Professional Documents
Culture Documents
net/publication/251781667
CITATIONS READS
3 292
4 authors, including:
Josef Lampl
Voith Hydro GmbH
3 PUBLICATIONS 6 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Helmut Benigni on 18 May 2015.
Introduction
The investigation of the flow through a horizontal shaft bulb turbine and the comparison with the results achieved at
a closed loop turbine test stand are the topic of the present paper. The draft tube highly influences the performance
of bulb turbines. To save building costs (e.q. civil parts), the whole machine configuration was optimised and
investigated in order to become shorter – this is of practical interest for the design of new machines but also for the
rehabilitation of existing ones.
During the last years computational fluid dynamics (CFD) have been used routinely within the R&D process of
hydraulic machinery. By means of CFD the development time of turbines can be reduced considerably and most of
the time- and cost-intensive experimental investigations can be skipped. Normally, commercial codes Reynolds-
averaged Navier-Stokes equations combined with a two-equation robust turbulence model are used for engineering
applications. For time saving purposes, the stationary mode with interfaces between stationary and rotary parts as
well as the use of a lot of averaging techniques are standard.
2
pstat_Runner=16°_GuideVane=45° / p_outlet_def
1.9
ptot(stgnFrame)_Runner=16°_GuideVane=15° / p_outlet_def
1.8 ptot(stgnFrame)_Runner=16°_GuideVane=25° / p_outlet_def
1.7 ptot(stgnFrame)_Runner=16°_GuideVane=35° / p_outlet_def
1.6 ptot(stgnFrame)_Runner=16°_GuideVane=45° / p_outlet_def
1.5
1.4
1.3
1.2
1.1
1
0.9
0.8
Energy recovery: kinetic
0.7 energy to static pressure
0.6
0.5
0 Inlet and guide vane 1 Runner 2 Draft tube 3
Interface stator/runner Interface runner/draft-tube
Domain standardised length
Fig. 1 – Total and static pressure along a 3-blade runner bulb turbine
In this work, the predictive quality of CFD has been carefully validated by the comparison of numerical and
experimental data of the existing geometry. Reasonable correlation was the motivation to optimise the hydraulic
shape by means of numerical flow simulation. To carry out single-phase calculations, but also to get accurate
information on cavitation and sigma values, the validated method of histogram analysis known from pump
development by the Astroe team [5] was used .
The runner geometry given has been realised in several machine configurations all over the world in three- and four-
blade runner configurations. For both installations different operating points were calculated in order to validate the
existing geometry and the runner-guide vane correlation for all runner positions. Thus, a wide range of measured
shells was investigated and not only the single operating points. The basis of those comparisons are measurements
of the machine configuration on a closed loop turbine test stand. The draft tube of a hydraulic turbine is a significant
part of the machine configuration, where kinetic energy is converted into static pressure – and this energy recovery
strongly effects efficiency. In Fig. 1 the mass flow averaged pressure (static and total pressure in stationary frame) is
lined out versus the length, where the domains of numerical simulation always run in a standardised way, between
zero and one. In the section of the draft tube the encreasing static pressure is clearly lined out. The operating point is
near the optimum of the 3-blade runner configuration with the optimum guide vane position of 35 degrees, which is
lined out in red. In general, the flow in the draft tube is characterised by self-excited unsteadiness, e.g. vortex
shedding or vortex rope, as well as by externally forced unsteadiness induced by rotor blades [4]. The model runner
diameter was D=340 mm with a specific speed of nq=201 min-1 near the best point for the 4-blade runner
configuration with 16 guide vanes.
1. Experimental analysis
The experimental data for the two configurations were carried out on a closed loop turbine test stand, which ensures
efficiency with measurement errors smaller than +/-0.2 % and repeatable single efficiencies of about +/- 0.1%. The
closed circuit allows to adjust any absolute pressure (0.1 bar up to 5 bar absolute) in the tailwater tank, to simulate
any cavitation conditions. The motor generator (turbine output) is equipped with a frictionless bearing assembly.
2. Numerical analysis
For the numerical simulation of the turbine the commercial CFD-software package ANSYS CFX 5.7 and CFX 10
were used. For the runner, the guide vane and the draft tube structural grids were generated, for the final complete
machine simulation also an additional inflow area was modelled with an unstructured grid. Nearly all calculations
were carried out on the parallel calculation cluster of the TU Graz (56 CPU´s, Compaq DEC Alpha ES45, 16GB
RAM/node, operating system UNIX) with the 64-Bit solver version. The total calculation time for all calculations
was approx. 40.000 CPUh. The process was automated and for most of the calculations the results reached the
residium target before the iteration maximum. For the turbulence modelling, finally, the SST-model (Menter [8],[2])
was chosen for the stationary points of the shell and the transient calculations. In the beginning of the project several
turbulence models (e.g. k-ε model of Launder and Spalding with logarithmic wall functions [7]) were tested for
stationary operating points and for transient simulation, calculations with new turbulence models (e.g. SST-SAS [9])
are ongoing. The settings used were already tested at the institute several times before, in order to simulate hydraulic
machines, and show results very close to reality [4].
The grid generation is a time-consuming part of the simulation. For the numerical simulation a runner grid with
332.400 nodes per passage, a guide vane with 218.700 nodes per passage and half a draft tube with approx. 600.000
nodes were used. These are all structured hexaeder grids with corresponding wall compressions. With a distance of 1
or 2 degrees a total of approx. 30 (!) runner-position (see Fig. 2 top/left and right) grids for the three- and four-blade
configuration was generated. The high number of runner positions results from operating points ranging from the
lowest (Q11 = 700 [l/s]) up to four times higher operating points (Q11 = 3750 [l/s]), which cover the whole range of
runner adjustments. The data assigned (‘step’ – surface files) were imported to Pro-engineer® and preprocessed,
grid generation was carried out with TASCgrid v2.12 . The enlargement of the blades on the hub, based on casting
requirements, was modelled as well, although the effect on the simulation is low, as the bulk of the flow, where
energy is converted, concentrates on higher radii. For all runner grids the tip clearence between the runner blade and
the shroud was modelled and realised with 15 layers (see Fig. 2 – top middle).
The turning of the guide vane was modelled with 5°-distances, starting with 10° and going up to 80°, and a grid for
the blade geometry including the casted enlargement of the hub was generated. Only for very high angles the casted
enlargement was neglected because of the needs of grid generation (see Fig. 2 bottom left and middle).
Fig. 2 – Top left: 4-blade runner nearly closed, top middle: tip clearence boundary layer, top right: runner fully open
bottom left: guide vane left 15° – bottom middle: 75° – bottom right: unstructured inlet region for full machine simulation
For the comparison of small and large inlet regions, a second series of grids for the whole passage was generated.
For full machine simulation, all 16 guide vanes were calculated. Therefore, an unstructured inflow area grid (with
the complete bulb) was generated with CFXBuilt 5.6 (see Fig. 2 bottom right).
Fig. 3 – Left: different draft tubes – right: grid structure, rotor stator interfaces displayed in red and pink (exploded view)
The draft tube was given special attention, as it is very important for the simulation of this type of machine. A fully
scripted structured hexaeder grid was generated with TASCgrid V12.1 for both investigated draft tubes, including
downstream water with a total of 600.000 nodes. The downstream water causes an abrupt growth of the
enlargement, an extension of the simulation area and thus well-defined flow conditions at the end of the draft tube,
as the hard boundary outflow condition is set at the downstream water exit. So, there is no influence of the outflow
condition on the flow distribution in the draft tube.The full machine simulation yields to a grid with 8025039 nodes
at 9313384 elements. Unlike the simulation with the full 360° guide vane, the runner and the draft tube, the shell
was carried out with a model with one guide vane (22.5°), one runner passage (90°) and half of the draft tube (180°).
Thus, a maximum of possible periodic and symmetric boundaries was used. With these settings the influence of the
pier is disregarded – grid size amounts to 1259265 nodes at 1195212 elements (only hexaeder).
Boundary conditions were set in the same way, according to the institute’s experience with other simulations. As the
solver gives mass flow highest priority, we used a mass flow boundary condition at the inlet and a pressure
boundary condition at the outlet. So, mass flow was constant and the pressure head resulted from the calculations,
which means vertical cuts through the shell. For the solver we used CFX10 – the latest software version.
For the sigma estimation the pressure on the blade was written out and spread on for histogram anlysis: This method
was developed by Astroe and was cross-checked several times [5]. pHistogram is the value, when the pressure at a
certain percentage of the blade surface exhibits pressures lower than pHistogram. This value is transferred to a sigma
denomination (see nomenclature). Theoretically, the conclusion for σHISTOGRAM =-σInstallation is, that this percentage of
the blade surface is cavitating. Experimental values: σCFD, HISTOGRAM, 0.005 is equivalent to 1.2 *σSTANDARD.
3. Results
This operating point with Q11 = 1700 [l/s] has the highest measured efficiency of all calculated operating points of
the 4-blade runner. In the diagram the efficiency measured is lined out with the grey line. It can clearly be seen, that
the best behaviour related to the cavitation shows with heads lower than 1.2 of the design head. Both, efficiency and
sigma are very close to the experimental results and show the same tendencies towards the bordering regions of the
shell. Best values are achieved with a runner position of 17 degrees.
1.0100 16
1.0000 Shell cut - Experiment
0.9900 14
LR15_eta_total
0.9800
0.9700 LR16_eta_total 12
0.9600
σHisto_0.002/σHisto_def
LR16.5_eta_total
0.9500 10
eta / eta_max_design
LR17_eta_total
0.9400
0.9300 LR18_eta_total 8
0.9200
LR15_sigma_(pHisto_0.002)
0.9100 6
0.9000 LR16_sigma_(pHisto_0.002)
0.8900 4
LR16.5_sigma_(pHisto_0.002)
0.8800
LR17_sigma_(pHisto_0.002)
0.8700 2
0.8600 LR18_sigma_(pHisto_0.002)
0.8500 0
0.5 1 1.5 2 2.5
H Def / H Design
Fig. 4 – Normalised efficiency and pHistogram for the4-blade runner configuration (near shell optimum)
3.1 Inlet variation
As far as this point is concerned, two variations of the inlet region are investigated – because of different space
requirements of the generator. The area factor between the two variants is 1.8. For both variants the pier (see Fig. 2
down / right) was neglected. Strong differences could only be detected at higher flowrates. In Fig. 5 the standardised
efficiency and the absolute delta of the efficiency are outlined for a Q11 of 3000. Around the investigated head
(H_Def / H_OperatingPoint =1) a difference of the total machine efficiency of 1 % could be estimated. This results
from higher losses with higher velocities of the fluid in the bulb passage.
1.01
1 7
0.99
0.98 6
0.97
0.96 5
LR31_eta_total_small_inlet
0.95 LR31_eta_total
eta / eta_max
In order to get an impression of how good the runner works, the meridional velocity (cm) was analised on different
planes around the runner – to be presented in this paper for three different operating points (Q11=800, Q11=1700 and
Q11=2800) for the 4-blade runner configuration.
The planes were situated in the different CFD-domains of the model (see Fig. 6 top left), the first plane (labelled
with 0.97) was situated near the stator-rotor (frozen rotor) interface in the stator domain on the normalised domain
length of 97%. The next three planes were situated in the runner domain at 10% (in front of the blades), 65% and
90% (after the runner) of the normalised domain length (labelled with 1.1, 1.65 and 1.9). The last plane was situated
after the frozen rotor interface of the runner-draft tube connection in the draft tube domain at 3% of the domain
length, labelled with 2.03.
All the planes are located on standardised domain lengths that thus are not planar in radial direction. Each of the
operating points has its local optimum, the flow rates differ by more than 300%.
For all operating points a homogeneous velocity distribution in front of the runner exists, which indicates a good
guide vane geometry and is lined out in Fig. 6 with the red and green lines corresponding to the coloured planes in
the picture on the top of Fig. 6. This could also be detected when – for each turbine component – the efficiency in
relation to the total efficiency was compared to other hydraulics.
After the runner there could be seen a cm-distribution with nearly the same value over the whole radius. This also
means, that the runner works fine over its height and that only little fluid transportation in radius direction is needed,
which can be found out by comparing the velocity in radial direction (not shown). After the hub, where the rotor-
draft tube interface is located, the cm is coloured in yellow. In the middle of the draft tube velocities are very low –
this area is also one of the most unsecure parts of the steady state simulation.
1.5 097_Q11=800_Runner=4_GuideVane=50 097_Q11=1700_Runner=17_GuideVane=35
Plane position: domain length
097_Q11=2800_Runner=31_GuideVane=25 1.1_Q11=800_Runner=4_GuideVane=50
0.97 1.1 1.65 1.9 2.03
1.4 1.1_Q11=1700_Runner=17_GuideVane=35 1.1_Q11=2800_Runner=_GuideVane=25
1.65_Q11=800_Runner=4_GuideVane=50 1.65_Q11=1700_Runner=17_GuideVane=35
1.3
1.65_Q11=2800_Runner=31_GuideVane=25 1.9_Q11=800_Runner=4_GuideVane=50
1.2 1.9_Q11=1700_Runner=17_GuideVane=35 1.9_Q11=2800_Runner=31_GuideVane=25
2.03_Q11=800_Runner.=4_GuideVane=50 2.03_Q11=1700_Runner=17_GuideVane=35
1.1
2.03_Q11=2800_Runner=31_GuideVane=25
1
Q11=800 Q11=1700 Q11=2800
0.9
cm / cm_max
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.25 0.5 0.75 1
Radius HUB to SHROUD
The steady state simulation exhibits a wide range of regions with low velocity fields after the hub (see Fig. 6 and
Fig. 7 top left). One of the important aspects for shorter draft tube construction are separation zones and the
correlated efficiency losses. In Fig. 3 bottom left some of the tested geometries are shown. Because of the
remarkable discrepancies [4] known, a transient simulation was carried out additionally.
For the transient simulation presented in this paper the SST-model of Menter was used, with an implicit time
discretion scheme and second order accurate time integration. A steady state calculation was used as initial guess for
the transient calculation of the full 360° model simulation. After 3 revolutions of the runner the result was
considered as periodic in time.
The file size was to large for Windows or Linux PC-systems and the post-processing on the UNIX machine was not
that easy, as the latest software version for our Unix systems is without graphical interface. So, for some additional
post-processing, third-party software (paraview.org) was used.
Experience from built-in situations shows, that separation effects of short building draft tubes (with large opening
angles) have an influence on the efficiency up to several percent of the total efficiency. With a high discharge
operating point in the transient simulation the calculation shows a separation near the wall. Also, strong changes of
the flow distribution in the center (compare Fig. 7 – three pictures top right) can be seen in the transient simulation.
In the center of the draft tube remarkable discrepancies could be found as compared to known experimental flow
distribution of a comparable bulb turbine configuration [4]. The calculation still underpredicts the mixing of the
runner wake flow with the main flow. Reflections on this effect are given in [4] and [10].
The hydraulic design reaches best efficiency when the separation and reattachment of the fluid in the draft tube is in
balance and so the diffusor has its best efficiency – the highest amount of transformation of velocity energy into
static pressure. At modern low pressure turbines the amount of retransferable energy is about 0.5 % to 5% of the
hydraulic energy [6].
stationary trn050 trn100 trn150
transient
separation
Fig. 7 – Velocity in a short draft tube, same operating point stationary and transient
For future research, transient calculations are to be carried out, including the new SST-SAS-model of Menter [9],
which should yield an improvement for the better reattachment of the flow which separates from the draft tube wall.
Nomenclature
1 ⎧⎪ ⎡ ρ ⎛ Q& ⎞ ⎤ ⎫⎪
2
Head H
H Net _ Head = ⎨ tot ,Inlet
p − ⎢ p + ⎜ ⎟ ⎥⎬
ρ ⋅g ⎪ 2 ⎜⎝ A ⎟⎠ ⎥ ⎪
stat ,outlet
⎢⎣ ⎦⎭
⎩
Sigma σHISTOGRAM pHistogram − ptot ,outlet
σ HISTOGRAM =
ptot ,inlet − ptot ,outlet
Area A
Specific speed nq Q
nq = n ⋅ 0.75
H Def
Efficiency eta η
QModel
Standardised discharge Q11 Q11 =
H Model ⋅ DModel
2
DModel
Standardised rotational speed n11 n11 = n Model
H Model
References
1. Avellan, F., “Flow investigation in a Francis draft tube: The Flindt Project”, Proceedings of the Hydraulic Machinery and
Systems 20th IAHR Symposium,Charlotte, August 7-9, 2000.
2. CFX, “Documentation of Ansys CFX5.7”, PDF-files of the documentaion on installation CD, Ansys Inc., 2004.
3. Gehrer, A., Egger A., Riener J., “Numerical and Experimental Investigation of the draft tube flow downstream of a bulb
turbine”, Proceedings of the 21st IAHR Symposium on Hydraulic Machines and Systems, Lausanne, September 9-12, 2002.
4. Gehrer, A., Benigni, H., Köstenberger, M., “Unsteady Simulation of the Flow Through a Horizontal-Shaft Bulb Turbine”,
Proceedings of the 22nd IAHR Symposium on Hydraulic Maschines and Systems, Stockholm, 2004.
5. Gehrer, A., Benigni, H., Penninger, G., “Dimensioning and Simulation of Process Pumps”, Karlsruhe Pump Users
Technical Forum, Preprints 15-4, VDMA, 2004.
6. Giesecke, J., Mosonyi, E., „Wasserkraftanlagen”, ISBN 3-540-64907-7, 2. Auflage, Springer Verlag, 1998.
7. Launder, B. E. , Spalding, D. B., „Lectures in Mathematical Models of Turbulence”, Academic Press, London and New
York, ISBN 0-12-438050-6, 1972.
8. Menter, F.R., “Two-equation eddy-viscosity turbulence models for engineering applications”, AIAA-Journal, 32 (8), 1994.
9. Menter, F.R., Egorov, Y., “A Scale-Adaptive Simulation Model using Two-Equation Models”, AIAA 2005-1095, 2005.
10. Ruprecht, A., Maihöfer, M., Heitle M., Helmrich, T., “Simulation of vortex rope in a turbine draft tube”, Proceedings of
the 21st IAHR Symposium on Hydraulic Machines and Systems, Lausanne, September 9-12, 2002.
The Authors
Helmut Benigni, Dr. Dipl.-Ing. Studies of mechanical engineering at the Technical University of Graz, specialisation in
numerical simulation, dissertation in optimisation of hydraulic machines.
Helmut Jaberg, Prof. Dr.-Ing. Studies of aeronautical engineering in Stuttgart, Munich and Southampton, head of a pump
development division and a business unit at KSB, head of the Institute for Hydraulic Fluidmachinery, Technical University of
Graz.
Benigni, Jaberg, Lampl, Franz 2006
Numerical simulation of a
horizontal shaft tube bulb turbine
HYDRO 2006
International Conference and Exhibition
Maximizing the Benefits of Hydropower
Porto Carras, Greece, 25-27 September 2006
1
2
Q
nq = n ⋅ 3
4
H
Head H
Pelton
Experimental analysis:
Francis
Closed turbine test stand, efficiency
errors smaller than +/- 0.2%
Kaplan
Repeatable single efficiency +/- 0.1%
Bulb turbine
Motor / generator: frictionless bearing
assembly
Source: VOITH Specific speed
2
BULB TURBINE 2
Benigni, Jaberg, Lampl, Franz 2006
pstat_Runner=16°_GuideVane=25° / p_outlet_def
2.1
characterised by self- pstat_Runner=16°_GuideVane=35° / p_outlet_def
Static_Pressure/Pdef_outlet, Total_pressure/Pdef_outlet
2
excited unsteadiness: 1.9
pstat_Runner=16°_GuideVane=45° / p_outlet_def
ptot(stgnFrame)_Runner=16°_GuideVane=15° / p_outlet_def
1.8 ptot(stgnFrame)_Runner=16°_GuideVane=25° / p_outlet_def
- Vortex rope (internal) 1.7 ptot(stgnFrame)_Runner=16°_GuideVane=35° / p_outlet_def
1.6 ptot(stgnFrame)_Runner=16°_GuideVane=45° / p_outlet_def
3
BULB TURBINE 3
Benigni, Jaberg, Lampl, Franz 2006
5
BULB TURBINE 5
Benigni, Jaberg, Lampl, Franz 2006
1.0100 16
1.0000 Shell cut - Experiment
0.9900 14
LR15_eta_total
0.9800
0.9700 LR16_eta_total 12
0.9600
σHisto_0.002/σHisto_def
LR16.5_eta_total
0.9500 10
eta / eta_max_design
LR17_eta_total
0.9400
0.9300 LR18_eta_total 8
0.9200
LR15_sigma_(pHisto_0.002)
0.9100 6
0.9000 LR16_sigma_(pHisto_0.002)
0.8900 4
LR16.5_sigma_(pHisto_0.002)
0.8800
LR17_sigma_(pHisto_0.002)
0.8700 2
0.8600 LR18_sigma_(pHisto_0.002)
0.8500 0
0.5 1 1.5 2 2.5
H Def / H Design
Highest measured efficiency
6
BULB TURBINE 6
Benigni, Jaberg, Lampl, Franz 2006
1
total[] pressure []
0.9
0.8
Histogramm analysis Cavitation model
0.7
1
Norm. total pressure
0.6
total []pressure []
0.9
Standardised
0.5
0.8
0.4
0.7
0.3
0.2
0.5
Standardised
0.1
0.4
0 0.3
0 10 20 30 40 50 60 70 80 90 100
Norm. total are a [%] 0.2
Standardized total area [%]
0.1
Numerical outlier, 0
0.001 0.01 0.1 1 10 100
Norm. total are a [%]
No area cavitation Log. standardised total area [%]
7
BULB TURBINE 7
Benigni, Jaberg, Lampl, Franz 2006
1.01
1 7
0.99
0.98 6
0.97
0.96 5
LR31_eta_total_small_inlet
0.95 LR31_eta_total
eta / eta_max
8
BULB TURBINE 8
Benigni, Jaberg, Lampl, Franz 2006
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
9
BULB TURBINE 9
Benigni, Jaberg, Lampl, Franz 2006
Separation Transient
10
BULB TURBINE 10
Benigni, Jaberg, Lampl, Franz 2006
Transient Calculation
1 1
0.9998
0.9996
0.998
0.9994
0.9992
H(Def)_value/H(Def)_value_max
0.999 0.996
0.9988
value/value_max
0.9986 Q_norm
0.994
0.9984
H(Moment)_norm
0.9982
0.998 0.992
H(Drall)_norm
0.9978
0.9976 eta_LA+LR+SR1+SR2_norm_Filter2
0.99
0.9974
H(Def)_norm_Filter2
0.9972
0.997 0.988
0 0.5 1 1.5 2 2.5 3 3.5 4
revolution
11
BULB TURBINE 11
Benigni, Jaberg, Lampl, Franz 2006
Summary
Excellent correlation between experimental and
simulation data
Machine optimisation for shorter draft tube
12
BULB TURBINE 12
Benigni, Jaberg, Lampl, Franz 2006
View publication stats
http://www.hfm.tugraz.at | http://www.koessler.com
13
BULB TURBINE 13