Abstract
This study identifies significant limitations in current methods for assessing thermal segregation in pavement construction. Existing methods capture uniformity but fail to distinguish between uniformly good and bad conditions, often missing critical quality issues. Field visits revealed a disconnect between identified severe thermal segregation profiles and actual pavement conditions, indicating that the current classification method may not accurately predict pavement quality. Severe segregation, as indicated by differential temperatures over 50°F, often represents only moderate issues. The proposed thermal segregation risk index is an effective approach, offering a better representation of thermal distribution. Evaluating thermal segregation based on differential temperature relative to the target discharge temperature is also more suitable. The study highlights the need for refined evaluation methods that consider absolute temperatures and relative differentials. These findings suggest the need for improved thermal segregation assessment methodologies to ensure reliable pavement quality evaluations.
Keywords
This study aims to evaluate projects classified by differential temperature assessment methods for current asphalt mixtures through field investigations. Since thermal segregation was first identified in the 1990s in relation to field characteristics, particularly the density of asphalt mixtures, there have been significant advancements in asphalt types and design methods ( 1 ). In addition, the hardware and software available for measuring and calculating thermal segregation, along with Department of Transportation (DOT) specifications, have seen remarkable improvements. However, the criteria for evaluating thermal segregation remain insufficient because of the weak correlation between thermal segregation results and field performance, as well as the lack of clarity in differential range statistics (DRS) specifications ( 2 ).
This study verifies thermal segregation, thermal profiling, asphalt mixture uniformity, and construction quality through field investigations. To establish robust evaluation criteria, this study utilized existing thermal profile datasets representing the current generation of asphalt mixtures. These datasets were thoroughly reviewed to assess compliance with existing criteria for moderate and severe thermal segregation across various mixture types.
A critical outcome of this analysis was the preliminary assessment and benchmarking of recurring thermal segregation. Field evaluation results also assessed the effectiveness of current quality control measures, revealing several limitations in the existing thermal segregation criteria. The findings from this study are intended to inform the establishment of new criteria for thermal segregation and updates to the related evaluation methods. By addressing these gaps, the study aims to enhance the reliability of thermal segregation assessments and improve overall pavement construction quality.
Research Literature
Throughout the 2010s, researchers investigated and fostered digital construction technologies, including those related to thermal segregation and profiling. The recent body of knowledge on and study of thermal segregation includes evaluations of the impacts of thermal and/or gradation segregation on laboratory and field performance, in some cases focusing specifically on warm-mix asphalt. Historical information has generally agreed well with the data obtained from the current literature.
Before the mid-1990s, it was believed that the paver screed, as shown in Figure 1, adequately blended the colder crust into the rest of the hot-mix asphalt (HMA). However, infrared sensors and thermal imaging cameras have shown otherwise. The cold HMA passes through the paver relatively intact, resulting in concentrated cold spots in the mat at regular intervals, aligned with each truckload ( 1 ).
Thermal segregation has a negative impact on density, and cold spots are generally more crack-susceptible and have poorer fracture properties ( 3 – 7 ). Even with warm-mix asphalt, thermal segregation influences density, high-temperature stability, low-temperature cracking, and tensile strength ( 7 ). Coarser mixes are more prone to aggregate segregation ( 8 , 9 ). However, not all thermal segregation is aggregate segregation ( 7 ).
A total consensus with respect to the utility of thermal profiling does not exist. One study reported that field-cored thermal-segregated samples did not have a statistically lower density. Although thermal-segregated samples generally had a lower fracture resistance, the correlation between the temperature differential and changes in fracture resistance was not strong ( 10 ).
Schematic description of the asphalt paver ( 11 ).
Reporting Thermal Segregation
Every specification reviewed for continuous thermal profiling utilized a 150-ft segment length. Current methods derive results from relative temperature within a profile, with a minimum of 180°F and areas with full width. The thermal segregation calculation calculates the relative temperature by subtracting the 1st percentile from the 98.5th percentile. The term “paver stop” refers to the area 2 ft behind and 8 ft in front of a location where the paver stopped for more than 1 min, yet the specification lacks clear requirements for this parameter. Variations in the data included what was used to determine the temperature differential ( 12 ).
In Texas, the current standard for evaluating thermal segregation categorizes a differential temperature in the profile as moderate segregation if it exceeds 25°F and severe thermal segregation if it exceeds 50°F ( 13 ). Some alternative thresholds exist for thermal segregation. Two documents classified high-severity segregation at temperature differentials from 43°F ( 5 ) and 38°F ( 6 ). Based on changes in air void (AV) content, one study proposed medium-level segregation beginning at a temperature difference of 14.4°F and high-level segregation beginning at a temperature difference of 32.4°F ( 7 ).
Benchmarking of Thermal Segregation for Current Mixes
This study focused on existing data collected with a paver-mounted thermal imaging system to identify what level of thermal segregation is normal for current mixes and paving practices. Table 1 presents a summary of the historical thermal profile data. Table 1 also shows that of the mix types with data available to researchers, the results obtained for the Superpave (SP)-C mix were substantially more numerous than the amount of thermal profile data available for the other mixes.
Available Full-Coverage Thermal Profile Data for Benchmarking
Note: SP = Superpave; SMA = stone mastic asphalt; CAM = Crack Attenuating Mixture; DG = Dense-Graded; TOM = Thin Overlay Mixture; PFC = Permeable Friction Course; TBPFC = Thin Bonded Permeable Friction Course.
Texas Department of Transportation’s (TxDOT’s) Superpave-B mixture approximates a standard 19 mm mixture, Superpave-C approximates 12.5 mm, and Superpave-D approximates 9.5 mm.
Table 1 presents the results showing that the extent of moderate thermal segregation, defined as temperature differentials exceeding 25°F and less than 50°F, ranged from approximately 19% to 58%. For most mix types, the results indicate that 19%–30% of the profiles exhibited moderate thermal segregation. The amount of severe thermal segregation, defined as temperature differentials exceeding 50°F, ranged from 0% to about 26%, with the results for most mix types showing results between 1% and 8% severe thermal segregation. Figure 2 shows the combined distribution frequency from the over 20,000 thermal profiles representing the historical data for different mix types. The data indicate that 67.2% of profiles had no thermal segregation, 28.5% had moderate thermal segregation, and 4.3% had severe thermal segregation.
Distribution of thermal profile results from all data.
Field Evaluation Location for Thermal Segregation
In the field evaluation, researchers conducted site visits to selected projects based on benchmark analysis results. The objective was to assess the current condition of existing pavement sections and determine whether locations that exhibited thermal segregation during placement showed any signs of distress or irregularities. Thirteen projects were chosen for site visits based on the review of benchmark analysis data. Table 2 provides the thermal profile results and project information, including the age of the selected projects (ranging from 2 to 9 years) and the mix types, such as SP-C, SP-D, and stone mastic asphalt (SMA)-D. The researchers specifically visited areas where severe thermal segregation was anticipated based on the differential temperature according to Tex-244-F ( 14 ) and conducted on-site verification to validate their findings.
Locations for Current Condition Surveys
Note: Let date=Official date a project is advertised and opened for contractor bids; SP = Superpave; SMA = stone mastic asphalt.
This study employed nondestructive techniques, such as a ground penetrating radar (GPR) survey, along with a digital video log, as depicted in Figure 3. In addition, site visits were conducted to assess the current condition of the pavement and determine whether thermal segregation identified in the thermal profile data was correlated with any areas exhibiting distress. Before the field condition survey, a detailed analysis of the temperature profile data was performed to identify locations with potential severe segregation, as illustrated in Figure 4. The entire project was meticulously evaluated using GPR and a digital video log, providing comprehensive insights. Visual inspections conducted during the site visits confirmed the presence of distress in selected areas where thermal segregation was frequently observed according to Tex-244-F.
Ground penetrating radar survey and visual condition survey using digital video equipment.
An example determination for survey locations of interest on the TX-2 project.
Field Evaluation Results: Phase I
In Phase I of the study, six projects with a high frequency (>5%) of severe thermal segregation, based on the benchmark analysis, and two projects with significantly lower segregation rates were selected for comparison.
Distress Observation at Severe Profiles
During site visits, areas where severe thermal segregation were measured at the time of construction based on the differential temperature according to Tex-244-F were visited, and on-site verification was conducted to validate current thermal segregation criteria. A total of 175 profiles in eight projects were identified before the site visits, ranging from 3 to 52 severe thermal profiles per project, as shown in Table 3, indicating the presence of severe thermal segregation. However, during the field evaluation, distresses associated with severe thermal segregation were expected to be found in 3–41 profiles per project (a total of 123 profiles), but none was detected. TX-1, which had already been seal coated, was excluded from the analysis. Although distress was found in TX-7, the location of the distress did not align with a location of known severe thermal segregation.
Summary of Field Evaluation Results: Phase I
The overall site visit revealed a lack of correlation between the severe thermal segregation profiles determined by the current classification method and the actual condition of the surveyed pavement. The field evaluation discrepancy suggests that the current classification method may not accurately predict the condition of the pavement.
Proposed Evaluation Criteria
In pavement projects, typically the contractor’s quality control personnel establish a rolling pattern, the final stage of road pavement work, considering factors such as the target delivery temperature of asphalt, base temperature, humidity, and wind speed. Thermal segregation evaluation aims to assess whether the temperature remains consistent when the asphalt is spread on the pavement. When the asphalt temperature remains constant on spreading and within the target laydown temperature range, there are no issues. However, problems arise when delays or unexpected issues occur during asphalt placement, leading to thermal segregation. Most projects follow the initially determined pattern based on the target delivery temperature, so any thermal segregation directly affects the project’s quality.
Figure 5 illustrates an example of thermal profiles from a project, with evaluation based on the temperature differential within the profile. In this project, while most profiles show no thermal segregation, two moderate thermal segregations are presented at one-third of the profile. Furthermore, moderate thermal segregation is more prominent in the latter two-thirds of the profile. In contrast to the differential temperature evaluations, the actual placement temperature across the profile is notably lower in the first one-third, ranging from 230°F to 260°F, while the latter two-thirds exhibit higher temperatures between 280°F and 315°F. A lower temperature than expected poses a problem because asphalt mix placed at lower temperatures results in lower density on compaction and higher porosity even after compaction. Moreover, materials with significantly low temperatures require more compaction effort, leading to substantial quality deterioration. Because of its inherent limitations, the current thermal segregation evaluation method based on differential temperature may not effectively pinpoint problematic areas.
An example of current evaluation profiles based on differential temperatures.
Thermal Segregation Evaluation Limits
The thermal segregation evaluation limits based on the current differential temperature, identified through the benchmark analysis and the initial field evaluation, are as follows.
Although thermal evaluation collects thousands of temperature data points within a single profile, evaluating a profile is based on the difference between only two points: the 1st and 98.5th percentiles. Many data points may lie close to either the 1st or 98.5th percentile, making it challenging to discern what occurs between these two percentiles.
When both the 1st and 98.5th percentiles are low (indicating an overall cold profile), it represents the most severe case. However, the current data approach assumes no thermal segregation because the profile is deemed uniform. The compaction pattern of asphalt remains consistent from the project’s outset, rendering sudden drops in temperature detrimental to the asphalt’s condition.
The impact of decreasing temperature on thermal segregation is non-linear, exacerbating with lower temperatures, an aspect currently overlooked in our analysis.
Proposed Evaluation Criteria
This study proposes evaluation criteria considering several identified limitations.
Instead of calculating relative temperature based on the difference between the 1st and 98.5th percentiles, the proposed method evaluates absolute temperature across all values between the 1st and 98.5th percentiles.
This study measured the bulk density of asphalt using a modified American Association of State Highway and Transportation Officials (AASHTO) T 19 method ( 15 ). Cooler asphalt has lower workability, leading to less material being placed per unit area, even with uniform compaction effort and final thickness. Thus, lower density results not only from compaction difficulty, but also from reduced material during placement. To capture this effect, bulk density was measured in the loose condition. HMA was divided into three portions and placed in a mold measuring 4 in. in diameter and 10 in. in height, with each layer leveled using a steel rod, as shown in Figure 6.
The bulk density of the asphalt was determined by weighing the material at different temperatures, as shown in Figure 7a. Figure 7b illustrates the relative bulk density (RBD) based on the temperature of 275°F. As the temperature increases, the height decreases. When materials are cooled below the recommended temperature (275°F), the RBD falls below 1. A RBD of less than 1 indicates lower bulk density before compaction, resulting in increased AVs with a predetermined rolling pattern. Conversely, materials compacted at higher than the recommended temperature will have lower AVs, leading to a RBD of greater than 1.
Researchers propose a new concept, the thermal segregation risk index (TSRI), which can be computed according to Equation 1 and is illustrated in Figure 8. The TSRI is calculated by dividing the bulk density obtained at a specific temperature by the bulk density at the target temperature minus 25°F, and then averaging all the data. To simplify the numbers, researchers subtracted 1 from the calculated value and then multiplied the result by 100. This index is valuable for evaluating the density of material discharged before compaction based on temperature, thus facilitating the assessment of thermal segregation.
A bulk density unit weight of asphalt using the modified American Association of State Highway and Transportation Officials (AASHTO) T 19.
An example of (a) the impact of temperature on bulk density and (b) relative bulk density (reference temperature 275°F).
An example of the thermal segregation risk index (TSRI) from laboratory results.
While this graph’s characteristics may vary depending on binder grade or material, corrected target temperature settings will ensure its utility in identifying thermal segregation patterns according to the designated compaction pattern. The reference temperature (RT) can vary based on the target discharge temperature (TDT). For this study, assuming a TDT of 300°F, the RT is set at TDT minus 25°F (275°F) and plotted accordingly.
The temperature criterion of 25°F was set after reviewing the historical data. The density profile data collected in the study have shown that when the temperature differential is greater than or equal to 25°F, almost 90% of the samples fail to meet the density criteria ( 1 ). Therefore, in the TSRI it was assumed that the criterion for dividing the severity in this study is based on conditions that are more favorable when the temperature is higher than 25°F below the TDT:
Figure 8 illustrates how the TSRI changes based on temperature. When materials are cooled below the recommended temperature, the TSRI will be less than 0. TSRI values between 0 and −3 (corresponding to a temperature differential between 25°F and 50°F from the TDT) indicate moderate segregation, as identified by lower bulk density. TSRI values less than −3 (when the temperature differential exceeds 50°F from the TDT) indicate severe segregation. TSRI values greater than 0 at temperatures higher than the recommended temperature result in lower AVs and desired quality.
Based on the literature review and field evaluation, this study found that thermal segregation occurs when the paver places cool spots or streaks in the mat at a relatively low density, rather than compaction issues caused by low temperature. Less material is discharged when the temperature is lower, resulting in lower density even with the same compaction pattern, as shown in Figure 9. Therefore, the TSRI was defined based on the density achieved at various temperatures rather than its effect on compaction at different temperatures.
Temperature differential sequence of events ( 1 ).
Field Evaluation Results: Phase II
Based on the field findings of the current Tex-244-F method (differential temperature) in determining thermal segregation, additional field surveys were conducted to evaluate thermal segregation. Five additional field trips focused on locations with small relative temperature differences but overall low temperatures.
Among the projects surveyed, thermal segregation was anticipated in all except for TX-9, which was chosen because of its consistent temperature profile. The field survey results are presented in Table 4. Contrary to the expectations based on the existing method, the field investigation revealed that most projects were not in good condition. Specifically, the locations of thermal segregation did not align with the locations of observed distress indicated in Table 4. In addition, although the temperature difference in the profile was small, the field investigation accurately identified locations with overall low temperatures and corresponding distress. This suggests that the existing method may not effectively capture the true extent of thermal segregation and its impact on pavement performance.
Summary of Field Evaluation Results: Phase II
In all projects, no distress was observed for the profiles currently classified as severe thermal segregation according to Tex-244-F. In addition, GPR surveys did not reveal any correlation between surface dielectric and differential temperature trends. However, in profiles where temperatures below 250°F (50°F lower than TDT) were predominantly distributed, several types of distress, such as raveling, patches, potholes, and pumping, were observed, as shown in Figure 10.
Field evaluation results: (a) TX-10 and (b) TX-12.
Re-evaluation Based on the Proposed Method
The following project demonstrates temperature evaluations under consistently low-temperature conditions. The project consists of eight profiles, totaling 1072 ft in length in the TX-10 area, using performance grade (PG)64-22 asphalt binder. The temperature differential assessment method categorizes this operation as exhibiting moderate segregation in one profile and good conditions in seven profiles, as shown in Figure 11. Despite the overall low measured temperatures (with a target compacting temperature of 295°F), the failure to detect severe thermal segregation is attributed to the relatively low-temperature differentials. Using the proposed approach, the project was evaluated by averaging the data between the 1st and 98.5th percentiles calculated using the TSRI. The calculation revealed that all eight profiles exhibited severe thermal segregation, aligning with actual field survey findings.
An example of comparison between differential temperature and the thermal segregation risk index (TSRI) on TX-10.
Another project example comprises 25 profiles spanning 3693 ft, utilizing PG64-22 asphalt binder. The differential temperature assessment method categorizes this operation as having 10 moderate segregation states, as shown in Figure 12. Despite the overall temperatures being lower than 240°F, as observed in the first profile of this project, a differential temperature of 21.8°F was recorded. In addition, despite the temperatures being close to the target compacting temperature for the overall project, the latter nine profiles are deemed to have moderate temperature segregation. This contrasts with the findings of actual field surveys. According to the proposed evaluation method, the first eight profiles of the project were assessed to have one severe and seven moderate temperature segregations. The remaining profiles were assessed to not have any concerns with thermal segregation. This aligns with the results of actual field surveys.
An example of comparison between differential temperature and the thermal segregation risk index (TSRI) on TX-12.
Discussion
This study focused on utilizing existing data gathered from a paver-mounted thermal imaging system and nondestructive techniques, such as GPR surveys, along with digital video logs, to assess the extent of thermal segregation at the time of construction and post-construction pavement condition. While the paver-mounted thermal imaging system effectively captures extensive information within a thermal profile, the challenge lies in simplifying the evaluation of these data. Currently, the assessment relies on evaluating relative temperature differences between two points (1st percentile and 98.5th percentile) in a profile without considering their absolute temperature or the slope between them.
These findings underscore that the current Tex-244-F method captures uniformity but fails to distinguish between uniformly good and uniformly bad conditions. For example, shows the Tex-244-F results with placement temperatures ranging from 232.7
Locations with Uniformly Low Temperatures Not Reported as Thermally Segregated
Note: Max. = maximum; Min. = minimum.
Locations with Less Uniformity but Higher Placement Temperatures
Note: Max. = maximum; Min. = minimum.
The case of TX-10 also supports the risk of differential temperature evaluation. Thermal segregation measurements were conducted over a total of 8.9 mi, revealing clear temperature discrepancies between the northern and southern sections compared to their absolute placing temperatures. The GPR values were analyzed by the dielectric constant (DC) value, as shown in Figure 13. The analysis indicated that high DC values (typically associated with low AVs) were more frequent in the southern section with high absolute temperatures, while low DC values (typically associated with higher AVs) were observed in the northern section with low absolute temperatures.
Dielectric constant analysis on TX-10.
Conclusions
In conclusion, this study highlights the limitations of current methods in assessing thermal segregation in pavement construction, proposes a new evaluation index, and refines the evaluation approach.
Existing techniques capture uniformity but fail to distinguish between uniformly good and bad conditions, leading to potential oversights in quality evaluation. The reliance on relative temperature differences overlooks crucial factors such as absolute temperature and slope between points.
Site visits revealed a disconnect between identified severe thermal segregation profiles and actual pavement conditions, suggesting that the current classification method may not accurately predict pavement quality. Severe thermal segregation, as classified by differential temperature, does not consistently indicate significant issues. Differential temperatures exceeding 50°F often represent moderate risk.
The proposed TSRI offers a more accurate approach than the evaluation method based on the difference between the 1st and 98.5th percentiles within individual profiles. The TSRI method provides a better depiction of thermal measurement distribution and risk of pavement distress.
For thermal segregation evaluation, assessing the differential temperature according to the TDT proves more suitable than relying solely on the relative temperature within individual profiles.
The study emphasizes the importance of refining evaluation methods to account for factors such as absolute temperatures alongside relative differentials, which could provide a more accurate depiction of segregation severity and risk of premature pavement distress.
Furthermore, the study emphasizes the need for additional TSRI analysis based on laboratory results for the mix type, as well as a review of historic data and in-field validation to support these findings and further refine the TSRI method, ensuring broader applicability and effectiveness. Overall, these insights highlight the importance of improving methodologies in thermal segregation assessment to enhance the reliability of pavement quality evaluations.
Footnotes
Author Contributions
The authors confirm the following contributions to this work: study conception and design: S.D. Sebesta, J. Kim; data collection: J. Kim, C. Bak; analysis and interpretation of results: S.D. Sebesta, J. Kim; draft manuscript preparation: J. Kim, S. Hu. All authors reviewed the results and approved the final version of the manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the Texas Department of Transportation and the Federal Highway Administration under TxDOT Research Project 0-7075.
The findings and opinions presented here are those of the authors and not necessarily those of the sponsoring agencies.