Multiple Polynomial Regression Approach to Design Optimization of an Exhaust Emission Reduction Device

The application of a multiple polynomial regression procedure for optimization of design and performance parameters has been presented. To illustrate the statistical method, data from a laboratory test of an exhaust emission reduction device was used. There were four factor variables for which the response was constructed. Among the factor variables two were design parameters; the remaining two were the operating parameters. The selected response variables were CO, NO, and HC's.

The assumptions for the analysis procedure are presented. Problems which could affect the validity of the statistical procedures for the design optimization are discussed. Experimental verification of the optimization results was performed. Overall agreement between the predicted level of pollutants reduction and measured values did not differ by more than ten percent.