High level of accu­racy

Simu­la­tion tools for aero­dy­namics

We are often asked whether, in view of the top-quality simu­la­tion tools, aero­dy­namic measure­ments are still neces­sary. Indeed, in the last 30 years there has been tremen­dous progress in the ability to predict flow phenomena. Whereas not too long ago it took days and weeks to calcu­late rela­tively simple, often more acad­emic cases on extremely expen­sive super­com­puters; thanks to the enor­mous improve­ment in the computing capacity of computers, huge progress in the devel­op­ment of compu­ta­tion loga­rithms and the contin­u­ally increasing expe­ri­ence in the use of new tools, a completely new situ­a­tion has arisen.

Today, even the proces­sors found in any PC are suffi­cient hard­ware. They are arranged to form so-called clus­ters (i.e. several proces­sors combined in groups). This creates very powerful computers which, compared to the former age of the super­com­puter, are very inex­pen­sive and easy to use and whose oper­a­tion is no great obstacle even for medium-sized compa­nies. Also with regard to compu­ta­tion proce­dures, there has been great progress, leading to an enor­mous improve­ment in the accu­racy of compu­ta­tion results even with complex geome­tries and oper­ating condi­tions. Mean­while there are advanced commer­cial compu­ta­tion programs for various areas of appli­ca­tion.

Whereas initially aero­dy­namic simu­la­tions were carried out mainly for aero­nau­tical engi­neering appli­ca­tions, today there is hardly an area of aero­dy­namics in which the most varied prob­lems are not being exam­ined with the use of simu­la­tion methods.

With regard to our topic of air flow around fans, air perfor­mance and effi­ciency are of partic­ular interest, whereby local flow phenomena such as slip, vortex gener­a­tion and uneven air flow reduce these vari­ables, and should thus be avoided. The advan­tage of simu­la­tion compared with measure­ment thus lies primarily in the fact that the flow field around the blades makes the prob­lems visible, so that they can be specif­i­cally elim­i­nated. In Fig. A, the flow slips at the cover plate and motor, for which reason this impeller has rather low effi­ciency. By means of a more favourable arrange­ment of the flow channel, slip is prevented (Fig. B), air perfor­mance and effi­ciency are thus consid­er­ably improved. This example shows that prob­lems which, in the time before simu­la­tions tools, could only be worked on using many time-consuming measure­ments on the basis of ‘trial and error’, can today be resolved with great accu­racy by means of simu­la­tion.

However, since consid­er­able simpli­fi­ca­tions are some­times neces­sary for the calcu­la­tion (espe­cially for calcu­lating turbu­lence), there may also be cases in which simu­la­tion does not provide satis­fac­tory accu­racy or for which a calcu­la­tion is not expe­dient on account of the exces­sive computing effort. And hence the answer to the initially posed ques­tion, that today and also in future, aero­dy­namic measure­ments are and will remain indis­pens­able. There will be a consid­er­able shift to measure­ments serving to check simu­la­tion results which are required for accep­tance of the final data and with which phys­ical vari­ables are deter­mined which are not (yet) acces­sible to a simu­la­tion, for example noise gener­ated by aero­dy­namic processes.