Mechanistic-Empirical Design Guide Modeling of Asphalt Emulsion Full Depth Reclamation Mixes

Transcription

1 Mechanistic-Empirical Design Guide Modeling of Asphalt Emulsion Full Depth Reclamation Mixes Todd W. Thomas SemMaterials, L.P South Yale Avenue, Suite 700 Tulsa, Oklahoma Telephone Fax Richard W. May SemMaterials, L.P South Yale Avenue, Suite 700 Tulsa, Oklahoma Telephone Fax August 1, 2006 Prepared for the 2007 Meeting of the Transportation Research Board Word Count (Abstract 250) Text Tables Figures 1250 TOTAL (7500 max.) 6582

2 Thomas and May 2 ABSTRACT Full Depth Reclamation (FDR) using asphalt emulsion is a cost-effective method for rehabilitating distressed roads. A reclaimer mills through the asphalt layers and aggregate base, adds asphalt emulsion, and then mixes the materials. The uncompacted, recycled mixture is spread and compacted by motor graders and rollers. The new base can support traffic on the same day. Finally, a surface treatment or overlay is placed on the newly formed base. The FDR process adds strength and flexibility to the pavement structure, as well as removes distresses to improve future performance. Current pavement design methods do not account for the unique properties of FDR. These mixes behave somewhat similar to granular bases in their early life. After curing, the mixtures exhibit visco-elastic stiffness and performance-related properties similar to asphalt concrete. The mixture behavior depends on the stabilized materials and the properties of the asphalt emulsion with which they were stabilized. The asphalt emulsion used in this study is designed for returning the pavement to traffic on the same day as construction and for a quicker time to overlay compared to other emulsified asphalts. Two aggregate combinations were studied. One combination consisted of 25% reclaimed asphalt pavement (RAP) and 75% base rock. The other combination consisted of 75% RAP and 25% base rock. Cement was added to some of the mixtures. This paper discusses dynamic modulus testing performed on FDR mixes and how the test results relate to the expected performance in a pavement structure by the Mechanistic- Empirical Pavement Design Guide (MEPDG) software.

3 Thomas and May 3 INTRODUCTION Background Many miles of asphalt pavements are in need of rehabilitation. Simply milling and overlaying these worn-out pavements can lead to premature failures due to weak bases or reflective cracking. Another need is to re-use existing asphalt pavement to minimize the use of new materials for a cost-effective paving solution. Full Depth Reclamation (FDR) is a technique being used by Departments of Transportation, city, and county agencies. A typical road where FDR can be used is shown in Figure 1. FIGURE 1 Typical pavement for FDR process, Custer County, Nebraska.. Full Depth Reclamation is a road rehabilitation technique in which the full thickness of asphalt pavement and a predetermined portion of underlying base rock are uniformly pulverized and blended to create an upgraded, homogenous base material (1). When asphalt emulsion is added, the overall process improves material shear strength and adds flexibility to the pavement, while removing and directly addressing existing distresses. The process facilitates longer-lasting and improved performance from the overlying new surface layer. The 1993 AASHTO Guide (2) is typically used by agencies to design the thickness of pavement layers. However, the new Mechanistic-Empirical Pavement Design Guide (MEPDG) is a new pavement design method requiring new laboratory tests, such as the mixture Dynamic Modulus (E*). Since very little is known about emulsion-treated FDR materials in the Design Guide, it is uncertain how the FDR bases will contribute to the new pavement designs. FDR Construction Once the mix design and design of the asphalt emulsion is completed, the emulsion is transported to the road construction site by a transport truck. The transport truck feeds the emulsion to the road reclaimer. The reclaimer pumps the emulsion from the delivery truck and meters the emulsion through a spray bar with nozzles into the mixing chamber. This chamber encloses the milling head, which simultaneously mills through the road and mixes the base material with the asphalt emulsion. Typically, the milling depth is six to eight inches (150 to 200 mm). The emulsion spray rate is proportioned to the forward speed of the reclaimer. The quantity of emulsion in gallons per feet (or liters per meter) is pre-determined by the laboratory mix design.

4 Thomas and May 4 Behind the road reclaimer, the fully processed base material is ready for breakdown rolling by a padfoot roller, which is followed by a motor grader to trim the pad marks. The motor grader is usually followed by a pneumatic roller and a steel drum roller for final compaction. The final product of the road reclaimer and rolling process serves as a uniform stable foundation for a suitable wearing course. This newly constructed base layer is typically allowed to be open to traffic the same day. Once the curing period, usually 2 to 7 days, is complete, the final surface is placed. The thickness of the final surface depends on the structural pavement design. CURRENT AND FUTURE PAVEMENT DESIGN METHODS Current Pavement Design Method In the U.S, nearly all state departments of transportation (DOT) use the 1993 AASHTO Guide (2) as the primary design method for determining the required pavement thickness and evaluating the overall cross-section. This particular design method uses an empirical relationship between the condition and roughness of the roadway referred to as the Present Serviceability Index (PSI) and the traffic load determined from the number of 80 kn Equivalent Single Axle Loads (ESAL) applied to the pavement. The required pavement is characterized in this relationship by a structural number (SN) that represents the sum of the contributions of each layer component. For each layer above the subgrade, the layer coefficient of each type of material, which is a function of the modulus, is multiplied by the proposed layer thickness. Since 1960, much research has been devoted to refining and updating this empirical method for changing technologies, such as material characterization, statistical reliability, drainage, and changing traffic characteristics like tire type, suspension systems, and tire pressures. However, the inherent weaknesses of the basic 1960 regression equation eventually dictated the need for a new approach. Mechanistic-Empirical Pavement Design Method Under the National Cooperative Highway Research Program (NCHRP), Project 1-37A was initiated to develop the framework for a new Mechanistic-Empirical Pavement Design Guide (MEPDG) (3). The researchers developed a greatly expanded capacity for describing the details of each axle configuration in the design traffic mixture and for accounting for the moisture and temperature impacts on the material properties. This method also allows the designer to enter the actual material properties for a layered-elastic analysis of a flexible pavement and it contains default transfer functions for predicting distress deterioration for fatigue, rutting, and low temperature cracking. More research is planned to improve these basic components and increase the flexibility of the new MEPDG for handling innovative materials. The work described in this paper is an example of how the MEPDG could be used in the future to incorporate a different type of layer, such as FDR made with emulsion, within the pavement cross-section. TEST SAMPLE PREPARATION A total of six blends were prepared by varying the RAP, limestone aggregate, cement, and emulsion type. Variations of aggregate and RAP materials were used to produce two of the different blends. The particle size distributions of these blends are shown in Figure 2. These blends represent different scenarios of reclamation: a thick section with more RAP and a thin section with less RAP produced from the existing asphalt concrete over the aggregate base. The third blend was a variation using cement RAP, Aggregate, and Cement Variations One blend incorporated 75% limestone aggregate and 25% RAP, and it is referenced hereafter as 25% RAP. This blend had a sand equivalent (SE) value of 39. Since small amounts of cement are sometimes blended with the base materials before the addition of the emulsion,

5 Thomas and May 5 mixes with 1% cement added to this blend were compared to mixes without cement (two blends). This blend represents an existing pavement with a thin asphalt concrete wearing course that can be reclaimed into a limestone aggregate base. The second mix was a blend of 25% limestone and 75% RAP. It is referenced as 75% RAP. This blend had an SE value of 72. No cement was used in this blend (one blend). The 75% RAP blend represents a pavement with a thicker existing asphalt concrete surface. Particle Size Distribution Percent Passing % limestone 25% limestone 0 1 Sieve size, mm FIGURE 2 Particle size distribution of materials. 100 Asphalt Emulsion Types Two asphalt emulsions (referred to as Emulsion A and Emulsion B) were used in the three blends for a total of six blends. The asphalt emulsions are cationic and differ slightly in the chemicals used to emulsify them. Each emulsion was composed of 65% asphalt and 35% water. The base asphalt was tested in a Dynamic Shear Rheometer (DSR), according to AASHTO T 315 (4); the shear modulus (G*) properties are shown in Table 1. G*/sin(delta) is the rutting criterion (1.0 kpa on unaged binder) used in the Performance Grade (PG) binder system. This unaged asphalt binder meets a PG high temperature of 58 C. TABLE 1 Shear Modulus Properties of Base Asphalt Temperature, C Shear modulus G*, kpa Phase angle delta, G*/sin(delta) Water Content Before the emulsion was added, the 25% RAP mixture was mixed with 4.9% water. This was 60% of the optimum moisture content of 8.1%, in accordance with ASTM D 1557 (Method C) (5). The 75% RAP mixture was mixed with 3% water. This water content was chosen based on experience since no distinguishable peak was obtained with this blend when tested by ASTM D These water contents represent the average expected moisture in a road for these types of materials.

6 Thomas and May 6 Laboratory Mixing and Compaction A high energy laboratory mixer was used for mixing the water and then the emulsion at various emulsion contents. Before compacting, the loose material was cured for 30 minutes at 40 C. The mixtures were then compacted in a Superpave Gyratory Compactor (SGC) set for 30 gyrations to a height of about 75 mm in a 150 mm diameter mold. The compacted specimens were extruded from the mold and cured for another 72 hours at 40 C. TESTING OF MIXTURES Indirect Tensile Strength After the specimens were cooled to 25 C, dry indirect tensile strength (ITS) and conditioned indirect tensile strength (after vacuum saturation and a 24-hour soak) were determined on the specimens, generally in accordance with ASTM D 4867 (6). Mixtures with cement were not tested for ITS properties. Table 2 shows the results for the 25% RAP blend, and Table 3 shows the 75% RAP blend results. An emulsion content of 5.5% was chosen for the 25% RAP dynamic modulus specimens. An emulsion content of 3.7% was chosen for the 75% RAP dynamic modulus specimens. TABLE 2 ITS Results for the 25% RAP Blend Emulsion A Emulsion B Emulsion, % ITS, psi (kpa) Retained ITS, psi (kpa) ITS, psi (kpa) Retained ITS, psi (kpa) 4 42 (290) 19 (131) 52 (359) 14 (97) 5 43 (296) 21 (145) 41 (283) 14 (97) 6 40 (275) 20 (138) 44 (303) 16 (110) TABLE 3 ITS Results for the 75% RAP Blend Emulsion A Emulsion B Emulsion, % ITS, psi (kpa) Retained ITS, psi (kpa) ITS, psi (kpa) Retained ITS, psi (kpa) 2 44 (303) 28 (193) 48 (331) 17 (117) 3 53 (365) 33 (228) 53 (365) 24 (166) 4 53 (365) 37 (255) 51 (352) 29 (200) Preparation of Dynamic Modulus Specimens Mixtures were generally prepared as discussed above, except they were compacted to a height of 170 mm in a 150 mm diameter mold. No moisture or emulsion adjustments were made for the mixtures containing 1% cement. For measuring Dynamic Modulus (E*), the center 100 mm of the lab-compacted specimens were cored. Next, the ends were removed by sawing to achieve a final test specimen that was 100 mm in diameter by 150 mm high with more uniform air voids than a non-cut specimen. The average percentage of air voids for each set of three replicate final test specimens is shown in Table 4. Typical specimens from the two blends are shown in Figure 3. TABLE 4 Average Air Voids of Saw-Cut Specimens Emulsion A 25% RAP 0% cement Emulsion A 75% RAP 0% cement Emulsion B 25% RAP 0% cement Emulsion B 75% RAP 0% cement Emulsion A 25% RAP 1% cement Emulsion B 25% RAP 1% cement Air Voids, %

7 Thomas and May 7 FIGURE 3 Compacted and saw-cut specimens (75% RAP on left, 25% RAP on right). Dynamic Modulus Three replicate E* specimens for each mix were tested in uniaxial compression with a servohydraulic loading frame, in accordance with the AASHTO provisional test method, TP-62. Each specimen was subjected to six frequencies of loading (25, 10, 5, 1, 0.5, and 0.1 Hz) at each of five temperatures (-10, 4.4, 21.1, 37.8, and 54.4ºC), beginning with the fastest frequency and the coolest temperature. For each loading condition, a sinusoidal load was applied to maintain a peak-to-peak controlled strain magnitude of 50 to 150 microstrain within the specimen. The elastic strain was measured with 100 mm gauge length extensometers placed at three points (each 120º) around the circumference of the specimen. TEST RESULTS Results of Dynamic Modulus Tests The E* testing results for each mixture were determined for all 30 testing conditions (six frequencies at five temperatures) for each of the three replicate specimens and the overall average for the mixture. The overall average values of E* for each of the six FDR mixtures are listed in Table 5. The values for the individual specimens are also averages based on the three strains measured around the circumference of the specimen. Level 1 of the MEPDG software, which allows the user to enter the actual measured mixture properties for asphalt concrete in lieu of using calculated values from a regression equation (Level 2) or default values (Level 3), constructs the master curves from the individual mixture values by shifting the data based on binder information. The binder complex shear modulus (G*) and phase angle (δ), measured with the DSR, are used to calculate the binder viscosity over a range of temperatures with the following equation: Viscosity = (G*/10)*(1/sin(δ))^ The viscosity information is then used to establish a temperature susceptibility relationship for the binder using the following equation: Log Log Viscosity = A + VTS (Log Temp (ºR)) where: A = intercept of the viscosity-temperature relationship, VTS = viscosity temperature susceptibility or the slope of the same relationship, R = is the temperature (degrees Rankine).

8 Thomas and May 8 Normally, the binder information is measured on material that has been aged in the Rolling Thin Film Oven (RTFO) to represent a condition that simulates the aging of the binder at the mixing plant. However, with the emulsions used in FDR, no aging is required or appropriate since the mixing is performed at normal air temperatures. The master curve is determined using a reference temperature, t r, of 21.1ºC (70ºF). A polynomial shift function, that relates the shift factor, a T, to temperature, T, (Log a T = AT 2 + BT + C), is used to fit the data using numerical optimizations of the following equation: Log (E*) = δ + (α / (1 + EXP(β + γ((log (F) + c(10^(a + VTS (Log Temp (ºR))) - Log(Viscosity 70ºF))))) where: δ, α, β, γ, and c = optimized equation fitting parameters, where δ is the minimum value of E*, δ+α represents the maximum value of E*, and β, γ, and c describe the shape of the sigmoidal function. F = testing frequency The other parameters are defined above. TABLE 5 Average Measured Dynamic Moduli Dynamic Modulus (E*) Results SemMaterials Performance Testing Laboratory Full-Depth Reclamation - Emulsified Mixtures Test Test Test Emulsion A Emulsion A Emulsion B Emulsion B Emulsion A Emulsion B Temp Temp Freq 25% RAP 75% RAP 25% RAP 75% RAP 25% RAP 25% RAP deg C deg F (Hz) 0% cement 0% cement 0% cement 0% cement 1% cement 1% cement ,631,033 1,607,227 1,877,152 1,642,191 1,422,209 1,688, ,568,279 1,549,970 1,794,282 1,570,239 1,390,740 1,631, ,507,123 1,501,440 1,726,378 1,513,243 1,342,276 1,578, ,375,721 1,379,146 1,583,306 1,386,901 1,247,410 1,479, ,324,569 1,330,985 1,524,769 1,330,696 1,211,821 1,440, ,213,684 1,209,222 1,383,890 1,199,946 1,142,717 1,355, ,226,276 1,128,885 1,395,063 1,177,043 1,166,800 1,342, ,139,928 1,048,600 1,279,916 1,084,426 1,112,055 1,275, ,042, ,738 1,179, ,392 1,036,579 1,202, , , , , ,705 1,093, , , , , ,176 1,043, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,085 98, ,947 98, , , ,768 65, ,922 63, , , , , , , , , ,606 98, , , , , ,855 82, ,837 82, , , ,087 37,981 88,122 38, , , ,279 31,900 76,296 28, , , ,495 23,843 62,179 20, , ,772 This sigmoidal shaped function is the basic model format of the asphalt concrete E* used in the MEPDG. The E* master curves of all six mixtures are plotted in Figure 4. A typical shift function curve is shown in Figure 5. The emulsified mixtures are more clearly delineated at the lower reduced frequencies (warmer temperatures and slower frequencies).

9 Thomas and May 9 At this lower end of the scale, the mixtures with 75% RAP have a lower modulus with 1.5 to 2% less air voids (and are more rutting susceptible) than the mixtures with 25% RAP. As expected, by adding 1% cement, these 25% RAP mixtures exhibit an even higher modulus at this lower range of the scale. At the upper reduced frequencies (cooler temperatures and higher frequencies), the 25% RAP, with binder A and 1% cement had the lowest modulus, implying that it is less susceptible to thermal cracking. Dynamic Modulus, E* (MPa) ) 100,000 10,000 1,000 A:75% RAP A:25% RAP B:75% RAP B:25% RAP B:25%RAP;1%c A:25%RAP;1%c E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 1.E+07 Reduced Frequency (Hz) FIGURE 4 Dynamic modulus master curves for all mixtures Log Shift Factor y = x x R 2 = Temperature (F) FIGURE 5 Typical Dynamic modulus shift function for master curves.

10 Thomas and May 10 Results of 1993 AASHTO Design Guide For an assumed section of 2 inches (50 mm) of HMA over 6 inches (150 mm) of FDR over 8 inches (200 mm) of aggregate base and a lean clay subgrade, the allowable design ESAL is 2,615,000. The analysis used a design life of 20 years, 95 percent reliability, an initial serviceability (P i ) of 4.4, a terminal serviceability (P t ) of 3.0, and a soil modulus of 16,000 psi (110 MPa), the default value in the MEPDG. A structural number of 3.50 is required. The structural coefficient used for HMA is 0.44, 0.25 for FDR, and 0.14 for the aggregate base. Since ESAL are not used in the new MEPDG, traffic inputs do not relate very well. Results of MEPDG Software Analysis The MEPDG software (version 0.800, ) was used to analyze the pavement structure described above with the E* results for all six mixtures. The assumed subgrade soil was lean clay, as used above. Since the MEPDG software does not currently have performance models for emulsion mixes, the default distress models for asphalt concrete were used for this analysis. Fatigue cracking models for these emulsion mixes are being developed. Table 6 shows the predicted performance results for the six different FDR mixtures placed in the same cross-section, subjected to 150 heavy trucks initially per day, with 4 percent annual growth, for 20 years of service. The total number of trucks accumulated over this 20 year service period with the 4 percent annual growth rate is 756,068. If these are considered to be all legally loaded tractor-trailers, this equates to approximately 2,000,000 ESAL, or about 76% of the allowable ESAL in the 1993 AASHTO analysis. The estimated performance was excellent for all of these cold-recycled asphalt concrete cases using the default models, with little variation due to the mix composition. TABLE 6 MEPDG Software Analysis Results NCHRP 1-37A Mechanistic-Empirical Pavement Design Guide Version Predicted Performance 20 years 50 mm AC / 150 mm FDR / 200 mm Aggregate Base / Lean Clay Subgrade Initial Traffic : 150 heavy trucks per day with 4 % annual growth Emulsion A Emulsion A Emulsion B Emulsion B Emulsion A Emulsion B 25% RAP 75% RAP 25% RAP 75% RAP 25% RAP 25% RAP Predicted Distress Unit 0% cement 0% cement 0% cement 0% cement 1% cement 1% cement Top-Down Longitudinal Cracking m/km Bottom-Up Fatigue Cracking % Thermal Transverse Cracking m/km Asphalt Layer Rutting mm Total Rutting mm ,068 Total Heavy Trucks accumulated over 20 Years SUMMARY AND CONCLUSIONS Full Depth Reclamation is process for improving worn out pavements by cold-mixing the existing asphalt pavement into the base and adding asphalt emulsion. The result is an upgraded, stabilized base that is ready for a final surface. Little is known how this engineered base can be used for full benefit in structural pavement designs. Since the new Mechanistic-Empirical Pavement Design Guide does not account for the added stability of emulsion FDR, pavement designers will not be able to fully quantify its structural benefit. To begin the process of understanding the added structural value of FDR, laboratory testing was performed to measure various properties. The dynamic modulus results showed some differences in the mixtures.

11 Thomas and May 11 As expected, at the lower frequencies (warmer temperatures), the mixtures with cement have higher E* values. These same mixes are not as distinctive at the higher end of the scale (colder temperatures). Emulsion B had higher E* values for both RAP mixtures, except that Emulsion A had higher E* values for the 75% RAP mixtures at the lower frequencies and higher temperatures. The 25% RAP mixtures had higher E* values than the 75% RAP mixtures in this particular case; however, this is probably impacted by the different emulsion contents. In general, the mixtures are much more clearly delineated at the lower reduced frequencies (warmer temperatures and slower frequencies). The distresses predicted by the MEPDG software for a typical application after 20 years of service were not severe. The default asphalt concrete performance models were used in the software, because further model development is needed to predict emulsion FDR performance. The mixtures with 1% cement show the least predicted distresses, though none of the predictions indicate poor performance. There was no difference in predicted thermal transverse cracking because this is a topdown distress and the asphalt concrete surface was the same for all cases. It is interesting however that the top-down longitudinal fatigue cracking did vary slightly because of the minor change in asphalt concrete base condition. Currently, due to a lack of research effort in this area, the MEPDG treats FDR as an unbound material with a default constant modulus of 69 MPa (10,000 psi). This value is unrealistic considering, for example, that the mixture with Emulsion A, 25% RAP, and no cement had an E* of 4415 MPa (640,000 psi) at 21 C and 10 Hz. Although the stress-strain behavior (modulus) of emulsified recycled asphalt mixtures does vary significantly with mix composition, binder type, and quantity of RAP, it has been clearly demonstrated that these FDR mixes would be better characterized as a less-aged asphalt concrete. In addition, there is much opportunity for further optimization of these types of materials. A rural highway, using these base materials and a thin bituminous surface, was reclaimed with Emulsion A without cement in the summer of 2003 and is performing well, carrying mostly car traffic and agricultural equipment. ACKNOWLEDGMENTS The authors gratefully acknowledge the contributions of Bill Criqui, Phil Blankenship, Lou Harper, Stephane Charmot, and Matt Carnal of SemMaterials, L.P. and Helen King of GHK, Inc. for their contributions to this paper. The authors also thank Mickey Blizzard and Brian Majeska of SemMaterials, L.P. for sponsoring this work. REFERENCES 1. Basic Asphalt Recycling Manual. Asphalt Recycling and Reclaiming Association, USA, 2001, p AASHTO Guide for Design of Pavement Structures. American Association of State Highway and Transportation Officials, Washington, DC, Development of the 2002 Guide for the Design of New and Rehabilitated Pavement Structures. Design Guide and Supplemental Documentation. NCHRP Project 1-37A, Transportation Research Board of the National Academies, Washington, DC, Standard Specifications for Transportation Materials and Methods of Sampling and Testing. American Association of State Highway and Transportation Officials, Washington, DC, Annual Book of ASTM Standards, Vol : Soil and Rock (I), ASTM, West Conshohocken, PA, Annual Book of ASTM Standards, Vol : Road and Paving Materials; Vehicle- Pavement Systems, ASTM, West Conshohocken, PA, 2006.

12 Thomas and May 12 LIST OF FIGURES AND TABLES FIGURE 1 Typical pavement for FDR process, Custer County, Nebraska FIGURE 2 Particle size distribution of materials. FIGURE 3 Compacted and saw-cut specimens (75% RAP on left and 25% RAP on right). FIGURE 4 Dynamic modulus master curves for all mixtures. FIGURE 5 Typical Dynamic modulus shift function for master curves. TABLE 1 Shear Modulus Properties of Base Asphalt TABLE 2 ITS Results for the 25% RAP Blend TABLE 3 ITS Results for the 75% RAP Blend TABLE 5 Average Measured Dynamic Moduli TABLE 6 MEPDG Software Analysis Results