Endüstriyel bir dizel hidro-işleme ünitesinin modellenmesi, en iyilenmesi ve kontrolü

Diesel hydroprocessing is an important refinery process which consists of hydrodesulfurization to remove the undesired sulfur from the oil feedstock followed by hydrocracking and fractionation to obtain diesel with desired properties. Due to the new emission standards to improve the air quality, there is an increasing demand for the production of ultra low sulfur diesel fuel. The first of the thesis is addressing the development of a reliable dynamic process model which can be used for real-time optimization and control purposes to improve the process conditions of existing plants to meet the low-sulfur demand. The overall plant model consists of a hydrodesulfurization (HDS) model for the first two reactor beds followed by a hydrocracking (HC) model for the last cracking bed. The models are dynamic, non-isothermal, pseudo-homogeneous plug flow reactor models. Reaction kinetics are modeled using the method of continuous lumping which treats the reaction medium as a continuum of species whose reactivities depend on the true boiling point of the mixture. The key modeling parameters are estimated using industrial data. Steady-state and dynamic model predictions of the reactor bed temperatures, sulfur removal, and diesel production match closely the plant data. The industrial Diesel Hydroprocessing plant operates with varying feed-stocks and large throughputs. Also, changing market conditions have significant effects on the diesel product specifications. In the presence of such a dynamic environment, the DHP plant has to run in the most profitable and safe way while satisfy the product requirements and not violating specified constraints. In the second part of the thesis, a hierarchical, cascaded model predictive control structure is proposed to be used for real-time optimization of the industrial DHP plant using developed models. The performance of the control structure is checked with closed loop simulations for both set point tracking and disturbance rejection cases.