Comparison of normal and lognormal distributions to analyze TCLP results for beneficial use permits

Beneficial use of foundry residual wastes is a critical element in reducing operating costs for the cast metal industry. Many regulatory decisions in evaluating beneficial use permits are based on the statistical analysis of Toxic Characteristic Leaching Procedure (TCLP) tests on by-product samples and comparison of these results with regulatory limits. Many of the statistical methods employed in this analysis assume that TCLP test results are normally distributed. This study analyzes the accuracy of this assumption and its analytical implications in evaluating foundry residuals for beneficial use. The paper examines data from TCLP tests for zinc content, conducted on residual sand from 33 foundries involved in a beneficial use per-mil application. It concludes that the TCLP test results ai-e not normally distributed. In addition, the analysis demonstrates that this erroneous assumption lends to an overstatement of the proportion of the residual sand with TCLP test results that exceed regulatory limits. For the data in the study, this overstatement was greater than 35% and may have a detrimental impact on approval of the proposed beneficial use. A lognormal distribution mole accurately modeled the TCLP test data and did not preclude regulatory approval for compliant foundries.