Luc Pibouleau

Separation sequence synthesis how to use simulated annealing procedure

Abstract In order to apply the simulated annealing (SA) algorithm, a separation-based coding procedure for complex columns sequence synthesis is presented in this paper. After the recall of the combinatorial aspect of separation sequence synthesis, the main steps of SA procedure are proposed. Then we have focused our attention on the definition of a local solution, that is an acceptable sequence structure, and on the evolution of this one to another acceptable solution. Due to the necessary feasible aspect of a reachable structure solution, we chose a structure-based coding for sequences. It consists in defining, for each structure, a list of elements describing it in a depth first manner and in transforming it by three elementary rules: creation of a complex separator, removal of a complex separator and transformation without changing the type or the number of separators. Many examples are explained, and the implementation into SA procedure is studied in detail.

A two-stage methodology for short-term batch plant scheduling: discrete-event simulation and genetic algorithm

Abstract In this paper, a two-stage methodology for solving jobshop scheduling problems is proposed. The first step involves the development of a discrete-event simulation (DES) model to represent dynamically the production system behavior, taking into account the main features inherent to the application field. Since most scheduling problems in batch processing belong to the family of problems classified as NP-complete, probabilistic optimization algorithms (such as simulated annealing, evolutionary algorithms) represent a good alternative for solving large-scale combinatorial problems (for instance, the traveling salesman problem). In the second step of our approach, we thus investigate genetic algorithms (GAs) for solving batch process scheduling problems: a GA has been developed for minimizing the average residence time to produce a set of batches in function of batch order in a multipurpose-multiobjective plant with unlimited storage. The evaluation of the objective function is provided by its coupling with the DES model embedded in the optimization loop. Computational results show that the use of this approach can significantly help to improve the efficiency of the production system. This paper is focused on semiconductor application, which is the first example treated in our laboratory, although the general approach adopted in this study is now extended to other fields of applications (e.g. fine chemistry with finite intermediate storage and unstable intermediates).

Multiobjective optimization for multiproduct batch plant design under economic and environmental considerations

This work deals with the multicriteria cost–environment design of multiproduct batch plants, where the design variables are the size of the equipment items as well as the operating conditions. The case study is a multiproduct batch plant for the production of four recombinant proteins. Given the important combinatorial aspect of the problem, the approach used consists in coupling a stochastic algorithm, indeed a genetic algorithm (GA) with a discrete-event simulator (DES). Another incentive to use this kind of optimization method is that, there is no easy way of calculating derivatives of the objective functions, which then discards gradient optimization methods. To take into account the conflicting situations that may be encountered at the earliest stage of batch plant design, i.e. compromise situations between cost and environmental consideration, a multiobjective genetic algorithm (MOGA) was developed with a Pareto optimal ranking method. The results show how the methodology can be used to find a range of trade-off solutions for optimizing batch plant design.

Process optimization by simulated annealing and NLP procedures. Application to heat exchanger network synthesis

This paper addresses the problem of automatically synthesize heat exchanger network that features minimum global cost. We present a new approach including HEN configuration choice by simulated annealing algorithm and NLP formulation to systematically optimize their operating conditions (temperatures and split rates). Particular attention is made to the size reduction of the NLP problem which has been the subject of a precedent paper (Athier et al., 1996). In this paper, the global methodology used to solve this problem is presented. The proposed approach is illustrated by two examples, the 5SP1 problem with forbidden matches and a more large scale problem, the 14SP1 one.

Design and retrofit of multiobjective batch plants via a multicriteria genetic algorithm

This paper addresses the development of a two-stage methodology for multiobjective batch plant design and retrofit, according to multiple criteria. At the upper level (master problem), a multiobjective genetic algorithm (MOGA) is implemented for managing the problem of design or retrofit and proposes several plant structures. At the inner level (slave problem), a discrete event simulator (DES) evaluates the technical feasibility of the proposed configurations. The basic principles of the DES are first recalled; then the following section develops a MOGA based on the combination of a single objective genetic algorithm (SOGA) and a Pareto sort (PS) procedure. Finally, a didactic example, related to the manufacturing of four products by using three types of equipment of discrete sizes, illustrates the approach. First, two criteria (investment cost and number of different sizes for units of the plant) are considered for designing the workshop. Then starting from the best solution with regard to investment cost found in the design phase, the plant is retrofitted for manufacturing a double production. Finally, assuming a double production at the design phase, the workshop is designed again. In terms on investment cost, this new solution yields a significant saving compared with the retrofitted plant. In fact, redesigning a new plant, may challenge the retrofitting choice. Secondly, an additional criterion concerning the number of production campaigns for reaching the steady-state or oscillatory regime is introduced, and the same approach (designing, retrofitting and redesigning) is carried out, leading to the same conclusion as in the bicriteria case.

Economic and environmental strategies for process design

Abstract This paper first addresses the definition of various objectives involved in eco-efficient processes, taking simultaneously into account ecological and economic considerations. The environmental aspect at the preliminary design phase of chemical processes is quantified by using a set of metrics or indicators following the guidelines of sustainability concepts proposed by IChemE (2001) . The resulting multiobjective problem is solved by a genetic algorithm following an improved variant of the so-called NSGA II algorithm. A key point for evaluating environmental burdens is the use of the package ARIANE™, a decision support tool dedicated to the management of plants utilities (steam, electricity, hot water, etc.) and pollutants (CO2, SO2, NO, etc.), implemented here both to compute the primary energy requirements of the process and to quantify its pollutant emissions. The well-known benchmark process for hydrodealkylation (HDA) of toluene to produce benzene, revisited here in a multiobjective optimization way, is used to illustrate the approach for finding eco-friendly and cost-effective designs. Preliminary biobjective studies are carried out for eliminating redundant environmental objectives. The trade-off between economic and environmental objectives is illustrated through Pareto curves. In order to aid decision making among the various alternatives that can be generated after this step, a synthetic evaluation method, based on the so-called Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) ( Opricovic & Tzeng, 2004 ), has been first used. Another simple procedure named FUCA has also been implemented and shown its efficiency vs. TOPSIS. Two scenarios are studied; in the former, the goal is to find the best trade-off between economic and ecological aspects while the latter case aims at defining the best compromise between economic and more strict environmental impacts.

Experiments in process synthesis via mixed-integer programming

Abstract Mixed-integer programming is the natural formulation of many process design synthesis problems. Up to now mixed-integer programming algorithms have been applied successfully for solving simplified synthesis problems involving Boolean variables, with the assumption of separability and linearity of constraints and/or objective function. This work aims at going still further in the solution of mixed problems with non-Boolean variables, nonlinear and nonseparable constraints and criteria. A general-purpose algorithm is presented and illustrated by two examples of processes. The synthesis of a heat exchanger network shows the capability of the algorithm to solve this problem via a simultaneous optimization of the structure and temperatures, without assuming a minimum temperature difference. The number of stages and the operating conditions of a multistage reactor for esterification of fatty acid are then optimized, under a purity constraint of the ester, computed by simulation. This second example shows that it is possible to handle nonlinear and implicit constraints.

A multiobjective optimization framework for multicontaminant industrial water network design

The optimal design of multicontaminant industrial water networks according to several objectives is carried out in this paper. The general formulation of the water allocation problem (WAP) is given as a set of nonlinear equations with binary variables representing the presence of interconnections in the network. For optimization purposes, three antagonist objectives are considered: F(1), the freshwater flow-rate at the network entrance, F(2), the water flow-rate at inlet of regeneration units, and F(3), the number of interconnections in the network. The multiobjective problem is solved via a lexicographic strategy, where a mixed-integer nonlinear programming (MINLP) procedure is used at each step. The approach is illustrated by a numerical example taken from the literature involving five processes, one regeneration unit and three contaminants. The set of potential network solutions is provided in the form of a Pareto front. Finally, the strategy for choosing the best network solution among those given by Pareto fronts is presented. This Multiple Criteria Decision Making (MCDM) problem is tackled by means of two approaches: a classical TOPSIS analysis is first implemented and then an innovative strategy based on the global equivalent cost (GEC) in freshwater that turns out to be more efficient for choosing a good network according to a practical point of view.

Optimization of preventive maintenance strategies in a multipurpose batch plant: application to semiconductor manufacturing

Abstract This paper addresses the problem of preventive maintenance (PM) strategy optimization in a semiconductor manufacturing environment, with the objective of minimizing maintenance costs. The approach developed takes into account the interaction of production and maintenance aspects. For this purpose, a discrete-event production-oriented simulator (MELISSA-C++) has been extended to incorporate equipment failures and maintenance operations, thus modeling residual breakdowns, occurring in a combined corrective/PM context. The usefulness of the simulation tool has also been demonstrated for the estimation of both direct and indirect maintenance costs, which are impossible to determine empirically due to the reentrant nature of product flows in a semiconductor manufacturing facility. The results obtained have confirmed the marked effect of equipment characteristics (bottleneck or non-limiting step) on maintenance cost evaluation. Following a tutorial example, typical results are presented and analyzed.

A mixed method for retrofiting heat-exchanger networks

Abstract This paper addresses the problem of automatically determining the optimal retrofit of an existing heat exchanger network, considering the placement/reassignment of existing exchangers to different process stream matches, their need for additional area, the placement of anew heat exchanger and the cost of stream repiping. A two level procedure, derived from a grassroot design model is proposed. The master problem, related to structural optimization, is carried out by a simulated annealing (SA) procedure. For each generated network, the required additional area for existing exchangers and the size of the new exchangers are optimized by a NonLinear Programming (NLP) algorithm. The purpose of this paper is to present the methodology and the resulting modifications to a previously developed Heat Exchanger Network Synthesis (HENS) model to treat this new aim. This approach is illustrated by two examples.