A steady, incompressible, turbulent flow field inside a propeller fan used in an air conditioner has been analyzed numerically using the single-equation Spalart-Allmaras turbulence model. It has been found that the formation of tip vortex starts from the blade tip's suction side at about one third of the axial chord's length aft of the rotor's leading edge. It is due to the rolling-up of the intense shear layer flow between the main axial flow and the suck-in inward flow caused by the large pressure difference between the pressure and the suction sides. The tip vortex passes through the blade passage in a curve reversed towards the direction of the blade's rotation. Its trace is partial to the tangential direction as it goes into the aft part of the blade passage covered by the shroud and, simultaneously, its trace in the radial direction is turned from the outward direction to the inward direction. The operating flow rates have an important effect on the axial position of the tip vortex's trace, while its effect on the radial position is negligible. At low flow rates, the vortex disappears at a location closer to the leading edge. The effect of the shroud's width on the tip vortex's trajectory is notable. For a fan with a wide shroud, the trace of the tip vortex moves upstream with a smaller radial influence region than that of a fan with a narrower shroud.
Issues related to numerical analyses of turbine stage stator/rotor interactions are discussed, with special focus on the selection of grid parameters to secure proper modelling of the stator wake's dissipation. Unsteady calculations of the flow through a high-pressure turbine stage, based on 3D URANS equations, are employed in the grid resolution analysis. Their results are compared with those obtained with other methods, both developed by the author or available in the literature. As a result, the 3D grid resolution of the order of 2000000 cells per one stator and/or one rotor passage has been determined as the necessary minimum for properly modelling the dissipation of the stator's wake on its way through the rotor's cascade, under the circumstances determined by the used CFD code. A generalization of this result is proposed.
The paper is devoted to parallel implementation of a compact discretization scheme combined with the Fourier pseudospectral method. The idle time of processors resulting from the method of computating derivatives using compact schemes is eliminated by proper ordering of subtasks and by performing useful computations when processors are waiting for data from their neighbors. The correctnes of the algorithm is confirmed by comparison of results of LES simulations with DNS data for flow in a 3D channel with periodic non-slip wall boundary conditions.
Mechanisms of formation of the tip leakage over shrouded and unshrouded rotor blades are described in the paper. The loss diagrams for these two types of leakage in a wide range of cascade inlet and outlet flow angles are also plotted. They are obtained in a theoretical way from a model of stream mixing with the help of simplifying assumptions concerning the load of the rotor profile. Results of numerical investigations based on a 3D RANS solver FlowER are also presented in the paper. They extend on the effects of geometrical and flow parameters of the cascade (stage or stage group) on the development of flow losses in the leakage-dominated region as well as on the interaction of tip leakage flow with secondary flows. The tip clearance size, the level of flow turning in the cascade, incidence angle and the effect of relative motion of the blades and endwall are considered here in the case of unshrouded free-tip blades. In the case of shrouded rotor blades the tip leakage mass flow rate and its direction on the re-entry to the blade-to-blade passage. Since the tip leakage non-uniformities are hardly dissipated within the blade row where they originate, the interaction of the tip leakage with the flow in the downstream stator is considered. Some investigations also take into account the effect of relative motion of the stator and rotor blades.
The aim of this work is to estimate the losses in steam flow through an LP steam turbine rotor and the whole stage. Two types of losses occur in steam flow, aerodynamic (profile, secondary flow, leakage) and thermodynamic (due to addition of heat caused by condensation). The presented numerical results are split into two groups. First, a comparison of three different calculation methods of steam flow is carried out. To this end, the geometry of an LP steam turbine's last rotor is chosen. The first examined method is the Streamline Curvature Method (SCM) used on the meridional plane with loss correlations, the other two being commercial and in-house CFD codes, solving the Reynolds-averaged Navier-Stokes equations for a 3D flow. The first two codes model equilibrium steam properties below the saturation line, while the latter models non-equilibrium steam properties. Finally, a comparison is made of the influence on loss prediction of various condensation models for the geometry of the penultimate stage, with the use of an in-house CFD code.
A peniche is designed to offset a half-span aircraft model from the wind tunnel wall boundary layer. This strategy of model mounting results in large influence on the measured aerodynamic coefficients, compared with full-span data. The negative influence is especially important in high-lift conditions leading to incorrect maximum lift behaviour. A very time-consuming set of python scripts was constructed to allow automatic meshing of the wing-body configuration of the DLR F11 high-lift model placed in the European Transonic Wind tunnel (ETW, Germany). Variations due to different concepts of model mountings (peniches) were included. A block-structured FLOWer solver (DLR, Germany) was used for all flow simulations, simplifying the mesh generation process by using the chimera overlapping grids technique. Preliminary results are available for a full-span configuration obtained with a symmetry condition at the mirror plane. Computations of the half-span model placed directly at the wall or mounted using a standard peniche are also presented.
Modelling of the continuum damage framework is developed for application in the elasto-viscoplastic Chaboche constitutive model. A brief description of the basic variant of the Chaboche model equations is given, followed by a discussion of the most important assumptions necessary to obtain evolution of the continuum damage model and its application to the open FE commercial program. A consistent presentation of the two proposed approaches is followed by numerical examples.
1997-2018 (C) CI TASK quarterly@task.gda.pl