Optimal Scheduling of Variable Speed Pumps Using Mixed Integer Linear Programming -- Towards An Automated Approach

TL;DR

This paper presents a novel MILP-based methodology for optimizing variable-speed pump scheduling in water distribution networks, enabling reliable global optimal solutions through linear approximations.

Contribution

It introduces a systematic approach to formulate and solve pump scheduling problems with VSPs as MILPs using piece-linear approximations, improving solution robustness.

Findings

The method reliably finds global optimal pump schedules.

The approach is demonstrated on a simple two-pump network.

Results show the formulation's robustness and practical applicability.

Abstract

This article describes the methodology for formulating and solving optimal pump scheduling problems with variable-speed pumps (VSPs) as mixed integer linear programs (MILPs) using piece-linear approximations of the network components. The water distribution network (WDN) is simulated with an initial pump schedule for a defined time horizon, e.g. 24 hours, using a nonlinear algebraic solver. Next, the network element equations including VSPs are approximated with linear and piece-linear functions around chosen operating point(s). Finally, a fully parameterized MILP is formulated in which the objective is the total pumping cost. The method was used to solve a pump scheduling problem on a a simple two variable speed pump single-tank network that allows the reader to easily understand how the methodology works and how it is applied in practice. The obtained results showed that the formulation is robust and the optimizer is able to return global optimal result in a reliable manner for a range of operating points.