CFD Information

I

If you've been solving stress problems using finite-element software, you'll eventually encounter a few that involve fluid flows. These situations are likely to require working with the FEA program's computational fluid dynamics (CFD) section. CFD was once the domain of Ph.D.s and supercomputers. But the hardware and software are so improved that a wider range of engineers now use it. "The guiding principles for getting accurate answers from a CFD program have not changed," says Ulises Gonzalez, a senior engineer for Pittsburgh-based Algor Inc. From his experience solving fluid problems, Gonzalez provides these observations and rules of thumb.

Fluid-flow problems fall into two categories: steady and unsteady. It is also important to note that some problems expected to be steady turn out to be oscillatory, and thus unsteady. CFD software packages generally use iterative methods to converge on the correct solution. Such methods are used to solve both steady and unsteady problems. In the solution of unsteady problems, time integration methods are used to advance the solution from one time-step to the next. In contrast, solutions to steady problems also use integration methods, but in what is called pseudo-time.

As an example of a steady problem, "Consider a fluid flowing through a cylindrical duct with a rubber orifice serving as a partial barrier to the flow," says Gonzalez. "This could be a biological application. Imagine a constant flow through this duct and how it deforms the orifice. But the question is, by how much?" he asks. Even this relatively simple problem involves a multiphysics approach. In this case, we must consider the coupled solutions of a CFD problem for the flow and a finite element stress analysis for the deformation of the orifice.

Consider the case in which the flow-rate through the duct is constant. CFD software can easily give a steady solution for the flow once the shape of the orifice is specified. But the shape is still in question. The first step in the solution to this multiphysics problem is to obtain the flow for the case of an undeformed, flat orifice.

"This first CFD solution can be used to obtain the pressure that the fluid causes on the flexible orifice. This pressure is subsequently used by the FEA portion to obtain the deformed shape of the orifice," says Gonzalez. Now we consider how the change in the orifice's shape affects the flow. "The orifice is now a curved surface with a hole. So the CFD solution using the new deformed orifice produces different pressure values, which will consequently result in yet another shape for the orifice," he says. Note how it appears that we are solving the transient problem of a moving orifice. But, instead, we are solving a steady problem in pseudo-time. When new pressures don't match those from the previous pseudo-time-step, we continue the process. "When they match to within an acceptable tolerance, the job is done," he says.

Oscillations in fluids produce more difficult problems. "Remember the Tacoma Narrows bridge catastrophe?" asks Gonzalez. "The wind speed was constant, yet the bridge kept oscillating. You might think it would get pushed downwind and stay there, like a sail. But it oscillated back and forth." Actually, the bridge was shedding unseen vortices in a periodic fashion. This subtle periodic flow caused the poorly designed bridge to oscillate at its natural frequencies until it collapsed.   (More from FindArticles.com)

Some of the sites given below are recommended by us:

Companies

Information