Pumping systems are considered among the largest energy consumers in many production facilities. As such, operations optimization of these systems is an important ingredient of a successful energy efficiency optimization effort. Some industries nowadays do have offline pump load management advisory systems that are based on analyzing predefined operating scenarios. However, the variability of a process and its effect on a pumping network's dynamics can make it difficult to predict and analyze every operating condition and to come up with optimal network settings beforehand, which can result in missed energy optimization opportunities. Online optimization of pumping systems operations can capture some of these missed opportunities, and as a result, an estimated saving of up to 3% in the energy/electricity consumption of big pumping systems can be attained.

The pumping systems on-line optimization problem addressed in this paper is a typical supply-demand constrained optimization problem under uncertainty. This paper addresses a modeling and optimization method that can be applied to a variety of configurations, including series, parallel, and hybrid systems, and can also account for the settings of controllable equipment such as control valves and adjustable speed drives.

A number of optimization formulations have been considered for the problem at hand. The initial approach used a mixed integer nonlinear program that required some remodeling to eliminate the non-linearity from the integer variables, but the resulting model required advanced solvers and did not always converge to the global optimal. The second attempt used linearization techniques to transform the problem into a mixed integer linear program, but the size of the model grew rapidly. A repeated simulation approach was finally used and has shown to give reliable results even using simple solvers such as MS Excel's standard solver.