____________________________________________________________________________________________ *Corresponding author: E-mail: doararomi@lautech.edu.ng; British Journal of Applied Science & Technology 4(21): 3039-3052, 2014 SCIENCEDOMAIN international www.sciencedomain.org Optimization of Process Variables for C-massecuite Exhaustion in a Nigerian Sugar Refinery M. O. Aremu 1 , D. O. Araromi 2* , J. A. Adeniran 2 and O. S. Alamu 1 1 Biochemical Engineering Laboratory, Department of Chemical Engineering, Ladoke Akintola University of Technology, Ogbomoso, Oyo State, Nigeria. 2 Optimization and Control Unit, Department of Chemical Engineering, Ladoke Akintola University of Technology, Ogbomoso, Nigeria. Authors’ contributions This work was carried out in collaboration between all authors. All authors write, read and approved the final manuscript. Received 20 th February 2014 Accepted 3 rd April 2014 Published 29 th May 2014 ABSTRACT Sucrose in the final molasses continues to be a source of major financial loss to sugar refineries worldwide. This study therefore aims at rectifying this anomaly. In this study, the final molasses exhaustibility was predicted using Adaptive Neuro Fuzzy Inference System (ANFIS) and Response Surface Methodology (RSM) based ondata generated from molasses sample collected from the recovery end of refining processes. The results show that both models are able to predict the final molasses exhaustibility with sufficient accuracy. The optimum sucrose recovery of 49.18% was achieved at the point when Brix 0 is 96.00%, Purity of 65.00% and pH of 4.50. Also, both models agree on the combination of purity and pH as the two factors interaction that have optimal effect on the sucrose recovery. The correlation coefficient (R 2 ) value obtained for ANFIS was 0.96 while that of RSM was 0.99. Thus, the RSM model has better prediction performance than ANFIS. Keywords: Adaptive Neuro Fuzzy Inference System (ANFIS); Response Surface Methodology (RSM); design expert; molasses exhaustion; purity; sucrose. Original Research Article