ABSTRACT
Multivariable optimization (MVO) is a powerful nonlinear steady-state flowsheet simulation technique used widely in the chemical process industry to optimize plant performance by increasing plant capacity and/or reducing energy usage. The user supplies the objective function(s), constraints, and variable limits based on operating heuristics, prior experience, and observed process behavior. In this paper we describe how we used MVO in conjunction with improved automation and statistical process monitoring to increase capacity and reduce energy consumption. This project was conducted over a two-year period in a methylamines facility that produces three products: monomethylamine (MMA), dimethylamine (DMA), and trimethylamine (TMA). It led to a 14% improvement in MMA capacity and a 7% improvement in TMA capacity. Energy consumption per pound of main product decreased by approximately 20% for the six-month period (April-September, 2000) when compared to a three-year average (1997-2000). It was accompanied by a 60% improvement in process capability and 31% improvement in product quality. We attribute this improvement to several factors: smoother transition between optimum set points determined by the process optimizer, improved disturbance rejection due to controller installation and tuning, and reduced operator intervention because of controlled operation at or near the optimum set points.
