Project 
Model/AQC Studied  

Strength Prediction Model 
Spalling Model 
Faulting Model 
Cracking Model  
3day Cyl. Compressive  3day Core Compressive  3day Cyl. Splitting Tensile  3day Core Splitting Tensile  Entrained Air Content  Joint Edge Spalling  Consolidation Around Dowels  Load Transfer Efficiency  Tie Bar Depth  Load Transfer Efficiency  
Meade, KS—Meade Municipal Airport 
Ö 
Ö 
Ö 
Ö 

Shewano, WI—Route 29 East 
Ö 
Ö 
Ö 
Ö 

St. Johns, MI—Route 27 North 
Ö 
Ö 
Ö 
Ö 

Ottumwa, IA—Route 23 South 
Ö 
Ö 
Ö 
Ö 

Des Plaines, IL—Route 58 West 
Ö 
Ö 
Ö 
Ö 

Benton Harbor, MI—I94 West 
Ö 
Ö 
Ö 
Ö 

Philo, IL—Route 130 South 
Ö 
Ö 
Ö 
Ö 

Omaha, NE—I80 East & West 
Ö 
Ö 
Prediction of 28day Flexural Strength The prototype PRS recommends using core testing as an indirect measure of in situ flexural strength.^{(1,2,3)} Cores are taken at some reference maturity equivalent (for example, 3 days at 22 °C). The mean compressive or splitting tensile strength of the cores from each sublot is adjusted to a 28day strength under standard laboratorycured conditions. This equivalent mean 28day compressive or splitting strength is then converted to a thirdpoint loading flexural strength using an approved interstrength relationship developed (prior to construction) from the specific concrete mixture and projectapproved aggregates, cement, and admixtures used during construction. The procedure was roughly outlined in the previous PRS study and evaluated during the prediction variability portion of this investigation, as described in appendix D of this report.^{(1,2,3)} The developed prediction procedure is necessary because the concrete fatigue model used in the PRS is based on 28day, standard cure flexural strength. The effects of flexural strength on fatigue life performance are welldocumented in the literature. Therefore, while no further field investigations of the effects of flexural strength on pavement performance were performed during the current project, a laboratory study was completed to better explore the errors associated with estimating a 28day standard cure flexural strength from early age strengths. Testing Procedure The previous PRS study compared core and cylinder compressive strength tested at ages of 7, 14, and 28 days for specimens cured under identical curing conditions.^{(1,2,3)} The study concluded that there is no statistical difference between core and cylinder compressive strengths at any of the three investigated maturities. However, the study did not include data from specimens tested at an equivalent 3day standard laboratory maturity. Therefore, additional laboratory testing was performed to establish whether there is a statistical difference between 102mm diameter core and 152mm diameter cylinder compressive and splitting tensile strength at 3 days. Different coarse aggregate types and cements were sampled from each of the four projects used for the strength prediction model study (see table 66). Each coarse aggregate was used with two cement contents to give a total of eight mixes. The concrete mixes were batched in accordance with the American Society for Testing and Materials (ASTM) C 192, Standard Practice for Making and Curing Concrete Test Specimens in the Laboratory. Similar to the previous PRS study, 12 cylinders were cast for each mix and moistcured for 3 days. Six cylinders were cored just before testing. The core lengths were trimmed to meet the required lengthtodiameter ratio of 2 to 1 established by ASTM C 42, Standard Test Method for Obtaining and Testing Drilled Cores and Sawed Beams of Concrete. Along with this standard, strength testing was performed in the laboratory in accordance with the following ASTM specifications:
Compressive strength (three cores and three cylinders) and splitting tensile strength (three cores and three cylinders) were determined on concrete made from the same batch and cured under identical conditions. In addition to the laboratory data collected, results of field samples obtained during the variability investigation were also summarized. For some of the field projects evaluated, cylinders were cast and cored to obtain a cylindertocore relationship. Because curing conditions were identical, the relationship is similar to the lab study. In addition, cores were drilled from the field pavement at some sampling locations. Because the variability of this relationship is slightly increased due to inconsistencies between specific batch constituents, the average cylinder and core strengths were evaluated and presented. Summary of Strength Results The results of the cylindertocore relationships are plotted in figures 73 and 74 for the compressive and splitting tensile strength data, respectively. On each graph, the bestfit linear equations are also presented. As described in the previous section, relationships were made between cylinders and cored cylinders, cylinders and cored slabs, and overall data sets. All three equations are given in the figures.
In the previous PRS study, it was concluded that there is no significant difference between core and cylinder compressive strengths, irrespective of mix design, coarse aggregate hardness, and coarse aggregate geometry.^{(1,2,3)} Since the data again appear to be clustered around the line of equality (1 to 1), paired ttests were done to determine if there is a statistically significant difference between matched pairs of core and cylinder compressive strengths. Twotailed ttests were performed on data sets of 44 cored cylinders, 44 cored slabs, and 88 overall pairs. In the analyses, the tvalues were calculated to be 0.049, 0.222, and 0.099 for the cored cylinders, cored slabs, and overall data, respectively. Therefore, the null hypothesis that there is no statistically significant difference between core and cylinder compressive strengths could not be rejected at the 5percent significance level for any of the compressive strength data sets. For completeness, paired ttests were also performed to confirm that there is a statistically significant difference between matched pairs of core and cylinder splitting tensile strengths. Twotailed ttests were performed on data sets of 43 cored cylinders, 36 cored slabs, and 79 overall pairs. In the analyses, the tvalues were calculated to be 3.209, 8.631, and 5.569 for the cored cylinders, cored slabs, and overall data, respectively. Therefore, the null hypothesis that there is no statistically significant difference between core and cylinder compressive strengths was rejected at the 5percent significance level for all of the splitting tensile strength data sets. Examination of the overall distribution of the compressive strength cylindertocore ratios showed an approximately bellshaped distribution around 1.0. The 88point relative frequency histogram is shown in figure 75. However, as expected, an examination of the splitting tensile strength cylindertocore ratios showed a very skewed curve distributed around 0.75. The 79point relative frequency histogram is shown in figure 76.
The current transverse joint faulting, transverse joint spalling, and transverse cracking distress indicator models require a 28day flexural strength. These current distress indicator models are described in detail in the section titled Summary of Current Distress Indicator Models. Effect of Entrained Air Content on Transverse Joint Spalling Introduction The effects of entrained air content on pavement performance were not welldocumented in the collected literature. However, under a previous PRS study, a laboratory materials study was conducted to evaluate the effects of air void system parameters on transverse joint scaling/spalling percentages.^{(1,2,3)} That research resulted in three different relationships in which joint spalling was a function of different combinations of entrained air content, void spacing factor, compressive strength, freezethaw cycles (at a depth of 76 mm below the pavement surface), and presence of deicing salt. (Note: These relationships and their inputs are presented in English units.) The first model was a function of entrained air content (not void spacing factor), as not all agencies collect hardened air content data. This developed relationship is presented in equation 14.^{(1,2,3)}
where
The second model estimated joint spalling as a function of both entrained air content and void spacing factor. This relationship is presented in equation 15.^{(1,2,3)}
where
The third model estimated joint spalling as a function of void spacing factor and not entrained air content. This relationship is presented in equation 16.^{(1,2,3)}
where
In order to use these relationships with confidence, they need to be verified under actual field conditions. Under the preceding PRS study, two alternatives were recommended for additional field investigation to accomplish this task.^{(1,2,3)} The first option recommended using inservice pavement sections that exhibited joint spalling due to deficient air void parameters and consolidation levels. The various air void parameters, effects of deicer salts, and the estimated number of freezethaw cycles could then be correlated with joint spalling. This information could be used to modify or verify the previous PRS study performance models. The advantage of this type of study is that a fieldcalibrated performance model could be incorporated into the PRS. The second option involved the construction and longterm monitoring of specially constructed test sections. The advantage of this option is that the developed performance model would be based on longterm field performance. The disadvantage is that results would not immediately be available (due to longterm monitoring) and certain pavement sections are essentially sacrificed ("destined" to fail because extremely low air contents would be used in order to observe accelerated deterioration). The research team decided to verify (or calibrate) the entrained air contentbased joint spalling relationships using field site evaluations. The testing procedures and model development are described in the following sections. Sampling and Testing Procedures For this research study, inservice pavements were used to calibrate the spalling performance models developed in a preceding PRS study.^{(1,2,3)} Sampling was performed at three pavement joint repair projects, as shown in table 66. The same three projects were used for the consolidation and air content studies for variability. A number of joints from each project, exhibiting varying degrees of joint spalling, were selected for the evaluation. Joint repair projects were investigated because they allowed the retrieval of a sufficient number of 102mmdiameter cores. At each joint evaluated, a sufficient number of cores were removed from between the dowel bars so that laboratory analyses of hardened air void parameters could be performed. The tests were run in accordance with ASTM C 457, Standard Test Method for Microscopical Determination of Parameters of the Air Void System in Hardened Concrete. In addition, cores were removed to determine the compressive strength of the concrete by ASTM C 42, Standard Test Method for Obtaining and Testing Drilled Cores and Sawed Beams of Concrete. As seen in equations 14 through 16, the laboratorydeveloped relationships were functions of the presence of deicer salt (calcium chloride).^{(1,2,3)} However, all three projects used to verify the relationship were in areas where deicer salts were regularly applied. Therefore, the SALT variable was effectively removed from the current investigation. In the previously developed equation, the use of salt was incorporated as "1" and the nonuse of salt was included as "0," thus eliminating the salt term. For the current study, SALT was always "1", so the SALT multiplier was simply added to the equation constant. The final factor necessary to develop/calibrate the joint spalling model was the number of freezethaw cycles experienced by the pavement. Because the previous study was controlled in the laboratory, the exact number of freezethaw cycles was determined with the use of thermocouples placed 76 mm below the surface of the deicer blocks. All the field sites evaluated in the current study were constructed more than 20 years ago, and the exact number of freezethaw cycles was unknown. Therefore, an estimate of the number of freezethaw cycles was calculated using the climatic model incorporated into the ClimaticMaterialsStructural (CMS) computer program.^{(64)} Atmospheric data obtained from weather stations located near the three projects for the past 3 years were entered into the CMS program. The program output includes the number of freezing and thawing cycles that occur at any depth in the concrete based on the amount of weather station data input. To mirror the previous laboratory study, a freeze occurred for each pavement when the temperature calculated at a depth of 76 mm below the surface stayed below 0 °C for at least 48 hours. The pavement is assumed to be thawed when the temperature at 76 mm below the surface rises above 0 °C. The CMS program was then used to predict the number of freezethaw cycles that occurred for each project for the years 1994, 1995, and 1996. Based on studies published by Dempsey, the resulting number of annual pavement freezethaw cycles for the projects evaluated seem reasonable. This average annual number of cycles was then multiplied by the age of the pavements studied to compute the total number of cumulative freezethaw cycles for each project. Transverse Joint Spalling Model Calibration The collected data were used to calibrate the previously developed model predicting the percent of joint that is spalled as a function of compressive strength, freezethaw cycles, and air void system parameters. For simplicity, the model incorporating entrained air content is preferred over the models based on other air void system parameters. The air content model would allow the use of plastic air content data to be used as opposed to costly linear traverse testing. The collected data for the model calibration is summarized in table 67. Figure 77 shows a plot of the predicted joint spalling using the previously developed equation with entrained air content as the only air void parameter (equation 14) versus the fieldmeasured joint spalling. Three new air void parameterbased regression equations were derived using the data collected for this investigation. Table 67. Data collected for variables related to percent joint spalling.
The first calibrated model, giving joint spalling as a function of entrained air content only, is presented as equation 17.
where
The collected data used in the development of equation 17 are presented in figure 78. A plot showing the sensitivity of equation 17 (function of only entrained air content) is presented in figure 79.
The second calibrated model, giving joint spalling as a function of both entrained air content and void spacing factor, is presented as equation 18.
where
The collected data used in the development of equation 18 are presented in figure 80.
The third calibrated model, estimating joint spalling as a function of void spacing factor and not entrained air content, is presented as equation 19.
where
The collected data used in the development of equation 19 is presented in figure 81.
The complete procedure making up the current transverse joint spalling model is described in detail in the section titled Summary of Current Distress Indicator Models. Effect of Percent Consolidation Around Dowels on Transverse Joint Faulting Introduction The effects of percent consolidation level around dowels on transverse joint faulting are not welldocumented in the literature. The three projects evaluated for the calibration of the spalling prediction model were also evaluated for consolidation level at doweled joints as shown in table 66. Different sections within each project exhibiting a range of distress (faulting or loss of load transfer) were selected for the evaluation of the effect of consolidation level on load transfer across doweled joints. The relationship was integrated into the current faulting prediction model, which is a function of concrete bearing stress (BSTRESS). Sampling and Testing Procedure To measure the in situ consolidation level of concrete pavements at doweled joints, density was measured from drilled cores as previously detailed in the consolidation variability study. For each project, several joints exhibiting varying degrees of deterioration were selected as part of the study. Cores were drilled through dowel bars, between dowel bars, and away from the joint to evaluate relative variability. Core densities were measured in a saturated surfacedried condition in accordance with ASTM C 642, Standard Test Method for Specific Gravity, Absorption, and Voids in Hardened Concrete, to evaluate variability in consolidation incorporating the effects of basket assemblies and dowel bars (above and below). The core removed away from a joint with the highest density within a sampled lot was assumed to be 100percent consolidated. The remaining core densities were compared to the 100percent consolidated density to determine the relative consolidation levels. This procedure incorporates overconsolidation effects when vibration time is extended over basket assemblies or when vibrators contact dowels for longerthannormal periods. Load transfer efficiency was directly measured using a fallingweight deflectometer (FWD), and the effective stress transfer efficiency was computed. Summary of Results The variation of consolidation level was determined to be fairly minor in the variability portion of this investigation. Therefore, further projects were not evaluated for the collection of additional data. However, based on the data collected for the three projects, a load transfer efficiency (LTE) versus percent consolidation relationship was determined and is presented in equation 20.
where
The data used to determine this relationship are plotted in figure 82. Also shown on the graph is the derived regression equation for the relationship. This relationship was incorporated into the current joint faulting model (described in the section titled Summary of Current Distress Indicator Models).
Tie Bar Depth Versus Load Transfer Efficiency To evaluate the feasibility that a model could be developed relating the tie bar depth to joint LTE, an FWD was used to measure longitudinal joint load transfer at the Omaha, Nebraska site as shown in table 66. As discussed in the section on tie bar variability, the tie bars at the Omaha project were consistently placed nearly 76 mm above slab middepth. However, as shown in figure 83 (for a consistent tie bar depth), the load transfer varied from as low as 30 percent to about 100 percent. Also shown in the figure are five locations where tie bars were placed significantly lower than average. The LTE was approximately 95 percent for these locations. Therefore, the collection of additional tie bar depth data from other construction sites was aborted after data collected at the Omaha project suggested that a reliable relationship was not likely.
