|FHWA > Engineering > Pavements > HIF-08-009 > Chapter 7. Instrumentation and Evaluation|
Construction of a Precast Prestressed Concrete Pavement Demonstration Project on Interstate 57 Near Sikeston, Missouri
Chapter 7. Instrumentation and Evaluation
As part of an evaluation of the FHWA PPCP demonstration projects, performance monitoring is conducted through condition surveys of the completed project and, whenever possible, instrumentation of the pavement section. For the I-57 demonstration project, an extensive instrumentation and monitoring program was conducted by the University of Missouri-Columbia through a separate research contract with MoDOT. This chapter will discuss the performance monitoring aspects as well as the instrumentation and monitoring program, including key findings from instrumentation and testing.
Project-Level ConditIon Survey
Soon after the pavement was opened to traffic, a project-level condition survey was performed to establish the initial as-constructed condition of the precast pavement. This initial project-level condition survey will be used for comparison with future condition surveys to identify any new distresses that occur over time. For this condition survey, a distress map was developed, noting four types of distresses visible after construction:
The distress map is shown in Figure 59 at the end of this chapter, and the as-constructed condition is summarized as follows:
Figure 52. Photo. Longitudinal crack near the pavement centerline extending away from one of the instrumentation blockouts.(18)
Figure 53. Photo. Typical minor spalling observed.
Figure 54. Photo. Typical condition of expansion joints 1, 2, 3, and 5 in service.
Despite the significant number of distresses observed, most should have little or no effect on pavement performance. The majority of the transverse cracks were sealed with epoxy at the precast plant to prevent infiltration of water and deicing salts (which are only used occasionally on this section of I-57). Both longitudinal and transverse cracks should also be adequately held closed by the prestressing in both directions. The deteriorated header material at the surface of either side of the expansion joint between sections 3 and 4 will likely need replacement in time, but is performing adequately at present. The dowels for this joint are intact, and as described below, deflection testing across this expansion joint showed load transfer as good as or better than the other expansion joints.
Minimization or elimination of slab cracking is one of the primary benefits of incorporating prestress into a precast pavement system. Despite prestressing in the I-57 demonstration project, both transverse and longitudinal cracking occurred, and the possible causes deserve some consideration, as discussed below.
Possible Causes of Transverse Cracking
The majority of the transverse cracks, which were almost exactly at mid panel for most of the cracked precast panels, and almost perfectly parallel to the panel edges, were observed at the fabrication plant. This indicates that some aspect of the fabrication process led to some or all of the transverse cracking that occurred.
One possible cause is thermal shock. As will be described below, instrumentation in the precast panels showed that the highest levels of strains occurred during the curing process.(18) Steam curing resulted in concrete temperatures in excess of 82 °C (180 °F) after casting the panels. The panels were generally removed from the forms in the early morning when ambient daily temperatures were at a minimum, resulting in significant thermal strains in the precast panels. High thermal strains are not necessarily the cause of cracking, however, but rather restraint of these thermal strain movements. If the precast panels are in an "expanded" state due to heat from the curing process and are subjected to rapid cooling due to low ambient temperatures, they will try to "contract." If this contraction is restrained, tensile stresses that are high enough to induce cracking can occur.
Restraint during the fabrication process (in the transverse or short axis dimension) could be caused by the casting bed itself, either from the sideforms that are fixed to the casting table, or from friction between the bottom of the precast panels and the casting table, or from the pretensioning strands prior to their being released from the anchoring abutments. The restraint could also have occurred when the panels were stacked for storage at the fabrication plant. Dunnage between the precast panels, combined with the weight of the panels bearing down on one another, could have provided high levels of restraint as the panels were "cooling" in the storage location. Unfortunately, concrete temperatures were not monitored after the panels were removed from the forms, and stacking order documentation was not compiled to see if this could be a potential cause. What was measured, however, were the thermal strains that occurred during the curing process.
Another possible cause or contributor to the transverse cracking is Poisson effect strains in the transverse direction resulting from release of the pretensioning strands in the longitudinal direction. Poisson effect strains are essentially strains in the transverse direction opposite to the compressive strains induced in the longitudinal direction from the pretensioning force. For concrete, a Poisson ratio of 0.2 is generally used, which means that tensile strains in the transverse direction will be roughly 20 percent of the magnitude of the compressive strains in the longitudinal direction. Since pretensioning induces compressive strains in the longitudinal direction, tensile strains can be expected in the transverse direction. This effect was confirmed from strain gage instrumentation in the panels, monitored during pretensioning release, as described below.(20) As before, however, these strains do not necessarily lead to cracking, unless they are restrained by external forces, as described above.
It should be noted that the steam curing process is well established within the precast concrete industry and at the CPI fabrication plant. Curing and pretensioning release processes employed by CPI are established and accepted practices, as is pretensioning long, flat precast panels. Precast panels of similar size were used on previous demonstration projects in Texas and California, but no cracking occurred (steam curing was used in California but not Texas). In all likelihood, a simultaneous occurrence of multiple conditions (thermal shock, Poisson effect, and others) led to the observed cracking. This experience will help to ensure these issues are addressed and do not occur on future projects.
It should also be noted that while much of the transverse cracking was observed at the fabrication plant, construction conditions may have exacerbated the cracking. As described in a previous chapter, the panels were installed over several days and epoxied together. Post-tensioning was not completed for several days following panel installation. As such, the epoxied panels may have acted as a long monolithic pavement slab with nothing to prevent transverse cracking from occurring in the panels themselves.
One possible measure which may have helped mitigate the extent of cracking is the inclusion of "temperature and shrinkage steel" reinforcement in the precast panels in the transverse direction. The ACI 318-95 Building Code recommends a minimum of 0.18 percent reinforcement (Grade 60) for structural slabs where the flexural reinforcement (pretensioning steel) extends in one direction only.(14) For the I-57 precast panels, this would have required 23 No. 5 or 16 No. 6 reinforcing bars per panel, which would not have increased the cost of the panels significantly (< 113 kg [250 lb] of additional reinforcing steel per panel). While the inclusion of reinforcing steel does not prevent cracking from occurring, it helps to hold cracks tightly closed, and it may also prevent cracks from propagating through the full depth of the panel.
Possible Causes of Longitudinal Cracking
While not as extensive as the transverse cracking in the pavement section, longitudinal cracking was also unexpected and had not been observed on previous demonstration projects. There are several possible causes of the longitudinal cracking, and in all likelihood, a combination of these factors resulted in the observed cracking.
As described previously, the longitudinal cracking near the crown of the pavement propagated away from blockouts cast into the precast panels for instrumentation of the post-tensioning strands. Squared corners used for these blockouts likely contributed to the observed cracking. A similar phenomenon was observed in a cast-in-place, post-tensioned pavement constructed near Waco, Texas, in 1985, as longitudinal cracking propagating away from squared corners of the post-tensioning blockouts was observed shortly after construction.(21)
Another potential cause or contributing factor to longitudinal cracking is curling of the pavement slab in the transverse direction. Temperature gradients in concrete pavement slabs, particularly during the early morning and early afternoon periods, can result in significant slab curling, which when resisted by the self-weight of the pavement slab, can cause bending stresses and either top-down or bottom-up cracking. In the transverse direction, there is essentially nothing restraining curling movement at the ends of the precast panels, and therefore curling stresses are likely. A finite element analysis performed by researchers from the Indiana Department of Transportation estimated that tensile curling stresses at the top of the PPCP slab near the pavement crown could be as much as 2 MPa (290 lbf/in2) when the pavement is subjected to a negative (top of slab cooler than bottom) temperature gradient.(22) This analysis did not account for the counteracting compressive stress from pretensioning, but is still of significant magnitude. Additionally, if these curling stresses truly were large enough to cause longitudinal cracking, it would be expected to occur along the full length of all four sections of pavement, and to date this is not the case.
Potentially adding to the curling effect is the eccentricity of the pretensioning strands through the center portion of the precast panels. Over this section (traffic lanes), the pretensioning strands were centered 100 mm (4 in.) from the bottom of the precast panel, which is below the centroid of the pavement cross section for this region. This eccentricity could potentially cause downward bending of the precast panels, reducing the effective compressive stress from pretensioning in the top of the pavement.
Another potential contributing factor to transverse panel curling is the draped pretensioning strands in the precast panels. As described in a previous chapter, because of the nonuniform panel thickness, it was necessary to drape the top pretensioning strands within the end portion (shoulder region) of the precast panels. Draping causes a pretensioning force near the ends of the panels that is not parallel to the bottom of the precast panels, and may have caused a slight upward curvature, similar to that caused by temperature curling. This curvature, when resisted by the weight of the precast panels, can result in tensile stresses in the top of the precast panels and top-down cracking.
Finally, another potential contributor to the longitudinal cracking could have been the sand used to fill the ruts in the permeable base during construction. If the sand was not perfectly leveled with the surrounding base, it could have created high points on which the precast panels rested, resulting in a bending stress in the panels.
As with transverse cracking, longitudinal cracking of this nature had not been observed on previous projects. And as with the transverse cracking, a simultaneous combination of circumstances (curling, squared blockout corners, pretensioning eccentricity) could have resulted in the observed cracking. Simple measures, however, such as rounding the corners of instrumentation, stressing pocket blockouts, and carefully examining potential curing stresses, may prevent the formation of such cracks on future projects.
As a result of the extensive cracking observed in the finished pavement section, MoDOT conducted deflection testing to check for anomalies in structural condition as a result of the cracking. The deflection testing was used to check load transfer across joints and cracks. Poor load transfer across a crack, in particular, could indicate a potential future structural failure.
FWD testing was conducted on September 12, 2006, approximately 8 months after the pavement had been opened to traffic. FWD deflection testing was conducted at the following locations:
Deflections were measured at three load levels for each test, and tests were conducted on both the "approach" and "leave" sides of each crack and joint. The results of the deflection testing, as analyzed and provided by MoDOT, are summarized in Table 9. Deflection testing of transverse cracks was conducted on all four post-tensioned sections, with five measurements from the first section, three from the second, one from the third, and four from the final section. As these results show, with only one exception, load transfer across the transverse cracks in the individual panels was greater than 88 percent. This indicates that, despite some of the cracks extending through the depth of the precast panels, the aggregate interlock, which is enhanced by the longitudinal post-tensioning, is adequate to prevent differential deflections on either side of the cracks.
For deflection testing of joints between individual panels, at least three joints were selected from each post-tensioned section. With the exception of 3 of the 19 joints tested (shown in italics in Table 9), all showed very good load transfer, with most measuring in excess of 85 percent. The three joints exhibiting below-average load transfer (the Long Term Pavement Performance Program recommends 60 percent as the threshold for load transfer restoration) had roughly the same load transfer for the approach and leave sides of the joint, indicating a possible void beneath the joint. Although it has not been verified with inspection data, it is believed that these three joints were likely shimmed during the panel installation process, as discussed in previous chapters. Shimming, which was used to correct deviation in centerline alignment of the pavement, likely kept the joints from closing completely during post-tensioning. This shimming likely prevented the epoxy in the joint from bonding the panels together, resulting in a "loose" keyway joint. Also of note, with the exception of two of the joints showing good load transfer, those with below-average load transfer exhibited significantly higher raw deflections than those exhibiting good load transfer, further confirming that voids were likely present.
The load transfer results for the expansion joints were also generally favorable. With the exception of measurements on the approach side of joints 2 and 3, load transfer measured greater than 80 percent. The cause of low load transfer at these two approach joints is not known for certain, but could be due to voids beneath the joint panels at these locations as grout was not used to fill voids beneath the pavement to prevent clogging the underlying permeable base. Past experience has shown, however, that precast panels tend to "settle" into flexible bituminous bases over time, which would help to eliminate any voids that may be present beneath the I-57 project. Fortunately, measurements on the "leave" side of these two joints showed good load transfer, indicating that if a void is present, it is likely only under the approach side of the joint.
MoDOT will continue to monitor the performance of the joints and cracks on this section of PPCP. If faulting, spalling, or other distresses are observed at these joints and cracks, additional investigation and possible mitigation will be required. At present, however, these joints and cracks do not present any performance problems for this section of pavement.
Summary of Instrumentation Program
The purpose of the instrumentation program conducted by MoDOT and the University of Missouri-Columbia was to monitor pavement performance throughout construction and in service to verify assumptions that were made during the design process and to evaluate the overall PPCP process. Key aspects of the instrumentation program were as follows:
The results of instrumentation and monitoring of these aspects will help to provide a better understanding of stresses in the PPCP system during the different phases of construction and in service. This includes a better understanding of prestress forces and the losses associated with pretensioning and post-tensioning. This information will help in the development of the design procedures for future PPCP projects in Missouri and elsewhere. Instrumentation also provided valuable information in assessing the possible causes of the distresses observed in the finished pavement, so that they can be avoided in the future.
Temperature sensors and strain gages were the primary devices used for instrumentation. Seven precast panels, six of them from one section of the pavement, were instrumented, as shown in Figure 55, below. Section 3 was selected for instrumentation primarily because it was not one of the end sections, which could be affected by the adjacent pavement. Both joint panels (A31 and A32), three base panels (B1, B2, and B3), and the anchor panel (C1) were instrumented. Additionally, base panel B4 from section 4 was also instrumented for comparison of behavior.
Strain gages embedded in the instrumented panels included 39 instrumented rebars, 14 vibrating wire strain gages, 4 strandmeters (for post-tensioning), and 38 thermocouples. Strain gages were oriented in both the longitudinal and transverse directions, both at mid-depth and at varying depths in the panels. Thermocouples were located primarily at mid-depth and close to the top and bottom of the precast panels to examine temperature gradients over the depth of the panels. Strandmeters were mounted to the longitudinal post-tensioning tendon at the pavement crown just prior to the post-tensioning operation for Section 3. Figure 56 shows the instrumentation layout for Panel A32, showing the locations of instrumented rebars, thermocouples, and vibrating wire strain gages. Figure 57 shows a typical instrumented rebar and vibrating wire strain gage in the forms prior to concrete placement.
Instrumentation was monitored during the panel fabrication process (until removal of the panels from the forms), and on-site for approximately 18 months. All wiring for the instrumentation was routed from a junction box in each instrumented panel (Figure 58) to a cabinet next to the project site that contained the data acquisition equipment (Figure 58) and a modem for remote data collection. For additional details on the instrumentation program and analysis of the results, see the University of Missouri-Columbia Master of Science theses for Dailey,(20) Davis,(16) and Luckenbill.(18)
Figure 55. Illustration. Overall layout of the instrumented precast panels from sections 3 and 4.(16)
Figure 56. Illustration. Instrumentation layout for panel A32 (R = instrumented rebar, T = thermocouple, V = vibrating wire strain gages).(20)
Figure 57. Photo. Instrumented rebar and vibrating wire strain gages for measuring concrete strain.
Figure 58. Photo. Junction box within an instrumented precast panel (left) and instrumentation cabinet next to the project site for collecting and uploading data (right).
Key Findings from Instrumentation Program
The instrumentation program provided a number of key findings. These findings confirmed certain assumptions made during the design process and showed where improvements could be made in the design process. Additionally, the instrumentation helped to ascertain the possible causes of some of the observed distresses.
As part of the instrumentation program, properties of the concrete mixture used for the precast panels were measured and monitored over time:
Table 10 provides a summary of the concrete properties monitored. Some of the key findings from monitoring of concrete properties are listed below:
Recommendations Based on Instrumentation Program
Figure 59. Illustration. Project Level condition survey after opening to traffic.