| Traditional methods for problem solving in chemical engineering have needed time and a basic understanding of the process. NeuroShell, an artificial neural network program, allows the engineer to model a unit which could be tedious and difficult without explaining the dynamics or the mathematical foundation of the process. In this thesis, NeuroShell was used in an attempt to simulate an ethane-ethylene distillation column based on first principles. It was also used in an attempt to discover the robustness of the model in which noisy data was used for its training. One can conclude, that NeuroShell models a distillation column well and the introduction of noise does not adversely affect the convergence of the network. |
