Optimal design of chemical processes at open-loop unstable steady states.

摘要

The design of chemical processes often involves the application of rules-of-thumb which can lead to significant overdesign. For many process designs, constraints are added to exclude operation near or within complex operating regimes, though in many cases the economic optimum occurs at an unstable steady state. The focus of this study is on overdesign in the exothermic CSTR, since, for a given conversion of reactant, the exothermic CSTR usually has a smaller volume, but introduces greater complexity such as hysteresis, periodic or chaotic behavior in process operations. This thesis presents results of an investigation to identify situations in which operation at an open-loop unstable steady state is desirable. A methodology for screening and designing a process to operate at an open-loop unstable steady state is presented. This involves an examination of the kinetic model and operational constraints, followed by a univariant search for the optimal design at both the stable and unstable operating points, and finally, the performance of a controllability study.; The propylene-glycol-ethers process is considered as a candidate process for operation at an unstable steady state. First, a typical design involving a series of PFTRs is examined using the ASPEN PLUS process simulator and its profitability is determined. Then, an alternative design, involving a single exothermic CSTR is introduced. The CSTR is examined using parameterization techniques, with care to avoid the operational constraints, and the final design is simulated and compared to the PFTR design.; The dynamic response of the CSTR to various disturbances was studied. A model-predictive controller was implemented using the controlled-random-search technique to determine the controllability of the process in response to various disturbances and set-point changes at the unstable steady state. Simulations were performed with and without process/model mismatch and compensation. This enabled the determination of adequate controllability in the design stages.; The methodology proposed applies to processes involving exothermic CSTRs only, but a broad class of reaction systems with recycle can be considered. The CSTR design was determined to be more profitable than the PFTR design; however, the gains in the selectivity were largely offset by the increased recirculation and separation costs. Also, the CSTR design was shown to be controlled easily for the disturbances and set-point changes considered.