Inverse modeling, Parameter estimation, Vapor compression cycle, Gray-box models, Component-based models
With the emergence of fault detection and diagnostics tools for air conditioning systems in the market, a fair and reliable evaluator, which tests the tools with a large array of data points from a variety of conditions and types of units, is needed. However, the number of data points necessary for reliable evaluation is too large to be generated through physical experiments. Also, existing forward models are difficult to employ for this application because of the requirement to have knowledge of many difficult-to-obtain component parameters and because of very long computation times. To address the issue, a gray box modeling approach is being developed to account for the effects of both operating conditions and faults on performance. This gray-box approach uses experimental data and inverse modeling to determine the values of parameters for each component of a vapor compression cycle. This has led to a fast and robust component-based model that is trained with a limited set of experimental data from normal and faulted conditions, and a few readily available geometrical measurements. Existing component modeling approaches have been simplified to reduce the number of parameters and computational costs. During parameter training for each component, optimization of a cost function is carried out to minimize residuals between experiments and simulation. This paper presents the models, training approaches, and validation results for individual components for a 3-ton R410A packaged air conditioner. Component models constructed included compressor, condenser, evaporator, fixed orifice expansion device, and refrigerant pipes. A companion paper presents the system-level modeling and validation, along with models and results for simulation of faults.