Preliminary design and scheduling of a batch agrochemical plant

Abstract

Several alternative designs for the manufacture of a new agrochemical were examined using combinatorial optimization. Decisions involved in these preliminary designs include how many pieces of equipment to buy, when to install them, the amount of intermediates and product to produce to satisfy uncertain demands, and when to operate the equipment. Because of the complexity and number of decisions to make, a mixed integer linear program (MILP) has been formulated and solved that maximizes expected net present value (ENPV). Case studies were run that concentrated on sharing of equipment and on staged expansions. Both shared and dedicated equipment cases were run with unlimited expansions, expansion in year two, expansion in year three, and no expansions. The shared case with unlimited expansions needed 44% less capital expense than the dedicated case with no expansions. A full range of deterministic cases was studied as well to illustrate the range of possible NPVs for the shared equipment case, resulting in NPVs from 40 to 110% of the ENPV. This preliminary analysis reduced multiple design options to a manageable number from the initial design phase to the final process design. For any future batch plant design, this systematic optimization approach should be used when complex tradeoffs are being considered.