Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete (2017)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

21 Dynamic modulus testing was conducted on small-scale cylindrical specimens while uncon- fined with axially applied loading using an AMPT in general accordance with AASHTO TP 79. Modifications to the specification included using a reduced set of temperatures (4.4°C, 21.1°C, and 37.8°C) and using small-scale cylindrical specimens. At each temperature, testing was con- ducted at loading frequencies of 25 Hz, 10 Hz, 5 Hz, 1 Hz, 0.5 Hz, and 0.1 Hz. To minimize the potential for damage, each specimen was reused throughout the testing regime and tested in an increasing order of temperature and a decreasing order of frequency at each temperature. 4.1 Outlier Analysis The results of dynamic modulus testing of the recycled materials were examined for outliers by visually inspecting the modulus values at a test frequency of 10 Hz at the three test tempera- tures (4.4°C, 21.1°C, and 37.8°C). Figure 12 shows an example of the data at 10 Hz and 21.1°C, grouped by project type. Figure 12 marks the dynamic modulus result from one test specimen. Visual observation of the dynamic modulus test results identified several data points that appeared to vary significantly from the results of similar project types. It is difficult to determine if these cases of data variabil- ity were caused by inherent variability in the construction processes or by a low quality test or specimen. Previous research has shown that dynamic modulus test results for recycled mixtures can be more variable than the results typically seen for asphalt mixtures (Diefenderfer and Link 2014). Also, the AASHTO TP 79 specification data quality indicators were developed for asphalt mixtures, and it is not yet clear if they are appropriate for recycled mixtures. The authors considered several methods by which to eliminate potentially abnormal data. First, the data were compared with the dynamic modulus data quality indicators suggested in AASHTO TP 79. This analysis showed that the deformation uniformity was the most commonly violated quality indicator; however, many of the results fell within the recommended ranges. The deformation uniformity parameter describes the variation in deformation between the multiple LVDTs mounted on the specimen (Bonaquist et al. 2003; Von Quintus et al. 2012). Bonaquist (2008) suggested that deformation uniformity values beyond the recommended ranges may be influenced by non-uniformity of the specimen. However, it was found that some data points that had very low deformation uniformity values (e.g., less than 10%) also appeared to be out- side the expected ranges. In addition, some data points had large deformation uniformity values (e.g., greater than 75%), but the dynamic modulus COV of specimens from the same project was quite low (e.g., also less than 10%). Because of these issues, using data quality indicators to identify potentially abnormal values was not pursued further. The specimen bulk density also was investigated to see if it played a role in the variability of the dynamic modulus test results. As C H A P T E R 4 Dynamic Modulus Test Results

22 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete with the data quality indicators, no consistent trend was apparent with respect to bulk density and dynamic modulus. Next, an outlier procedure was performed using a quartile analysis. For each stabilizing/ recycling agent and chemical additive type, the difference between the calculated third and first quartile was determined and identified as the interquartile range (IQR). The IQR was added to the third quartile value and subtracted from the first quartile value to calculate an upper and lower fence from which to consider outliers. This process identified most of the same outliers as were identified by visual observation. The quartile analysis was conducted independently for the 10 Hz data at the three test temperatures; the resulting data were termed trimmed data. Figure 13 shows an example of the trimmed data at 10 Hz and 21.1°C. 4.2 Significance Testing of Dynamic Modulus Data at 10 Hz Using the trimmed data, a two-tailed, two-sample Student’s t-test assuming unequal variance was used to determine if differences in the mean dynamic modulus with respect to recycling processes, stabilizing/recycling agents, and chemical additive combinations were statistically sig- nificant. The t-test for each comparison was performed separately at each test temperature using data from a test frequency of 10 Hz and considered a significance level (a) of 0.05. Table 3 and Table 4 show the results of significance testing for recycling processes and stabilizing/recycling agent combinations, respectively. From Table 3, it can be seen that the differ- ences in the mean dynamic modulus values of the different recycling processes are not significant at 4.4°C. At 21.1°C and 37.8°C, however, the difference in the value of the mean dynamic modulus for CCPR versus FDR is significant at both temperatures. Considering CCPR versus CIR, the dif- ference in the mean dynamic modulus values was found to be significant only at 21.1°C; and the difference in the mean dynamic modulus considering CIR versus FDR was found to be significant only at 37.8°C. Generally, the differences become more significant with increasing temperature. E = emulsified asphalt, F = foamed asphalt, C = hydraulic cement, and L = lime. Figure 12. Dynamic modulus results at 10 Hz and 21.1çC.

Dynamic Modulus Test Results 23 Data points identified as outside the IQR are shown with no shading. E = emulsified asphalt, F = foamed asphalt, C = hydraulic cement, and L = lime. Figure 13. Trimmed dynamic modulus data at 10 Hz and 21.1çC. a) 4.4°C CIR FDR CCPR 0.8764 0.8659 CIR 0.2085 b) 21.1°C CIR FDR CCPR 0.0075 0.0018 CIR 0.1989 c) 37.8°C CIR FDR CCPR 0.0929 0.0000 CIR 0.0000 Shading highlights significant differences. Table 3. Recycling process statistical comparisons at (a) 4.4çC, (b) 21.1çC, and (c) 37.8çC. Table 4 shows that the difference in the mean dynamic modulus considering emulsified asphalt versus foamed asphalt was not significant except for the mixtures at 4.4°C temperature using FDR. When considering the presence of cement in addition to the bituminous stabilizing/recycling agents, three of the nine comparisons using emulsified asphalt (three processes at three temperatures) and none of the three comparisons using foamed asphalt (CIR at three temperatures) showed a signifi- cant difference in the dynamic modulus. The difference in dynamic modulus values for mixtures using cement as a chemical additive in mixtures stabilized/recycled with emulsified asphalt were found to be significant only for CIR at 21.1°C and 37.8°C and for CCPR at 21.1°C. When con- sidering the presence of lime in addition to emulsified asphalt for CIR, the difference in the mean dynamic modulus was found to be statistically significant for the 21.1°C and 37.8°C temperatures. Table 4 shows that the difference in the mean dynamic modulus when comparing the com- bination of emulsified asphalt plus cement versus foamed asphalt plus cement was significant only for one of the six comparisons (FDR and CIR at three temperatures): FDR at a temperature

24 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete of 4.4°C. The difference in the mean dynamic modulus for foamed asphalt CIR versus foamed asphalt CIR plus cement was found not to be statistically significant at any of the three tem- peratures considered. Table 4 shows that the difference in the mean dynamic modulus for all stabilizing/recycling agents without cement versus with cement was statistically significant for five of the nine com- parisons: CCPR at 21.1°C, CIR at 21.1°C and 37.8°C, and FDR at 4.4°C and 37.8°C. For the three CIR comparisons considering all stabilizing/recycling agents with cement versus all stabilizing/ recycling agents with lime, the differences in the mean dynamic modulus values were found to be statistically significant only at 4.4°C. In summary, the main systematic and significant difference in dynamic modulus values identi- fied was between mixtures that used a stabilizing/recycling agent with a chemical additive versus mixtures without a chemical additive. This difference was more apparent at higher temperatures. No discernible trends were identified between the use of emulsified asphalt or foamed asphalt as stabilizing/recycling agents; the only statistically significant difference in dynamic modulus values was found at the 4.4°C test temperature for FDR. Also, no discernible trend was identi- fied between the use of hydraulic cement or lime as the chemical additive; the only statistically a) 4.4°C Emulsion vs. Foam Emulsion vs. Emulsion + Cement Emulsion vs. Emulsion + Lime Emulsion + Cement vs. Foam + Cement Foam vs. Foam + Cement Cement vs. No Cement Cement vs. Lime CCPR - 0.2873 - - - 0.2873 - CIR 0.7285 0.1862 0.2422 0.7151 0.8320 0.2055 0.0042 FDR 0.0016 0.4958 - 0.0016 - 0.0422 - b) 21.1°C Emulsion vs. Foam Emulsion vs. Emulsion + Cement Emulsion vs. Emulsion + Lime Emulsion + Cement vs. Foam + Cement Foam vs. Foam + Cement Cement vs. No Cement Cement vs. Lime CCPR - 0.0203 - - - 0.0203 - CIR 0.7356 0.0108 0.0000 0.1496 0.0804 0.0039 0.1970 FDR 0.7105 0.7840 - 0.7105 - 0.9558 - c) 37.8°C Emulsion vs. Foam Emulsion vs. Emulsion + Cement Emulsion vs. Emulsion + Lime Emulsion + Cement vs. Foam + Cement Foam vs. Foam + Cement Cement vs. No Cement Cement vs. Lime CCPR - 0.0513 - - - 0.0513 - CIR 0.0641 0.0049 0.0068 0.2183 0.3671 0.0036 0.0582 FDR 0.9938 0.1466 - 0.9938 - 0.0095 - Shading highlights significant differences. Table 4. Stabilizing/recycling agent statistical combinations comparisons at (a) 4.4çC, (b) 21.1çC, and (c) 37.8çC.

Dynamic Modulus Test Results 25 significant difference in dynamic modulus values was found at the 4.4°C test temperature (only the CIR process had projects using both hydraulic cement and lime as chemical additives). 4.3 Master Curve Analysis To study the influence of the recycling process and the stabilizing/recycling agent and chemical additive combinations on the dynamic modulus over the complete temperature/frequency spec- trum, the measured dynamic modulus data for each specimen (after applying the outlier analysis) were averaged for each project. These averages were used to create a master curve at a reference tem- perature of 21.1°C. Figure 14 shows an example of a dynamic modulus master curve created from the average measured dynamic modulus data at three test temperatures (4.4°C, 21.1°C, and 37.8°C) from a CIR project using foamed asphalt from San Jose, California. Generally, the master curves of the cold-recycled materials followed a sinusoidal shape as is seen for typical HMA mixtures. From the master curves, data envelopes (bounded by the maximum and minimum aver- age dynamic modulus data) were developed to compare project and material types by visual observation. The data envelopes were developed because the statistical analysis only indicates the existence of a difference between project or material types, not the direction of the dif- ference. The master curves for similar project or material types were grouped together. The groupings included recycling process, stabilizing/recycling agent, and presence of chemical additive. Figure 15 and Figure 16 show the dynamic modulus data envelopes for the CIR, CCPR, and FDR processes without discriminating between the presence of a chemical additive or the type of stabilizing/recycling agent (foamed or emulsified asphalt). Figure 15 shows the data using a log- linear plot that emphasizes the data at higher reduced frequencies (corresponding to lower test temperatures). Figure 16 shows the data using a log-log plot that emphasizes the data at lower reduced frequencies (corresponding to higher test temperatures). Master curve reflects averages of measurements from three specimens. Figure 14. Dynamic modulus master curve from CIR project using foamed asphalt (San Jose, California, Project 13-1124).

26 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete From Figure 15 and Figure 16 it can be seen that the FDR mixtures had a greater stiffness than the CIR and CCPR mixtures at lower reduced frequencies (less than approximately 10 Hz), which corresponds to higher test temperatures. The stiffness of the CIR mixtures overlaps the lower portion of the FDR envelope at frequencies less than approximately 10 Hz but then exceeds the stiffness of the FDR mixtures at higher reduced frequencies (greater than approximately 10 Hz), which correspond to lower test temperatures. The stiffness of the CCPR mixtures also overlaps the FDR envelope throughout the range of reduced frequencies; however, the CCPR mixtures 0 200,000 400,000 600,000 800,000 1,000,000 1,200,000 1,400,000 1,600,000 0 2,000 4,000 6,000 8,000 10,000 12,000 0.0001 0.01 1 100 10000 D yn am ic M od ul us (p si ) D yn am ic M od ul us (M Pa ) Reduced Frequency (Hz) CCPR CIR FDR Figure 15. Dynamic modulus master curve data envelopes for mixtures produced by FDR, CIR, and CCPR using log-linear scale. 1,450 14,500 145,000 1,450,000 14,500,000 10 100 1,000 10,000 100,000 0.0001 0.01 1 100 10000 D yn am ic M od ul us (p si ) D yn am ic M od ul us (M Pa ) Reduced Frequency (Hz) CCPR CIR FDR Figure 16. Dynamic modulus master curve data envelopes for mixtures produced by FDR, CIR, and CCPR using log-log scale.

Dynamic Modulus Test Results 27 data envelope minimum values are less than the minimum values of the other recycling process types throughout the range of reduced frequencies. Most highway agencies and design procedures typically assign a lesser layer coefficient (or less modulus) to FDR; therefore, it may be surprising that the FDR data envelopes show a greater stiffness than either CIR or CCPR. In an effort to compare the performance of in-place recycled mixes to traditional asphalt base mixes, cores from three in-service base mix projects from Virginia were collected. The base mixes had a NMAS of 25.0 mm and were tested under the same conditions of the in-place recycled mixes. Figure 17 again shows the dynamic modulus envelopes of the various in-place recycled mixes but adds the envelope for the three base mixtures. At low frequencies (analogous to higher temperatures) the base mix has similar modulus values to those of mixtures produced by CIR and CCPR, whereas at lower-middle frequencies the base mix is more comparable to the FDR mixture. However, at high frequencies (analogous to lower temperatures) the base mix exhibits higher moduli than any of the recycled mixes. Figure 18 and Figure 19 show the dynamic modulus data envelopes for the recycled mixtures with respect to stabilizing/recycling agents without discriminating between the recycling process or the presence of a chemical additive. It can be seen that there is much overlap in the data for the two stabilizing/recycling agent types. Mixtures using foamed asphalt were found to be stiffer at lower reduced frequencies (less than approximately 20 Hz), corresponding to higher test tem- peratures, whereas mixtures using emulsified asphalt were found to be stiffer at higher reduced frequencies (greater than approximately 20 Hz), corresponding to higher test temperatures. Figure 20 and Figure 21 show the dynamic modulus data envelopes for the recycled mixtures with respect to the presence of a chemical additive; the data envelopes are grouped by mixtures containing no chemical additive, lime, or cement regardless of stabilizing/recycling agent or recy- cling process. (Only the CIR process recycled with emulsified asphalt contained lime as a chemical additive.) The dynamic modulus data envelope for the mixtures that included cement overlaps the envelopes for the mixtures containing lime and no chemical additive at low and intermedi- ate reduced frequencies (less than approximately 10 Hz and 100 Hz, respectively). At these same Figure 17. Dynamic modulus master curve data envelopes for mixtures produced by FDR, CIR, and CCPR, as well as asphalt base mixes, using log-log scale.

28 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete 0 200,000 400,000 600,000 800,000 1,000,000 1,200,000 1,400,000 1,600,000 0 2,000 4,000 6,000 8,000 10,000 12,000 0.0001 0.01 1 100 10000 D yn am ic M od ul us (p si ) D yn am ic M od ul us (M Pa ) Reduced Frequency (Hz) Emulsified Asphalt Foamed Asphalt Figure 18. Dynamic modulus master curve data envelopes for mixtures using emulsified asphalt and foamed asphalt as stabilizing/recycling agents using log-linear scale. 1,450 14,500 145,000 1,450,000 14,500,000 10 100 1,000 10,000 100,000 0.0001 0.01 1 100 10000 D yn am ic M od ul us (p si ) D yn am ic M od ul us (M Pa ) Reduced Frequency (Hz) Emulsified Asphalt Foamed Asphalt Figure 19. Dynamic modulus master curve data envelopes for mixtures using emulsified asphalt and foamed asphalt as stabilizing/recycling agents using log-log scale.

Dynamic Modulus Test Results 29 0 200,000 400,000 600,000 800,000 1,000,000 1,200,000 1,400,000 1,600,000 0 2,000 4,000 6,000 8,000 10,000 12,000 0.0001 0.01 1 100 10000 D yn am ic M od ul us (p si ) D yn am ic M od ul us (M Pa ) Reduced Frequency (Hz) No Chemical Additive Lime Cement Figure 20. Dynamic modulus master curve data envelopes for mixtures with no chemical additive, with lime, and with cement, using log-linear scale. 1,450 14,500 145,000 1,450,000 14,500,000 10 100 1,000 10,000 100,000 0.0001 0.01 1 100 10000 D yn am ic M od ul us (p si ) D yn am ic M od ul us (M Pa ) Reduced Frequency (Hz) No Chemical Additive Lime Cement Figure 21. Dynamic modulus master curve data envelopes for mixtures with no chemical additive, with lime, and with cement using log-log scale.

30 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete reduced frequency ranges, mixtures containing lime were found to be stiffer than mixtures hav- ing no chemical additive. At higher reduced frequencies (greater than approximately 100 Hz), however, the stiffness of the mixtures with no chemical additive overlapped with the stiffness of those mixtures containing lime and also were generally found to be stiffer than the mixtures containing cement. Mixtures containing cement may be stiffer at lower reduced frequencies (higher test tempera- tures) because the cement has hydrated and produced bonds that are stiffer than the asphalt binder at higher temperatures. As the binder in the mixture from the RAP and/or the stabilizing agent softens, the presence of a stiff hydration product would be noticeable. Like those contain- ing cement, mixtures containing lime also seem to stiffen at lower temperatures as compared to mixtures that contain no chemical additive. The effect is less pronounced with lime than it is with cement. The master curve envelopes also show that mixtures containing a chemical additive are generally less temperature dependent than those containing no chemical additive. The lower temperature dependency for mixtures containing a chemical additive may be caused by the presence of a non-viscoelastic material as part of the stabilizing mechanism. It is also worth noting that a wider range of stiffness values occurs for the recycled materials that contain cement as an additive. This result could reflect the inclusion of all project types and recycling processes in the stiffness comparison. The increased stiffness at lower reduced frequencies for mixtures containing cement is also noteworthy in that many design procedures report that the chemical additives are available for recycling agent dispersion or improved strength at early ages. Given that these materials were tested at 12–24 months after construction, it is clear that this strength increase continues beyond the initial age of the recycled layer. 4.4 Analysis of Fitting Parameters The four fitting parameters from Equation 2 (a, b, d, and g) can be used to reconstruct any master curve. Table 5 shows the fitting parameters describing the average master curve for each project site. A statistical analysis of the fitting parameters was performed to quantify any observed trends from within the data. A two-tailed, two-sample Student’s t-test assuming unequal variance was used to determine if differences in the fitting parameters with respect to recycling processes, stabilizing/recycling agents, and chemical additive combinations were statistically significant at a significance level (a) of 0.05. Table 6 shows the results of the statistical testing of the master curve shape parameters for the three recycling processes considered. In Table 6, the effects of the different stabilizing/recycling agents and recycling agent/chemical additive combinations are included with averaging for each process. From the comparison, the difference in the fitting parameters for any of the recycling processes was not found to be statistically significant. Table 7 shows the results of statistical testing of the master curve shape parameters for the different stabilizing/recycling agents and recycling agent/chemical additive combinations. Not all combinations were considered because in some instances an insufficient number of projects were available for a particular condition, and thus the t-test could not be performed. The results in Table 7 show that the differences in the beta (b) and delta (d) parameters for CIR projects including emulsion versus those containing emulsion plus cement were statistically significant. The b parameter indicates that the difference in horizontal position of the turning point shown in Figure 11 is statistically significant. Also, the d parameter indicates that the difference in mini- mum modulus value is statistically significant.

Location Project ID Alpha (α) Beta ( ) Delta ( ) Gamma ( ) Kansas 13-1093 4.3194 -1.5444 1.9507 -0.2626 Ontario 13-1111 4.3670 -1.4334 1.9662 -0.3523 Ontario 13-1112 4.7391 -1.4374 1.6832 -0.3480 Ontario 13-1113 4.3712 -1.3721 1.9571 -0.3465 Ontario 13-1114 3.5732 -0.9951 2.7638 -0.4074 Alberta 13-1115 0.6706 1.5158 4.9562 -1.9025 Alberta 13-1116 4.2084 -1.8426 1.8577 -0.3202 Alberta 13-1117 4.1272 -1.6234 1.7471 -0.3104 California (San Jose) 13-1124 4.2913 -1.8334 1.9469 -0.3212 Colorado 13-1127 4.2620 -1.9634 1.9202 -0.3105 California (Los Angeles) 14-1001 4.3253 -1.7761 1.8988 -0.3072 California (Los Angeles) 14-1002 4.1005 -2.0384 1.7765 -0.3895 California (Los Angeles) 14-1003 4.3227 -1.7923 1.9877 -0.3247 West Virginia 14-1011 4.2673 -1.9638 1.9110 -0.2650 Delaware 14-1025 3.5213 -0.8741 2.8475 -0.4030 Delaware 14-1026 4.3564 -1.1814 2.1968 -0.3505 Delaware 14-1027 3.4839 -0.9139 2.9062 -0.4087 Delaware 14-1028 4.5154 -1.2669 2.0818 -0.2449 Utah 14-1055 4.4060 -1.4641 2.0232 -0.3092 Georgia 14-1057 4.3148 -2.1788 1.9247 -0.2066 Washington State 14-1058 4.3487 -1.8578 2.0427 -0.2781 Colorado 14-1062 4.2884 -1.6808 1.9354 -0.2670 Maine (Lyman) 15-1001 4.3257 -1.1786 1.8657 -0.3054 Maine (Corinna, Exeter) 15-1002 4.3500 -1.2693 1.9350 -0.2535 Table 5. Dynamic modulus master curve fitting parameters.

32 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete Table 7 also shows that, for CIR mixtures using emulsion versus emulsion plus lime, the dif- ference in the g parameter, representing the rate of change between the minimum and maximum modulus values, was statistically significant. Like the analysis of the dynamic modulus values, the analysis of the fitting parameters shows that differences in foamed asphalt versus emulsified asphalt were not statistically significant. Also, the differences in the fitting parameters for CIR projects using lime versus cement as a chemical additive were not statistically significant. 4.5 Analysis of Phase Angle Black Space diagrams are useful in showing the relationship between stiffness and phase angle as measured during the dynamic modulus test. The phase angle represents the lag between the applied stress and the resultant strain in the specimen during the dynamic modulus test. A phase angle of zero degrees implies a purely elastic material and a phase angle of 90° corresponds to a purely viscous material. Figure 22 shows a Black Space diagram for three base asphalt Process Comparison Fitting Parameter Alpha (α) Beta ( ) Delta ( ) Gamma ( ) CIR vs. CCPR 0.2411 0.5104 0.0805 0.1289 CIR vs. FDR 0.4414 0.8379 0.5104 0.4791 CCPR vs. FDR 0.3629 0.9763 0.3498 0.4039 Table 6. Dynamic modulus master curve fitting parameters statistical comparisons by process. Process Stabilizing/Recycling Agents and Recycling Agent/Chemical Additive Combinations Fitting Parameter Alpha (α) Beta ( ) Delta ( ) Gamma ( ) CIR Emulsion vs. Foam 0.1574 0.9676 0.1089 0.4491 Emulsion vs. Emulsion + Cement 0.9641 0.0031 0.0610 0.9890 Emulsion vs. Emulsion + Lime 0.1332 0.3517 0.2066 0.0448 Cement vs. Lime 0.2134 0.0713 0.1942 0.3854 FDR Emulsion vs. Foam 0.3481 0.7052 0.4708 0.3903 Shading highlights significant differences. Table 7. Dynamic modulus master curve fitting parameters statistical comparisons by stabilizing/recycling agents and recycling agent/chemical additive combinations.

Dynamic Modulus Test Results 33 mixtures produced in Virginia. The results show that the phase angle peaks at lower stiff- ness values and reaches a maximum value and then decreases. If labels for the test tem- peratures also appeared in Figure 22, it would be seen that the phase angle increases (and the modulus decreases) as the test temperature increases. The phase angles reach their maximum values at higher test temperatures because the asphalt mixtures exhibit more viscous behavior. At the highest test temperatures, the aggregate structure begins to play a larger role in the mate- rial behavior and the phase angle peaks and begins to decrease. The relationship between phase angle and stiffness for the various pavement recycling tech- niques and additives was investigated visually using a Black Space diagram. The Black Space diagram for a typical asphalt mixture shows a peak phase angle value (see the right side of Figure 22) that is caused by the behavioral influence of the aggregate structure at higher tem- peratures. At lower temperatures, the mixture volumetrics and binder stiffness control the behavior (Biligiri et al. 2010). The highest values of phase angles in this plot are associated with the inflection points in the master curve. It can be seen in Figure 22 that these occur around a stiffness of 1,000–1,200 MPa. Figure 23 shows the Black Space diagram for all recycled mixtures after removing outlier dynamic modulus values. While there is much overlap between the recycling types, it appears that the phase angle is generally the least for FDR and greatest for CIR; the phase angle for CCPR is between the other two recycling process types. This pattern indicates that the FDR mixtures tested generally had the most elastic response. This result was expected because the FDR mixtures contained a lower proportion of asphalt binder and they used a combination of RAP and unbound materials. Given the differences shown in Figure 23 between the recycling processes, further investigation was performed with respect to recycling agents and chemical additives. Figure 24 shows the Black Space diagram for emulsified asphalt and foamed asphalt CIR mix- tures. These mixtures contained no chemical additives. In Figure 24 it can be seen that the phase angle versus stiffness data overlap for the two recycling agents shown. This pattern suggests that CIR mixtures using emulsified asphalt or foamed asphalt should have similar viscoelastic behavior. Figure 22. Black Space diagram for three asphalt base mixtures produced in Virginia.

34 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete Figure 23. Black Space diagram for mixtures produced by FDR, CCPR, and CIR. Figure 24. Black Space diagram for emulsified asphalt and foamed asphalt CIR mixtures having no chemical additive.

Dynamic Modulus Test Results 35 Figure 25 shows the Black Space diagram for emulsified asphalt CIR mixtures having lime, cement, or no chemical additive. Mixtures including a chemical additive generally have a more elastic response in that their phase angles are smaller than those mixtures with no chemical additive. Figure 25 also shows that those mixtures using cement as a chemical additive gener- ally had a smaller phase angle than the mixtures using lime. This observation was expected because cement exhibits little viscoelastic behavior. Of particular interest in this figure is the lack of an inflection point in the mixtures containing a chemical additive. The inflection point is indicative of the transition point at which the aggregate begins to control the stiffness (i.e., the point at which the binder’s viscosity no longer controls the mixture’s stiffness). The lack of an inflection point in Figure 25 likely reflects the limited temperatures in the test procedure. If higher temperatures were evaluated, it is probable that an inflection point would also appear for the mixtures containing chemical additives. It is not clear if the lack of an inflection point in Figure 25 indicates that the recycled mixtures maintain a predominately elastic response at higher temperatures. Figure 26 shows the Black Space diagram for foamed asphalt CIR mixtures having cement or no chemical additive. (No foamed asphalt CIR mixtures having lime as a chemical additive were sampled.) The presence of cement reduces the phase angle in a manner similar to that shown for emulsified asphalt CIR mixtures and suggests a more elastic response. Figure 27 shows the Black Space diagram for emulsified and foamed asphalt FDR mixtures with cement as a chemical additive. It can be seen that the one emulsified asphalt FDR mixture with cement had slightly greater phase angle values than the foamed asphalt FDR mixtures with cement. This result may have been caused by the differences in stabilization mechanisms for emulsified mixtures versus foamed mixtures. Asphalt Academy (2009) states that emulsified mixtures are held together in a coating process whereas foamed mixtures utilize a spot-welding process to hold the recycled particles together. If this is true, the emulsified asphalt mixture may be expected to have a more viscous response if the interparticle bonding is more dependent on the coating process. Furthermore, this would create an increased surface area throughout the mixture in which a softer binder is present. If blending between the emulsified asphalt and the RAP binder occurs, a composite binder layer would form that is stiffer than the base binder Figure 25. Black Space diagram for emulsified asphalt CIR mixtures.

36 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete Figure 26. Black Space diagram for foamed asphalt CIR mixtures having cement or no chemical additive. Figure 27. Black Space diagram for emulsified and foamed asphalt FDR mixtures with cement.

Dynamic Modulus Test Results 37 used in the emulsion and softer than the RAP binder. In the foaming process, the spot-welding effect allows the RAP binder stiffness to control the mixture performance, hence the more elastic response. In the described emulsion-composite case, the composite layer would control, result- ing in a less stiff material. If a composite is not formed initially (i.e., if the RAP binder is simply coated with the softer emulsified binder), the molecular dynamics of binder blending over time would lead to a composite layer that might result in a stiffer mix overall. Unfortunately, the data set is not representative of the same mix tested over time, so this theory cannot be validated. Some interaction with the cement chemical additive also is likely. This variable could not be eliminated because all foamed asphalt FDR mixtures included cement as a chemical additive. Figure 28 shows the Black Space diagram for the two emulsified asphalt FDR mixtures in the study. One mixture has no chemical additive and the other includes cement. The presence of cement is noticeable in that the phase angle for the FDR mixture with cement is smaller than that of the mixture having no chemical additive, suggesting a more elastic behavior. 4.6 Relationship between Stiffness and Mixture Properties The density of a cold-recycled material is a commonly cited property that is included as a quality measure in many agency specifications (Stroup-Gardiner 2011). It is not known, how- ever, whether density is always a good predictor of future performance or if it is just an easily measured surrogate property. Table 8 shows the coefficient of correlation (R2) between the bulk density of the dynamic modulus specimen and the dynamic modulus result at 10 Hz for the three test temperatures. The results are grouped to show the correlation for all projects of a particular recycling process and also separated by recycling/stabilizing agent and chemical additive combi- nation. In Table 8, cells that contain a dash indicate that too few results were available for com- parison after outliers were removed. Table 8 shows that (1) in some cases, density and stiffness are highly correlated and (2) in general, the correlation improves with decreasing temperature. Figure 28. Black Space diagram for emulsified asphalt FDR mixtures having no chemical additive and no cement.

38 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete The R2 value was greater than 0.5 in only 7 of the 33 comparisons, however, indicating that the density describes more than 50% of the variability in the measured stiffness in only 21% of the comparisons. Figure 29 shows the relationship between density and stiffness with respect to recycling process. Table 9 shows the coefficient of correlation (R2) between the percentage of mix design den- sity of the dynamic modulus specimen and the dynamic modulus result at 10 Hz for the three test temperatures. As with Table 8, cells that contain dashes indicate that too few results were available for comparison after outliers were removed. Table 9 also shows that the percentage of mix design density and stiffness is highly correlated in certain cases and that the correla- tion generally increases with decreasing temperature. The R2 value was greater than 0.5 in only 9 of the 30 comparisons, indicating that the percentage of mix design density describes more than 50% of the variability in the measured stiffness in only 30% of the comparisons. Figure 30 shows the relationship between percentage mix design density and stiffness with respect to recycling process. A correlation analysis was performed between design properties and performance properties for the cold-recycled materials. Design properties include the mixture design characteristics and information collected at time of construction (e.g., construction equipment, weather conditions, compaction). A limited number of these properties were known across all of the cold-recycled material types. Performance properties are the dynamic modulus master curve parameters and dynamic modulus values at specific temperature and loading rate combinations. Process Recycling/ Stabilizing Agent Chemical Additive Test Temperature 4.4°C 21.1°C 37.8°C CIR All 0.48 0.04 0.04 Emulsion None 0.46 0.31 0.10 Emulsion Lime 0.55 0.47 0.33 Emulsion Cement 0.66 0.71 0.09 Foam None 0.87 0.69 0.84 Foam Cement 0.73 -- -- CCPR All 0.11 0.09 0.26 Emulsion None -- -- -- Emulsion Cement 0.13 0.09 0.01 FDR All 0.30 0.01 0.01 Emulsion None 0.10 0.36 -- Emulsion Cement 0.37 0.19 0.42 Foam Cement 0.34 0.01 0.10 Dashes indicate too few data were available for comparison after outliers removed. Table 8. Coefficient of correlation (R2) between bulk density and dynamic modulus at 10 Hz.

Dynamic Modulus Test Results 39 Figure 29. Relationship between specimen bulk density and stiffness. Process Recycling/ Stabilizing Agent Chemical Additive R2 at Test Temperature 4.4°C 21.1°C 37.8°C CIR All 0.51 0.12 0.03 Emulsion None 0.36 0.21 0.04 Emulsion Lime 0.49 0.06 0.02 Emulsion Cement 0.66 0.64 0.09 Foam None 0.96 0.88 0.94 Foam Cement 0.04 -- -- CCPR All 0.07 0.66 0.10 Emulsion None -- -- -- Emulsion Cement 0.14 0.66 0.80 FDR All 0.25 < 0.01 < 0.01 Emulsion None 0.10 0.36 -- Emulsion Cement 0.37 0.19 0.42 Foam Cement 0.19 < 0.01 0.08 Dashes indicate too few data were available for comparison after outliers removed. Table 9. Coefficients of correlation (R2) between percentage maximum density and dynamic modulus at 10 Hz.

40 Material Properties of Cold In-Place Recycled and Full-Depth Reclamation Asphalt Concrete A bivariate correlation was performed between the design and performance properties. Correla- tion coefficients between these two groups are summarized in Table 10. Very few strong correlations were observed; nearly all of the correlation coefficients were less than 0.3 in absolute value terms. At 4°C, 25 Hz and 20°C, 10 Hz, the modulus values were more strongly correlated with bulk density. A modestly strong correlation also was observed between lower shelf and dry additive. The intercorrelations within each category of properties are presented in Table 11 and Table 12. As expected, modulus values had higher intercorrelations among themselves (E* @ 4°C, 25 Hz versus E* @ 20°C, 10 Hz versus E* @ 40°C, 1 Hz), as shown in Table 11. Gradation parameters (P200, Cu, Cz) also had higher intercorrelations, as shown in Table 12. Overall, the limited set of cold-recycled materials at the end of this study prevented any stronger statistical conclusions between mixture and construction parameters and expected material behavior. Percentage of Mix Design Density D yn am ic M od ul us a t 1 0 H z ( M Pa ) 21.1°C 37.8°C 4.4°C Figure 30. Relationship between percentage mix design density and stiffness. Stabilizer Content P200 Cu Cz OL Thickness Curing Time Dry Additive Depth of Recycling Bulk Density Max. E* 0.09 0.037 0.06 0.135 0.061 0.023 -0.029 0.044 -0.091 Min. E* -0.043 0.218 0.004 0.212 0.089 0.035 0.32 0.225 -0.074 Beta ( ) 0.092 0.073 0.128 0.205 -0.071 0.116 0.142 0.072 0.061 Gamma ( ) 0.016 -0.211 -0.156 -0.069 -0.18 0.084 -0.056 -0.174 0.254 EA -0.141 -0.144 -0.247 0.013 -0.084 0.011 0.107 -0.123 0.169 E* @ 4.4°C, 25 Hz -0.022 -0.239 -0.276 -0.17 -0.268 -0.064 -0.165 -0.141 0.766 E* @ 21.1°C, 10 Hz -0.055 -0.059 -0.208 0.113 0.123 -0.092 0.097 0.159 0.382 E* @ 37.8°C, 1 Hz -0.033 -0.129 -0.129 0.459 0.391 -0.027 0.428 0.536 -0.112 Table 10. Correlation coefficients of design versus performance parameters.

Dynamic Modulus Test Results 41 Max. E* Min. E* Beta ( Gamma ( EA E* @ 4°C, 25 Hz E* @ 20°C, 10 Hz E* @ 40°C, 1 Hz Max. E* 1 0.085 0.518 -0.016 -0.688 -0.178 -0.152 -0.031 Min. E* 1 0.243 -0.488 0.052 0.017 0.221 0.298 Beta ( ) ) 1 -0.162 -0.318 -0.159 -0.265 -0.117 Gamma ( ) ) 1 0.37 0.309 0.123 0.052 EA 1 0.413 0.344 0.152 E* @ 4.4°C, 25 Hz 1 0.811 0.286 E* @ 21.1°C, 10 Hz 1 1 0.717 E* @ 37.8°C, 1 Hz Table 11. Intercorrelation of performance parameters. Bulk Density Stabilizer Content P200 Cu Cz OL Thickness Curing Time Dry Additive Depth of Recycling Bulk density 1 0.069 -0.181 -0.151 -0.412 -0.49 0.013 -0.391 Stabilizer content 1 0.455 0.681 0.463 -0.221 0.595 -0.083 P200 1 0.479 0.248 -0.225 0.165 0.287 Cu 1 0.237 0.097 0.004 0.055 Cz 1 0.359 0.386 0.638 OL thickness 1 -0.288 0.51 Curing time 1 0114 Dry additive 1 1 -0.478 -0.495 0.139 0.067 0.617 0.381 -0.335 0.721 Depth of recycling Table 12. Intercorrelation of design parameters.