Harmonic balance Navier‐Stokes aerodynamic analysis of horizontal axis wind turbines in yawed wind

Abstract

Multimegawatt horizontal axis wind turbines often operate in yawed wind transients, in which the resulting periodic loads acting on blades, drive-train, tower, and foundation adversely impact on fatigue life. Accurately predicting yawed wind turbine aerodynamics and resulting structural loads can be challenging and would require the use of computationally expensive high-fidelity unsteady Navier-Stokes computational fluid dynamics. The high computational cost of this approach can be significantly reduced by using a frequency-domain framework. The paper summarizes the main features of the COSA harmonic balance Navier-Stokes solver for the analysis of open rotor periodic flows, presents initial validation results on the basis of the analysis of the NREL Phase VI experiment, and it also provides a sample application to the analysis of a multimegawatt turbine in yawed wind. The reported analyses indicate that the harmonic balance solver determines the considered periodic flows from 30 to 50 times faster than the conventional time-domain approach with negligible accuracy penalty to the latter. KEYWORDS harmonic balance Navier-Stokes equations, horizontal axis wind turbine aerodynamics, NREL Phase VI wind turbine, NREL 5 MW wind turbine, yawed wind aerodynamics 1 INTRODUCTION Wind energy is a key low-carbon energy source, playing a crucial role in lowering global greenhouse gas emissions, and it is now viewed as one of the most cost-effective climate change mitigation technologies. Current utility-scale horizontal axis wind turbines (HAWTs) already feature fairly high aerodynamic efficiencies; however, because of the spatial and temporal variability of the environmental conditions, HAWTs often experience unsteady flow conditions, which induce fatigue and lower the energy harvest. Several of such regimes can be viewed as periodic, and typical examples include the blades rotating (a) through stratifications of the atmosphere associated with the atmospheric boundary layer, with the wind velocity varying by as much as 6 m/s over a 100-m rotor diameter, 1 (b) through the variable pressure field due to the downwind tower, and (c) in yawed wind, occurring when the freestream wind velocity is no longer orthogonal to the turbine rotor. 2 In all these cases, the fundamental frequency of the periodic excitation is a multiple of the rotor speed.Regarding yawed wind, utility-scale HAWTs typically feature yaw control systems that monitor the wind direction and turn the entire nacelle to realign wind and rotor normal. 2 Yaw actuators, however, adjust the nacelle position after a relatively long time interval from the yaw misalignment detection, and thus, fatigue due to yaw misalignment can be significant. The yawed wind condition also reduces the power produced by the turbine, and the reduction increases nonlinearly with the cosine of the yaw misalignment. Above certain wind speed-and turbine-dependent yaw misalignment thresholds, dynamic stall also occurs, and this aggravates further unsteady loads because of hysteretic force and mom...

Load balance and parallel I/O: Optimising COSA for large simulations

“…This paper outlines work undertaken [5] to take a parallelised HB CFD application, COSA [6], and enable it to scale efficiently to large numbers of cores. COSA is a finite volume NS code featuring a steady solver, a TD solver for the solution of general unsteady flows [7] [8], and a HB solver for the rapid solution of highly nonlinear unsteady periodic flows [9] [10] [11]. The general parallelisation of COSA is efficient and scales well, but the load balance and data input/output (I/O) functionality did not match the rest of the application and stood in the way of large scale research deployment.…”

“…the multi-block grid) for the simulation. In the case of complex geometries such as wind turbine rotors [11], state-of-the-art multi-block grid generators lack the flexibility needed to easily generate multi-block grids whose blocks have the same or at least comparable number of cells, and instead yield grids whose blocks differ significantly in size. Thus, additional lengthy grid pre-processing operations, often requiring the use of several different grid pre-processors, are needed to obtain a grid with equal block sizes.…”

“…The former benchmark case, similar to the analyses in [11], refers to the simulation of the unsteady periodic flow field past a wind turbine rotor in yawed wind retaining four complex harmonics in the truncated Fourier series used to approximate the sought periodic flow field. The latter benchmark case refers to the analyses of the unsteady periodic flow field past an oscillating wing (concurrent plunging and pitching motion) for renewable power generation, it uses four complex harmonics, and it is similar to the TD analyses reported in [8].…”

“…The smaller simulation (800 blocks) has a restart file of approximately 3 GB with nine flowtec files of around half a GB each.In the case of the restart file, the Nh solution snapshots q at the centres of the cells of each grid block is written by the serial code with the instructions inFigure 5, which are executed within a loop over all grid blocks. The parameter npde denotes the number of partial differential equations (PDEs), and equals 7, corresponding to one PDE for the conservation of mass, three PDEs for the momentum conservation in all three directions, one PDE for the conservation of energy and two PDEs for the shear stress transport turbulence model used by COSA [[8][11].…”

Compressible Navier-Stokes Analysis of Floating Wind Turbine Rotor Aerodynamics

“…These codes are extremely fast and, thus, ideally suited to industrial design, but their predictions may be affected by significant uncertainty when dealing with complex three-dimensional (3D) rotor flows. For example, the uncertainty affecting the prediction of low-fidelity codes for yawed flows is discussed in [3], and studies providing quantitative comparisons of BEMT and high-fidelity Navier-Stokes (NS) Computational Fluid Dynamics (CFD) predictions are cited therein. The use of BEMT-based engineering codes for FOWT analysis entails additional uncertainty, due to the, generally non-uniform, time-dependent entrainment velocity associated with the rigid-body motion of the whole turbine, which introduces further complexity in rotor unsteady aerodynamics.…”