|FHWA > Engineering > Pavements > Concrete > TechBrief: Performance of Sealed and Unsealed Concrete Pavement Joints|
Printable version of TechBrief (.pdf, 0.5 mb)
TechBrief: Performance of Sealed and Unsealed Concrete Pavement Joints
January 2009, Revised May 2009
This TechBrief presents the results of a nationwide study of the effects of transverse joint sealing on performance of jointed plain concrete pavement (JPCP). This study was conducted to assess whether JPCP designs with unsealed transverse joints performed differently from JPCP designs with sealed transverse joints. Distress and deflection data were collected from 117 test sections at 26 experimental joint sealing projects located in 11 states. Performance of the pavement test sections with unsealed joints was compared with the performance of pavement test sections with one or more types of sealed joints. It should be noted that most of the test sections studied were 10 years or less in age at the time of data collection.
The sealing of transverse contraction joints in JPCP has been standard practice throughout much of the United States for many years. Its widespread use is due to the common belief that sealing joints improves concrete pavement performance in two ways: by reducing water infiltration into the pavement structure, thereby reducing the occurrence of moisture-related distresses such as pumping and faulting; and, by preventing the infiltration of incompressibles (i.e., sand and small stones) into the joints, thereby reducing the likelihood of pressure-related joint distresses such as joint spalling and blowups.
Transverse joints in jointed concrete pavement (JCP) are typically created by making an initial sawcut to force controlled cracking, followed by a second, wider sawcut to produce a reservoir for the joint sealant material. This traditional approach of sawing and sealing transverse contraction joints is estimated to account for between 2 and 7 percent of the initial construction cost of a JCP. Moreover, these sealed transverse joints require resealing one or more times over the service life of the pavement, leading to additional costs in terms of labor, materials, operations, and lane closures.
Recently, several State departments of transportation (DOTs) have been questioning conventional transverse joint sawing and sealing practices. These agencies contend that the benefits derived from sealing do not offset the costs associated with the placement and continued upkeep of the sealant over the life of the pavement. As a result, they have been experimenting with different sawing and sealing alternatives, for example:
To address the question of the effect of joint sealing on JCP performance, the FHWA sponsored a study to collect and examine field performance data from a wide variety of in-service concrete pavement joint sealing experiments across the United States. This TechBrief presents the results of this nationwide study. It should be noted that most of the test sections studied were 10 years or less in age at the time of data collection.
As part of the field testing program, distress and deflection data were collected on 117 different test sections at 26 pavement projects located in 11 States (Arizona, Colorado, Florida, Georgia, Illinois, Indiana, Iowa, Minnesota, New York, Ohio, and Wisconsin). The locations of these projects are shown in figure 1. These projects represent a range of concrete pavement designs, sealant material types, and transverse joint configurations.
A standard, systematic, and comprehensive field data collection effort was employed in the study to obtain consistent and meaningful data. At each selected site, a visual distress survey was performed in general accordance with the procedures and distress definitions found in the Distress Identification Manual for the Long-Term Pavement Performance Program (Miller and Bellinger 2003), developed under the Strategic Highway Research Program. For each observed distress, the location, size, and severity were identified and drawn approximately to scale on distress survey maps. Because the focus of this study was on the effects of sealed or unsealed joints on performance, particular attention was paid to the following distresses:
For sections with sealed transverse joints, a primary goal of the field survey activities was to assess how well sealed the joints were and how well the sealant was performing. To determine how well the sealant was performing, the entire length of each joint was assessed to estimate the joint length that was performing well and the joint length that exhibited joint sealant failures. The lengths of the following three types of joint failures were estimated separately:
The total joint length with sealant failure was calculated as the sum of the measured lengths of adhesive, cohesive, and absence failures. To normalize the total failure length computed for each joint, this total length of sealant failures was converted into a percentage of total joint length.
As part of the joint sealant field evaluation process, an estimate also was made of the total joint lengths containing “coarse-grained” and “fine-grained” incompressible material. Coarse-grained particles were defined as stones with a diameter of 50 to 100 percent of the observed joint width, while fine-grained particles were defined as those with a diameter less than 50 percent of the observed joint width. The total recorded length of each incompressible type within a joint was converted into a percentage of total joint length.
Deflection testing was conducted with a falling weight deflectometer (FWD) to assess the support conditions and load transfer efficiencies of the various pavement structures included in the study. For this study, a standard sensor spacing and a multiple-drop load sequence were employed. The FWD testing was performed by the participating State DOTs (i.e., each participating DOT conducted the testing using its own equipment and operator) following a prescribed testing protocol to provide consistency in the way the deflection data were collected.
To allow for joint, edge, and midslab testing without repositioning during testing, the positions for five sensors along the longitudinal sensor bar were -305, 0, 305, 610, and 914 mm (-12, 0, 12, 24, and 36 in.) from the center of the load plate. Optional placements for two additional sensors, if available, were 1,219 and 1,524 mm (48 and 60 in.) from the center of the load plate. The FWD drop loads were approximately 40, 53, and 80 kN (9,000, 12,000 and 18,000 lbf), recorded as load levels 1, 2, and 3, respectively. A testing pattern consisting of a sequence of drop load levels 3, 1, 2, and 3 was used at each test position.
Within each test section, a group of four consecutive slabs was tested. To complete this testing efficiently, the FWD made two passes over the four-slab group. For the first pass, down the center of the outer traffic lane, the FWD was positioned so that the load plate was placed equidistant between the lane–shoulder and lane–lane longitudinal joints. Along the first pass, five transverse joint-leave tests and four center-slab tests were taken. For the transverse joint approach tests, the load plate was placed as close as possible to the joint. For the center-slab tests, the FWD was maneuvered so that the load plate was as close to the geometric center of the slab as possible. The second pass consisted of an edge pass in which the FWD load plate was positioned as close as possible to the outer lane–shoulder joint. Along the second pass, 10 corner tests and 5 outer edge tests were taken. A summary of the deflection testing locations in each four-slab group is outlined in figure 2.
To aid in the analysis of the FWD data, temperatures were collected at various depths within the pavement by way of a multidepth temperature probe placed in a hole 13 mm (0.5 in.) in diameter drilled into the slab.
An analysis of the collected faulting data found that the only sites where significant faulting occurred were the three sites with undowelled joints: St. Cloud, Minnesota (MN 7), Pewaukee, Wisconsin (WI 1), and Green Bay, Wisconsin (WI 2). At the MN 7 site, the average joint faulting in the hot-pour–sealed section was 4.73 mm (0.19 in.), and the average joint faulting in the unsealed section was 3.77 mm (0.15 in.). Constructed in 1954, this section is by far the oldest experimental site surveyed for this study. Also, although the available project data do not provide a record of past restoration work, the appearance of the pavement surface (i.e., the visibility of large aggregates) suggests that the pavement was diamond ground at some time in the past. There is also some question as to whether or not the unsealed section was truly unsealed over its entire 50 years or more of service.
At the Pewaukee (WI 1) site, average joint faulting for the three silicone-sealed sections was 6.16 mm (0.24 in.), and the average joint faulting for the one unsealed section was 4.79 mm (0.19 in.). This is the second oldest site in the study, constructed in 1983.
At the Green Bay (WI 2) site, the average joint faulting in the undowelled section with preformed sealant was 4.19 mm (0.16 in.), and the average joint faulting in the undowelled, unsealed section was 4.81 mm (0.19 in.). At the same site, the average joint faulting in the dowelled section with preformed sealant was 0.77 mm (0.03 in.), and the average joint faulting in the dowelled, unsealed section was 0.96 (0.04 in.). This is the third oldest site in the study, constructed in 1988.
Of the remaining projects (all of which contain dowelled joints), average joint faulting levels exceeded 1 mm (0.04 in.) at only one site: Spring Valley, Minnesota (MN 3). At the MN 3 site, the average joint faulting in the silicone-sealed section was 1.31 mm (0.05 in.), the average joint faulting in the section with preformed sealant was 0.99 mm (0.04 in.), and the average joint faulting in the section with unsealed joints was 1.16 mm (0.05 in.). At all of the remaining sites, the average joint faulting levels were less than 1 mm (0.04 in.).
A statistical analysis of the collected faulting data also was completed to determine if there was any statistical difference between faulting on sealed joints (i.e., hot-poured, preformed, and silicone) and faulting on unsealed joints. Using a 95 percent confidence interval, no significant difference in average joint faulting was detected between sealed joints of any kind and unsealed joints. It should be noted that with few exceptions the magnitudes of average joint faulting were very low, most less than 1 mm (0.04 in.). The notable exceptions were the few sites with undowelled joints.
Joint Sealant Failures
There are two sites at which all three categories of sealant (silicone, preformed, and hot pour) were used: Mesa, Arizona (AZ 1) and Athens, Ohio (OH 1). At both of these sites, a considerable degree of joint sealant failure had occurred in the test sections with at least one of these types of sealant. At the AZ 1 site, the average percentages of joint length with one or more types of sealant distress (adhesive failure, cohesive failure, or absence) were 8 percent, 41 percent, and 50 percent for the silicone, preformed, and hot-pour sections, respectively. At the OH 1 site, the average percentages of joint length with one or more types of sealant distress (adhesive failure, cohesive failure, or absence) were 38 percent, 18 percent, and 57 percent for the silicone, preformed, and hot-pour sections, respectively. Thus, the hot-pour sections had the greatest amount of sealant failure at both sites, while at one site, silicone failed by a smaller percentage than preformed, and at the other, preformed failed by a smaller percentage than silicone. At some of the other sites, the sealed-joint section or sections experienced little or no sealant failure of any kind, while at others, the sealed-joint section or sections experienced significant to substantial sealant failure.
Low-severity joint spalling was analyzed by site and by joint sealant type. A detailed statistical analysis of the spalling results indicates that the low-severity spalling at unsealed-joint test sections was significantly greater than the low-severity spalling at each of the three types of sealed-joint test sections at the same sites. An overall confidence level of 95 percent was used for this statistical analysis.
Medium- and high-severity joint spalling was analyzed by site and by joint sealing type. The mean differences between medium- and high-severity joint spalling in the unsealed-joint test sections and that in each of the three kinds of sealed-joint test sections were negative, indicating that medium- and high-severity joint spalling in the three types of sealed-joint test sections was greater than that in the unsealed-joint test sections. However, this difference was statistically significant only in the case of preformed sealant.
For this study, deflection testing was conducted on 89 different pavement sections. Deflection testing could not be conducted on all 117 sections due to site constraints. All data from deflection tests taken at the transverse joints (mid-slab and corners), slab interior, and along the longitudinal edge were used in the analyses, as described below.
1. Analysis of transverse joint deflections at the mid-slab location: The transverse joint deflections at the mid-slab location (approach or leave side of joint) were first used to determine the joint load transfer efficiency. The transverse joint deflection load transfer efficiency is computed as the simple ratio of unloaded and loaded slab deflections, and provides a measure of the competence of dowel bar and/or aggregate interlock interactions to effectively transfer edge loadings between adjacent slabs.
Transverse joint deflections also were used to determine the normalized total joint deflection, dynamic edge foundation support, and transverse edge slab support ratios. The normalized total edge deflection is computed as the simple addition of unloaded and loaded slab deflections, normalized to a common load level of 40 kN (9,000 lbf). The normalized total edge deflection should remain relatively constant regardless of available deflection load transfer, provided that slab thickness, elastic modulus, and foundation support remain constant. The total edge deflection can be used as a relative indicator of the overall edge structural capacity of a test section as well as an input for the backcalculation of edge foundation support.
Incremental analysis of transverse edge deflection response was conducted to provide a means of differentiating slab curling from poor foundation support. In those cases where temperature curling alone was responsible for poor support, incremental slab support should increase over that computed based on individual load levels, provided at least two load levels produced sufficient total edge deflection to close any curl-induced voids.
The uniformity of support under the transverse edge, termed the transverse edge slab support ratio, is computed as the ratio of the backcalculated incremental edge and interior dynamic foundation k-values (Crovetti 1994). In general, incremental transverse edge slab support ratios less than approximately 0.75 are indicative of slabs with poor edge support due to foundation densification/pumping and/or temperature curling (Crovetti 1994).
2. Analysis of interior slab deflections: The interior deflections collected at the center-slab test location were used to determine the dynamic foundation support k-value. Interior deflections were first used to compute the deflection basin area, which was then used to backcalculate an initial estimate of the radius of relative stiffness of the pavement system. A slab size correction factor for the estimated radius of relative stiffness was computed, and the effective slab length was computed based on the actual slab dimensions and the transverse joint load transfer efficiency. After computation of slab size correction factors, the adjusted radius of relative stiffness and the dynamic interior foundation k-value were computed.
As with the edge deflections, an incremental analysis was also conducted to provide a means of differentiating slab curling from poor foundation support. For those cases where the slab temperature gradient (top temperature – bottom temperature) is excessively positive and foundation support stiffness is high, the center of the slab may be lifted off the foundation. In these cases, the maximum deflection and the deflection basin area term increase, resulting in a reduced backcalculated foundation k-value. If, however, at least two of the load levels used during testing were sufficient to create maximum surface deflections exceeding the depth of curling-induced voids, incremental analysis should indicate an increased dynamic foundation k-value and a decreased effective slab thickness as compared to values backcalculated from individual load/deflection pairs.
3. Analysis of longitudinal joint deflections: The longitudinal joint deflections were used to determine the normalized total edge deflection, dynamic edge foundation support, and longitudinal edge slab support ratios. For slabs with asphalt concrete or gravel shoulders, the total edge deflection is equal to the maximum deflection recorded at the longitudinal edge. For slabs with portland cement concrete (PCC) shoulders (tied or untied), the normalized total longitudinal edge deflection is computed as the simple addition of unloaded and loaded slab deflections, normalized to a 40-kN (9,000-lbf) load level.
Incremental analysis of longitudinal edge deflection response was conducted to provide a means of differentiating slab curling from poor foundation support. The incremental normalized total longitudinal edge deflections were used to compute incremental longitudinal edge slab support. In those cases where temperature curling alone was responsible for poor support, incremental slab support should increase over that computed based on individual load levels, provided at least two load levels produced sufficient total longitudinal edge deflection to close any curl-induced voids. The uniformity of support under the longitudinal edge, termed the longitudinal edge slab support ratio, is computed as the ratio of backcalculated incremental longitudinal edge and interior dynamic foundation k-values. In general, incremental longitudinal edge slab support ratios less than approximately 0.75 are indicative of slabs with poor longitudinal edge support due to foundation densification/pumping and/or temperature curling.
4. Analysis of corner deflections: The corner deflections were used to determine the normalized total corner deflection, dynamic corner foundation support, and corner slab support ratios. For slabs with asphalt concrete or gravel shoulders, the total corner deflection is equal to the maximum deflection recorded on the loaded slab corner plus the deflection recorded on the unloaded slab corner. For slabs with PCC shoulders (tied or untied), the normalized total corner deflection is computed as the addition of unloaded and loaded corner slab deflections, normalized to a 40-kN (9,000-lbf) load level, modified by the longitudinal joint load transfer.
Incremental analysis of corner deflection response was conducted to provide a means of differentiating slab curling from poor foundation support. In those cases where temperature curling alone was responsible for poor support, incremental slab support should increase over that computed based on individual load levels, provided at least two load levels produced sufficient total longitudinal edge deflection to close any curl-induced voids. The uniformity of support under the slab corner, termed the corner slab support ratio, is computed as the ratio of backcalculated incremental corner and interior dynamic foundation k-values (Crovetti 1994). In general, incremental corner slab support ratios less than approximately 0.75 indicate slabs with poor corner support due to foundation densification/pumping and/or temperature curling (Crovetti 1994).
A review of the computed slab support ratios provides the following observations:
The slab support ratios along the transverse joints are believed to be useful in identifying condition of the slab support that might be attributable to differences in joint features such as presence or absence of sealant or type of sealant. In general, poor slab support is no more or less common in the unsealed sections than in the sealed sections.
Summary of Findings
A summary of the most significant findings from this study is included below. When evaluating the findings, the age of the test sections studied should be considered. As indicated previously, most of the test sections studied were 10 years or less in age at the time of data collection.
Crovetti, J. A. 1994. Design and Evaluation of Jointed Concrete Pavement Systems Incorporating Free-Draining Base Layers, Ph.D. dissertation, University of Illinois at Champaign-Urbana.
Miller, J. S., and W. Y. Bellinger. 2003. Distress Identification Manual for the Long-Term Pavement Performance Program, 4th rev. ed. (Report No. FHWA-RD-03-031). Federal Highway Administration, Washington, DC.
The Concrete Pavement Technology Program (CPTP) is an integrated, national effort to improve the long-term performance and cost-effectiveness of concrete pavements. Managed by the Federal Highway Administration through partnerships with State highway agencies, industry, and academia, CPTP’s primary goals are to reduce congestion, improve safety, lower costs, improve performance, and foster innovation. The program was designed to produce user-friendly software, procedures, methods, guidelines, and other tools for use in materials selection, mixture proportioning, and the design, construction, and rehabilitation of concrete pavements.
PDF files can be viewed with the Acrobat® Reader®