Optimization of a batch reactor process using statistical analysis

Abstract

A batch reactor is used to remove an environmental concern (EC) from the wastewater stream of an industrial process at Novelis, Inc.’s, Oswego, NY, facility. Between May 2017 and February 2018 major process changes occurred across the machine centers that produce the wastewater. This led to expected decreases in the concentration of EC in the wastewater, how much wastewater was produced, and several other changes. As a result, the process engineer responsible for the reactor decided that the amount of Chemical A and Chemical B, the reactants, added to each batch would be reduced. Additionally, batches are run less frequently, often with smaller volumes. These changes were not expected to have any effects on reactor efficiency. However, after some time it was noted that the reaction was consistently taking longer to complete than previously and the final concentration of B was often higher than before the changes were made. With multiple changes having been made around the same time, to both the source of the wastewater and the reactor, it is difficult to understand which changes, or combination of changes, caused this shift. This report details a series of statistical analyses which were used to gain a better understanding of the connections between the changes observed, the resulting shifts, and optimization of the reactor operation. Through the use of factor analysis it was found that the seven potentially relevant input variables could be reduced to three components. Through ANOVA four variables were determined to have significant impacts on both the length of time for the reaction to complete and the final outcome of the reaction. Based on these findings adjustments were made, additional data was collected, and new analyses were run. From this second round of analyses the following change was recommended: the 42 liters of A and 36 liters of B should be used for every cubic meter of wastewater to be treated.