Minimization of economical losses due to utility disturbances in the process industry

• A method for modeling utilities at process industrial sites is suggested.
• An optimization problem for minimizing losses due to utilities is formulated.
• The resulting MIQP optimization problem is relaxed to a QP.
• Two examples with different dimensions are given.
• The research was conducted in collaboration with process industrial companies.

A process industrial site may consist of several production areas, some producing intermediate products for further refinement in other areas, and some producing end products. The areas may share the same utilities, such as steam and cooling water, which means that the areas could be connected both by the flow of products through the site and by the use of the same utilities. Management of utility disturbances thus becomes an interesting topic. In this paper, a simple approach for modeling utilities is suggested and used to formulate a mixed-integer quadratic program (MIQP) that aims at minimizing the total economic loss at the site, due to utility disturbances. The optimization problem is reformulated as an ordinary quadratic program (QP), where auxiliary variables are utilized to avoid the use of integer variables. For suitable choices of the optimization weights, the solutions to the MIQP and the QP are in many cases equal. Two examples are given, where one is a small example inspired by a real site at the specialty chemicals company Perstorp, and the second is a larger problem that aims to show the advantage of the QP formulation when the number of areas, and thus the number of integer variables, becomes large.