In this Flow example, I have an inlet pipe diameter 2 times larger than its corresponding outlet pipe diameter with a turbulent flow rate. Below is a descriptive picture of this pipe with two different configurations: (A) at a small angle of 3°, which causes a very long transition, and (B) a larger angle of 30° resulting in a short fitting.

Pipe A

Pipe B

Which configuration results in a smaller, thus better, pressure drop? Better yet, what is the best angle to reduce the pressure drop to a minimum. It is very easy with configurations (or a design table) to create a fitting with a transitional angle at every degree from 3 to 30. We can leverage these configurations within SW Flow Simulation with Cloning and Batch Runs to complete the task.

What we find out is that the best angle is 11°, as shown in the graph of the results below (only values until 13 are shown). This is because if the angle is too small, then the overall pipe length is longer resulting in frictional pressure losses. Larger angles creates a sharp velocity increase and the resulting pressure drop goes up. Isn't this a perfect reason to use a virtual test bench to find out what the best design is without spending the money and taking the time to build the physical test!

Could I have done this same task with the Parametric Study tool in Flow Simulation? Not exactly, that tools requires that you have a set value defined by a Goal that you are hoping to converge upon, thus the Parametric Study tool is not an optimization tool.

Copyright © 2010 Dassault Systèmes SolidWorks Corp. All rights reserved.

Do not distribute or reproduce without the written consent of Dassault Systèmes SolidWorks Corp.

Optimization for CFD is not as straight-forward as it is for FEA. To do a Design of Experiments optimization algorithm there are a number of required runs in order to determine the optimal solution, thus it would require a bit of computational time.

A short search for other CFD products optimization capabilities confirms the issues of shape optimization (see bottom of link); and those who do are "high-end," like Hypermesh, and do not compete in the same space as SW Flow Simulation.