Physical Phenomena

When using CFD to study a fluid dynamics problem there are numerous details that should be considered in order to ensure useful results. Some of these issues are not obvious and are the focus of the articles in this section titled Physical Phenomena.

A question that often arises is at what Reynolds number is a computational model likely to be accurate? The article Reynolds Number Restrictions in CFD addresses this question by providing a discussion of both high and low Reynolds numbers where limitations may seriously affect a simulation. Another important question is whether or not it is necessary to use numerical approximations that satisfy the basic fluid conservation laws of mass, momentum and energy. Generally, it would be thought that satisfying these laws is a good thing, but as the article To Conserve or Not explains, this is not always the case.

Along similar lines, it is not always recognized that there are two, not one, conditions for a fluid to be incompressible; and there is also more than one possible specification for a pressure or outflow boundary, depending on the physical situation that is to be modeled. These topics are covered in the articles The Incompressibility Assumption and several articles that discuss Boundary Conditions.

An always present consideration is whether the numerical method to be used is computationally stable. Or, if it is stable, is it excessively dissipative? These topics are addressed in several articles in the CFD-101 collection. In particular, the topic of stability is presented from several points of view that include the traditional von Neumann Fourier Series approach, a non-rigorous, but often more useful approach based on truncation error analysis and finally some simple tests that may provide answers with very little effort. The purpose of presenting these different approaches is to encourage users of CFD to always be open to possible alternative ways of solving problems.

Beyond these specific modeling techniques there are more general considerations necessary before undertaking a CFD project. For instance, should the approach be based on a Lagrangian or Eulerian methodology for describing a fluid, or would it be possible and/or useful to introduce discrete particles. What might be the best way to describe freely moving fluid boundaries or moving solid boundaries? Many of these issues are discussed in this series of articles under the title of CFD-101.