Test Rig Effect on Performance Measurement for Low Loaded Large-Diameter Fan for Automotive Application

作     者：Manuel Henner Bruno Demory Franqois Franquelin Youssef Beddadi Zebin Zhang 

作者机构：Valeo Thermal Systems- 8 La VerriOre 78321 France 

出 版 物：《Journal of Mechanics Engineering and Automation》 (机械工程与自动化（英文版）)

年 卷 期：2014年第4卷第12期

页      面：924-936页

学科分类：080903[工学-微电子学与固体电子学] 080706[工学-化工过程机械] 0809[工学-电子科学与技术（可授工学、理学学位）] 08[工学] 0807[工学-动力工程及工程热物理] 080501[工学-材料物理与化学] 0805[工学-材料科学与工程（可授工学、理学学位）] 080502[工学-材料学] 

主　　题：Fan performance measurements test rig validation torque uncertainty. 

摘      要：Large diameter fans with low solidity are widely used in automotive application for engine cooling. Their designs with small chord length help reducing the torque on the electrical motor and providing a good aerodynamic compromise between several operating conditions, some of these being at high flow rate. Their global performances are measured according to the ISO standard DP 5801, which allows comparison of results from different facilities. However, some variations in global performances are observed when considering results from two different test rigs. On a fan selected for the purpose of this study, up to 6 % of efficiency is lost on the worst case. As efficiency is more than ever a key factor to select a component, some experimental and numerical investigations were conducted to analyze the fan behavior on each facility. Two sets of measurement and simulation are performed and compared. Geometries considered for the domain of computation include the test rig plenum, the torquemeter, the ground and a large domain for the atmospheric conditions. The exact fan geometry with tip clearance and under-hub ribs is also considered. Numerical results show a good agreement with experiment in both cases when convergence is reached and for low flow rate when computations are switched to unsteady mode. Comparisons show that simulations are able to capture the different fan behaviors depending on the confguration and those efficiency losses previously observed are correctly predicted. These results are further analyzed to perform some post-processing. Blade loading remains identical for both cases but disparities appear in the wake and its interaction with the surrounding. Tiny details that are often neglected during experiment and/or simulation appear to be the cause of slight variations. Position of the torquemeter and shape of the plenum are among the parameters that various and that have cumulative effects. Efficiency being a ration of pressure and torque, variations ar