Application of Uncertainty Quantification to the study of the exhaust plume impact on ship superstructures

The work carried out during the PhD led to the implementation of a procedure for studying the propagation of uncertainty on the results of fluid dynamics simulations in the naval field on the impact of exhaust fumes on sensitive superstructures. The use of uncertainty analysis (Uncertainty Quantification) can provide planning and operative information to establish which are the most significant input variables and how they affect the final results. It is therefore possible to establish which are the most critical operating conditions. The UQ was applied to fluid dynamics simulations through the creation of an interface procedure between the CFD code (CFX®) and the open source software Dakota. The activity therefore first assessed the contributions of the input variables by carrying out preliminary simulations. The results showed that, according to what has been reported in the literature on the development of plumes subject to transverse currents, the variables that most influence the trajectory and impact of the exhaust fumes are the modulus of wind speed and fumes, which can be summarized as the R parameter and the direction of the main flow. In relation to this last quantity it was found that in the case in question the directions of the wind coming both from the stern and from the bow are significant. Subsequently, a literature test case of a single plume immersed in a prevailing flow was readjusted to set up an Uncertainty Quantification procedure and test different methods for analyzing the propagation of uncertainty. In particular, the following were assessed: the approach to constructing a surrogate model using sampling techniques and the Gaussian Process; the Polynomial Chaos (PCE) method. Finally, the PCE method was applied for the UQ on the available ship geometry by choosing the wind speed (in module and direction) and the external temperature as input variables.