As a contribution to a better understanding of cavitation erosion mechanisms, a compressible inviscid finite volume flow solver with barotropic homogeneous liquid-vapor mixture cavitation model is applied to ultrasonic horn set-ups with and without stationary specimen, that exhibit attached cavitation at the horn tip. Void collapses and shock waves, which are closely related to cavitation erosion, are resolved. The computational results are compared to hydrophone, shadowgraphy and erosion test data. At the horn tip, vapor volume and topology, subharmonic oscillation frequency as well as the amplitude of propagating pressure waves are in good agreement with experimental data. For the evaluation of flow aggressiveness and the assessment of erosion sensitive wall zones, statistical analyses of wall loads and of the multiplicity of distinct collapses in wall-adjacent flow regions are applied to the horn tip and the stationary specimen. An a posteriori projection of load collectives, i.e. cumulative collapse rate vs. collapse pressure, onto a reference grid eliminates the grid dependency effectively for attached cavitation at the horn tip, whereas a significant grid dependency remains at the stationary specimen. The load collectives show an exponential decrease towards higher collapse pressures. Erosion sensitive wall zones are well predicted for both, horn tip and stationary specimen, and load profiles are in good qualitative agreement with measured topography profiles of eroded duplex stainless steel samples after long-term runs. For the considered amplitude and gap width according to ASTM G32-10 standard, the analysis of load collectives reveals that the distinctive erosive ring shape at the horn tip can be attributed to frequent breakdown and re-development of a small portion of the tip-attached cavity. This partial breakdown of the attached cavity repeats at each driving cycle and is associated with relatively moderate collapse peak pressures, whereas the stationary specimen is rather unfrequently stressed at the end of each subharmonic oscillation cycle by the violent collapse of the complete cavity.

Three-dimensional (3D) simulations with ansys cfx 16.1 as well as measurements of the cavitating flow in a low specific speed centrifugal pump (n_q = 12 min⁻¹) are performed for different operation conditions and varying surface roughness. Surface roughness is considered by wall functions in the flow simulations. Good agreement between measured and calculated head is achieved for noncavitating flow. Net positive suction head (NPSH_3%) rises toward overload due to incidence, flow separation, and vapor zones at the volute tongue. The NPSH_3% rise is slightly higher for rough walls according to measurements and significantly overestimated by the wall function approach, irrespective of the roughness level in the simulation. A low-Reynolds number approach at the volute tongue leads to a more accurate prediction of NPSH_3% than wall functions, at the cost of high computational effort.

A local loss analysis (LLA) based on entropy production is presented for the numerical three-dimensional (3D) simulation of isothermal centrifugal pump flow. A finite volume method and a statistical turbulence model are employed. Wall functions for direct and turbulent entropy production in isothermal flow are derived, implemented in a node-centered finite volume scheme as a postprocessing procedure, and validated on an attached channel flow as well as on separated flow in an asymmetric diffuser. The integrity of the entropy wall function is demonstrated by a loss balance for a wide range of boundary layer resolution in terms of nondimensional wall distance y+≈1 to ≈200. Remaining differences to the total pressure loss are traced back to the particular turbulent wall function for the flow solution within the finite volume solver and vanish toward a wall resolution of the viscous sublayer, i.e., y+≈1. LLA together with the new entropy wall function is applied to highly unsteady isothermal flow in a single-blade pump as well as to part-load operation of a conventional multiblade pump which reveals distinctive flow structures that are associated with entropy production. By these examples, it is demonstrated how efficiency characteristics of centrifugal pumps can be attributed to local loss production in particular flow regions.

Three-dimensional (3D) unsteady Reynolds-averaged Navier–Stokes (URANS) flow simulations are conducted to investigate the highly unsteady flow field at part load operation of a centrifugal pump. By the availability of unsteady flow field measurement data in the impeller wake region, a thorough validation of the simulation method is performed. Grid independence of the results is ensured. Unsteady characteristics in terms of head and shaft power as well as transient blade loads are evaluated to assess the unsteady pump performance. Significant mis-loading of the blading is revealed when one blade passes the volute tongue and associated with the strong unsteady and 3D flow field in the impeller-volute tongue region. Negative radial velocity in the tongue region is the origin of a vortex at the blade pressure side and a subsequent pressure drop that leads to even temporally negative blade loading. The results provide a detailed insight in the complex part load flow field that might be utilized for an improved pump design. As a valuable secondary outcome, a comparison of results obtained by two widely used computational fluid dynamics (CFD) codes for pump flow simulation is provided, i.e., the commercial code ansyscfx and the branch foam-extend of the open source software openfoam. It is found that the results of both methods in terms of unsteady characteristics as well as local ensemble-averaged velocity field are consistent.

Cavitation leads to rapid degassing of fluids. Up to date, there is a lack of model approaches of cavitation-induced degassing. The aim of the present study is to gain a more thorough knowledge of the process. Therefore, the relation between cavitation intensity and air release is investigated experimentally for an orifice flow as function of cavitation number. For this, shadowgraphy imaging is used to visualize regions of steam and air volume downstream of the orifice. Analysis of the images shows a strongly nonlinear scaling behavior for both cavitation intensity and air release as a function of cavitation number. Three distinct regimes could be identified for cavitation-induced gas release. While an exponential scaling was found at high cavitation intensities, degassing rates appear to be nearly constant in the intermediate cavitation number range. Empirical scaling laws are given here that may serve as first model approach for the prediction of cavitation induced air release behind flow constrictions.