Integration of superstructure-based optimization and semantic models for the synthesis of reactor networks

Semantically enabled process synthesis and optimisation

“…The objective of this reaction is to maximise the outlet concentration of desired product B while minimising by-products C and D. As a matter of reference, the best reported solution (Labrador-Darder et al 2009), to our knowledge, is the outlet concentration of 3.66 with CSTR + PFR reactors with no recycle and by-pass flows and with total volume of 28. 77…”

“…More recent approaches focus on stochastic search based optimisation with perhaps Simulated Annealing (Marcoulaki, E. and Kokossis 1996;Marcoulaki, E. C. and Kokossis 1999;Mehta and Kokossis 2000;Linke and Kokossis 2003a), Tabu Search (Wang, C et al 1999;Linke and Kokossis 2003b;Cavin et al 2004) and Ant Colony (Dorigo et al 1999;Jayaraman et al 2000;Dorigo and Blum 2005) being the most widely applied. The latest advances in stochastic search based on cascading of population and inflection of solutions is particularly attractive as it provides readily access to optimisation solutions at every stage of the process, the search known as the Cascade Algorithm (Labrador-Darder et al 2009;Kokossis, A et al 2011;Cecelja, F et al 2014). …”

Process Design, Integration and Optimisation: Advantages, Challenges and Drivers

“…Reactors' network -the operating mode for the reactors (batch versus continuous) is selected, together with their type and the overall performance of the reaction train; the variability of raw materials and several scenarios regarding the production rate should delimit an 'operation window' for the state and operating parameters. The non-linearities induced by external recycles from the separation trains should be addressed, to emphasise possible problems with reactors' control and stability (Labrador-Darder et al , 2009). A rough estimation of the utility allocation could be done, based upon the heat of the reaction.…”

Reactor Network Synthesis Using Coupled Genetic Algorithm with the Quasi-linear Programming Method

“…Labrador et al 26 presented a novel framework for the optimization and synthesis of complex reactor networks that combined superstructure-based optimization, semantic models, and analytical tools. Pontes and Pinto 27 presented a MINLP model to synthesize Fenton reactor networks for phenol degradation.…”