Académique Documents
Professionnel Documents
Culture Documents
Strickland
Assoc. Professor of Mechanical Engineering.
A Vortex Model of the Darrieus
Mem.ASME
Turbine: An Analytical and
B. T. Webster
Graduate Assistant. Experimental Study
An aerodynamic prediction model has been formulated for two- and three-
dimensional Darrieus turbines using a vortex lattice method of analysis. Ex-
T. Nguyen
periments were conducted on a series of two-dimensional rotor configurations in a
Graduate Assistant. water tow tank. The agreement between analysis and experiment was in general
College of Engineering, found to be good. This model should allow one to make accurate predictions of
Texas Tech University, instantaneous aerodynamic blade forces and to characterize the near wake flow
Lubbock, Texas 79409
behind the rotor.
Introduction
Several aerodynamic performance prediction models have assessments of the environmental impact of large scale rotors
been formulated for the Darrieus turbine. The models of on downstream areas.
Templin [1], Wilson and Lissaman [2], Strickland [3], and Alteration of the simple momentum models to alleviate all
Shankar [4] have all been used to predict the performance of of the listed objections presently appears to be unlikely. On
three-dimensional Darrieus rotors. Each of these models (the the other hand, further development of the "vortex model"
latter three being virtually identical) are based upon equating appears to be quite worthwhile. Several of these vortex
the forces on the rotor blades to the change in streamwise models for vertical wind machines have been developed in the
momentum through the rotor. The overall performance can past, although none are applicable to the three dimensional
be predicted reasonably well with these models under con- curve-bladed Darrieus rotor. Models which typify previously
ditions where the rotor blades are lightly loaded and the rotor developed vertical axis vortex models are those due to Fanucci
dp to wind speed ratios are not high. [5j, Larsen [6], Wilson [7], and Holmes [8].
While these models are moderately successful at predicting
overall performance trends they are inadequate from several While none of the previously developed vortex models are
standpoints. Accurate performance predictions for large tip applicable to the curve-bladed Darrieus turbine, they do
to wind speed ratios cannot be made because the momentum provide valuable insight into concepts which can be used in a
equations used in these models become invalid. This situation three-dimensional vortex model for the Darrieus turbine.
deteriorates with increasing rotor solidity. Predicted blade Thus, the present work is simply a logical extension of
loads are inaccurate since these models (1) assume a quasi- previous work. There are two major extensions which are
steady flow through the rotor, (2) assume a constant required. The first of these is the extensions from a two-
streamwise velocity as a function of streamwise position in the dimensional vortex structure to a three-dimensional structure.
vicinity of the rotor, and (3) assume that the flow velocities The vortex models of Fanucci [5], Wilson [7], and Holmes [8]
normal to the freestream direction are zero. It is doubtful that are strictly two-dimensional models. The vortex model of
meaningful information concerning the near wake structure Larsen [6] is not strictly two-dimensional if the vortices
of the rotor can be obtained from the present models. This trailing from the rotor blade tips are considered. In any event
information may be important with regard to the placement the vortex structure for the curve-bladed Darrieus is much
of rotors in close proximity to each other and in making more complex than any of those considered previously. The
second major extension requires that aerodynamic stall be
considered in the model. Both Fanucci [5] and Holmes [8]
assumed that the rotor blades were always at angles of attack
sufficiently small such that aerodynamic stall was not en-
countered. The giromill [6, 7] has articulating blades which
operate at angles of attack that are less than the stall threshold
Contributed by the Fluids Engineering Division for presentation at the levels. Since a portion of the curve-bladed Darrieus rotor near
Winter Annual Meeting, New York, N. Y., December 2-7, 1979, of THE the hub is stalled a major part of the time, aerodynamic stall
AMERICAN SOCIETY OF MECHANICAL ENGINEERS. Manuscript received by the
Fluids Engineering Division, February 21, 1978. Paper No. 79-WA/FE-6. must be considered.
Aerodynamic Model
The general approach requires that the rotor blades be induced at a point in the flow field by a single vortex filament
divided into a number of segments along their span. The can be obtained from the Biot-Savart law which relates the
production, convection, and interaction of vortex systems induced velocity to the filament strength. Referring to the case
springing from the individual blade elements are modeled and shown in Fig. 2 for a straight vortex filament of strength T
used to predict the "induced velocity" or "perturbation and length / the induced velocity vp at a point p not on the
velocity" at various points in the flow field, the induced or filament is given by [10]
perturbation velocity at a point is simply the velocity which is
superimposed on the undisturbed wind stream by the wind
r
=e (cos d, + cos d2) (1)
machine. Having obtained the induced velocities, the lift and 4-rrh
drag of the blade segment can be obtained using airfoil where the unit vector e is in the direction of r x 1. It should be
section data. noted that if the point p should happen to lie on the vortex
A simple representation of the vortex system associated filament that equation (1) yields indeterminate results since e
with a blade element is shown in Fig. 1. The airfoil blade cannot be defined and the magnitude of \p is infinite. The
element is replaced by a "bound" vortex filament sometimes velocity induced by a straight vortex filament on itself is in
called a "substitution" vortex filament [9] or a "lifting line" fact equal to zero [12].
[10]. The use of a single line vortex to represent an airfoil In order to allow closure of the proposed vortex model, a
segment is a simplification over the two-dimensional vortex relationship between the bound vortex strength and the
model of Fanucci [5] which uses three to eight bound vortices velocity induced at a blade segment must be obtained. A
positioned along the camber line. The use of a single bound relationship between the lift L per unit span on a blade
vortex represents the flow field adequately at distances greater segment and the bound vortex stength TB is given by the
than about one chord length from the airfoil [9]. The Kutta-Joukowski law [11]. The lift can also be formulated in
strengths of the bound vortex and each trailing tip vortex are terms of the airfoil section lift coefficient C,. Equating these
equal as a consequence of the Helmholtz theorems of vorticity two expressions for lift yields the required relationship
[11]. As indicated in Fig. 1, the strengths of the shed vortex between the bound vortex strength and the induced velocity at
systems have changed on several occasions. On each of these a particular blade segment.
occasions a spanwise vortex is shed whose strength is equal to
the change in the bound vortex strength as dictated by TB=1-C,CUR (2)
Kelvin's theorem [11].
Each portion of a vortex filament making up the shed Here the blade chord is denoted by C and UK is the local
vortex system is convected in the flow at the local fluid relative fluid velocity in the plane of the airfoil section. It
velocity. Therefore, the vortex filament will be distorted in a should be noted that the effects of aerodynamic stall are
number of ways as it moves through the fluid. As a first automatically introduced into equation (2) through the section
approximation it can be assumed that the vortex filament lift coefficient.
remains straight with its ends being convected at their Kinematic Considerations. The curve-bladed rotor which is
respective local fluid velocities. Therefore, the vortex filament depicted in Fig. 3 is assumed to possess blades which are bent
may stretch, translate, and rotate as a function of time. into a parabolic shape along their spans. The parabola closely
The fluid velocity at any point in the flow field is the sum of approximates the ideal Troposkein shape which minimizes
the undisturbed wind stream velocity and the velocity induced blade bending stresses [13].
by all of the vortex filaments in the flow field. The velocity The relative velocity of a blade element with respect to the
n uuiciicia I ui c
Ft dimensionless normal force filament
AF = rotor frontal area F, tangential force /,, length of blade element
c = unit vector in the chordwise Ft dimensionless tangential force L lift
direction h distance from vortex filament n unit vector in normal direction
C = airfoil chord length H rotor height N number of blades
cd = drag coefficient i unit vector in x direction NE number of element ends
c, = lift coefficient i index AT number of time steps
c„ = normal coefficient J unit vector in j'direction NTI number of time increments
C
P = rotor power coefficient J index per revolution,
c,e = tangential coefficient k unit vector in z direction Q field variable
unit vector
= unit vector k index r
F„ = normal force 1 unit vector along vortex r radius to blade quarter chord
i i i i i i
airfoil lift and drag coefficients C, and Cd by:
C, =C/ sin a—Cd cos a Uy/Uos
Cn = - C , cos a-Cd sin a (10) Fig. 5 Two dimensional rotor performance (N 1, C/R = 0.15, Re
The torque produced by a single element can be written in 40,000)
nondimensional form by
rF, leCrF+ 1 i i i i i
T+ = (11) O 3 D Vortex Model (VDART 3)
\/2pRAFUl RAf 0.5 -
Simple Momentum Model [2,3,4]
where AF is the total projected frontal area of the rotor. The
average power coefficient for the entire rotor during a single 0.4 • Sandia Experiment [13]
•
revolution is given by
1 PJ1 1 l\tl - 1 TJ
(12)
0.3 / o ° \
Q
-
" NTI 1 1 U
rl \
<» 7 Q
\
Q
B \
where NTI is the number of time increments per revolution of
0.2 • -
B
the rotor and UT is the tangential blade velocity at the rotor 0 \
equator. 0.1 • 7 a
\ -
a \
Experimental Measurements 0.0 ^ - ^ ® -
In order to check the validity of the analytical model a
comparison between predicted and experimentally determined i i i i i
i i i i i i
airfoil lift and drag coefficients C, and Cd by:
C, =C/ sin a—Cd cos a Uy/Uos
Cn = - C , cos a-Cd sin a (10) Fig. 5 Two dimensional rotor performance (N 1, C/R = 0.15, Re
The torque produced by a single element can be written in 40,000)
nondimensional form by
rF, leCrF+ 1 i i i i i
T+ = (11) O 3 D Vortex Model (VDART 3)
\/2pRAFUl RAf 0.5 -
Simple Momentum Model [2,3,4]
where AF is the total projected frontal area of the rotor. The
average power coefficient for the entire rotor during a single 0.4 • Sandia Experiment [13]
•
revolution is given by
1 PJ1 1 l\tl - 1 TJ
(12)
0.3 / o ° \
Q
-
" NTI 1 1 U
rl \
<» 7 Q
\
Q
B \
where NTI is the number of time increments per revolution of
0.2 • -
B
the rotor and UT is the tangential blade velocity at the rotor 0 \
equator. 0.1 • 7 a
\ -
a \
Experimental Measurements 0.0 ^ - ^ ® -
In order to check the validity of the analytical model a
comparison between predicted and experimentally determined i i i i i