Comments on “optimum Design of a Two Step Planar Diffuser: A Hybrid Approach”

This letter is in response to the Engineering Applications of Computational Fluid Mechanics, Vol. 4, No. 3, pp. 415-424 (2010) article “Optimum Design of a Two Step Planar Diffuser: A Hybrid Approach” by Suman Ghosh et al. The topic of the paper is interesting and there is not any major problem in the mentioned paper, the comments listed below are intended to suggest some ways to derive the same results more easily and in some cases to suggest a way to derive better results. 1. The authors of paper have used a CFD package, namely Gambit & Fluent and combined it with an optimization package which uses Genetic Algorithms (GAs) in its optimization process. The authors have tried to combine them using C programming language to have fully automatic software. In fact all of these processes are provided in NUMECA software, this software can optimize turbomachines and channels (or diffusers) by combining GAs, Artificial Neural Networks (ANN) and CFD so using this software in this paper is much easier. 2. The surname of Mohammadi in Mohammadi and Pironneau (2001) is wrongly written as Mohamma. 3. At the end of the paper the authors have used a software, namely Labfit for modeling the optimized value of “s” and “h” with respect to input values of Re and Ar. We do not know the basis of Labfit for modeling (GAs, etc.), but it was better that the authors used modern ANN, namely GMDH for modeling the outputs (Nariman-Zadeh et al., 2003; Jamali et al., 2009). We have used the data of Table 1 in the paper for GMDH modeling and the details for “s” are as follows