Axial-ow turbines for low head microhydro systems

K.V. Alexander
a,
*
, E.P. Giddens
b,1
, A.M. Fuller
a
a
Department of Mechanical Engineering, University of Canterbury, PB 4800, Christchurch, New Zealand
b
Department of Civil Engineering, University of Canterbury, Christchurch, New Zealand
a r t i c l e i n f o
Article history:
Received 19 December 2007
Accepted 22 March 2008
Available online 25 July 2008
Keywords:
Microhydro
Hydropower
Renewable energy
Home energy
Low head
Remote power generation
a b s t r a c t
This paper describes the design of four different specic speed microhydro propeller turbines operating
at heads between 4 m and 9 m, and their application to a wider range of heads and outputs by scaling.
The features are specically tailored for ease of manufacture and uniquely resistant to debris blockage.
Test machines are described and test results given; hydraulic efciencies of over 68% have been achieved
in all test models despite the fact that these turbines blades are planar, further simplifying manufacture.
Theoretical models show how closely these at blades can be made to approach the ideal blade shapes.
Outline drawings are given with key dimensions for each reference model, along with the equations for
scaling to arbitrary sites. These turbines are the axial ow members of a family of turbines developed to
cover the microhydro range from 2 m to about 40 m of head.
2008 Elsevier Ltd. All rights reserved.
1. Introduction
This paper is one of a group of papers describing a University of
Canterbury project that sets out to provide a properly congured
range of microhydro systems, from radial- to axial-ow designs,
based on a modular concept and aimed in particular at third-world
sites where regional workshops might be capable of undertaking
much of the manufacture themselves. With that in mind, the scope
of this paper is a subset of the microhydro project at the University
of Canterbury. This papers scope is specically the propeller, or
axial-ow, turbines. Future papers will cover the Universitys
radial- and mixed-owdesigns. A particular goal is to provide well-
grounded but lowest-cost options for the communities who pos-
sess an adequate hydro resource and access to basic fabrication
facilities, but who are conned to less reliable, sustainable, or
economical power generating technologies due to a lack of design
knowledge. Earlier papers have given an overviewand justication
of the modular approach [1,2] and the analysis leading to the most
economic choice of penstock [3]. The selection of the most eco-
nomic penstock dictates the turbine size. The combination of dis-
charge, available head, and a xed generator speed allow
calculation of the sites specic speed. If that value falls in the range
of this papers designs, then one of the turbines described in this
paper is likely a suitable choice for development.
Microhydro is typically used to describe sites of output below
about 20 kW. Above that is minihydro, where the scale of in-
vestment is such that professional input is proportionally smaller
and there is some advantage in building custom-made systems. The
area of particular interest for the project discussed here is specic
speeds above those of the Pelton wheel. Pelton wheels are simpler
to implement than reaction turbines of similar power output, and
are therefore quite popular relative to other types of microhydro
machines. However, their applicability is limited to high-head sites.
A number of Pelton wheel solutions are already available [4]. This
project aims to complement these and other existing microhydro
solutions by providing efcient designs for a broader range of sites.
The exact scope of this project in terms of head and discharge is
shown in Fig. 1.
The development of turbines suitable for operation in the
bounded region of Fig. 1 has been the main task of the project,
including Francis (radial-ow), mixed-ow, and propeller (axial-
ow) machines.
2. Turbine forms and scaling
It is well known that different forms of turbine are required
for different conditions; the classic range of turbine forms rele-
vant to this project is shown in Fig. 2. The forms are classied by
their specic speeds, where in this instance the specic speed is
dened as
* Corresponding author. Tel.: 64 3 3667 001x7385; fax: 64 3 364 2078
E-mail addresses: keith.alexander@canterbury.ac.nz (K.V. Alexander), amf88@
student.canterbury.ac.nz (A.M. Fuller).
1
Recently deceased.
Contents lists available at ScienceDirect
Renewable Energy
j ournal homepage: www. el sevi er. com/ l ocat e/ renene
0960-1481/$ see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2008.03.017
Renewable Energy 34 (2009) 3547
N
S

N

P
r
p
H
5=4
(1)
Note this denitions departure from Nechlebas version with the
use of an indirectly calculated powerdthat absorbed by the run-
nerdrather than the power measured directly by the dynamom-
eter [5, p. 71]. As with Nechlebas, and other practical forms of
specic speed, it is not a dimensionless number. In this case, the SI
units are
N
S

Newtons
1=2
seconds
3=2
meters
3=4
Therefore, its use is restricted to comparison with machines char-
acterized in the same way. It is worth noting here an interpretation
of the termspecic speed which may help clarify its meaning and
relevance for comparison of hydraulic machines. To paraphrase the
source, The specic speed of any turbine equals the speed of
a geometrically similar turbine working under the head of 1 m,
when the latter turbine has such dimensions that it delivers under
the head of 1 m a unit of power [5, p. 70]. A denition of N
S
based
on runner power was adopted because the four reference turbines
covered in this paper were developed in parallel with matching
bearing and transmission designs. As part of each machines de-
velopment, the efciency of transmission components has been
measured, which allows isolated analysis of hydraulic performance.
Gross Head, Discharge, Specific Speed at 1550 RPM and Electrical Power Delivered
P
e
l
t
o
n

W
h
e
e
l

N
s

=

3
6
2
5

k
W

E
l
e
c
t
i
c
a
l

O
u
t
p
u
t
.
0
.
2

k
W

E
l
e
c
t
i
c
a
l

O
u
t
p
u
t
.
H
i
g
h
e
s
t

P
r
a
c
t
i
c
a
l

N
s

=

6
0
0
1
10
100
1 10 100 1000
Discharge (l/s)
G
r
o
s
s

S
i
t
e

H
e
a
d

(
m
)
.
Pelton Wheel Region
Minihydro Region
Microhydro Region
Power too low to consider
Specific Speed too high to achieve
Fig. 1. The site head and discharge ranges dening the microhydro and neighbouring regions.
Notation
1
subscript denoting leading edge station
2
subscript denoting trailing edge station
f(.) function of stated variables
g gravitational acceleration (m/s
2
)
h
subscript denoting location on blade/hub intersection
(Fig. 5)
H
R
head ratio, site/reference
H turbine head (m)
j counter
k constant for free vortex ow
L
R
length ratio, site/reference
_ m mass ow rate (kg/s)
N runner rotational speed (rev./min)
N
S
specic speed
P
i
hydraulic power into the turbine from the penstock
(W)
P
r
power out of the runner (W)
P
R
power ratio, site/reference
P
d
power measured at the dynamometer (W)
Q ow rate (l/s or m
3
/s as noted)
Q
R
discharge ratio, site/reference
r general radius (mm)
ref
subscript denoting parameter of a reference machine
site
subscript denoting parameter of machine scaled to
match a real site
t
subscript denoting location on blade tip (Fig. 5)
T
r
runner torque (Nm)
v
a
axial ow velocity (m/s)
v
c
circumferential ow velocity (m/s)
a angle CCW around scroll casing, from inlet/casing
intersection, as viewed from generator end, for use
with Eq. (26) (degrees)
b blade angle relative to the impeller plane (degrees)
d deviation angle (degrees)
h
t
turbine hydraulic efciency (does not include h
m
)
h
m
combined bearing, shaft seal, and transmission
efciency
p 3.14159.
q angle clockwise around runner axis from the setup
point O (degrees)
r density of water (kg/m
3
)
u runner rotational speed (rad/s)
j blade set-up angle (degrees)
K.V. Alexander et al. / Renewable Energy 34 (2009) 3547 36
To facilitate comparison based on hydraulic performance alone, the
power measured by the dynamometer during testing has been
divided by h
m
, so that the power output reported and used in Eq. (1)
is the power extracted from the uid by the runner, P
r
. This leads to
slightly high estimates of power and specic speed when com-
paring to designs characterized by their output at the dynamom-
eter shaft, but has been chosen for the sake of simplicity in this
analysis and reusability of the data. P
i
, the hydraulic power into the
turbine, is therefore the maximumpower available, and will always
exceed a power conversion machines output due to efciencies
less than unity.
P
i
rQgH _ mgH (2)
The relationship between the available hydraulic power, efcien-
cies, and measured output at the dynamometer is
P
d
h
m
h
t
rQgH h
m
h
t
P
i
h
m
P
r
(3)
To obtain the power transmitted by the runner for hydraulic per-
formance comparison, divide Eq. (2) by h
m
to yield
P
r

P
d
h
m
h
t
P
i
(4)
Substituting Eq. (2) in Eq. (4) and the result into Eq. (1) gives the
explicit form used in this project.
N
S
N

h
t
rgQ
_
H
3=4
(5)
In this project, the range of specic speeds is from 60 (high head,
small ow, radial-ow) to 540 (lowhead, large ow, axial-ow). For
a given specic speed the geometrical form of the turbine remains
the same, but the size may be scaled (Fig. 3). This allows the design
for each specic speed to be scaled to match the available penstock
ows and to deliver, fromthe appropriate heads and discharges, the
4 or 5 power bands to cover the 0.220 kW range. When scaling
from a reference machine, the resultant hydraulic efciency can be
predicted from an empirical function of the physical size (or dis-
charge, depending on the model used) of so-called majoration ef-
fects [6, p. 765].
Furthermore, these several sizes of the same form can be
arranged to deliver at a xed N. This is apparent from Eq. (1) where
Q and H may be increased while keeping N
S
and N constant. A
constant N enables the generation of AC in the standard frequency
of 50 Hz using directly driven 4-pole generators, regardless of in-
stallation size.
Once a survey is completed for a particular site, using the
methods in [2] and [3], Q and H are known parameters. Now,
suppose there is a reference model of known P
r
, N, H, Q, N
S
, and size.
For the purposes of illustration, assume the reference machine runs
at 1500 rev./min and produces 1 kW at the runner shaft. To pro-
ceed, it is necessary to scale that reference machine to an available
site. Fig. 3 represents the process of scaling to a geometrically
similar machine, that is, a machine with the same value of N
S
, but
the calculations to produce such a graphical tool are as follows. The
project constraints of the scaling operation will be N
site
1500 to
satisfy the generator and N
S;site
N
S;ref
to preserve the validity of
the scaling operations. N
S,site
, although it is a property of the ma-
chine, can be estimated fromthe sites parameters. H, Q, and length
ratio, L
R
, are to be determined. First, compute the site/reference
ratios from the given values for P and N.
8
0

m
m
1
0
0

m
m
1
2
5

m
m
1
5
0

m
m
1
7
5

m
m
2
0
0

m
m
2
2
5

m
m
2
5
0

m
m
3
0
0

m
m
P
e
lt
o
n

N
S
=
3
6

=
7
0
%

N
S
=
6
0
N
S
=
9
3
N
S
=
1
7
6
N
S
=
2
4
2
N
S
=
3
5
5
N
S
=
5
4
4
0
.
3

k
W

E
l
e
c
t
r
i
c
1
5

k
W

E
l
e
c
t
r
i
c
3
5

k
W

E
l
e
c
t
r
i
c
:

M
i
n
i

H
y
d
r
o
1
.
3

k
W

E
l
e
c
t
r
i
c
3

k
W

E
l
e
c
t
r
i
c
5

k
W

E
l
e
c
t
r
i
c
1
10
100
1 10 100 1000
Discharge, Q (l/s)
T
u
r
b
i
n
e

H
e
a
d
,

H

[
m
]
Turbine Options
Penstock Nominal Diameters:
Fig. 3. Axial-ow turbine selection chart, N
S
lines are at constant h
t
70%.
Fig. 2. The classic gure of turbine forms, with their specic speeds based on the
denition in Eq. (1).
K.V. Alexander et al. / Renewable Energy 34 (2009) 3547 37
P
R

P
r;site
P
r;ref

1kW
P
r;ref
(6)
N
R

N
site
N
ref

1500 rev:=min
N
ref
(7)
Next, calculate H
R
, Q
R
, and L
R
, which, when multiplied by the ref-
erence models values of H, Q, and any linear dimension, will yield
the site machines values to achieve the stated constraints. Since
scaling is explicitly constrained by dimensional similarity, N
S
is the
same for site and reference, and Eq. (1) can be solved for H
R
be-
tween the two.
H
R

_
N
R

P
R
_
_
0:8
(8)
Eq. (8), the discharge ratio, is a manipulated form of Eq. (5), for
geometrically similar machines. Here, assume that h
t
is constant
during this operation, despite having already stated that h
t
f(Q
R
),
or alternatively: h
t
f(L
R
). Efciency majoration is included after
scaling as a function of L
R
in Eq. (11).
Q
R

P
R
H
R
(9)
Eq. (10) relates ow and head variations to the physical scale ratio
from reference to site.
L
R

Q
R

H
R
_

(10)
The site machines hydraulic efciency can be majorated from the
reference machines h
t
as a function of the size ratio between the
two using an empirical relation [6, p. 765, Eq. (10)].
h
t;site
1
_
1 h
t;ref
_
L
0:25
r
(11)
In line with the modular concept, three different axial-ow (pro-
peller) turbine designs have been developed, with a fourth in the
nal stages, each of a different specic speed. This paper focuses on
the turbines in the four higher specic speed ranges (lower heads)
of N
S
176, 242, 355 and 544. The actual machines built of each
design have been inconsistent in terms of speed and power output,
and have been sized to t a real-world target site and to compli-
ment the dynamometer and plumbing facilities available for test-
ing. Their specications are shown in Table 1. For this reason, the
four test models have been scaled, using the exact procedure pre-
sented above, to an output of 1 kWand a speed of 1500 rev./min to
facilitate comparison between the designs as shown in Table 2.
These normalized results, compiled in Table 2, are the so-called
references. These can be scaled to cover heads from 2 m to 20 m
depending on their power output, as seen in Fig. 3.
3. Inlet theoretical design
This projects objective of producing minimal-maintenance de-
signs with long-term resistance to clogging manifests itself in the
unique design of most key components. Eliminating the possibility
of clogging has necessitated the removal of any ow-spanning
structure. Hence, stay vanes and inlet guide vanes are left out al-
together, runner blade shapes are strongly inuenced by their
vegetation-shedding performance, and ow-straightening vanes
downstream of the runner have been designed to be self-cleaning.
The main impact of this consideration on the designs presented
here is that the conditions at the runner are not constrained by
guide vanes immediately upstream, but rely solely on the static
geometry of the casing. As the machines are intended to run at
a single setting, with no adjustability, the nal geometry of each
Table 1
Lab model specications
Performance and sizing for best models
Specic speed Achieved 176 242 355 544
Design parameters P
r
, kW 3.07 1.96 1.41 2.82
H
t
, m 9.198 7.07 6.31 3.39
N, rev./min 1605 1989 2986 1491
Q, l/s 46.2 38 32.4 123.9
h
t
h
m
, % 70.6 70.8 62.7 60.8
h
m
, % 95.5 95.0 89.2 89.2
h
t
, % 73.9 74.5 70.3 68.2
Runner parameters Runner OD, mm 155 127 101 225
Hub OD, mm 100 75 61 136
Setup angle, degrees 30 30 26 20
Casing dimensions Draft tube included angle, A, degrees 16 6 16 7
Vortex ange-to-outlet throat, B, mm 241 370 120 335
Main casing radius, RC
o
, mm 151 152 102 252
Scrolled casing radius exponential factor, RCp 0.0024 0.0020
Runner casing ID, D, mm 155 128 102 227
External draft tube length, E, mm 396 465
Inlet-to-runner centerline offset, F, mm 73 71 140 400
Outlet cone-to-draft tube gap, G, mm 6 13 5 0
Main casing internal length, H, mm 362 368 170 1000
Inlet centerline-to-vortex ange offset, K, mm 235 217 44 685
Vortex ange OD, L, mm 250 220 190 256
Inlet ID, P, mm 155 162 170 300
Vortex ange ID radius, RR, mm 5 5 5 5
Vortex ange-to-main casing clearance gap, S, mm 32 44 60 80
Outlet throat ID, T, mm 103 97 78 227
Vortex ange OD radius, RV, mm 5 5 5 8
Inlet diffuser plan width, W, mm 177 397
Inlet diffuser full length, X, mm 339 1455
Outlet cone included angle, Z, degrees 16 11 16 0
K.V. Alexander et al. / Renewable Energy 34 (2009) 3547 38
machine has been heavily inuenced by experimentation to obtain
a satisfactory efciency. Due to the complex three-dimensional
nature of the ow through the machine, applying conservation of
angular momentum does not reliably predict conditions at the
runner. However, the subsequent derivations will show the theo-
retical starting point for inlet ow prediction.
The turbine casing is a volume just upstream of the runner,
designed with an offset inlet such that the ow acquires angular
momentum about the runner axis while dissipating any circum-
ferential asymmetry before entering the runner. The spinning ow
approaching the runner has been assumed to form free-vortex ow
to satisfy radial equilibrium, meaning no radial velocity compo-
nents are present, and axial velocity is constant throughout the
uid. The latter property allows axial velocity to be written as
a function of known parameter Q, in a manipulated form of
Q V Area.
v
a

Q
p
_
r
2
t
r
2
h
_ (12)
Next, the circumferential ow velocity, v
c
, at any radius from the
runner axis is given by the well-known free-vortex relation, derived
from the line-vortex solution to radial equilibrium [7, p. 153]
v
c

k
r
(13)
This is the starting point for calculating runner blade geometry as
a function of site parameters, design constraints, and casing ge-
ometry. Eq. (13) is attractive because a known value for k will allow
runner design to begin.
The torque at any increment of radius r
j
is produced by reducing
the local circumferential velocity v
cj
to zero (or close to it). There-
fore, the torque at the jth increment of radius is given by
T
j
_ m
j
v
cj
r
j
(14)
The torque on the runner as a whole is simply the difference of the
angular momentum of the uid upstream and downstream. The
mass ow rate in terms of cross-sectional area through the runner
and axial velocity (which is constant pursuant to the free-vortex
assumption) is given by
_ m
j
rv
a
2pr
j
_
r
j
r
j1
_
(15)
Now, substitute Eq. (15) into Eq. (14).
T
j
rv
a
2pr
j
_
r
j
r
j1
_
v
cj
r
j
(16)
This difference equation can be rewritten as a denite integral,
where the difference term becomes the innitesimal radial in-
crement, dr. Remove all constants fromthe integral, substitute k for
v
cj
r
j
, and integrate radially from hub to tip:
T
r
rv
a
2pk
_
rt
rr
h
rdr rv
a
2pk
_
r
2
2
_rt
rr
h
krv
a
p
_
r
2
t
r
2
h
_
(17)
Combining all terms following k gives
T
r
k _ m (18)
On its own, Eq. (18), like Eq. (13) is not very helpful because k is
unknown. However, the shaft power calculated in this manner
equated with the shaft power calculated from potential energy and
efciencies allows a solution for the axial and circumferential ow
velocities just upstream of the runner.
The power produced by the turbine runner, P
r
is given by the
product of the runner torque, T
r
, and the runner rotational speed
P
r
T
r
N
60
2p (19)
But the power P
r
to be absorbed by the runner is the potential
energy in the ow, reduced by the turbine efciency
Table 2
Table 1 scaled to reference power and speed
Specs when lab models are scaled to a speed/power reference
Scaled design parameters P
r
, kW 1 1 1 1
N, rpm 1500 1500 1500 1500
N
S
176 242 355 544
H
t
, m 5.56 4.31 3.17 2.25
Q, l/s 25 32 46 66
L
R
0.832 1.035 1.411 0.81
h
t
0.727 0.747 0.727 0.665
Scaled runner Runner OD, mm 129 131 143 182
Hub OD, mm 83 78 86 110
Scaled casing dimensions Draft tube included angle, A, deg 16 6 16 7
Vortex ange-to-outlet throat, B, mm 201 383 169 271
Main casing radius, RC, mm 126 157 144 204
Scrolled casing radius exponential factor, RCp 0.0024 0.0020
Runner casing ID, D, mm 129 132 144 184
External draft tube length, E, mm 329 481
Inlet-to-runner centerline offset, F, mm 61 73 198 324
Outlet cone-to-draft tube gap, G, mm 5 13 7 0
Main casing internal length, H, mm 301 381 240 810
Inlet centerline-to-vortex ange offset, K, mm 196 225 62 555
Vortex ange OD, L, mm 208 228 268 207
Inlet ID, P, mm 129 168 240 243
Vortex ange ID radius, RR, mm 4 5 7 4
Vortex ange-to-main casing clearance gap, S, mm 26 46 85 65
Outlet throat ID, T, mm 86 100 111 184
Vortex ange OD radius, RV, mm 4 5 7 6
Inlet diffuser plan width, W, mm 250 322
Inlet diffuser full length, X, mm 478 1179
Outlet cone included angle, Z, deg 16 11 16 0
K.V. Alexander et al. / Renewable Energy 34 (2009) 3547 39
P
r
_ mgHh
t
(20)
Using Eq. (18), (19), and (20), and rearranging to solve for k
k
gHh
t
2p
_
60
N
_
(21)
Substituting Eq. (21) back into Eq. (13) gives the equation for the
circumferential velocity, v
c
, at any radius, r, without knowing the
value of k a priori.
v
c

1
r
gHh
t
2p
_
60
N
_
(22)
Now both axial and circumferential ow velocity into the runner
are known. From this, the angle of the ow can be determined at
any radius into the runner annulus, allowing the design of blades to
match.
4. Runner blade theoretical design
Remaining consistent with this projects idea of simple con-
struction, the runner is made of a cylindrical hub with at blades
attached at a specied angle, j. Each blade is set up as shown in
Fig. 4, so that the hub radius through a reference point O also lies
in the plane of the blade. Therefore, at point O, the blade also lies
at the set-up angle, j, to the impeller disk. As a consequence of this
arrangement, the blade angle, b, varies from j where q 0, to
something less than j at any angle q, in either direction, around the
hub from point O. b is dened by the relation
b tan
1
tanjcosq (23)
This relation shows that at q 0, b j, and conrms that re-
gardless of the value of j, b goes to zero when absqgoes to 90
degrees.
The runner is designed to spin in the casing at a speed com-
patible with the generator, nominally 1500 rev./min. The hub size
can be varied by the designer taking into account several issues
namely:
1. The circumferential ow at the hub, as given by Eq. (22), must
be less than the velocity ur at the hub in order to obtain a non-
negative angle of attack along the runner leading edge.
2. Too large a hub-to-tip ratio, while addressing the rst issue,
results in a high axial ow velocity in the annulus, creating
excess friction losses and abrupt expansion losses as the hub
truncates just downstream of the blades.
3. Too high a hub-to-tip ratio also creates short-span, highly
loaded blades with insufcient area to resist cavitation; this
requires numerous blades
While addressing the points listed above, a hub-to-tip ratio of
about 0.6 has been found appropriate. With the hub-to-tip ratio set,
a velocity triangle can be drawn at any point on the leading edge of
a runner blade, based on the circumferential velocity, v
c
, the runner
velocity, ur, and the axial velocity, v
a,
as shown in the velocity tri-
angle diagram, Fig. 5. Fromthis, the shock-free (angle of attack 0)
leading edge blade angle may be determined. The required angle
(relative to the impeller plane) anywhere along the leading edge is
b
1
tan
1
_
v
a
ur v
c
_
(24)
The trailing edge angles are found from the exit velocity triangle
shown again in Fig. 5, which embodies the constraint that the
exiting ow should have no rotation (i.e. v
c
0). Due to losses ex-
perienced in the ow between the leading and trailing edges of the
runner, the exiting ow will not exactly follow the blades cam-
berline at the trailing edge [5, p. 325]. In order for the no-rotation
constraint to be obeyed in reality, the runner blades need to over-
turn the ow by a deviation angle, d, of about 5 degrees relative to
the runner compared to what the simplied two-dimensional
theory would dictate. To be precise, d will vary with r, and can be
calculated across the trailing edge [8, p. 31]. Therefore, the equa-
tions for blade angle relative to the impeller plane, at the hub and
tip, respectively, are
b
2
tan
1
v
a
ur 0
d (25)
Given Eqs. (23), (24), and (25) (which, together, embody the at-
blade concept), the free-vortex inlet assumption and the no-rotation
exit constraint (with the inclusion of d), the task is to nd a blade
shape where b is correct for all points along the blades leading and
trailing edges. Realistically, if b can be satised at the hub and tip at
leading and trailing edges, all intermediate points along the leading
and trailing edges can be found to coexist on a at blade, as well.
This is achieved by choosing a value for j, substituting the result of
Eqs. (24) and (25) into Eq. (23) as b, and solving for the angle q from
the reference point, where the angle b will exist. The four corners
of the blade can thus be dened as shown in Fig. 6. Likewise, values
of q for any value of r along the leading or trailing edge can be
calculated.

Runner Blade
Blade
Setup
Jig
Hub
Reference Point O
Hub
Fig. 4. The set-up arrangement for attaching the blade to the runner hub.
Fig. 5. Runner velocity diagram.
K.V. Alexander et al. / Renewable Energy 34 (2009) 3547 40
Using the equations above, several choices are available as to
how to form the blades. These are based on the choice of j. The
options are:
1. Choose the blade setup angle j to equal the largest angle of any
part of the blade. This angle is at the leading edge at the hub.
This point then is at the reference point O. The other corner
points are then found from Eq. (23) around the annulus as
shown in Fig. 6. The result is a blade that is rather small and has
a swept back trailing edge.
2. Choose the blade setup angle j to be steeper than the largest
angle required to match incoming ow at any point along the
leading edge. This is shown in Fig. 7. Again all corner angles can
be achieved, but the blade has an even smaller chord and
a similar swept back trailing edge.
3. Choosethebladesetupanglejtomatchtherequiredangleat the
tip at the leading edge. As shown in Fig. 8 this leaves the angle at
the hub at the leading edge too small; it may even leave the
trailing edge angle too small. The gure shows that at the hub,
the blade angle is lower than both the required leading edge and
Hub
Correct LE Angle
Correct TE Angle
0
5
10
15
20
25
30
35
40
Setup Angle = 34.1
Setup Angle = 34.1
Tip
Correct LE Angle
Correct TE Angle
0
5
10
15
20
25
0 20 40 60 80 100
0 20 40 60 80 100
Point
"O"
Blade
Hub

Fig. 6. The case where the blade set-up angle is equal to the required angle at the leading edge at the hub (at point O). All corner angles can be achieved.
Hub
Correct LE Angle
Correct TE Angle
0
5
10
15
20
25
30
35
40
Setup Angle = 41.0
Tip
Correct LE Angle
Correct TE Angle
0
5
10
15
20
25
0 20 40 60 80 100
0 20 40 60 80 100
Setup Angle = 41.0
Point
"O"
Hub
Blade

Fig. 7. The case where the blade set-up angle is bigger than the required angle at the leading edge at the hub. Again all corner angles can be achieved, but the blade is narrow.
K.V. Alexander et al. / Renewable Energy 34 (2009) 3547 41
trailing edge angles. (See below, as in many cases this has shown
to be the best choice due to compromises for vegetation shed-
ding.) Theswept leadingedgesheds vegetationintheow, but as
shown in the plots in the gure, makes the leading edge mis-
match worse, at least in this theoretical analysis.
4. Choose the blade setup angle j to be greater than the required
angle at the tip at the leading edge. As shown in Fig. 9 this
leaves the angle at the hub at the leading edge too small, but
with the peculiar-shaped triangular blade shown, the trailing
edge angle can be achieved. Again in the interests of achieving
Tip
Required LE Angle
Required TE Angle
0
10
20
30
Blade with Setup Angle = 19.2
Hub
Required LE Angle
Required TE Angle
0
10
20
30
40
0 20 40 60 80 100
-60 -40 -20 0 20 40 60 80 100
Blade with Setup Angle = 19.2
Trailing Edge Extension
Swept Leading Edge Extension
Hub

Calc
Blade
Swept Leading Edge Extension
Trailing Edge Extension
17
Fig. 8. The case where the blade set-up angle is equal to the required angle at the leading edge at the tip. Only tip corner angles can be achieved; the calculated blade is unable to
connect with the hub. A swept leading edge and trailing edge are added.
Hub
Reqired LE Angle
Required TE
Angle
0
20
30
40
-60 -40 -20 0 20 40 60 80 100
Blade with Setup Angle 23.8
Swept Leading Edge Extension
Trailing Edge Extension
Tip
Required LE Angle
Required TE
Angle
0
10
20
30
0 20 40 60 80 100
Blade with Setup Angle 23.8
Hub
Calc
Blade
Swept Leading Edge Extension
Trailing Edge Extension
Fig. 9. The case where the blade set-up angle is bigger than the required angle at the leading edge at the tip. Only tip corner angles can be achieved; the calculated blade can just
connect with the hub at its TE. A swept leading edge and trailing edge.
K.V. Alexander et al. / Renewable Energy 34 (2009) 3547 42
a sensible blade shape, a swept leading edge and extended
trailing edge can be added, and a workable solution achieved.
Plots in Figs. 8 and 9 show the resulting angle mismatches
which occur at the hub.
It is notable that there are actually several different congura-
tions which can be used to achieve suitable ow-matching angles
along both the leading and trailing edges of the runner blade. The
method chosen for this project, item 3 from the list above, is a de-
parture from the theory used to predict the ideal blade shapes,
when in fact it produces blade angles close enough to the ideal
values to give acceptable efciency and excellent vegetation
shedding performance. The leading edge swept to more negative q
(upstream) leads to an increase in angle of attack and the expec-
tancy of some turbulence on the suction side just downstream of
the leading edge. The added area at more positive q (downstream)
after the trailing edge overturns the uid trajectory leaving the
runner, potentially resulting in energy being returned to the exiting
ow. Details of the blade design methodology, including the choice
of setup conguration used throughout this project and the gov-
erning practical constraints, are given in Section 6.2.
5. Turbine casing design
A major thrust of the turbine development in this project has
been a radical departure from traditional Kaplan inlet congura-
tion. The new designs avoid the use of guide vanes or any other
ow-spanning structure that might catch leaves or vegetation. This
is necessary to eliminate the daily ritual of removing leaves and
blockages common to many microhydro sites. The lack of guide
surfaces has necessitated careful experimentation with casing ge-
ometry to achieve desired conditions at the runner.
The traditional Kaplan turbine uses guide vanes to give rotation
to the ow immediately upstream of the runner. In the case of
microhydro, the scale is sufciently small that vegetation catches
on the leading edges of such guide vanes and remains there,
blocking the ow. A particular motivation in this project has been
the solution to this problem. The general layout of the casing design
is shown in Fig. 10 and the design considerations have been the
following:
1. The casing must convert the energy in the ow into an appro-
priate circumferential velocity. To achieve this, the inlet has been
offset to one side. While this has allowed the removal of guide
vanes for a single set of operating conditions, it has required
considerable experimentation to arrive at an effective solution.
2. For good efciency it is important to create an axi-symmetric
circumferential velocity eld into the runner. To achieve this,
a narrowdisk-shaped owspace has been created between the
vortex ange and casing end-plate. In this region, the end
spacing (dimension S, Figs. 11 and 12) has been adjusted
empirically to achieve a steady increase in circumferential
velocity as the ow moves radially inward toward the runner.
In the bulk internal volume of the casing and in this region, any
signicant differences in circumferential velocity at a given
radius around the circumference will tend to be damped out
due to the uids viscosity.
3. A further device required in the higher specic speed cases, has
been to set up the inlet some distance from the runner so the
ow is required to undergo several revolutions before entering
the disk-shaped space, and is characterized by dimension K
in Figs. 11 and 12. This gives the larger asymmetrical ows an
opportunity to even out before entering the runner.
4. The ow eld must have a suitable rotational speed to drive
a runner at the required 1500 rev./min for the generator. To
achieve this, the size of the inlet piping and its offset have been
adjusted to achieve the circumferential velocity required for
the impeller and the casing diameters needed. The starting
point for the inlet size and its offset has been the required k in
Eq. (13) and the ow velocity from the penstock. While these
allow the calculation of an initial value of r (corresponding to
dimension F in Figs. 11 and 12) for the offset, nal dimensions
have been arrived at by experiment.
5. The draft tube efciency becomes increasingly signicant as the
specic speed gets higher; this component must also cope with
more energetic ow in the runaway condition. Design and ver-
ication of the draft tube has been an integral part of the ex-
perimental program. For the N
S
176 and 242 machines (low
N
S
), the majority of the draft tubes expansion is external to the
maincasing. Bycontrast, for theN
S
355and544machines (high
N
S
), the draft tube expansion occurs within the connes of the
main casing, exhaust elbows and other such downstream
plumbingbeingequal indiameter tothat at the maincasingexit.
6. A further task has been to design a casing structure than can be
easily built in regional workshops. This has been achieved with
casing concepts that use largely at plate with simple curves,
and components that can be made on a lathe.
7. The design has had to be such that the turbine can easily be
dismantled to clear any obstructions that manage to lodge in
the runner. This has led to the form of the casing being that of
the back-pull-out pump, where the shaft bearings and run-
ner assembly can be withdrawn without disturbing the inlet
and outlet pipe connections. To accomplish this, a spacer cou-
pling is necessary for the direct-connected transmission to the
generator and a design consideration is minimising the length
Penstock/
Flow in
SECTION B-B
SCALE 1 : 5
Draft tube/
Flow out
Runner assembly
Vortex flange
Shaft
Fig. 10. General layout sketch of the turbines.
K.V. Alexander et al. / Renewable Energy 34 (2009) 3547 43
SECTION A-A
G
B
H
S
E
K
A
R
V
T
R
R
D
L
Z
P
F
R
C
o
A
A
Fig. 11. Low N
S
casing dimensioning and orientation guide.
Z

R
R
R
V
R
C
o
H

B

F

S

G

W
L
D
K
T
A
SECTION A-A
A
A
P
X
Fig. 12. High N
S
casing dimensioning and orientation guide.
K.V. Alexander et al. / Renewable Energy 34 (2009) 3547 44
of the spacer coupling. Dowelling, or some equivalent, is also
important to ensure correct realignment on re-assembly.
8. The design also has to consider the forces of the adjacent pipe
work on the turbine and allowance made for sufcient strength
of the casing. An alternative is for exible pipe couplings which
are to be preferred, though are more expensive.
9. Consideration has had to be given to the runaway condition
which occurs, for example if there is a loss of electrical load to
the generator. In this instance there is the potential for cavi-
tation in the draft tube and destructive forces in the turbine. In
each design this eventuality has been addressed by a conr-
mation test, to verify the integrity of the system left in the
runaway condition for several hours.
10. Robustness of themachine is important tocopewithhandlingin
transit to the site and for installation with limited resources for
lifting. Depending on size, the machine may have to be dis-
mantled into small enough components for transport on foot.
11. Pressure testing of the turbine should be undertaken after
manufacturing and before installation. Proof testing in the
operational mode is normally out of the question.
Key tasks of the research programme have been the construc-
tion, modication and testing of casings to achieve these aims, the
adjustment of the disk-shaped space and verication of the ap-
propriate ow around the lip and into the runner. Fig. 13 shows
a sample of computational owmodelling undertaken for the entry
into the runner of the N
S
355 machine.
Generic casings for low and high N
S
machines are given in Figs.
11 and 12 respectively, with dimensions in Tables 1 and 2. Note that
the main difference is that the low N
S
designs have a simpler
spliced-pipe inlet to the main casing, whereas the high N
S
designs
have a less restrictive rectangular diffuser type inlet, with a fabri-
cated circle-to-square ange adapter at the upstream penstock
interface. In addition, the N
S
355 and 544 machines casings have
a scroll shape versus the tubular shape of the other two designs.
This introduces the RCp parameter seen in Tables 1 and 2, where
the N
S
355 and 544 casings radius from the runner axis is
RC RC
o
exp
RCpa
(26)
Table 1 contains the casing dimensions of those models tested in
the laboratory, while Table 2 contains the same four machines
scaled to the same 1 kW power output and 1500 rev./min for
comparison and use as reference points for further scaling. It is
noted that the actual manufacturing drawings are necessarily more
complex and detailed. These gures only give the dimensions of
essential features necessary for the casings functionality.
6. Compromises due to project context
The microhydro designs from this paper emphasize some con-
ceptual differences between large hydro and microhydro, mainly in
the form of compromises that are made to keep the schemes eco-
nomical and reliable with minimal maintenance.
6.1. Reduced efciency due to simple construction
One driver of this project is that the turbines should be easily
manufactured in regional workshops in third-world countries. This
has resulted in runners (Fig. 14) and casings designed for simple
fabrication using mild steel plate. Volumes and passages are
therefore not as streamlined as would be possible with more so-
phisticated manufacturing techniques. Some changes trickle down
to affect other components. For example, eliminating the com-
plexity of trash racks or other maintenance-intensive supporting
structures has led to the less-than-ideal blade shape illustrated in
Fig. 8 being utilized throughout. The requirement to freely pass
debris has likewise led to the elimination of intake umes, guide
vanes, wicket gates and the like. To be sure, some turbine efciency
has been compromised in the design process, but given the context,
a few percent of efciency has been deemed allowable as long as
the nal scheme produces the originally predicted power reliably.
Runner
centreline
Hub boundary
Leading edge
Trailing edge
Fig. 13. Axi-symmetric CFD modelling of ow into runner, showing meridional com-
ponents upstream of runner.
Fig. 14. The runners of some of the development turbines in this project: N
S
355 (left) and N
S
242.
K.V. Alexander et al. / Renewable Energy 34 (2009) 3547 45
Even with these manufacturing limitations, it has been an object
of the development program that every turbine should achieve an
h
t
of 70% efciency. The N
S
544, the least efcient model, operates at
a maximum h
t
of 68%, while the N
S
242, the most efcient model,
operates at a maximum h
t
of 74% (Fig. 15).
6.2. Runner blade practical design
In developing runners for this project it was decided early on
that options 1 and 2 from the list in Section 4 gave blade shapes
that, while possible, were unsatisfactory in that their chord was too
short, the area per blade too small, and many more blades would be
required to reduced blade loading to levels below those causing
cavitation. The starting point for experimentation has been option
3 shown in Fig. 8. The arguments in favour of this choice are:
A CFD study of the 3D ow into the runner annulus shows that
there is a considerable radial component tothe ownear the hub
where the leading edges are located (see Fig. 13). The effect is to
reduce both v
a
and v
c
in Fig. 5, lowering the required leading
edge angle at the hub to a value much closer to what is actually
built. As a consequence the blade angle mismatch at the hub is
not as signicant as shown in the theoretical model of Fig. 8.
These blades have been found by experiment to work as
required [9, p. 141143].
They are a reasonable area, so that a small number of blades is
sufcient; this makes manufacture easier.
The hub-to-tip ratio is reasonable, being close to that of Kaplan
turbines.
Efciencies of 70% are achievable; the losses introduced by the
compromises are perhaps less of a problem than building and
mounting multiple, small, odd-shaped blades.
The added logarithmic spiral leading edge works effectively in
shedding vegetation, though it may cause some small stalled
regions near the hub so some losses will result.
The mismatch of the angles at the trailing edge effectively adds
more deviation (d) to the blades and consequently has the
potential to create a small reverse circumferential velocity in
the core of the discharge ow. Again this loss appears small.
Experimental work with leaves and the swept leading edge has
shown that a constant spiral angle of about 17 degrees is required
as shown in Fig. 8. The leading edges should be sharpened, with
material taken off the suction face; the sharp edges can be seen in
some cases in the photographs of Fig. 14, the same as is represented
on the leading edge of the prole shown in Fig. 5. The process of
shedding vegetation is that any leaves wrapping themselves over
the leading edge are either cut immediately or they slide out to the
tip where they catch between the tip and the casing and are sliced
up by the sharp edge.
Propeller Turbine Model Hydraulic Efficiency
50.0%
60.0%
70.0%
80.0%
0 100 200 300 400 500 600
Specific Speed, Ns
H
y
d
r
a
u
l
i
c

E
f
f
i
c
i
e
n
c
y
Ns 176
Ns 242
Ns 355
Ns 544
Fig. 15. Tested model efciency across propeller turbine range.
-100
-80
-60
-40
-20
0
20
40
60
80
0 20 40 60 80 100 120 140
Labaratory Models
-100
-80
-60
-40
-20
0
20
40
60
80
0 20 40 60 80 100 120 140
References: 1 kW, 1500 RPM
Ns 176
Ns 242
Ns 355
Ns 544
Ns 176
Ns 242
Ns 355
Ns 544
Fig. 16. Runner blade true shapes for the four specic speeds.
K.V. Alexander et al. / Renewable Energy 34 (2009) 3547 46
An important consideration in the manufacture of the runners is
the correct attachment angle of the blades. In Fig. 4, the xture for
mounting the blades is shown, as is the point on the blade that
must be lined up on the radius through the reference point O on
the hub. The dimensions for the runners for the four specic speeds
are shown in Table 1, and the templates for the blades at laboratory
model and 1 kW reference scale are shown in Fig. 16. Note that
although the hub-to-tip ratio is not constant across all four designs,
the runners swept diameter increases with specic speed at con-
stant power output, conrming the prediction that higher specic
speed machines, and therefore lower power-density machines, are
necessarily larger when compared to low-specic speed machines.
6.3. Higher cost due to low power density of low head
For sites of a given power, as the site head reduces, the ow
must increase. This is indicated by Eq. (2). This means that as site
head reduces, the penstocks [3] and turbines need to become larger
to carry the increased ow, and although the penstocks are shorter,
the larger components inevitably make the installations more ex-
pensive. So, on a per kilowatt basis, the installations for these low
head sites are quite likely more expensive than higher head sites.
This is one of the reasons why this low head area has remained
undeveloped. Note that in the commercial sector, low head appli-
cations such as tidal barrages and marine stream power generation
cannot typically compete with higher head hydro plants. One no-
table exception may be Toshibas Hydro-eKIDS scheme, marketed
for use in irrigation channels [10]. Therefore, it must be appreciated
that at a lowhead site, the installation is going to be more costly (in
$/kW terms) than at a higher head site of equivalent power.
7. Conclusions
These four turbine types have been built and tested, at one scale.
The test unit efciency results for each machines nal conguration
are given in Fig. 15 showing that a nite number of designs may be
used to cover a broad range of sites. Nominally, there is sufcient
information in this paper for these propeller turbines to be built, in
essence covering 28 of the possible 38 units shown in Fig. 3.
However, at this stage working drawings are only available at the
four scales of the tested models and the one full-scale machine that
has been implemented in practice. Work is proceeding on further
designs.While it is expected that slightly better efciencies could be
achieved with more renement in the runner design, this paper
presents practical, working, and economic turbine solutions for low
head microhydro systems, especially when they are matched with
the modular components described in the other papers in this
series [2,3,11].
References
[1] Blakely RJ, OConnor KF. Present and potential use of micro-hydro-electric
schemes in remote locations. October. Auckland: New Zealand Energy
Research and Development Committee 1981.
[2] Alexander KV, Giddens EP. Microhydro: cost-effective, modular systems for
low heads. Renewable Energy 2008;33(6):137991.
[3] Alexander KV, Giddens EP. Optimum penstocks for low head microhydro
schemes. Renewable Energy 2008;33(3):50719.
[4] New Zealand. Demonstration project prole: remote area power supply: micro
hydro/diesel hybrid, Project summary 22. Wellington: Energy Efciency &
Conservation Authority; 1994.
[5] Nechleba M. Hydraulic turbines: their design and equipment. Prague: ARTIA;
1957.
[6] White FM. Fluid mechanics. 5th ed. Boston: McGraw-Hill; 2003.
[7] Dixon SL. Fluid mechanics: thermodynamics of turbomachinery. 2nd ed.
Oxford: Pergamon Press; 1975.
[8] Hothersall R. Hydrodynamic design guide for small Francis and propeller
turbines. Vienna: United Nations Industrial Development Organization; 2004.
[9] Parker GJ, Faulkner SA, Giddens EP. A low head turbine for microhydropower.
RERIC International Energy Journal 1993;15(2).
[10] <http://www.toshiba.co.jp/f-ene/hydro/english/newtech/newproducts/doc5.
htm>.
[11] Bryce P, Giddens EP. Malfunction protection for electrical equipment used in
micro hydro plants. Water Power & Dam Construction Supplement 1985;
37(11):58.
K.V. Alexander et al. / Renewable Energy 34 (2009) 3547 47