UNCERTAINTY QUANTIFICATION IN CFD SHIP RESISTANCE EVALUATION

Ships are the largest moving devices ever created by humans. In the design phase, it is important to define the main characteristic for a ship to run; one of the key features that must be addresses is the power estimation to reach the design speed. Over the years, different attempts to define this value can be listed, once the ships moved from both the muscular and the wind energy to a mechanical propulsion plant. In this scenario, based on Froude’s findings about the similarity between a ship and its scale model, the Experimental Fluid Dynamics (EFD) began its activity in the second half of the XIX century. By that time, the need for a constantly updated and freely accessible guidance grew up in importance among the scientific community. These requirements have been addressed by the International Towing Tank Conference (ITTC). From the second half of the XX century, a new resource became available in order to replicate the EFD experiences; indeed, based on the computing available sources, the Computational Fluid Dynamics (CFD) began its estimations. Both the EFD and CFD approaches, accounts for uncertainties connected with the results obtained. The Uncertainty Assessment (UA) issue began in the last couple of decades of the XX century. With the respect to the CFD, the uncertainty assessment can be estimated based on the following two main phases: the Verification and the Validation. With the respect to a towing tank test, in current research a single numerical set-up has been defined to compare the numerical results with the experimental one with the attempt to define the accuracy for a CFD calculation compared with an EFD result. In order to generate a database where all the results can be stored, 142 simulations have run for 22 different hulls tested at different speeds. The results obtained have been used to define some statistical correlation among the differences recorded and the other input parameters like the hydrodynamic coefficients and the main dimension for the models. With the respect to the UA, two different hulls have been tested at different speeds and configuration for a total number 72 different simulations with the attempt to test the literature available approaches to get the numerical uncertainty assessment. These approaches have been divided into two families, as follows: three and multiple solutions (at least four) families. Finally, based on the findings from current research, the possible answer to the question “how precise is a CFD resistance calculation” can be described as follows: in the towing tank test, if a blind CFD simulation has run, a difference of below 5% can be experienced with a confidence on 80%.