Fluid-structure coupled computations of the NREL 5 MW wind turbine by means of CFD

Highlights

• Fluid-structure coupled CFD simulations of the NREL 5MW reference wind turbine are performed.

• A new high fidelity framework coupling the CFD tool box OpenFOAM and a geometrically non-linear beam solver is used.

• The increase of accuracy due to the fluid-structure coupling comes without a significant increase of the computational costs.

• The comparison of rigid and fluid-structure coupled CFD simulations reveals significant effects of the blade deformations.

• Especially for complex conditions as yawed inflow, large deviations can be observed for the forces in the outer blade part.

Abstract

This paper presents a fluid-structure coupled simulation tool for high-fidelity simulations of wind turbine rotors. Coupling the open source Computational Fluid Dynamics (CFD) code OpenFOAM and the inhouse structural solver BeamFOAM, the developed tool allows the analysis of flexible wind turbines blades by means of CFD without a significant increase in computational costs. To demonstrate the capabilities of the coupled solver, the aero-elastic response of the NREL 5 MW reference wind turbine is computed for various conditions and specific results are compared to findings of other authors. The solver framework is then used to investigate the effect of blade deformations on aerodynamic key parameters such as power, thrust and sectional forces. It is shown, that the structural deformations have a clear influence on the aerodynamic rotor performance. Especially for the case of yawed inflow, significant implications can be observed in terms of loads and local induction factors. Compared to the fluid-structure coupled framework, the rigid CFD solver underpredicts the forces acting on the blades for most of the cases. Consequently, the presented results are expected to contribute to improve the correction models used in aerodynamic models of lower fidelity like those based on the Blade Element Momentum theory.

Introduction

Over the last decades wind power emerged as an important source of renewable electricity. In 2020, more than 800 GW of wind power capacity are expected to be installed worldwide, which accounts to the second largest share of the renewable electricity production after hydro power [1,2]. However, after years of solid growth and economic success, supporting subsidies are globally decreasing and wind energy has increasingly to compete directly with fossil fuel powered power plants. In order to decrease the costs of energy, the wind energy community responds with continuously increasing rotor sizes. Following that, several wind turbines of the 8 MW class were commercially available on the market in the end of 2016 [1]. Reaching blade lengths of 80 m or more, these machines rely on advanced, lightweight blade designs, which are getting increasingly flexible and therefore more prone to aero-elastic influences [3]. To make the next generation of wind turbines a success, it will be more important than ever to use accurate numerical models to better predict the aerodynamic loads and the structural response of the wind turbine blades.

At the same time, the industrial wind turbine design still relies on the computational efficient Blade Element Momentum (BEM) theory for aeroelastic load calculations. Lacking any three-dimensional and unsteady effects and being based on tabulated 2D airfoil data, numerous correction models are employed in order to improve the shortcomings of the underlying simple BEM theory. However, the ever increasing rotor size gives rise to doubts on the general validity of correction models, which were mostly derived based on results obtained from comparatively small experimental wind turbines. Since aerodynamic models of lower fidelity like the BEM theory are expected to remain the method of choice for industrial load calculations in the foreseeable future [4], numerical models of higher fidelity like Computational Fluid Dynamics (CFD) are increasingly used to investigate aerodynamic phenomena relevant for large rotors. Besides large Reynolds number effects and uncertainties due to the use of thick airfoils, also the effect of large blade deformations and their implication on the aerodynamic and aero-elastic rotor performance is not yet completely understood [5]. In order to investigate the non-linear interaction of large, flexible wind turbine blades with the incoming wind on a high fidelity level, complex fluid-structure coupled simulation frameworks are required. These simulation frameworks, also denoted as Fluid-Structure Interaction (FSI) methods, can be used to investigate complex aero-elastic phenomena, which clearly fall outside the validated range of BEM based tools.

For that reason, recently an increasing number of research groups within the wind energy community started to develop and use fluid-structure coupled high fidelity analysis frameworks. In the following, the most important contributions addressing CFD based FSI for wind energy are shortly introduced. One of the first attempts to perform a 3D FSI simulation of a wind turbine at full scale was made by Hsu and Bazilevs in 2012 [6]. They used a finite-element-based flow solver, coupled to a NURBS-based isogeometric structural analysis (IGA) tool to perform time-accurate simulations of the NREL 5 MW wind turbine. In 2013, Corson et al. used the commercial package AcuSolve, coupled to a linear modally reduced finite element model, to investigate the aero-elastic response of the Sandia 100 m reference blade design [7]. One year later, another attempt was made by Yu and Kwon, conducting time-accurate aero-elastic simulations of the complete NREL 5 MW wind turbine under yawed and sheared inflow conditions [8]. They used an in-house unstructured CFD code coupled to a non-linear Euler-Bernoulli beam model, performing the coupling once per full rotor revolution. In the same year, Carrión et al. presented fluid-structure coupled steady-state computations of the MexNext experimental wind turbine using the compressible Helicopter Multi-Block (HMB), developed at Liverpool [9]. Next, Imiela et al. performed steady-state FSI simulations on the NREL 5 MW turbine by coupling the compressible CFD code TAU and the multibody simulation software SIMPACK [10]. Furthermore, Horcas et al. presented quasi-steady fluid-structure coupled simulation results obtained by using the commercial code FINE Turbo. The structural response was predicted by a linear structural solver which was included within the flow solver [11]. In 2016, Heinz et al. published their work on coupling the incompressible CFD code EllipSyS3D to the wind turbine simulation code HAWC2 [12]. The fluid-structure coupled framework HAWC2CFD was used to investigate the NREL 5 MW reference wind turbine under several inflow and operating conditions. Moreover they performed a study on aeroelastic instabilities caused by vortex-induced blade vibrations on the DTU 10 MW reference wind turbine [13]. In addition, Sayed et al. presented a coupling of the compressible CFD solver FLOWer and the structural solver Carat++, which was utilized to investigate the performance of the DTU 10 MW reference turbine [14].

However, most of the mentioned works investigated the effect of elastic blade deformations for frontal inflow conditions and mainly focused on a general quantification of the effects. According to the authors knowledge, an in-depth investigation of the occurring effects involving an analysis of the local induction factors, especially for yawed inflow, has not been discussed so far. The evaluation of the latter can lead to new insights, especially interesting for complex load cases like yawed inflow. Since the skewed wake correction models, used for yawed inflow in BEM methods, are based on the induction factors at each blade section [15,16], fluid-structure coupled simulations have the potential to support current efforts based on CFD-only simulations to improve BEM correction models like the work of Rahimi et al. [17]. In this work, a high-fidelity framework for flexible, rotating wind turbine blades is presented and used for fluid-structure coupled computations on the NREL 5 MW reference wind turbine [18]. Aiming at giving new insight into the effect of blade deformations on the aerodynamic performance of wind turbines, several simulations are performed for different rotor configurations. In Section 2, the numerical methods developed and used in the frame of this work are described. Section 3 introduces the simulated wind turbine geometry and provides information about the numerical setup used for the performed simulation. The obtained results and findings are then presented and discussed in Section 4. The conclusions are summarized in Section 5.