Abstract:
Process industry has been one of the most energy consuming sector. In order to reduce energy consumption, efficient energy process is vital. Heat exchanger is one of the abundantly used equipment in process industry. Plate fin heat exchanger mostly used in process industries also got substantial share of research for realization it’s optimum design and operation. The studies have been focused on maximizing the heat transfer rate and minimizing the pressure drop [Yidan Songa et al., 2015], minimize the total volume, CO2 emissions and cost [Lixia Kanga., 2015], optimizing shape of fins of the plate and fins heat exchanger [Chunbao Liu et al., 2017]. In this study, a plate and fin heat exchanger model of a gas furnace of a tile factory was modeled in Aspen Exchanger Design & Rating (EDR) environment. The EDR was linked with an excel sheet and MATLAB to transform the model from a steady state to a dynamic mode. Several hundred scenarios were generated by inserting artificial uncertainty in the steady-state values of the process conditions such as inlet hot temperature, inlet cold temperature, and fouling resistance. Then Genetic Algorithm (GA) was applied to derive the optimum combination of the inlet flow rate of the hot and cold streams keeping minimization of the outlet temperature of the hot stream as the objective function. The datasets comprised of optimum operating conditions and their corresponding output were used to develop an Artificial Neural Networks (ANN), model. The ANN model was also used as a surrogate in SOBOL and Fourier amplitude sensitivity testing (FAST) based sensitivity analysis framework to find hierarchy in the input variables in terms of their impact on the model output.
