Beruflich Dokumente
Kultur Dokumente
Search
Collections
Journals
About
Contact us
My IOPscience
This article has been downloaded from IOPscience. Please scroll down to see the full text article.
2010 IOP Conf. Ser.: Earth Environ. Sci. 12 012063
(http://iopscience.iop.org/1755-1315/12/1/012063)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 178.138.35.116
The article was downloaded on 24/07/2013 at 14:55
IOP Publishing
doi:10.1088/1755-1315/12/1/012063
Abstract. High accuracy performance prediction method based on entire flow passage for a
Francis turbine is presented. The performance is predicted by solving unsteady ReynoldsAveraged Navier-Stokes equations combined with Reynolds- Stress model and Bubble twophase flow model. The prediction accuracy was evaluated to compare with the model test
results for three different specific speed turbines. By comparing the simulation results with
model test results for efficiency characteristic, pressure fluctuation characteristic and cavitation
characteristic, it was proved that numerical model test presented in this paper could predict
important characteristics of Francis turbine with high accuracy not only quantitatively but also
qualitatively. As a result, it was concluded that numerical model test is going to be a more
realistic estimation tool for hydraulic performance of Francis turbine and hence contributes to
cost reduction in the development of Francis turbine.
1. Introduction
At present, research and development of hydraulic turbines involves two stages; (1) design optimization by
numerical flow simulation and (2) verifications of optimized turbine performance by model test. Since model
test requires manufacturing of model turbine, it forms a large portion of the total research and development cost.
Significant cost reduction could be achieved if conventional performance verifications by model test are replaced
by numerical simulation which can predict main performances of hydraulic turbine such as efficiency, pressure
fluctuation and cavitations with high accuracy.
As numerical simulation is a strong tool to reduce development cost of hydraulic turbine, many researches
related to prediction of hydraulic turbine performance by numerical simulations have been done. Reports show
that numerical simulation methods such as choice of numerical scheme, computational grids quality, handling of
boundary conditions, choice of turbulence model and cavitation model are important factors that influence the
prediction accuracy of hydraulic turbine performance.[1][2] For example, numerical scheme with higher order
discretization and computational grid with high quality is desired to achieve solutions with high accuracy and to
maintain stability of numerical simulations respectively; numerical error of boundary conditions can be
minimized by solving entire flow passage from casing to draft tube. Moreover, many types of turbulent models
and cavitation models have been developed by academic researchers to cater for different nature of flow to be
simulated, but there is trade-off, namely the exchange of time and robustness in return for prediction accuracy.
Despite understanding of the factors that influence the prediction accuracy of hydraulic turbine, there is little
study showing numerical simulation methods that can predict main performances of hydraulic turbine with high
accuracy. Therefore, this paper presents an optimal combination of numerical simulation methods that can
predict main performances of Francis turbine with high accuracy. Entire flow passage from casing to draft tube
of three different specific speed Francis turbines were simulated. Finally, by comparing numerical simulation
results with model test results, the practical use of numerical simulation to replace conventional model test for
hydraulic turbine performance verifications, namely numerical model test is widely discussed.
IOP Publishing
doi:10.1088/1755-1315/12/1/012063
P = Pth ( Pf + Pm )
(1)
Pth = g (Q Q ) H th
P
=
= h v m
g Q H
H
h = th
H
(
Q Q )
v =
Q
P
m =
Pth
N
Q
n=
q=
H
H
(2)
(3)
(4)
(5)
(6)
(7)
P, Q, N and H are the output, discharge, rotational speed and effective net head respectively. is water
density and g is gravitational acceleration. Hth represents the theoretical head acts on runner blades. Pf and
Pm are losses in power due to disk and seal friction, and Q is leakage of discharge around runner.
Turbine efficiency () is finally predicted by using hydraulic efficiency (h), volumetric efficiency (
)
and
mechanical efficiency (m). The prediction of an operating range is done by use of unit speed (n) and
v
unit discharge (q) of eq.(7), which are derived from given actual rotational speed (N), Q, and predicted H.
Prediction procedure
The performance prediction requires to predicth, v and m in the above equations under several operating
conditions which are determined by given Q, N and guide vane opening (a). To put it more concretely, the
prediction procedure is expressed as follows:
In the above manner, thev and m can be calculated using the simple equations (JSME, 1999) depending
on only the discharge specific speed at the optimum point of the turbine, in which the correct factors are included
for discharge in v and for the rotational speed inm. [3] On the other hand as H is defined as the total pressure
difference (input head) before and behind a turbine, the H can be directly calculated from 3-D turbulent flow
simulations as well as the theoretical head Hth (output head) that is derived from the torque acting on blades.
Three-dimensional turbulence flow simulation
There are two kinds of turbulence flow simulation for practical usable for hydro-turbine. One of them is
based on Reynolds Averaged Navier-Stokes (RANS) equation. Another is Large Eddy Simulation (LES) that
directly computed the turbulence eddies of a scale larger than the computational grid without time average.
Although LES has the potential to predict the essential unsteady flow, such as the draft tube flow under the offdesigned condition, it needs huge computational time and also have some problem for handling the surface
roughness. Therefore LES is not good choice for the performance prediction though it is good choice for the
IOP Publishing
doi:10.1088/1755-1315/12/1/012063
d 1 dR 2
d 2 R 3 dR
P P
R + R 2 + = crit
dt i ri dt
2 dt
L
dt
Where
(8)
ri is the distance to bubble i center, P is the pressure around the cavitation bubble calculated by the
RANS equation and Pcrit is the pressure inside the cavitation bubble calculated by the equation as follow.
Pcrit = PV
2T
R
4
R
t
(9)
The void fraction f g in the RANS equation is determined by the following formula based on the radius R of
the cavitation bubbles calculated by (8) and (9) formula.
4
f g = n R 3
3
(10)
In addition, the initial bubble density n used as the generation source of cavitation was assumed to be
constant all over the computational domain, though real cavity flows have a distributed bubble density.
2.2 Numerical method
The discretization of the basic equations was done by finite volume method. Convective terms of RANS
equation were approximated by the third order MUSCL scheme and differential term with respect to time was
approximated by second order time difference. As for numerical algorithm to solve the algebraic finite volume
equations, the PISO method was used. Due to applying the sliding mesh interface, the relative motion between
the rotational part and the stationary part is simulated.
In addition , as for the prediction of cavitation performance, the Courant number was kept less than 0.3 to
stabilize the calculation since the time scale in a growth and collapse of cavitation bubbles is small 3 - 5 order
as compared with a mainstream phenomenon.
2.3 Computational grid and boundary condition
To verify the validity of the numerical model test, three kinds of Francis turbines were selected as typical
examples. Configurations and specific speed of these turbines are shown in Fig. 1.
IOP Publishing
doi:10.1088/1755-1315/12/1/012063
(a) Ns 80 turbine
(a) Ns 80 turbine
3. Experiment
Validation data for prediction results by the numerical model test was measured on a test rig at
Hydraulic research laboratory in Toshiba Corporation. The test rig for Ns 150 turbine is shown in Fig.3 as a
typical example. The model test was conducted on the basis of IEC standard and the following performance
data were measured.
Efficiency
Pressure fluctuation
CavitationEfficiency break down
Moreover the photographs in the flowfield of the runner outlet were taken to verify the vortex rope and
the cavitation predicted by the numerical model test.
IOP Publishing
doi:10.1088/1755-1315/12/1/012063
Ns 80 turbine
Ns 150 turbine
Ns 220 turbine
IOP Publishing
doi:10.1088/1755-1315/12/1/012063
The prediction results for the model turbine efficiency performance are indicated in Fig. 4. Besides the
performance predicted by the steady calculation, which is neglected the transient term in the fundamental
equations of the numerical model test, is written together for reference. Moreover, in order to study the ratio of
each component loss within the total hydraulic loss, the prediction result of the Ns 150 turbine is shown in fig.4
as a typical example. In these figures, the performance data are normalized by the unit discharge (n0), turbine
efficiency(0) and hydraulic loss corresponding to the optimum operating point and it shows the characteristic
on the effective net head for the design point, respectively.
(a) Ns 80 turbine
IOP Publishing
doi:10.1088/1755-1315/12/1/012063
Fig. 6 Pressure fluctuation amplitude comparison between prediction and model test
IOP Publishing
doi:10.1088/1755-1315/12/1/012063
Therefore it is thought that the cavitation characteristic can be predicted by using the numerical model test
presented here.
B
Ns 80 turbine
(Note) q/q0=1.09
Ns 150 turbine
(Note)
q/q0=1.13
Ns 220 turbine
(Note)
q/q0=1.17
A. = 0.05
B. = 0.10
(a) = 0.10
(b) =0.05
Iso-surface of void fractionfg=0.50for Ns 150 turbine under critical cavitation test
Fig. 9 Verification of cavitation flow field
IOP Publishing
doi:10.1088/1755-1315/12/1/012063
Fig. 10 Component hydraulic losses for Ns 150 turbine under critical cavitation test
5. Conclusions
The high accuracy prediction method based on the whole flow passage model was applied for three kinds of
Francis turbines and the prediction accuracy was evaluated. The results obtained are as follows:
(1) As for the efficiency characteristic, the interaction effect between the runner and the draft tube become
strong with the increase of specific speed value and the hydraulic loss is overestimated in the steady
calculation since the occurrence region for the vortex rope and the wake is time-averaged. Therefore, it is
necessary to predict the efficiency characteristic by using the unsteady calculation basically.
(2) As for the pressure fluctuation characteristic, it is confirmed that the source for the pressure fluctuation, such
as the vortex rope and the interblade vortex, can be captured and the amplitude of the pressure fluctuation is
fairly good agreement with the model test results.
(3) As for the cavitation characteristic, it is verified that the critical cavitation coefficient can be predicted high
accuracy and it is specified that the main cause of the efficiency drop is the increasing of the draft tube loss.
The unsteady simulation based on the numerical model test needs huge computational time to calculate a lot of
the operating points desired for the hydraulic performance development at this time. However it has been
reached to the technical level that the main turbine performances as have been noted are able to predict fairly
good accuracy. Therefore it is thought that the numericl model test is going to be a more realistic estimation
tool for hydraulic performance of Francis turbine with increasing the processing speed of a computer near
further.
Acknowledgments
The virtual model test for a Francis turbine was completed by the help of project members in hydraulic
research laboratory of Toshiba Corporation and the support of members in Toshiba Information System
Corporation. The authors would like to acknowledge Kiyoshi Matsumoto, Suzuki Toshiaki, Nakamura Takanori,
Akira Shinohara for helpful discussion and for suggesting the problem and Kunie Ochiai, Oo Thanda for
supplying the large scale computer system for the virtual model test.
IOP Publishing
doi:10.1088/1755-1315/12/1/012063
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
Kurosawa S et al 2006 Turbulent Flow Simulation for the Draft Tube of a Kaplan Turbine Proc. 23rd IAHR
Symp. (Yokohama, Japan)
Cervantes M et al 2005 Proc. of Turbine-99 Workshop 3 on Draft Tube Flow (Porjus, Sweden)
JSME-S008 1999 Performance Conversion Method for Hydraulic Turbines and Pump-Turbines JSME
(Tokyo, Japan)
Fu S et al 1987 Modeling Strongly Swirling Recirculating Jet Flow with Reynolds-Stress Transport
Closures 6th Symp. on Turbul. Shear Flows (Tulouse, France)
Kim S et al 1995 Near-Wall Treatment Using Wall Functions Sensitized to Pressure Gradient ASME
FED Separated and Complex Flows 217
Kurosawa S et al 2003 Numerical Prediction of Critical Cavitation Performance in Hydraulic Turbines
Proc. 4rd ASME-JSME Joint Fluids Engin. Conf. (Hawaii, USA)
Peter R et al 1995 Automatic Structured Grid Generation Computational Fluid Dynamics
10