15 th International Conference on Experimental Mechanics ICEM15 1 ABSTRACT In the present paper, experimental results for a 5 kW capacity steam ejector with variable primary nozzle geometry are presented and compared to numerical simulations. The variable geometry was achieved by applying a movable spindle in the primary nozzle. Operating conditions were considered in a range that would be suitable for an air-conditioning application, with thermal energy supplied by vacuum tube solar collectors. The numerical results were obtained using CFD. The experimental entrainment ratio varied in the range of 0.1 to 0.5 depending on operating conditions and spindle tip position. It was found that the primary flow rate can be successfully adjusted by the spindle. CFD and experimental primary flow rates agreed well, with an average relative error of 7.7%. CFD predicted the secondary flow rate and entrainment ratio with good accuracy only in 70% of the cases. INTRODUCTION Air conditioning represents a growing portion of electricity consumption in both developed and developing countries. Its increase over the last years is so dramatic that in many European countries the peak of electricity consumption is switching from winter to summer.Thus, an efficient solution to reduce electricity demand would be the use of refrigeration systems that are powered by solar thermal energy. Ejector cooling seems to be an attractive technology because of its structural simplicity and low capital cost when compared to e.g. an absorption refrigerator. Although the coefficient of performance (COP) of an ejector cycle is relatively low, ejectors don’t have moving parts, thus require little maintenance and have a long lifespan. In order to achieve a better ejector cooling performance and, thus better economic perspectives, a number of research works have been carried out in the past few years. Several experimental works have been published, and the most recent results can be found in: Aphornratana and Eames (1997); Chunnamond and Aphornratana (2004a); Eames et al. (2007); Yapici et al., 2008); Huang and Chang (1999); Sankarlal and Mani (2007); Chang and Chen (2000); Eames (2002); Cizungu et al. (2001); Godefroy et al. (2007); Sun (1999) and Hernandez et al. (2004). Alternatively, mathematical models can be used for analysing the performance of an ejector or the entire refrigeration cycle. A review of available models can be found in He et al. (2009). The major advantages of modelling, compared to experimental studies, are its reduced cost and the capacity of producing a large amount of performance data in a short time. Nevertheless the theoretical studies with experimental validation contribute to an important added value to the numerical investigations. !" # $%& ’ ()*( # +(( , # (&&( ’ $* , # -*( ./% , # 00 ’ (001 , !"# , *21$*%1 0 3(1 %)(%$*%1# %()*(14 0 11(%/$# 11(%/$# 56 ,# 7 $ %&’()()