This report summarizes a research project that investigated the use of Marshall properties for acceptance purposes. Since the Marshall properties are physically related, they can be expected to be statistically correlated. It is therefore necessary to determine whether correlations should be considered when developing acceptance plans. The research consisted of 3 major phases; a laboratory analysis, field data collection and computer simulation analyses. A laboratory analysis was conducted to establish whether correlations are present among asphalt content, gradation, and the Marshall values for stability, flow and air voids. Another aspect of the laboratory analysis investigated 3 methods for determining maximum specific gravity (MSG) for air voids determination. Field data were also collected from 5 paving projects. Finally, computer simulation was used to evaluate the performance of 7 methods for determining the payment factor for the Marshall properties. It is recommended that ASTM D-2041 be used to establish MSG values for both the job mix formula and for field quality control testing. It is also recommended that the quality control testing. It is also recommended that the quality index approach be used to estimate individual PWL values for each of the Marshall properties. These PWL values can then be used with a payment schedule to determine individual payment factors for the 3 properties. The overall Marshall payment factor is then the average of the 3 individual factors. Keywords: Acceptance Plans, Marshall Properties, Computer Simulation, Quality Assurance, Multiple Price Adjustments, Correlation, Statistical Quality Control.

The laboratory phase of a three phase research effort conducted to field validate a multiple price adjustment system for bituminous airport pavements using the Marshall properties, stability, flow and air voids, is presented. The purpose of the laboratory phase was to identify whether correlations exist among the Marshall properties within individual tests. To consider the use of these properties in a multiple price adjustment system, it was first necessary to identify these correlations. The experimental design consisted of 4 different aggregate gradations and 6 different asphalt contents for a total of 24 combinations. A total of 12 replicates were tested for each combination for a total of 288 Marshall test specimens, A number of statistical analyses were conducted on the laboratory test results. An analysis of variance was conducted to determine whether time, i.e., order of testing, had an effect on the results. Correlation coefficients among the Marshall properties, i.e., stability with flow, stability with air voids and flow with air voids, were calculated for each of the 24 combinations. The results of the analysis indicate correlations that are consistent enough to violate an assumption of statistical independence among the properties. The results indicate that the effect of correlations among the properties should be taken into consideration in developing a multiple price adjustment approach based on the Marshall properties. Keywords include: Acceptance plans, Price Adjustment, Marshall Properties, Statistical Quality Control, and Correlation.