(32d) Analysis of Bayesian Linearity (BL) to Improve Performance for API to DP Integration

Using of a mixture of Bayesian statistics, data mining, and modeling it was determined how absolute temperature (AT), population density, (PD) and equipment coefficient (EC) affect yield. This fundamental chemical engineering problem was the focus of our group’s research, the Wooster Technical Foundations (WTF). We have asked questions that have brought significant insight in developing solutions with Bayesian Linearity (BL).

 We discovered that a reaction that occurs at cold average temperatures of 285°K significantly produces a low response of ~20% compared to the yield that occurs at temperatures averaging 295°K. This relatively small difference in temperature is an important, but not the sole contributor to lower yield: we also observed a pollution balance effect in which an inverse correlation exists between the temperature and the relative population density, even at average temperatures below 285°K, so that one can overcome yield by a higher population density. The temperature to yield factor is well known in many fields but also one additional factor was discovered, which was classified as EC. The EC factor can also improve yield even at low temperatures and low populations densities. Using data mining techniques, along with the fundamental mathematics, experimental methods applied in other fields and a computational method, we developed a model demonstrating how performance of DS/DP integration is improved though seemingly simple factors of population balances, temperature differential and equipment.