| 초록 |
This research proposes a methodology for scaling up tubular reactors using economic, physical, and chemical constraints. Herein, a CFD simulation of a rigorous mathematical model based on an experimental single tubular reactor was executed with a validation error of 2.1%. The model was then employed in scaling up the lab-scale reactor to an industrial scale by simulating 51 different reactor configurations with design and operating variables. Next, data from these simulations were used to estimate the desired product yield, reactor equipment costs, pressure drops, catalyst deactivation, and temperature control efficiency. These results were employed in developing a deep neural network to establish a complex non-linear relationship between the reactor configurations and the four decision variables. Furthermore, a genetic algorithm was integrated with the model to optimize the reactor configurations. As a result, the optimal reactor design were computed. |
