Vous êtes sur la page 1sur 10

1

Proceedings of the 37
th
International & 4
th
National Conference on Fluid Mechanics and
Fluid Power
FMFP2010
December 16-18, 2010, IIT Madras, Chennai, India
FMFP2010________

DESIGN DEVELOPMENT OF OPTIMUM DRAFT TUBE FOR
HIGH HEAD FRANCIS TURBINE USING CFD

Vishal Soni
Hydraulic Engineer,
CFD Analysis Center, R & D, Jyoti Ltd.,
Vadodara, Gujarat, India
cfx@jyoti.com /
soni_vishal2002@yahoo.com


Amit Roghelia
Hydraulic Engineer,
CFD Analysis Center, R & D,
Jyoti Ltd.,
Vadodara, Gujarat, India
cfx@jyoti.com /
amitroghs@yahoo.com
Jaymin Desai
Hydraulic Engineer,
CFD Analysis Center, R & D,
Jyoti Ltd.,
Vadodara, Gujarat, India
cfx@jyoti.com /
desai_jaymin@yahoo.com


Vishal Chauhan
Hydraulic Engineer,
CFD Analysis Center, R & D,
Jyoti Ltd.,
Vadodara, Gujarat, India
cfx@jyoti.com /
vishu_343@yahoo.co.in


ABSTRACT
In hydro turbines, the importance of draft
tube is well known. Its function is to create
positive suction pressure utilizing unused
kinetic energy at the exit of turbine runner
and hence to increase the effective net head
on turbine unit. Till date there are no
numerical methods or design procedure
available for design of draft tubes, however
they are designed based on the intuition and
experiment of designers and are confirmed
with experimental test study. In order to
have good performance of draft tube, the
magnitude of pressure recovery should be
greater. In context to this, in the present
paper, for a high head Francis turbine,
various designs of bend type curved draft
tube are made using conventional approach
and their CFD simulations are carried out at
Best Efficiency Point (BEP) and at off duty
points of turbine having same flow physics
but different geometry of draft tube.
Modification of design is made by changing
the hydraulic parameters of suction cone,
elbow and exit diffuser. From these different
trials, an optimum draft tube is selected by
considering a tradeoff between the higher
values of C
P
and q
H
.

Keywords: Draft tube, CFD, BEP, Pressure
Recovery and Hydro Turbines.

INTRODUCTION
In the development program at Jyoti Ltd.,
project 'Swayam' was initiated by CFD
Analysis Centre for indigenous design of
turbine. Along the phase of development,
many interesting & knowledge imparting
Proceedings of the 37th National & 4th International Conference on Fluid Mechanics and Fluid Power
December 16-18, 2010, IIT Madras, Chennai, India.
FMFP10 - TM- 20
2
facts were discovered & analyzed, which in
the process lead to this research paper.
In order to utilize the available energy of
water to its fullest, ideally there should not
be any loss / reduction of energy of water in
the hydraulic passage of hydro turbines.
Draft tube which is a tube like structure
diverging in nature placed at outlet of
runner, plays an important role in utilizing
this unused kinetic energy of flow. Its main
purpose is to create positive suction pressure
at the downstream of turbine runner and
hence to increase the net head of turbine unit

LITERATURE REVIEW
Design of an optimum draft tube is a
challenging task since several kind of flow
phenomenon takes place simultaneously
inside it at BEP and at off design conditions.
Gubin, 1973 [1] had carried out extensive
study to examine the flow characteristic
downstream of runner for reaction turbine
and it was observed that the flow
downstream of reaction type runner is
having certain amount of swirling [2], [3].
As we know, the runner is the only rotating
component in hydro turbines, for maximum
efficiency of unit, ideally according to basic
equations of hydraulics Eq. 1, the flow
leaving the runner should not have any
circulation i.e. I
2
=0.

gH
) V (V

r2 u2 r1 u1
H

=
(1)
Where,
I
1
= V
u1
r
1
, &
I
2
= V
u2
r
2
Hence it can be said that flow leaving the
runner should have zero outlet whirl
component for best performance of turbine
i.e. V
u2
=0. But hydrodynamic investigation
of flow downstream of reaction type runner
with variable angle guide vanes shows that
maximum efficiency of turbine occurs when
I
2
>0 i.e. when there is certain amount of
positive circulation after runner [4]. Possible
reason for better operation of draft tube with
certain amount of swirl is because of the
presence of circulation forces the flow is
forced towards the wall and hence reducing
the boundary layer thickness and causes the
diffuser to operate with full flow across its
entire section [5].
The mean flow with velocity V
3
emerging
out of runner is unused/ not utilized if runner
is directly exiting the flow to tail race outlet.
Moreover if the runner is situated above tail
race then the part of potential energy
equivalent to height of draft tube is lost.
Draft tube will convert the kinetic energy to
pressure energy as the flow pass through it.
For optimum design of such structures, till
date there are no such analytical/ numerical
methods available and it is designed based
on the experience and intuition of designers
and often confirmed with experimental study
[1], [4] and lately by CFD.
The shape of Draft tube is gradually
diverging and hence it resembles to the
design of a diffuser therefore it can be
deduced that the parameter which can define
the performance of a draft tube is pressure
recovery coefficient C
P
[6] and it is given by
Eq. 2,
2
3
sDTInlet sDTOutlet
P
V 0.5
) P (P
C

=

(2)
By applying Bernoullis equation at inlet and
outlet of draft tube, it can be written as,

|
|
.
|

\
|
=
loss
2
5
2
3
s a
3
h
2g
V
2g
V
h P
w
P

(3)
From above Eq. 3, it can be said that the net
head on turbine increases when the
difference of kinetic energy head (V
3
2
/2g -
V
5
2
/2g) is higher i.e. when the velocity V
5
at
exit of draft tube is smaller. To take care of
this the diffuser tubes are chosen and hence
the kinetic energy difference can be
increased by either increasing diffuser angle
of draft tube or its length. Corresponding to
this there will be losses due to flow
separation, swirling of flow and vortex flow
along the length of draft tube and especially
3
in curved type draft tubes there will be
secondary flows and backpressure in flow
after bend.
To reduce the excavation cost, nowadays
curved types of draft tubes are used. As
shown in Fig.1 is a curved type of draft tube
with three main parts (i) the suction cone
(attached to the outlet of runner), (ii) elbow
(which changes the flow direction from
vertical to horizontal) and, (iii) the exit
diffuser connecting the elbow to tail race.
Influence of Hydraulic Design of
Individual Draft tube Components on
Fluid Flow
There is significant influence of hydraulic
parameters of suction cone on the
performance of curved draft tubes. Figure 1
shows the schematic 2D view of components
of a draft tube.

Fig.1 Components of Draft tube.
Generally, there is non-uniform distribution
of velocity at entrance section of draft tube
and the flow has certain amount of vorticity.
Moreover, asymmetric flow in initial suction
cone gives rise to flow separation even with
cone angle 2u at which flow separation
should not occur [1],[7]. Hydraulic losses in
draft tube suction cone depends on several
data like divergence angle 2u, wall
roughness, relative area at the inlet (F
2
) and
outlet (F
eo
) and characteristic of flow at
entrance of draft tube. On one hand
increasing the height of suction cone helps
to obtain a uniform flow after turbine runner,
on other hand after attaining certain height
of cone, there will be increased hydraulic
losses due to shorter radius of curvature of
elbow. So these losses completely ruin the
advantage of increasing height [1].
Over 20% or more of total hydraulic losses
are contributed by the elbow portion of draft
tube guiding the flow 90 from one direction
to other [1].
Generally nowadays a convergent-divergent
kind of elbow designs are adopted with
initial part as divergent and remaining part
with slight convergent to minimize the
hydraulic losses induced due to separations
and eddies generated in the curvatures of
draft tube elbow (i.e. on inner surface) [5].
In such hybrid designs, in the initial part
negative pressure gradient is developed and
hence to decrease the losses small
divergence is provided so as to raise the
pressure at end. In the remaining portion
when the flow changes it direction, positive
pressure gradient is established and to
reduce the loss the flow is intentionally
accelerated by reducing the cross sectional
area of elbow.
The purpose of exit diffuser is to connect the
elbow with the tail race with further
recovery of pressure from flow downstream
of an elbow. Since it is diffusing in shape
both in plan and in elevation view, care
should be taken by designer while designing
this part to minimize the separation and
hence the backpressure of flow.

METHODOLOGY
The specification of model turbine for which
optimum design of draft tube is required is
tabulated in Tab.1

D
3




Fig.2: 2D Sketch of draft tube with
Nomenclatures.

h
e

D
2

Inner Surface
Outer Surface

h
sc

R
o

R
i

h
eo

h
d

B
d

B
e

L
e
L
d

4
Tab.1 Specifications of Turbine.
Specific Speed, Nsq (MKW) 101
RPM 600
Based on several available turbine models of
our company and based on Gubin [1], first
design of draft tube was initiated and those
non dimensional parameters of initial draft
tube design are tabulated in Tab.2. Various
dimensions of elbow type draft tube are
shown in Fig.2.
Tab.2: Initial Non-Dimensional Design
parameters of suction cone and exit diffuser.
Non-Dimensional
Parameter
Value
D
3
/D
i
0.696
D
2
/D
i
0.904
h
sc
/D
i
1.193
B
e
/D
i
1.728
B
d
/D
i
1.902
h
d
/D
i
0.646
L
d
/D
i
2.041
9.5
Design of an elbow was initiated by
assuming suitable elbow outlet to inlet area
ratio (F
eo
/F
2
) and axial length to suction cone
Outlet diameter ratio (L
ax
/D
2
), Gubin [1].
Also a convergent-divergent hybrid elbow
design adopted with an assumption of
optimal law of variation of cross sectional
area as function of length for an elbow
(Fig.3). The transition from circular to
rectangular cross-section of an elbow is
accomplished by intermediated oval cross-
sections.
From [1], [4] and [5] it is confirmed that
there is specific importance of each and
every component of draft tube to the
performance of a turbine unit. Keeping this
thing in mind, a new approach was evolved
in which it was decided to modify each part
individually and to check the performance of
draft tube by combining them with several
permutations and combinations.


Fig.3. Cross-sectional Area vs. Length of
Convergent-Divergent Elbow.

First of all considering the design of an
elbow and exit diffuser fixed, it was decided
to vary the height of suction cone. This task
can be accomplished in two ways either by
keeping its divergence angle (2) constant or
by varying it and keeping suction cone
diameters D
3
and D
2
constant. In present
study, D
3
and D
2
are kept constant and the
divergence angle 2 is varied.
Optimization of suction cone needs to check
its performance at different height and hence
the height of suction cone i.e. h
sc
was
increased or decreased from initial
assumption with a fixed amount and is
equivalent to 0.109Di.

Tab.3. various variants of Suction cone
height along with their Type.
h
sc
/D
i
Type
0.652 T1
0.761 T2
0.870 T3
0.978 T4
1.087 T5
1.193 T6
1.304 T7

From these, seven different lengths of
suction cones as shown in Tab.3 were
identified and complete system CFD
simulations was carried out at three points
one at BEP, one at part load and one at
overload. Out of them, suction cone with
5
good performance was chosen for rest of
designs.
Similarly keeping optimum suction cone and
elbow designs fixed, the angle of exit
diffuser was varied i.e. decreased from
initial assumption firstly by an amount of
0.5 and then by 1 . From this, exit diffuser
with six different angles was found out and
is as shown in Tab.4 and corresponding CFD
simulations were taken at off design and
BEP. It was also assumed that the length of
exit diffuser in all cases were constant.
Figure 4 gives 2D over view of various
variants of suction cone along with elbow.

Tab.4. various variants of exit diffuser
angles along with their names.

o Type
5 D1
6 D2
7 D3
8 D4
9 D5
9.5 D6


NUMERICAL SIMULATION
CAD models of spiral casing, guide vanes
and draft tubes were made on Pro|E software
while the runner design was made by using
Bladegen software. For meshing of spiral
casing, draft tubes and guide vanes, ANSYS
ICEM CFD was used in which TCL/TK
subroutines were included for automatic
mesh generation of guide vanes at different
openings. This subroutines helps to reduce
the mesh generation time to great extent.
Prism layers were added with y
+
~1 in draft
tubes to capture the near wall effect using
SST turbulence model. Same meshing
parameters were used for meshing of various
variants of draft tubes to ensure comparison
with them. Turbo grid software was used for
Hexameshing of runner. A steady state 3D
viscous fluid flow in complete turbine
system was simulated using ANSYS CFX
software and its details are given in Refs.
[10]. Incompressible fluid water was used as
working fluid at atmospheric temperature.

RESULTS AND DISCUSSION
Initial design of draft tube was simulated by
CFD at BEP and on investigating this initial
design a large amount of swirl was found at
downstream of runner. It was believed that
the presence of such large swirling was due
to inefficient design of draft tube and hence
various modifications in designs of draft
tubes were made and further simulated at
BEP but the results were not welcoming.
Thorough investigation of flow leads to a
conclusion that it was the outlet whirl
velocity downstream of runner which was
high and hence the swirl angle. Since it is
necessary to study the effect of runner and
draft tube together [1], [4], a revision in
design of runner were initiated and the outlet
blade angles for runner were modified.
Again the initial design of draft tube was
simulated using entire system and modified
runner.

Fig.6 Various Variants of Suction Cone
along with Elbow Sections.


Suction Cone
Elbow
6

Fig 4. Dimensionless circumferential velocity vs. Dimensionless Radius





Fig.5. Dimensionless Axial Velocity Vs. Dimensionless Radius


7
From comparison between various
parameters of initial and modified runners in
Tab. 5 its clear that due to the large swirling
of flow at inlet of draft tube, there were
more hydraulic losses and hence the
magnitude of static pressure recovery was
less. In modified design of runner, the
magnitude of pressure recovery and
efficiency are quite good.
Tab.5 Comparison of modified and initial
runner
Runner
Type
ND
u
C
C

ND
a
C
C

C
P
q
h

Initial 0.254 0.354 35.7 40.7 93
Modified 0.0268 0.346 4.47 80.1 95.5

Fig.7 Swirling streamlines of Initial Runner
Figure 4 and Fig.5 shows the comparison
between the mean value of C
u
and C
a
of
initial and modified Runner at entrance
section to draft tube suction cone with
radius. In this figures, the velocity is made
dimensionless by a factor C
ND
=Runner
Angular Speed * Runner outlet radius and
radius by outlet radius of Runner, R
3
.
The central region of graph between non-
dimensional radius values of -0.26 to +0.26
is flat because of the presence of runner hub
(nipple refer Fig.1).
Figure 7 and Fig. 8 shows the streamlines
just downstream of initial and modified
runner and it can be seen that in modified
runner design, the swirling is quite low and
the streamlines are flowing smoothly. From
this study it was now clear that the modified
runner is the one which can be used for
further simulations.

Fig. 8 Swirling Streamlines of Modified
Runner
Further CFD analysis of draft tube at part
load, BEP and over load with modified
runner and various suction cone variants was
carried out and the corresponding results of
C
P
vs. | and q
H
vs. | are given below in Fig.
9 and fig. 10.


Fig. 9 Pressure recovery Vs. Discharge for
various suction cone variants

From Fig.9 it can be seen that draft tube with
suction cone variant T2 and T3 have good
linear variation of C
P
at different flow rate
but correspondingly Fig.10 shows that at
part load and at over load, the efficiency of
T3 is very good. Hence it can be concluded
that of all different variants, suction cone T3
can be used for further design procedures.
After selecting the optimum runner, suction
cone and elbow the exit diffuser is
optimized. To complete the task, as stated
earlier six different variants of exit diffuser
were identified with different diffuser
angles. CFD analysis results of various
variants of exit diffuser along with optimum
8
suction cone i.e. T3 at three different flow
rates are shown in shown in Fig.11 and
Fig.12.

Fig. 10 Hydraulic efficiency Vs. Discharge
for various suction cone variants

It can be seen from Fig 11 and Fig.12 that
for a draft tube with suction cone Type T3
and exit diffuser Type D4 at different
discharge, the magnitude of pressure
recovery and hydraulic efficiencies are quite
good. Also, there is smooth variation of C
P

and q
H
at different values of discharge.

Fig. 11 Pressure recovery Vs. Discharge for
draft tube with various exit diffusers and
with suction cone of type T3

So from this it can be concluded that
compared to rest other exit diffuser designs,
combination of T3-D4 type of draft tube is
optimum one and at BEP the value of
efficiency is 96.00%.
From Fig.10 and 12, it can be seen that at off
design points there is reduction of efficiency
of turbine unit compared to the BEP.
Reasons of lowering of efficiency at off
design points can be described by the fact
that at part load about 70-80% of total
kinetic energy at downstream of runner is
due to the whirl component of velocity while
at BEP, the flow downstream of runner is
axial. It is due to this reason; there is
deterioration of efficiency at part load [4].


Fig. 12 Hydraulic efficiency Vs. Discharge
for draft tube with various exit diffusers and
with suction cone of type T3

CONCLUSION
In the case discussed above, an optimum
design of draft tube is found out by using
CFD as a tool. Initial design of draft tube
made by assuming suitable dimensions
suggests a large amount of swirl at the inlet
of draft tube. There was no effect of
modification of design of draft tube on the
performance of turbine unit and hence the
runner was modified. Finally by several
permutations and combination of different
part of draft tube like suction cone, elbow
and exit diffuser an optimum design was
find out which guaranteed best pressure
recovery and hydraulic efficiency of a
turbine unit at BEP and part load conditions.
Hence it can be concluded that by
modifying various parts of draft tube and by
using CFD as a tool, we can find out the
optimized design of a component during
initial design stage.


ACKNOWLEDGMENTS
The Authors of this paper are greatly
thankful to Jyoti Ltd. for their immense
support throughout the research work and
their CFD center for utilizing their latest
software/hardware resources.
9
NOMENCLATURE
o Divergence angle of exit
diffuser
B
d
Width of exit diffuser
BEP Best Efficiency Point
B
e0
Width of elbow outlet
C
a
Axial velocity
CFD Computational fluid
dynamics
C
ND
(Runner angular velocity )*
(Runner outlet radius)
C
P
Coefficient of pressure
C
u
Circumferential velocity
D
2
Diameter of suction cone
outlet
D
3
Diameter of suction cone
inlet
F
2
Cross-sectional area of elbow
inlet
F
eo
Cross-sectional area of elbow
outlet
| Dimensionless flow rate-
Q/Q
BEP

G Gravitational acceleration
Swirl angle
q
h
Hydraulic efficiency
H Net head
h
e
Height of elbow
h
eo
Height of elbow outlet
h
loss
Hydraulic losses in draft tube
h
s
Suction head
h
sc
Height of suction cone
L
ax
Axial length of an elbow
L
d
Length of exit diffuser
L
e
Length of elbow from axis
N
sq
Specific speed
P
a
Atmospheric pressure
P
3
Static pressure at draft tube
inlet
P
sDTInlet
Static Pressure at draft tube
inlet
P
sDTOutlet
Static Pressure at draft tube
outlet
u Divergence angle of suction
cone
Q Mass flow
Density
r
1
Inlet radius of runner
r
2
Outlet radius of runner
R
i
Inner radius of an elbow
R
o
Outer radius of an elbow
V
3
Velocity at inlet of draft tube
V
5
Velocity at exit of draft tube
V
u1
Circumferential velocity
component at inlet of runner
V
u2

Circumferential velocity
component at outlet of runner
y
+
Non-dimensional distance of
1st node from surface of wall
I
1
Circulation at inlet of runner
I
2
Circulation at outlet of runner

REFERENCES
[1] Gubin, M.F 1973. Draft tubes of Hydro-
Electric Stations. Amerind Publishing Co.
Pvt. Ltd., New Delhi.
[2] Susan-Resiga, R., Muntean, S., Stein, P.
Avellan, F., Axisymmetric Swirling Flow
Simulation of the Draft tube Vortex in
Francis Turbines at Partial Discharge. In 24
th

IAHR Symposium on Hydraulic Machinery
and systems, October 27-31, Foz Do
Iguassu, Brazil.
[3] Susan-Resiga, R., Ciocan, G.D., Anton,
L., Avellan F., 2006. Analysis of Swirling
Flow Downstream a Francis Turbine
Runner, Journal of Fluids Engineering. 177-
189.Vol. 128, January 2006.
[4] Barlit, V.V., 2007. Hydraulic Turbines
(Hydraulic Theory, Computations and
Experimental Investigations), Vol-1, Vol-2.
Maulana Azad National Institute of
Technology, Bhopal.
[5] Mosonyi, E., 1987. Water Power
Development Low Head Power Plants, Third
ed. Akademiai Kiado, Budapest. pp. 647-
653, 669-679.
[6] Japikse, D., Caines, N.1998. Diffuser
design Technology. Edward Brothers
Incorporated, United States of America.
[7] Marjavaara, B.D., 2006. CFD Driven
Optimization of Hydraulic Turbine Draft
10
tubes using Surrogate Models, PhD thesis.
Lule University of Technology, Sweden.
[8] Mauri, S., Kueni, J.L., Avellan, F., 2004.
WereLegendre Separation in a Hydraulic
Machine Draft tube, Journal of Fluids
Engineering. 976-980.Vol. 126, November
2004.
[9] Idelchik, I.E., 2008.Handbook of
Hydraulic Resistance, Sixth ed. Jaico
Publishing House, India.
[10] ANSYS Inc, ANSYS CFX Reference
Guide, Release 12.0, April 2009.
[11] Avellan, F., 2001. Flow investigation in
a Francis Draft tube: The Flindt Project. In
20
th
IAHR Symposium on Hydraulic
Machinery and systems, 6-9 August,
Charlotte.
[12]Miyagawa, Sano, Kunimatsu, Aki,
Nishi., 2006. Mitigation of Draft tube Flow
Instability with Auxiliary Parts in High Head
Pump-turbines. In: Yokohama, 23
rd
IAHR
Symposium, Yokohama, Japan.
[13] Kovalov, N.N, 1965. HydroTubines
Design and Constructions. The National
Science Foundation, Washington, U.S.A.
[14] Nechleba, M. 1957. Hydraulic
Turbines, their Design and Equipments,
Constable & Co. London.

Vous aimerez peut-être aussi