CFD Models for the Analysis of Rotor-Only Industrial Axial-Flow Fans

D3 - Design and Optimisation

The actual performance of industrial fans has not yet reached the level claimed as possible by a considerable amount of up-to-date literature in which organized campaigns of CFD calculations are performed using optimisation algorithms. On one side this is due to a well-established know-how often considered as “satisfactory” by the manufacturers. On the other side there is still a lack of confidence in the technical and economic advantages deriving from time consuming optimisation techniques.
In this context, local and total performance of an actual rotor-only axial flow industrial fan have been evaluated by experimental tests and CFD calculations. Different degrees of the overall model complexity and amplitudes of the computational domain are considered in the latter to find the best compromise between accuracy of the results and saving of computational effort and costs. In particular, 4 different computational domains were analysed:
*A single blade passage without tip clearance and idealized meridional geometry;
*A single blade passage and idealized meridional geometry;
*A single blade passage with tip clearance and actual meridional geometry (including bell-mouth entry, electric motor and other ancillary components);
* Entire fan within an idealized fan test-rig assembly.
All computational domains were discretised taking advantage of the innovative polyhedral grid meshing capability of CFD commercial codes. The importance of the proper grid size for scientific and industrial applications is discussed also accounting for the turbulence model selection. When the analysis is carried out at the maximum level of detail, more than 1,500,000 cells are needed to get reliable results for each fluid volume containing the impeller blade channel. When, instead, industrial applications are considered meshes with 600,000 cells are acceptable if two-equations one-layer turbulence models are used.
Results show that single blade channel stationary calculations with an idealized meridional passage show the better predictions of the total fan performance and efficiency in the stable operation range, also because of a “compensation” between the excessive dissipation that is predicted and the geometry idealization. Instead, reliable analysis of the rotor flow field can be performed only by modeling in detail the actual meridional geometry. In addition, calculations of the entire fan domain on a very rough grid show capability of performance and efficiency prediction that appears very attractive for fan industry.