Anda di halaman 1dari 8

Available online at www.sciencedirect.

com

ScienceDirect
Energy Procedia 112 (2017) 43 50

Sustainable Solutions for Energy and Environment, EENVIRO 2016, 26-28 October 2016,
Bucharest, Romania

Experimental and numerical investigation of a Bnki turbine


operating far away from design point
Andrei Dragomirescua,*, Mihai Schiauaa
a
University Politehnica of Bucharest, Department of Hydraulics, Hydraulic Machinery and Environmental Engineering,
313 Splaiul Independentei, Bucharest 060042, Romania

Abstract

This paper presents the results of a combined experimental and numerical study on a Bnki turbine with horizontal nozzle,
operating at heads and discharges that go below 5% and 30%, respectively, of the rated values. The experimental results suggest
that, even for such dramatic decreases in the values of the operating parameters, the turbine could still attain a peak efficiency of
more than 55%, although at the expense of a significant decrease in speed. To get an insight into the flow through the turbine,
numerical simulations were carried out. The results obtained are in good agreement with the experiments. They suggest that far
away from design point the energy transfer inside the runner suffers a significant change when compared with the normal
operation, most of the energy being now transferred on the second pass.
2017Published
2017 The Authors. Published
by Elsevier by Elsevier
Ltd. This Ltd.
is an open access article under the CC BY-NC-ND license
Peer-review under responsibility ofthe organizing committee of the international conference on Sustainable Solutions for Energy
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review
and under responsibility
Environment 2016. of the organizing committee of the international conference on Sustainable Solutions for Energy
and Environment 2016
Keywords:cross-flow water turbine; Bnki; part load; efficiency; energy transfer; CFD

1. Introduction

Small hydroelectric power plants are a common solution to the power needs of small communities that are not
connected to the grid. Presently, many such plants use cross-flow turbines (CFTs) Bnki and its variants due to
the advantages that these turbines possess: simple construction and maintenance, low price, flat efficiency curve,

* Corresponding author. Tel.: +4-074-056-9986; fax: +4-021-402-9865.


E-mail address:andrei.dragomirescu@gmail.com

1876-6102 2017 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the international conference on Sustainable Solutions for Energy and Environment 2016
doi:10.1016/j.egypro.2017.03.1057
44 Andrei Dragomirescu and Mihai Schiaua / Energy Procedia 112 (2017) 43 50

and good behavior with partial loads. In particular, the efficiency that remains almost constant down to about 20%
of the rated discharge can yield better annual performance than other turbine types. Another characteristic of CFTs
is that the water passes through the runner blades twice, most of the energy about 66% of it being transferred on
the first pass.
Due to their advantages, CFTs have been extensively studied. Key parameters that influence the efficiency were
investigated experimentally [1, 2]. Numerical simulations were carried out to get a better insight into the flow
through CFTs and to provide a basis for further studies aimed at improving turbine design [3, 4]. Based on such
simulations, attempts were made to optimize turbine performance by properly changing nozzle and runner
geometry [5, 6]. New procedures for the optimal design of CFTs were proposed [7, 8]. The geometry of a jet
deflector installed inside the runner of a CFT was optimized with CFD tools [9]. Studies were devoted to
investigating the beneficial effects of an air layer injected inside the turbine chamber [10, 11]. CFTs with modified
designs for usage in aqueducts and conduits that carry water from a water source to a tank were proposed and
studied both experimentally and numerically [12, 13]. To harvest energy at low heads, in rural areas, a simplified
and cheaper design of a CFT having an open inlet duct (without nozzle and guide vane) and a draft tube was
investigated [14, 15]. The design and manufacturing plan of a Bnki turbine for very low heads was presented [16].
In this paper, results of a combined experimental and numerical study on a Bnki turbine with horizontal nozzle,
operating at heads and discharges significantly lower than the rated ones, is presented. The purpose of the study was
twofold: (i) to investigate how the flow through the turbine is influenced by a dramatic change in operating
parameters and (ii) to verify whether Bnki turbines could remain relatively efficient even when operating far away
from design point.

2. Experimental installation

The experimental installation is presented in Figure 1. Water is supplied to the turbine from a pump station through an
upstream conduit foreseen with a Venturi tube for discharge measurement. After passing through the turbine, the water
falls freely into a channel that goes beneath and returns to the pump station. The turbine is coupled to a hydrodynamic
brake for simulating different loads and measuring the torque (denoted hereinafter as shaft moment for consistency with
other notations). A flap installed inside the turbine nozzle allows the variation of the discharge.
The turbine (Fig. 2) was designed for a head of 22 m, a discharge of 250 l/s, and a speed of 770 rpm. The runner
has 19 blades. Its inner and outer diameters are of 165 mm and 250 mm respectively. The runner (and nozzle)
breadth is of 300 mm. The runner shaft sits on two ball bearings.
Since the turbine is part of a system that is directly supplied with water by a pump, the turbine can operate only
with the head and discharge corresponding to the intersection between system curve and pump curve. This particular
situation is similar to those investigated by Sammartano et al. [12] and Sinagra et al [13]. Moreover, due to
limitations that appeared in time, the maximum head that can be still obtained in the experimental installation is of
slightly less than 15 m when the nozzle flap is at 1/8 of full opening.
All measurements were carried out at fully opened flap. The differential pressure at the Venturi tube and the
pressure at turbine inlet were measured with a differential pressure transducer (DPT) and a gauge pressure
transducer (GPT) having measuring ranges of 0100 mbar and of 02.5 bar respectively. Both transducers have
an accuracy of 0.5% of full scale (FS). They were powered by a 0-30 VDC stabilized power supply and were

Fig. 1. Experimental installation.


Andrei Dragomirescu and Mihai Schiaua / Energy Procedia 112 (2017) 43 50 45

connected to a GL800 data logger for data acquisition. The speed was measured with a Testo 470 digital tachometer
with a measuring range of 19,999 rpm and an accuracy of 0.02% of reading (RD).
Since it is difficult to manage the flow parameters at turbine exit, the head was calculated based only on the total
pressure at turbine inlet. During the measurements, the head values varied roughly between 0.720.92 m. The
discharge remained almost constant from 175 rpm to 220 rpm and then decreased slightly with increasing speed.
Overall, the discharge varied only very little within the range 67.268.3 l/s. Therefore, the turbine operated at less
than 5% of its rated head and less than 30% of its rated discharge. Because the efficiency of the hydrodynamic brake
decreases with decreasing speed, measurements were not possible below 174.9 rpm. At zero torque (i.e. no braking),
the maxim runner speed was of 268.4 rpm.

3. Numerical simulations

All numerical simulations were performed on the computational domain presented at the right hand side of
Figure 2. The domain consists of the turbine nozzle, the runner, and a draft tube. The two-phase flow of water and
air was considered unsteady and turbulent. Since during the experiments the discharge remained almost constant, a
constant water mass flow rate of 226.7 kg/s was imposed at nozzle entry, corresponding to a flow rate of 68 l/s. The
same quantity of water was required to leave the computational domain at draft tube exit. The equations describing
the flow continuity and momentum equation, equations for tracking the air-water interface, and closure equations
for turbulent quantities were integrated in space and time with the Finite Volume Method implemented in the
commercial code Ansys Fluent. The Reynolds Stress Model was chosen to provide the closure equations for the
turbulent quantities. The runner movement was simulated with the Sliding Mesh Technique. Simulations were
carried out for runner speeds ranging from 0 rpm to 275 rpm, with time steps that decreased from 0.5 ms down to
0.1 ms with increasing runner speed. Each simulation was advanced in time until the total pressure at nozzle entry
and the moment coefficient of the blades got stabilized around average values for 10 complete revolutions of the
runner. The convergence criterion at each time step was the drop in all scaled residuals below 10 -3.
A preliminary mesh sensitivity analysis was carried out on three meshes with different densities. The main
parameters of these meshes mesh labels, minimum and maximum cell sizes, and number of cells are
presented in Table 1. Non-uniform structured grids refined at the walls were used at nozzle entry, inside the
blade channels, and in the main part of the draft tube. The end of the nozzle, close to runner entry, the interior
of the runner and the region of the draft tube close to runner exit were meshed with quadrilateral unstructured
grids. Details of the meshes obtained are presented in Figure 3. Simulations for the mesh sensitivity analysis

Fig. 2. Sketches of turbine and computational domain.


46 Andrei Dragomirescu and Mihai Schiaua / Energy Procedia 112 (2017) 43 50

Table 1. Main parameters of meshes used for the mesh sensitivity analysis.

Mesh ID Min. cell size (mm) Max. cell size inside runner (mm) Max. cell size inside nozzle (mm) Number of cells
mesh 05 0.5 2.5 5 61,375
mesh 04 0.4 2.0 4 84,690
mesh 03 0.3 1.5 3 128,382

Fig. 3. Details of the meshes used for the mesh sensitivity analysis: a) mesh 05, b) mesh 04, c) mesh 03.

were carried out only for a runner speed of 175 rpm. To assess the convergence, the Grid Convergence
Method (GCM) [17] was used, taking as representative grid size the minimum cell size. As key variables, the
total pressure at nozzle entry, pti, and the moment coefficient of the blades, cm, were chosen, since these are
required to assess the performance and efficiency of the turbine. Variations of these varia bles depending on the
number of grid cells are presented in Figure 4. For each key variable, the extrapolated value, the approximate
relative error, the extrapolated relative error, and the grid convergence index (GCI) were calculated. The
extrapolated value is the value of the key variable obtained by Richardson extrapolation using the values
obtained on mesh 03 and mesh 04. The approximate relative error is the absolute value of the difference in key
parameter values obtained on mesh 03 and mesh 04 divided by the key parameter value obtained on mesh 03.
The extrapolated relative error is the absolute value of the difference in extrapolated value and key parameter
value obtained on mesh 03 divided by the extrapolated value. GCI was calculated with the formulas presented
in [17], based on the results obtained on mesh 03 and mesh 04. The results obtained with GCM are
summarized in Table 2. It can be seen that, for both variables, the approximate and extrapolated relative errors
as well as the grid convergence index are very small. Considering the results obtained, it was decided to
complete the simulations on mesh 04, which offered a good trade-off between accuracy and CPU time.

4. Results and discussion

Figure 5 presents comparisons between experimental and numerical values of turbine head and moment. Towards
175 rpm, the heads, H, obtained by numerical simulations agree very well with the measured ones (Fig. 5a). Above
220 rpm, the simulations start to slightly overestimate the head. This is due to the fact that all simulations were run
with a constant discharge of 68 l/s, while the measured discharge slightly decreased below this value at speeds
higher than 220 rpm. The results of the numerical simulations suggest that, under the particular operating conditions
of the investigated turbine, the head could decrease down to about 0.4 m as the runner speed decreases.
The moments, M, were extracted from the results of the numerical simulations based on the moment coefficients
calculated by Fluent. The runner peripheral velocity, the total area of the longitudinal cross-section of the runner,
and the runner radius were used as reference velocity, reference area, and reference length respectively. As
Figure 5b shows, above 175 rpm there is a good agreement between the moment values obtained experimentally and
Andrei Dragomirescu and Mihai Schiaua / Energy Procedia 112 (2017) 43 50 47

Fig. 4. Variations of total pressure at inlet (a) and moment coefficient (b) depending on number of mesh cells.

Table 2. Results of the mesh sensitivity analysis.

Key Value on Value on Value on Extrapolated Approximate Extrapolated GCI


variable mesh 03 mesh 04 mesh 05 value relative error relative error
pti 6,878 Pa 6,960 Pa 6,559 Pa 6,865 Pa 1.19% 0.19% 0.24%
cm 1.2 1.207 1.189 1.196 0.64% 0.33% 0.41%

numerically. The numerical results suggest that the starting moment (or starting torque) of the turbine could be of
about 80 Nm. However, this finding should be treated with care since it could not be verified experimentally.
The shaft power, Ps, depending on runner speed is presented in Figure 6a. As expected, the values obtained by
numerical simulations are higher than the measured ones, since the numerical algorithms could account neither for
mechanical losses nor for all volumetric losses. Due to the dramatic decrease in head and discharge with respect to
the rated values, the turbine could deliver only less than 0.3 kW.
Figure 6b shows that, in the range of runner speeds investigated experimentally, the total efficiency, K, remains
below 32%. It should be noted, however, that this value includes the efficiency of the hydrodynamic brake used to
measure the shaft moment. An extrapolation based on a third order polynomial fit of the experimental data suggests
that the maximum total efficiency could be of roughly 55% at about 90 rpm. The reason for choosing a third order
polynomial to fit the data was the simplicity of the function coupled with the fact that it was possible to intercept the
efficiency axis at K = 0 (since the efficiency is 0 when the runner speed equals 0 rpm) while providing a very good
value of the determination coefficient:R2 = 0.9942.Hydrodynamic brakes have designs that are usually similar to fluid
couplings, for which the efficiencies are in the range from 95 to 97% [18]. Since the hydrodynamic brake used in the
experiments is old and no data is available for it, it was assumed that its efficiency could be of about 95%. Under this
assumption, the turbine efficiency can be assessed at about 58%. According to the results of the numerical simulations,
the hydraulic efficiency of the turbine, Kh, could reach 92%, at about 75 rpm. During turbine operation, a small quantity
of water by-passes the runner through the two clearances of about 4 mm between the disks that support the runner
blades and the turbine casing. There is also a small leakage through the clearances between the nozzle flap and the

Fig. 5. Variations of measured and calculated turbine head (a) and moment (b).
48 Andrei Dragomirescu and Mihai Schiaua / Energy Procedia 112 (2017) 43 50

Fig. 6. Variations of measured and calculated shaft power (a) and efficiency (b).

casing. To account for these discharge losses, a turbine volumetric efficiency of about 95% could be assumed. The pair
of roll bearings that support the runner has a mechanical efficiency of around 99% [19]. Additional mechanical losses
are caused bythe friction between the shaft and its seals and by the viscous friction between therunner disks and the
water that by-passes the runner. Therefore, the mechanical efficiency of the turbine was assumed to be of 98%.
Considering the assumptions made for the volumetric and mechanical efficiencies of the turbine and accounting also
for the brake efficiency, a value of roughly 81% would result for the total turbine efficiency. Although this value is
realistic for modern cross-flow turbines operating at their rated point, it seems rather overestimated for the turbine
investigated in this study. There are different causes for the difference between the values of the total turbine efficiency
of 58% and 81% estimated based on the experimental data and based on the results of the numerical simulations
respectively. The simulations did not account for the high roughness of the internal walls of the turbine caused by
corrosion. This roughness could significantly decrease the hydraulic efficiency. The volumetric losses are actually very
difficult to assess since it is not possible to disassemble the welded turbine casing in order to properly evaluate all the
clearances that can cause the losses. The misalignment of the turbine and brake shafts could also cause mechanical
losses in the bush pin type flange coupling that couples the turbine and the brake. Additional losses are caused by the
viscous friction between the coupling (whose dimensions are not negligible) and the surrounding air.
The results presented in Figure 6b show also that the peak efficiency remains relatively high at the expense of a
dramatic decrease in runner speed.
A very interesting behavior shows the moment on a runner blade. Figure 7 depicts the blade moment variations
depending on angular position M measured clockwise starting from nozzle lip, for two runner speeds: 75 rpm, where
the efficiency reaches the maximum in the numerical simulations, and 175 rpm, the lowest speed for which
measurements were still possible. At M = 0 the blade moment is negative, but starts to increase rapidly as the blade
enters the water jet created at nozzle exit. Afterwards, the moment oscillates around an average value up to an angle
of about 75. As the blade approaches the nozzle tongue, located at M = 90, the moment shows a second rapid
increase and reaches a local maximum at an angle which is slightly larger than 90. The blade moves then away
from the nozzle and the moment starts to decrease as the blade exits the water jet. The interaction between the water

Fig. 7. Blade moment variations depending on angle with respect to flap lip at two impeller speeds.
Andrei Dragomirescu and Mihai Schiaua / Energy Procedia 112 (2017) 43 50 49

Table 3. Average moments on a runner blade and energies absorbed during the first and second pass at two runner speeds.

n(rpm) T (s) Mavg (Nm) E1 (J) E2 (J) E (J) E1/E E2/E


75 0.7995 1.866 4.633 7.101 11.73 0.3948 0.6052
175 0.3425 0.4507 0.7113 2.134 2.845 0.25 0.75

jet and the blade stops completely when the blade reaches an angle of about 110, which does not depend on runner
speed. This can be easily explained by the fact that, at this angle, due to the runner geometry, the blade behind the
current one reaches the nozzle tongue position and the water jet cannot reach anymore the current blade. Hence, it
can be considered that the first pass of the water takes place at angles comprised between 0 and 110. After leaving
the nozzle tongue behind, the blade enters shortly into an idle state, suggested by the fast decrease in blade moment,
until a local minimum is reached. This idle state is shorter at higher runner speeds. A new rapid increase in blade
moment marks the beginning of the second pass. As it can be seen, the peak blade moment is reached not in the first
pass, but in the second one, as opposed to cross-flow turbines operating at or close to the rated point. Finally, a last
decrease in blade moment indicates that the blade approaches the end of the second pass, which takes place at M =
240 independently of runner speed. Beyond this angle, the blade goes into an idle state, where the blade moment
becomes practically zero, since the blade interacts no more with water but only with the surrounding air. Very close
to nozzle lip, the blade interacts again with the water jet. However, the interaction starts on the suction side, as it
will be seen in Figure 8. Consequently, the blade moment shortly drops below zero.
Integration of the blade moment over the angular ranges of each pass yields the energies E1 and E2 absorbed by the
blade during the first and the second pass, respectively. The results are summarized in Table 3, with E being the sum of
E1 and E2. The values obtained indicate that, when the cross-flow turbine operates far away from design point, the
energy transfer inside the runner suffers a significant change: the energy is no more absorbed by the runner
predominantly during the first pass, but during the second pass. It seems that the runner starts to behave more like an
undershot wheel. The average blade moments, Mavg, were also calculated. It can be seen that the products between
these values and the number of runner blades (19 blades) are consistent with the results presented in Figure 5b.
An explanation for the change in energy transfer inside the runner can be found in Figure 8, which shows plots of
volume fractions of water and air inside the turbine at different runner speeds. When the runner is blocked (Fig. 8a),
the image is typical for a cross-flow turbine: both in the first and in the second pass the blade channels are
completely filled with water. However, Figures 8b, 8c, and 8d show that, as the runner speed increases, air pockets
form behind the blades during the first pass. The size of the air pockets increases with increasing speed until the
water barely comes into contact with the blades. This phenomenon can be explained based on the fundamental
equation of turbo machinery. A dramatic drop in turbine head is expected to lead to a drop in tangential velocity at
nozzle exit. The decrease in tangential velocity coupled with the increase in peripheral velocity as the runner speed
increases is accompanied by an increase in absolute velocity angle. Therefore, as the runner speed grows, the water
attacks the blades increasingly closer to radial directions. Practically, the water just falls between the blades. During
the second pass, however, the blade channels are better filled with water. Hence, most of the energy is transferred
during the second pass and it is this energy transfer that makes possible the increase in runner speed.

Fig. 8. Volume fractions of water (in black) and air (in white) inside the turbine at four impeller speeds: a) 0 rpm, b) 75 rpm, c) 175 rpm, d)
275 rpm.
50 Andrei Dragomirescu and Mihai Schiaua / Energy Procedia 112 (2017) 43 50

5. Conclusions

The results presented in this paper are based on numerical simulations validated by experimental data. They
suggest that cross-flow turbines can still operate with decent efficiencies even far away from design point, although
at the expense of a dramatic decrease in speed. In the case of the turbine investigated in this study, the peak
efficiency could be of roughly 55%.
The results also suggest that, when the turbine head drops significantly, the water enters the runner at high angles
of the absolute velocity. Consequently, the blade channels do not fill completely with water during the first pass.
Instead, air pockets form behind the blades. The size of the air pockets increases with increasing runner speed until
the water barely comes into contact with the blades. This phenomenon diminishes the energy absorbed by the blades
during the first pass. The runner starts to behave more like an undershot wheel, most of the energy being transferred
from the water during the second pass.

Acknowledgment

This work was supported by the Romanian National Authority for Scientific Research, MEN-UEFISCDI, in the
frame of the research program Partnerships in Priority Domains PN II, project code PN-II-PT-PCCA-2013-4-1901.

References

[1] Desai VR, Aziz NM. Parametric Evaluation of Cross-flow Turbine Performance. J Energ Eng-ASCE 1994; 120(1):17-34.
[2] Joshi CB, Seshadri V, Singh SN. Modifications in a Cross-flow Turbine for Performance Improvement. Indian J Eng Mater S 1995; 2(6):261-267.
[3] Carija Z, Sinozic M, Fucak S, Mrsa Z, Cavrak M. Fluid Flow Simulation of a Crossflow Turbine. In: Katalinc B, editor. Annals of DAAM for
2009 and Proceedings of the 20th International DAAM Symposium. Vienna; 2009. Vol. 20, p. 1921-1922.
[4] de Andrade J, Curiel C, Kenyery F, Aguilln O, Vsquez A, Miguel A. Numerical Investigation of the Internal Flow in a Banki Turbine. Int J
Rotating Mach 2011; 2011:1-12.
[5] Acharya N, Chang-Gu K, Bhola T, Young-Ho L. Numerical analysis and performance enhancement of a cross-flow hydro turbine. Renew
Energ 2015; 80:819-826.
[6] Choi YD, Lim JI, Kim YT, Lee YH.Performance and Internal Flow Characteristics of a Cross-Flow Hydro Turbine by the Shapes of Nozzle
and Runner Blade. Journal of Fluid Science and Technology 2008; 3(3):398-409.
[7] Sammartano V, Arico C, Carravetta A, Fecarotta O, Tucciarelli T.Banki-Michell Optimal Design by Computational Fluid Dynamics Testing
and Hydrodynamic Analysis. Energies 2013; 6(5):2362-2385.
[8] Sammartano V, Arico C, Sinagra M, Tucciarelli T.Cross-Flow Turbine Design for Energy Production and Discharge Regulation. J Hydraul
Eng 2015; 141(3).
[9] Croquer SD, de Andrade J, Clarembaux J, Jeanty F, Asuaje M.Use of CFD Tools in Internal Deflector Design for Cross Flow Turbine
Efficiency Improvement. In: Proceedings of the ASME Fluids Engineering Division Summer Meeting 2012. Rio Grande; 2012. p. 315-325.
[10] Choi YD, Shin BR, Lee YH.Air Layer Effect on the Performance Improvement of a Cross-Flow Hydro Turbine. The KSFM Journal of Fluid
Machinery 2010; 13(4):37-43.
[11] Choi YD, Wei Q, Chen Z, Singh PM. Effect of Air Layer on the Performance of an Open Ducted Cross Flow Turbine. The KSFM Journal of
Fluid Machinery 2015; 18(1): 11-19.
[12] Sammartano V, Arico C, Sinagra M, Tucciarelli T.Cross-Flow Turbine Design for Energy Production and Discharge Regulation. J Hydraul
Eng 2015; 141(3).
[13] Sinagra M, Sammartano V, Arico C, Collura A. Experimental and Numerical Analysis of a Cross-Flow Turbine. J Hydraul Eng 2016; 142(1).
[14] Chen Z, Singh PM, Choi YD.Performance Improvement of Very Low Head Cross Flow Turbine with Inlet Open Duct. The KSFM Journal
of Fluid Machinery 2014; 17(4):30-39.
[15] Choi YD, Wei Q, Chen Z, Singh PM. Effect of Air Layer on the Performance of an Open Ducted Cross Flow Turbine. The KSFM Journal of
Fluid Machinery 2015; 18(1):11-19.
[16]Popescu C.Design and Manufacturing of a Low Head Banki Turbine. Appl Mech Mater 2013; 371:672-676.
[17] Celik IB, Ghia U, Roache PJ, Freitas CJ, Coleman H, Raad PE.Procedure for Estimation and Reporting of Uncertainty Due to Discretization
in CFD Applications. J Fluids Eng 2008; 130(7).
[18] Peligrad N. Fluid Coupling and Torque Converters (in Romanian). Technical Press, Bucharest, 1985.
[19] Filipoiu ID, Tudor A. Mechanical Transmissions (in Romanian). University Politehnica of Bucharest, Bucharest, 1995.

Anda mungkin juga menyukai