excel program can be modified and calibrated to lab results to better fit the state's subbase material. Triaxial tests can also be used to verify the Hogg Method results.
4. It is recommended that Portland cement should be used as an additive if there is heavy truck traffic and the pavement is wide enough to support detours or lane closures for at least the seven day recommended curing time without traffic loads.
5. Asphalt emulsion could be investigated further on another state road. Increasing the depth of the layer that contains emulsion from three inches to five or six inches could improve performance, but it will increase the cost. We could analyze our predictions against the condition survey data from dTIMS to see how close the predictions were to the MED.
6. RIDOT leases the MED program from AASHTO on a yearly basis and should be renewed.
Earlier versions of the MED were slow, but new software provides not only important prediction information but also runs in less than five minutes. The information that can be derived from MED can be highly useful if dTIMS, GPR, triaxial tests and/or FWDs are used in conjunction with the MED.
7. Collecting the data for the MED inputs took two weeks for all the test sections. All the information to do a Level 1 analysis is available but it is scattered throughout the RIDOT Departments. A library of pavement material, subbase resilient moduli, AADTT, cores, and GPR can improve collection time and prevent unnecessary extra testing.
8. If the RIDOT's falling weight deflectometer and ground penetrating radar are to be used to determine the pavement stiffness and pavement thickness, these machines need to be calibrated on a regular basis and receive scheduled maintenance.

Conclusions
Overall, the Portland cement test section has the highest predicted performance, but is difficult to construct due to curing time and traffic detours on Route 165. Calcium chloride and the control sections were least expensive, but needed to have a thick base course which would increase cost. Geo-grid is typically used in poor drainage areas with high water tables on highways and has a high cost. Asphalt emulsions can work if either the asphalt emulsion is mixed to a deeper depth or an additional inch and a half of base course is used.
It should be interesting to see in the coming years how well the MED predictions for the test sections compare with the future distresses.