(473a) Global Optimization of Mass and Property Integration Networks

Mass integration has been widely used to reduce the pollution and to increase the economic efficiency of industrial processes. Most of the methodologies reported to solve the mass integration process have been characterized in terms of the composition of streams. However, sometimes the process equipment and the waste discharged to the environment impose specific constraints in terms of the properties of the streams; even when such properties can be sometimes calculated from the composition of streams, this is a difficult task because of the large number of components that may be present in the process streams. This work presents a mathematical programming model for the optimal design of mass integration networks based on properties of streams. This new formulation is based on a superstructure that considers in plant treatment units to improve the quality of the streams in order to satisfy the constraints given by the process units and the environment in terms of properties. The model consists of mass and property balances, together with logical disjunctions to model the optimal selection of the proper treatment unit. The disjunctions contain only linear relationships because the efficiency and cost of the available treatment units are known previous to the optimization process. A recycle-reuse configuration is considered, which enables the possibility to send, partially or completely, streams from any treatment unit to another unit, to any process sink or to the final waste. Because neither the properties nor the flowrates of these streams are known, the property balances at the mixing points contain bilinear terms, giving rise to a non convex MINLP problem. A global optimization algorithm was implemented to guaranty convergence in the solution of the MINLP problem. The objective function was to minimize the total annual cost of the network, including fresh sources cost, annualized property treatment system cost and piping cost, while satisfying simultaneously process and environmental constraints for properties such as composition, toxicity, theoretical oxygen demand, pH, density and viscosity. The solution of the proposed model shows the advantages of considering simultaneously process and environmental constraints over a sequential approach in which they are considered separately. It is also shown how the computational effort depends directly on the size of the problem and on the number of intervals used to strengthen the bounds of the relaxation that is implemented as part of the solution strategy.