Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
173 CHAPTER 4: EVALUATION OF THE EFFECTS OF DIFFERENCES IN TEST RESULTS ON ACCEPTANCE OUTCOMES The extensive statistical comparisons in Chapter 3 indicate contractor and state DOT tests of HMAC properties are likely different. These comparisons indicate that it is highly likely that variability of state DOT tests are significantly (α =0.01) larger than the variability of contractor tests. These comparisons also indicate deviations of state DOT test results from target values or specification limits are likely larger than deviations of contractor test results. However, it is less likely that differences between state DOT and contractor deviations from target values are statistically significant. To provide an assessment of the effects of these observed differences between state DOT and contractor tests, evaluations of acceptance outcomes using HMAC statistics for Georgia, Florida, Kansas, North Carolina and California were conducted. Statistics for contractor and state test results were applied to acceptance procedures to compute theoretical outcomes. These outcomes were compared and differences computed. The results from these computations are summarized in Table 62. Details for all the acceptance outcome computations are contained in Appendix F and illustrated in Table 63 for Kansas computations.
174 Table 62. Comparison of Acceptance Outcomes with DOT and Contractor Test Results % Greater Chance of Property PF<100% with Georgia DOT Statistics Exceeding Spec. Limits with Florida DOT Statistics Exceeding Spec. Limits with Kansas DOT Statistics Exceeding Min. Spec. Limits with North Carolina DOT Statistics Exceeding Spec. Limits with Caltrans Statistics % Asphalt 0.8 6.0 - - 7.8 % Passing 1/2 Sieve 0.1 - - - 2.7 % Passing 3/8 Sieve 0.3 - - - 5.4 % Passing #4 Sieve 2.7 - - - 1.5 % Passing #8 Sieve 12.3 7.6 - - 5.6 % Passing #30 Sieve - - - - 5.9 % Passing #200 Sieve - 5.7 - - 4.7 Void Content - 12.5 13.9 - - Mat Density (% Gmm) - 3.6 Coarse Mix 4.2 Fine Mix 10.3 12.6* 9.1** - * Contractor and NCDOT Retest  Nuclear Gage ** Contractor and NCDOT Comparison - Cores
175 Table 63. Kansas DOT Acceptance Outcomes Computation Voids Content Statistics s 2DOT = 0.643 s 2 CONT. = 0.318 s 2 are SD â DOT = 0.322% â CONT. = 0.262% â are NSD Probability of exceeding upper or lower specification limits Limits = ± 1% from 4% target Probability = 24.9% with DOT statistics Probability = 11.0% with contractor statistics 13.9% greater chance of exceeding specification limits with DOT statistics Mat Density Statistics* s 2DOT = 3.016 s 2 CONT. = 1.674 s 2 are SD â DOT = 1.429% â CONT. = 1.642% â are SD Probability of exceeding lower* specification limits (LSL) LSL = 92% of Gmm for mainline paving > 2 inches thick LSL = 91% of Gmm for mainline paving ⤠2 inches thick LSL = 90% of Gmm for shoulder paving Probability = 20.5% with DOT statistics Probability = 10.2% with contractor statistics 10.3% greater chance of exceeding LSL with DOT statistics * Mat density is controlled with one sided limits, ie, only a minimum or lower limit is specified. Data from three types of paving were combined by using the variable â = X-LSL.
176 Statistics (â and s2) were selected from the largest available data bases for each state. For example, the following statistics were selected from Table 22 for Florida, and used for the computation of acceptance outcomes for asphalt content: s 2FDOT = 0.084 s 2CONT = 0.062 â FDOT = 0.016% â CONT = -0.012% Statistics were applied to each state DOTÂs acceptance procedures to compute the probability of certain outcomes. The DOTÂs in Florida and Kansas and California use the percent within limits (PWL) procedure, and the outcomes computed were probabilities that upper and lower (two-sided) or lower (one-sided) specification limits would be exceeded. As shown in Table 63, the Kansas DOT specification limits for voids content are ± 1% from the 4% target. The probabilities that the upper or lower specification limits might be exceeded were computed as 24.9 and 11.0%, respectively, using Kansas DOT and contractor statistics. Therefore, there is a 13.9% greater chance of exceeding specification limits with Kansas DOT statistics. Mat density has a one-sided or lower limit. There is a 10.3% greater chance of exceeding the lower specification limit with Kansas DOT statistics. Similar computations were used for the Florida and North Carolina DOTÂs and Caltrans. The North Carolina DOT specifies a minimum of 92% Gmm mat density, below which a pay reduction is assessed. The North Carolina DOT allows the use of nuclear gages or cores to measure mat density and computations are included for both.
177 Computations for the Georgia DOT are different. Pay adjustments are applied based on average absolute differences from target values. Limits for pay adjustments, both reductions and bonuses, are a function of the number of test results for a LOT: typically 2 to 6. For comparisons, probabilities for pay factors less than 100% were computed with criteria for numbers of tests equal to 3. The absolute value distributions and computation procedures suggested in Parker, et al (18) were used for these computations. For asphalt content, a LOT pay reduction (PF<100%) will be applied by the Georgia DOT if the average absolute deviation from a target value of the average of 3 tests exceeds 0.46%. With Georgia DOT and contractor statistics the probabilities that a pay reduction will be applied are 0.9 and 0.1%. Therefore, there is an 0.8% greater chance of getting a pay reduction for asphalt content with DOT statistics. The percentages in Table 62 indicate that outcomes with contractor DOT test results are always more favorable. More favorable outcomes with contractor statistics are expected since contractor test results are, typically, closer to targets and significantly less variable. However, apparent differences in test results may not be necessary for difference in pay. As presented in Chapter 3, contractor and New Mexico DOT statistics are quite similar, but acceptance outcomes using contractor statistics were more favorable (approximately 2% greater pay). The range of percent chances in Table 62 is 0.1 to 13.9%.