Back to the programme printer.gif Print




Delegates are invited to meet and discuss with the poster presenters in this topic directly after the session 'Aerodynamics and rotor design' taking place on Wednesday, 12 March 2014 at 09:00-10:30. The meet-the-authors will take place in the poster area.

Emmanuel Branlard DTU, Denmark
Co-authors:
Emmanuel Branlard (1) F P Ewan Machefaux (1) Mac Gaunaa (1) Niels Troldborg (1) H.H. Brandenborg Sørensen (2)
(1) DTU, Roskilde, Denmark (2) DTU Compute, Lyngby, Denmark

Printer friendly version: printer.gif Print

Abstract

Validation of vortex code viscous models using lidar wake measurements and CFD

Introduction

The application of vortex methods to wind energy have received regained interest over the last decade[10]. The physical modelling and the computational time requirement of vortex code may be seen as intermediate between the ones of Blade Element Momentum(BEM) codes and the traditional computational fluid dynamics (CFD) codes[3]. Yet, a wide range of formulations are possible in vortex methods and different level of complexity may be achieved[6, 11]. The current study will investigate wake descriptions from vortex methods by comparison with lidar-measurements and actuator-line large eddy simulation(LES) CFD.


Approach

Inviscid panel methods and lifting line codes have been extensively used for aircraft and helicopters modelling[20, 12, 15]. Wind energy application differs from the fact that low induction in the wake keeps the vorticity close to the rotor. The accuracy of the near-wake description is thus of greater importance than for helicopter and aircrafts. Performance of lifting line codes have shown to give results within the same range of accuracy than BEM codes[16, 9, 5, 3] but it is usually difficult to prove better performance. The physical modelling of such vortex codes is of greater complexity but they usually suffer from their singular behavior and results are usually dependent on the choice of the viscous core model[22]. A possible improvement of these vortex codes lay in the introduction of viscosity. Purely inviscid vortex methods will fail at predicting accurately far-wake behaviors[3] where viscous diffusion plays a major role. A consistent description of viscosity from near-wake to far-wake is thus expected to improve the accuracy of vortex methods while at the same time resolving problems associated with their singular behavior. Vortex methods were successfully applied for simulation of viscous flow[6], showing in some cases more fidelity than standard CFD[2]. Such results are attributed to the mesh-free Lagrangian characteristics of some vortex-methods which thus do not suffer from numerical diffusion and dispersion. The introduction of viscosity in vortex methods can be done in various ways[1]. The approach chosen for this study consists in implementing several of these methods, analysing the influence of their main parameters and comparing their performance with respect to actuator-line CFD results and lidar wake measurements.



Main body of abstract

A vortex code has been recently implemented at DTU. The purpose of this code is twofold. It was implemented in view of its application to aeroelastic simulation of wind turbines and was coupled to DTU aero-servo-elastic code Hawc2[21]. The vortex code is also intended to be used as a research tool to study various vortex configurations and study the contributions of different vortex components separately. The first condition requires the code to support long time-series simulation for standard load cases applications while remaining as computationally inexpensive as possible. The second conditions requires the code to stay as general as possible so that it could be used to attempt to improve BEM models or investigate concepts such as kites, tip-rotors, tip-vanes, shrouded rotors, vertical axis wind-turbines, etc. The approach and choices made to satisfy these two conditions will be briefly described in this paper. The vortex code uses lower order vortex elements such as vortex segments, vortex particles, constant source and doublet panels[11]. Lifting and non-lifting bodies can be present in the flow. Bodies are allowed to move and deform and may be controlled by an elastic code such as Hawc 2. The lifting bodies may have a thick representation or a surface representation in which case the non-flow through condition is used to determine the forces on the body[11]. Other lifting bodies may be represented as lifting lines, in which case 2D tabulated profile coefficient data and an iterative scheme are used to determine the aerodynamic loads(see e.g. [23]). The wake consists of vorticity segments and particles and an hybrid-wake scheme[24] is used to achieve constant computational time per time-step.

A succinct presentation of the main possible implementations of viscosity in vortex methods will be presented. The one chosen for investigation in this study will be described in more details. The methods chosen are the random walk method[4, 18], the particle strength exchange method[8, 7] and the viscous core spreading method[1, 19]. Sensitivity analyses based on the different parameters influencing these methods will be performed. The challenges and scope of each of them will be outlined.

Two recent lidar measurement campaigns performed at Risø were also sources of motivation for the implementation of such viscous models to the vortex code. These measurement campaigns consisted in nacelle-mounted lidars pointing downstream for the investigation of wind turbines operating in single, double-wakes or partial wakes. The staring pattern of the continuous wave lidars were optimized to cover an important cross section of the wake in a short time. Several downstream wake cross sections were scanned successively. This measurement campaign will be described in a third part of this article. The presentation of the actuator line CFD simulations used for comparison will follow. These simulations were performed using the in-house fluid dynamic code EllipSys3D[17]. Comparison of CFD and lidar measurement data were already done in a previous work[13] so that the focus of this study will be on the vortex code results.

The following part of the article presents the performance of the vortex code and CFD code for simulations parameters that matches the running conditions of the wind turbine. Turbulence is artificially introduced in both codes using divergence free turbulence boxes derived from boxes obtained using the Mann model[14]. The performances of the different viscous models implemented will be confronted to the CFD results. Results from both codes will also be compared with measurements for 10min statistics. The velocity profile in the different measured wake cross sections will be compared for different operating condition of the turbine: in terms of thrust coefficient and for single-wake or double-wake situations. CFD and vortex methods aero-elastic simulations using Hawc2 may be considered. Indeed the meandering of the wake depends on large scale turbulence but also on the yaw variations of the wind turbine regulated by the controller[13].



Conclusion

This study presents a vortex code which is coupled to an aero-servo-elastic tool for simulation of different wind turbines and wind turbines concepts. The vortex code can include influence of tower and nacelle and is not restricted to wind-turbine application. The baseline inviscid model used in this study is a lifting line model since comparison with actuator line simulation is done. Nevertheless, higher order formulation are possible and the possibility of including boundary layer models is under investigation. Improvements of such vortex code using viscous modelling is suggested in this study since a better modelling of the wake by vortex methods should imply better performance evaluation at the rotor plane. Various viscous models were implement and tested against CFD results. The choice of vortex viscous models is a compromise between implementation complexity, computational time and performance. Results from both the vortex and CFD codes were compared to lidar measurements. Comparison with experimental data requires the use of turbulence boxes which match the experimental configuration and preferably the use of a aero-elastic simulation with a yaw controller. Indeed a large uncertainty is introduced by the choice of turbulence box and the yaw variation. Turbulence and shear were not modelled in potential terms but added to the incoming flow. Future work should study the influence of including a careful modelling of both turbulence and shear within vortex codes. The present study offers new opportunities for numerical simulations of wind turbines in configuration were standard BEM codes are beyond the range of their applicability.



Learning objectives
Long term aero-elastic simulations of wind turbines using vortex methods are possible by using different numerical techniques. Vortex methods are not by nature inviscid methods. Different viscous models exists the performance of some of them being evaluated in this study. Wake measurements are of high value for vortex methods.



References
[1] Barba, L. A. Vortex Method for computing high-Reynolds number flows
: Increased accuracy with a fully mesh-less formulation. PhD thesis, Cali-
fornia Institute of technology, 2004.
[2] Beale, J. T., and Majda, A. Rates of convergence for viscous splitting
of the navier-stokes equations. Mathematics of Computation 37 (1981),
243–259.
[3] Branlard, E. Wind turbine tip-loss corrections: Review, implementa-
tion and investigation of new models. Master’s thesis, Risø-DTU, Siemens
Energy Inc. (available at DTU’s library), September 2011.
[4] Chorin, A. J. Numerical study of slightly viscous flow. Journal of Fluid
Mechanics 57 (1973), 785–796.
[5] Cline, S. Potential flow modelling or wind turbines. Master’s thesis,
University of Victoria, 2011.
[6] Cottet, G.-H., and Koumoutsakos, P. Vortex methods: theory and
practice. Cambridge University Press, 2000.
[7] Degond, P., and Mas-Gallic, S. The weighted particle method for
convection-diffusion equations. part 1: The anisotropic case. Mathematics
of Computation 53, 188 (1989), 509–525.
[8] Degond, P., and Mas-Gallic, S. The weighted particle method for
convection-diffusion equations. part 1: The case of an isotropic viscosity.
Mathematics of Computation 53, 188 (1989), 485–507.
[9] Gupta, S., and Leishman, J. Comparison of momentum and vor-
tex methods for the aerodynamic analysis of wind turbines. 43rd AIAA
Aerospace Scieces and Exhibit (2005).
[10] Hansen, M. O. L., Sø rensen, J. N., Voutsinas, S., Sø rensen, N.,
and Madsen, H. A. State of the art in wind turbine aerodynamics and
aeroelasticity. Progress in Aerospace Sciences 42 (2006), 285–330.
[11] Katz, J., and Plotkin, A. Low-Speed Aerodynamics, 2nd Edition. Cam-
bridge Aerospace Series(No. 13). Cambridge University Press, February
2001.
[12] Leishman, J. G. Principles of Helicopter Aerodynamics - Second Edition.
Cambridge Aerospace Series(No. 12). Cambridge University Press, april
2006.
[13] Machefaux, E., Larsen, G., Troldborg, N., and Rettenmeier, A.
Single wake meandering, advection and expansion - an analysis using an
adapted pulsed lidar and cfd les-acl simulations. In Prodeedings of EWEA
2013 (2013), EWEA - The European Wind Energy Association.
[14] Mann, J. Wind field simulation. Prob. Engng. Mech. 13, 4 (1998), p269–
282.
[15] Maskew, B. Program vsaero theory document. Tech. rep., Nasa contrac-
tor report 4023, 1987.
[16] Sant, T. Improving BEM-based aerodynamics models in Wind turbine de-
sign codes. PhD thesis, DU Wind - Delft University Wind Energy Research
Institute, 2007.
[17] Sørensen, N. General purpose flow solver applied to flow over hills. PhD
thesis, Risø-DTU, 1995.
[18] Spalart, P. R. Numerical simulation of separated flows. Tech. rep.,
NASA Technical Memorandum 84328, 1984.
[19] Squire, H. The growth of a vortex in a turbulent flow. Aeronautical
Quaterly 16 (1965), 302–306.
[20] Stepniewski, W., and Keys, C. Rotary-Wings Aerodynamics. Dover
Publication, 1984.
[21] T.J.Larsen, A. H. HAWC2 - User manual. DTU-Risø-R-1597, 2007.
[22] Uzol, O. Free-wake calculations of wind turbine wake-rotor interactions:
The effect of vortex filament core size, expansion and stretching on perfor-
mance prediction. Tech. rep., Science of making torque from the wind -
Oldenbug 2012 proceedings, 2012.
[23] van Garrel, A. Development of a wind turbine aerodynamics simulation
module. Tech. Rep. ECN-C–03-079, ECN, 2003.
[24] Voutsinas, S. G. Vortex methods in aeronautics: how to make things
work. International Journal of Computational Fluid Dunamics (2006).