To scale model test data there is a technical and economical need for a correction method, which fulfills five tasks: It should be (i) physically based, (ii) understandable and easy to apply, (iii) universal, i.e. applicable to centrifugal as well as to axial machines of different specific speed. More over the method should (iv) account for the aerodynamic quality of the machine and should (v) be reliable not only at the point of peak efficiency but also at off peak condition. Up to now, no method meets all five tasks. To fill that gap, a method developed at TU Darmstadt together with FLT (Forschungsvereinigung für Luft- und Trocknungstechnik e. V.) is introduced, validated, and critically discussed in comparison with other methods.