Numerical simulation of DIC drying process on matlab distributed computing server

Abstract

Instant controlled pressure drop, also known as DIC, is one of the drying techniques that has been used for texturing, extracting and drying various food products. Mathematical model has been used to explain the drying process, although most of the studies focused on the statistical regression model approach. Due to the limitations of regression model, which neglects the fundamental of dehydration process, this paper presents the development of mathematical models to detect, solve and visualize the three-dimensional (3D) heat and mass transfer in DIC drying process. Finite difference method (FDM) with central difference formula is used to discretize the mathematical models. A large sparse of system of linear equation (SLE), which represents the actual drying process, is solved by using some numerical methods, such as jacobi (JB), red black gauss seidel (RBGS), alternating group explicit with douglas (AGED) and brian (AGEB) methods. Based on the numerical results, high execution time and high computational complexity have been shown. In order to reduce the execution time and computational complexity, the parallel algorithm based on domain decomposition technique has been implemented on the MATLAB distributed computing server (MDCS). The parallel algorithm of the numerical methods was evaluated and compared based on the parallel performance metrics: execution time, speed up, efficiency, effectiveness, temporal performance and granularity. From the parallel performance metrics, it was found that the PAGEB approach had better performance, followed by PAGED, PRBGS and PJB methods.