Correlation allowances in model tests results - A delicate balance between performance, accuracy and commercial interests?

Model tests at ballast and design draught are used to convert the sea trial results from the ballast trial draught to the contractual design draught. Correlation allowances in model test results and their effect on the trial performance prediction are of major importance. Nowadays it is not only typical to verify the contract speed but also the EEDI certification requires a verification of the speed power performance of the vessel. The use of a to favorable C _A-value may lead to attractive performance figures, but also leads to higher fuel consumption figures than expected. Furthermore the design point of the propeller is affected, which leads to a too low light running margin and in some cases to erosive cavitation.
During a study, large spreading in the values of the correlation allowances for design draughts have been found for merchant vessels tested at different model test institutes, but at ballast trial draught the spreading is much less. Can it happen that some institutes select favorable correlations allowances on the basis of inaccurate trial data of shipyards? Or should we accept a large spreading in correlation allowances and have these indeed been confirmed by sea trials at design draught? This paper will present a discussion using the experience of a large full scale trial database as well as the accuracy of model and full scale tests.