Effect of Sample Size and Methods on Percent within Limits for Quality Control and Assurance

摘要

The construction quality of a lot is often measured by collecting and testing a few samples. The statistics of an acceptance quality characteristic (AQC) of the lot determined based on the samples is used to make an inference on the overall construction quality. The most commonly used quality measure to quantify the quality is percent within limits (PWL). For performance-related construction specification, the PWL value of a lot is related to its expected future performance. While a contractor is paid an incentive if the measured quality is better than acceptable quality limits (AQL), the contractor may be penalized if the measured quality is less than AQL. Therefore, the accuracy of PWL estimation is important to minimize over/under payment to the contractor for the produced quality. In this paper, an effort has been made to determine the effect of sampling plans and sizes on the PWL estimation of a lot. Air void contents after construction were used as the AQC with two sided specification limits. Different sampling plans include: (a) random sampling with replacement, (b) stratified sampling, and (c) random sampling without replacement. For each sampling plan, samples of sizes 3, 5, 10, and 20 were used to evaluate the combination (i.e., sampling plan and sample size) for estimating PWL. The estimated PWL values were compared with the true PWL of the lot. The results show that stratified sampling plan represents the true construction quality of a lot better for lower sample sizes.