In any design, the information upon which the design is based is uncertain due to biases, random errors, large variances, missing data, and other factors that cannot be controlled. In process design, these uncertainties are often ignored or treated deterministically. Various strategies have been proposed for estimating and rectifying the data. Among these strategies are statistical techniques such as least squares and maximum likelihood, optimization techniques such as linear programming and Lagrange multipliers, and filtering techniques. All of these techniques focus on the data itself rather than on the use of the data within the design process. A new approach to data reconciliation is developed using influence methods and interval analysis to eliminate gross and systematic errors in the data and to reconcile the remaining data. This new methodology will enhance the quality of the data available to the designer. The research will result in algorithms and computer programs that will be applicable in a number of disciplines.
