Pulp and Paper Network Analysis by Genetic Algorithm, Mathematical Programming and Process Simulation, 2001 Engineering / Finishing & Converting Conference Proceedings

S. Shafiei, P. Floquet, S. Domenech, J. Paris--System closure in pulp and paper mills consists of optimising internal reutilisation of used water to reduce fresh water consumption and enhance energy and materials efficiency while respecting process constraints. The optimum closure target is specific to each mill because of great differences in process configurations. North American mills have achieved much progress over the years by applying conventional process analysis and engineering techniques. Significant gains can still be accomplished by implementing innovative closure strategies; process integration methods and tools can yield interesting results in such endeavours. A water-pinch ® type analysis, such as those which have been used with success in the petrochemical industry for optimisation of wash water networks, are rarely applicable to the essentially aqueous pulp and paper processes. An approach based on the characterisation of individual operations by water demands and sources can be utilised. The search for novel network configurations as a function of specified process constraints and objective functions is done by means of genetic algorithm coupled with linear or nonlinear programming (LP or NLP) depending on the problem formulation and, with process simulation. Potential solutions are derived from an initial superstructure, which embodies all possible source- demand connections, by an evolutive process akin to natural selection and adaptation. Each possible solution is defined by a unique chromosome encoding existing and non-existing connections and characterised by a value of the objective function optimised by LP or NLP, the adaptation index. At each iteration, the solutions with highest index are more likely to be retained and new solutions are generated by chromosome combinations (crossover) or change (mutations) according to established rules. Eventually a set of solutions (a population) with highest indices is evolved. Process impacts of closure strategies are determined by simulation on the platform CADSIM plus PAPDYN ® .