Trying to increase the yields or profit or efficiency (less pollution) of chemical processes is a central goal of the chemical engineer in theory and practice. Certainly sound training in chemistry, business and pollution control help the engineer to set up optimal chemical processes. However, the ever changing demands of customers and business conditions, plus the multivariate complexity of the chemical business can make optimization challenging.
Mathematical tools such as statistics and linear programming have certainly been useful to chemical engineers in their pursuit of optimal efficiency. However, some processes can be modeled linearly and some can not. Therefore, presented here will be an industrial chemical process with potentially five variables affecting the yield. Data from over one hundred runs of the process has been collected, but it is not known initially whether the yield relationship is linear or nonlinear. Therefore, the CTSP multivariate correlation coefficient will be calculated for the data to see if a relationship exists among the variables.
Then once it is proven that there is a statistically significant relationship, an appropriate linear or nonlinear equation can be fitted to the data, and it can be optimized for use in the chemical plant.