© Photo | Prof. Dr.-Ing. Wolfgang Elmendorf

“Measure­ments have the last word”

Prof. Wolf­gang Elmen­dorf on the poten­tial and limi­ta­tions of measure­ments and calcu­la­tions for product data.

Where do product data come from, and how depend­able are they?

Both depend on the approach taken by the busi­ness. There have been cases in which data was regu­larly extrap­o­lated for years on the basis of measure­ments made only once. Of course that’s not a depend­able solu­tion, but it’s also an excep­tion. The clas­sical approach is to deter­mine the data on test stands, because in the end, measure­ments have the last word when it comes to reli­a­bility.

What require­ments have to be fulfilled by such test stands?

First, test stands have to be set up according to stan­dards; the primary stan­dard applic­able to fans is DIN EN ISO 5801. Test stands that comply with this stan­dard provide a suit­able basis for compar­isons. Second, an organization’s quality manage­ment spec­i­fies the processes for performing and analyzing measure­ments; in today’s indus­trial land­scape that is governed for the most part by DIN ISO 9001. Then the processes that lead to the results are also repro­ducible.

In your expe­ri­ence, does that really lead to compa­rable data?

As a rule, yes. If the compar­ison result devi­ates signif­i­cantly, then the setup condi­tions were not complied with completely — often because they’re unknown. So the setup has to be clearly defined and docu­mented.

How real­is­ti­cally can the oper­ating condi­tions even be repre­sented in tests and calcu­la­tions?

The ques­tion to begin with is: What do I need the results for? If I want to compute complex fan geome­tries, then with some expe­ri­ence I know that I can make certain simpli­fi­ca­tions that will only have a marginal effect on the results.

Prof. Dr.-Ing. Wolf­gang Elmen­dorf

Prof. Dr.-Ing. Wolf­gang Elmen­dorf heads the labo­ra­tory for flow machines at Heil­bronn Univer­sity. His work centers around the devel­op­ment of fan test stands, exper­i­mental fluid dynamic analyses, theo­ret­ical fluid dynamics calcu­la­tions, and the opti­miza­tion of fluid flow systems.

For example, there are many fans in which the blades are not spaced evenly for aeroa­coustic reasons. That has no effect on the air perfor­mance. So if I’m inter­ested in venti­la­tion results in a compu­ta­tion, I can also space the blades evenly. But if I’m inter­ested in aeroa­coustic results, then I have to take the complexity into account. The hard and fast rule is: A fan never works alone. It’s always in an envi­ron­ment. To assess the true oper­ating behavior, I need to know the char­ac­ter­is­tics of both the fan and the system.

And then come results that are close to reality?

It may happen that not all factors can be taken into account in a compu­ta­tion. But these short­com­ings can also arise in measure­ments since you can’t always make a one-to-one simu­la­tion of the orig­inal customer situ­a­tion. If the customer wants to know the exact behavior, the only way to do so is usually to test the fan in the device.

The exper­i­mental setup for every measure­ment has to be clearly defined and docu­mented.“

In coop­er­a­tion with the customer, the oper­ating behavior in a special instal­la­tion situ­a­tion can be predicted very reli­ably in this way. But it can be computed only with an unrea­son­able amount of effort.

What is better, measuring or computing?

Both together! In some fields, you can make better progress with measure­ments and in others with compu­ta­tion. For example, better computing methods allow for faster product cycles. The reli­a­bility of modern compu­ta­tional methods is outstanding and continues to improve. But without vali­da­tion from measure­ments they aren’t produc­tive. And at the latest, the customer will demand this veri­fi­ca­tion.

What should one watch out for when comparing measure­ments and compu­ta­tions?

With a compu­ta­tion you will not exactly match the measure­ment. Often you’ll system­at­i­cally over- or under­es­ti­mate para­me­ters such as pres­sure increase or effi­ciency — and then notice that the results only undergo a parallel shift. So if I achieve improve­ments in the compu­ta­tion by changing geometric para­me­ters, I can transfer those one-to-one to the measure­ments. In this way, compu­ta­tion can iden­tify factors I can use to make improve­ments — much faster than with measure­ments.

How much can such results improve the effi­ciency of equip­ment?

In venti­la­tion appli­ca­tions, aero­dy­namic opti­miza­tion plays a crucial role. There is still poten­tial for improve­ment here, even though it’s getting smaller all the time. You have to focus on the effi­ciency chain. The impeller is a major factor, but all the other compo­nents play a role as well. The impeller can be viewed in terms of various phys­ical causes of effi­ciency losses such as fric­tion and static pres­sure. The effi­ciency can be increased simply by improving the inlet nozzle or the outlet area, with effects on both the aero­dy­namic perfor­mance and on the aeroa­coustic side. Geometric details also offer poten­tial. However, the result of an analysis can some­times be that it’s impos­sible to make full use of the aero­dy­namic poten­tial, for example because of required space limi­ta­tions.

Do analysis methods also offer poten­tial?

Math­e­mat­ical opti­miza­tion strate­gies are being applied with increasing frequency. They involve taking the results of fluid dynamics calcu­la­tions and using math­e­mat­ical methods to derive the optimum config­u­ra­tion of the para­me­ters. That’s very labo­rious since it’s a multi-vari­able problem. And such methods typi­cally also require vali­da­tion by exper­i­ment in the end.

How impor­tant are measure­ment results to customers?

The customer needs the measure­ments but also needs help in inter­preting and imple­menting them. It has to be a matter of course for manu­fac­turers to provide clar­i­fi­ca­tion in their state­ments by means of stan­dards-compliant recording of measured quan­ti­ties and clear iden­ti­fi­ca­tion of para­me­ters. Open deal­ings between the parties are called for. Customers have be clear about what they want. That sounds simple but is actu­ally astound­ingly compli­cated because when they design their prod­ucts, customers often don’t know the require­ments of their own customers yet.

Required fields: Comment, Name & Mail (Mail will not be published). Please also take note of our Privacy protection.