Plantwide design and control of tubular reactor systems

This work presents quantitative comparisons of the steady-state economics and dynamics of alternative tubular reactor systems. These alternative designs use different tubular reactor configurations: single adiabatic, multi-stage adiabatic with interstage cooling, multi-stage adiabatic with cold-shot cooling and single cooled. The design considers the entire plantwide process: reactor(s), heat exchangers, gas recycle compressor, preheat furnace, condenser and separator. The chemistry is the exothermic, irreversible, solid catalytic, gas-phase reaction A + B → C, which is carried out in a packed-bed tubular reactor.; Design and optimization procedures have been developed for each alternative design. The steady-state design results indicate that the single-cooled reactor configuration is the best from the standpoint of steady-state economics because the cooled reactor operates at higher average temperature and the adiabatic-temperature-rise limitation is removed.; The dynamics of the four alternative systems are studied by dynamic simulation. All of the adiabatic systems were found to be openloop unstable (producing limit-cycle behavior) because of the positive feedback of energy between the reactor and the feed preheating system or because of interaction. The optimum cooled reactor system with moderate activation energy is openloop stable because cooling the reactor makes it less sensitive to changes in the inlet flow and temperature.; All of the systems could be stabilized by using conventional PI feedback controllers, except for the economically optimum 7-stage adiabatic system with cold-shot cooling, reducing the number of stages to three gave a system that could be stabilized.; Conventional PI control systems were used in this study, using either Ziegler-Nichols or Tyreus-Luyben tuning rules. The performance of PI control was compared with that of model predictive control (MPC) in the cooled tubular reactor system. Conventional PI control performed better than MPC and was much easier to apply since MPC requires the selection of many tuning parameters (weighting parameters).; The cooled reactor system with a “hot” reaction (higher activation energy and specific reaction rate) was also studied. The optimum design of this system was openloop and closedloop unstable. A suboptimal design, using a smaller reactor tube diameter and more heat transfer area, was both openloop and closedloop stable.