New Development of VRF System Utilizing the Big Data Technology

Liu Hua

All industries have met the age of the big data. The key of the big data technology is to mine the massive data professionally, and discover its technical and commercial value. GREE has built an efficient and fully-functional big data system, which collects over 100G of data every day. GREE has applied the big data technology to the lifecycle of products, such as the design, debugging, fault diagnosis and prediction.

Based on the big data analysis of over 200 thousand VRF samples in China, it is found that during 60% of total operating hours, the VRF system operates at a load ratio of less than 30%. To improve the system performance of VRF under very low part-load conditions, the compressor with double cylinders has been developed for VRF system. The system COP improves by 130% for the 10% part-load condition, and the lowest cooling capacity reaches as low as 5% of rated capacity.

The big data technology is applied to automatic fault diagnosis and prediction. When a fault is detected in a VRF unit, its related operational parameters are compared to other units that are operating under the same conditions, and the parameters that are deviated from their average levels are found. Causes of faults are diagnosed and solutions are provided automatically. Moreover, through the analysis of the MAP diagram of the compressor, the unit that operates beyond the reliable condition boundary gets a health evaluation, and its possibility of fault in the future is predicted.

Based on the big data analysis, there are large differences between actual operating status of VRF system and the current standards. For example, the current VRF standard doesn’t consider the energy efficiency under 25% load ratio, which is neglected during normal product design. In practice VRF system operates inefficiently at extremely low part-load conditions. Therefore, some new standards of VRF system need to be drafted and released in order to improve the efficiency of VRF system.

KEYWORDS: Big data, Variable Refrigerant Flow, Part load, Fault diagnosis, Standard formulation