Multi-Regime Shape Optimization of Fan Vanes for Energy Conversion Efficiency using CFD, 3D Optical Scanning and Parameterization

Abstract

An enhanced reverse engineering procedure was developed for roof fan redesign. An original numerical workflow for robust shape optimization based on maximum energy conversion efficiency was developed. It operates using a sample of multiple operating regimes coupled with CFD simulations. The initial shape solution was originally obtained in point cloud form by optical 3D scanning and subsequent B-spline based parameterization of shape. The CFD simulation of the scanned shape using 3D RANS based software was shown to agree very well with the measured features, experimentally obtained in our lab with the actual initial-shape fan. By manipulating the control points of parametric curves, the developed evolutionary optimization workflow was subsequently able to create shape-optimized vanes. This original procedure was applied to cases of constant-thickness and profiled single curvature vanes, both for single-regime and robust multi-point operating conditions. The corresponding increase in efficiency gained by our computational procedure was correlated with respective velocity and pressure distributions and suppression of flow separation. The novel numerical procedure developed here therefore provides a numerical framework for generic object geometry to reshape itself autonomously. The change in shape ensures maximum energy conversion efficiency for a given composition of operating regimes. The gain in efficiency with optimized vane shapes proves to be significant in the wide range of flow rates around the best efficiency point.

3D shape optimization of fan vanes for multiple operating regimes subject to efficiency and noise-related excellence criteria and constraints

“…Case 1 is the reference case, with vanes optimized as single curved (or 2D) surfaces (Milas et al, 2014). For comparison with other designs, the results for the 2D optimized fan for a single flow rate designated Case 1 are presented first.…”

“…The CFD model was validated by comparison with the experimental results available for the initial fan design with z = 14 flat vanes. The details of the experimental setup and experimental uncertainty are presented in Milas et al (2014), who also illustrate that the CFD prediction follows the trend line of the experimental results. The agreement between the CFD model and the experiment was very good for a design flow rate ofV = 1000 m 3 /h.…”

“…An unstructured (tetrahedral) mesh is adequate for the local refinement near the vane leading edge and the vane-shroud corner areas. Testing of the CFD results using grids of different resolutions (Milas et al, 2014) showed that grid-independent results are obtained with 100,000 finite volumes inside one impeller segment of D 2 = 325 mm impeller diameter. For simulations where the number of vanes is different from the initial condition (z = 14), the number of finite volumes was changed proportionally.…”

“…If the improvement in efficiency is not significant during several consecutive generations, the process is terminated. In previous study (Milas, Vučina & Marinić-Kragić, 2014), about 1,000 iterations were required to reach the optimal solution for the 2D shape optimization case, while in this paper for the 3D shape optimization cases 15,000 iterations were usually required. In this paper, the genetic algorithm (Goldberg, 1989) was used to achieve a nearoptimal solution and the simplex method was used to achieve the final optimal solution in order to increase the convergence speed.…”

“…Optimization for multi-point operating regimes was accomplished by generating a statistical sample of the assumed overall lifespan operational conditions of the fan. The CFD model was validated by extensive laboratory tests by Milas, Vučina, and Marinić-Kragić (2014) to confirm its compliance with the actual performance before being coupled with the optimizer.…”