With the increasing use of Computer Aided Engineering, it has become vital to be able to evaluate the accuracy of numerical
models. This research poses the problem of selection of the most accurate and relevant correlation solution to a set of corridor variations.
Specific methods such as CORA, widely accepted in industry, are developed to objectively evaluate the correlation between monotonic functions, while the Minimum Area Discrepancy Method, or MADM, is the only method to address the correlation of non-injective mathematical variations, usually related to force / acceleration vs. displacement problems. Often, it is not possible to differentiate objectively various solutions proposed by CORA, which this paper proposes to answer. This research is original, as it proposes a new innovative correlation optimization framework, which can select the best CORA solution by including MADM as a subsequent process.
The paper and the methods are rigorous, having used an industry standard driver airbag computer model, built virtual test corridors and
compared the relationship between different CORA and MADM ratings from 100 Latin Hypercube samples. For the same CORA value
of ‘1’ (perfect correlation), MADM was capable to objectively differentiate between them. The paper has recommended the MADM
settings n=1; m=2 or n=3; m=2 for a congruent relationship with CORA. As MADM is performed subsequently, this new framework
can be implemented in already existing industrial processes and provide automotive manufacturers and Original Equipment
Manufacturers (OEM) with a new tool to generate more accurate computer models.