Cosentino, Bleakley, Sajjadi, and Petersen

1

Evaluating  the  Laboratory  Compaction  Techniques  of  Reclaimed   Asphalt  Pavement   Words: 4321 Words Counted Twice in Figures and Tables plus Title Page Words = -343 Tables: 0 = 0 Figures: 13 = 3250 Total Word Count: 7228 Corresponding Author: Paul J. Cosentino, Ph.D., P.E., Florida Institute of Technology, Melbourne, FL 32901. Phone 321 674-7555, FAX 321 674-7565, [email protected]. Albert M. Bleakley, Ph.D., P.E., Florida Institute of Technology, Melbourne, FL 32901. Phone 321 674-8048, FAX 321 674-8251, [email protected]. Amir M. Sajjadi, Graduate Research Assistant, Florida Institute of Technology, Melbourne, FL 32901. Phone 321 674-8048, FAX 321 674-7565, [email protected]. Andrew J. Petersen, E.I., Project Engineer, Creech Engineers, Inc. 4450 W. Eau Gallie Blvd. Suite 232, Melbourne, FL 32934; 321.255.5434 Fax: 321.255.7751, http://www.creechinc.com/ ; [email protected]

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

2

ABSTRACT   Reclaimed Asphalt Pavement (RAP) is a byproduct of roadway resurfacing. A limited amount can be recycled into new hot mix asphalt; the rest is stockpiled. Some states allow the use of RAP/aggregate blends as base course material. Due to its low strength and susceptibility to creep deformation the Florida Department of Transportation (FDOT) excludes RAP as pavement base course for high traffic areas. The research objective was to determine whether the strength characteristics of RAP could be improved through compaction, to make it base suitable in high traffic areas. Modified Proctor, vibratory, and gyratory compaction data were compared. Four RAP sources were used. Specimens compacted by the three methods were tested using Limerock Bearing Ratio (LBR), Unconfined Compressive, and Indirect Split Tensile strength. LBR is Florida’s variation of the California Bearing Ratio. Specimens were compacted to either a density or to a compaction energy level. Vibratory compaction produced the lowest densities and strengths. Modified Proctor produced higher densities and strengths than vibratory but the LBR strengths for all RAP types were consistently below FDOT standards. Gyratory compaction produced the highest densities and strengths. Gyratory RAP specimens were two to four times stronger than modified Proctor specimens at the same density. Compaction method did not have as significant an effect on creep, although gyratory compacted samples did produce less creep than modified Proctor compacted samples.

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

3

INTRODUCTION The pavement milling process produces reclaimed asphalt pavement (RAP) a portion of which can be recycled directly into new hot-mix asphalt pavement. The remaining portion, often hauled off site and stockpiled, is available for other highway uses. Although RAP, a granular material, possesses good drainage characteristics and adequate shear strength, it has low bearing strength and under constant stress produces excessive creep. The Florida Department of Transportation (FDOT) bases bearing strength on the limerock bearing ratio (LBR) test conducted in accordance with Florida Method (FM) 5-515 (1). The LBR, found by dividing the stress at 0.1-inch (2.54 mm) deflection by 800 psi (5600 kPa), produces values, which are 25 % higher than California Bearing Ratios. OBJECTIVE The research objective was to evaluate the modified Proctor, vibratory and gyratory compaction of RAP through LBR and one-dimensional creep testing. RAP and limerock, cemented coquina and clayey sand were used for the testing; with 50/50 blends of RAP and limerock included. Limerock served as the control material.

LITERATURE     RAP, which is not reused on site, is typically transported to asphalt plants where it is crushed and stockpiled. Crushing produces a more uniform product. RAP not crushed is termed milled RAP to differentiate it from crushed RAP. Sandin (2) performed a statewide variability evaluation of Florida RAP and determined that the average statewide asphalt content of crushed RAP is lower than that of milled RAP. Montemayor (3) conducted Proctor compaction tests on RAP and found that it did not produce well-defined maximum density curves. He concluded that specifying density, as a percentage of the modified Proctor maximum density, might not be realistic for RAP. Cosentino and Kalajian (4) subjected RAP to static pressures of 212, 400, 700, and 1000 psi (1484, 2800, 4900, and 7000 kPa), and found that densities increased approximately 4% per loading interval, while LBR increased significantly with each interval. The densities achieved were higher than those achieved with modified Proctor compaction. Gyratory compaction (ASTM D6925), used in asphalt mix design, has also been used on various soils. Gyratory compaction of Florida base and subbase soils produced dry unit weights closer to field compaction than either modified Proctor or vibratory compaction. Gyratory compaction is most sensitive to the number of gyrations (5, 6, 7), less sensitive to the gyration rate, and not affected by the gyration angle (5, 6).

RAP  AND  AGGREGATE  SOURCES   Four milled and crushed RAP stockpiles were sampled from three asphalt plants. APAC Florida plants in Jacksonville and Melbourne, Florida, plus the V. E. Whitehurst and Sons plant in Gainesville, Florida, were sampled. Both crushed and milled RAP were obtained from the APAC Melbourne facility, while only crushed RAP was obtained from the APAC Jacksonville plant and milled RAP from the Whitehurst plant. These sources were referred to as Melbourne Milled, Melbourne Crushed, Whitehurst Milled and Jacksonville Crushed. The four RAP sources were compared to limerock, cemented coquina, and clayey sand. The limerock and cemented coquina were obtained from FDOT approved base course sources since RAP was being evaluated for potential use as base material (8).

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

4

COMPACTION  METHODS   The gyratory compaction machine could be set for either compactive effort (number of gyrations) or density (sample height). Therefore, modified Proctor specimens were made first and then gyratory specimens were made to match the modified Proctor densities. Regardless of vibration time, vibratory compaction produced densities below modified Proctor and gyratory compaction and therefore could not be matched.

STRENGTH  AND  CREEP  TESTING   The strength testing included LBR, unconfined compression (UCC) and indirect tensile (IDT) testing. The creep testing was performed using one-dimensional confined and unconfined creep tests conducted at constant stress of 12 psi (84 kPa). Similar size samples were produced from all three compaction devices. Samples obtained from LBR testing were 6.00-inch (152.4 mm) diameter and 4.58-inch (116.4 mm) high, which is the size of the LBR mold. Samples from gyratory compaction were 5.91-inch diameter (150.00 mm) and samples from vibratory compaction were 6.00-inch diameter (152.4 mm), however their sample heights varied. LBR specimens were not soaked to enable direct comparisons between specimens from all compaction methods. The gyratory and vibratory molds do not allow soaking. UCC and IDT tests were conducted on ejected modified Proctor and gyratory specimens to eliminate mold geometry effects. Unconfined creep tests were performed to investigate the creep of ejected modified Proctor and gyratory specimens. Due to the geometry of the vibratory mold, it was not possible to eject or test vibratory specimens.

DENSITY  RESULTS   Modified Proctor Modified Proctor moisture-density relationships from the four RAP stockpiles based on 3rd order polynomials, are shown in Figure 1. Cohesionless granular soils may produce an ‘S-curve’ moisture-density relationship, typically yielding high densities at both low and high moisture contents (9). Three of the four RAP sources produce this ‘S-curve’. RAP sampled from APAC Jacksonville followed the typical moisture density parabola probably due to the higher percentage of material passing the #200 sieve, which may produce cohesive effects.

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

5

130

Dry Density (pcf)

125 120 115 110 105 100 0

2

4

6

8

10

12

Moisture Content (%) Melbourne Crushed

Melbourne Milled

Whitehurst

Jacksonville

FIGURE 1 Dry density versus Moisture Content for Modified Proctor Compaction of RAP (1 pcf = 0.157 kN /m3). The highest densities were produced by the APAC Jacksonville crushed RAP which had a specific gravity (Gs) of 2.604 and the largest percent fines (7%), the second highest were from the Whitehurst RAP with a Gs of 2.576, while the APAC Melbourne crushed Gs was 2.508 and milled Gs was 2.524. The higher crushed Melbourne RAP densities are probably due to higher fine content. Gyratory Moisture-density relationships at 75 gyrations are presented in Figure 2. RAP was tested at target moisture contents of 3%, 6%, and 9% to determine if moisture content had an effect on gyratory compaction. Moisture contents below 9% were used to prevent excess water from leaking into the gyratory compactor. Based upon the results (Figure 2), three of the four samples showed peak-dry densities. Whitehurst Milled RAP did not. Polynomial fits were placed through each set of data. Melbourne crushed RAP polynomial fit peak density occurred at 5.1 %, Melbourne milled RAP polynomial peak occurred near 6.1%, and APAC Jacksonville RAP polynomial fit peak dry density occurred near 4.6%. The large difference in densities of Melbourne milled RAP from 3% to 5.5% moisture may be the result of sample variability since these materials were obtained at two different times. Peak densities ranged from near 122 and 126 pcf (19.1 and 19.7 kN/m3) for all four sources.

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

6

130 128 126

Jacksonville, Crushed

Dry Density (pcf)

124 122 120 Whitehurst, Milled

118

Melbourne, Crushed

116

Melbourne, Milled

114 112 110 0

2

4

6

8

10

12

Moisture Content (%)

FIGURE 2 Dry Density versus Moisture Content for Gyratory Compaction of RAP (1 pcf = 0.157 kN/m3). Vibratory     The vibratory compaction tests were conducted following ASTM D4253. Vibratory compaction did not produce well-defined optimum moisture contents. Moisture-density plots, (Figure 3), show that three of the four materials produced peak densities using linear and 2nd order polynomial fits, at very high moisture contents, while the Whitehurst milled RAP produced a curve similar to modified curves. Moisture contents higher than 9 % could not be produced with the vibratory equipment because water vibrated out of the mold during compaction.

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

7

130 125

APAC Melbourne, Crushed

Dry Density (pcf)

120

APAC Jacksonville, Crushed

115 110 Whitehurst, Milled

105 100 95

APAC Melbourne, Milled

90 0

2

4

6

8

10

12

Moisture Content (%)

FIGURE 3 Dry Density versus Moisture Content for Vibratory Compaction of RAP (1 pcf = 0.157 kN /m3). As shown in Figure 3, correlations between dry density and moisture content for vibratory compaction were inconclusive and varied by source. The Whitehurst milled RAP appears to show a conventional relationship, where dry density peaked at approximately 111.0 pcf (17.4 kN/m3) at an optimum moisture content of 4.0%. The remaining three samples initially decrease in density as moisture content increased, and then increased as moisture content increased above 4.0%. Second order polynomial trend lines, which are shown, yielded the highest regressions. Crushed RAP produced higher densities at the higher moisture contents than milled RAP. Similar to the modified Proctor results, APAC Jacksonville crushed, with the highest % fines and specific gravity, produced the highest densities.

STRENGTH  TESTING  RESULTS   Limerock  Bearing  Ratio     Following modified Proctor, vibratory and gyratory compaction testing unsoaked LBR tests were conducted on 116 samples from all four RAP stockpiles. Unsoaked tests, which would produce higher LBR’s than soaked tests, were conducted because the vibratory and gyratory mold geometries prevented soaking. Modified  Proctor-­‐  LBR  Correlation   The unsoaked LBR values obtained from the modified Proctor compaction are presented as LBR versus dry density in Figure 4. A total of 32 tests were conducted: 9 with Melbourne crushed, 8 with Melbourne milled, 7 with Whitehurst milled, and 8 with Jacksonville crushed RAP.

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

8

25

Unsoaked LBR

20

y = 0.58x - 51.6 R² = 0.57

15 10 5 0 100

105

110

115

120

125

130

Dry Density (pcf)

Melbourne Crushed

Melbourne Milled

Whitehurst

Jacksonville

FIGURE 4 Unsoaked LBR versus Dry Density for Modified Proctor Compaction (1 pcf = 0.157 kN /m3). Unsoaked LBR values rose as dry density increased. The general trend among all data points indicated a linear relationship, which produced a regression coefficient of 0.57. LBR increased 0.58 per 1 pcf (0.157 kN/m3) increase in dry density. The highest unsoaked LBR of 24 from the modified Proctor was well below FDOT’s specified LBR of 100 for base course or of 40 for subbase material. The error typically associated with sample variation and LBR testing most likely produced the large scatter. Vibratory-­‐LBR  Correlation   The LBR values obtained from vibratory compaction are presented as LBR versus dry density in Figure 5. The vibratory LBR’s were all well below the FDOT specified subbase LBR of 40. A total of 32 tests were conducted: 10 with Melbourne crushed, 7 with Melbourne milled, 8 with Whitehurst milled, and 7 with Jacksonville crushed RAP. For the vibratory compaction method, LBR values slightly increased as dry density increased. The general trend from all points indicates a weak linear relationship with a regression coefficient of 0.17. LBR increased 0.24 per 1 pcf (0.157 kN/m3) increase in dry density, which is less than half the rate obtained from the modified Proctor.

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

9

40 35

Unsoaked LBR

30 25

y = 0.24x - 13.27 R² = 0.17

20 15 10 5 0 90

95

100

105

110

115

120

125

Dry Density (pcf) Melbourne, Crushed

Melbourne, Milled

Whitehurst, Milled

Jacksonville, Crushed

FIGURE 5 Unsoaked LBR versus Dry Density for Vibratory Compaction (1 pcf = 0.157 kN/m3). Gyratory-­‐LBR  Correlation   The LBR versus dry density results from the gyratory compaction samples are presented in Figure 6. A total of 52 tests were conducted: 12 with Melbourne crushed, 8 with Melbourne milled, 14 with Whitehurst milled, and 18 with Jacksonville crushed RAP. The LBR values from gyratory compaction were significantly higher than those achieved with the other compaction methods. LBR increased linearly 2.75 per 1 pcf (0.157 kN/m3) increase in dry density (Figure 6). This rate is much higher than the rates associated with either modified Proctor or vibratory compaction. The R2 of 0.62 is similar to the coefficient produced by the Modified Proctor testing.

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

10

120 100

y = 2.75x - 258.87 R² = 0.62

Unsoaked LBR

80 60 40 20 0 100

105

110

Melbourne, Crushed Whitehurst, Milled

115

120

125

130

Dry Density (pcf) Melbourne, Milled Jacksonville, Crushed

FIGURE 6 Unsoaked LBR versus Dry Density for Gyratory Compaction (1 pcf = 0.157 kN/m3). A graph of the LBR versus number of gyrations was prepared to determine whether the extra compactive effort from additional gyrations had any influence on the LBR (Figure 7). LBR increased linearly 0.41 per gyration, with a strong positive linear correlation coefficient of 0.70. Two outliers, one at 100 gyrations, and one at 150 gyrations, produced LBR values approximately 30 below the trend line. These outliers were considered to be from experimental error in the LBR test. Four specimens compacted at 150 gyrations, produced LBRs at or greater than 100. All samples compacted for 75 or more gyrations produced LBR values greater than 40. Five samples with densities over 125 pcf (19.6 kN/m3) produced unsoaked LBR values slightly greater than 100, while samples with densities above 116 pcf (18.2 kN/m3) produced LBR values over 40. Samples compacted for less than 25 gyrations did not achieve an LBR of 40. In conclusion, gyratory compaction of RAP produced the highest LBR’s of the threecompaction methods.

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

11

110 100

Unsoaked LBR

90 80

y = 0.41x + 36.74 R² = 0.70

70 60 50 40 30 0

25

Melbourne, Crushed

50

75 100 Number of Gyrations

Melbourne, Milled

125

Whitehurst, Milled

150

175

Jacksonville, Crushed

FIGURE 7 Unsoaked LBR versus Number of Gyrations for Gyratory Compaction of Samples.

COMPARING  GYRATORY  TO  MODIFIED  PROCTOR  STRENGTHS   A comparison of modified Proctor and gyratory strengths at the same densities was performed. Testing included LBR, UCC and IDT tests. Both UCC and IDT testing was on extruded samples.

LBR  Results   LBR versus density values from gyratory and modified Proctor tests are shown in Figure 8. Although both methods show an increase in LBR with density, gyratory compaction yielded significantly higher LBR values than the modified Proctor method at similar densities. FDOT’s State Materials Office (SMO) confirmed these results, by performing LBR tests on two modified Proctor-compacted samples, three gyratory compacted samples compacted with 75 gyrations and four gyratory compacted samples which matched the maximum modified Proctor density. As seen in Figure 8, gyratory compaction yielded LBR values several times higher than modified Proctor compaction at similar densities. The linear regression for the modified Proctorcompacted samples indicated a 0.58 increase in LBR for every 1.0 pcf (0.157 kN/m3) increase in dry density with a regression coefficient of 0.45 (medium correlation). The linear regression line for the gyratory samples indicated a much larger LBR increase of 2.71 for every 1.0 pcf (0.157 kN/m3) increase in dry density with a regression coefficient of 0.40 (medium correlation). The slopes and regression coefficients shown in Figure 8 differ from those shown in Figures 4 and 6 due to the additional data provided by FDOT SMO.

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

12

120

Gyratory y = 2.65x - 245.63 R² = 0.58

100

Unsoaked LBR

80 60

Modified Proctor y = 0.58x - 51.28 R² = 0.45

40 20 0 100

105

110

115

120

125

130

135

Dry Density (pcf) FIT Gyratory

FIT Mod. Proctor

FDOT Gyratory

FDOT Mod Proctor

FIGURE 8 Unsoaked LBR versus Dry Density Comparison between Modified Proctor and Gyratory Compaction for all RAP Stockpiles (1 pcf = 0.157 kN/m3).

UCC  Results   UCC tests were performed on samples extruded from both modified Proctor and gyratory molds. ASTM standards specify a 2:1 length to diameter ratio to minimize end effects, however the size of the modified Proctor and gyratory molds prevented this. Both the modified Proctor and gyratory samples were approximately 4.5-inches (114.3 mm) tall by 6-inches (152.4 mm) in diameter. Since both molds had the same relative proportions, the results were comparable. The peak value on the stress versus deflection plot was used as the UCC strength. Five specimens were compacted using modified Proctor energy, four with the gyratory energy at 75 gyrations, and three with gyratory energy heights set to match the modified Proctor densities. UCC strength versus dry density results are shown in Figure 9. Linear trend lines were fitted to the data. The regression coefficients are not shown because of the limited number of data points. Similar to the strength variations found from LBR tests, gyratory specimens yielded UCC strengths three to four times higher than the modified Proctor specimens and dry density increased as the number of gyrations increased. Figure 9 shows the UCC strength of the gyratory compacted specimens increased linearly from about 25 to 105 psi (172 to 723 kPa) as density increased, while the UCC strength of the modified Proctor specimens remained constant at about 20 psi (138 kPa).

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

13

140

UCC Strength (psi)

120

Gyratory y = 7.37x - 807.1

100 80 60 40

Modified Proctor y = -1.38x + 186.5

20 0 112

114

116

118

120

122

124

126

Dry Density (pcf) Gyratory

Modified Proctor

FIGURE 9 UCC Strength versus Dry Density from Modified Proctor and Gyratory RAP Specimens (1 psi = 6.985 kPa, 1 pcf = 0.157 kN/m3).

IDT  Results   IDT tests were also performed on specimens extruded from modified Proctor and gyratory molds. IDT versus dry density results are shown in Figure 10. A total of 10 tests were conducted, four compacted using modified Proctor, three compacted at 75 gyrations, and three compacted with the gyratory compactor to match the modified Proctor densities. As was the case for both the LBR and UCC, gyratory testing yielded IDT strengths two to three higher than modified Proctor specimens. Some modified Proctor specimens crumbled during mold extraction, producing zero tensile strength. IDT strengths of specimens compacted by both methods increased linearly as density increased.

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

14

2 1.8 y = 0.10x - 11.04

IDT Strength (psi)

1.6 1.4 1.2 1 0.8

`  

0.6 0.4

y = 0.08x - 8.73

0.2 0 114

116

Gyratory

118

120 122 Dry Density (pcf)

124

126

Modified Proctor

FIGURE 10 IDT Strength versus Density from Modified Proctor and Gyratory RAP Specimens (1 psi = 6.985 kPa, 1 pcf = 0.157 kN/m3).

Strength  Comparison  Summary   It was hypothesized, that the constant pressure and kneading action during gyratory compaction restored some of the asphalt bonding to produce these large increases in RAP strength. For comparison, LBR and IDT tests were performed on three readily available pavement materials, without asphalt. Three samples of limerock (LR), cemented coquina (CC), and clayey sand (CS) were prepared for each test. Figure 11 shows the ratio of strengths between the gyratory and the modified Proctor compaction. Additional test results from 50/50 blends of Melbourne milled RAP and Limerock are included. This figure shows that the strength ratio decreases from the upper levels near 4 to about 2 with the blends and finally to 1 for conventional soils. These ratios are near 1 indicating that gyratory compaction does not improve strengths of these conventional soils.

TRB 2013 Annual Meeting

Paper revised from original submittal.

Gyratory  to  Modified  Proctor  Strength  Ra4o  

Cosentino, Bleakley, Sajjadi, and Petersen

15

4.5  

LBR  

4  

IDT  

3.5  

UCC  

3   2.5   2   1.5   1   0.5   0   100%  RAP   50%  RAP/50%   100%  LR   LR  

100%  CC  

100%  SC  

Material  

FIGURE 11 Ratio of Gyratory to Modified Proctor Compaction for RAP and Conventional Soils.

UNCONFINED  CREEP  BEHAVIOR  OF  GYRATORY  AND  MODIFIED  PROCTOR   SPECIMENS     Unconfined creep tests were conducted on ejected gyratory and Proctor specimens to determine whether gyratory compaction affected creep performance as well as strength. Unconfined specimens were tested to eliminate any effects resulting from the different geometry of the Proctor and gyratory molds. The 4-inch (101.6 mm) diameter, 4.584-inch (116.43 mm) high modified Proctor specimens were prepared first. Gyratory specimens with slightly different diameters of 3.94-inches (100 mm) and heights of 4.72-inches (120 mm) were then prepared (about 0.15% difference). Tests were conducted on limerock and Melbourne milled RAP. Data, the average from two tests on each specimen, are presented in terms of displacements versus the log time in order to allow the slope of the plots (Δε/Δ log(t)), termed the creep strain rate (CSR), to be used for evaluation purposes. Creep  of  Limerock     Although extensive confined creep testing on conventional aggregates produced undetectably low levels of creep (9), these unconfined specimens showed extremely low but measurable creep. Figure 12 shows the average modified Proctor and gyratory results, each from three specimens. Although different initial displacements occurred, they had the same CSR (Figure 12). The equations are in the form of y = mx +b, where the slope, m, is the CSR and the intercept, b, represents the displacement at 1.0 days. This displacement includes sample seating and initial elastic movements. The initial displacement differences in Figure 12 were attributed to specimen seating. CSR demonstrates that the two compaction methods resulted in the same rate of creep. This result is consistent with the previous strength testing which found essentially no difference between gyratory and modified Proctor specimens of materials without asphalt.

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

16

0.0018   0.0016  

Strain  (in/in)  

0.0014   0.0012  

ε  =  0.00002log(t)  +  0.0015  

0.001   0.0008   0.0006   0.0004  

ε  =  0.00002log(t)  +  0.0002  

0.0002   0   0.01  

0.10  

1.00  

10.00  

Time  (days)   Average  Modified  Proctor  

Average  Gyratory  

FIGURE 12 Average Gyratory and Modified Proctor Unconfined Creep Displacement versus Log Time Results for Limerock. Creep  of  Melbourne  Milled  RAP   To clarify the creep of RAP, the gyratory and modified Proctor results from RAP and 50/50 blends of RAP and limerock were compared (Figure 13). All four data sets produced similar initial strains. The initial slopes for RAP data sets were also very similar (Figure 13). Modified Proctor specimens produced about 50 % more creep strain at seven days (Figure 13). Both sets of RAP data produced tertiary creep or the beginning of failure, near 0.4 % strain (i.e., 0.02-inch (0.51 mm) displacement/ 4.584-inch (116.4mm) sample thickness) or 1 day. Once tertiary creep started for the modified Proctor RAP specimens, the CSR became nonlinear on the log time plot. No regression equations were included since the RAP specimens exhibited the tertiary creep. The onset of tertiary creep for RAP at a low 0.4% axial strain was from the worst-case unconfined loading condition. Actual base courses have some confinement so creep rates in roadway applications would be less than these unconfined creep values, but more than those observed in the one-dimensional testing by Cosentino et al, (9).

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

17

1.20%  

Strain  (in/in)    

1.00%   0.80%   0.60%   0.40%   0.20%   0.00%   0.01  

0.1  

Average  Gyratory  MRAP   50/50  RAP  Limeorck  Proctor  

Time  (days)  

1  

10  

Average  Modified  Proctor  MRAP   50/50  RAP  Limerock  Gyratory  

FIGURE 13 Gyratory and Proctor Strain versus Log Time from Unconfined Creep of MRAP. Summary  of  Unconfined  Creep   Similar to strength testing results, gyratory compaction had a noticeable effect on deflection and CSR for the 100% RAP, a much less noticeable effect on the 50%/50% blend, and no effect on 100% limerock. The observed differences for RAP were not as pronounced as those from strength tests. The most significant result of these tests was that blending had a much greater impact on deflection and CSR than the compaction method. Both the gyratory and modified Proctor RAP specimens entered tertiary creep after approximately one day of unconfined 12 psi (83 kPa) stress. The improvements in CSR and total deflection from blending are more evident when the different blends are plotted on the same graph.

CONCLUSIONS     Vibratory compaction, which did not work well with RAP, consistently producing lower densities and strengths than either modified Proctor or gyratory compaction. Modified Proctor compaction produced both higher densities and LBR values than vibratory compaction, but lower densities and LBR strength than gyratory compaction. Gyratory compaction of RAP produced much higher strengths than the modified Proctor or vibratory compaction methods. Based on the linear correlations presented, the LBR from gyratory compaction was three to four times higher at the same density than samples compacted with modified Proctor. No significant LBR differences were observed between modified Proctor and gyratory compacted specimens of limerock, cemented coquina or clayey sand. Modified Proctor compaction of RAP yielded unsoaked LBR values less than 40. Gyratory compaction at 75 gyrations consistently produced unsoaked LBR values greater than 40. Compacting RAP with 150 gyrations yielded unsoaked LBR values of 100 or greater in four of nine specimens. It may be possible to achieve soaked LBR’s greater than 40 with RAP and allow its use as a subbase in the pavement system

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

18

RAP compacted using Gyratory compaction achieved 3 to 4 times more UCC and IDT strength than RAP compacted by the modified Proctor method at similar densities. Based on the consistent results from LBR, UCC and IDT testing, it was concluded that the constant pressure and kneading action of the gyratory compaction reestablished some of the adhesion in RAPs asphalt binder and may allow its use in more areas of the pavement system. Gyratory compaction of RAP and RAP/aggregate blends produced less creep than modified Proctor compaction at the same density; however, the difference in CSR, or amount of creep, was not as pronounced as the differences in strength previously noted. There was a 50 % reduction in creep between RAP and the 50/50 RAP limerock blends. To control creep of RAP it must be blended with a high quality base material.

RECOMMENDATIONS   To verify these findings, large-scale field compaction sites with RAP and RAP blends should be developed to evaluate the strength and creep differences between RAP and RAP blends. These materials should be compacted using static, vibratory and pneumatic rollers. Compaction trains consisting of static and pneumatic and/or vibratory and pneumatic rollers should be used in addition to the individual pieces of equipment. The results from this work would lead to construction specifications that would produce more uses for RAP and RAP blends.

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

19

ACKNOWLEDGEMENTS   Our sincere thanks to Dr. David Horhota, Dan Pitocchi, David Webb, Jose Hernando and the remaining FDOT SMO laboratory technicians that helped complete this work. We also thank the dedicated work force from the Florida Institute of Technology of Thaddeus J Misilo III, Babacar Diouf, Steve Craig, Etienne Wolmarans, and R. Quincy Sy.

TRB 2013 Annual Meeting

Paper revised from original submittal.

Cosentino, Bleakley, Sajjadi, and Petersen

20

REFERENCES   1. Florida Department of Transportation (FDOT). Standard Specifications for Road and Bridge Construction. SMO, Gainesville, Florida, 2010. 2. Sandin, C. Laboratory Evaluation of the Variability of Florida’s RAP Materials for Use in Earthwork Applications. M.S. Thesis, Florida Institute of Technology. 2008. 3. Montemayor, T. A. Compaction and Strength-Deformation Characteristics of Reclaimed Asphalt Pavement. M.S. Thesis, Florida Institute of Technology. 1998. 4. Cosentino, P. J. and E. H. Kalajian. Developing Specifications for Using Recycled Asphalt Pavement as Base, Subbase, or General Fill Materials. Final Report. Contract BB-892. FDOT SMO. 2001. 5. Ping, W. V., M. Leonard, and Z. Yang. Laboratory Simulation of Field Compaction Characteristics (Phase I). Final Report. FDOT SMO. Final Report No.: FL/DOT/RMC/BB-890 (F), 2003. 6. Ping, W. V., M. Leonard, and Z. Yang. Evaluation of Laboratory Compaction Techniques for Simulating Field Soil Compaction (Phase II). Final Report. FDOT SMO. Final Report No.: FL/DOT/RMC/BB-890(F), 2003. 7. Zhang, J. Evaluation of Mechanistic-Based Compaction Measurements for Earthwork QC/QA. M.S. Thesis, Iowa State University. 2010. 8. Lambe T. W. and R. V. Whitman. Soil Mechanics, John Wiley & Sons, 1969. 9. Cosentino, P. J., E. H. Kalajian. A. M. Bleakley, B. S. Diouf, T. J. Misilo III, A. J. Petersen, R. E. Krajcik, and A. M. Sajjadi. Improving the Properties of Reclaimed Asphalt Pavement for Roadway Base Applications, FDOT Final Report: Contract BDK81 Work Order 977-02, August, 2012.

TRB 2013 Annual Meeting

Paper revised from original submittal.