2002 Wwece Ion Cardaav

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/237447045

PREDICTION CAPABILITIES IN VERTICAL-AXIS WIND TURBINE


AERODYNAMICS

Article · January 2002

CITATIONS READS

8 1,385

3 authors, including:

Farooq Saeed
Imam Abdul Rahman bin Faisal University
50 PUBLICATIONS   602 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Farooq Saeed on 05 February 2016.

The user has requested enhancement of the downloaded file.


The World Wind Energy Conference and Exhibition, Berlin, Germany, 2-6 July 2002

PREDICTION CAPABILITIES IN VERTICAL-AXIS WIND TURBINE AERODYNAMICS

I. Paraschivoiu, F. Saeed and V. Desobry


J.-A. Bombardier Aeronautical Chair, École Polytechnique de Montréal
C.P. 6079, Succ. Centre-ville, Montréal (Québec) Canada, H3C 3A7
Phone: (514) 340-4711 ext.: 4583, Fax: (514) 340-5917,
E-mail: ion.paraschivoiu@meca.polymtl.ca

ABSTRACT: The paper summaries the progress made in the development of the different state-of-the-art numerical
tools for simulating the flow field around the vertical-axis wind turbine as well as the dynamic-stall phenomenon. More
specifically, the paper highlights the current capabilities and limitations of aerodynamic models with emphasis on double-
multiple streamtube model as a design tool (CARDAAV code). The paper also highlights some of the recent
improvements for an enhanced performance of VAWT through a variable pitch for straight-bladed rotor to regulate
dynamic-stall regime. Furthermore, a new inverse design tool has been developed to optimize VAWT performance using
a Genetic Algorithm in conjunction with the CARDAAV code. The strength of the numerical tool based on the DMS
model lies in its ability to analyze secondary effects that interfere with aerodynamic calculations and the dynamic stall.
Keywords: Blade Aerodynamics, Dynamic Stall, Models (Mathematical), Numerical Methods, Navier-Stokes Equations,
and Wind Turbine (VAWT).

1 INTRODUCTION model was developed by Strickland [3] who considered


the swept volume of the turbine as a series of adjacent
The Vertical-Axis Wind Turbine (VAWT) offers a streamtubes. Paraschivoiu [4] developed an analytical
mechanically and structurally simple method of model, the double-multiple streamtube model (DMS
harnessing the energy of the wind. The blade elements model), that considers a multiple-streamtube system
operate in both unstalled and stalled conditions with divided into two parts where the upwind and downwind
dynamic stall providing the rotor inherent power components of the induced velocities at each level of the
regulation. The J.-A. Bombardier Aeronautical Chair at rotor are calculated by using the principle of two actuator
Ecole Polytechnique has developed computer codes to disks in tandem.
predict aerodynamic loads and performance of a Darrieus
rotor including viscous, stochastic wind and dynamic-stall 2.1 Codes Based on the DMS Model
effects [1]. A fundamental study of the dynamic-stall The DMS model has been implemented in CARDAAV
phenomenon has also been performed using the unsteady computer code in which VAWTs can be designed to meet
Navier-Stokes equations in a streamfunction-vorticity the desired specifications. The code can also be used to
formulation. satisfy certain desired parameters, for example, the
The development of the different state-of-the-art influence of the blade geometry, the airfoil type, and the
numerical tools for simulating the flow field around the presence of struts and aerodynamic spoilers on the
VAWT as well as the dynamic-stall phenomenon are Darrieus turbine thus making CARDAAV a very
based on methods that can be divided into four major attractive and efficient design and analysis tool.
categories: momentum models, vortex models, local- Three computer code variants based of the DMS model
circulation models and viscous models. In the momentum have been developed: CARDAA, CARDAAV and
model, the rotor is replaced by an actuator disk and then CARDAAX. The first variant, CARDAA, uses two
the flow through the rotor is determined by equating the constant interference factors in the induced velocities, for
forces on the rotor blades to the time rate of change in each half-streamtube. It uses semi-empirical dynamic-
momentum through the rotor. The vortex models are stall model and has the capability to analyze and to
based on the vorticity equation and use Biot-Savart law to predict the power output of a VAWT whose blades
calculate induced velocities over the blades. In the local- contain step changes in airfoil chord length and section
circulation model, the blade is represented by a profile. This code is also used for the prediction of
superposition of imaginary blades of different spans with VAWT dynamic structural response. The CARDAAV
elliptical circulation distribution in the same manner as code considers the variation of the interference factors as
used for rotary wings. All these three methods consider a function of the azimuthal angle on the upwind and
the fluid to be inviscid. A viscous model solves a 2D/3D downwind half of the rotor. In addition, a number of
steady, laminar Navier-Stokes equations in primitive secondary effects are also taken into account, namely the
variable approach. blade geometry and profile type, the rotating tower and
the presence of struts and spoilers. Two semi-empirical
dynamic-stall models (Gormont and Indicial models) are
2 MOMENTUM MODELS included in CARDAAV code. Finally, since the airflow
over Darrieus turbine is cut twice during one cycle, with
The first approach to analyze the flow field around a the result that the downwind zone of the rotor is strongly
VAWT was developed by Templin [2] who considered disturbed by the upwind zone, the third variant,
the rotor as an actuator disk enclosed in a simple CARDAAX, incorporates a simplified physical model
streamtube where the induced velocity through the swept that takes the streamtube expansion effects into account
volume of the turbine is assumed to be constant. An and allows a more realistic modeling of the upwind/
extension of this method to the multiple-streamtube downwind flow field asymmetries inherent in a VAWT.
3 STOCHASTIC MODEL rotation on the power output. Three different
modifications to the local effective angle of attack were
Turbulent flow effects have been identified as one of studied with the aid of the CARDAAV code. In the third
the major source of rotor blade fatigue and several studies modification (mod3), being the most significant as well as
have examined the effect of atmospheric turbulence on a most practical of the three, the magnitude of the local
rotating wind turbine blade [5, 6]. A new code named effective angle of attack during the rotation was set as a
CARDAAS, which refers to the use of stochastic wind, sinusoidal function whose maximum amplitude was set
was developed to predict aerodynamic loads on Darrieus equal to the sum of the local effective angle of attack and
rotor by taking the fluctuating nature of the wind into the maximum difference between the local effective angle
account [6]. The stochastic wind model simulates the of attack and the local stall angle. As such the rotor blade
turbulent velocity fluctuations by incorporating both the always operates below the local stall angle and the
streamwise, lateral components and fluctuations in DMS evolution of the effective angle of attack during a
model. The 1-D variations of this turbulent wind are complete rotation is smooth and continuous. In the first
introduced by creating time series of the wind velocity at two modifications (mod1 & mod2), the local effective
a fixed point upwind of the rotor by assuming that the angle of attack is kept just below the local stall and,
wind speed is constant in a plane perpendicular to the hence, results in sharp jumps in the local effective angle
mean wind direction. In the 3-D wind model a wind of attack that are highly undesirable from a practical point
speed time series is generated to represent the variation of view. Moreover, although the power output obtained
about the mean velocity in the longitudinal and vertical by the third modification is less than that of the first two,
directions. it results in a smooth and continuous change in the
resulting aerodynamic loads and hence less fatigue. The
third modification is, in our opinion, a good compromise
4 VISCOUS MODELS between maximum gain in effectiveness and represents a
practical way to realise a variable pitch for a straight-
The flow field of a rotating machine, especially the bladed rotor and regularization of dynamic stall in power
wake, is dominated by the viscous and rotational curve vs. wind speed.
characteristics. Two computer codes based on the
viscous flow field models have been developed by
Rajagopalan [7, 8]. The numerical model is based on the 7 VAWT DESIGN OPTIMIZATION USING A
steady laminar incompressible Navier-Stokes equations in GENETIC ALGORITHM
a cartesian domain (VFFVAT code - Viscous Flow Field
of Vertical-Axis Turbine) or in cylindrical coordinates A new Darrieus VAWT design and analysis tool has been
using a control volume approach (3DVF code). developed. It is based on the coupling of CARDAAV
code and a genetic algorithm (the GENIAL code
developed by Henrik Widell [11]). Thus, with this tool,
5 DYNAMIC-STALL MODELS optimization of the wind turbine performance for a given
set of constraints and an objective function can provide
In the case of Darrieus wind turbine, when the design parameters for maximizing the ability of the wind
operational speed approaches its maximum, all the blade turbine to extract the wind energy for a given power. The
sections exceed the static stall angle, the angle of attack variables considered in this optimization are the
changes rapidly and the whole blade operates under geometrical and operational parameters such as: height,
dynamic-stall conditions increasing the unsteady blade aspect ratio, chord length, number of blades, airfoils,
loads and structural fatigue. Dynamic stall significantly rotation speed, etc. The use of a genetic algorithm is
influences the global performance of wind turbines. particularly suitable for the combined analysis of these
The semi-empirical models were employed in our variables (see Fig. 1). Furthermore, the tool can also be
computer codes to evaluate the performance of the wind used for parametric studies since a large number of
turbine, however, they are limited to the type of airfoil evaluations are possible due to the fast execution time of
and motion used in the experiment from which they were the CARDAAV code.
derived. In addition, they cannot predict the evolution of
the boundary layer and pressure distribution that are
essential for studying dynamic stall. Hence, a new code 8 RESULTS AND DISCUSSION
for simulating the dynamic stall around Darrieus turbine
has been developed that solves the Navier-Stokes The performance prediction of Sandia 17-m wind turbine
equations using the finite-element method in a (Fig. 2) using CARDAA code was improved by
streamfunction-vorticity formulation [9]. Since 3-D considering the variable interference factors (CARDAAV
simulation would be very expensive a 2-D simulation has code) and by including streamtube expansion effects
been adopted. (CARDAAX code). CARDAA overpredicts the power
coefficient peak while CARDAAV and CARDAAX
results show good agreements when compared to
6 STRAIGHT-BLADE VARIABLE-PITCH VAWT experimental data. In addition, these results are very
close to vortex model results (VDART3), [12]. The
One of the major problems in the Darrieus rotor is to simulation of the dynamic-stall hysterisis loop using
regulate dynamic-stall regime for maximizing the 3DVF code with indicial model, CARDAAV predictions
performance of the wind turbine. A study [10] was in the upwind and downwind regions of the turbine agree
carried out to examine the effect of changing the local well with experimental data [13], (Fig. 3).
effective angle of attack from its original value during
Input variables and their
boundary

Initial Population

Call CARDAAV and


evalution of Objective
Function

Yes
Sort of the best
Last Generation chromosome between
the populations
Figure 4 Comparison of normal force coefficient at tip-
No speed ratio of 2.86 and turbulence = (27%, 25%)

Selection of Figure 4 compares the normal force coefficients for the


chromosomes from later steady state and stochastic loads, using a 3-D wind
population and
generation of child simulations (CARDAAS 3D), viscous code 3DVF and
chromosomes experimental data at a tip-speed ratio of 2.86 including
Gormont’s dynamic-stall model. 3-D models predict
almost the same ensemble-averaged values of normal
Insertion of the new force coefficient although the instantaneous values may
chromosomes (new have significant variations. For low turbulence intensity
population)
however, the deviation between the ensemble-averaged
values and steady-state predictions becomes much
Figure 1 Flow chart for the Genetic Algorithm smaller than in the case of high turbulence intensity.

Figure 2 Comparison of different analysis codes

Figure 5 Numerical simulation of dynamic stall

In the above aerodynamic codes the model used for


dynamic stall is based on semi-empirical methods. The
simulation of the computed flow structure around an
airfoil in Darrieus motion using Navier-Stokes solver in a
streamfunction-vorticity formulation is given in Fig. 5 in
Figure 3 Normal force coefficient vs. angle of attack at term of streamline evolution. Results using Johnson-
tip-speed ratio of 2.49 King turbulence model can predict the region where the
dynamic stall may occur. In the upwind zone of the unsteady Navier-Stokes equations. The differents codes
rotation, the dynamic stall occurs between 26 deg and 195 developed so far represent valuable tools for simulating
deg of the azimuthal angle θ. This dynamic stall the flow field around the turbine and dynamic-stall
generates vortices that result in peaks in the normal and phenomenon. Although comparison of theoretical results
tangential force coefficients. with experimental data have shown good agreements,
Figures 6 and 7 illustrate the effect of blade effective further work is under consideration to develop improved
angle of attack regulation on the power output of a models of the wind encountered by the rotor, the
straight-blade VAWT. It can be seen that the three atmospheric flow across a site, the wake that affects
modifications (mod1-3 in Fig. 7) yield more power for blades in addition to the development of optimization
higher wind speeds than the case when the VAWT is tools.
operating under static or dynamic stalls. Amongst these
modifications, only mod3 is physically feasible (Fig. 6)
because of the continuity of the blade pitch angle. ACKNOWLEDGEMENTS

The authors gratefully acknowledge the support of


NSERC through their Grant 1442 and Bombardier
Foundation.

REFERENCES

[1] Paraschivoiu, I., "Wind Turbine Design with emphasis


on Darrieus Concept," Polytechnic International Press,
Montreal, Québec, Canada, 2002, 438p.
[2] Templin, R.J., "Aerodynamic Performance Theory
for the NRC Vertical Axis Wind Turbine," NRC of
Canada, Report LTR-LA-10, 1974.
[3] Strickland, J.H., "The Darrieus Turbine: A
Performance Prediction Model Using Multiple
Streamtubes," Sandia National Laboratories Report,
SAND 75-041, 1975.
[4] Paraschivoiu, I., "Aerodynamic Loads and
Performance of the Darrieus Rotor," J. of Energy, Vol.,
Figure 6 Variation of blade effective angle of attack No., pp. 406-412, 1981.
[5] Veers, P.S. "Modeling Stochastic Wind Loads on
Vertical Axis Wind Turbines," Sandia National
Laboratories SAND83-2909, Albuquerque, NM, 1984.
[6] Brahimi, M.T., and Paraschivoiu, I., "Darrieus Rotor
Aerodynamics in Turbulent Wind," ASME, J. of Solar
Energy Engineering, Vol. 117, May 1995, pp. 128-136.
[7] Rajagopalan, R.G., "Viscous Flow Field Analysis of
a Vertical-Axis Wind Turbine," Proceedings of the 21st
Intersociety Energy Conversion Engineering Conference,
vol. 2, San Diego, CA, Aug. 25-29, 1986, pp. 1242-1247.
[8] Paraschivoiu, I., Masson, C., Rajagopalan, R.G.,
"Predictions and Experiments of the VAWT Viscous
Flow Field," AIAA Paper 87-1429, 8 pp., 1987.
[9] Tchon, K.F., "Simulation numérique du décrochage
dynamique sur un profil d'aile en mouvement de rotation
Darrieus," Thèse de doctorat, Dépt. de Génie Mécanique,
École Polytechnique de Montréal, 1993.
[10] Staelens, Y., Saeed, F., and Paraschivoiu, I.,
"Modifications to a Darrieus-Type VAWT for Improved
Power Generation," to be presented at the 22nd Wind
Figure 7 Power output as a result of blade pitch
Energy Symposium, Reno, NV, USA, Jan. 2003.
regulation
[11] Widell, H., “Genial 1.1, a Friendly Function
Optimizer based on Evolutionary Algorithms,” 1997,
http://www.hjem.get2net.dk/widell/
9 CONCLUSIONS
[12] Strickland, J.H., Webster, B.T., and Nguyen, T., "A
Vortex Model of the Darrieus Turbine: An Analytical
Computer codes have been developed to predict
and Experimental Study," Sandia National Laboratories
aerodynamic loads and performance of a Darrieus rotor
Report, SAND79-7058, 1980.
including viscous, stochastic wind, and dynamic-stall
[13] Akins, R.E., "Measurements of Surface Pressure on
effects. Three dynamic-stall models are included in our
an Operating Vertical Axis Wind Turbine," Sandia
wind turbine load/performance computer codes:
National Laboratories, SAND89-7051, Albuquerque,
Gormont, MIT and Indicial models. A fundamental study
NM, 1989.
of dynamic stall has been performed by solving the

View publication stats

You might also like