You are on page 1of 8

CFD-supported Design of a Diagonal Turbine

Nathan Ledergerber, Thomas Staubli Christian Liess


Hochschule Luzern, Technik & Architektur 86, rue des Vosges
Technikumstrasse 21
CH-6048 Horw 68220 Buschwiller
Switzerland France


Introduction
This paper describes a CFD-supported design process of a diagonal turbine. The focus lies in the numerical
simulations. To minimize the costs the process relied on computational fluid dynamics. CFD was used to improve
the hydraulic design of the turbine based on simple streamline theory. Furthermore, a structural analysis using the
forces calculated in the simulation was performed.

1 Verification of the Simulation Method
1.1 Background
In a first step the design process was tested on the basis of data of an already existing diagonal turbine. For
validation of the CFD simulations the model measurement were compared to simulation results. If the simulation
predicts global values such as hydraulic efficiency, torque, head and power output well, then the simulation method
can be considered reliable. To speed up the project it was decided to use only the best efficiency point (BEP) for
validation.
1.2 Simulation Domain Setup
In Fig. 1 the cross section and the partitions of the simulation domain are shown. The partition of the inlet-section
has been simplified to utilize the rotational periodicity. Fig. 2 displays the different partitions in the pre-processing.
The geometry was further simplified by neglecting the gap between the runner blade tip and the shroud side.

Fig. 1 Cross section simulation domain

Fig. 2 Modeled simulation domain

1.3 Boundary- and Initial Conditions
Fig. 3 and Fig. 4 contain all the model measurement data. To simulate the best efficiency point, the guide vane
opening (GVO) and the angle of the runner blades have to be positioned correctly. The angle = 2 degree describes
the optimal runner blade position and the angle = 17.4 degree is the ideal GVO. To simulate the BEP, the rotation
speed n of the runner was set to 1380 rpm. The flow rate Q in this point is 0.29567 m
3
s
-1
. According to the
measurement the head H in the simulation should be about 22.8 m. The expected hydraulic efficiency is 91.04%.


Fig. 3 Model measurement data

Fig. 4 Torque on runner blade axis

1.4 Numerical Analysis
Table 1 and Table 2 contain the most important numerical values of two simulations in the BEP. One of these
simulations consists of only 30 degree of the whole simulation domain, the other is a 360 degree simulation. The
result of the comparison between Table 1 and Table 2 shows that a simplified domain is sufficient to simulate the
global values of a diagonal turbine in the BEP. It was therefore possible to reduce the simulation time by 80 %.

Formula Symbol Numerical Value Unit Denotation
Q 0.296 m
3
s
-1
flow rate
H 20.98 m head (total pressure difference between in- and outlet
P
hyd
60.67 kW hydraulic input
P
mec
56.34 kW mechanical output
93.2 - hydraulic efficiency
torque
z
390 Nm torque on runner axis
torque
a
5.7 Nm torque on one blade axis
T
11
9 Nm subsection torque
Table 1 Simulation with constant mass flow (30 degree domain)

Formula Symbol Numerical Value Unit Denotation
Q 0.296 m
3
s
-1
flow rate
H 20.94 m head (total pressure difference between in- and outlet
P
hyd
60.53 kW hydraulic input
P
mec
56.22 kW mechanical output
92.9 - hydraulic efficiency
torque
z
389 Nm torque on runner axis
Table 2 Simulation with constant mass flow (360 degree domain)

In both cases the head H differs by 8 % to the measurement. This difference is caused by the missing bearing
friction and the simplification of the spiral case. The hydraulic efficiency was approximately 2 % higher than
expected. This effect is also a result of the missing bearing friction and the lack of the gap at the runner blade tip.
The analysis of the torque T
11
verifies that this simulation method delivers sufficient reliable results for the design
process of a new turbine.

The comparison between the measurement data and the numerical simulations is a global estimate. Details of the
flow field and wall pressure distributions were not compared, since no such data were available. For validation of
part or overload load operation more than one operating point needs to be compared.



2 Development Procedure
2.1 Schematic
The development procedure schematic in Fig. 5 describes the CFD-supported design process of a turbine. The start
of this iterative cycle is a simulation of the flow in the distributor. The results of this numerical simulation are the
velocity components downstream of the trailing edge. If these values were not in the desired range the distributor
geometry was adjusted and the flow field had to be calculated again.


Fig. 5 Development cycle

After this first iteration loop of distributor simulations the velocity components and therefore the angle of attack
were hold constant. According to these calculated velocity components and the preset data like the head and the
mass flow, the leading- and the trailing edge angles of the runner blade could be defined by Eulers formula. This
blade design then is the starting point of the second iterative process. Since the Euler equation describes a
frictionless system, several adjustments of the runner blade geometry were necessary. If the main criteria head,
hydraulic efficiency, risk of cavitation, blade load, torque and the downstream velocity components are in the
requested range, the final design and the manufacturing of the model can be started.
The next step will be to test the model turbine in the MHyLab laboratory, since it is necessary to compare the results
of the simulation with the measurement for a final confirmation that the simulation predicts the correct values. If the
results will be satisfying the development process ends here, otherwise further geometry adjustment will be needed.

3 Numerical Simulation of the Distributor
3.1 Geometry
The distributor conditions the flow upstream of the runner blade. The hydraulic profile of the chosen guide vanes is
for manufacturing reasons a very simple design. Because of this fact also the flow simulations are facilitated.
3.2 Simplification
After simulations with the full domain and a one channel domain the results have shown that this case can be
simulated using rotational periodicity. The full domain is displayed in Fig. 6 and the simplified one in Fig. 7.

Fig. 6 Full domain with velocity analysis lines

Fig. 7 One channel domain with rotational periodicity
inlet
periodicity 1
periodicity 2
hub
main blade
splitter blade
outlet
velocity analysis lines

3.3 Setup of the Simulations
Several simulations with different distributor geometries were performed. The trailing edge angle, the leading edge
shape and the blade bending were varied parameters. This process is described in the section Development
Procedure as first iteration loop. In addition to the geometrical boundaries a fixed flow rate was set at the inlet and
an averaged static pressure at the outlet. The constraint on the hub, main blade, splitter blade and the shroud was a
no slip condition.
3.4 Results
To analyse the calculated data the velocity components on the velocity analysis lines, which are shown in Fig. 6,
have been averaged over the circumference. Through these averaged values it was possible to calculate the
downstream flow angle. The results of these various simulations are shown in Fig. 8. Each graph describes a
downstream flow angle characteristic of a different geometry version.

0
5
10
15
20
25
30
35
40
45
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Span [-]
A
n
g
l
e

[

]
Fig. 8 Downstream flow angle in relation to the span

3.5 Visual Analysis
In Fig. 9 and Fig. 10 the flow in one of the distributors is visualized. Fig. 9 shows that the deflection angle of the
flow is less than the trailing edge angle.
Fig. 10 shows a flow separation at the inner trailing edge. These vortices are caused by the low pressure area on the
inner side of the trailing edges. The vortices and the low pressure are indicators that the deflection in this distributor
might be not ideal. Due to these findings the distributor geometry has been adjusted and calculated again.


Fig. 9 Downstream flow angle Fig. 10 Vortices at the inner trailing edge
Flow Separation
Low Pressure Area

After several trials it was possible to design a blade geometry which generates the desired downstream flow angles.
The adjustable parameters are described in the section Setup of the Simulations. On the basis of the final
downstream flow angles the first geometry of the runner blade was designed.


4 Numerical Simulation of the Runner Blade
4.1 Geometry Design
With given head, mass flow and the velocity components upstream of the runner blades, it is possible to define the
leading- and the trailing edge angles of the runner blade by Eulers formula. This blade design is the starting point of
the second iterative process which is described in the section Development Procedure.

2 2 1 1 u u
c u c u Y =
Eq. 1 Eulers Formula
Y = specific energy
u = circumferential speed of the runner
c
u
= circumferential velocity component of the flow

4.2 Setup of the Simulations
The simulations discussed in this section include the distributor, the runner and the draft tube section. All the
simulations in the second iteration loop were set up with different runner blade geometries but the same boundary
conditions. The flow rate at the inlet is still fixed to the design point of the turbine. The runner section is defined as
rotating domain in which the shroud side is a counter rotating wall. The gap between the runner blade and the shroud
side is neglected. A further boundary condition is a constant averaged pressure at the outlet.
Fig. 11 displays a 360 degree total computational domain which consists of the three single domains. The results in
this section were calculated with a 45 degree domain.

Fig. 11 Entire computational domain

Fig. 12 The diagonal turbine with inlet and outlet sections
Distributor
Runner
Draft Tube


4.3 Results
The result of the first simulation shows that the downstream angle of the distributor does not match to the leading
edge angle of the runner blade. Also the downstream angle of the runner blade was not in the desired range. Since
the Euler equation describes a frictionless system, several adjustments of the runner blade geometry were necessary.
In Fig. 13 and Fig. 14 examples of the meridional (cm) and the circumferential (cu) velocities are shown. Based on
these values it is possible to calculate the head H. In Fig. 14 the remaining swirl at the outer radius of the draft tube
is apparent.

0
2
4
6
8
10
0.15 0.16 0.17 0.18 0.19 0.20 0.21 0.22
spherical radius [m]
v
e
l
o
c
i
t
y

[
m
/
s
]
cm 1
cu 1

Fig. 13 Meridional- and circumferential velocity components
before the runner
-2
0
2
4
6
8
10
0.15 0.16 0.17 0.18 0.19 0.20 0.21 0.22
spherical radius [m]
v
e
l
o
c
i
t
y

[
m
/
s
]
cm 2
cu 2

Fig. 14 Meridional- and circumferential velocity components
after the runner
remaining swirl

Furthermore, the pressure distribution on the runner blades was investigated. This analysis was performed on five
constant span sections which are displayed in Fig. 16. Fig. 15 shows the static pressure on these sections. Using this
analysis method it is easy to locate the low pressure areas which are indicators for misaligned flow and risk of
cavitation.

-1.0E+05
-5.0E+04
0.0E+00
5.0E+04
1.0E+05
1.5E+05
2.0E+05
2.5E+05
50 60 70 80 90 100 110
Theta []
P
r
e
s
s
u
r
e

[
P
a
]
Pressure_CS_01
Pressure_CS_03
Pressure_CS_05
Pressure_CS_07
Pressure_CS_09
Zero-Line
Fig. 15 Pressure on runner blade sections

Fig. 16 Constant span sections on runner blades

Visual analysis was conducted to check the angle of attack. The pressure distribution on the blades, especially on the
leading- and the trailing edge were analysed visually as well.
According to these different analyses the runner blade geometry was adapted several times. The final version of the
runner blade was also calculated for other operating points. Several calculations were done with blade angles of -5, 0
+5 and +10 degrees. One result of these simulations was the torque characteristic on the runner blade axis.
The main results of the finally simulated turbine geometry are listed in Table 3. In Fig. 17 and Fig. 18 the flow
through the turbine is visualized through streamlines.

Formula Symbol Numerical Value Unit Denotation
Q 0.283 m
3
s
-1
flow rate
H 14.6 m head (total pressure difference between in- and outlet
P
hyd
40.4 kW hydraulic input
P
mec
37.75 kW mechanical output
93.4 - hydraulic efficiency
torque
z
324.2 Nm torque on runner axis
torque
a
-1.7 Nm torque on one blade axis
Table 3 Main results of the final geometry


Fig. 17 Streamlines 1

Fig. 18 Streamlines 2


5 Continuative Work
The next step after the final definition of the hydraulic profile was to design all the components for the model
turbine. This work and also the tests of the new diagonal turbine model were performed by MHyLab in their
laboratory. More information on the performance of MHyLab is discussed in the paper Laboratory development of
diagonal turbines for medium head (25-100 m) Small Hydropower Plants.
6 Conclusion
To minimize the costs of development of a diagonal turbine the design process was based on computational fluid
dynamics. A first estimate of the geometry was found with simple streamline theory. This initial geometry was then
used for iteration with CFD to improve the hydraulic design of the turbine. The simulation method was validated
checking the results of the numerical simulation with measurement of a comparable turbine. The comparison
showed that a 30 degree domain is sufficient to simulate the global values of this turbine in the best point. The
simulation of the new diagonal turbine was divided into two separate iteration loops. The first iteration comprised
the simulation and the design of the distributor. The outcome of this process were the velocity components
downstream of the distributor. An initial runner channel geometry was designed using these velocity components in
combination with Eulers formula. Since the Eulers equation describes a frictionless system, there several
adjustments of the runner blade geometry were needed in a second iteration loop. The final design based on CFD
simulations of the new diagonal turbine was found with a total of seven iterations.



Acknowledgement
This study was made possible by a grant of swisselectric research: www.swisselectric-research.ch.



References
1. Herbert Sigloch, Strmungsmaschinen Grundlagen und Anwendungen, ISBN 3-446-14049-2
2. Aline Choulot, Cdric Cottin, Vincent Denis, Bruno Reul, Laboratory development of diagonal turbines for
mediumhead (25-100 m) Small Hydropower Plants Hidroenergia conference on Small Hydropower, Lausanne, Session 4B,
2010



The Authors
BSc Nathan Ledergerber graduated in mechanical engineering from Hochschule Luzern in 2008. Since then he is research
assistant at Competence Centre Fluid Mechanics and Hydro Machines.
Prof. Dr. Thomas Staubli graduated in Mechanical Engineering from the Swiss Federal Institute of Technology (ETH) in
Zrich. After two years of post-doctoral research in the field of flow induced vibration at Lehigh University, Pennsylvania, he
worked in experimental fluid mechanics at Sulzer Hydro (now Andritz Hydro) in Zrich. He then headed the Hydromachinery
Laboratory at the ETH Zrich. During this period he directed research projects in the field of hydraulic machinery. Since 1996 he
is professor in Fluid Mechanics & Hydro Machines at the Hochschule Luzern.
Prof. Dr. Christian Liess studied mechanical engineering at the Technical University Munich, Germany. He did research on gas
turbine flows at the Von Karman Insitute for Fluid Dynamics in Belgium and obtained his Ph.D. from the Free University of
Bruxelles. After some years of research on vortex flows at the Max-Planck-Institute for Flow Research in Gttingen, Germany,
he joined the company Voith in Heidenheim and worked on the development of hydraulic turbines and pumps. From 1991 till his
retirement he teached fluid mechanics and turbomachinery at the University of applied sciences in Konstanz, Germany.

You might also like