Rickard Bensow

Implicit LES Predictions of the Cavitating Flow on a Propeller

We describe an approach to simulate dynamic cavitation behavior based on large eddy simulation of the governing flow, using an implicit approach for the subgrid terms together with a wall model and a single fluid, two-phase mixture description of the cavitation combined with a finite rate mass transfer model. The pressure-velocity coupling is handled using a PISO algorithm with a modified pressure equation for improved stability when the mass transfer terms are active. The computational model is first applied to a propeller flow in homogeneous inflow in both wetted and cavitating conditions and then tested in an artificial wake condition yielding a dynamic cavitation behavior. Although the predicted cavity extent shows discrepancy with the experimental data, the most important cavitation mechanisms are present in the simulation, including internal jets and leading edge desinence. Based on the ability of the model to predict these mechanisms, we believe that numerical assessment of the risk of cavitation nuisance, such as erosion or noise, is tangible in the near future.

On the justification and extension of mixed models in LES

A reformulation of the large eddy simulation (LES) equations is presented, based on an alternative decomposition of the subgrid stress tensor leading to modified Leonard, cross, and Reynolds terms, which are all individually frame indifferent. The new Leonard tensor, identical to the scale similarity model proposed by Bardina et al(J. Bardina, J. H. Ferziger and W. C. Reynolds, 1980, AIAA Paper 80-1357), is computable and thus becomes an integral part of the LES equations, so this formulation clearly emphasizes the difference between Reynolds averaged Navier–Stokes (RANS) and LES. The remaining modified cross and Reynolds terms can be regrouped further into a reconstructable part and a true subgrid component, where we here use an approximate deconvolution model and a subgrid viscosity model for the respective parts. This structure justifies the use of a dissipative model term in mixed models based on, e.g., the scale similarity model. The reformulated LES model is tested on two basic flows, the Taylor Gre...

Numerical investigation of the flow over an axisymmetric hill using LES, DES, and RANS

The flow around an axisymmetric hill, mounted in a channel with a fully developed approach flow, is investigated. The flow contains complex structures such as a turbulent boundary layer with several unsteady separations and reattachments. It is highly three-dimensional due to both streamwise and spanwise pressure gradients on the leeside of the hill. The shallowness of the separation region makes the flow a very demanding test case for any computational fluid dynamics model. Three different strategies are used in this study: Reynolds-averaged Navier–Stokes (RANS), large eddy simulation (LES), and detached eddy simulation (DES). The computed flow, in terms of velocity and pressure profiles, compared with measurement data and the results show that LES and DES are indeed capable of handling this complicated flow in a correct way whereas RANS clearly fails to predict several important flow features. Furthermore, the influence of the size of the computational domain, the grid resolution and the inflow boundary conditions is also studied. It is found that the pressure field is sensitive to the location of the inlet and the DES model is very sensitive to the inlet boundary condition on the eddy viscosity. To significantly improve the predictions, it is believed that the near-wall resolution must be increased substantially, in particular in the spanwise direction, or a better wall handling has to be incorporated.

LES of unsteady cavitation on the delft twisted foil

In this paper, the cavitating flow around the Delft twisted hydrofoil with unsteady inflow condition is numerically simulated using Large Eddy Simulation in combination with a volume of fluid implementation to capture the liquid-vapor interface and Kunz’s model for the mass transfer between the phases. Main cavitation mechanisms, including periodic shedding of main and secondary cavities, side- and re-entrant jets, as well as the cavity extent and the lock-in effect between the inflow variation and the cavity are compared with experimental observations.

Implicit and Explicit Subgrid Modeling in LES Applied to a Marine Propeller

The flow around a four-bladed marine propeller in homogeneous inflow and in non-cavitating conditions is investigated using Large Eddy Simulation, LES. Explicit, using a k-equation eddy viscosity model, and implicit subgrid modeling are compared for both the standard LES formulation as well as a mixed formulation containing the, so called, scale similarity term. A wall-modeled approach is used on a relatively coarse grid, containing 5.5 million cells, for the full propeller in order to mimic a future applied computation including the ship hull. The implicit modeling is of particular interest in cavitation simulation, where the interaction between an explicit subgrid model and the liquid-vapor interface may cause numerical and modeling problems. All simulations yield fairly similar results, although the implicit LES gives better prediction of the global performance of the propeller. The agreement with experimental data is good close to the propeller, but the simulated flow structures diffuses quickly at the present grid resolution.

3D Unsteady Computations for Submarine-Like Bodies

Results from a computational study using Unsteady Reynolds Averaged Navier Stokes (URANS) models and Large Eddy Simulation (LES) of flows past submarine-like bodies are here presented. The aims are to evaluate URANS and LES for high-Re number hydrodynamic flows, to investigate the influence of the turbulence and subgrid turbulence modeling, and to discuss some features of submarine hydrodynamics. For this purpose we have chosen to examine the flow past a prolate spheroid at 10° and 20° angle of attack at a body length Re number of 4-106, and the flow past the DARPA-2 Suboff bare hull and fully appended hull configurations at a body length Re number of 12-106. For both cases experimental data is available for comparison. One finite element and one finite volume flow solver has been used - both with the capability of employing a range of turbulence models and with the capacity of using unstructured and hybrid grids. Better agreement between predictions and experimental data is obtained with LES than with the URANS models, but at a considerably higher price, due to the finer grids and finer temporal resolution in LES.

Numerical Simulation of an Oscillating Cylinder Using Large Eddy Simulation and Implicit Large Eddy Simulation

In this work, we use large eddy simulation (LES) to study the influence of grid and subgrid model on the lift and drag force predictions of a fixed cylinder undergoing streamwise sinusoidal oscillations in a steady flow, resulting in a varying Reynolds number, Re, within the range 405 <= Re <= 2482. This benchmark case is a first step toward studying engineering applications related to flow-induced vibrations. We examine the influence of both grid resolution and the subgrid model using implicit and explicit LES. The methodology used, LES based on a finite-volume method capable of handling moving meshes, are found to provide force predictions that agree well with experimentally measured data, with respect both to the overall flow development and force magnitude.

DISCONTINUOUS/CONTINUOUS LEAST-SQUARES FINITE ELEMENT METHODS FOR ELLIPTIC PROBLEMS

Least-squares finite element methods (LSFEM) are useful for first-order systems, where they avoid the stability consideration of mixed methods and problems with constraints, like the div-curl problem. However, LSFEM typically suffer from requirements on the solution to be very regular. This rules out, e.g., applications posed on nonconvex domains. In this paper we study a least-squares formulation where the discrete space is enriched by discontinuous elements in the vicinity of singularities. The weighting on the interelement terms are chosen to give correct regularity of the solution space and thus making computation of less regular problems possible. We apply this technique to the first-order Poisson problem, show coercivity and a priori estimates, and present numerical results in 3D.

Development and application of optimisation algorithms for propeller design

Propeller design is a comprehensive task in finding the best trade-off between competing objectives and constraints. It requires a multi-disciplinary evaluation of the propeller performance based on various input parameters. Thus optimisation algorithm applied to this type of problem require consideration of the problem to solve. The purpose of this paper is hence the improvement of commonly used population-based algorithms (NSGA-II and PSO) towards the application of marine propeller design. The extension to three algorithms are outlined utilising meta models, adapted constraints and modified constraints handling. The proposed algorithms are applied on a real-life marine propeller example. Results of 13 optimisations are compared in terms of optimisation convergence, constraints compliance and Pareto optimality and show advantageous performance of the developed cavitation constraints and the meta-model extended NSGA-II.

Large-Eddy Simulation of an Oscillating Cylinder in a Steady Flow

In this work, large-eddy simulation is used to study the flow around a circular cylinder undergoing streamwise sinusoidal oscillations. This benchmark case is a first step toward studying engineering applications related to flow-induced vibrations. Both the flow physics, which correlate the flow development with the time varying loading of the cylinder at two different oscillation frequencies, as well as a validation of the fluid structure interaction methodology through comparison with experimental data for the same configuration are described. With the methodology used, large-eddy simulation based on a finite volume method capable of handling moving meshes gives force predictions that generally agree well with experimentally measured data, both with respect to the overall flow development as with force magnitude.