CFD software gets easier but users need to be aware

The latest developments in computational fluid dynamics software are to make it easy enough for the non-CFD specialist engineer to use. This does not mean, however, that it can be applied to difficult problem by any muggins in the design office and with the expectation of meaningful results. There are many inherent assumptions in CFD analysis and we are advised that non-experts would be wise to limit themselves to applying it to design tasks fairly similar to others previously analysed and experimentally validated by experts. The two leading players, Fluent, and CD adapco, have both announced major enhancements to improve ease of use, but in depth questioning by Eureka reveals that their software should still be used with some caution. Fluent has just announced a development called, Flowizard. A PR person, whom we shall not name, told us that it would even be suitable for a surgeon to use to analyse blood flow in and around the human heart. Since we are aware that blood flow is both turbulent and laminar, according to where it is flowing in the body, and what stage it is at during the pumping cycle, we asked that this claim be referred to one of the company's technical experts. Ben Simpson, from Fluent Europe responded, "Transition of flow from laminar to turbulent and vice versa is very difficult (some might say impossible or impracticable) to do in any CFD package. The usual approach even in a sophisticated package such as Fluent is to model the flow with either a low Reynolds number turbulence model or run the problem twice, once laminar and once turbulent to get a handle on the importance of the turbulence. This type of problem should not be done in Flowizard. I would recommend using Fluent 6 to investigate such applications." We then put the same problem to Stephen Ferguson, technical marketing manager of CD adapco, which has just announced their ability to do away with CAD translation, by embedding Star CD in leading edge CAD packages such as Catia 5, SolidWorks, ProEngineer, and Unigraphics. Meshing is automatic and the software is associative, so that changing the CAD model allows the automatic update of the CFD analysis. However, applied to the heart problem Ferguson said, "The software is not good at transitions, no CFD package is. CFD is best used as a tool to compare fairly similar designs. It will never get rid of experimental work." Accepting that the capabilities of any other CFD package are limited, Flowizard aims to save much time in comparing the effects of small changes to designs, by automating the analysis process, where possible. For example, it asks whether the flow is turbulent or laminar, and offers "I don't know" as an option. The software then calculates a Reynolds number based on the boundary conditions and cell size, and decides whether the flow is highly turbulent, mildly turbulent or laminar, and selects what it thinks is the most appropriate option. It also offers a slider bar that asks the question as to whether the user wants a quick solution or an accurate one. The more accuracy requested by the user, the finer the mesh. If reaching a solution proves difficult, and convergence criteria are not met, the software will advise the user that the solution is not complete, and offer to run on for a specified time period, or look at the results anyway. If the user is having problems, a more experienced user can join the session and check the set up. In addition, Flowizward files can be read into Fluent for a closer examination by an experienced user if required. Ben Simpson advises that compressible flow, cavitation and transient phenomena are not suitable for Flowizward and, "Should be tackled using Fluent 6 and by a trained, experienced CFD user." To that observation, we feel we must add the opinion of Brian Spalding FRS, emeritus Professor of Heat Transfer at Imperial College, who once told Eureka, that much of the problems with using CFD arise from imperfections in the physical understanding of fluid flow, especially under extreme conditions. It has, for example, been suggested that collapsing bubbles are sometimes capable of producing the conditions suitable for nuclear fusion. Such problems remain at the very leading edge of physics and hotly debated. Fluent Europe CD adapco Group Pointers * Flowizard greatly automates Computational Fluid Dynamics but it may be wise to restrict its use to problems similar to those already analysed and verified * Latest versions of Star CD can be embedded in leading edge CAD packages and are associative, so that changes to the CAD model can be made to automatically update solutions * CFD in no way does away with the need for experimental verification