Renewable Energy 121 (2018) 745e756

Contents lists available at ScienceDirect

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

Evaluation criteria for velocity distributions in front of bulb hydro

turbines
Roman Gabl a, b, c, *, Daniel Innerhofer a, Stefan Achleitner a, Maurizio Righetti c,
Markus Aufleger a
a
University of Innsbruck, Unit of Hydraulic Engineering, Austria
b
The University of Edinburgh, School of Engineering, Institute for Energy Systems, UK
c
Free University of Bozen, Faculty of Science and Technology, Hydraulic and maritime constructions and hydrology, Italy

a r t i c l e i n f o a b s t r a c t

Article history: General guidelines are available for the design of intake structures in river power plants. Nearly all
Received 5 February 2017 existing criteria are limited in scope to a (rectangular) control section near the trash rack. In this section,
Received in revised form a homogeneous flow with negligible wall influence is defined as the ideal condition. 3D numerics can
4 January 2018
simulate the complete velocity field up to the turbine, and therefore inform investigations of different
Accepted 11 January 2018
inflow structure variations. This paper presents a review of six existing criteria and a modification of the
Fisher-Franke criterion. All criteria are tested for both theoretical pipe flow conditions and artificial
biased velocity distributions, for which different simplified obstacles in front of a turbine are investigated
Keywords:
Kaplan
with the help of the 3D numerical software ANSYS-CFX. The best results could be achieved using the
Hydro power evaluation of the kinetic energy flux coefficient as well as the new modified criterion. Both can be
Design process recommended for the geometry optimisation of the intake structure.
Velocity distribution © 2018 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
3D numerical simulation (http://creativecommons.org/licenses/by/4.0/).
ANSYS-CFX

1. Introduction
1.1. Overview
An intake structure is the upstream connection of a river or
reservoir to a hydraulic system, in which the water is used for
drinking, cooling, or producing energy. For energy production,
different types of turbines are used, depending on the available
energy head and discharge [1]. Liu et al. [2] summarise the im-
provements over the last years in the field of hydraulic turbines,
which have overall led to a very high rate of energy production
efficiency with hydro power turbines. Further investigation has
focused on Kaplan turbines, which are used for comparable small
heights, and are sometimes used in combination with large dis-
charges as the first choice. These can be installed without addi-
tional structures, for example as tidal turbines [3e7]. However,
under normal conditions, the intake structure guides the water to
the rotor blades. Different investigations focus on the draft tubes
* Corresponding author. University of Innsbruck, Unit of Hydraulic Engineering,
Technikerstrasse 13, 6020 Innsbruck, Austria.
E-mail address: roman.gabl@icloud.com (R. Gabl).
(downstream the turbine) for vertical [8e12] and horizontal axis
turbines [5,13,14]. Energy generation can be further enhanced by
inlet guide vans [1,15]. For example, Fernando and Rival [16]
investigated different intake structures for very low head condi-
tions and Ferro et al. [17] focused on the influence of inlet guide
vans for mini hydro turbines, which should allow the use of hydro
power with low investment costs. In general, a well-designed
structure can help to reduce the construction costs associated
with hydro power plants, and should ensure good inflow condi-
tions, which help to guarantee turbine durability and high energy
production efficiency [18e21].
The main goal of intake structure design is to achieve good
hydraulic performance. Therefore, head loss should be minimised,
and the pressure line must decrease continuously with increasing
velocity in the flow direction [18,21]. The design should ensure that
as much water as possible can flow to the turbine, including as few
impurities as possible. Therefore, trash racks are installed to
remove floating debris or other incoming objects from the inflow
[22], as well as fish [23]. Based on built-in components [24,25] and
reservoir considerations [26e29], the amount of suspended sedi-
ment should be reduced as much as possible. An additional vital
topic is the prevention of swirls and air entrainment in the

https://doi.org/10.1016/j.renene.2018.01.027
0960-1481/© 2018 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
746 R. Gabl et al. / Renewable Energy 121 (2018) 745e756
structure [30e33]; these can lead to turbine damage or efficiency
reductions [20]. Under special conditions, ice plugging [34] may
also need to be considered. For water-pump intakes, additional
criteria must be applied [35,36], particularly if multiple intakes are
planned [37].
In addition to the experience and good practice of the designer,
two different primary tools can be used to investigate and optimise
such construction: (a) numerical simulations and (b) scale model
laboratory tests. The latter is often used to investigate sediment or
check if free surface vortices will occur. This method is limited by
scale effects [38,39], as well as the fact that preparing and testing
different geometries can be very costly and time intensive. Partic-
ularly for significant variations in geometry, the use of 3D numer-
ical tools is a vital alternative. Various previous studies and
validation experiments showed good method usability and accu-
racy [40e44]. A further advantage of the 3D numerics is that
distributed pressure and velocity results are available up to the
turbine at a very high resolution. This allows for the further veri-
fication and evaluation of existing criteria for fully developed pipe
flow conditions upstream of turbines, which are used as target
values for the optimisation of intake structures.
1.2. Key aspects
In general, distorted inflow conditions can cause reductions in
energy production and lead to vibration-induced turbine damage
[19]. Godde [45] compared different existing evaluation criteria
(Sec. 2.2). The main weakness in previous work was that nearly all
studies were limited to the cross section near the trash rack. This
control section is accessible in both scale model tests and nature.
For these areas, homogeneous flow conditions and very small wall
influences are characteristic. Ideal conditions, which must be
indicated by design criteria, differ in cases of downward optimi-
sation in this section; therefore, wall effects are no longer negli-
gible. This paper shows which existing criteria can be used for such
pipe flow conditions. It also presents a new modified criterion that
allows the control section to be moved near the turbine. In a second
step, artificial biased velocity distributions (VDs) are used to further
investigate each criterion. This allows for the extension of the ge-
ometry optimisation process up to the turbine and helps reduce
losses and increase energy production.
2. Methodology
2.1. Concept
Based on a literature review, seven widely used and well known
requirements from different turbine manufacturers are summar-
ised in Section 2.2 and a new modified criterion is presented in
Section 2.3. In a first step, a data set based on different theoretical
VDs in fully developed pipe flow conditions (Section 2.4) is used to
evaluate those criteria. The second part of the paper is focused on
biased VD in front of an exemplary simplified bulb turbine.
Therefore, different theoretical obstacles, which are presented in
Section 2.5, are investigated with the help of 3D numerical simu-
lations using ANSYS-CFX. A comparison of these standardised dis-
turbances with real inflow conditions can improve understanding
of each individual criterion and classify the actual VD in future
projects. The direct connection with possible reductions in energy
production is an on-going research topic. Therefore, further in-
vestigations and experiments on real turbines are needed.
2.2. Existing criteria
The inflow condition criteria for Kaplan turbines are not
standardised, and so each turbine manufacturer can individually
specify such criteria. The most common requirements can be
classified into the following groups of criteria [19,45]:

C0 e General conditions: Vortices, flow separation, air
entrainment, and rotation in the flow should be avoided.

C1 e Discharge: The complete section is typically divided in half
or in quadrants, and the local maximum deviation of discharge
in these parts should be smaller than 5% of the total discharge.

C2 e Angle of the velocity vector: The maximum deviation
measured from the axial direction should be smaller than 5+.

C3 e Cross flow: The orthogonal velocity components should
not extend 5% of the average velocity.

C4 e Velocity distribution: The deviation in local velocity
magnitude should be in a range of 5% or 10% of the average
value.

C5 e kinetic energy flux coefficient a: The correction factor for
the velocity head with respect to kinetic energy should be as
small as possible.

C6 e Fisher-Franke-criterion: The mean velocity of testing
sections is normalised by the global mean velocity. This value
should fall between the defined upper and lower boundaries,
depending on the size of the testing section (Table 1) required to
meet the criterion.
The kinetic energy flux coefficient a corrects the real kinetic
energy Ekin;real of an investigated control section in relation to the
theoretical value Ekin;theo . This value is multiplied with the velocity
head u2m =ð2$gÞ in the basic Bernoulli's equation to take into account
a non-uniformity of the velocity profile, and is calculated based on
the local velocity u and mean value um as presented in Eq. (1)
[46,47].
Z  3
Ekin;real 1 u
a¼ ¼ $ dA (1)
Ekin;theo A um
A
The coefficient a is always larger than or equal to 1, which
represents a completely uniform velocity profile. For turbulent flow
conditions, a value of 1.2 is typical. In the case of laminar conditions
a equals 2 (parabolic profile) [46]. These correction terms must be
considered, particularly for the calculation of local head losses with
different sections before and after loss [42]. For the present work,
the kinetic energy flux coefficient a (criterion C5) is also used to
check the density of evaluation points in the control section. This is
needed to generate the data set based on the theoretical velocity
distributions (Sec. 2.4).
Fisher and Franke [19] proposed criterion C6 to identify
acceptable velocity profiles based on various scale model tests.
Therefore, different testing areas of the control section in the intake
structure are cut out, and a local mean value of the velocity in this
smaller testing area is calculated. This local mean value um;part is
normalised by the global mean value um for the complete control
section. All testing areas result in a value of 1 [-] for a perfect flow
condition. Different boundaries Blo and Bup are defined depending
on the percentage of each investigated local testing area with
respect to the total control section. Table 1 shows the boundary
values defining the conditions (Eq. (2)) that satisfy criterion C6 in
Table 1
Boundary conditions for Fisher-Franke-criterion [45].
Apart =A [%] 0 20 50 100
upper value Bup [] 1.25 1.1 1.05 1.05
lower value Blo [] 0.75 0.9 0.95 0.95
 R. Gabl et al. / Renewable Energy 121 (2018) 745e756 747

the literature [19,45]. Specific values are similar to criteria C1 and (rectangular) trash rack sections.
C4. In between, a linear interpolation can be applied (Figs. 4e6). Based on the original criterion C6 by Fisher and Franke [19], a
Hence, the boundaries are larger for smaller tested areas. new criterion C6mod is proposed, which is only used to evaluate
complete sectors (Fig. 1 (c)) as testing areas. The only degrees of
um;part
Blo   Bup (2) freedom that are used are the starting and ending values for the
um angle q in polar coordinates. Thereby, near-wall effects are included
in each testing area, in a comparable way to the complete control
section. The results and advantages of this modification will be
shown in Section 3.1 for theoretically perfect velocity distributions
2.3. Modified Fisher-Franke-criterion
and in Section 3.2 for biased VDs.

The control section for the investigation was located by Fisher

2.4. Theoretical VDs
and Franke [19] downstream of the trash rack and upstream of the
turbine. The control section for this investigation is typically
Existing criterion C6 specifies constant VDs as the ideal condi-
located near the trash rack. At that location, accessibility is ensured
tion; the criterion is focused on the intake section near the trash
for scale model tests, particularly for real projects. Consequently,
rack. Hence, the velocities observed in this area are smaller
any changes downstream of this area are not included in these
compared to those near the turbine, and overall the boundary ef-
observations. The measurement of a detailed velocity distribution
fects can be neglected. Close to the turbine, and in the case of pipe
(VD) in front of a nature-scale bulb turbine is very cost-intensive.
flow, this assumption cannot be maintained. To test the presented
With the help of 3D numerical simulations, the complete inflow
criteria, four different theoretical VDs are used (see Table 2 and
channel can be evaluated and optimised in detail. Especially for
Fig. 2) to generate a data set of velocities. The first, called VD1, is a
small usable heads, the reduction of losses can help increase energy
uniform VD in the complete control section. As mentioned above,
production. To fully use the advantages of numerical simulations,
this is not realistic for pipe flow but is used as a reference. More
the control section should be defined as close to the turbine as
suitable for VDs in front of turbines are the other three distribu-
possible. In this case, the investigated cross section area changes to
tions, which are a laminar (VD2) distribution and two different
a circular profile, in contrast with a typical rectangular trash rack
exemplary turbulent profiles [46,48]. In all cases, rotational sym-
cross-section [47]. Fig. 1 shows such a pipe cross-section, as well as
metry is assumed. Therefore, the local velocity u depends only on
different exemplary testing areas that can be used. Normally for the
the distance r to the centre line where the maximum value of umax
Fisher-Franke criterion, C6 rectangular testing areas are randomly
occurs. To ensure that each VD has the same discharge Q, an
cut out of the complete control section. If the same method is used
additional correction factor b (¼ umax =um ) is introduced, which
for circular planes, the analysed area can be located partly outside
connects the mean velocity um of VD1 with the specific maximum
the cross section. As an example, Fig. 1 (a) shows one testing area in
umax for each VD (Table 2).
which only the grey shaded part contains non-zero values. To avoid
For further investigations and the use of these VD, a discharge Q
this reduction, the logical step is to change the coordinate system
of 3.5 m3/s is defined, as well as a maximum radius R of a pipe with
from Cartesian to polar (Fig. 1 (b)). Then, the complete randomly
1.3 m (similar to the simplified model, which will be presented in
chosen investigated testing area will be located inside the flow
Section 2.5). These typical values for a bulb turbine lead to a mean
field. As shown in Section 3.1, based on such evaluations areas, the
velocity um of 0.6592 m/s, which is also used to generate the
targets for criterion C6 cannot be reached for fully developed ve-
exemplary distribution in Fig. 2.
locity distributions in pipes (Sec. 2.4). This is mainly caused by pipe
As a preliminary step to test the criteria, the chosen VD func-
flow boundary effects, which are often negligible for bigger
tions are used to generate a discrete data set of local velocity u
values; this data set is comparable to numerical results or mea-
surements. Therefore, a polar approach is used, and the complete
pipe is divided into constant steps Dr for radius r, which ranges
from 0 to the pipe radius R, and Dq for angle q (from p to p). The
grid generated in this way contains a specific number N of uni-
formly distributed locations, which cover the complete control
section. An additional variable f is introduced to investigate
different grid densities. If the value of f is equal to 1, each centi-
metre and degree has one point (Eq. (3)). Fig. 2 was generated with
a grid density f equal to 0.2.

1 p
Dr ¼ ½m and Dq ¼ ½rad (3)
100$f 180$f

Table 2
Theoretical VDs.

case specification local velocity b

VD1 constant uc ¼ umax;c ¼ um 1
VD2 laminar   2  2
ul ¼ 1  Rr $umax;l
VD3 turbulent  17 60
ut1 ¼ 1  Rr $umax;t1 49

VD4 turbulent  101 231

ut1 ¼ 1  Rr $umax;t2 200
Fig. 1. Different types of testing areas in relation to a full pipe control section.
748 R. Gabl et al. / Renewable Energy 121 (2018) 745e756

Fig. 2. Theoretical VDs generated based on density value f ¼ 0.2 [-] and mean velocity um ¼ 0.6592 m/s.

The kinetic energy flux coefficient a (criterion C5, Section 2.2) is The runner and its possible effects on the flow in front of the
calculated as an evaluation criterion of the quality of the generated turbine are neglected in the chosen test set-up to reduce the
data set of velocities u. The numerical integration is performed with calculation time. All other components are used based on a real
MATLAB [49] and Table 3 shows the a-value for each VD (Table 2) in world case. Fig. 3 presents the dimensions of the investigated ge-
relation to parameter f. The results for VD1 and VD2 are by defi- ometry and the default values of the obstacle in front of the turbine.
nition 1 and 2, respectively, which can be achieved by using a f The main pipe in front of the turbine has a diameter D of 2.6 m, and
larger than 1. The changes for the other two VDs are also very small; its discharge Q is defined as 3.5 m3/s. Both values are based on a real
as a compromise based on calculation speed, the accuracy value of turbine investigated at the University of Innsbruck to ensure real-
f is set to 5 for the evaluations in Section 3.1. istic flow conditions (um ¼ 0.6592 m/s, ReD ¼ 1:2$106 ). The length
of the upstream pipe allows uniform flow conditions to be achieved
2.5. Biased VD in front of the obstacle. Downstream of the simplified turbine, a
pressure outlet at 105 Pa ensures that the complete model is free of
In a second step, a simplified model of a bulb turbine is cavitation under any investigated conditions.
investigated with the help of a 3D numerical simulation. The The Shear Stress Transport (SST) turbulence model is chosen to
software ANSYS-CFX [48] is chosen, because this commercial solve the closure problem of the Reynolds-averaged Navier-Stokes
software is well validated for free surfaces [26,50], pipe flow (RANS)-equation [4e6, 14, 43, 55]. This well known turbulence
[8,43], and further investigations [51,52]. The optimisation of approach combines a k-ε with a k-u-model with the help of a
turbines [3,11,53,54,44,55,56] and turbomachinery blades [57,58] blending function and allows to use the advantages of both tur-
is a typical application area of this software. The complex meshing bulence models [62,63]. For all conducted numerical simulations
capabilities and full geometry parametrisation allows rapid vari- the steady-state solver delivers stable results. The investigated fluid
ation, which is the main reason this software is chosen. The is simplified as incompressible water with a constant temperature.
simulated velocities are imported and evaluated in MATLAB [49], In a first verification step, the selected numerical model is tested
like the theoretical VDs. This numerical computing environment is based on quality standards, and a high independence of all user
exemplarily used to solve complex equations [59], conduct tran- inputs (calculation grid and boundary definitions) could be ach-
sient analyses [60], or optimise blade geometry [54,61]. ieved. All numerical simulations are conducted with double pre-
cision to reduce rounding errors. Additional monitor points for
pressure, velocity, and global loss are defined as target values.
Table 3
These are observed throughout the calculation process to ensure
Evaluation based on criterion C5 for VD1 to VD4, depending on the grid density
variable f. that steady state conditions in the complete domain are reached,
and a stable solution is provided for post-processing.
f N aVD1 aVD2 aVD3 aVD4
The chosen obstacle is a local limitation of the cross section,
0.1 4.7Eþ02 1.011 2.020 1.192 1.167 which causes a back flow region and (as a result) a non-uniform VD.
0.2 1.9Eþ03 1.003 2.005 1.117 1.092 This type of set-up with simplified obstacles is similar to that in
0.5 1.2Eþ04 1.000 2.001 1.079 1.053
1 4.7Eþ04 1.000 2.000 1.068 1.042
some of the tests conducted by Fisher and Franke [19] to develop
2 1.9Eþ05 1.000 2.000 1.063 1.036 the original criterion C6. The biased VD passes five chosen control
5 1.2Eþ06 1.000 2.000 1.060 1.033 sections E1 to E5 (Fig. 3), which are later used for the application of
10 4.7Eþ06 1.000 2.000 1.059 1.032 the criteria. The last section E1 (in flow direction) is located directly
20 1.9Eþ07 1.000 2.000 1.059 1.031
at the narrowing cross-section. For the investigation presented in
 R. Gabl et al. / Renewable Energy 121 (2018) 745e756 749

Fig. 3. Configuration for the generation of the biased VDs based on simplified obstacles.

Section 3.2 the main variation parameters are the end distances of 3. Results
the obstacle to the downstream control section E1 (represented by
the value EK) and the centre line (EA). In general, when these two 3.1. Theoretical VD
values are smaller, they have a larger influence on control section
E1. The thickness of obstacle DK has a minor effect on flow condi- The four simple theoretical VDs defined in Section 2.4 are used
tions, and is therefore fixed with 0.5 m, which allows a good mesh to generate a data set of velocities u. All of the criteria are fulfilled
quality for the complete variation of parameters EA and EK. for VD1, which is only included as a reference. The criteria C0 to C3
In this study, parameter EA or EK (Fig. 3) is fixed in four different (referring to the description in Section 2.2) are met for all theo-
positions, while the other is varied independently. In the case of retical VDs by definition because of the perfect rotational symmetry
varying EA (Table 4; first four rows), typically 27 single simulations and absence of cross flow.
are used to evaluate a range between nearly 0 (obstacle is in the Criterion C4 compares the local observed velocity u to the mean
middle of the pipe and approximately half of the section is blocked) value um of the complete section. The deviation should not extend
and 1:04$R (¼1.35 m), where no obstacle influences the flow. This beyond a range of 5%e10% for um. For further consideration, only
value is chosen to be larger than the pipe radius R to ensure that the the smaller value of 5% is used, but it can similarly be calculated for
geometry generation in ANSYS, which includes obstacle subtrac- a higher percentage. To find the boundaries bC4;u and bC4;l , within
tion, does not lead to a singularity. This state is used to check the which the criterion is fulfilled, the maximum allowed deviation is
simulation and analyse the process under uniform flow conditions. set equal to the theoretical VD definition. An example for VD2 is
The upper boundary for the variation of EK is 9:5$R which is equal calculated as follows:
to 12.35 m. The obstacle is not moved in the direction of the control   r 2 
section E1 closer than one pipe radius R. For each fixed value of EA,
ul ¼ 1 $ um $bVD2 ¼ ð1±0:05Þ $um (4)
18 single simulations are conducted. Table 4 summarises the input R |fflfflfflfflfflffl{zfflfflfflfflfflffl} |fflfflfflfflfflffl{zfflfflfflfflfflffl}
values for the geometry study investigated with 3D numerical umax;l boundary
simulation. The complete VD of each chosen combination is
exported, including the velocities in all three directions (u, v and w) sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
and location (x, y and z-coordinate). r bVD2  1H0:05 2  1H0:05
0 ¼ ¼ (5)
The presented criteria (Sec. 2.2), including the new modified R bVD2 2
|fflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflffl
ffl}
criterion C6mod (Sec. 2.3), are evaluated for usability based on these bC4;u  or bC4;l
biased VD. The main goal is to find the conditions for which each
Given the non-slip conditions of real fluid in a pipe, only a
criterion is barely fulfilled, thereby obtaining deeper insight into
limited portion of the control section meets criterion C4. Table 5
how each criterion reacts to an artificial obstacle in the inflow
presents this narrow range for r=R defined by the lower boundary
section. In a further step, the direct connection between efficiency
bC4;l and upper boundary bC4;u (Eq. (5)) as well as the difference
(produced electrical energy) and the criteria should be investi-
between them. The range for VD2 is small in comparison to VD3 and
gated. This is part of ongoing research work.
VD4. The conclusion can be drawn that this criterion C4 is not
applicable for real pipe flow conditions.
Table 4 The sensitivity of criterion C5 (kinetic energy flux coefficient a)
Overview of variation of variables EA and EK.
with respect to the density of evaluation points has been assessed
Identifier EA EK range in Table 3. It can be observed that criterion C5 is practically
EAvarEK2R var 2$R 0 to 1:04$Ra
EAvarEK4R var 4$R 0 to 1:04$Ra
EAvarEK6R var 6$R 0 to 1:04$Ra Table 5
EAvarEK8R var 8$R 0 to 1:04$Ra Upper and lower boundary ratios meeting C4 e 5%.

EKvarEA000R 0$R var R to 9:5$R boundary VD1 VD2 VD3 VD4

EKvarEA025R 0:25$R var R to 9:5$R
bC4;u 1 0.7246 0.8308 0.8583
EKvarEA050R 0:58$R var R to 9:5$R
bC4;l 0 0.6892 0.6591 0.6145
EKvarEA075R 0:75$R var R to 9:5$R
a
All values  R indicates no obstacle. bC4;u  bC4;l 1 0.0354 0.1717 0.2438
750 R. Gabl et al. / Renewable Energy 121 (2018) 745e756
independent for f values higher than 5. Therefore, this f value is
chosen for further evaluations.
As presented in Section 2.3 and shown in Fig. 1 (a), the typical
testing area for C6 is a rectangular probe out of the complete sec-
tion. Obviously, the satisfaction of criterion C6 will depend both on
the size and on the position of the testing area. In order to check
this dependence, a series of 10000 testing areas randomly spanning
the control section have been generated in MATLAB. Such a
randomly chosen area is only used if the marked grey area Apart in
Fig. 1 (a) is bigger than 0.5% of the complete section A. For each
testing area, the mean value um;part of velocity u is calculated. This
value is normalised by the global mean value um and compared
with the given boundaries (Table 1). Fig. 4 shows the results of this
evaluation based on criterion C6 with rectangular testing areas. As
expected, VD1 also meets criterion C6 perfectly, but all other VD fail
to do so. In particular, for the laminar flow condition VD2, the
distribution is far outside the given boundaries. Additionally, tur-
bulent VD3 and VD4 cannot meet criterion C6. Switching the co-
ordinate system for the investigation from Cartesian to polar does
not change this result, as shown in Fig. 5. In this latter case, again
the evaluation areas are randomly chosen as given in Fig. 1 (b). In
both approaches, the testing area can be focused solely either on
the higher-velocity centre or near the wall. Especially for VD2, the
results fall outside both boundary conditions. For the turbulent
VDs, the wall effects with slower velocities cannot meet the crite-
rion. Hence, a modification of the criterion is needed to allow for
the evaluation of real flow conditions in a pipe. Criterion C6mod only
uses complete segments of the profile, and therefore always in-
cludes partitions of the control section. The similar test for the four
theoretical VD based on criterion C6mod is shown in Fig. 6. In this
particular case, all theoretical VD fully meet the criterion. There-
fore, this approach can be further tested on biased VD.
3.2. Biased VD
As a second step, the presented criteria (Sec. 2.2) are tested on
artificial biased VDs. Therefore, a numerical model is used that
includes a simplified bulb turbine and variable obstacles (Sec. 2.5).
The evaluation of the biased VD using criterion C0 is focused on
the investigation of a back-flow region. Therefore, each location in
the control sections E1 to E5 is checked for whether the flow di-
rection of the local component u is oriented concordant to the main
flow direction. The number of reversed flow points is divided by the
total number (approximately 1600) and presented in Fig. 7
depending on the parameters EA and EK. Hence, a value of
Fig. 4. Evaluation of the criterion C6 for the theoretical VD1 to VD4 e rectangular evaluation section as shown in Fig. 1 (a); 10000 iterations; Apart > 0.5% A e including upper and
lower boundaries (Table 1).
0 indicates a VD without back-flow at this specific control section,
which should at least be reached for simulations without an
obstacle (EA=R > 1, upper row). In the lower row of Fig. 7 the results
for the variation of EK are presented. The vertical line indicates the
location where the chosen obstacle parameter EK intersects the
evaluated control section. If the obstacle ends upstream of the
control section, then significant back-flow is introduced, particu-
larly for small values of EA. This evaluation concept is also used for
the other criteria, with a small modification for C5, for which the
evaluated a-value is used.
For the evaluation of criterion C1, the frequently used quadrant
method C10:25 (Fig. 8) is used, as well as the evaluation of half of
section C10:5 (Fig. 9). Therefore, either a quadrant or half of the
cross section is chosen as the testing area. The starting angle for this
testing area is varied with a step of 1, so that the complete control
section is evaluated. The relative discharges, Q0:25 and Q0:5 are
calculated from the velocity distribution obtained by numerical
integration. These discharges are made dimensionless, dividing by
flow rate Q. If the deviation is larger than 5%, the section does not
fulfil the criterion. The evaluation of C1 in Figs. 8 and 9 shows that
the criterion is either fulfilled or nearly all evaluation segments are
outside of the boundary.
In the evaluation of both criteria in Figs. 10 and 11, similar
behaviour is shown. In particular, it must be highlighted that sec-
tion E1 is not effective for the application of criteria C2 and C3,
because narrowing (which starts in this section) inevitably induces
non-zero components in the plane orthogonal to the main direc-
tion. For the other control sections (E2 to E5) the criteria are fulfilled
for VD without an obstacle (EA=R > 1).
For criteria C2 and C3, a velocity component velrot is needed,
which is the component orthogonal to the main flow direction u
calculated based on Eq. (6).
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
velrot ¼ v2 þ w2 (6)
The velocity u is used for criterion C4, for which only the smaller
boundary of 5% of um is tested (Fig. 12). Similar to the evaluation of
the theoretical VD in Section 3.1, only a few points are in the range
of the criterion, which is nearly independent of the used obstacle.
In Fig. 13, the kinetic energy flux coefficient a (criterion C5) is
presented. Additionally, for the case with no obstacle (EA=R > 1), the
value in the control section E1 is higher than it is expected to be for
a VD. This effect, which is caused by narrowing is similar to C2 and
C3. The highest values are reached for obstacles are close to the
control section (the lower row of the Fig. 13).
The evaluation of the Fisher-Franke criterion is limited to
 R. Gabl et al. / Renewable Energy 121 (2018) 745e756 751

Fig. 5. Evaluation of criterion C6 for theoretical VD1 to VD4 e polar evaluation section as shown in Fig. 1 (b); 10000 iterations; Apart > 0.5% A e including upper and lower
boundaries (Table 1).

Fig. 6. Evaluation of criterion C6mod for theoretical VD1 to VD4 e only complete segments are used as evaluation sections as shown in Fig. 1 (c); 10000 iterations; Apart > 0.5% A e
including upper and lower boundaries (Table 1).

Fig. 7. Evaluation of criterion C0 depending on the variation of EA and EK e 0 indicates that no back flow occurs and 1 represents a complete revised flow direction.
752 R. Gabl et al. / Renewable Energy 121 (2018) 745e756

Fig. 8. Evaluation of criterion C10:25 based on quadrants depending on the variation of EA and EK e 0: criterion fully fulfilled and 1: all quadrants have a Q-deviation bigger than 5%.

Fig. 9. Evaluation of criterion C10:5 based on one half, depending on variations in EA and EK e 0: criterion fully fulfilled and 1: all sections have a Q-deviation bigger than 5%.

modified criterion C6mod , which uses only complete segments
(Fig. 1 (c)). The results in Fig. 14 are based on the main velocity u and
10000 iterations, similar to the evaluation of the theoretical VD in
Fig. 6. Criteria C6mod and C1 show comparable behaviour, especially
for larger values of EA between the obstacle and the centre line.
C6mod , C1, and C5 also indicate that an obstacle is present in front of
the evaluated control section.
4. Discussion
Different criteria are evaluated to assess the VD in front of a
horizontal bulb hydro turbine. Compared to a typically used control
section near the trash rack, in such a pipe, near wall effects on flow
conditions are not negligible. The testing of general conditions
(criterion C0) can be fully implemented in a 3D numerical simu-
lation. However, it is difficult to make judgements based on crite-
rion C0, when two nearly identical (already optimised) inlet
structures are to be compared. Hence, further criteria are needed.
In a first step theoretical VD for pipe flow conditions, (Section
2.4 and 3.1) are investigated. Criteria C0 to C3 (Sec. 2.2) are fulfilled
by definition. For further evaluations, velocity data sets are gener-
ated with the help of a polar grid. The density of used locations is
evaluated with criterion C5. It could be found that a comparably
high number of data points are needed, so that the results are in-
dependent of the chosen grid spacing (Table 3). Criteria C4 and C6
are not recommendable to be used under pipe flow conditions. A
 R. Gabl et al. / Renewable Energy 121 (2018) 745e756 753

Fig. 10. Evaluation of criterion C2 depending on the variation of EA and EK e 0 indicates that all angles fulfilled the criterion.

Fig. 11. Evaluation of criterion C3 depending on the variation of EA and EK e 0 indicates that all velrot components fulfilled the criterion.

new criterion is introduced, defined as C6mod (Sec. 2.3). This crite-
rion uses complete sectors, instead of the rectangular testing areas
recommended in C6. Therefore, both the higher velocities in the
centre and the lower values near the wall are represented in the
area. This modification allows the application of the well-known
Fisher-Franke criterion to be extended to identify ideal pipe flow
conditions.
Artificial biased VDs are investigated with the help of ANSYS-
CFX (Section 2.5 and 3.2). This numerical study is limited to one
exemplary turbine geometry; it is assumed that the effects of this
choice on the presented findings are negligibly small. Further tests
on this would be advisable, as would those on other simple ob-
stacles causing the disturbance.
It should be highlighted that the choice of location for the
control section can affect the usability of the individual criterion.
For example, the VD in control section E1 (Fig. 3) is influenced by
narrowing, which occurs directly downstream. Even for ideal
conditions, criteria C2 (Fig. 10) and C3 (Fig. 11), as well as a-value
(C5, Fig. 13) are fulfilled for this control section E1.
Two of the mentioned criteria, C2 and C3, focus on velocities
orthogonal to the main flow direction. Despite the fact that C5 uses
only the main flow direction, it can also produce a good indication
that the orthogonal components should be considered. Hence,
criterion C5 needs only one integration over the complete cross
section, it is a good choice for a first classification of flow conditions.
The advantage of both criterion C1 (Figs. 8 and 9) and new criterion
754 R. Gabl et al. / Renewable Energy 121 (2018) 745e756

Fig. 12. Evaluation of the criterion C4 depending on the variation of EA and EK e 0 indicates that all single main flow components fulfilled the criterion.

Fig. 13. Evaluation of the criterion C5 depending on the variation of EA and EK e a-value.

C6mod (Fig. 14) is the given definition of boundaries for which they
are fulfilled. Comparing these two criteria, criterion C6mod has a
higher variability of investigated testing areas, which also allows
for the use of smaller testing areas. In contrast, C1 is limited to
quadrants or halves of the control section, which are also included
in criterion C6mod .
5. Conclusions
In conclusion, the combination of an evaluation of kinetic en-
ergy flux using coefficient a (criterion C5) and new criterion C6mod
is a good choice for evaluating pipe flow conditions in front of a
bulb hydro turbine. The a-value (Eq. (1)) compares real kinetic
energy Ekin;real with the theoretical value Ekin;theo and provides one
value for the complete investigated control section. Based on the
Fisher-Franke-criterion [19], the local mean velocity um;part of an
individual testing areas are normalised by the global mean value um
and compared with boundary values (Eq. (2), Table 1). In case of a
pipe flow condition, the testing areas have to be limited to complete
sectors (C6mod ) to include the near wall effects correctly.
In combination with a 3D numerical simulation, these criteria
can facilitate the selection of geometry options up to a turbine in
future projects (i.e., the criteria can indicate whether the geometry
is better or worse than others). Furthermore, the conducted
 R. Gabl et al. / Renewable Energy 121 (2018) 745e756 755

Fig. 14. Evaluation of criterion C6mod depending on the variation of EA and EK e 0 indicates that all sections fulfilled the criterion.

variations, which are shown in Figs. 7e14, also allow for classifi- f point density of discrete velocity distribution ½
cation in comparison with simplified obstacles (as bad/as good as r mass density of water z 997 [kg/m3]
an obstacle with such a dimension).
The next step will be to directly connect the results for each
criterion with power generation at the turbine, comparable to the References
work of Fernando and Rival [16]. This is on-going research [64] that
[1] E. Dick, Fundamentals of Turbomachines, Springer, 2015, https://doi.org/
will allow for the monetary quantification of biased VDs, allowing 10.1007/978-94-017-9627-9_9. Netherlands.
the complete optimisation of construction costs in relation to the [2] X. Liu, Y. Luo, B. Karney, W. Wang, A selected literature review of efficiency
gained benefits based on higher production. improvements in hydraulic turbines, Renew. Sustain. Energy Rev. 51 (2015)
18e28, https://doi.org/10.1016/j.rser.2015.06.023.
[3] J.N. Goundar, M.R. Ahmed, Design of a horizontal axis tidal current turbine, Appl.
Acknowledgement Energy 111 (2013) 161e174, https://doi.org/10.1016/j.apenergy.2013.04.064.
[4] J.H. Lee, S. Park, D.H. Kim, S.H. Rhee, M.-C. Kim, Computational methods for
performance analysis of horizontal axis tidal stream turbines, Appl. Energy 98
The presented research was supported by the Tiroler Wissen- (2012) 512e523, https://doi.org/10.1016/j.apenergy.2012.04.018.
schaftsfonds (UNI-0404/921) and by the Austrian Science Fund [5] Y. Luo, Z. Wang, X. Liu, Y. Xiao, C. Chen, H. Wang, J. Yan, Numerical prediction
(FWF) under Grant J3918. The authors also want to thank the tur- of pressure pulsation for a low head bidirectional tidal bulb turbine, Energy 89
(2015) 730e738, https://doi.org/10.1016/j.energy.2015.06.022.
bine manufacture Geppert GmbH for the practical knowledge and [6] W. Shi, M. Atlar, R. Norman, B. Aktas, S. Turkmen, Numerical optimization and
help. experimental validation for a tidal turbine blade with leading-edge tubercles,
Renew. Energy 96 (2016) 42e55, https://doi.org/10.1016/j.renene.2016.04.064.
[7] S. Waters, G. Aggidis, Tidal range technologies and state of the art in review,
List of symbols Renew. Sustain. Energy Rev. 59 (2016) 514e529, https://doi.org/10.1016/
j.rser.2015.12.347.
A area ½m2  [8] K.K. Adane, S.J. Ormiston, M.F. Tachie, Numerical investigation of flow recir-
culation in a draft tube, J. Hydraul. Res. 46 (1) (2008) 15e20, https://doi.org/
B boundary value ½ 10.1080/00221686.2008.9521839.
Ekin kinetic energy ½Joule [9] T. Ciocan, R.F. Susan-Resiga, S. Muntean, Modelling and optimization of the
D diameter of the pipe ½m velocity profiles at the draft tube inlet of a Francis turbine within an operating
range, J. Hydraul. Res. 54 (1) (2016) 74e89, https://doi.org/10.1080/
g gravitational acceleration z 9.81 [m/s2] 00221686.2015.1119763.
N numbers of grid points in section ½ [10] P.P. Jonsson, B.G. Mulu, M.J. Cervantes, Experimental investigation of a Kaplan
Q discharge [l/s] or [m3/s] draft tube Part II: Off-design conditions, Appl. Energy 94 (2012) 71e83,
https://doi.org/10.1016/j.apenergy.2012.01.032.
r radius ½m [11] D. Jost, A. Skerlavaj, A. Lipej, Improvement of efficiency prediction for a Kaplan
R radius of pipe ½m turbine with advanced turbulence models Strojniski vestnik, J. Mech. Eng. 60
Re Reynolds number ½ (2) (2014) 124e134, https://doi.org/10.5545/sv-jme.2013.1222.
[12] B.G. Mulu, P.P. Jonsson, M.J. Cervantes, Experimental investigation of a Kaplan
t time ½s draft tube Part I: Best efficiency point, Appl. Energy 93 (2012) 695e706,
u velocity in the main direction½m=s https://doi.org/10.1016/j.apenergy.2012.01.004.
um mean value of u ½m=s [13] P. Duquesne, Y. Maciel, C. Deschnes, Unsteady flow separation in a turbine
diffuser, Exp. Fluid 56 (8) (2015), https://doi.org/10.1007/s00348-015-2030-7,
umax maximum value of u ½m=s
1432e1114.
v velocity in the y-direction½m=s [14] Y. Li, G. Song, Y. Yan, Transient hydrodynamic analysis of the transition pro-
w velocity in the z-direction½m=s cess of bulb hydraulic turbine, Adv. Eng. Software 90 (2015) 152e158, https://
a kinetic energy correction factor [] doi.org/10.1016/j.advengsoft.2015.08.006.
[15] R. Koirala, H. Prasad Neopane, O. Shrestha, B. Zhu, B. Thapa, Selection of guide
b correction factor for VD ½ vane profile for erosion handling in Francis turbines, Renew. Energy 112
q angle in cross section plane [ ] (2017) 328e336, https://doi.org/10.1016/j.renene.2017.05.033.
756 R. Gabl et al. / Renewable Energy 121 (2018) 745e756
[16] J. Fernando, D. Rival, Characterizing the influence of upstream obstacles on
very low head water-turbine performance, J. Hydraul. Res. (2014), https://
doi.org/10.1080/00221686.2014.917809.
[17] L.M.C. Ferro, L.M.C. Gato, A.F.O. Falcao, Design and experimental validation of
the inlet guide vane system of a mini hydraulic bulb-turbine, Renew. Energy
35 (9) (2010) 1920e1928, https://doi.org/10.1016/j.renene.2010.01.020.
[18] American Society of Civil Engineers, Guidelines for Design of Intakes for Hy-
droelectric Plants, Committee on Hydropower Intakes of the Energy Division
of ASCE, Ron Ott, (committee chairman), New York, 1995.
[19] R.K. Fisher, G.F. Franke, The impact of inlet flow characteristics on low head
hydro projects, in: Brian W. Clowes (Ed.), Proc. Int. Conf. on Hydropower,
Portland, 1987, pp. 1673e1680. Portland.
[20] J. Knaus, Swirling Flow Problems at Intakes - IAHR Hydraulic Structures
Design Manual, 1987. Rotterdam, Balkema.
[21] D.L. Vischer, W.H. Hager, Dam Hydraulics, Wiley, Chichester, 1999.
[22] E. Egusquiza, C. Valero, A. Estevez, A. Guardo, M. Coussirat, Failures due to
ingested bodies in hydraulic turbines, Eng. Fail. Anal. 18 (1) (2011) 464e473,
https://doi.org/10.1016/j.engfailanal.2010.09.039.
[23] N. Rajaratnam, C. Katopodis, M. Sadeque, N. Pokharel, Turbulent flow near
vertical angled fish screen, J. Hydraul. Eng. 136 (11) (2010), 945951, https://
doi.org/10.1061/(ASCE)HY.1943-7900.0000269.
[24] H. Allahyonesi, M. Hossein Omid, A. Hamzeh Haghiabi, A study of the effects of
the longitudinal arrangement sediment behavior near intake structures,
J. Hydraul. Res. 46 (6) (2008) 814e819, https://doi.org/10.1080/
00221686.2008.9521925.
[25] B. Gems, M. Wo € rndl, R. Gabl, C. Weber, M. Aufleger, Experimental and nu-
merical study on the design of a deposition basin outlet structure at a
mountain debris cone, Nat. Hazards Earth Syst. Sci. 14 (2014) 175e187,
https://doi.org/10.5194/nhess-14-175-2014.
[26] G. De Cesare, A. Schleiss, F. Hermann, Impact of turbidity currents on reservoir
sedimentation, J. Hydraul. Eng. 127 (1) (2001) 616, https://doi.org/10.1061/
(ASCE)0733-9429(2001)127:1(6).
[27] M. Müller, G. De Cesare, A. Schleiss, Continuous long-term observation of
suspended sediment Transport between two pumped-storage reservoirs,
J. Hydraul. Eng. 140 (5) (2014), 05014003, https://doi.org/10.1061/(ASCE)
HY.1943-7900.0000866.
[28] M. Müller, G. De Cesare, A. Schleiss, Flow field in a reservoir subject to pumped-
storage operation in situ measurement and numerical modeling, J. Appl. Water
Eng. Res. (2016), https://doi.org/10.1080/23249676.2016.1224692.
[29] A. Schleiss, M. Franca, C. Juez, G. De Cesare, Reservoir sedimentation,
J. Hydraul. Res. (2016), https://doi.org/10.1080/00221686.2016.1225320.
[30] H. Sarkardeh, A.R. Zarrati, R. Roshan, Effect of intake head wall and trash rack
on vortices, J. Hydraul. Res. 48 (1) (2010) 108e112, https://doi.org/10.1080/
00221680903565952.
[31] F. Suerich-Gulick, S. Gaskin, M. Villeneuve, E. Parkinson, Characteristics of free
surface vortices at low-head hydropower intakes, J. Hydraul. Eng. 140 (3)
(2014) 291e299, https://doi.org/10.1061/(ASCE)HY.1943-7900.0000826.
[32] F. Suerich-Gulick, S. Gaskin, M. Villeneuve, E. Parkinson, Free surface intake
vortices: theoretical model and measurements, J. Hydraul. Res. (2014), https://
doi.org/10.1080/00221686.2014.896425.
[33] Y. Wang, C. Jiang, D. Liang, Comparison between empirical formulae of intake
vortices, J. Hydraul. Res. 49 (1) (2011) 113e116, https://doi.org/10.1080/
00221686.2010.534279.
[34] Y. Wang, M. Politano, H.-C. Ho, M. Muste, F. Michell, J. Stallings, Assessment of
ice plugging of a cooling water intake by a numerical model, J. Hydraul. Res.
52 (1) (2014) 81e92, https://doi.org/10.1080/00221686.2013.854844.
[35] S. Li, Y. Lai, L. Weber, J. Matos Silva, V.C. Patel, Validation of a three dimen-
sional numerical model for water-pump intakes, J. Hydraul. Res. 42 (3) (2004)
282e292, https://doi.org/10.1080/00221686.2004.9728393.
[36] S. Li, J.M. Silva, Y. Lai, L.J. Weber, V.C. Pate, Three-dimensional simulation of
flows in practical water-pump intakes, J. Hydroinf. 8 (2) (2006) 111e124.
[37] N. Yildirim, K. Tastan, Critical submergence for multiple pipe intakes,
J. Hydraul. Eng. 135 (12) (2009) 1052e1062, https://doi.org/10.1061/(ASCE)
HY.1943-7900.0000127.
[38] V. Heller, Scale effects in physical hydraulic engineering models, J. Hydraul.
Res. 49 (3) (2011) 293e306, https://doi.org/10.1080/00221686.2011.578914.
[39] F. Suerich-Gulick, S. Gaskin, M. Villeneuve, E. Parkinson, Free surface intake
vortices: scale effects due to surface tension and viscosity, J. Hydraul. Res.
(2014), https://doi.org/10.1080/00221686.2014.905503.
[40] L. Khan, E. Wicklein, M. Rashid, L. Ebner, N. Richards, Computational fluid
dynamics modeling of turbine intake hydraulics at a hydropower plant,
J. Hydraul. Res. 42 (1) (2004) 61e69, https://doi.org/10.1080/
00221686.2004.9641184.
[41] A. Nakayama, N. Hisasue, Large eddy simulation of vortex flow in intake
channel of hydropower facility, J. Hydraul. Res. 48 (4) (2010) 415e427.
[42] R. Gabl, S. Achleitner, J. Neuner, M. Aufleger, Accuracy analysis of a physical
scale model using the example of an asymmetric orifice, Flow Meas. Instrum.
36 (2014) 36e46, https://doi.org/10.1016/j.flowmeasinst.2014.02.001.
[43] R. Gabl, B. Gems, M. Plo € rer, R. Klar, T. Gschnitzer, S. Achleitner, M. Aufleger,
Numerical simulations in hydraulic engineering, Comput. Econ. (2014)
195e224, https://doi.org/10.1007/978-3-319-05933-4_8.
[44] H. Benigni, M. Mosshammer, H. Jaberg, M. Ho €lbling, J. Mayrhube, Flow anal-
ysis and operational optimisation of a two-unit bulb turbine hydro plant with
significant power difference: simultaneous CFD calculation of both turbines,
in: Proc. of the Hydro 2017: Shaping the Future of Hydropower, vol. 03(10),
2017, pp. 1e8.
[45] D. Godde, Experimentelle Untersuchungen zur Anstro € mung von Rohrturbinen
e ein Beitrag zur Optimierung des Turbineneinlaufes (Doctoral thesis), Techn.
Univ. München, 1994 (in German).
[46] A.J. Ward-Smith, Internal Fluid Flow e the Fluid Dynamics of Flow in Pipes
and Ducts, Clarendon Press, Oxford, 1980.
[47] R. Gabl, S. Achleitner, D. Innerhofer, M. Aufleger, Adapted evaluation criterion
for the velocity distribution in front of turbines, in: Proceedings of the 36th
IAHR World Congress: Deltas of the Future and what Happens Upstream,
2015, pp. 3193e3198.
[48] ANSYS-CFX: Users Manual Version 14.0, ANSYS, Inc.
[49] MATLAB: Users Manual Version R2016a, MathWorks.
[50] A.G. Andersson, P. Andreasson, T.S. Lundstro €m, CFD-modelling and validation
of free surface flow during spilling of reservoir in down-scale model, Eng.
Appl. Comut. Fluid Mech. 7 (1) (2013) 159e167, https://doi.org/10.1080/
19942060.2013.11015461.
[51] R. Edgar, S. Cochard, Z. Stachurski, Double-layer orthogonal-offset photovol-
taic platforms, Appl. Energy 147 (2015) 478e485, https://doi.org/10.1016/
j.apenergy.2015.03.002.
[52] B. Lotfi, B. Sundn, Q. Wang, An investigation of the thermo-hydraulic per-
formance of the smooth wavy fin-and-elliptical tube heat exchangers utilizing
new type vortex generators, Appl. Energy 162 (2016) 1282e1302, https://
doi.org/10.1016/j.apenergy.2015.07.065.
[53] Se Park, Su Park, S.H. Rhee, Influence of blade deformation and yawed inflow
on performance of a horizontal axis tidal stream turbine, Renew. Energy 92
(2016) 321e332, https://doi.org/10.1016/j.renene.2016.02.025.
[54] L.A. Teran, F.J. Larrahondo, S.A. Rodrguez, Performance improvement of a 500-
kW Francis turbine based on CFD, Renew. Energy 96 (2016) 977e992, https://
doi.org/10.1016/j.renene.2016.05.044.
[55] D. Li, H. Wang, Z. Li, T.K. Nielsen, R. Goyal, X. Wei, D. Qin, Transient charac-
teristics during the closure of guide vanes in a pump-turbine in pump mode,
Renew. Energy 118 (2018) 973e983, https://doi.org/10.1016/
j.renene.2017.10.088.
[56] D. Nan, T. Shigemitsu, S. Zhao, T. Ikebuchi, Y. Takeshima, Study on perfor-
mance of contra-rotating small hydro-turbine with thinner blade and longer
front hub, Renew. Energy 117 (2018) (2018) 184e192, https://doi.org/
10.1016/j.renene.2017.10.046.
[57] E. Benini, R. Biollo, R. Ponza, Efficiency enhancement in transonic compressor
rotor blades using synthetic jets : a numerical investigation, Appl. Energy 88
(2011) 953e962, https://doi.org/10.1016/j.apenergy.2010.08.006.
[58] W. Zhang, Z. Zou, J. Ye, Leading-edge redesign of a turbomachinery blade and
its effect on aerodynamic performance, Appl. Energy 93 (2012) 655e667,
https://doi.org/10.1016/j.apenergy.2011.12.091.
[59] R. Absi, An ordinary differential equation for velocity distribution and dip-
phenomenon in open channel flows, J. Hydraul. Res. 49 (1) (2011) 82e89,
https://doi.org/10.1080/00221686.2010.535700.
[60] K. Amiri, B. Mulu, M. Raisee, M. Cervantes, Unsteady pressure measure-
ments on the runner of a Kaplan turbine during load acceptance and load
rejection, J. Hydraul. Res. 54 (1) (2016) 56e73, https://doi.org/10.1080/
00221686.2015.1110626.
[61] F. Song, Y. Ni, Z. Tan, Optimization design, modeling and dynamic analysis for
composite wind turbine blade, Proc. Eng. 16 (2011) 369e375, https://doi.org/
10.1016/j.proeng.2011.08.1097.
[62] F.R. Menter, Two-equation eddy-viscosity turbulence models for engineering
applications, AIAA J. 32 (8) (1994) 1598e1605, https://doi.org/10.2514/
3.12149.
[63] F.R. Menter, M. Kuntz, R. Langtry, Ten years of industrial experience with the
SST turbulence model, in: K. Hanjalic, Y. Nagano, M. Tummers (Eds.), Pro-
ceedings of the Fourth International Symposium on Turbulence, Heat and
Mass Transfer, Antalya, Turkey, 2003.
[64] D. Innerhofer, J. Lochschmidt, J. Lampl, R. Wührer, B. Brinkmeier, M. Aufleger,
Anstro €mung von Kompaktturbinen/Inflow at compact turbines,
€
Osterreichische Wasser- Abfallwirtsch. 67 (7) (2015) 307e314, https://
doi.org/10.1007/s00506-015-0249-4 [in German].