Chilled water plants are complex and dynamic systems that are difficult to accurately model. Core to an accurate analysis of a chilled water plant is the component models used for the chilled water plant equipment. This paper documents techniques to create calibrated simulation models of electric chillers from manufacturers' and short-term monitored data. The authors present two techniques: (1) standard linear regression for well-formed data sets and (2) a reference-curve method that can be used with small or limited data sets through the application of a library of well-formed regression models. These techniques, incorporated into a publicly available automated tool, have been applied to hundreds of chillers. This paper describes the authors' experience with these techniques and provides insight into the lessons learned in their application.