Optimal chiller loading including transients
Manuel R. Arahal, Manuel G. Satué, Manuel G. Ortega
TL;DR
This work reframes Optimal Chiller Loading/Sequencing as a dynamic optimization that accounts for transient effects during on/off switching. It extends the decision variable set to include mass flows and outlet temperatures, and solves the resulting high-dimensional problem with SPSA, using forward grey-box models to compute costs and constraints. Experimental results on a 3-chiller plant show that incorporating transients reduces switching and lowers energy consumption compared to a static approach, with practical execution times suitable for non-real-time planning. The approach offers a scalable, physics-informed framework for reducing cooling energy use in multi-chiller plants.
Abstract
Scheduling and loading of chillers in a multi-chiller plant is considered. A new framework is introduced considering an extended set of independent variables for the optimization problem of energy consumption. In this way the number of decision variables is increased, providing extra degrees of freedom to optimize cooling plant operation. The dynamic effects due to transients arising from switching on and off of units are usually not considered in the literature dealing with Optimal Chiller Loading/Sequencing which is restricted to the static case. In this paper, these effects are treated in a way that results in a manageable optimization problem. A Simultaneous Perturbation Stochastic Approximation solution is deployed for the problem and the proposed method is compared with a similar but static approach showing the benefits in terms of reduced energy consumption.