Research And Application Of Advanced Control For Delayed Coking Process

Delayed coking (DC) is a widely used thermal process for bottom of the barrel upgrading. As the supply of crude oil tends to be insufficient and the quality of which tends to be heavier and poorer, DC plays a more important role in modern refinery. DC is one of the most difficult refinery units to control, which is characterized by the disturbances caused by semi-continuous operation, high temperature and inevitable coking. The productive capacity of DC in China holds the second position in the world. But the practice of advanced control for this process is still backward. In this dissertation, by comprehensively using the theory and technology of process modeling, control theory and artificial intelligence, and by exploring the inherent mechanism of process and operation experience, research is carried out on the issues related to the advanced control for coking furnace, which is the key equipment of the DC process. The main work and research results are as follows.1. An adaptive state feedback predictive control scheme is presented for the temperature control of DC furnace. By taking chamber temperature and temperature-pressure compensated fuel gas flowrate as state variables and using measured variables for states feedback, the disturbances that enters into fuel loop and chamber temperature loop can be suppressed simultaneously, which can not be achieved by conventional PID cascade control. A simplified nonlinear model on the basis of first principles of the furnace is developed to obtain the state space model. The online model adapting method is proposed to accommodate the dynamic change of process characteristics because of tube coking and load changes. Simulations based on HYSYS and industrial implementation shows the effectiveness and feasibility of the proposed control strategy.2. A large inverse response for the outlet temperature can be caused by a sudden increase of the injected steam of a particular velocity, and usually following a violation of high temperature limit. By analyzing the mechanism of this abnormal event, a rule-based monitoring method that uses data from multiple sensors and an expert control scheme based on heat balance calculation are proposed. The results of site implementation show that the monitoring method is effective and the expert control is capable of giving appropriate feedfoward compensation, and thus, preventing the outlet temperatures from violating their high limits.3. The inevitable coke deposition on inside tubes of the DC furnace is one of the main factors that limits the run length of the DC process, and increases the consumption of fuel and power energy, lowers productivity. An on-line tube coking thickness estimation method that based on the calculation of thermal resistance is developed. The tubes are lumped according to the place of fire-side tube wall thermal couples. Real time plant data and heat transfer model are used to determine the overall thermal resistance and coking thickness in different lump zones. Applying to a coking furnace by using historical data shows that the coking thickness can be effectively predicted. The coking prediction helps to improve operation, and enhance the safety of the furnace.4. A pass balance control scheme based on steady state energy balance to maintain the pass outlet temperature to be same for multi-pass fired heater is proposed and the convergence of the control scheme is analyzed. Steady state control, which is realized by choosing long control interval, is used to avoid the dynamic coupling of different pass streams caused by frequently adjustment of pass flowrate. For furnaces fire fuel gas and fuel oil and take fuel gas as the main fuel, an ideal zone control for fuel gas is proposed to save the expensive fuel oil and leave a margin for fuel gas to deal with disturbances. Simulations and field implementation to a crude furnace show that the proposed control schemes are very effective and easy to implement. An improved thermal balance control scheme is applied to a furnace in a molecular sieve dewaxing unit which burns three kinds of fuel.5. The challenges of networked control system (NCS) are the network induced time delay and data packet dropout. A compensation scheme that designs compensators separately for the feed-forward and feedback channel in the framework of predictive control is proposed to deal with these problems. For the feedback channel, timestamp is attached to the sampled data. Then the available newest output measurement is identified using the timestamp. The feedback correction term in predictive control is modified. For the feed-forward channel, a receiving buffer located before actuator node is added to transform the random delay to a constant delay, which is incorporate in predictive model. Simulation results show the effectiveness of the scheme.