In naphtha cracking center process, naphtha materials are craked in furnaces to a smaller hydrocarbons and finally the monomers such as ethylene and propylene are produced through the downstream process units for example quench tower, reactors, refriegerators, and separation towers. Among them, acetylene converter has a role to convert the acetylene to ethylene because the acetylene do very harmful effects to the polymer-grade ethylene. Therefore the acetylene converter operation is very important for the purity of ethylene.
In the acetylene converter, selective hydrogenation reactions such as acetylene hydrogenation and ethylene hydrogenation are occurring. In addition to these main exothermal reactions, undesired reactions such as green oil formation and hydrocarbon cracking reactions occur and cause harmfulness on the product quality, reactor economics, and safety.
Because of safety consideration and the tight constraint on the purity of the ethylene at the outlet of the acetylene converter, there exists a possibility of optimization through the modeling of the complicated mechanisms. In this work, a kinetic model structure based on the open reaction rate expressions has been selected and the kinetic parameters have been adjusted to fit the operation data. During this process, a catalyst deactivation model was developed as a hyperbolic type with the estimated amount of green oil produced. This was validated using the catalyst activity profile estimated from the given operation data. But because the target of that model is complete tracking of outlet acetylene concentration, the deviations of the predicted values from the operation data of the other important variables such as outlet hydrogen, ethane concentrations, and outlet temperature are inevitable. Thus a hybrid model combining the first principles model and a neural network was used to compensate the offsets. This hybrid model shows better performance than the first principles model and can be used f...