Session one - Hydraulic

6th International Conference on Innovation in Hydraulic Efficiency Measurements, July 30 –

August 1 2006, Portland, Oregon, USA

Sebastian MUNTEAN Romanian Academy Timisoara Branch, Romania

[email protected]
Liviu Eugen ANTON Politehnica University of Timisoara, Romania
[email protected]
Alexandru BAYA Politehnica University of Timisoara, Romania
[email protected]
Viorel CAMPIAN Eftimie Murgu University of Resita, Romania
[email protected]
Daniel BALINT Politehnica University of Timisoara, Romania
[email protected]
Romeo SUSAN-RESIGA Politehnica University of Timisoara, Romania
[email protected]

Abstract
Turbine discharge determination is an important and a difficult problem. Index tests methods
generally offer orientative values and absolute methods are difficult to perform. Consequently,
a numerical method that helps evaluating the discharge values is welcome. The paper presents
numerical simulation of the flow considering a 3D steady, inviscid and incompressible flow in
order to determine discharge of Kaplan turbines. In fact, by numerical simulation it is
determined the Winter-Kennedy pressure drop for a number of operating points and discharge
evaluation using Winter-Kennedy formula. Two experimental methods are employed to
validate computed discharge values. Numerical results are in good agreement with
experimental data, and thus the first method can be successfully applied.

Introduction
The measurement of the discharge in a hydroelectric power plant is required for acceptance
test, or to evaluate the performance in operation. The choice of measurement method is
imposed by some limitations, such as hydroelectric plant design, plant operation conditions,
installation and special equipment costs and desired accuracy.

Most of hydropower plants are equipped with pressure taps in spiral casing for discharge
measurement using the Winter – Kennedy method. However, for various reasons either the
calibration was not performed, or the pressure taps are not in the best locations or are
temporary out of service. As a result, difficulties are encountered when attempting discharge
measurement. The major problems with the Winter – Kennedy method are stability and
predictibility of the relation between pressure drop and turbine discharge within the whole
operating range.

From the design and operating practice of low head hydraulic turbines it is well known the
major influence of intake and spiral casing inlet geometry on operations characteristics.
Therefore, plant designing and operation conditions can induce major perturbations even in
discharge measurement. Permanent modifications of inlet conditions caused instability in
discharge measuring process by Winter – Kennedy method, low repeatability and low
accuracy, [2]. When inlet conditions are steady, discharge measurements using Winter –
Kennedy method leads to high precision determinations and a very good repeatability, [2].

Present paper deals with the calibration of Winter - Kennedy method using numerical
simulation. Practical application on a low head Kaplan turbine is considered. The computed
pressure distribution allowed to establish correct position of the pressure taps in spiral casing
and the relation between pressure drop and discharge value. In order to verify numerical
results, experimental studies were made in situ, using thermodynamic method and Winter –
Kennedy method.

Computational domain. Equations and boundary conditions.

Figure 1 presents the cross section through the Kaplan turbine as well as the computational
domain considered in the present study. The inlet section of computational domain
corresponds to the power plant inlet section.

Figure 1. Three-dimensional computational domain extended from the power plant inlet to
the runner reference plane.

The outlet section is conventionally chosen in the runner reference plane. Normally, one
would consider a computational procedure that couples the steady absolute flow in spiral
case/distributor domain with the relative flow in the runner [9].
 Figure 2. 3D structured mesh with 2 millions hexahedral cells.

A structured 3D mesh is generated. In the inflow region we consider a relatively coarse mesh,
and the mesh is further refined downstream as the flow accelerates. A particular attention is
paid to the discretization near the stay/guide vanes, to correctly represent the local large
velocity gradients. The mesh has approximately 2 million computational cells.

A 3D steady, inviscid and incompressible flow is considered, thus we solve the steady Euler
equations:

∇ ⋅V = 0 (1)

ρ (V ⋅ ∇ )V = −∇p (2)

On the inlet section we prescribe a constant total pressure. The value of the total pressure is
adjusted to obtain the maximum operating flow rate for the Kaplan turbine under
consideration. On the outlet section, the swirling flow structure is compatible with the so-
called pressure radial equilibrium. This condition is derived from the radial component of the
Euler equation,

∂ Vr Vu ∂ Vr ∂V V 2 1∂p
Vr + + Vz r − u = − . (3)
∂r r ∂θ ∂z r ρ ∂r

If the radial velocity component is negligible, Vr ≈ 0 , one obtains the pressure radial

equilibrium condition,

1 ∂ p Vu2
= . (4)
ρ ∂r r

This condition has been successfully employed on the draft tube inlet section when computing
the runner flow [7], and it has been validated experimentally [8]. A reference pressure is
conventionally set to zero at the hub on the outlet section, since condition (4) defines the
pressure only up to an additive constant.

Computations are performed in twenty-one operating points, see Figure 3. The parameters of
the operating points investigated are presented in Table 1.

Figure 3. Kaplan turbine hill chart and the operating points investigated.

Table 1 Parameters for the operating points investigated.

OP No. a0^ [-] Q^ [-] H^ [-] OP No. a0^ [-] Q^ [-] H^[-]
1 0.621 0.601
2 0.744 0.816 12 0.744 0.564
3 0.867 1.046 13 0.867 0.740 0.875
4 0.990 1.191 14 0.990 0.886
5 1.114 1.492 15 1.114 1.102
16 1.240 1.402
6 0.621 0.475 1.234
7 0.744 0.691 17 0.744 0.418
8 0.867 0.894 18 0.867 0.635
9 0.990 1.049 19 0.990 0.724
10 1.114 1.299 20 1.114 0.886
11 1.240 1.645 21 1.240 1.139
1.05 0.7

Discharge evaluation through Kaplan turbine

The Kaplan turbine is equipped with two pairs of Winter-Kennedy taps. Each Winter-
Kennedy pair contains two pressure taps. In our case, the first pair (includes taps no. 1 and no.
2, see Figure 4) is displaced at 150° relative to the spiral casing tongue while the second one
(contains taps no. 3 and no. 4) at 100°. One pressure tap from each pair is displaced on the
spiral case ceiling (taps no. 1 and no. 4) whilst the second one on the side surface (taps no. 2
and no. 3), see Figure 4(right).

Figure 4. Position of Winter-Kennedy taps on Kaplan turbine spiral casing: (left) top view and
(right) axonometric view.

Using numerical simulation performed on Kaplan turbine the Winter-Kennedy pressure drop
is computed from pressure field in all twenty-one operating points. Applying a non-linear
least square procedure on numerical pressure drop both multiplier k and exponent n from
equation (5) are computed, see Table 2.

Q = k ⋅ ∆p n (5)

300

250

200

150
Q

numerical data
Q=174.391*dp^0.50409
100 numerical data
Q=185.060*dp^0.49579

50

0
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2
dp

Figure 5. Numerical results and non-linear curve fitting for both pair taps.
 Table 2 Winter-Kennedy parameters (k, n) computed using non-linear least square
procedure.
k n
First pair (taps no. 1 and no. 2) 174.391 0.50409
Second pair (taps no. 3 and no. 4) 185.060 0.49579

According to the IEC 41 recommendations, the best choice for exponent value is 0.5. In our
case, choosing the recommended exponent value lead to ±0.82% and ±0.85% uncertainties in
discharge values, respectively. Based on pressure drop measured in situ on second pair (taps
no. 3 and no. 4) the following values are obtained: k=186.8 and n=0.5. Unfortunately, the
experimental data on the first pair (taps no. 1 and no. 2) cannot be measured since one tap was
clogged.

300

250

200

150
Q

experimental data
numerical data
100 Q=185.060*dp^0.49579

50

0
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2
dp

Figure 6. Comparison between numerical results and experimental data.

The ratio between the flow in the right and left part of spiral case (there is an inlet pier) is
analyzed. On the inlet section constant total pressure is imposed. Although in numerical
investigation the discharge changes from 0.418Q^ to 1.615 Q^ and head from 0.7H^ to
1.234H^ respectively. In our case, the ratio between the flow in the right and left part of spiral
case (there is an inlet pier) is 44.5%±0.5 and 55.5%±0.5, see Figure 4.Thanks to a very stable
inflow condition considered here, the operating head and discharge have a negligible
influence on the above-mentioned ratio.
Figure 7. Position of the control sections on Kaplan spiral case in order to compute the ratio
between the flow in the right and left part.

Validation the numerical results against experimental data

In Figure 8 are presented discharge dependencies of distributor opening considering 4 values
of turbine’s head, extended on all operation regimes. Selected points for numerical simulation
are represented such as: H^min=0.7; H^=0.875; H^n=1.05; H^max=1.234. In order to
evaluate turbine behavior in all head range H ^ ∈ [0.7,1.234 ] and for all range of guide vane

opening have been determined polynomial coefficients for every head value, like as:

( )
Q = c0 + c1 ⋅ (a 0 ) + c2 ⋅ a0
2

Coefficients’ values c0, c1, c2 for upper mentioned heads are presented in Table 3.
Table 3. Coefficients values for parabolic fit at four heads investigated.

Nr. Crt. H^ [-] c0 c1 c2

1 0.700 0.006441 0.04962 0.0006674
2 0.875 0.325000 0.06953 0.0007981
3 1.050 0.529800 0.10490 0.0008853
4 1.234 0.718200 0.15990 0.0009145
 250

H^=0.7 (numeric)
H^=0.7 (parabolic fit)
200
H^=0.875 (numeric)
H^=0.875 (parabolic fit)
H^n=1.05 (numeric)
H^n=1.05 (parabolic fit)
150 H^max=1.234 (numeric)
H^max=1.234 (parabolic fit)
Q^

100

50

0
0 50 100 150 200 250 300 350 400 450 500
a0^

Figure 8. Discharge (Q) versus guide vane opening (a 0) evaluated by numerical simulation.
( H^min=0.7; H^=0.875; H^n=1.05; H^max=1.234)

It can be mentioned that guide vane opening was considerate independent of runner
blades position, mainly not “on cam” condition. In order to validate numerical simulation
results two experimental methods were used:
1. Thermodynamic method (MT) used during operation on cam, at head H^=1.04;
2. Index test measurements (MOR), used for 3 operating off-cam regimes, for
H^=1.01
250
H^min=0.7 (numeric)
H^=0.875 (numeric)
H^n=1.05 (numeric)
200 H^max=1.234 (numeric)
H^=1.04 (experimental, MT)
H^=1.01 (experimental, MOR)

150
fi=1.1009
Q^

fi=−5.6079
100
fi=−7.5558

50

0
0 50 100 150 200 250 300 350 400 450 500
a0^

Figure 9. Comparison between numerical results and experimental data

Figure 9 presents o good agreement between numerical results and experimental data of
thermodynamic method. Experimental data from MOR method for three propeller regimes
(off cam operating Kaplan turbine) are also presented. Large magenta circles from Figure 9
indicate operation on cam. A good agreement with numerical results is obtained.

Conclusions
The paper presents a methodology for calibrating the Winter-Kennedy method using
numerical simulation. First, the full three-dimensional flow upstream the Kaplan turbine
runner is computed. We have developed a methodology for accurately describing the complex
3D geometry, as well as for building a suitable structured 3D mesh. A significant step forward
has been made to reduce the time devoted to the problem definition (geometry and mesh), in
order to be able to apply the present approach to design improvement and optimization.

With calibration of the Winter-Kennedy taps, done by numerical simulation, the parameters
(multiplicative constant k and exponent n) are easily determined. In our case, a non-linear
function fitting is used to evaluate both multiplier and exponent for calibration the Winter-
Kennedy method. The position of the Winter-Kennedy taps plays an important role on the
value of pressure drop. From the measuring point of view, it is recommended to achieve as
large as possible pressure drop on the prototype.

The two experimental methods validated numerical method of determining discharge values
of a low head of a Kaplan turbine, so that computed method can be successfully applied. A
good agreement between numerical results and experimental data is obtained. The Winter-
Kennedy method for discharge measurement using pressure drop between high and low
pressure zone is cheaper and time for preparation in incomparably shorter than for other
methods. Determination on which method should be used for a particular situation could be
done considering the main aim of the measurement, the price, the consuming time and
possible consequences of result uncertainties.

Acknowledgments
The authors acknowledge the support from the Romanian National University Research
Council grant CNCSIS A Consortium 33/2006.

References
[1] ***, International code for field acceptance tests to determine the hydraulic performance of
hydraulic turbines, storage pumps and pump turbines, Publication 41, 3rd edition, 1991
[2] Kercan V, Djelic V., Rus T., Vujanic V, “Experience with Kaplan turbine efficiency
measurements – current-meter anr/or index test flow measurements”, Proceeding of the 1st
International Group for Hydraulic Efficiency Measurements (IGHEM1996), Montreal, Canada,
1996.
[3] Lemon D., Bell P.W., “Measuring hydraulic turbine discharge with acoustic scintillation flow
meter”, Proceedings of the 1st International Group for Hydraulic Efficiency Measurements
(IGHEM1996), Montreal, Canada, 1996..
[4] Muciaccia F.F., Walter R.B., “Evaluation of the benefits of turbine refurbishment by means of
Index Test Method Reliability of results and Problems in Applications”, Proceedings of the 3rd
International Group for Hydraulic Efficiency Measurement (IGHEM2000), Kempten, UK,
2000.
[5] Rus T, Djelic V., “Absolute flow measurements on HPP Ozbalt using 320 current-meters
simultaneously”, Proceeding of the 3rd International Group for Hydraulic Efficiency
Measurements (IGHEM2000), Kempten, UK, 2000.
[6] Karlicek R.F., ”Test Equipment and Results from 25 Hydraulic Turbine Tests using
Thermodynamic Method”, Proceedings of the 1 st International Group for Hydraulic Efficiency
Measurement (IGHEM1996), Montreal, Canada, 1996.
[7] Muntean S., Balint D., Susan-Resiga R., Anton I., Darzan C., 3D Flow Analysis in the Spiral
Case and Distributor of a Kaplan Turbine, Proceedings of the 22nd IAHR Symposium, June 29
– July 2, 2004, Stockholm, Sweden, PaperA10-2, 2004.
[8] Dahlhaug O.G., A study of swirling flow in draft tube, PhD Thesis Norwegian University of
Science and Technology, Trondheim, Norway, 1997.
[9] Balint D., Muntean S., Susan-Resiga R., Accurate evaluation of Kaplan turbine efficiency by
improving the subdomain numerical coupling approach, Proceedings of International
Conference “Energy and Environment”, Bucharest, Romania, 2005. (on CD-ROM)
[10] ***, (2001), FLUENT 6. User’s Guide, Fluent Inc.
[11] ***, (2001), Gambit 2. User’s Guide, Fluent Inc.