Showing papers by "Sreenivas Jayanti published in 2017"

Shape optimization of flow split ducting elements using an improved Box complex method

Abstract: Iterative search methods, such as the Box complex method, can be used for inverse shape design problems. In the present article, an improved version of the Box complex method is proposed specifically for computational fluid dynamics-based optimization of fluid flow ducting elements. The original Box complex method is improved by (1) assigning non-uniform weights for the estimation of the centroid, (2) using a reduced reflection factor for accelerated convergence, and (3) introducing measures to prevent premature breakdown of the iterative process. The success of the improved Box complex method over the original Box complex method is demonstrated on two benchmark functions and by applying it to two fluid flow problems of engineering. The improved method is shown to substantially accelerate the convergence with approximately 50% reduction in computational effort for the T-junction and manifold problems.

Flow Control in T-Junction Using CFD Based Optimization

Abstract: The present work proposes a design methodology to achieve a desired single phase flow distribution in a T-junction. The desired flow control is achieved by contouring the flow path of the duct using three bi-quadratic Bezier curves. A Bezier curve is one of the parametric curves which is defined by a set of a few control points. These curves possess some interesting properties which render them as a good choice for the representation of smooth curves. A bi-quadratic Bezier curve is defined by five control points of which four are fixed in the present work (two control points to mark initial and final positions and another two points adjacent to them for tangent continuity). Different curves, which are nothing but the contours depicting the flow path, are obtained by moving the remaining one control point for each curve. Since in two-dimensional simulation, each control point is defined by two coordinates, in total, there are six optimization design variables for the three Bezier curves. The present work determines the optimal values of these design variables using CFD based optimization. The optimization is carried out using Box’s complex method with the objective function being the minimization of the percentage absolute error in the flow rate in the branch outlet. Constraint surfaces for the design variables are described such that the Bezier curves do not intersect each other. Numerical simulation of the geometry yields the value of objective function. The geometry creation and numerical simulation are performed in GAMBIT and FLUENT respectively in batch mode by integrating MATLAB, GAMBIT and FLUENT to enable automated optimization. Two dimensional numerical simulations were performed on a T-junction for three cases with branch outlet flow rates of 30, 50 and 70 % of the incoming flow rates. The optimal geometry for all the studied cases is determined within an flow rate error of 3 %.