Endüstriyel bir hidrokraker reaktörünün modellenmesi

The aim of this study is to model an industrial hydrocracker reactor. As preliminary work, the industrial hydrocracker reactor was simulated by the Aspen HYSYS Hydrocracker. Next, a first principle model was developed in which the discrete lumping approach was adopted. Since the model consists of a set of ordinary differential equations and algebraic equations which have to be solved simultaneously, a code was written by using MATLAB. Parameter estimation was performed to match model predictions with real plant data. Under constant conversion operating conditions model estimates matched the plant data closely. In order to investigate the effect of catalyst deactivation on parameters, the constant-conversion model was updated using data corresponding to operating days with different catalyst activity. While significant changes in the product distribution parameters and heat of reaction parameters were not observed, rate constant parameters correlated well with catalyst deactivation. Since the reactor inlet temperatures are the major independent and adjustable process variables, their effect on plant products and conversion was also studied. For the constant conversion training data, the model showed low sensitivity of conversion to reactor inlet temperatures. Next, a new model was developed by estimating the parameters using operating data at two different conversion levels. Simulations showed that an overall 3.1 °C increase in inlet temperatures along the reactor affected the conversion by 0.75 % which is a remarkable value for an Hydrocracker unit. Finally, the predictive ability of the model was questioned by analyzing the plant data. It was observed that increase in inlet temperatures resulted in the same conversion for some days of operation. This means that in order to keep conversion constant the reactor inlet temperatures are adjusted by the existing advanced control system to reject disturbances such as feed properties, catalyst deactivation or other operation conditions. The developed model cannot predict such changes. However its parameters can be updated on-line to match different plant data that includes disturbance effects. Alternatively, a fundamental or empirical disturbance model can be integrated into the developed model in order to predict the behavior of the process under realistic disturbances.