Journal of Wind Engineering and Industrial Aerodynamics, 15 (1983) 369--380 369

Elsevier Science Publishers B.V., Amsterdam -- Printed in The Netherlands

FEATURES TO AID OR ENABLE SELF STARTING OF FIXED PITCH LOW SOLIDITY

VERTICAL AXIS WIND TURBINES

J.R. BAKER
Electricity Division, New Zealand Ministry of Energy

SUMMARY

This paper reviews the cause of the inability of low solidity fixed pitch
vertical axis wind turbines to self start, and investigates a number of ways of
overcoming this problem. One approach is only applicable to variants of the
Musgrove VGVAWT and involves changes of blade geometry. Another approach is to
utilize blades having aerofoil sections that strongly exhibit the laminar
separation bubble phenomenon below the stall and that have low immediate post-
stall drag. Evidence is given that VAWT's can be made to consistently self start.

NOTATION
Cp turbine efficiency coefficient ( = power/ ~ p A v3
Ct thrust coefficient
E net energy produced by blade element in one revolution
q dynamic pressure
r blade radius
Re Reynolds number
V wind velocity experienced by blade element
Vd wind speed through plane of rotor
angle of attack of blade element
tip speed ratio ( = re/ Vd )

0 angle between span line of blade and the vertical

azimuthal angle of blade
co turbine rotational speed

I. INTRODUCTION

G.R. Watson showed in his paper "The self starting capabilities of low
solidity fixed pitch Darrieus rotors"[8] that the inability of Darrieus type
VAWT's to self start arises from the existance of a band of tip speed ratios
below the operating condition for which the net amount of energy collected by
each blade each revolution is negative. To achieve self starting a machine must
be altered in such a way that a net positive quantity of energy is produced at
all tip speed ratios up to the operating condition.

016%6105/83/$03.00 © 1983 Elsevier Science Publishers B.V.

 370

2. I N A B I L I T Y TO SELF START

Figure i p r e s e n t s the g e o m e t r y of a two d i m e n s i o n a l slice t h r o u g h a VAWT~ ahd

Figure 2 graphs S h a r p e ' s compilation[7] of data for the N A C A 0012 at a R e y n o l d s
number of 80,000.

It can be seen f r o m Figure 2 that at an angle of attack above 2 ° the section

has a p o s i t i v e Ct and will p r o d u c e forward thrust. This force w h e n m u l t i p l i e d by
the b l a d e radius r gives the instantaneous torque that the blade element is
c o n t r i b u t i n g to the turbine. B e y o n d about 15 ° (in Sharpe's compilation) ~he
v a l u e of Ct goes negative. A b l a d e element generates n e g a t i v e torque betwe<~
angles of a t t a c k of 15 ° and 65 ° . A b o v e 65 ° the torque is p o s i t i v e v i r t u a l l y ~<~
180 ° . Due to the symmetry of the N A C A 0012 section the Ct(~) curve repeats
itself in r e v e r s e from 180 ° to 360 ° .

Figure 3 graphs the angle of a t t a c k as a f u n c t i o n of azimuthal p o s i t i o n S r

various tip speed ratios. The b l a d e angles of attack that give rise to nega<:ive
Ct's and h e n c e n e g a t i v e torques are marked. B e t w e e n starting at a tip sp<.e:i
ratio of zero and a t t a i n i n g the o p e r a t i n g condition a blade runs up through tke
range of tip speed ratios that involve a c o n s i d e r a b l e dwell time in n e g a t i v e C%
regions. At these tip speed ratios the net energy p r o d u c e d by a blade arounl one
r e v o l u t i o n is negative. The net energy produced by a l'lade element in on{
revolution, E, is given by

E = ~ r q Ct(~) d @

Figure 4 shows E as a f u n c t i o n of tip speed ratio and highlights the 'dea<:

band'. A VAWT will start and a c c e l e r a t e to a tip speed ratio of about 0.7, when
unless energy is added from some other source, the m a c h i n e will not go faster.
Once the m a c h i n e achieves a tip speed ratio g r e a t e r t h a n about 2.2 i~ wi]!L
a c c e l e r a t e away and p r o d u c e power. (These values oll the limits of the dead band
d e p e n d on the p a r t i c u l a r values of Ct in S h a r p e ' s compilation.)

3. ALTERED GEOHETRY
C o n s i d e r i n g e q u a t i o n i, it w o u l d appear that two of the variables carl be
m a n i p u l a t e d to alter E. They are the section Ct(~) c h a r a c t e r i s t i c s and,
implicitly,~(@). This latter can be a l t e r e d by a l t e r i n g the g e o m e t r y of tm~
VAWT, and the former by the choice of blade section.

3.1 Initial O u t w a r d Tilt

If the blades of a straight b l a d e d VAWT are t i l t e d outwards im the f a s h i o or
M u s g r o v e ' s VGVAWT, then the angle of a t t a c k is dierctly a f f e c t e d by tn~ ansie <,f
 371

tilt (the angle b e t w e e n the span line of the blade and the vertical) and is

given b y

-i
O( = sin ( Vd ~ n ¢ cos9 ) (2)
V

where

V = ( ( r~- V d c o s @ )2 + ( Vd s i n @ c o s O )2 )~ (3)

The angle of attack e x p e r i e n c e d b y blades at v a r i o u s angles of tilt when at a

tip speed ratio of 1.5 is graphed in Figure 5- It can be seen that tilt causes a
reduction of the amount of time dwelt in the n e g a t i v e - C t - p r o d u c i n g region.

Tilt introduces an element of spanwise flow into what was p r e v i o u s l y an essent-

ially two d i m e n s i o n a l flow state. It will be shown later that the existance of
spanwise flow affects the Ct(~) curve however in this situation the effect is
small. The spanwise flow arises at the expense of the flow component normal to
the chord line. The flow p a r a l l e l to the chord, m a i n l y r~, remains unaltered. As
a consequence the angle of attack is d i m i n i s h e d and the blade spends a c c o r d i n g l y
less time badly stalled around the p e r i m e t e r of the turbine.

Figure 6 shows the effect of tilt on E as a f u n c t i o n of tip speed ratio. The

graph suggests that a straight b l a d e d VAWT w i t h an angle of tilt of 60 ° should

be able to self start. In fact consistent self starting has b e e n o b s e r v e d on a
turbine in Gisborne[5]. The m a c h i n e w o u l d start an winds above 5m/s for tilts
b e t w e e n 60 ° and 65 ° . Too m u c h tilt reduces the frontal area of the m a c h i n e to
~he extent that not enough power is p r o d u c e d to over come friction.

3.2 [~]ade F o r w a r d Lean

A n o t h e r mode of altered g e o m e t r y considered was that resulting from ].eaning the
blades of s straight b l a d e d VAWT forwards. This time the angle of lean results
in a reduction of the r ~ component of tile resultant wind. A wind tunnel study of
the NACA 0012 at low Reynolds ntmfoers and at v a r i o u s angles of y a w p e r f o r m e d by
the author[l] found that at angles of y a w greater than 20 ° the stall b e h a v i o u r
<of the section was considerably altered. Some results are shown in Wigures 7 and
8. The greater the angle of y a w the greater the stall angle and the smaller %he
subsequent negative Ct peak. No graph of E as a f u n c t i o n of tip speed rat!o is
presented for VAWT's i n c o r p o r a t i n g blades having angles of forward lean as in
thi,s case the airflows are h i g h l y three dimensional and an attempt at a two
dimensional simulation w o u l d be meaningless. Clearly yaw changes the Ct(~) curve
in desirable ways and should aid self starting.
372

A practical straight bladed VAWT utilising both the tilt and yaw effects mig~t
consist of one with a hinge of the Musgrove type rotated so that the bla~e ~,:
longer tilted outwards in the plane of the strut, but instead leaned f o r w s r d a.:
it tilted. This would allow it to retain the speed limiting spring that i: ~
valuable feature of Musgrove's machine. The offset hinge was suggested b~y
Mr. V.A.L. Chasteau, the author~ supervisor on the masters thesis from w~Jci
this paper came.

4. SELECTION OF DESIRABLE AEROFOIL SECTION FEATURES

One aspect of the performance of the NACA 0012 at sub-critical Reynolds
numbers that was apparant in the author's wind tunnel stud~ was the low angle of
attack at which it stalled. With Re in the millions the NACA 0012 stalls nea~
16 ° . For Re in the low hundred thousands the section stalls before it reaches
i0 ° and Ct instantly fails to a large negative value (about -0.06). O~e way of
aiding E in equation i would be to find an aerofoil section that has a more
desirable Ct(~) curve at the low Re's experienced by VAWT blades on starbing.
Such a section would stall late and ~ts Ct coefficient would then fall only to a
small negative value. Not relying on altered geometries such a section would be

applicable to the curved bladed Darrieus type turbine.

4.1 Delta Planform

Wings of delta planform have this characteristic and are used in the small
water pumping VAWT's built by RIKAN ENGINEERING LTD. of Napier. These machines
self start and have considerable low speed torque. They are essentially low tip
speed ratio devices, their maximum efficiency occurring at a tip s~eed rati~ a
little more than 3. They appear to have maximum Cp's between 0.25 and 0.3 ir
their variable pitch form, however are able to self start with blades rigidl~
held perpendicular to the struts. Using aerodynamic lift instead of irag, th,~~
would appear to bridge the gap between drag dependent rotors such as the

Savonius, and high speed VAWT's.

4.2 The Separation Bubble

At the high Re's associated with aeroplanes, boundary layer ~ransition from
laminar to turbulent flow occurs before the suction peak. At the low Re's
experienced by VAWT blades on starting, the boundary layer at the suction peak
is invariably laminar, and separation occurs there. Once the flow has separated
it generally becomes turbulent and in certain circumstances can reattach to form
a separation bubble[3]. The now attached turbulent boundary layer negotiates the
remaining pressure rise. At Re's in the io~ hundred thousarlds a separation
bubble exists on the NACA 0012 at all angles of attack up to the stall. Stalling
occurs when the turbulent flow is no longer able to reattach.
 373

Given that laminar separation is inevitable at the suction peak for aerofoils
at low Re's, a section that encourages the formation of a separation bubble will

stall later than one that does not. To aid reattachment of the separated flow a
section should have a rounded nose and camber near the leading edge[4].

Figure 9 compares the Ct(~) curves of the NACA 0012 and the G5 420[6]. The
G~ 420 (see Figure i0) reaches a higher m a x i m u m u n s t a l l e d Ct and has a
relatively much smaller negative stalled region.

4.3 Post-Stall Drag

It can be seen from figure 9 that the G6 420 exhibits a larger negative Ct
peak after the stall than the N A C A 0012. The n e g a t i v e Ct's after the stall are a
direct result of the existance of negative (suction) p r e s s u r e in the wake of the
stalled aerofoil. The stronger the suction, the more negative the Ct peak.

The strength of the suction is closely related to the acuteness of the forced
deviation of the flow of air along the pressure side of the aerofoil as it
leaves the t r a i l i n g edge[4]. The strength of the suction can be reduced by smoo-
thing the flow around the t r a i l i n g edge. The t r a i l i n g edge of tbe G~ 420 is more
acutely angled than that of the N A C A 0012 for positive angles of attack, and the

larger n e g a t i v e Ct after the stall is the result. The author built a VAWT using
blades h a v i n g the W o r t m a n n F X 6 3 - 1 3 7 section, w h i c h strongly encourages the

formation of a separation bubble, but w h i c h also has a very sharply angled

trailing edge (see figure i0). The m a c h i n e w o u l d not self start and exhibited a
noisy stall from p o s i t i v e angles of attack, a c c o m p a n i e d by a sudden deceleration.
In c o m p a r i s o n the stall of a similar m a c h i n e with N A C A 0012 blades was almost
silent. Figure ii shows E as a function of tip speed ratio computed for two
turbines that should self start, and for a conventional N A C A 0012 VAWT.

5. A S S Y M E T R I C BLADE SECTIONS AND VAWT E F F I C I E N C Y

The question arises as to what effect the use of an assymetric section has on
the p e r f o r m a n c e of a VAWT. As the power that can be e x t r a c t e d from the w i n d is a
function of V 3, even a small difference b e t w e e n v e l o c i t i e s experienced by the
u p s t r e a m and d o w n s t r e a m blades will have a significant effect on the relative
amounts of power available to each side. The v o r t e x flow model described by
C.J. Duremherg[2] gives the v e l o c i t y difference across the three bladed turbine
it analyses as about 0.~5. That is, the d o w n s t r e a m blades receive w i n d at a
speed of 0.55 of that on the u p s t r e a m side. (This number was estimated by the
author from D u r e m b e r g ' s Figure 4.) The d o w n s t r e a m blades t h e n have available
0.553 or 0.17 of the p o w e r at the u p s t r e a m side. By v i r t u e of running in slower
air the blades on the d o w n s t r e a m side will be o p e r a t i n g at a higher tip speed
374

ratio, in this case about twice that of the upstres~n blades. If the upstr<~am

blades are operating at or near maximum efficiency, then the downstres~i I~i~des
will be considerably less efficient. Ntunbers are hard to determine in this ~:a~e
but the author calculates that the downstream blad(~s will produce betweer ',i'
and 10% of the power of the upstre~n blades. This is probably an e~treme ~ ~
as the three bladed turbine of the model is inefficient, having a maximm~i (J'~ ol~
0.25. If however a blade section is used that preferentiaY]y extracts power ~r~
the upwind side, then this effect wi]l be heightened, and not to t,he d e t ~ i m ~ t
of the turbine as there would be lJttie ~ensib]~ power o~ the d o w n s t r ~ i s:,<~ <
utilize.

6. CONCLUSIONS
In theory and in practice it has been found possible to construct VAWT's in
such a way that they self start. The features to aid self starting that are
referred to in this paper are not mutually exclusive and could be combinea i~
one machine. Such a turbine would use blades having sections with good low
Reynolds number performance, and these would be assisted by some degree of
change to the turbine geometry.

ACKNOWLEDGEMENTS

I should like to thank my supervisor Mr. V.A.L. Chasteau for his help with ~ne
thesis from which this paper is extracted. I should also like ~o thank New

Zealand Electricity for their financial suppor~ of this paper.

REFERENCES
i. J.R. Baker, The self starting of vertical axis wind turbines, Masters ffhesis
in Mechanical Engineering at the University of Auckland, (1982)
2. C.J. Duremberg, Unsteady aerodynamics of vertical axis wind ~urbines, ]s~
British Wind Energy Association Workshop paper, (1979)
3. D. Gault, A correlation of low speed airfoil section stalling characteristic
tics with Reynolds number and airfoil geometry, NACA TN 396~, (1957
4. S.F. Hoerner, Fluid dynamic lift, published by the author (1975]
5- Correspondence between the author and Peter Holdsworth, Te Karaka, ]isbo~'ne
6. F.W. Riegels, Aerofoil sections, London, Butterworths ( 1 9 6 1
7. D.J. Sharpe, of the Mechanical, Aeronautical and Production Engineering
School of Kingston Polytechnic U.K. supplied ~his data
8. G.R. Watson, The self starting capabilities of low solidity fixed pitc~
Darrieus rotors, ist British Wind Energy Association Workshop paper, (1979)
 375

va

V COS
/

Vsin
I Vd

Figure 1 Turbine geometry

O,Z~O
Thrust coefficient Ct
0,35

,0,30

0,25
, U n s t a U e d positive Ct region
:o.~o j ~,,,,+. ~, A'+='"
High angle of

.O,lO ~ f /
, /~ / Stalled negative Ct / /

-0,05
Angle of attack oc
• lo'.2o" ~o'. ~.o'.5o'. 6 ~ . 7.o'. q¢_ ~ ' . 1 ~ ' . Uo" 12o" 13.o" 1~o" 1so'.16.o'. 17o".

Figure 2 Sharpe's compilation of Ct(~ ) for the NACA 0012

376

.ls0. =0.0

• Angle of a t t a c k 0,5
(x
90" /

Ct is negative at the $ ",4~,

angles of attack in these bands

.-90"

-~50" Azimuthal angle

60" 120" 1E 240" 3O0"

Figure 3 Oc((~) for various tip speed ratios

Net energy per cycle E

(arbitrary units)

~ '~-The dead band /

-1

Tip speed ratio

0.4 0,B 1,2 1.6 2.0
i i

Figure /, E(~) highlighting the dead band

 377

-8o"

-Angle of attack oc

,60"- - - -~- . . . . .

Et is negafive for oc
between 1/,,° and 60* e=°"

20" A

Figure 5 cx.(~) for various angles of tilt e, ~=1,5

Net energy per cycle E

=60"

e:~o °

-I =0 °
Tip speed rafio ~9
0,/~ 0,8 t2 1,6 2,0 2,4

Figure 6 E(/~) for various angles of tilt

378

Thrust coefficient
Ct

Symbot Reynotds number

o 40,000
n 80,000
<> 160,000

Angte of attack o<

4O" 6O" 8O" ioo" 12o" I~.O. 16o"

Figure 7 Ct(o~) for yaw=0, NACA 0012

Thrust coefficient
Ct

--0,02 Symbot Reynolds number

o 40,000
-0,04 u 80,000
o 160,000
O

-0,06¢ 2O. 4~ 6O.

Angle of attacko~
80= 10O' 12~ 14~ 16~ 18~

Figure 8 Ct(c~) for yaw=30", NACA 0012

 379

0.3 Thrust coefficient Ct

O'j~ NACA0012
\

Ang|e of ati~ack oc
60" lZO" 180" 2~,0~ 300'

Figure 9 Ct (~-1 for NACA 0012 and G~J ~20

C_ NAEA 0012

5E ~20

FX 63-137

Figure 10
380

-20

Net energy per cycle

16 E

(arbitrary units)

12 5iJ 420

NACA 0012
e=60 ~

NACA 0012

-4 Tip speed ratio

0,~. 0,8 1,2 1,6 2,0 2,4

Figure 11 E(~) for various configurc~itions