368 J. ENERGY VOL. 2, NO.

Fluid Dynamics of Diffuser-Augmented Wind Turbines

Barry L. Gilbert,* Richard A. Oman,t and Kenneth M. Foreman J
Grumman Aerospace Corporation, Bethpage, N. Y.

The diffuser-augmented wind turbine (DAWT) is one of the advanced concepts being investigated to improve
the economics of wind energy conversion systems (WECS). Application of modern boundary-layer control
techniques has reduced the surface area requirements of an efficient diffuser by an order of magnitude. Many
parameters that affect the performance of the diffuser system have been examined in small-scale wind tunnel
tests with a family of compact diffusers, using screens and centerbodies to simulate the presence of a turbine.
Flowfield surveys, overall performance, the effect of ground proximity, and the prospects for further im-
Downloaded by Universitats- und Landesbibliothek Dusseldorf on September 14, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.47988

provement are described. The baseline configuration is a conical, 60 deg included angle diffuser with an area
ratio of 2.78 controlled by two tangential injection slots. This first-generation DAWT can provide about twice
the power of a conventional WECS with the same turbine diameter and wind. Economic estimates show that this
DAWT can be as much as 50% cheaper than conventional WECS for the same rated power.

Nomenclature that are reliably free of extensive flow separation, and

Cp — exit static-pressure coefficient = (P4 — P0) /q0 maintain effective performance characteristics, but possess
CT = screen factor (turbine disk-loading coefficient) = H/q2 divergence included angles much greater than the con-
H = total pressure drop across the turbine (screen) ventional 6 to 12 deg. The problem is to prevent boundary-
Pj = static pressure at / layer separation in very strong adverse pressure gradients.
q{ — dynamic pressure at / One way of accomplishing this is by the tangential injection
r = augmentation ratio of high-energy air available from the external wind (see Fig.
V- = mean velocity at / 2). The additional momentum supplied to the boundary-layer
fluid helps the main flow overcome the severe adverse
Subscripts
pressure gradient and frictional losses in the wall region. If
0 = quantities evaluated or normalized far upstream of
the turbine
2 = quantities at the entrance to the turbine
4 = quantities at the diffuser exit
Introduction

T HE major unresolved technical problem in the design of

conventional wind energy conversion system (WECS)
turbines for very large power ratings is in the blade dynamics
of large diameter rotors. ] Because of this fact, integration of
wind generators into a national or regional power grid is
inhibited by the unacceptable reliability of very large units or
the economic liability of many smaller units of comparable
total power output. Many of these capital and performance
Fig. 1 Conceptual field installations of a DAWT with 6-m-diameter
restrictions of WECS may be reduced or eliminated by en-
turbine.
closing the wind turbine rotor in a suitably shaped duct.
The energy available to a horizontal-axis turbine rotor of
fixed size is increased by installing it at the entrance to a STATION NUMBER
0 123 4
diffuser (see Fig. 1). By recovering exhaust kinetic energy, the I
diffuser produces a greatly reduced pressure behind the
turbine relative to that behind a free turbine. This causes more
mass to flow through the diffuser-augmented wind turbine INLET
(DAWT) with at least as much pressure change as across the SLOT
conventional turbine. Because the wind power available to the
turbine is the product of flow rate and pressure drop, much
greater power output is possible than for a conventional BOUNDARY L A Y E R
WECS at the same turbine size and free wind speed. LOSING MOMENTUM
The diffuser design must have the smallest possible RE-ENERGIZED
BOUNDARY
structural cost, while still maintaining an appreciable LAYER
subatmospheric pressure at the turbine exit plane and a large
pressure recovery within the diffuser. This requires diffusers
Received July 10, 1978. Copyright © American Institute of
Aeronautics and Astronautics, Inc., 1978. All rights reserved.
Index categories: Wind Power; Subsonic Flow; Aerospace
Technology Utilization.
*Research Scientist, Fluid Dynamics Laboratory, Research ENLARGED VIEW
Department. UNDISTURBED OF B O U N D A R Y
fDirector, Fluid Dynamics Laboratory. AIRFLOW LAYER CONTROL FLOWS
JStaff Scientist, Fluid Dynamics Laboratory, Research Depart- Fig. 2 Flow streamlines (upper half) and velocity profiles (lower
ment. half)inaBLC DAWT.
 NOV.-DEC. 1978 FLUID DYNAMICS OF DIFFUSER-AUGMENTED W I N D TURBINES 369

properly done, this technique has the effect of preventing flow FLOW R E T A R D A T I O N
UPSTREAM
separation from the wall (the primary cause of failure of large
angle diffusers). There are several aspects of modern TOTAL PRESSURE
12.5 'LOSS
technology that offer significantly greater advantages to these
shrouded systems than have been considered in the past. 2 The 10
INLET
most important of these are: 1) the promise of boundary-layer (ACCELERATION) i 7 5
-
control to reduce the surface area (i.e. length) requirements of
an efficient diffuser by an order of magnitude; and 2) the i 5
realization (first indicated by Igra 3 and since confirmed by ! 2.5
our experiments) that there is a beneficial augmentation of a
shrouded wind turbine that results from the turbulent -2-1 2 3 4 5 1 3
momentum interaction between the turbine/diffuser wake A X I A L DISTANCE F R O M A X I A L DISTANCE FROM
and the accelerated peripheral flow. This momentum ex- SCREEN (DIFFUSER SCREEN (DIFFUSER
change results in a greatly reduced exit plane pressure that has LENGTHS) LENGTHS

been as beneficial as the effect of the diffuser itself. There is Fig. 3 Typical centerline axial static- and total-pressure surveys in
also the promise of synergistic advantages derived from the 60-deg included angle diffuser, CT = 0.47.
design features of a static diffuser structure which dampens
Downloaded by Universitats- und Landesbibliothek Dusseldorf on September 14, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.47988

gusts, reduces wind shear and directional problems, lowers

cut-in wind speed by increasing the axial velocity at the
turbine (thereby increasing the annual on-line factor), and
eliminates tower wakes. Finally, an increasing concern for the
cost and development risk associated with very large rotors
leads to significant advantages for a system that produces
larger rated power from a rotor that is limited in size by
aeroelastic design constraints.
The overriding criterion for evaluating diffuser con-
figurations is the cost of the energy they would produce in
production quantities. This is reflected primarily in the capital
cost comparison for competing concepts, with secondary a)
influences from other cost factors such as operating wind
range, annual effective power production, aesthetics, and
maintenance requirements.Cost estimates for components of
both the DAWT and conventional systems in production
quantities are, unfortunately, but necessarily, highly
speculative at this time. An attempt has been made to use the
best available data 4 ' 5 and to use estimating methods that are
easily adjusted for changes in cost data. Cost and practicality
projections are the major factors that determine the selection
and sequence of the DAWT configurations tested and
analyzed.
The complete program is given in Ref. 6. Although in-
dependently derived, many of the technical decisions con-
cerning the diffuser design closely parallel the work of Igra. 3
This paper is limited primarily to the experimental work
conducted on boundary-layer control (BLC) conical diffusers.
A complete economic analysis may be found in Ref. 6.

Experimental Method
Experiments have been conducted in a low-speed low- Fig. 4 Boundary-layer controlled (BLC) diffuser models: a) (left to
turbulence-level free jet tunnel facility. The core region of a right) 40-, 60-, and 80-deg included angle models each with an area
29.2-cm-diameter free jet flow is used as the test section in ratio of 2.78; b) baseline configuration, 60-deg diffuser with an area
which uniform wind conditions are simulated. A velocity of ratio of 2.78 (screen installed CT = 0.37).
13.0 m/s (43fps) was used for the tests. Models were mounted
slightly downstream of the nozzle exit plane.
Since it is impractical to build a family of wind turbines for derived from an ideal turbine, 0.593 of the total available free
an exploratory investigation of small-scale diffuser models, wind power in a streamtube with the same area as the turbine.
turbine energy extraction was simulated by screens that As given in Ref. 6,
dissipate energy at the turbine station. The turbine per-
formance is simulated by the total pressure change across the r=HV2/q0V0 (0.593}
screen. The power extracted per unit area of screen is the
product of the change in total pressure and local velocity. where His the total pressure drop across the turbine (screen).
The instrumentation system employed a transconductance The magnitude of H is a function of the particular screen
differential pressure transducer. Measurements were derived used and the local dynamic pressure and is measured directly
from the combined use of a single static-pressure probe and a in the system. Figure 3 shows the standard technique used to
single total-pressure probe, each of which could pass through determine the system augmentation. The local dynamic
the screens used. For the axial- and radial-pressure surveys, pressure q2 is <?inlet + q0. The screen factor, C r , is computed as
the probes were mounted on a motor-driven traversing — H/q2. Measurements of static, dynamic, and total pressure
mechanism. This device produced an electrical output usually were made. Repeated axial surveys at different radial
proportional to its position so that pressure vs spatial position positions indicate that r increases with radial distance from
could be recorded. the axis. The weighted average over the entire cross section of
The augmentation ratio is defined as the ideal power ex- the turbine simulator gives the total diffuser augmentation
tracted by the turbine, normalized by the optimum power ratio, f.
 370 GILBERT, OMAN, AND FOREMAN J. ENERGY

Table 1 Diffuser configuration parameters that have been tested

Diffuser Additional
Number of Area ratio (diffuser exit/ included Disk tests below
segments screen inlet) angles, deg loading, CT footnoted
1 4.94 40,90 0.47 c
2.55,1.93 40 0.47 c
2.78,2.05,1.93,1.85,1.63 60 0.47
1.69 80 0.47

2 4.94 90 0.47 c
3.92 40 0.47
2.78 40 0.47 c —e
2.78 60 0.37,0.47, b,f-j
0.55,0.64,
0.93
2.78 80 0.37,0.47, b,g,h,j
0.55,0.64,
Downloaded by Universitats- und Landesbibliothek Dusseldorf on September 14, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.47988

0.93
2.49 40,60,80 0.47 a,c
2.37 60 0.47 a
1.28 40 0.47

3 4.94 40 0.47,0.55, b,d

4.43 40 0.64 a
4.06 40,60 b

4 4.94 40 0.47,0.55, b
0.64
4.94 60 0.47 b
4.43 40 0.47 b
a
Inlet slot height/radius,%,0,1.25,2.5,5.
b
Auxiliary slot height/local radius,%,3.36,2.52,1.92,1.76,1.68,1.63,0.36.
c
Positioning screen at diffuser inlet with and without screen ring.
Internal turbulent tape trips.
e
Trailing-edge flaps.
f
Leading- and trailing-edge configurations.
g
Ground plane and/or shear flow study.
h
Position of auxiliary slots.
'Centerbody configurations.
J
Yaw sensitivity.

Figure 4 shows two BLC diffusers. The upper photo shows Diffuser
three diffusers with included angles of 40, 60, and 80 deg and Many parameters have been identified that influence the
the same exit-to-turbine disk ratio (2.78) and slot sizes. The diffuser system performance. Investigations were conducted
40-deg diffuser is of an earlier construction employing rolled to determine the effect of several of these. Although it was
stainless steel sheet, and the other two are spun aluminum. impossible to span the entire parameter space for each
The decrease in material from 40 to 80 deg is evident. The configuration, data were compiled that give an indication of
second photograph is a three-quarter view of the 60-deg the important trends.
diffuser showing the mounting details of the screen ring, inlet, Over 100 variations of wide-angle segmented diffusers that
and auxiliary slots. employed the boundary-layer control mechanism were
constructed and tested. A partial summary of these is given in
Experimental Results Table 1. A 60-deg included angle diffuser with a screen factor
The system augmentation is primarily a result of the in- (disk loading) of 0.63 is the best configuration tested, giving a
creased mass flow rate through the turbine. This increase in mass-weighted average-augmentation ratio of 1.89. This
entrance velocity may be thought of as the combination of configuration has an area ratio of 2.78 and uses an inlet and
two major pressure effects. The first of these is the static- auxiliary slot for boundary-layer control. It is expected that
pressure rise in the diffuser. The efficiency of this element is with further model refinements it will be possible to improve
dependent upon such factors as the inlet configuration, upon this value.
number, and location of boundary-layer energizing slots, size The first important result was learning how to start the flow
of the slots, and turbine disk loading. along the diffuser wall against the large adverse pressure
The second effect is due to the accelerated flow around the gradient. If the flow in the diffuser was not properly started,
diffuser structure producing an exit plane static-pressure it behaved like a jet issuing from a plane wall; i.e. a vena
reduction. The importance of this effect may be emphasized contracta flow. The solution was simply to move the screen up
by realizing that a subatmospheric exit pressure requires that into a constant area section upstream of the diverging section,
the exiting fluid continues to expand to the far downstream and to provide an initial inlet slot around the screen section
conditions. The corresponding further rise in static pressure is which added enough momentum to the boundary layer to
accomplished without the need (and cost) of constructing allow it to start turning the angle of the diffuser wall as if it
more diffuser structures. For 60 and 80 deg, typical values for were a Coanda flow process. It appeared to be necessary to
the exit pressure coefficient are -0.58 and -0.63, respec- provide a short attachment surface of little or no divergence
tively. However, a DAWT receives the net benefit from the downstream of the screen prior to where the diffuser wall
diffuser of only that pressure reduction below the equivalent began to rapidly diverge or else jet flow would form.
CP4 value for a conventional rotor of -1/3—that is, Figure 5 shows the augmentation obtained with various
something like -0.25 to -0.3. angle diffuser configurations as a function of their area ratio.
 NOV.-DEC. 1978 FLUID DYNAMICS OF DIFFUSER-AUGMENTED WIND TURBINES 371

CT = 0.47

• 40° DIFFUSER
• 60° DIFFUSER 40° - 2 SEGMENTED DIFFUSER
30° - 2 SEGMENTED D I F F U S E R

3 4 0.6 0.7 0.8

EXIT A R E A / T U R B I N E A R E A (A 4 /A 2 ) DISK LOADING, C T
Downloaded by Universitats- und Landesbibliothek Dusseldorf on September 14, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.47988

Fig. 5 Trend of centerline augmentation with area ratio for diffusers Fig. 6 Average induced dynamic pressure ratio for 60- and 80-deg
constructed of various numbers of segments. BLC diffusers at various disk loadings.

These models were constructed from a different number of

sections, each with a slot between them. It is apparent that the + 80° BLC DIFFUSER
maximum augmentation seems to be at about an area ratio of • 60° BLC DIFFUSER
three. Curves are drawn through the experimental data to • EMPIRICAL OPTIMUM 60°
A EMPIRICAL OPTIMUM 80°
indicate trends. §1-8
Increasing the area ratio beyond three always leads to a
degradation of performance because the flow separates from £1.7
01
the diffuser. In some cases, the separation point actually 5

moves upstream when sections with increased wall area are 6 1.6

added to the rear of the diffuser. Larger area ratios may be

S1.5
attainable, but they would require better implementation of
CE
the slot flow for boundary-layer control.
SSl-4
For an area ratio of 2.78, 60- and 80-deg included angle
diffusers were tested with various simulated turbine disk 1.3
loadings. This was accomplished by changing the screen
solidity in the same diffuser configuration. An increase in 0 .4 .5 .6 .7 .8 .9
turbine pressure drop usually causes the injection slots to LOCAL DISK LOADING, Cj
perform better because the increased pressure differential Fig. 7 Dependence of augmentation on disk loading.
drives more fluid through each slot. However, an increase in
screen blockage appears as a greater resistance to the in-
coming fluid, so there is a corresponding decrease in mass reduce the inlet slot substantially before the decrease in system
flow through the screen. augmentation is no longer compensated for by the increase in
Figure 6 shows the dynamic pressure ratio induced up- turbine size.
stream of the screen simulators at various turbine loadings. In Decreasing the turbine size with respect to the diffuser
this range of loadings, the functional relationship is nearly increases the system augmentation. The inlet slot fluid is not
linear. Figure 7 shows the augmentation ratio calculated from only used to energize the boundary layer so that it may
the measured average velocities in 60- and 80-deg diffusers. negotiate the starting turn into the diffuser but the inlet then
The empirical prediction for the optimal disk loading coef- also behaves like an ejector, with high energy fluid in the
ficient is in very good agreement with the loading predicted by- peripheral slot drawing more fluid through the turbine. If
ideal one-dimensional momentum theory. 6 Slightly higher increasing the diffuser size is less expensive than increasing
values are obtained from the momentum theory because the the turbine diameter, promoting the ejector system may
flow is not one dimensional and is frictional. become more economically credible.
This departure is larger for the 80-deg than for the 60-deg A series of tests were conducted using 60-deg included angle
diffusers. Since this is primarily a Reynolds number effect, it diffusers to determine the effectiveness of the slots. When a
is expected that there would be closer agreement between no-slot configuration was compared to the same diffuser with
experiment and theory for larger scale models. After the an inlet slot area that was 19% of the total diffuser inlet area,
relatively flat range of maximum augmentation, there is a loss an increase of 24% in augmentation was found (note that the
of augmentation with an increase in freestream disk loading increase in inlet slot was at the expense of turbine area).
because the induced velocity into the turbine decreases faster Figure 8 shows the augmentation possible from single-
than the increases in the energy extraction by the turbine. segment diffusers as a function of area ratio. The trend seems
There are several conclusions that may be made regarding to indicate the best performance near an area ratio of three.
the number, size and arrangement of diffuser segments, and This is partially a result of the base-pressure effect, still to be
the characteristics of inlet and auxiliary slots. Decreasing the discussed, and also the failure of the fluid to remain attached
initial inlet slot by increasing the screen diameter in the same to the diffuser walls for longer distances without aid.
diffuser configurations is equivalent to increasing the ap- For an area ratio of 2.78, the addition of a single auxiliary
parent blockage which produces a slight decrease in the mass slot seems to have a minor influence as shown in Fig. 9.
flow. However, the induced velocity, and therefore available Moving the slot forward (from 52% of diffuser axial length to
local power density, increases towards the periphery. As a 39%) increases the augmentation 15%. Further forward
result, the decrease in system augmentation due to the reduced movement (to 30% length) again decreases the augmentation
slot size was nearly compensated for by the increase in turbine because the second section becomes too long for the re-
area, such that the total available energy to the turbine energized fluid to remain attached. In this case, to maintain
remained nearly constant. It appears possible, therefore, to the same length, a second auxiliary slot is indicated. Alter-
 372 GILBERT, OMAN, AND FOREMAN J. ENERGY

2.0 0.64r 80° BLC DIFFUSER

60° BLC DIFFUSER
INLET AND 40° BLC DIFFUSER
ONE A U X I L I A R Y CT = 0.47
SLOT
D CT = 0.55
O C = 0.47
1.6-

INLET SLOT ONLY

1.2

1.0

1.5 1.7 1.9 2.1 2.3 2.5 2.7 2.9

3 4 5
EXIT A R E A / T U R B I N E A R E A (A 4 /A 2 )
EXIT A R E A / T U R B I N E A R E A (A 4 /A 2 )
Downloaded by Universitats- und Landesbibliothek Dusseldorf on September 14, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.47988

Fig. 8 Effect of slot configuration on average augmentation in 60- Fig. 10 Exit plane pressure reduction for various BLC diffuser
deg BLC diffusers. angles as a function of area ratio.

q_

_: 0.641- • 80° BLC DIFFUSER

2 SEGMENTED A 60° BLC DIFFUSER
60° D I F F U S E R S • 40° BLC DIFFUSER
CT-0.47 60° & 8 0 ° - 2 SEGMENTED

o = 2.78
o
§1.61-
0.56-

o 0.52-

INLET SLOT
ONLY
0.48-

1.2

NO SLOTS X 0.44-

DC 0.40-

0.4 0.5 0.6 0.7 0.8 0.9 1.0

0.00 0.20 0.40 0.60 0.80 1.00
" DISK LOADING COEFFICIENT, CT
SLOT STATION
D I V E R G I N G SECTION LENGTH Fig. 11 Exit plane pressure reduction for various BLC diffuser
Fig. 9 Effect of the auxiliary slot position and length on augmen- angles as a function of disk loadings.
tation (length of vector symbol, from inlet to exit, is proportional to
slot lengt^).
profile independent of the specific diffuser performance
details. The total system augmentation is due to the diffuser
natively, the diffuser could be made shorter. Also shown are characteristics and also (and in some cases, more so) to the
two cases designed to test the effect of the slot length, that is, base pressure reduction.
the loss of energy in the slot. For virtually no length of slot, Initially, it was expected that CP4 would be a function of
there was an 8% increase in augmentation compared to a 5% the diffuser area ratio. For very small area ratios, as seen in
loss in the exceptionally long slot. In a few cases tested, there Fig. 10, this seems to be the case. However, for a particular
did not seem to be too much effect resulting from the change diffuser angle and disk loading, an area ratio is reached where
in slot height. Cp remains essentially constant until at very large area
No general statement may be made regarding the "correct" ratios, the Cp effect actually decreases due to a substantial
number of segments. Diffusers with the same angle and area separated region. The area ratio and value where Cp becomes
ratio but constructed from a different number of segments asymptotic is dependent on diffuser angle. Smaller angle
were compared. For some area ratios, the best performance diffusers have asymptotic lower subatmospheric pressures at
was obtained with two segments, whereas an increase to three larger area ratios than larger angle diffusers.
resulted in degraded augmentation. At larger area ratios, When the value of Cp is compared to the turbine (screen)
three sections seemed best with poorer performance at both disk loading coefficient for a particular area ratio (Fig. 11) a
two and four sections. The optimum configuration for given strong dependence is found. The disk loading which gives the
angle and area ratio will probably be determined as a tradeoff lowest base pressure seems to be a function of the con-
between the number of sections and the exact positioning of figuration employed. The value of loading that gives the best
auxiliary slots. base pressure for a given diffuser does not seem to be too
different from that found to give the best system per-
Exit Plane Pressure Reduction formance, thereby further emphasizing the benefits of CP4 .
The exit plane static-pressure coefficient, CP4, is a very
important parameter in the system optimization. The exit Ground Plane Study
plane subatmospheric pressure provides additional system A proposed design concept calls for a three-point
augmentation by translating the entire diffuser pressure suspension where the diffuser exit lower lip is supported from
 NOV.-DEC. 1978 FLUID DYNAMICS OF DIFFUSER-AUGMENTED WIND TURBINES 373

conical forebodies). Flow visualization techniques indicate

5 1.6 USING KAMAN TURBINE
that the fluid remained firmly attached to the centerbody and
1 4 -DAWT MORE
COST TRENDS quickly flowed in behind it. This produced only a very small
I -
Q COSTLY wake region which was undetectable by exit plane pressure
r 1.2 USING GE T U R B I N E surveys. However, at the next tested centerbody size (area
COST T R E N D S
A ratio of about 12%) with short afterbodies, the flow reverted
to a highly transitory flow. By increasing the afterbody
0.8
DAWT LESS
length, or using a conical forebody, the severely oscillating
.COSTLY flow becomes a stable annular jet moving along the diffuser
wall producing a large diverging conical wake.
WECS T U R B I N E DIAMETER, M
i 0.4
10 20 30 40 50 60 70 80 90
Two other separated flow configurations were achieved for
.0'—— the 8% centerbody. When a rounded forebody was combined
0 10 20 -30 40 50 60 70 with a conical afterbody, the flow separated from the af-
a: DAWTTURBINE DIAMETER, M
terbody and produced a 7% decrease in augmentation. When
Fig. 12 Cost comparison of DAWT (turbine and diffuser) and
conventional WECS for equal rated power.
a conical forebody was combined with a rounded afterbody,
the flow separated from the forebody and did not reattach
itself, but continued to diverge for a 65% decrease in local
Downloaded by Universitats- und Landesbibliothek Dusseldorf on September 14, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.47988

the ground to reduce tower and structural cost. It is very augmentation.

important that the proximity to the ground does not degrade The presence of a centerbody requires that the centerline
the system performance. With the trailing edge of the diffuser mass flow be diverted into a larger annular ring. For the
on the ground, the upper portion of the diffuser showed a smaller bodies, the mass flow at the test location remains
very slight increase in mass flow when compared to the same essentially unchanged, implying no loss of augmentation.
diffuser configuration with no ground plane. However, the However, there will be a larger mass flow at the radius of the
lower section, nearer to the ground, exhibited substantial centerbody than without a body. For larger diameter bodies,
improvement, with augmentation increased by 10 to 14%. there is a loss of diffuser performance manifested as a
Even when the exit was half a turbine diameter above the decrease in flow rate. This interpretation must be examined
ground, the lower section still showed a 6% better more fully by complete radial surveys.
augmentation than for the no-ground plane case. Two final points are important. In the system con-
This increase may be due to the distortion of the exit plane figurations that have been found to give the best per-
static-pressure profile. However, no evidence was apparent formance, flow visualization techniques indicate three distinct
from exit plane static-pressure surveys. The important point, conical regions of low-velocity fluid. These may be traced
which should be emphasized, is that there is a slight increase directly to the pins that support the center screen ring. It is
in DAWT system performance in the presence of a ground unclear, however, whether the losses are inherent in the
plane. system and simply manifest themselves at the weakest point or
The other important consequence of the ground is that the whether they are purely a cylindrical wake effect. If the latter,
wind velocity profile far upstream of the system is not streamlining the support rods in larger scale models should
uniform. An atmospheric boundary layer was simulated in the improve the situation.
jet facility by a honeycomb mesh cut with a linear profile to The other point is that in all cases tested there did not seem
produce a nearly linear variation in dynamic pressure. The to be any indication of severe diffuser flow variations with
turbine disk intercepted a velocity profile which far upstream small changes in either pitch or yaw. It is expected that as the
varied by 14%. The augmentation calculations were based inlet to the turbine as well as the diffuser are improved, there
upon the integrated available power in this intercepted disk. should be even less tendency for yaw-induced distortion.
In the presence of the shear flow, there was a very slight
decrease of augmentation. At all measured local positions, Closure
there was a loss of about 1% of the augmentation compared The work reported here with small-scale DAWT con-
to the same configuration mounted in a free jet flow. figurations has been extended to larger scale. This tested
A test of the combination profile of the ground plane and scaling laws by increasing the Reynolds number by a factor of
shear flow results in a flowfield that is a simple superposition 35, but still used the screen concept to simulate the turbine.
of the individual profiles. Again, the upper part of the dif- Several slot-scaling philosophies were evaluated to gain a
fuser showed little effect, while the lower section indicated better understanding of slot flow as it enters a large adverse
about a 7.5% increase in augmentation when based upon an pressure gradient. Preliminary examination of these data
integrated value of q0. The local augmentation was about indicates that the results and conclusions from the small-scale
11% greater than the free jet diffuser augmentation. The tests are reproducible; however, the finished data are not
important result, however, is that there appears to be no ready at this time.
decrease in the augmentation effect of the DAWT in the Future work includes replacing the screen simulator with a
proximity to the ground. real turbine. We have decided to use a fixed-pitch three-
Centerbody
bladed turbine with relatively smooth stall characteristics.
Additional computer analysis by the method of singularities
Another important aspect of a real turbine geometry is the of the base pressure effect is planned. A theoretical model for
presence of a centerbody. A performance comparison was the boundary-layer control mechanism for wide-angle dif-
made with and without centerbodies at a single representative fusers is also being developed.
point (at 60% of the radius). Seven different centerbodies
were tested ranging from an area of centerbody to area of
turbine of 1.9 to 30.9%. Two forebody designs were used, a Economics
conical and a rounded shape, each 1.25 cm long and of the The dominant factor in the evaluation of DAWT con-
same diameter as the afterbody. The afterbodies were circular figurations is the cost of the energy produced. The DAWT can
cylinders of various lengths from 1.25 to 10 cm and all tests be considered economically viable only when the diffuser
were conducted in a 60-deg two-segment diffuser with a 6.3- component costs, in production quantities (~100/year), are
mesh/cm screen. offset by improved productivity, such that the overall elec-
When the area of the centerbody was less than about 8% of trical generation system cost is reduced. While the compact
the total disk area, a very slight degradation of performance baseline diffuser design produced only a fraction of the
was apparent (less for longer centerbodies and more for augmentation believed possible with a much larger diffuser, it
 374 GILBERT, OMAN, AND FOREMAN J. ENERGY

presents a more cost effective solution for wind energy benefits of DAWT in the intermediate size range are
conversion. The diffuser cost estimates were made by somewhat obscured by the significant uncertainties of current
Grumman on the basis of a finite-element structural analysis turbine cost estimates. The indirect benefits of DAWT's,
for a full-scale baseline diffuser design (2.78 area ratio, 6 = 60 including a potentially greater factor for usable annual wind
cleg, r— 1.89). The total cost of the diffuser and of the rotor energy patterns, tower cost reduction, elimination of tower
elements has been normalized by the turbine diameter squared wake, and reduction of wind shear and direction problems,
to facilitate comparison of the two systems. probably mean that DAWT's have the possibility of becoming
At this time, operationally reliable rotor diameters greater more economical than conventional WECS, regardless of size
than 65 m are highly speculative. The most straightforward or turbine costing trends.
advantage of the DAWT is that it offers an alternative to
replication of conventional WECS when the desired unit
output would require these large turbine rotors. The cost
Acknowledgments
comparison is shown in Fig. 12 for equal rated power output This work was performed under U.S. Energy Research and
on the basis of the turbine and diffuser elements of the system Development Administration (ERDA) Contract E(l 1-1) 2616.
only. That is, differences in the costs of the electrical
generator, tower, foundation, control system, and References
mechanical shaft linkage or transmission system components 1
Savino, J.M. (ed.), Proceedings, NSF-NASA Workshop on Wind
Downloaded by Universitats- und Landesbibliothek Dusseldorf on September 14, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.47988

have not been considered in these cost comparisons, nor have Driven Generator Systems, Rept. NSF/RA/W-73-006, Dec. 1973.
any other system economics of scale. 2
Betz, A., "Energiemsetzungen in Venturidusen," Natur-
The DAWT can be as much 50% cheaper than a WECS for wissenshaften, Vol. 10, S. 3, 1939, pp. 160-164.
3
turbine diameters greater than about 35 m. For smaller Igra, O., "Shrouds for Aerogenerators," Dept. of Mechanical
turbine sizes, the DAWT can be anywhere from marginally Engineering, Ben Gurion University of Negev, Israel, Rept. 2, March
cheaper to significantly more expensive depending on which 1975.
4
cost trend one actually experiences. Thomas, R.L., "Large Experimental Wind Turbines - Where We
A usable annual wind pattern factor has not been included Are Now," NASA TMX-71890, March 1976.
5
"Wind Generator System, Final Design Review, NASA-LeRC
in the graphic results. But, because of the inherent wind Contract NAS 3-19404," Kaman Aerospace Corporation, July 1975.
speedup features of the diffuser section, there can result a 5- 6
Oman, R.A., Foreman, K.M., and Gilbert, B.L., "Investigation
50% greater annual power conversion for the DAWT com- of Diffuser-Augmented Wind Turbines, Parts I & II," ERDA Tech.
pared to the WECS. Therefore, there is a probable real Rept. COO-2616-2, Jan. 1977.
economic advantage for the DAWT, regardless of which
turbine cost estimate is appropriate.

Summary
The experimental fluid mechanics of boundary-layer
controlled compact diffusers were investigated for power STATEMENT OF OWNERSHIP, MANAGEMENT AND CIRCULATION
(Required by 39 U.S.C. 3685)

augmentation of wind turbines. By an innovative use of

external wind, the surface area of efficient diffusers has been
reduced by an order of magnitude, producing a cost-effective
WECS.
The major advantage of a DAWT advanced concept is in
very large power production units where conventional WECS
encounter severe blade dynamics problems. It is possible to
induce much larger amounts of air through a DAWT than
passes through a conventional WECS because of low exit
pressure and large diffuser pressure recovery. We have
already demonstrated that a DAWT can provide about twice
the power of a conventional WECS with the same turbine
diameter and wind speed.
Many parameters that affect the diffuser system per-
formance have been examined in small-scale wind tunnel tests
with a family of compact diffusers, using screens and cen-
terbodies to simulate the presence of a real turbine. The most
important results are the following.
1) A constant area section behind the turbine is necessary to
start the diffusion process.
2) Maximum augmentation occurs at a ratio of about 3.
3) The empirical maximum is in agreement with the one-
dimensional momentum theory prediction.
4) The optimum configuration will be a tradeoff between
cost and benefit of additional auxiliary slots for boundary-
layer control.
5) The DAWT system exhibits substantial improvement in
simulated ground effect.
6) There is a small degradation of performance with
realistic centerbodies.
Preliminary economic studies using and comparing the
most recent turbine component cost trends and Grumman-
generated diffuser cost estimates and performance
measurements indicate that DAWT should have lower specific
power costs than WECS for very large (MW ranges) and small
turbine diameter (<100 kW) sizes. The possible direct
 This article has been cited by:

1. Devon Jedamski, Kenneth D. VisserComputational Analysis of a Diffuser Using USM3D for Diffuser Augmented Wind
Turbines . [Citation] [PDF] [PDF Plus]
2. Costin Ioan Coşoiu, Andrei Mugur Georgescu, Mircea Degeratu, Dan Hlevca. 2013. Numerical predictions of the
flow around a profiled casing equipped with passive flow control devices. Journal of Wind Engineering and Industrial
Aerodynamics 114, 48-61. [CrossRef]
3. C I Coşoiu, A M Georgescu, M Degeratu, L Haşegan, D Hlevca. 2012. Device for passive flow control around vertical
axis marine turbine. IOP Conference Series: Earth and Environmental Science 15:6, 062031. [CrossRef]
4. Tudor Foote, Ramesh AgarwalOptimization of Power Generation from Shrouded Wind Turbines . [Citation] [PDF]
[PDF Plus]
5. Yuji Ohya, Takashi Karasudani, Akira Sakurai, Ken-ichi Abe, Masahiro Inoue. 2008. Development of a shrouded wind
turbine with a flanged diffuser. Journal of Wind Engineering and Industrial Aerodynamics 96:5, 524-539. [CrossRef]
Downloaded by Universitats- und Landesbibliothek Dusseldorf on September 14, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.47988

6. Dr Gerard J W van Bussel. 2007. The science of making more torque from wind: Diffuser experiments and theory
revisited. Journal of Physics: Conference Series 75, 012010. [CrossRef]
7. Yuji OHYA, Takashi KARASUDANI, Akira SAKURAI, Masahiro INOUE. 2006. Development of a High-Performance
Wind Turbine Equipped with a Brimmed Diffuser Shroud. TRANSACTIONS OF THE JAPAN SOCIETY FOR
AERONAUTICAL AND SPACE SCIENCES 49:163, 18-24. [CrossRef]
8. H. Grassmann, F. Bet, M. Ceschia, M.L. Ganis. 2004. On the physics of partially static turbines. Renewable Energy
29:4, 491-499. [CrossRef]
9. Yuji Ohya, Takashi Karasudani, Akira Sakurai, Masahiro Inoue. 2004. Development of High-Performance Wind Turbine
with a Brimmed-Diffuser-Part 2-. JOURNAL OF THE JAPAN SOCIETY FOR AERONAUTICAL AND SPACE
SCIENCES 52:604, 210-213. [CrossRef]
10. D.G. Phillips, P.J. Richards, R.G.J. Flay. 2002. CFD modelling and the development of the diffuser augmented wind
turbine. Wind and Structures 5:2_3_4, 267-276. [CrossRef]
11. Yuji Ohya, Takashi Karasudani, Akira Sakurai. 2002. Development of High-Performance Wind Turbine with Brimmed
Diffuser. JOURNAL OF THE JAPAN SOCIETY FOR AERONAUTICAL AND SPACE SCIENCES 50:587, 477-482.
[CrossRef]
12. M. O. L. Hansen, N. N. S�rensen, R. G. J. Flay. 2000. Effect of Placing a Diffuser around a Wind Turbine. Wind
Energy 3:4, 207-213. [CrossRef]
13. C.A.J. FLETCHER, A.J. HONAN, J.S. SAPUPPO. 1983. Aerodynamic platform comparison for jet-stream electricity
generation. Journal of Energy 7:1, 17-23. [Citation] [PDF] [PDF Plus]
14. Clive A.J. Fletcher. 1981. Computational analysis of diffuser-augmented wind turbines. Energy Conversion and
Management 21:3, 175-183. [CrossRef]
15. M.L. LOYD. 1980. Crosswind kite power (for large-scale wind power production). Journal of Energy 4:3, 106-111.
[Citation] [PDF] [PDF Plus]
16. B.L. GILBERT, K.M. FOREMAN. 1979. Experimental demonstration of the diffuser-augmented wind turbine concept.
Journal of Energy 3:4, 235-240. [Citation] [PDF] [PDF Plus]