HI-02-18-2 -- Development and Testing of a Reformulated Regression-Based Electric Chiller Model PDF

This paper presents the de elopment of a new electric chiller model and the testing of it and three public domain electric chiller models. These models enable analysis for the optimization of the design, optimization of the controls, commissioning, and diagnostics for chilled water plants. The four models were tested for their accuracy in predicting power over the range of operation and evaluated for their ease of calibration from readily available data sources for existing and new equipment. In testing and evaluation, each of the three existing public-domain models was found to have limitations. Of particular concern was the inaccuracy of these models in predicting the power usage of chillers with variable condenser water flow and centrifugal chillers operating with variable- speed drives at low loads. This paper describes a reformulation of one of these models to specifically address these issues. The validation testing included both manufacturer-supplied and field-monitored data. In all of the tests the newly reformulated model effectively produced lowest error values for power prediction across a range of chillers including centrifugal, reciprocating, screw, and scroll compressors. Relative to the other models, this new model predicts power more accurately for variable condenser water flow and low-load variable- speed applications.

Units: Dual