1 Abstract—This paper summarizes the key results of testing work performed by three organizations (EPRI, SCE, and BPA) on a total of twenty seven air conditioning units in order to better understand and thus characterize their behavior for power system simulations. The diversity of the tested air conditioner units included sizes (tonnage), compressor technology (reciprocating and scroll), type of refrigerant (R-22 and R-410A), efficiencies (between 10 and 13 SEER), and vintage (new and old). A common test plan was developed by the three organizations. The tests were then performed independently by each of the three organizations. The EPRI work was sponsored by APS and SRP. This effort was part of the current load modeling effort going on in WECC under the Load Modeling Task Force. The key findings of this work are presented here together with a description of the testing methodology. All three organizations found very similar results despite testing a variety of different sizes and manufacturer units. The key results presented are associated with the stalling behavior of the units at different outdoor temperatures, the behavior of thermal overload tripping, contactor dropout, and the behavior of the units in response to different emulated types of system events. Index Terms— residential air conditioners, motor stalling, thermal overload protection I. INTRODUCTION HE demand for residential air conditioners (RAC) has grown significantly over the years. According to the latest Department of Energy (DOE) survey available for the year 2001, about 80% of US households have either central or room air conditioners [1]. Although today’s units are 30% more efficient than those in the 1970s, there has been a 228% increase in the number of RAC during the period of 1978- 2001 [2]. This increase in the number of RAC has posed problems for utilities in recent years. Particularly, many utilities in Western Electricity Coordinating Council (WECC) have experienced delayed voltage recovery following faults in the transmission system. In addition, it is believed that the increased level of low inertia motor loads (e.g. RAC) tend to decrease the damping of large power oscillations on major inter-ties such as the California-Oregon Intertie (COI). Particularly in the case of delayed voltage recovery, attempts to simulate these events using conventional load models (i.e. polynomial and exponential form of static load models and motor models that do not adequately capture the stalling behavior) were not successful. This delayed voltage recovery is attributed to stalling of RAC. It was evident that a suitable representation of these single-phase RAC in dynamic studies is critical for successfully emulating the delayed voltage recovery phenomenon in dynamic simulations. The first step in improving dynamic models is to understand the behavior of RAC when subjected to under voltage conditions. WECC-wide efforts were undertaken to characterize behavior of RAC when subjected to voltage and frequency variations. Arizona Public Services (APS) and Salt River Project (SRP) contracted Electric Power Research Institute (EPRI) to test eleven RAC and develop a more representative load model in GE PSLF ® . Southern California Edison (SCE) and Bonneville Power Administration (BPA) tested ten and six units, respectively, at their own test facilities. A detailed test plan was jointly developed by SCE, BPA, and EPRI. A summary of the testing work performed and results obtained by EPRI, SCE, and BPA is presented in this paper. A brief overview of the test procedure and setup is given in section II and Appendix A. A summary of the tests performed and results obtained by EPRI, SCE, and BPA are given in section III. The test procedure had voltage and frequency variation tests. Only under voltage tests were performed by EPRI while BPA and SCE performed under voltage and frequency variation tests. II. TEST PROCEDURE A schematic of the test setup showing wiring of a typical RAC (Fig. 21) is given in Appendix A. The indoor (air- handler that houses the evaporating coil) and outdoor (condenser unit that houses the compressor) units of the RAC were kept in separate environmentally controlled chambers to simulate actual residential loading conditions. The tests were performed at three different outdoor temperatures: 80°F, 100°F, and 115°F. The indoor temperature was kept constant at 70°F with 70% humidity. Three types of undervoltage disturbances were used to test the behavior of the RAC. These are described below: A. Rectangular Voltage Sag Rectangular voltage sags were generated by a drop in voltage that was held for 30 seconds (Fig. 1). This type of sag was used to determine the stalling threshold voltage and contactor drop out voltage. Results of Residential Air Conditioner Testing in WECC A. M. Gaikwad, R. J. Bravo, D. Kosterev, S. Yang, A. Maitra, P. Pourbeik, B. Agrawal, R. Yinger, and D. Brooks T