Quantifying uncertainties in numerical predictions of dynamic cavitation

Cavitation on marine propellers is an important issue due to its negative effects on many aspects of their operation. Therefore, accurate prediction of cavitation is important to ensure better propeller design. Estimating the cavitation behavior numerically is a difficult task due to the high computational cost of simulations and various numerical uncertainties. This study is carried out in order to estimate the parameter and discretization uncertainties and combine them into a single value using an example 2D foil as the test case. Angle of attack and the cavitation number are selected as input parameters and their influence on force coefficients and sheet cavity properties is studied. Sobol indices are also obtained to measure the relative importance of the input and discretization uncertainties. It is seen that uncertainty in the angle of attack has a much greater influence on the force coefficients than the uncertainty in the
cavitation number or grid discretization uncertainty. On the other hand, the cavitation number uncertainty is dominant over the grid and angle of attack uncertainties for the length and volume of the cavity sheet. According to the results obtained, applying only the grid refinement studies is not sufficient for the estimation of the numerical uncertainties for this kind of CFD problems. It is proposed that assessing both parameter and discretization uncertainties for the presented and other similar applications with epistemic uncertainties should be applied more often.