Introduction
Segregation in asphalt mixtures can be described as a concentration of coarse materials in some areas and fine materials in others, which results in non-uniform mixes that do not duplicate the original design, grading, or asphalt cement [
1]. Once this occurs, the segregated area of a pavement is likely to develop localized premature distresses such as fatigue cracking, rutting, raveling, pothole, etc. Numerous researchers and engineers have investigated the topic for decades. The National Cooperative Highway Research Program (NCNRP) Report 441 [
2] is one of these efforts, in which segregation was further identified as 1) gradation segregation, 2) temperature segregation, and 3) aggregate-asphalt segregation (a.k.a., drain-down in stone matrix asphalt).
Gradation segregation results in either coarse aggregate-rich or fine aggregate-rich spots and is the most common problem encountered; therefore, many remedies have been introduced that result in significant reductions of the problem. Widely practiced remedies to reduce the chances of gradation segregation include multiple pile truck loading from the storage bin at the plant, use of material transfer vehicle (MTV) from truck bed to paver, and so on. The second type, temperature segregation (TS), is a recently-found phenomenon thanks to the popular use of high-precision portable infrared thermal cameras in the paving sites. Several state agencies and researchers have investigated this latest phenomenon to find causes and possible effects on the performance of asphalt pavements [
2–
16]. While many studies reported similar causes of the TS, such as differential cooling on the haul trucks, lack of remixing before laydown, inappropriate paver operation, etc., concerning the effects on the quality and performance of pavements, some studies found correlations while others did not. The lack of agreement on the influence of TS to pavement quality and performance could have been resulted from inconsistent definition of TS and dissimilar ranges of asphalt mat temperatures investigated by different researchers.
Louisiana standard specifications for roads and bridges [
17] require specific operational details such as the truck loading practice, discharge manner, use of MTV, paver requirements, etc., to prevent the gradation segregation. Some of these practices such as the use of MTV have been commonly found to be effective in minimizing the TS by researchers [
4–
8]. However, the actual occurrence of the TS in Louisiana asphalt pavements has not been investigated until recently nor the circumstances under which, how often, how long, and how severe the TS occurs. Moreover, better understanding the ultimate link between TS and asphalt pavement performance via mechanical properties of asphalt mixtures will enable Louisiana pavement engineers to tailor the solutions to mitigate the problem.
Objectives and scope
The primary objective of this study was to determine the impacts of temperature segregation on the quality and mechanical properties of asphalt mixtures as defined by density, fracture resistance, stiffness, and rutting performance of asphalt mixtures. Seven asphalt rehabilitation projects across Louisiana were investigated to include various factors that might be related to the temperature segregation phenomena. These factors are listed below.
Asphalt mix: wearing course (WC), binder course (BC), and shoulder mix;
Asphalt binder grades: PG64-22, PG70-22, PG76-22, and PG82-22 rubber modified asphalt;
Mix production technique: hot-mix asphalt (HMA) and warm-mix asphalt (WMA);
Nominal maximum aggregate sizes (NMAS): 12.5 mm and 19 mm.
Methodology
Field projects and asphalt mixtures
Seven asphalt rehabilitation projects across Louisiana were selected for evaluation. Table 1 presents details of these projects with the ten asphalt mixtures included in this study, which were staged as Phase I and Phase II. Phase I included field projects completed during the first year of the study: LA30 WC, LA1058 WC, US165 WC, and LA1053 BC. Phase II included field projects completed during the second year of the study: LA1053 WC, LA411 WC, LA940 BC, LA940 WC, LA1 Shoulder, and LA1 BC. Out of the ten mixtures investigated, four were HMA and six were WMA mixes. The thickness of WC layer ranged from 38 mm (1.5 in.) to 50 mm (2.0 in.), while that of BC and Shoulder were 50 mm (2 in.) and 100 mm (4 in.), respectively. The WC layer mixes used 12.5 mm NMAS, while the BC and Shoulder mixes used 19.0 mm NMAS. The target laydown temperatures of these mixtures ranged from 135°C to 149°C.
Pavement density (Voids in Total Mix–VTM), Semi-Circular Bending (SCB), Loaded Wheel-Tracking (LWT), and Indirect Tensile Dynamic Modulus (IDT|E*|) tests were performed on field cores to measure the volumetric and mechanistic properties of compacted thermally-segregated asphalt mixtures.
Temperature segregation measurements
Figure 1(a) shows the multi-sensor infrared bar (IR-bar) system used to record temperatures of uncompacted asphalt mat immediately behind the paver during the first year of the study (Phase I). The IR-bar system consists of 12 IR sensors arranged on a bar with 0.3 m distance between sensors, an odometer measuring trip distance, a GPS receiver, and an operand for controlling the system and displaying the real-time thermal profiles of asphalt mats. The IR sensors measure the real-time temperature of every 0.1×0.3 m2 area continuously with a measurement accuracy of 1.1°C (2°F) to create a thermal profile of an entire project. Figure 1(b) shows a hand-held portable thermal camera used, in addition to the IR-bar system, during the second year of the study (Phase II) to capture stationary thermal images of a specific area. This portable thermal camera enabled the research team to measure mat temperatures over time until the first breaking roller compacts the area.
Table 2 presents the TS categories that were established to group asphalt mat areas with different levels of TS severity: from none to very high. As seen in the first column of the table, each group had 13.9°C (i.e., ±6.9°C) temperature range with each mid-point shown in the parenthesis. Thus, when the target temperature is 148.9°C, an asphalt mat with temperature between 142.0°C and 155.8°C has the “none” severity level of TS and is designated by the group name “Target.” Note that the temperature range of 13.9°C corresponds to the temperature differential of 25°F, which was the primary temperature unit used during the study. Temperature ranges of the successive severity levels are defined by a successive 13.9°C drop at a time from the target temperature, and then, simultaneously applying ±6.9°C to that value. For example, when the target temperature is 148.9°C, the temperature range of the Low severity group is defined as (148.9–13.9)±6.9°C (i.e., from 128.1°C to 142.0°C). In this way, the following groups of medium, high, and very high severity levels are found to have the temperature ranges of 121.1±6.9°C, 107.2±6.9°C, and 93.3±6.9°C, respectively, and their group names are given as combinations of the word “Target” and corresponding temperature differentials in Fahrenheit as shown in the last column.
Laboratory tests
Field cores were taken within the first two weeks of overlay placements from the areas with different severity levels of TS, as specified in Table 2, and then tested in the laboratory for the volumetric and mechanical property measurements.
The bulk specific gravity of asphalt mixtures was measured in accordance with AASHTO T166; fracture resistance of asphalt mixtures was measured by the SCB test in accordance with ASTM D8044-16; rutting resistance of asphalt mixtures was evaluated by the LWT device in accordance with AASHTO T324; and the dynamic modulus of asphalt mixtures was measured in the indirect tensile mode (IDT |E*|) at three test temperatures (i.e., –10°C, 10°C, 30°C) and at five different frequencies (i.e., 10, 5, 1, 0.5, 0.1 Hz) similar to the test method used by Kim et al. [
18].
Analysis procedures
Temperature uniformity
Temperature uniformity of all pavement sections was evaluated using two different parameters: an average standard deviation and %Severe.
The average standard deviation in mat temperature was computed for each of the ten pavement sections. To do so, the entire length of the thermal profile was first divided into a number of 45-m long segments. Within a single 45-m long segment, which includes approximately 5400 temperature readings, a standard deviation of the temperature readings was calculated. Then, the average of the standard deviations in all segments were computed and analyzed to show the overall quality of paving operations in keeping the consistent laydown temperatures throughout each of the ten pavement sections.
The second parameter, %Severe, is the sum of percentages of TS severity level that fall into medium, high, and very high groups. A typical overlay project that is approximately 3,600-m long had 432,000 temperature readings in its Pave-IR thermal profile. These individual temperature readings were individually classified into corresponding TS severity level, and then the percentages of the five TS severity levels were computed. The %Severe parameter takes the three worst TS severity levels (i.e., medium, high, and very high) into account to evaluate the quality of paving operations.
Statistical analysis
When appropriate, a paired t-test was conducted to compare the two mean values (i.e., temperature, density, and mechanical properties) of different TS severity levels. The two-tail conditional probability (p-value) of 0.05 was used as the null hypothesis test (i.e., m0≠ m1) criterion. A multiple comparison among different mean values using the Tukey’s range test procedure was also conducted to classify significantly different means of different TS severity levels into different groups.
Results and discussion
Temperature variations
Throughout the study, two distinctive patterns of temperature segregation were observed: a cyclic temperature segregation (CTS) due to continuous cooling of asphalt mixtures during the normal paving operation and irregular temperature segregation (ITS) due to un-planned work stoppages during the paving operation [
10,
11].
Table 3 presents the average standard deviation in temperature and its relationship to varying levels (category) of construction factors, such as the MTV type, contractors, nominal maximum aggregate size, and so forth. Also, the sample size associated with the categories in each of the five factors are shown. The number of 45-m segments shown in Table 3 gives an estimate of profile length and sample size used to conduct Tukey analysis for finding differences and similarities among different categories in a factor. From this statistical analysis, it was observed that the use of Light or Full MTV improves temperature uniformity compared to No MTV. Also, Full MTV delivered significantly better temperature uniformity than Light MTV as its grouping, C, differs clearly from that of Light MTV, B. For the temperature uniformity in four different Contractors, no significant differences were observed in their grouping, i.e., all four were grouped as A. All projects used nominal maximum aggregate size as 12.5-mm for WC and 19-mm for BC. The 19-mm aggregate may have more surface area open to environment, which can cause rapider cooling. In Table 3, it is evident that the temperature uniformity of 19-mm NMAS mixtures was worse than that of 12.5-mm NMAS mixtures. It can be noted, however, that the difference in the average standard deviations of the two NMAS is merely 1.1°C, although the Tukey test grouping shows two different letters, A and B. For the target temperature and ambient temperature, no significant differences were found among different levels.
Table 4 presents the combined percentage of severe temperature segregation levels, which includes medium, high, and very high severity levels as shown in the footnote. Due to limitations in sample size, only three construction factors (NMAS, target, and ambient temperatures) were considered in the Tukey analysis for their effects on the occurrence of temperature segregations. As shown in the table, it was found that none of the three construction factors have statistically significant effects on the occurrence of severe temperature segregations.
Figure 2 further presents the results of two different temperature uniformity analyses: standard deviation and percent severity levels of TS concerning the MTV use. It should be noted that the additional severity levels such as “hot” and “extreme” are presented in Fig. 2(b). These two additional levels were introduced to adequately fragmentize the entire temperature data into equally spaced groups, which were not experienced in the Phase I study. The hot severity level includes the temperature segregations that are at least 13.9°C hotter than the target range, i.e., (Target+13.9)±6.9°C.
Figure 2(a) shows a decreasing trend in average standard deviation with the lowest value of 4.4°C for Full MTV section. In Fig. 2(b), Full MTV section shows 60 percent of Low severity level TS. It must be understood that in Fig. 2(a), the deviations are calculated from average temperature of the complete thermal profile, which is a relative measure of the quality of paving. When the deviation is low, the asphalt mixture could have been paved uniformly at High severity level or uniformly at Low severity level. On the other hand, Figure 2(b) shows the entire distribution of the TS into separate severity levels. Therefore, the two measures shown in Figs. 2(a) and 2(b) together tell a more precise story about the effects of MTV use in temperature uniformity and about the paving work in general. For the No MTV section, it was observed that the entire section has the highest average standard deviation at 10.7°C, while Light MTV and Full MTV sections have lower values at 9.2°C and 4.4°C, respectively.
Also, the No MTV section has the greatest number of severity levels all the way from hot to extreme, showing that the mixture temperature during the paving varied a greater deal, suggesting that the overall quality of the paving was questionable. While the Full MTV section showed the lowest standard deviation among the three, it also showed that a significant portion (nearly 60 percent) of the entire TS falls in the low severity level and more (i.e., medium, high, very high, and extreme). On the other hand, the Light MTV section shows less percentage of the TS (slightly over 20 percent) falls in the same severity levels. This observation alone can mislead to a conclusion that the light MTV provides better paving quality. However, Fig. 2(b), in fact, shows a considerably large percentage of TS (about 28 percent) is marked as hot, meaning that the asphalt mixtures left the asphalt plant significantly “overheated” than the mixtures used in Full MTV section. Hence, during the paving operation, the mixtures in Light MTV section mostly stayed in the hot and none severe TS levels until eventually cooling down to low and more severe TS levels. Moreover, the number of severity levels present in the Light MTV bar of Fig. 2(b) is larger than that of the Full MTV bar, which only shows three levels at none, low, and medium. Thus, it can be concluded that the Full MTV section has the most uniform temperatures throughout the entire section and so does the quality, in general. It is also noteworthy that this more uniform and better quality paving was achieved without having to overheat the asphalt mixtures in the plant, which is beneficial for the producer in reducing the fuel cost and potential issues in environmental protection.
Density variations
Effects of temperature segregation on field density
Figures 3(a) and 3(b) show density measurements of ten mixtures at various TS severity levels. The averaged air voids content shown in these charts have sample sizes ranging from two to eleven with less than 20 percent of coefficient of variation. Overall, the bar charts show an increase in air voids with increase in severity level. The maximum allowable specification limit is 8 percent.
Among the Phase I projects in Fig. 3(a), LA30 WC is the only project that shows significant increases in air voids of segregated specimens. The Tukey comparison analysis of the mean air voids of different TS severity levels shows that the LA30 WC samples were the only mixtures that showed clearly different letter groupings among different TS severity levels. Increasing trends of air voids are graphically visible in other projects, too, but the differences between and among different TS severity levels are not statistically significant, as their letter groupings all resulted in a single group, A. Note that the temperature segregation severity levels of Phase I projects were determined at laydown using the Pave-IR system immediately behind the paver.
Unlike the Phase I projects, the temperature segregation severity levels were determined at the time of the first breaking roller compaction using the IR camera. Among the Phase II projects in Fig. 3(b), LA411 WC, LA940 BC, LA940 WC, and LA1 BC showed clear distinctions in air voids between the Target and segregated samples, especially when the TS severity level is worse than Target-75. According to the Tukey comparisons, no difference was detected between the Target and segregated samples up to the Target-75 severity level in five of the six projects (LA1 Shoulder is the only exception). Then, as the severity worsened, four of the six projects started to show clear distinctions between their Target and Target-100 samples. In fact, among the five projects that include Target-100 samples, LA1 Shoulder was the only project that did not show increased air void in Target-100 samples. This could have resulted in due to the temperature gradient of asphalt layer in depth. The layer thickness of LA1 Shoulder is 100 mm, while the layer thicknesses of other projects were either 38 mm or 50 mm. The IR thermal sensors and cameras measure the temperature of asphalt mats on the surface, but the temperature in a certain depth of the mat is normally higher and rises as the depth goes deeper. Thus, it may be possible that the actual in-depth layer temperature of the 100 mm thick LA1 Shoulder at the time of compaction was closer to the target temperature than was measured on the surface.
The clear contrasting observations between the Phase I and Phase II projects in Figs. 3(a) and 3(b) suggest that the TS must be evaluated at compaction rather than laydown operation. Also, the layer thickness of asphalt mats should be taken into account for the evaluation.
Random sampling vs. targeted sampling
Figure 4 presents the density differentials (DD) of the field cores measured from the minimum specification limit of 92 percent. The quality acceptance (QA) core DD data are shown along with the field core DD at different TS levels for comparisons between the random sampling process used in QA sampling and the targeted sampling process followed throughout this study. The average DD of QA samples is 2.3 percent more than the required minimum value of 92 percent, which is an excellent achievement. The average DD of None severity level TS samples is 1.9 percent still in excellent level. However, the DDs of TS samples tend to decrease as the TS level gets worse; DD values of Low, Medium, High, and Very High (VH) TS levels were recorded as 1.2, 0.3, 1.5, and −1.6 percent, respectively. This observation suggests that the pavement qualities assessed by the current QA sampling protocol could deviate from the true quality due to the inherent risk associated with the random sampling process. The thermal imaging techniques can be used to guide the project inspectors to determined targeted sampling locations for more effective quality control and quality acceptance during the construction.
Effect of construction factors on density differentials (DD)
Table 6 presents the average DD expressed in percentage and the Tukey analysis grouping among different categories of construction factors, i.e., NMAS, target temperature, and ambient temperature. As shown with the Tukey Grouping, NMAS and ambient temperature did not affect the average DD. On the other hand, the target laydown temperatures appeared to affect the average DD among the three categories.
Table 7 shows the maximum DD within the categories of the three construction related factors. Similar to the average DD, NMAS and ambient temperature showed no statistically significant effects on the maximum DD, while the target laydown temperature appeared to have some effects on the maximum DD. The mixtures with a laydown temperature of 135°C were Foamed WMA, 143°C were Latex-modified asphalt, and 149°C were HMA. Although the number of observations are small, it would suggest that the foamed WMA and HMA make no difference in DD, while the latex modified asphalt mixtures can cause higher DD when TS occurs.
Fracture resistance
Limited core samples from Phase I projects (i.e., LA30 WC, US165 WC, and LA1053 BC) and Phase II projects (i.e., LA940 BC, LA1 BC, and LA1 IP) were tested for fracture resistance evaluation using the SCB test. LA DOTD 2013 specification requires a minimum Jc value of 0.5 kJ/m2 for asphalt mixtures designed for low volume roads. Figures 5(a) and 5(b) present the SCB Jc values of all six mixtures and their Tukey analysis results.
For all Phase I projects, a decreasing trend in Jc values was observed as the TS severity level worsened: LA30 WC showed a decrease of 0.17 kJ/m2 in Jc value from Target to Target-50 samples; US165 WC showed 0.22, 0.31, 0.25 kJ/m2 reductions in Jc for Target-25, Target-50, and Target-75, respectively; and LA1053 BC showed a slight decrease of 0.04 kJ/m2 in Jc value from Target to Target-50 samples. A similar trend was observed for Phase II projects with more profound decreases in Jc values: for LA940 BC, Jc values decreased by 0.30 and 0.32 kJ/m2 for Target-75 and Target-100, respectively; for LA1 Shoulder, Jc values decreased by 0.18 and 0.42 kJ/m2 for Target-75 and Target-100, respectively; and for LA1 BC, a huge decrease of 0.76 kJ/m2 from Target to Target-100 specimens was observed. The differences in the magnitude of Jc value reduction between Phase I and Phase II projects are caused by the different ranges of the TS severity levels included; i.e., up to Target-75 TS samples were included in Phase I and up to Target-100 TS samples were included in Phase II. When combined together, on an average, 0.22, 0.17, 0.24, and 0.50 kJ/m2 of Jc reductions were observed for Target-25, Target-50, Target-75, and Target-100 TS samples, respectively.
On the other hand, the Tukey comparisons of SCB Jc values shown with the letter groupings in Figs. 5(a) and 5(b) indicated that these reductions are not always significant. The Target-25 sample showed statistically significant reduction in Jc value, but with only one observation (i.e., pink bar). Two of three Target-50 samples (i.e., orange bars) in Phase I showed that the reductions are not statistically significant. Two of three Target-75 samples (i.e., green bars) showed the reductions are significant, while the remaining one appeared to be on the border line with the double-letter grouping of A/B. All three Target-100 samples (i.e., blue bars) showed the reductions are statistically significant. Overall, it can be concluded that the effects of TS up to Target-50 (i.e., medium severity) level on the SCB Jc values may not be significant, however, the effects become clearer and more significant when the TS level becomes as bad as Target-75 and worse (i.e., high and very high).
Rutting resistance
The rutting resistance of a selected field core samples from LA940 BC and LA1 Shoulder layers was evaluated using the loaded wheel tracking (LWT) device. Figures 6(a) and 6(b) present the average rut depth measurements of the two pavement samples for 20,000 passes of wheel loading. Also shown are the maximum allowable rut depth of 10 mm at 20,000 passes specified in the LA Standard Specifications and average air voids (VTM) of tested samples.
In LA940 BC, a distinctively higher rut depth of the Target-100 sample was observed compared to the rut depth of Target and Target-75 samples. The difference started from the early stage of the test and kept increasing until the terminal stage. Terminal rut depth values were 2.6, 2.9, and 5.4 mm for Target, Target-75, and Target-100, respectively. It is noteworthy that the higher rut depth of 5.4 mm for the Target-100 sample is still significantly below the specification limit. It is also interesting to note that the trend in terminal rut depth values of the three TS level samples roughly matches the trend of their corresponding air voids content.
In LA1 Shoulder, a slightly reversed trend in rut depth measurements was observed: the terminal rut depth value of 8.1 mm for the Target sample is higher than 6.1 mm rut depth for the Target-100 sample. Nonetheless, these rut depth values are still lower than the specification requirement, and the pavements are expected to perform well in service resisting the rutting. Interestingly, the rut depth increase patterns of the two samples were almost the same up to 10,000 wheel passes and only diverted from each other beyond that point, with the Target sample showing clearer signs of stripping around 15,000 wheel passes. VTMs of these two samples were almost the same. Overall, with the limited observations on the LWT rut depth evaluations of the TS samples, it is not clear whether the rutting resistance of asphalt pavements is affected by the measured temperature segregation, but the air voids (or density) of compacted asphalt pavements may be more responsible for the rutting.
Stiffness variations
Selected core samples from LA1053 WC, LA411 BC, and LA1 IP were tested for the IDT |E*| to determine the effects of TS on the stiffness of compacted asphalt pavements. For comparisons, the |E*| values of TS samples were normalized by the |E*| value of the Target sample at three temperatures (i.e., –10°C, 10°C, and 30°C) and at 0.1 Hz. Figures 7(a), 7(b), and 7(c) present the normalized IDT |E*| values of LA1053 WC, LA411 WC, and LA1 Shoulder samples, respectively.
Figure 7(a) shows that the stiffness of Target-50 sample was decreased by 12, 5, and 14 percent at –10°C, 10°C, and 30°C, respectively; while that of Target-75 sample gained mixed results except at 30°C where 13 percent reduction in stiffness was recorded. With up to 20 percent of typical experimental variability of IDT |E*| tests, the observed stiffness reductions did not seem significant. Statistical comparisons using the t-tests also showed that the differences were not significant by returning the two-tailed p-values much higher than 0.05. It is noteworthy that the density differentials of both Target-50 and Target-75 samples were not substantially large, i.e., 0.4 and 1.4 percent higher than that of Target samples, respectively.
In Fig. 7(b), similar mixed observations were made with Target-50 samples of LA411 WC where stiffness at –10°C and 10°C increased slightly and suddenly decreased by 27 percent at 30°C. The t-test results returned p-values of 0.78, 0.46, and 0.01 for –10°C, 10°C, and 30°C, respectively. The density differential of the Target-50 sample was 1.7 percent. On the other hand, the Target-100 sample showed consistently higher stiffness reductions across all three temperatures with a density differential of 6 percent. The t-test results indicated that these reductions are statistically significant with all p-values much lower than 0.05.
In Fig. 7(c) of LA1 Shoulder, the Target-100 sample showed consistently lower stiffness than the Target sample did at all three temperatures. Specifically, the stiffness was significantly reduced by about 40 percent at 30°C, t-test of which returned a p-value of 0.00. However, the other two t-tests between means of Target and Target-100 at –10°C and at 10°C showed statistically insignificant difference with respective p-values of 0.67 and 0.21. It should be noted that the density differential between the two samples was merely 0.1 percent. Overall, the observations clearly suggest that, as long as the final density (or air voids) of the finished pavements is within the acceptable specification limit, the stiffness of medium to high level temperature segregated pavements may not be adversely impacted. However, when the temperature drops to very high TS severity level (Target-100) and gets compacted, the stiffness of that area would be definitely lower than that of normal pavements, and be more prone to premature distresses.
Summary and conclusions
The objective of the study was to evaluate the effects of temperature segregation (TS) on the initial quality and the performance of asphalt pavements, as measured by the core density and laboratory measured mechanical properties such as fracture and rutting resistance.
Seven asphalt rehabilitation projects, which include ten different asphalt mixtures across Louisiana were selected for this research. Varying levels of construction related factors such as contractors, use of material transfer vehicle (MTV), ambient temperature, target laydown temperature, and nominal maximum aggregate size (NMAS) were considered during the study. The study was carried out through two paving seasons. Two infrared (IR) thermal imaging techniques were used for temperature measurements: the Pave-IR system was used for continuous temperature monitoring on all seven projects under Phase I and Phase II, while a handheld portable IR camera was used on Phase II projects, in addition. The Pave-IR system measured temperature of asphalt mixture near the screed behind the paver, while the IR camera measured temperature at a stationary work stoppage location until the first breaking roller compacts the area.
Field cores from varying levels of TS locations were obtained and tested for the density and mechanical properties in the laboratory. Bulk specific gravity testing was conducted to measure pavement density in accordance with AASHTO T 166. Semi-circular bending test (ASTM D8044) was performed to measure the fracture resistance at intermediate temperature, while loaded-wheel tracking test (AASHTO T324) was conducted for rutting resistance measurements. In addition, Indirect Tensile Dynamic Modulus (IDT|E*|) test was performed to measure the mixture stiffness.
Observations and findings of the study are summarized below:
1) According to the temperature uniformity analysis, the use of MTV significantly improved the uniformity of asphalt mixture temperature across the uncompacted mat. Pavement sections where full-size MTV with 20-ton storage capacity was utilized showed significantly better consistency than the sections where no MTV and/or light MTV was utilized. Larger aggregate mixtures defined by the nominal maximum aggregate size (NMAS) appeared to have higher temperature variability across the mat than smaller aggregate mixtures. Other factors, i.e., ambient temperature, contractors, and target laydown temperature did not affect temperature uniformity significantly.
2) Laboratory test results showed mixed trends in relationships to the temperature segregations:
- For the density, a strong correlation between the density and temperature segregation was found in Phase II projects where the temperature segregation was measured right before compaction rather than immediately behind the paver.
- Fracture resistance values showed a decreasing trend in most projects, and showed significant decrease in values as high as 0.76 kJ/m2 of Very High severity temperature segregation core samples.
- For the rut depth measured by the LWT device, it is not clear whether the rutting resistance of asphalt pavements is affected by the measured TS, but the air voids (or density) of compacted asphalt pavements may be more responsible for the rutting.
- IDT |E*| values of High severity TS samples at 30°C showed significant stiffness reductions around 35 to 40 percent, while the reductions were not significant at lower temperatures (e.g., −10°C and 10°C).
Based upon the analysis and findings presented, the following conclusions can be drawn:
1) Temperature segregation, as measured at the time of compaction, affects mixture properties depending on its level of severity.
a) TS from 0 to 28°C: Do not affect the density and mechanical properties;
b) TS from 28°C to 42°C: May not affect the mechanical properties depending on the associated density differentials
c) TS from 42°C to 56°C: Significantly reduce the density and mechanical properties.
2) TS measured right before compaction in Phase II projects correlated well with decrease in density, fracture resistance, dynamic modulus, and increase in rut depth.
The IR-bar system used for the study appeared to be a helpful device in monitoring the temperature uniformity across the asphalt mat immediately behind the paver, providing a vital quality control information in real-time during the paving process. However, the ultimate relationship between the temperature segregations measured at laydown to the quality and performance of the pavements could not be confidently established throughout the study, since many other uncertainties are still involved in the process between the laydown and the actual compaction of the asphalt mat. As observed, on the other hand, much better correlations were established between the temperature segregations measured right at compaction and the quality and performance of the pavements. Therefore, temperature segregation must be redefined as the non-uniform temperature distribution in the uncompacted asphalt mat, measured just before the first breakdown compaction, which causes significant reductions in pavement quality and performance.
Higher Education Press and Springer-Verlag GmbH Germany, part of Springer Nature
PDF
(994KB)
