Synthesis of heat-integrated distillation sequences in non-conventional processes
Special Symposium - EPIC-1: European Process Intensification Conference - 1
EPIC-1: Poster Session (EPIC - Poster) - P2
Keywords: Distillation, heat integration, process synsthesis, nonlinear programming
The optimal synthesis of heat integrated distillation sequences has received great attention in the past and, currently, is still an important research area in Process System Engineering. Several strategies have been proposed, presenting distinct ranges of applicability and requiring different numerical schemes for their solution:
* based on Mixed-Integer Linear Programming (MILP): these are suitable when all of the product streams correspond to pure components (Andrecovich & Westerberg, 1985); under these circumstances, a branch expansion of all sharp separation alternatives is considered, and simplified linear relations employed.
* based on Mixed-Integer Non-Linear Programming (MINLP); these enable the synthesis of the more general non-sharp separation problem, and consider the optimization of a superstructure where the mixing, fractioning and bypassing of streams are key aspects (Aggarwal & Floudas, 1992).
* based on Generalized Disjunctive Programming (GDP); with a similar range of applicability to MINLP strategies, they can allow a more compact problem formulation by including logical expressions that enable a given equipment to assure different tasks (Yeomans and Grossmann, 2000).
Although more general, MINLP/GDP strategies can face critical convergence difficulties when non-sharp separation models are considered, especially when these refer to highly non-ideal liquid-vapour equilibriums (LVE). Under these circumstances, the construction of approximate models is often adopted (Aggarwal & Floudas, 1992) as a way of reducing the scale and non-linearity of the original formulations. However, since these simplified models are obtained through regression exercises, they also exhibit some limitations, and assumptions must be made concerning the non-distribution of non-key components.
The current work focus on the difficulties that classical strategies may exhibit, when addressing complex processes of non-conventional nature, for instance:
* in the presence of highly non-linear separation models, as a consequence of complex LVE and of the necessary distribution on non-key components.
* the feed stream is characterized by a great number of byproducts, present in vestigial concentrations, but with crucial importance for a set of purity constraints.
In both of these cases, the use of discrete numerical schemes and the construction of approximate models can be hindered. In order to allow the optimization of such processes, a new strategy is introduced that includes three main aspects:
* considers a problem reformulation, with the purpose of enabling a more simplified approach, by grouping sets of real species in (virtual) pseudo-components.
* deals with non-sharp separations, using mathematical approaches of easy implementation, initially developed for more simple and less general problems.
* exploits the capacity of optimizing large-scale non-ideal models, through continuous formulations, to reduce the risks involved in the previous aspect.
In order to illustrate the robustness of this procedure, a complex non-conventional case-study, drawn from industry, will be considered; this regards the purification phase of the aniline production process implemented in Quimigal. The results obtained allowed the identification of a new heat-integrated separation structure, capable of drastically reducing the current operating costs and, simultaneously, enabling a more pure commercial product.
References
Aggarwal, A. & Floudas, C.A. (1992). "Synthesis of heat integrated nonsharp distillation sequences", Computers & Chemical Engineering, 16, 89.
Andrecovich, M.J. & Westerberg, A.W. (1985). "An MILP formulation for heat-integrated distillation sequence synthesis", AIChE Journal, 31, 1461.
Yeomans, H. & Grossmann, I.E. (2000). "Disjunctive programming models for the optimal design of distillation columns and separation sequences", Industrial & Engineering Chemistry Research, 39, 1637.
Presented Thursday 20, 13:30 to 14:40, in session EPIC-1: Poster Session (EPIC - Poster) - P2.
