Charge inventory system modeling and validation for unitary air conditioners

Todd Michael Harms, Purdue University

Abstract

Charge inventory, the accounting of distributed fluid mass in closed systems, is important to the modeling of vapor compression systems. Public domain simulation models used to predict the performance of unitary equipment are currently unable to accurately determine charge inventory. Several issues have been identified as sources of error in these models including incomplete internal volume accounting, absent refrigerant-oil diffusion effects, and void fraction uncertainty. The goal of this work is to improve the accuracy of charge inventory system modeling. This includes identifying and evaluating the relative importance of the various modeling issues. Initially, detailed models of the heat exchangers were developed to isolate charge inventory modeling issues from the overall system dynamics. Three commercially available air conditioners were fully instrumented and tested to provide inputs to the component models and to provide model validation. The component models were further developed into two types of system models: a detailed, local model and a simplified, global model. The simplified model was compared to the detailed system simulation to examine the balance between execution speed and prediction accuracy. The present results suggest that the most important charge inventory issue is proper void fraction determination. This was true for both the component modeling results and the system modeling results. For the component models, the Baroczy void fraction correlation gave the best agreement with the measured charge inventory. However, models that incorporate a dependence on the Froude number give the most accurate description of void fraction over a wide range of conditions. As such, the combination of the Yashar model for separated flow and the Taitel and Barnea model for slug flow is recommended for broad use in charge inventory system modeling. This combination of void fraction models performed well in the system models.

Degree

Ph.D.

Advisors

Braun, Purdue University.

Subject Area

Mechanical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS