Modeling of polycaprolactone production from ε-caprolactone using neural network

Abstract

In this paper, extensive study of ring-opening polymerization ε-caprolactone (ε-CL) using lipase Novozym 435 as catalyst in flask level and reactor level were conducted. The polymerization rates increase with an increase in time up to 4 h after which there has been a steep decrease for all temperature from 50 to 100°C in the flask level. The conclusion out of flask level and reactor level study is that a uniform trend is obtained at 70°C. A multilayer feed-forward neural network (FANN) model was trained with an error back-propagation algorithm. Reaction time, temperature were used as the input parameters and molecular weight is the output for the flask level study where as reactor impeller speed was also included for reactor level study. Two FANN models with modeling performances of 2-10-1 in the flask level and 3-9-1 FANN1 and 2-13-1 FANN2 (excluding reactor impeller speed) for the reactor level study were obtained. © 2012 Springer-Verlag.