Sie sind auf Seite 1von 9

Renewable Energy 35 (2010) 19201928

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Design and experimental validation of the inlet guide vane system of a mini
hydraulic bulb-turbine
L.M.C. Ferro a, b, *, L.M.C. Gato b, A.F.O. Falcao b
a
b

bal, Polytechnic Institute of Setu


bal, Campus do IPS, Estefanilha, 2910-761 Setu
bal, Portugal
Department of Mechanical Engineering, Escola Superior de Tecnologia de Setu
IDMEC, Instituto Superior Tecnico, Technical University of Lisbon, Av. Rovisco Pais,1049-001 Lisboa, Portugal

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 20 July 2009
Accepted 15 January 2010
Available online 2 March 2010

The paper presents a fast design method for the inlet guide vanes of low-cost mini hydraulic bulb
turbines. The guide vanes are positioned between two conical surfaces with a common vertex and have
constant thickness distribution, except close to the leading and the trailing edges. The conical-walled
inlet guide vane row is designed using a quasi-three-dimensional calculation method, by prescribing
the angular-momentum distribution along the span at the outlet section of the guide vanes.
The meridional through-ow is computed by a streamline curvature method and the blade-to-blade
ow by a singularity surface method. The stagger angle and the vane camber are computed to full
the required design circulation and zero-incidence ow at the leading edge. The nal vane shape is
a single-curvature surface with straight leading and trailing edges. To validate the design method,
a conical-walled inlet guide vane row nozzle-model with six xed vanes was designed, manufactured
and tested in an airow rig. Traversing measurements along the circumferential and radial directions
were made with a ve-hole probe.
The experimental results are compared with the prescribed design conditions and with numerical
results from the three-dimensional inviscid and viscous ow computed with the FLUENT code.
 2010 Elsevier Ltd. All rights reserved.

Keywords:
Axial ow turbine
Design method
Guide vane row
Singularity method
FLUENT code

1. Introduction
There is still a huge potential for power generation from miniscale hydro installations with outputs ranging between a few
kilowatts and 1 MW. This is mainly due to a coincidence of two
factors: the strategic interest in the use of renewable energy
sources and the recent advances in electronics which enables
output control of mini turbo-generators to follow the demand and
meet the (voltage and frequency) standards of the electrical
network at low cost. When available, mini-scale hydro power
systems represent a good technological solution to produce electrical power to disperse communities specially in less developed
regions of the world. The development and the installation of
locally manufactured hydro plants in those regions are appreciated
by the governments and bring further large social impact.
Also, in Europe [1], where the potential for large hydro power
schemes is almost exhausted, mini-hydro power has still a large

* Corresponding author. Department of Mechanical Engineering, Escola Superior


de Tecnologia de Setubal, Polytechnic Institute of Setubal, Campus do IPS, Estefanilha, 2910-761 Setubal, Portugal. Tel.: 351 265790000; fax: 351 265721869.
E-mail addresses: lferro@hidro1.ist.utl.pt (L.M.C. Ferro), luis.gato@ist.utl.pt (L.M.
C. Gato), falcao@hidro1.ist.utl.pt (A.F.O. Falcao).
0960-1481/$ see front matter  2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2010.01.020

potential to develop, which will allow EU small hydro industry to


increase. The remaining potential for small hydro power (SHP) in
2005 for EU, Switzerland and the candidate countries was estimated
to be 49 TWh/year [1]. The installed capacity of SHP in EU was
12.3 GW in 2007 (including 2.8 GW for mini-hydro power systems)
[2]. The gross production for the same year was 41.4 TWh (10.5 TWh
for mini-hydro systems).
The paper presents a fast design method for the inlet guide
vanes of low-cost mini hydraulic bulb turbines. The guide vanes are
bounded by two coaxial conical surfaces with a common vertex.
The vanes have constant thickness with rounded leading edge and
cusped trailing edge. The mean lines of the vane sections on the two
unfolded conical surfaces are arcs of circumference. The conical-walled
inlet guide vane row is dened using a quasi-three-dimensional
design method, by prescribing constant angular-momentum
distribution along the span at the exit of the guide vanes. The
meridional through-ow is computed by a streamline curvature
method and the blade-to-blade ow by a singularity surface
method. For the inner and the outer casing sections, the stagger
angle and the vane camber are computed to full the required
design circulation and zero-incidence ow at the leading edge. The
nal shape of the vane is obtained by moving a straight line along
the mean line of the two referred sections. This produces a single-

L.M.C. Ferro et al. / Renewable Energy 35 (2010) 19201928

Nomenclature
A
c
Ctot
Cm
D
DH
f
g
H
h
I
k
ku
kVm
N
N*
P
p
ptot
ptotN
Q
q
(r, q, z)
(r, f, q)
(r, j)
s
S1
S2
t

cross sectional area


blade chord
total pressure coefcient
constant of k  3 model
diameter; tip runner diameter
hub runner diameter
maximum camber height
acceleration due to gravity
available head
linear displacement of the probe measured from the
inner casing
turbulence intensity
turbulent kinetic energy
dimensionless tip speed velocity
dimensionless meridional velocity
number of blades
N/sin f
wetted perimeter of cross-section
static pressure
total pressure
total pressure of undisturbed upstream ow
ow rate
source intensity
cylindrical coordinate system
spherical coordinate system
coordinate system on the unfolded conical surface
circumferential pitch
blade-to-blade surface
meridional surface
blade thickness

curvature surface vane shape with straight leading and trailing


edges. To validate the design method, a 0.95 m inlet diameter
conical-walled guide vane row was designed, manufactured and
tested in an airow rig. Traversing measurements along the
circumferential and radial directions were made with ve-hole
probe. Measurements include the radial and circumferential
distributions of velocity and pressure at the exit of the guide vanes.
The experimental results are compared with the prescribed design
conditions and with numerical results from three-dimensional
analysis with FLUENT code for inviscid ow and viscous ow.
2. Quasi three-dimensional design method
A classical analysis method for the inviscid ow through a turbomachine was developed by Wu [3]. According to Wu the solution
of the three-dimensional ow eld (for pressure and velocity) is
obtained by means of the iterative solution of the two-dimensional
ow dened on two families of stream surfaces. Three-dimensional
streamlines can then be obtained as intersections of the two families
of stream surfaces. For a closed turbomachine, where the ow is
bounded by two wall surfaces of revolution, one of the families must
include these two surfaces, it being expected that the intermediate
surfaces are approximately surfaces of revolution. The other family
includes the surfaces of the vanes and a set of intermediate surfaces
whose shapes do not differ much from the mean blade surface. Wu
[3] named the rst family as S1 and the second family as S2. The
solution of the ow using the above three-dimensional approach is
computationally expensive, difcult to use, and usually produces

U
V
Vm
Vr
Vref
Vz
Vq
V*
(x, y, z)

1921

tip velocity of the runner


velocity
meridional velocity component
radial velocity component
reference velocity 4Q/[p(D2  D2H)]
axial velocity component
circumferential velocity component
dimensionless velocity
cartesian coordinate system

Greek symbol
b
ow angle measured from meridional direction
G
vortex circulation
3
turbulent dissipation rate
l
stagger angle
r
density
s
velocity angle relative to the chord
q
circumferential direction
F
dimensionless ow coefcient
f
conical angle
4
angle between the meridional vertical plane and the
meridional displacement probe plane
U
rotor angular speed
Subscript
1, 2
avg
ax
hyd
m

inlet and outlet, respectively


average
axial
hydraulic
meridional
relative to design conditions

double-curvature blade surfaces. Wu method can be simplied,


using a quasi-three-dimensional approach called through-ow
analysis approach, where (i) all surfaces in the S2 family become
identical (two-dimensional axisymmetric ow or meridional ow),
and (ii) the S1 surfaces are assumed to be surfaces of revolution
generated from the streamlines of the ow computed on the S2
surfaces. In the two-dimensional ow dened on the S2 surfaces, the
inuence of the blade thickness tq (measured in the circumferential
direction) is taken into account by a blockage factor, 1  tq/s, where s
is the circumferential pitch of the blade cascade.
The through-ow analysis approach allows the solution of the
three-dimensional ow to be obtained by the iterative solution of
the two-dimensional ows on the S1 and the S2 surfaces. The
solution of the ow on the S2 surface requires the knowledge of
the distribution of angular momentum rVq; this distribution is
computed by the solution of the ow on the S1 surfaces. However,
the calculation of the blade-to-blade ow (S1 surfaces) requires
the knowledge of the geometry of the streamlines on the S2
surface, as well as the distance between two adjacent streamlines
surfaces, which are obtained from the solution of the meridional
ow.
The ow on the S1 and S2 surfaces can be solved numerically
using different methods like the nite difference method (matrix
method) [4], the streamline curvature method [5,6], the nite
element method [7] and the singularity method (boundary
elements) [8,9]. In the present work, a streamline curvature method
is used for the solution of the meridional ow and a singularity
method is utilized for the blade-to-blade ow.

1922

L.M.C. Ferro et al. / Renewable Energy 35 (2010) 19201928

3. Design of inlet guide vanes

Vx1  iVy1
The through-ow analysis approach described in Section 2 is
applied to the design of a conical inlet guide vane system of a mini
hydraulic turbine. The bladed duct has six vanes. The vanes are
placed between two coaxial conical wall surfaces with a common
vertex, which coincide with the inner and the outer walls of the
guide vane system. The cone angles of the inner and the outer walls
are f 15 and f 32 , respectively. Upstream of the guide vanes
is a cylindrical annular duct and the ow at the inlet section of the
conical zone is assumed to be uniform without swirl (rVq 0).
Three steps need to be considered in the design of any turbine:
(i) specication of the design variables: ow rate Qh, available head
Hh and rotor angular speed U; (ii) denition of the velocity
diagrams upstream and downstream of the blade rows; (iii)
calculation of the blade geometry: camber and thickness distributions and cascade chord-to-pitch ratio c/s. The specication of the
design variables, available head Hh, ow rate Qh and angular speed
U can be done starting from one of them (for example Hh) and
computing the other two variables from dimensionless values of
the tip speed velocity, ku U/(2gHh)1/2, and the meridional velocity,
kVm Vm =2gHh 1=2 , obtained from the experience accumulated
from different manufacturers of hydraulic turbines [10,11]. A hubto-tip ratio DH/D 0.428 is chosen for the rotor. The radial distribution of the angular momentum rVq at the exit from the guide
vanes (inlet section of the rotor) can then be calculated from the
known values of Hh, Qh and U.
3.1. Computation of the meridional velocity eld
The meridional ow eld through the guide vanes is computed
by the streamline curvature method as described in Ref. [12]. The
following distribution of the angular momentum is prescribed
along the median surface of the vanes: the angular momentum is
constant along the span at the trailing and the leading edges;
between the leading and the trailing edges the angular momentum
follows a cosine law distribution

rVq

K
1  cospx0 ;
2

 N*
q N * 1 z*
;
4p z 1  z* N*

(2)

and the second due the thickness variation of the conical layer

Vr23

qN *

2pr

Vj23

)

 X
N
h
i
1
1
*
1 * 

cos nN j  j0 L ;
2
r
n1

N
h
i
qN * X
sin nN* j  j0 L;
2pr n 1

(3a)

(3b)

with


1  * an 12  * nN*
r
H r
;
2an
r
1
an
n2 N *2 n 0; 1; 2; .;
4

1

N* N/sin f, r* r/r0, z* z/z0, where (r, j) are coordinates along the


radial and circumferential directions on the unfolded conical
surface. Similar equations can be derived for the velocity induced by
the vortex lines [14]. The velocity and the pressure elds over the
surface are computed using a panel method [15] with linear
elements and constant strength distributions of sources and
vortices. To compute the velocities induced by a constant distribution on the panel, Equation (2) is rewritten by splitting the induced
velocity into the velocity due to the central element of the cascade
and the velocity due to the remaining elements. The nal equation is

Vx1  iVy1

Z
z0  z
1
1

ln 2

2p z01  z 2p
1
1 A

dz0 ;
z  z0

 * N*
*
@1 N 1 z  *
N
z0 1 2 z 1  z *
z0 2

(4)

where z02 and z01 denote the panel ends. The integrals in Equations
(3a), (3b) and (4) are computed by numerical gaussian integration.

(1)

where K is the maximum angular momentum at the trailing edge


and x0 is a dimensionless distance measured along a straight line
containing the common vertex (x0 0 at the leading edge and x0 1
at the trailing edge). It is noted that the x0 derivative of the above
cosine law distribution is zero for x0 0 and x0 1, thus respecting
the zero load condition at the leading and trailing edges of the
vanes.
3.2. Computation of the blade-to-blade velocity eld
The method described in [13] for the velocity potential due to
a cascade of N equally spaced source and vortex lines spanning two
conical surfaces with a common vertex is used for the computation
of the ow eld through the conical cascade of guide vanes. The
singularity lines are arcs of circle with the center at the common
vertex of the conical surfaces. The walls are at f f1, f f2
(0 < f1 < f2). The expressions for the velocity potential induced by
these rows of sources and vortices when the angular distance
between the conical walls becomes innitely small are also presented in [13]. The induced velocities at a point z(x, y) due to
a source of intensity q or a vortex of circulation G located at z0(x0, y0)
are obtained from the potential velocity derivative and may be split
into two components: the rst due a two-dimensional conical
cascade

Fig. 1. Isometric perspective of the inlet guide vanes system showing the inner casing
and the outer casing conical surfaces.

L.M.C. Ferro et al. / Renewable Energy 35 (2010) 19201928

A panel method code based on Eqs. (3) and (4) is used to optimize the pressure distribution over the blade surface by modifying
the camber and the stagger angle of the vanes at the inner casing
(f 15 ) and the outer casing (f 32 ).
The meridional distributions of rVq and Vm at the exit section of
the guide vane system are computed from the solution of the
meridional ow using the streamline curvature method as referred
in Section 3.1. The spanwise distributions of the meridional and the
circumferential components of the velocity at the exit section of the
guide vanes are used as input for the panel method program.
The stagger angle is adjusted so that the prescribed rVq value, i.e.
the circulation G around the section, is obtained. The section
camber is then optimized so that the pressure distribution on the
suction surface is smooth without suction peaks near the leading
edge. The boundary of the vane sections is discretized into 200
elements. Constant strength distribution of sources is adopted at
each panel, as well as constant intensity vortex distribution is
imposed along the section contour.
Once the vane sections at the inner and the outer casings are
known, the median vane surface is generated by connecting the
leading edges of the two proles with a straight line, and moving it
through the mean line of each prole up to the trailing edge. This
ensures that the median surface of the vane is a single-curvature
surface.
The relative position of the vane sections along the circumferential direction is adjusted to minimize the slope of the leading and
the trailing edges relative to the radial direction. The angles of the
leading and trailing edge relative to the radial direction are made
equal in magnitude with opposite signs. The mean lines of intermediate sections are obtained by the intersection of the conical
surfaces (15 < f < 32 ) with the straight line that generates the
vane mean surface. An isometric perspective of the conical-walled
inlet guide vane row is plotted in Fig. 1. The geometric characteristics of the vane sections are shown in Table 1, for different f angles.

Table 1
Characteristics of the vane sections.

f( )

c(m)

fmax/c

tmax/c

l( )

15.00
19.25
23.50
27.75
32.00

0.333
0.341
0.354
0.371
0.393

0.113
0.103
0.092
0.082
0.071

0.0150
0.0146
0.0141
0.0135
0.0127

33.55
28.96
24.85
20.88
17.00

The number of terms to be kept in the series of Equation (3)


depends on the value of r* and increases as r* approaches unity.
The resulting linear system of equations is solved by a Gauss
elimination direct method.
3.3. Computation of the aerodynamic parameters
For economical reasons, simple shapes are adopted in the design
of the guide vanes. The mean lines for the inner casing and outer
casing vane-proles are arcs of circumference on the unfolded
conical surfaces.
The rst step consists in a conformal transformation from
a rectilinear straight (cylindrical) cascade to a conical cascade. This
transformation allows to change from the Cartesian plane z(x, y) to
the unfolded conical plane z(r, j),

Ns
ln z;
2psin f

1923

(5)

where s 2pr0/N is the pitch of the cylindrical cascade, with r0


p
r1 r2 (here the subscripts 1 and 2 indicate the inlet and the outlet
sections of the conical cascade, respectively). For a rectilinear
straight cascade with circular-arc proles the inlet and the outlet
velocity conditions are related to the stagger angle l as follows [16]:
s1 s2 0, s2 b2 l and s1 l, where s1and s2 are the angles of
the velocity vector relative to the chord of the prole and
b arctanVq/Vm. For the inner casing and the outer casing sections,
the above conformal transformation is used to dene the chord
length, the shape of the mean line and the camber that will be
changed in an ultimate phase of the design.
Constant 5 mm thickness distribution is prescribed for the vane
section (t/c z 1.5%), except near the leading edge (elliptical distribution of thickness) and trailing edge (parabolic distribution of
thickness) [14]. The nal shape of the proles is obtained
combining the thickness and camber distribution as for the NACA
sections [17].

4. Experimental facility
An experimental blow-down rig was used to validate the whole
turbine design. Air is the working uid. The turbine is composed by
an upstream annular duct with 950 mm outer diameter, an inlet
conical guide vane system, a 0.5 m tip diameter rotor and a 2 m
long, 6 angle diffuser, Fig. 2. There is a plenum chamber between
the fan and the turbine diffuser. The ow rate is measured by
a calibrated nozzle with an ellipsoidal longitudinal section and
a 400 mm diameter throat.

guide vanes
950 mm

diffuser
6

500 mm

rotor blades
3380 mm
Fig. 2. Schematic representation of the turbine model.

L.M.C. Ferro et al. / Renewable Energy 35 (2010) 19201928

The use of air, instead of water, has the advantages of allowing in


practice a much less expensive open circuit rig (rather than a closed
one) to be used, with a much smaller energy consumption and
simpler and less expensive instrumentation. The drawback is the
larger cinematic viscosity of air, which could in part be compensated, in terms of Reynolds number reproducibility, by a larger
rotational speed (about 2500 rpm as compared with 750 rpm for
water). Of course, no cavitation studies could be performed in air.
The measurement of the velocity and pressure distributions at
the outlet sections of the guide vane system were made with a ve
holes probe: one central hole and four equidistant lateral holes
(two holes with the center on a plane normal to the axis of the
probe right and left holes and other two on the axis of the probe
upper and lower holes) [14]. The probe has two degrees of
freedom: a linear displacement along the probe axis and a rotation.
The radial and yaw movements of the probe are produced by
a mechanical system with a step displacement of 0.01 mm and an
angular step of 0.1. The probe axis is placed on a meridional plane
with its axis at an angle of 20 relative to the radial direction. The
probe is rotated until the pressure difference between the left and
right holes vanish. The velocity direction, the total pressure and the
dynamic pressure are obtained by linear interpolation from the
probe calibration curves.
5. Computation of inviscid and viscous three-dimensional
ow using the FLUENT code
The FLUENT 6.2 [18] code was used to compute the threedimensional ow through the guide vanes. FLUENT solves the
integral equations for the conservation of mass, momentum and
turbulence species, turbulence kinetic energy k and turbulence
dissipation rate 3 (for viscous turbulent ow), using the controlvolume-based technique. The domain is divided into discrete
volumes by a computational mesh. The computational meshes
were generated using the GAMBIT 2.2 code [19]. The mesh may be
either unstructured, with tetrahedrical elements, or structured
with hexahedrical elements.
The numerical calculations used the segregated method with
SIMPLEC [20] pressure-velocity coupling algorithm. QUICK [21]
interpolation scheme for momentum and a second-order upwind
scheme for the pressure are applied. The turbulence eld is simulated by the standard two-equation k  3 turbulence model [22]

1.5

0.6

0.5

0.4

(rV)*

1924

0.3

0.2
Design
Three-dimensional inviscid
Three-dimensional viscous
Test case 3
Test case 4

0.1

0.0
0.0

0.2

0.4

0.6

0.8

1.0

h/hmax
Fig. 4. Spanwise distribution of circumferentially averaged angular momentum.

with either the standard wall functions (logarithmic law), as


dened by Launder and Spalding [23], or a near wall modeling
approach, known as Enhanced Wall Treatment [2426].
The boundary conditions used at inlet and outlet sections are
the same for inviscid ow and viscous ow. At the inlet section,
a pressure-inlet type condition is used, with total pressure, turbulence parameters and velocity direction prescribed at this section.
For viscous ow, turbulence quantities, k and 3, must be specied at
inlet section. For a uid entering a duct, FLUENT suggests a uniform
value of the turbulence quantities to be prescribed at the inlet
section. A uniform distribution of the turbulence intensity, I,
dened as the ratio of the root-mean-square of the velocity uctuation, u0 , to the mean velocity, Vavg, is prescribed at inlet section.
The turbulent kinetic energy k is given by k 3/2 (VavgI)2 and the
dissipation rate 3 by

3 Cm3=4

k3=2
;
0:07 Dhyd

(6)

1.2

1.5

1.1
1.0

1.0

1.0

0.5

0.5

0.9

V*z

0.7

Cp

Cp

0.8

0.5

0.0

0.0

0.6

0.4
Design
Three-dimensional inviscid
Three-dimensional viscous
Test case 3
Test case 4

0.3

-0.5

-0.5

0.2
0.1
-1.0
0.0

0.2

0.4

0.6

x/c

0.8

1.0

-1.0
0.0

0.2

0.4

0.6

0.8

1.0

x/c

Fig. 3. Pressure coefcient distribution around (a) the inner and (b) outer casing vane
sections.

0.0
0.0

0.2

0.4

0.6

0.8

1.0

h/hmax
Fig. 5. Spanwise distribution of circumferentially averaged axial velocity component.

L.M.C. Ferro et al. / Renewable Energy 35 (2010) 19201928

0.0

hi h1

-0.1

V*r

-0.3

-0.5
-0.6
0.0


0
hmax 
cos 20  cos fi ;

2cos 20

(7)

where h is the linear displacement of the probe measured from the


0
inner casing wall, h1 1 mm and hmax 137.5 mm. The values of fi
are given by

-0.2

-0.4

1925

Design
Three-dimensional inviscid
Three-dimensional viscous
Test Case 3
Test Case 4

0.2

0.4

0.6

0.8

1.0

h/hmax
Fig. 6. Spanwise distribution of circumferentially averaged radial velocity component.

where Dhyd 4A/P is the hydraulic diameter, A the cross-sectional


area, P the wetted perimeter of the cross-section and Cm 0.09 is
an empirical constant of the k  3 turbulence model [22].
At the outlet section, a pressure-outlet type condition with
radial equilibrium is used, the specied pressure being the static
pressure at the minimum radius of the boundary.
Wall boundary conditions are employed to limit the uid and
solid regions. The no-slip condition (zero velocity) is enforced by
FLUENT at stationary walls for viscous ow. Periodic boundary
conditions are applied with no pressure discontinuities across the
periodic surfaces. When computing the ow across periodic
boundary adjacent to a uid cell, FLUENT uses the ow conditions
at the cell adjacent to the opposite periodic surface.

6. Results
Measurements with the ve-hole probe described in Section 4
were made at the outlet section of the guide vanes, at 16 meridional planes with a circumferential amplitude of 65 larger than
the 60 angular amplitude of the pitch of the cascade. The rst
reading line is placed at 4 13 , where 4 is the angle measured in
a plane normal to the turbine axis, with 4 0 when the radial
direction coincides with the vertical direction. There are 14 equally
ve-degree-spaced lines from 4 13 to 4 78 with two intermediate lines at 4 21.5 and 4 25.5 . At each reading line, the
span of the probe displacement is 137.5 mm. A total of N 21 points
are considered at each measurement line with a cosine law distribution for the position of the probe,

Vref

1.0

1.0

0.8

0.8

Test case 1
Test case 2
Test case 3
Test case 4
Three-dimensional inviscid
Three-dimensional viscous (k- standard)
Three-dimensional viscous (k- ENH)

0.2
0.0
0.0

0.2

0.4

0.6

h/hmax

0.8

4Q

p D2  D2H

1.2

0.4

;

(9)

0.6
Test case 1
Test case 2
Test case 3
Test case 4
Three-dimensional viscous (k- standard)
Three-dimensionl viscous (k- ENH)
Three-dimensional inviscid

0.4
0.2

1.0

(8)

where D 0.5 m is the diameter of the rotor tip and DH 0.107 m is


the diameter of the rotor hub. The angular momentum is made
non-dimensional by the velocity Vref and the rotor outer radius.
The computational domain for the three-dimensional ow is
composed by two cylindrical ducts, one upstream and another
downstream of the conical duct that contains the six guide vanes.
The total axial length is 1.26 m. The outer diameters of the
upstream and downstream cylindrical ducts are 0.946 m and 0.5 m,
respectively. The axial vane-chord cax is equal to 0.289 m for the
inner conical surface and 0.326 m for the outer surface. The
distance between the domain inlet section and the mid-span vane
leading edge is z 0.6 m or z/cax 0.30 where z is the axial
coordinate. The dimensionless axial distance z/cax from the trailing
edge to the outlet section of the domain is 1.08 at the outer conical
surface and 1.24 to the inner one. The nal mesh is generated using
the turbo option of the GAMBIT program. The numerical simulation
for inviscid ow was performed with an unstructured tetrahedrical
mesh with 1,886,484 elements. A hexahedrical mesh with
1,344,000 elements was used for the solution of the viscous ow.
The design pressure coefcient distributions, Cp (V/Vref)2  1
around the contour of the inner and outer casing sections are
plotted in Fig. 3 and compared with FLUENT numerical results for
three-dimensional inviscid and viscous ow. Except in the vicinity
of the leading and trailing edges, these distributions show

0.6

180+  2  20
i  1 ;
N1

with i 1, ., N  1.
A total of 16  21 336 measuring points were taken for four
different test condition: two without rotor blades and ow rates
Q Qh, and Q 1.56 Qh, another one with F 0.110, U 2151 rpm
and the last one with F 0.155, U 1455 rpm, where F Q/(UD3)
is the dimensionless ow coefcient. The design values are
Qh 3.36 m3/s, (rVq)h 2.491 m2/s and U 2500 rpm. The velocity
is non-dimensionalized by the mean axial velocity at the rotor inlet
section

V*
z

Vz

1.2

fi f1

0.0
0.0

0.2

0.4

0.6

h/hmax

Fig. 7. Axial velocity distributions at the measurement section for (a) 4 23 and (b) 4 43 .

0.8

1.0

1926

L.M.C. Ferro et al. / Renewable Energy 35 (2010) 19201928

0.0

0.0

-0.1

-0.1

-0.2

-0.2

V*r

V*r

-0.3
Test case 1
Test case 2
Test case 3
Test case 4
Three-dimensional inviscid
Three-dimensional viscous (k- standard)
Three-dimensional viscous (k- ENH)

-0.4
-0.5
-0.6
0.0

0.2

0.4

0.6

0.8

-0.3
-0.4
-0.5
-0.6
0.0

1.0

Test case 1
Test case 2
Test case 3
Test case 4
Three-dimensional viscous (k- standard)
Three-dimensional viscous (k- ENH)
Three-dimensional inviscid

0.2

0.4

0.6

0.8

1.0

h/hmax

h/hmax

Fig. 8. Radial velocity distributions at the measurement section for (a) 4 23 and (b) 4 43 .

a monotone ow acceleration that increases the vane pressure load


with the axial distance. As a consequence of the prescribed design
law of Equation (1), the vanes introduce small deection in the ow
up to about 30% of the vane chord. At the inner casing, the agreement between the design results and FLUENT calculations is very
good. Poorer agreement is observed at the outer casing. Differences
between design and FLUENT results might be due to larger design
inlet upstream velocity than that in the real three-dimensional
ow.
Figs. 46 show comparisons between (i) the design values given
by the solution of the meridional ow, (ii) the experimental values
and (iii and iv) the numerical results from FLUENT. These are circumferentially and mass-ow averaged values. Those gures plot
the spanwise distributions of the dimensionless axial and radial
velocity components, V*z Vz/Vref, Vr* Vr/Vref, and the angular
momentum (rVq)* (2rVq)/(DVref). Results in Figs. 46 show fairly
good agreement between the experimental, the design and the
numerically obtained results. It may be seen that the design
method is able to produce the desired ow deection distribution
with very good accuracy. Numerical results from FLUENT show that
viscous effects practically do not affect the Vz and Vr velocity
components, and produce a small reduction in the circumferential
component Vq that is more relevant near the inner casing wall.
Close the inner casing, the experimental and numerical results for
viscous ow show smaller ow deection than the design
condition.
The spanwise distributions at 4 23 and 4 43 for the three
velocity components V*z , Vr* and V*q are plotted in Figs. 79. The

1.2

Three-dimensional inviscid
Three-dimensional viscous (k- standard)
Three-dimensional viscous (k- ENH)
Test case 1
Test case 2
Test case 3
Test case 4

1.0

1.2
1.0
0.8

0.6

V*

V*

0.8

numerical results from FLUENT for the inviscid three-dimensional


ow and for the viscous ow with k  3 standard model (k  3
standard) and with Enhanced Wall Treatment (k  3 ENH) are also
shown. A fairly good agreement between the experimental and
numerical results at 4 43 is observed. The agreement between
the wake position on experimental and numerical results for
4 23 is good, although the numerical values of the velocities at
the wake are too low in comparison with the experimental ones.
The comparison between numerically obtained results with standard k  3 and k  3 ENH only show very small differences for the
circumferential velocity prole at 4 23 in the vane wake zone.
The experimental isolines of the axial velocity component V*z
and angular momentum (rVq)* are plotted in Fig. 10 at the
measurement plane and are compared with numerical results for
inviscid and viscous ow. A fairly good agreement is observed
between the isolines for the axial velocity component. The realuid effects close to the inner and outer casing walls predicted by
k  3 standard model could not be conrmed by the measurements
due to the coarse measurement grid and the loss of accuracy of the
probe close the conical walls (especially at the inner wall). The
isolines of angular momentum computed by FLUENT also show
a good agreement with experimental results, with predicted values
greater than the design value (0.475) close the outer casing and the
upper surface of the vanes. Fig. 11 represents the total pressure
2 , where p
coefcient Ctot ptot  ptotN =12rVref
totN is the static
pressure of undisturbed upstream ow, at the measurement plane.
The values of Ctot are almost constant and close to zero, which
shows that the losses through the inlet guide vanes are small. As

0.6

0.4

0.4

0.2

0.2

0.0
0.0

0.2

0.4

0.6

h/hmax

0.8

1.0

0.0
0.0

Test case 1
Test case 2
Test case 3
Test case 4
Three-dimensional viscous (k- standard)
Three-dimensional viscous (k- ENH)
Three-dimensional inviscid

0.2

0.4

0.6

0.8

h/hmax

Fig. 9. Circumferential velocity distributions at the measurement section for (a) 4 23 and (b) 4 43 .

1.0

L.M.C. Ferro et al. / Renewable Energy 35 (2010) 19201928

1927

Fig. 10. Comparison of numerical and experimental axial velocity and angular momentum isolines at measurement section.

expected, the inviscid ow predictions give a Ctot 0, except for


a very small region at the wake of the blade where numerical
diffusion is apparent. At the blade wakes and close to the inner and
outer casing walls, the viscous-ow FLUENT code predicts a non-

negligible total pressure loss due to the presence the boundary


layers. This agrees qualitatively with experimental data. It should
be noted that the dimension of the probe tip and the interaction of
the probe with the wall do not allow accurate experimental results

Fig. 11. Comparison of numerical and experimental total pressure coefcient distribution at measurement section: (a) experimental results, (b) inviscid ow and (c) viscous ow
(k  3 standard).

1928

L.M.C. Ferro et al. / Renewable Energy 35 (2010) 19201928

close to the walls. In fact, the probe was unable to detect the
boundary layer at inner casing surface. Fig. 11 also shows the thin
wakes of the blades.
7. Conclusions
A guide vane system of a mini hydraulic turbine was designed
using a quasi-three-dimensional method, combining the streamline
curvature method for the solution of the meridional ow and a panel
method for the blade-to-blade ow. A conical-walled inlet guide vane
with six xed vanes was manufactured and tested in an airow rig.
Experimental results for the velocity and pressure distributions
obtained with a ve-hole probe were compared with the design
values and the numerical values obtained with FLUENT for inviscid
and viscous ows. The experimental results show that the nal shape
of the vanes produces with good approximation the prescribed
angular momentum distribution at the exit of the guide vane system
although small deviations between the experimental and the design
values appear for the angular momentum close the inner casing. The
FLUENT code yielded numerical results in fairly good agreement with
the experimental results, and conrmed it as a useful tool for the
analysis of the ow through a guide vane system. The inviscid quasithree-dimensional method presented in the paper was found to be an
adequate design tool for this kind of applications.
Acknowledgements
This work was supported by IDMEC and the Portuguese Foundation for Science and Technology under contract PTDC/EME-MFE/
66608/2006.
References
[1] Thematic Network on Small Hydropower (TN SHP). Proposals for a European
strategy of research, development and demonstration (RD&D) for renewable
energy for small hydropower. Available at:. Brussels: MHyLab and ESHA
http://www.esha.be; 2005. Technical report.
[2] EUROPA Eurostat Environment and Energy European Comission. Available
at:
http://nui.epp.eurostat.ec.europa.eu/nui/
setupModifyTableLayout.do
[Accessed July 2007].
[3] Wu CH. General through-ow theory of uid ow with subsonic or supersonic
velocity in turbomachinery of arbitrary hub and casing shapes. NACA TN 2302.
Cleveland: Lewis Fligth Propulsion Laboratory; 1951.

[4] Marsh H. A digital computer program for the through-ow uid mechanics in
an arbitrary turbomachine using a matrix method, R&M 3509. London:
Aeronautical Research Center; 1968.
[5] Katsanis T. Use of arbitrary quasi-orthogonals for calculating ow distribution
in the meridional plane of a turbomachine. NACA TN D-2546. Cleveland: Lewis
Research Center; 1964.
[6] Katsanis T. Use of arbitrary quasi-orthogonals for calculating ow distribution
on a blade-to-blade surface in a turbomachine, NACA TN D-2809. Cleveland:
Lewis Research Center; 1965.
[7] Krimerman Y, Adler D. The complete three-dimensional calculation of the
compressible ow eld in turbo impellers. J Mech Eng Sci 1978;20(3):14958.
[8] Schlichting H. Calculation of the frictionless incompressible ow for a given
two-dimensional cascade (direct problem). NASA TMX 56364. Langley
Research Center; 1958.
[9] Martensen E. The calculation of the pressure distribution on a cascade of thick
airfoils by means of a Fredholm integral equation. NASA TT-F-702. Gottingen:
Max-Planck-Institute for Fluid Research and Aerodynamic Experimental
Station; 1959.
[10] Raabe J. Hydro power. Dusseldorf: VDI-Verlag; 1985.
[11] Nechleba N. Hydraulyc turbines, their design and equipment. Prague: Artie;
1957.
[12] Denton JD. Throughow calculations for transonic axial ow turbines. J Eng
Power Trans ASME 1978;100(2):2128.
[13] Falcao AFO, Ferro LMC. Flow eld due to a row of vortex and source line
spanning a conical annular duct. J Eng Math 1989;23:20318.
[14] Ferro LMC. Numerical and experimental study of the ow through an axial
hydraulic turbine (in Portuguese). PhD thesis, Instituto Superior Tecnico,
Technical University of Lisbon, Lisbon; 2009.
[15] Hess JL, Smith AMO. Calculation of potential ow about arbitrary bodies. Prog
Aerosp Sci 1967;8:1138.
[16] Scholz N. Aerodynamics of cascades; 1977. AGARDograph No. 220.
[17] Abbott I, von Doenhoff A. Theory of wing sections. New York: Dover Publications Inc.; 1959.
[18] Fluent Incorporated. Fluent 6.2 users guide. Lebanon: Centerra Resource Park;
2005.
[19] Fluent Incorporated. Gambit 2.2 users guide. Lebanon: Centerra Resource
Park; 2005.
[20] Vandoormaal JP, Raithby GD. Enhancements of the SIMPLE method for predicting incompressible uid ows. Numer Heat Transfer 1984;7(2):14763.
[21] Leonard BP, Mokhtari S. ULTRA-SHARP Nonoscillatory convection schemes for
high-speed steady multidimensional ow. NASA TM 1-2568 (ICOMP-90-12).
Cleveland: Lewis Research Center; 1990.
[22] Launder BE, Spalding DB. Lectures in mathematical models of turbulence.
London, England: Academic Press; 1972.
[23] Launder BE, Spalding DB. The numerical computation of turbulent ows.
Comput Method Appl M 1974;3(2):26989.
[24] Wolfstein M. The velocity and temperature distribution of one-dimensional
ow with turbulence augmentation and pressure gradient. Int J Heat and
Mass Tranfer 1969;12:30118.
[25] Chen HC, Patel VC. Near-wall turbulence models for complex ows including
separation. AIAA J 1988;26(6):6418.
[26] Jongen T. Simulation and Modeling of Turbulent Incompressible Flows. PhD
thesis, EPF Lausanne, Lausanne; 1991.

Das könnte Ihnen auch gefallen