Title: Performance of Variable Speed Centrifugal Chillers
Abstract: Centrifugal water chiller performance is rated for full-load conditions at peak condenser water temperatures850 F (29.4° C)-and for specified partial load conditions (Integrated Part~Load Value or IPLV). Most centrifugal chillers are designed to operate at fixed speed, ~'ith load control provided by compressor inlet guide vanes and/or hot gas b:y-pass. The cooling water temperature available for a centrifugal chiller ~ill vary 'ith the seasons, the time of day, and changes in the weather. As the cooling water temperature drops, it becomes feasible to reduce the speed of the centrifugal compressor. This paper describes the performance improvements that follow from reducing compressor speed in relation to cooling water temperature and c.hiller load. The interaction of speed control and inlet guide vane control is discussed. Both single stage and multistage compressor applications of variable speed drive are considered. LlTRODUCTION Both single and multi~stage centrifugal compressors are widely used for liquid cooled chiller applications ranging from about 100 ton (351.7 kW) to 10,000 ton (35,169 kW). In most cases, the compressors are designed to run at constant speed. With such a wide range of application, the system requirements can be equally diverse. They range from installations with constant condenser water temperature regardless of the load, to installations that operate continuously at 100% load ~ith varying condenser water temperature. Most systems operate between these e>..'tremes. The HVAC industry uses a unique operating line, known as the AR1 [1]load line, to model compressor operation for comfort cooling applications. In addition, an averaged performance number, known as the IPLV, is used as an index representative of chiller performance at full and part~load conditions. The AR1load line defines condenser water temperature as a function of chiller load, and has remained unchanged for many years. The IPLV calculation has undergone changes since its original introduction, and is now based on many assumptions of building type, operating hours, and Atlanta's annual dry-bulb temperature profile. The condenser water temperature used for the different load points is not based on weather data; rather, it is a linear distribution between 60° F(15.6°C) at 0% load and 85° F(29.4°C) at 100% load. Many in the industry believe the ARJ load line and IPLV calculation are too conservative, not showing the actual water temperature available to a chiller at different operating conditions. Review of the Atlanta case will demonstrate the origin of these concerns. While the mean wet-bulb temperature coincident with the full-load design condition is actually 74 F (23.3°C), a wet-bulb temperature of 77° F (25°C)[2]-a value observed less than I% of the time in the summer-is used to establish a design condenser water temperature of 85° F(29.4°C). Opponents argue that, in the need to be simple and conservative, ( i) it neglects night time usage of chillers distorting the true load profile; and ( ii) it uses arbitrary condenser water temperature distribution as a function of load. They argue, while it is correct to use the higher wet-bulb temperature to design controls to protect the chiller from surging even during that 1% of the time it is observed, it is not correct to use it for the purpose of comparing performance of different machines. Instead, the mean coincident wet-bulb temperature should be used. According to the standard, at 10% load, the cooling water temperature is assumed to be 62.5° F (16.9°C); the lowest it will ever be as manufacturers promise operation only down to 10% load. H one generates an AR1-t:y-pe load line using the true mean wet-bulb temperature distribution and reasonable tower performance, the water temperature at 10% load would be even less than 55° F (l2.8°C). In fact, many customers demand operation down to 55° F (12.8°C). This means that their real operating line is far steeper than the AR1load line. Furthermore, the new operating line would weigh the lower temperature end of the part-load application heavier than the current one does.
Publication Year: 1996
Publication Date: 1996-01-01
Language: en
Type: article
Access and Citation
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot