Choice of facility model types in solving a linear statistical optimization problem for a continuous process

Static optimization of a chemical industrial process is usually formulated as a linear programming problem, the static models of facilities being in some way linearized. The paper discusses rational linearization of the models for ensuring the desired modeling accuracy. Cases are indicated where piecewise-linear models are useful and when “ambiguous” models are preferable in which the facility output has a set of values with a fixed value of the input.