This paper evaluates the applicability of the hexahedral block structured grids for marine propeller performance predictions. Hydrodynamic characteristics for Potsdam Propeller Test Case (PPTC), namely thrust and torque coefficients, were determined using numerical simulations in two commercial solvers: Ansys Fluent and STAR-CCM+. Results were attained for hexahedral and tetrahedral hybrid grids equivalent in terms of cell count and quality, and compared to the experimental results. Furthermore, accuracy of Realizable k- $ϵ$ and SST k- $\omega$ turbulent models when analyzing marine propeller performance was investigated. Finally, performance characteristics were assessed in cavitating flow conditions for a single advance ratio using Zwart–Gerber–Belamri and Schnerr and Sauer models. The resulting cavitation pattern was compared to cavity extents and shape noted during measurements. The results suggest that hexa and hybrid grids, in certain range of advance ratios, do provide similar results; however, for low and high ratios, structured grids in conjunction with Realizable k- $ϵ$ model can achieve more accurate results.

In the present paper, a Random Forest classifier is used to detect leak locations on two different sized water distribution networks with sparse sensor placement. A great number of leak scenarios were simulated with Monte Carlo determined leak parameters (leak location and emitter coefficient). In order to account for demand variations that occur on a daily basis and to obtain a larger dataset, scenarios were simulated with random base demand increments or reductions for each network node. Classifier accuracy was assessed for different sensor layouts and numbers of sensors. Multiple prediction models were constructed for differently sized leakage and demand range variations in order to investigate model accuracy under various conditions. Results indicate that the prediction model provides the greatest accuracy for the largest leaks, with the smallest variation in base demand (62% accuracy for greater- and 82% for smaller-sized networks, for the largest considered leak size and a base demand variation of $\pm 2.5\%$). However, even for small leaks and the greatest base demand variations, the prediction model provided considerable accuracy, especially when localizing the sources of leaks when the true leak node and neighbor nodes were considered (for a smaller-sized network and a base demand of variation $\pm 20\%$ the model accuracy increased from 44% to 89% when top five nodes with greatest probability were considered, and for a greater-sized network with a base demand variation of $\pm 10\%$ the accuracy increased from 36% to 77%).

Conventional cavitation assessment methodology in industrial and scientific applications generally depends on cavitation models utilizing homogeneous mixture assumption. These models have been extensively assessed, modified and expanded to account for deficiencies of their predecessors. Unfortunately, none of the proposed models can be classified as the universal solution for all engineering applications, with usage mainly directed by experience or general availability of the models. In this study we propose a workflow through which the empirical constants governing the phase change of the Kunz mixture cavitation model can be calibrated for a given application or a series of problems, with machine learning as a tool for parameter estimation. The proposed approach was validated on a three-dimensional propeller test case with results in excellent agreement for the case in question. Results for thrust and torque were within 2% with cavity extents differing by up to 20%. This is a significant improvement when compared to previously proposed parameters. Despite the lack of generalization due to the limited nature of the dataset on which the model was trained, the proposed parameters entail acceptable results for similar cases as well. The overall methodology is applicable to other problems as well and should lead to more accurate cavitation predictions.