267226 Methodology for the Analysis and Design of Chemical Conversion Procesess Applied to the Methanol Synthesis
A broad range of methods for process design appears in the literature which can be classified based on the chosen strategy for representing the design alternatives or the design space. These can range from methods that attempt to enumerate the alternatives in an explicit space, a coordinated search in the space of design decisions, evolutionary methods, superstructure optimization, targeting, problem abstraction, and combinations of these Biegler (1997). In the field of reactor design the methods that have made considerable contributions can be classified into attainable region methods or targeting, and optimisation methods involving either construction of superstructures that are formulated as MINLP problems or state space design representation that exploit analogies from dynamic optimization techniques.
Attainable region methods have their origin in the early work of Horn (1964), and is a representation of the attainable concentration space when reaction and mixing take place. The theoretical framework of the attainable region approach has been extensively further developed by Glasser et al. (1987) , Hildebrandt and Glasser (1990) , Feinberg and Hildebrandt (1997). Given a chemical reaction, the method constructs graphically the consecration space based on the fact the all points in its boundaries are represented by the convex hull of combination of ideal reactor (PFR and CSTR) trajectories. The constructed convex hull contains the entire attainable region in the concentration space and given the reaction kinetics one can predict upper bounds of the performance of reaction system. The attainable region can be considered as a fundamental contribution in the area of reactor design since it provides critical insight into the properties of reaction and mixing. However the applicability of the method has some limitations inherently imposed by the graphical approach considered restricting the designer to analyse problems in two or three dimensions.
Superstructure approaches were developed by several authors , Achenie and Biegler (1986) worked on an idea originated by Jackson (1968) and on the basis of one dimensional dispersion model they formulate the reactor synthesis problem as a NLP optimisation problem with split fractions, source points and sink points that determine the mixing pattern, this network has been extended also to non isothermal reactors. Kokossis and A. (1990) proposed a reactor representation that utilizes as a basis the continuous stirred tank reactor (CSTR) and approximates the plug flow reactor as a cascade of CSTRs. A reactor network superstructure is presented in which all possible structural configurations of interconnected reactors are embedded. Marcoulaki and Kokossis (1999) formulated the reactor synthesis problem in a same way and attempted to solve it with stochastic optimisation approaches such as simulated annealing. Many contributions can be found also in the area of determinist global optimisation of superstructure formulations Esposito and Floudas (2002). The combinatorial nature of the MINLP superstructure formulations result in a computational intractable problems and most of the aforementioned contributions focused on developing efficient optimisation algorithms and less focus has been given to the practical implementation in real industrial applications.
Within this contribution we present an alternative methodology for reactor design considered within the context of conceptual process design. The main idea is to represent basic operations, reaction, fluid mixing and heat transfer, and separation through a continues state space model describing the reaction path.A fluid element is tracked along the reaction path and optimal conditions are determined that satisfy a design objective subject to process and product quality constraints. By encapsulating fundamental phenomena in a composite model rich enough to resemble the feasible design space and conducting a systematic search we are given the possibility to identify regions where improved and innovative reactor configuration are located. We illustrate the methodology by considering the methanol synthesis loop,a process of industrial interest. A multistage case for methanol synthesis with interstage product removal is further considered and its economic feasibility is assessed.
Keywords: process, design, reactor design, optimization, methanol synthesis
Glasser, D., Crowe, C., Hildebrandt, D., 1987. A geometric approach to steady flow reactors: the attainable region and optimization in concentration space. Industrial Engineering Chemistry Research 26 (9), 1803–1810.