Modeling and optimization for energy efficient large scale cooling operation

Optimal chiller loading (OCL) is described as a means to improve the energy efficiency of a chiller plant operation. It is formulated as a multi-period constrained mixed integer non-linear optimization problem to optimize the total cooling load distribution through accurate chiller models. OCL is solved as a set of quadratic programs using sequential programming algorithm (SQP) in MATLAB. Based on application of the methodology to chiller systems at UT Austin and a semiconductor manufacturing facility, OCL can result in an annual energy savings of about 8%. However, the savings may reduce considerably in case of additional physical constraints on overall plant operation. With the addition of thermal energy storage (TES) to the system, OCL can reduce the daily cooling costs in the case of time varying electricity prices by 13.45% on an average. The energy efficiency of a chiller plant as a function of its chiller arrangement is studied by using fitted chiller models. If all other variables are kept same, chillers operating in parallel consume up to 9.62% less power as compared to when they are operated in series. Otherwise, chillers may operate up to 12.26% more efficiently in series depending on their chilled water outlet temperature values. The answer to the optimal chiller arrangement can be straightforward in some cases or can be a complex optimization problem in others.