Jess Michelsen

An improved SIMPLEC method on collocated grids for steady and unsteady flow computations

A modified SIMPLEC scheme for flow computations on collocated grids has been developed. It is demonstrated that the standard SIMPLEC scheme [1] is inconsistent when applied on collocated grids. Hence, for steady computations the computed solution depends on the velocity underrelaxation parameter f u , whereas the solutions of unsteady computations for small time steps are polluted by unphysical wiggles. A revised scheme is proposed that extends the capability of the SIMPLEC method to cope with collocated grids in a general and consistent way. The efficiency of the new scheme is demonstrated by computing flows past a circular cylinder and an airfoil.

Drag Prediction for Blades at High Angle of Attack Using CFD

In the present paper it is first demonstrated that state of the art 3D CFD codes are capable of predicting the correct dependency of the integrated drag of a flat plate placed perpendicular to the flow. This is in strong contrast to previous 2D investigations of infinite plates, where computations are known to severely overpredict drag. We then demonstrate that the computed drag distribution along the plate span deviate from the general expectation of 2D behavior at the central part of the plate, an important finding in connection with the theoretical estimation of drag behavior on wind turbine blades. The computations additionally indicate that a tip effect is present that produces increased drag near the end of the plate, which is opposite of the assumptions generally used in drag estimation for blades. Following this several wind turbine blades are analysed, ranging from older blades of approximately 10 meter length (LM 8.2) over more recent blades (LM 19.1) around 20 meters to two modern blades suited for megawatt size turbines. Due to the geometrical difference between the four blades, the simple dependency on aspect ratio observed for the plates are not recovered in this analysis. The turbine blades behave qualitatively very similar to the flat plates and the spanwise drag distributions show similar tip effects. For the turbine blades this effect is even more pronounced, because the tapering of the blades makes the tip effect spread to a larger part of the blades. The findings are supported by visualizations of the wake patterns behind the blades.

Detached-Eddy Simulation of Flow Around the NREL Phase-VI Blade

The Detached-Eddy Simulation model implemented in the computational fluid dynamics code, EllipSys3D, is applied on the flow around the NREL Phase-VI wind turbine blade. Results are presented for flow around a parked blade at fixed angle of attack and a blade pitching along the blade axis. Computed blade characteristics are compared with experimental data from the NREL/NASA Ames Phase-VI unsteady experiment. The Detached-Eddy Simulation model is a method for predicting turbulence in computational fluid dynamics computations, which combines a Reynolds Averaged Navier-Stokes method in the boundary layer with a Large Eddy Simulation in the free shear flow. The present study focuses on static and dynamic stall regions highly relevant for stall regulated wind turbines. Computations do predict force coefficients and pressure distributions fairly good and results using Detached-Eddy Simulation show considerably more three-dimensional flow structures compared to conventional two-equation Reynolds Averaged Navier-Stokes turbulence models, but no particular improvements are seen on the global blade characteristics.Copyright

AERO-ACOUSTIC COMPUTATIONS OF WIND TURBINES

A numerical algorithm for acoustic noise generation is extended to 3D flows. The approach involves two parts comprising a viscous incompressible flow part and an inviscid acoustic part. In order to simulate noise generated from a wind turbine, the incompressible and acoustic equations are written in polar coordinates. The developed algorithm is combined with a so-called actuator-line technique in which the loading is distributed along lines representing the blade forces. Computations are carried out for the 500kW Nordtank wind turbine equipped with three LM19 blades.Copyright

Aerodynamic Drag Prediction of Helicopter Fuselage

is investigated for one test case using different computational grids. The results are compared with wind-tunnel measurements and show good agreement with the reference data when the ine uence of the wind-tunnel strut is taken into account. Thecomputational e uid dynamics model developed consists ofa parallel multiblock, multigrid Reynolds-averaged Navier ‐Stokes codeand a three-dimensionalhyperbolic grid generator. Block-structured grids are generated from a CAD base. From a numerical point of view, convergence was achieved quickly in all of the cases computed, thus showing efe ciency and low CPU cost.

Numerical study of swirling flow in a cylinder with rotating top and bottom

A numerical investigation of oscillatory instability is presented for axisymmetric swirling flow in a closed cylinder with rotating top and bottom. The critical Reynolds number and frequency of the oscillations are evaluated as function of the ratio of angular velocities of the bottom and the top (ξ=Ωbottom∕Ωtop). Earlier linear stability analysis (LSA) using the Galerkin spectral method by Gelfgat et al. [Phys. Fluids, 8, 2614 (1996)] revealed that the curve of the critical Reynolds number behaves like an “S” around ξ=0.54 in the co-rotation branch and around ξ=−0.63 in the counter-rotation branch. Additional finite volume computations, however, did not show a clear S behavior. In order to check the existence of the S shape, computations are performed using an axisymmetric finite volume Navier-Stokes code at aspect ratios (λ=H∕R) 1.5 and 2.0. Comparisons with LSA at λ=1.5 show that the S shape does exist. The S shape of the stability diagram predicted by LSA is thus confirmed by a finite-volume based Nav...

Navier-Stokes predictions of the NREL phase VI rotor in the NASA Ames 80-by-120 wind tunnel

The application of an incompressible Reynolds Averaged Navier-Stokes solver to cases from the NREL/NASA Ames wind tunnel test is described. Six cases of the NREL PHASE-VI rotor in the upwind configuration under zero yaw and zero degrees tip pitch are computed. Favorable comparison of the computed results with measurements in the form of shaft torque, root moments, spanwise force distributions, and pressure distributions are shown. The good agreement documents the feasibility of 3D CFD computations in connection with prediction of the performance of new rotors. Additionally it is shown how CFD computations can be used to determine the three dimensional effects in rotor flows.© 2002 ASME

Detached-Eddy simulation of decaying homogeneous isotropic turbulence

The current paper describes how a formulation of Detached-Eddy Simulation is used to simulate decaying homogeneous isotropic turbulence. A formulation of DES based on the standard k-H RANS model is used and calculation is done to calibrate the LES regions of DES. The paper describes how initial conditions are created together with other numerical aspects such as how different differentiation schemes behave. Simulation results are compared to the experimental data of Comte-Bellot and Corrsin. The paper shows a way of calibrating the model to a specific numerical scheme and result show that the simple model used in the DES formulation is capable of modelling the dissipation of turbulent energy.

Unsteady fluid dynamics in a semi-closed cavity

Results are presented for unsteady, time-true simulations of a shear driven cavity started impulsively from rest. The cavity is partially occluded by slits on its upper side, thereby forming a Helmholtz resonator type of geometry. Simulations with Reynolds-averaged Navier-Stokes solvers at the laminar Reynolds number Re = 3,000, for a spanwise aspect-ratio equal to 3, showed that the flow is unsteady and a-periodic over a normalised time of one hundred units. Taylor-Gortler spanwise vortices appear early in the simulation (flow time 3 or 4 units); these vortices have a relatively short time span, and completely disappear at a flow time equal to 90. This study also reports about the findings on the secondary eddies (upstream and wall vortices), on the vortex filaments that are drawn from the cavity floor, and on the mass transport properties of the flow.Copyright