Elsevier

Renewable Energy

Volume 97, November 2016, Pages 230-242
Renewable Energy

CFD simulation of a floating offshore wind turbine system using a variable-speed generator-torque controller

https://doi.org/10.1016/j.renene.2016.05.061Get rights and content

Highlights

  • CFD simulation of OC3-Hywind with inertial rotor and variable-speed control.

  • DDES used, which predicts separation on blades during upstream motions.

  • Less overall platform motions predicted compared to NREL OC3 Phase IV results.

  • Less generator power and torque predicted compared to NREL OC3 Phase IV results.

Abstract

Prediction and control of rotor rotational velocity is critical for accurate aerodynamic loading and generator power predictions. A variable-speed generator-torque controller is combined with the two-phase CFD solver CFDShip-Iowa V4.5. The developed code is utilized in simulations of the 5 MW floating offshore wind turbine (FOWT) conceptualized by the National Renewable Energy Laboratory (NREL) for the Offshore Code Comparison Collaboration (OC3). Fixed platform simulations are first performed to determine baseline rotor velocity and developed torque. A prescribed platform motion simulation is completed to identify effects of platform motion on rotor torque. The OC3’s load case 5.1, with regular wave and steady wind excitation, is performed and results are compared to NREL’s OC3 results. The developed code is shown to functionally control generator speed and torque but requires controller calibration for maximum power extraction. Generator speed variance is observed to be a function of unsteady stream-wise platform motions. The increased mooring forces of the present model are shown to keep the turbine in a more favorable variable-speed control region. Lower overall platform velocity magnitudes and less rotor torque are predicted corresponding to lower rotor rotational velocities and a reduction in generated power. Potential improvements and modifications to the present method are considered.

Introduction

Offshore wind capacity in the United States is becoming a reality. Initial manufacturing has begun, in mid-2015, on the Block Island Wind Farm [1], expected to be in service in late 2016. The US Department of Energy (DOE) has drafted a plan in which the US will utilize 86 GW of offshore wind power by 2050 [2], an aggressive goal considering the US uses no offshore wind power as of year-end 2014. The US has this amount, and much more, available to it [3]. However the majority of offshore capacity available to the US comes from deeper waters where FOWT are required [4]. Siting concerns regarding noise, visual aesthetics, shipping lanes and ecology have made waters farther from shore attractive, and FOWT technology is being aggressively pursued.

Designing for FOWT introduces a level of complexity not seen in onshore designs due to platform motions that, in turn, produce unsteady aerodynamic loading at the blades. Predictions of power require accurate predictions of loading, both aerodynamic and hydrodynamic. Most FOWT simulations to date have used the blade-element momentum theory (BEM), explained in detail in Ref. [5], to determine aerodynamic loading on the rotor and Morison’s equation [6] to determine hydrodynamic loading on the platform. The certified wind turbine simulator code FAST from NREL [7], widely used in both the industry and research communities and compared to herein, uses BEM and Morison’s equation. BEM is a 2-dimensional quasi-steady method utilizing empirically determined lift and drag coefficients and other correction models, such as dynamic stall and wake models. BEM was developed for analysis of flow perpendicular to the rotor plane. As such BEM, and its current corrections, may not be appropriate for general purpose offshore simulations given the varying yawed inflow conditions, dynamic stall, and potential for rotor-wake interaction [8], [9]. BEM, as designed, also does not consider the tower geometry and requires a correction model to account for the presence of the tower in wake deficit and blade-tower aerodynamic disruption [10], [11]. Morison’s equation is a 1-dimensional, semi-empirical function developed to determine hydrodynamic loading, requiring experimentally derived added mass and drag coefficients for any given geometry. Morison’s equation assumes the diameter of the structure is small relative to incident wavelength such that wave diffraction effects can be neglected. This is not appropriate for many FOWT platforms, notably buoyancy stabilized platforms such as barges [12]. The use of computational fluid dynamics (CFD) can help to overcome the limitations of BEM and Morison’s equation.

With CFD the governing Navier-Stokes equations are discretized spatially and temporally into algebraic equations and solved. CFD can intrinsically solve in 3-dimensions, requiring empirical corrections only to determine turbulent characteristics, and can provide details of flow physics that BEM and Morison’s equation cannot. With computational resources becoming more readily available, especially parallel-computing resources, finer resolution of both time and space discretization can be accomplished, allowing CFD to scale in a way that correction models may not be able to. 3-dimensional aerodynamic CFD simulations of a wind turbine require a relative rotational motion between a blade or rotor and the surrounding fluid. This presents a challenge to the usage of CFD as many solvers require static grids and cannot model dynamic geometric situations, such as an accelerating rotor. Techniques such as overset or “chimera” meshing [13] and sliding-mesh [14] have been employed for the purposes of platform motion and rotor rotation relative to the tower. The most notable application of CFD to date are simulations based on NREL’s onshore unsteady phase VI experiments [15], [16], [17]. In these experiments the rotational velocity of the turbine was prescribed, making the dataset excellent for code validation. The rotational velocity of the rotor and developed aerodynamic torque cannot be de-coupled, however, especially considering underlying platform motions. The component of velocity provided by rotor rotation to the blade is usually the dominating component of overall magnitude, particularly on the outboard span of long blades like those used on FOWT. To properly predict generated power, stemming from aerodynamic power developed by the rotor, requires an inertial model of the drivetrain to predict rotor acceleration. Most current CFD simulations of FOWT have used prescribed rotor rotational velocities, with or without platform motions. Prescribed rotor rotation velocity and platform pitch oscillations were used with overset grids in Ref. [8] and with sliding-mesh in Ref. [18]. Both of these studies also produced predictions using BEM and compared, showing differences between the two methods. The rotor velocity of a FOWT was predicted using an inertial drivetrain model, along with a variable-speed generator (VS) control software scheme, and overset CFD in Ref. [19] with a fixed platform and single-phase computation. Rigid-body 6 degrees of freedom (6-DOF) platform motions and mooring forces were predicted using overset CFD in Ref. [20], where rotor power was investigated but improperly compared to generator power. Prediction of rotor rotational velocity is crucial in calculating proper aerodynamic loading and thrust, especially important for FOWT where the platform is free to pitch and requires careful controller calibration. The present study extends upon [20], using the crowfoot mooring system developed within and predicting, rather than prescribing, rotor rotational velocity for appropriate comparison to generator power. The authors are unaware of any study, to date, where rotor rotational velocity and FOWT platform motions are simultaneously predicted.

The objective of this study is to develop a FOWT simulation tool that combines an inertial drivetrain model and VS controller, a mooring-line model, and overset CFD with advanced turbulent modeling and a high-resolution gridset. The developed tool is applied to the OC3-Hywind, a model developed by NREL and simulated by multiple expert entities in the Offshore Code Comparison Collaboration (OC3) [21], [22]. Simulations of increasing complexity are performed and results are compared with results produced by NREL during the OC3 [23] using the industry recognized wind turbine simulator FAST [7]. Time histories of predicted platform and rotor motions are analyzed along with predictions of developed and generated power. The effects of platform pitching velocity on blade pressure is examined in pressure coefficient plots.

Section snippets

Geometry

The OC3-Hywind, shown in Fig. 1, is a variable-speed, variable collective-blade-pitch-to-feather controlled spar-buoy FOWT model based on the full-scale Hywind model developed by Statoil of Norway [24]. It utilizes a 3-bladed, 125 m diameter rotor located at a hub height of 90 m. The turbine sits atop a cylindrical, ballast stabilized spar-buoy platform. Detailed geometric specifications are available in Refs. [21], [22].

Fig. 1 shows the three coordinate systems used. The earth-fixed system (X,

Load cases

Four simulations (Cases 1–4) are performed. A summary of simulated cases is presented in Table 3. All cases use 8 m/s steady, unidirectional incoming wind, approximately 70% rated wind velocity. Case 1 and Case 2 are presented as baseline cases without platform motion and use the input conditions from the OC3’s load cases 2.1a and 2.1b, respectively, detailed in Ref. [34]. In both of these cases no platform motions occur and hydrodynamic loading and waves are both disabled, although all

Overview of flow field

Several features key to FOWT wake modeling and simulation are observed in the flow field of the near and far wake, which is visually depicted in Fig. 7(a) through (d). Here (a), (b), and (c) present 3-dimensional views of the turbine and Fig. 7(d) shows contours of streamwise velocity at the central vertical cross section of the system. Tip vortices are the dominant feature in Fig. 7(a), (b), and (c), visualized with isosurfaces of the Q-criterion [36]. These vortices generate a helical

Conclusions

An inertial rotor model with a VS generator-torque controller is coupled with high resolution CFD and a mooring force model to predict motion and generated power of FOWT. The developed code is utilized in four simulations of the OC3-Hywind FOWT using the OC3’s LC 2.1a, 2.1b, and 5.1 wind and wave conditions. Results are compared to the publically available OC3 results of NREL using their FAST software. Simulations utilize an incremental approach for verification of the method. OC3 LC 2.1a,

Future work

The present results suggest a modification of the VS controller scheme. The VS controller is designed to maximize power capture below rated rotor rotational speed. At, or beyond rated velocity, however, requires a method of releasing torque from the system to avoid generator overload. A blade-pitch controller will be added in the future to the developed code for analysis of rated conditions and beyond. Combined with the Mann wind model, recently implemented into CFDShip-Iowa [38] OC3 LC 5.2

Acknowledgements

The present study is funded by the National Science Foundation, Division of Chemical Bioengineering, Environmental, and Transport Systems, under award number 1066873. All simulations presented were performed using the Idaho National Laboratory’s Falcon HPC, an SGI ICE with 16,416 cores. Gratitude is offered to both of these organizations for their help and resources. NREL’s accomplishments in all phases of the OC3, several of which are compared to in this study, are also acknowledged. Jason

References (39)

  • T. Burton

    Wind Energy Handbook

    (2011)
  • J.R. Morison et al.

    The force exerted by surface waves on piles

    J. Pet. Technol.

    (1950)
  • J.M. Jonkman et al.

    FAST User’s Guide

    (2005)
  • D. Matha et al.

    Variations in ultimate load predictions for floating offshore wind turbine extreme pitching motions applying different aerodynamic methodologies

  • T. Sebastian et al.

    A comparison of first-order aerodynamic analysis methods for floating wind turbines

  • C. Bak et al.

    Influence from blade-tower interaction on fatigue loads and dynamics

  • S. Quallen et al.

    An investigation of the blade tower interaction of a floating offshore wind turbine

  • D. Matha et al.

    Challenges in Simulation of Aerodynamics, Hydrodynamics, and Mooring-line Dynamics of Floating Offshore Wind Turbines

    (2011)
  • A. Sánchez-Caja et al.

    Computation of the incompressible viscous flow around a tractor thruster using a sliding-mesh technique

  • Cited by (0)

    View full text