Paper accepted for presentation at 2009 IEEE Bucharest Power Tech Conference, June 28th - July 2nd, Bucharest,

Romania
1

Cross-flow Water Turbines Control Under Grid

Disturbances
M. Andreica, Member, IEEE, S. Bacha, Member, IEEE, D. Roye, I. Munteanu, Member, IEEE,
A.I. Bratcu, Member, IEEE, J. Guiraud

uG , uL generator-side and grid side converters’ duty

Abstract—The cross-flow water turbine generation system is, cycles
as any distributed grid connected generator, sensitive to all kind
of power grid disturbances. The goal of this paper is to show how
II. INTRODUCTION
this system, by the means of its controls, is able to reject polluting
harmonics. The fault ride through following voltage gaps is also
proven. Experimental results obtained with a hybrid real-time
test rig validate the correct operation of this system under grid
L arge scale power grids, weak grids or microgrids can pass
through temporary instabilities dues to grid disturbances
such as voltage sags, harmonics, voltage drops or frequency
disturbances.
variations. These events can disturb or even stop the
Index Terms—cross-flow water turbine, resonant controller,
generation operation. Renewable energy resources which are
voltage sags, hybrid real-time simulation, harmonics. often connected to the distribution power grid are very
sensitive to this kind of phenomena and the grid code is very
I. NOMENCLATURE restrictive.
Different types of advanced solutions are proposed in the
CFWT cross-flow water turbine
PMSG permanent magnet synchronous generator literature for the low voltage ride through of distributed
DCM direct current machine generators [1]-[4].
PHIL power-hardware-in-the-loop This paper focuses on the CFWT- based generation systems
TT , PT turbine torque and power operating under grid disturbances. A suitable control and
management is needed to reject the inherent perturbations.
w , λ , Cp water speed, turbine tip speed ratio and
Hence this study proposes the control of the CFWT generation
power coefficient system so as to achieve the mitigation of grid disturbances
RT , HT turbine radius and height impact. The considered grid disturbances are voltage sag and
ΩG , TG generator rotational speed and torque polluting harmonics.
iqG , idG dq generator currents The generation system analyzed is an electrical power
generation system based on a new CFWT concept: the Achard
U DC DC-link voltage turbines piled-up in towers [5]–[8]. This concept can be used
I RDC DC-link protective load current in a modular structure and does not need important civil work
CDC , RDC capacity and resistance on the DC-link investments.
The CFWT generation systems are variable-speed systems
L inductance of the output filter
and this characteristic requires the presence of a power
RPL polluting load resistance
electronics interface between the generator and the grid. This
iPL three-phased current drawn by the polluting kind of system has multiple similarities with the wind energy
load conversion systems; therefore the control schemes are in some
iG , iL 2 generator and grid three-phased currents way close and similarities certainly exist [9]. However, as the
iL1 grid line currents before the connection point work presented here contributes to fulfilling a precise final
of the polluting load goal – the realization of a ½ scale real-world prototype, the
vL grid three-phased voltages CFWT and the primary energy resource particularities must be
taken into account. Therefore, the low level control loops and
k p , ki gains of the current resonant controller the management of the delivered power have to be adapted
Ts sample time and conceived consequently.
ω 2 k +1 odd harmonics frequencies The generation system's architecture is composed of the
CFWT tower, a permanent magnet synchronous generator
(PMSG) and two back-to-back converters (see Fig.1). The
The financial support for this work comes from the ANR (The French generated electric power is injected to a power grid. Thus, the
National Research Agency). control structures are meant to ensure the proper flow of the
M. Andreica, S. Bacha, D. Roye, I. Munteanu, A. I. Bratcu and J. Guiraud
are with the Grenoble Electrical Engineering Laboratory (G2ELab), ENSE3, BP electrical power.
46, 38402 Saint Martin d'Hères, France, (e-mails: {andreica, bacha, roye,
munteanu, bratcu, guiraud} @g2elab.grenoble-inp.fr).

978-1-4244-2235-7/09/$25.00 ©2009 IEEE

 2

PMSG DC-link Power

AC DC Grid
iG C DC RDC i L1 L iL 2
3~ > > >
i3~ U DC
i3~ i PL
DC AC

>
TG , ΩG
R PL
CFWT Currents Voltage Current
tower sense (d,q) PWM G sense PWM L ~ sense
uG uL
iqG (d, q) 3~
Currents Currents
control control
Water ∗ ∗
iqG i3~
flow
Speed Speed Voltage
sense control Ω∗G control ∗
U DC
ΩG
Fig. 1. CFWT generation system and its control structures [10]

A controlled resistive load is used on the DC-link to

B. CFWT tower
dissipate the fault over energy and a polluting nonlinear load
is used to harmonically disturb the grid. In order to physically The CFWT tower's model is based on the extracted
implement the voltage dips, a power amplifier is controlled by mechanical power equation [5] and the approach is similar to
an I/O xPC Target. the wind turbines modelling [9].
A CFWT generation system is considered to be sensitive to For one turbine the extractible power PT , depends on the
momentary interruptions, voltage sags and strong harmonics water flow speed cubed and some other turbine parameters:
perturbations. Thus it is important to prove that the controls PT = 0.5 ⋅ ρ ⋅ C p ⋅ S ⋅ w3 (1)
are robust enough to ride through the fault. where ρ is the water density, C p is the non-dimensional power
Experimental results are shown to illustrate the correct
operation of the CFWT-based generation system under the coefficient (expressing the turbine's efficiency), S is the
considered grid disturbances. surface swept by the turbines blades and w is the water flow
A power-hardware-in-the-loop (PHIL) physical test rig is speed. The surface S is computed based on the turbine radius,
used in order to test the fault ride through of the CFWT RT, and on the turbine height, HT, as S = 2 RT HT . The power
generation system. The CFWT behavior is physically coefficient is a function of the tip speed ratio, λ, defined as
simulated by a torque-controlled direct current machine λ = RT ⋅ ΩG / w , where ΩG is the turbine shaft rotational
(DCM), whereas the directly-driven PMSG and its associated speed. Fig. 2 presents a CFWT power coefficient
power electronics are designed for the real-world application. characteristic. Note that in the case of a CFWT the
hydrodynamic efficiency, Cp, usually does not overpass 0.35
III. THE GENERATION SYSTEM AND ITS CONTROL [12]. The CFWT tower is composed of 4 piled-up elementary
turbines.
A. Control aspects The turbines are placed on the same shaft, but mechanically
Fig.1 presents the system structure and the control loops. shifted with an angle of 2π/(3*4) each versus the next one [6]-
The CFWT tower is composed of four three-bladed Achard [7]. The main reason of such placement is the alleviation of
turbines, a PMSG and two three-phased back-to-back the pulsating turbine torque [5] and an improvement of the
converters. The system is grid-connected. efficiency of a single turbine. Experimental results obtained on
For this kind of configuration (as presented in Fig. 1.) the a real 1/3 scale Achard turbine prototype [6]–[7] helped at the
optimal operating mode (a maximum power injected to the simplified model design.
grid) [10] is achieved via the current control loops, the low- All the four turbines are identical thus have the same
level main controls, the speed control and the DC-link voltage characteristics; the output torque increases proportionally with
regulation. the number of turbines in the tower. The turbine torque also
Nevertheless, the grid currents can be perturbed as a result exhibits specific behavior due to hydrodynamic conditions
of grid disturbances; therefore the transferred energy quality is (e.g, mechanical torque pulsations).
negatively affected. The grid-side converter control loops have Modularity and flexibility are important characteristics of
to react promptly at this type of instabilities such as to ensure this system; therefore it can be adapted for either river flow or
a continuous operation of the CFWT. Power transfer to the marine current applications.
grid (with losses deducted) is the main role of the inverter but Concerning the grid disturbances impact analysis, the
in some cases it can also ensure the active filtering [11]. CFWT is supposed emerged in a regular river flow. So this
system is a non-intermittent source of energy.
 3

0.3 UL(p.u.) UL(p.u.)

I II
Power coefficient (-) 0.2

0.1
ULn ULn

0
0 0.5 1 1.5 2 2.5 3 3.5 4
Tip speed ratio (-)
Fig. 2. CFWT power coefficient characteristic
0.1ULn 0.1ULn
C. Power Grid Voltage Sag Ride Through t (s) t (s)
600ms 600ms 2.5s
Regarding the generation system operation under voltage
sags, and after their passage, DC-link extra energy dissipation Fig. 3. Voltage sags shapes
is required in this case, in order for the fault not to be sensed at Equation (2) is the transfer function of the employed resonant
the PMSG level. controller, aimed at rejecting odd harmonics of order 3
Slow system dynamics and hydrodynamic constraints through 11.
forbid the rapid acceleration or the brutal braking of the
5
CFWT tower. DC-link voltage control loss can be shattering 2 ⋅ ki ⋅ Ts [ z 2 − cos(ω2 k +1 ⋅ Ts ) ⋅ z ]
for this kind of system as the generator side converter will no CR ( z ) = k p + ∑ 2
k =1 z − 2 ⋅ cos(ω 2 k +1 ⋅ Ts ) ⋅ z + 1
(2)
longer be able to operate correctly. The PMSG can lose its
speed and current control. A three-phased software phase-locked loop (PLL) has been
In order to ensure the correct fault ride through a resistive used in order to synchronize the controlled currents with the
dissipation load exists as a protection on the DC-link and it is power grid voltages.
automatically on/off switched by an analogic hysteresis-based A polluting (diode bridge) load is connected in parallel to
controller (as shown in Fig. 1.). the grid in order to experiment the operation of the CFWT
This controller switches on at the moment when the PI DC system under harmonically polluted grid. The goal is to prove
voltage controller reaches its saturation limits and is no longer that no disturbance is “felt“ by the PMSG therefore neither by
acting. the CFWT.
Two voltage sags shapes of a depth of 0.9. p.u. are
considered and tested as shown in Fig. 3. IV. EXPERIMENTAL TEST RIG
D. Power Grid Harmonics Rejection A DCM simulates in real-time the dynamic behavior of the
The choice of the current controller type was made in order CFWT tower. The turbines model is fed by a regular water
to guarantee harmonic grid perturbations rejection. flow and is implemented using dSPACE hardware (DS1005).
Accordingly, a multi-frame resonant PI control approach [11], The DCM is directly driving a PMSG which is connected to
[13] is adopted for the three-phased grid currents control. the power grid via two back-to-back converters (see Fig. 4).
The integral term of this controller contains resonant poles
at the frequencies to be rejected.
Emulated Power Grid

xPC Target I/O

Power
Amplifier

CFWT torque emulator Conversion chain

DCM

PMSG

TMS320F240

Polluting
DS 1005
Load
Fig.4. Experimental test rig schematics
 4

The power grid is emulated via a power amplifier [14] The DC-link voltage rises and the on-off controller acts by
which is controlled using an xPC Target real-time system [10]. closing the switch of the dissipative load. The energy surplus
Control algorithms are implemented using the same goes into this protective load. Meanwhile, the PMSG does not
dSPACE hardware. "see" the fault, its rotational speed and its active current have
Similitude considerations have been made to correctly no significant transient. The grid injected current increases as
scale-down the real-world system to the emulated one. A scale the sag occurs. After the voltage sag all variables regain their
factor of m is given to the generator speed to perfectly scale stable evolution. The fault ride through is assured.
the test rig with the real desired application. The simulator Fig. 6 illustrates the response of the system when the power
launching time must be the same as the one of the CFWT grid is polluted with harmonics. It is shown that no
generation system (see [10] and the Appendix). perturbation passes towards the DC-link, U DC , which remains
constant. The PMSG rotational speed, ωG , and the generator
V. EXPERIMENTAL RESULTS
active current, iqG , also keep their stable curves.
The physical test rig in Fig. 4 is used to experimentally
validate the fault ride through of the CFWT generation These results show that the resonant controller is robustness
system. of the PI resonant controller as it rejects the perturbations
Fig. 5 (a) and (b) show the variables evolution under a introduced by the power grid. The output controlled currents
voltage sag of the shape I: a rectangular sag of 0.9 p.u. and iL1 remain sinusoidal and the CWFT generating system
during 0.6s. behavior is not affected by the polluting load.
Fig. 5 (c) and (d) show variables evolution under a voltage
sag of the shape II: rectangular drop to 0.9 p.u. during 0.6s
then return to 1 p.u. in 2.5s.

a) b)
3 i L 2a (5A/V) 2 v La (25V/V)
1 U DC (75V/V )

4 Ω G (50rad/s/V )

2 iqG (5A/V)

4 I RDC (5A/V)

3 v La (25V/V)

c) d)
3 i L 2a (5A/V) 2 v La (25V/V)
1 U DC (75V/V)

4 Ω G (50rad/s/V )

2 iqG (5A/V)

4 I RDC (5A/V)

3 v La ( 25V/V )

Fig. 5. System variables evolution under voltage sags

 5

a) b)
1 i L 2a (5A/V)
3 Ω G (50rad/s/V)
1 iqG (5A/V)

4 iPLa (5A/V) 4 iPLa (5A/V)

3 U DC (75V/V )
2 iL1a (5A/V) 2 i L1a (5A/V)

Fig. 6. Power grid harmonics rejection

Fig. 7 illustrates the response of the phase a controlled

current iL1a in the same conditions as above; all grid harmonic VII. APPENDIX
perturbations are rejected. CFWT system data
Radius: 0.25 m
Output rated power: 2.3 kW at water speed 2.3 m/s
Maximum Cp: 0.31, optimal λ: 2.4
Tower inertia: 2.47 kg⋅m2
PMSG: 5 kW, 3000 rpm, 135V, 4 poles
4 iPLa (5A/V)
Laboratory test rig
DCM: 6 kW, 3000 rpm, inertia 0.0275 kg⋅m2
Similitude scale factor m: 9
Inverters: 5 kW, 10 kHz, 3-leg, IGBT-based
DC-link voltage UDC = 450 V
2 i L1a (5A/V ) Grid voltage: 127/230 V

Control parameters
Sampling time: Ts=100 μs
Resonant controller parameters: ki=2000, kp=24.5

VIII. REFERENCES
3 iL*1a (5A/V)
[1] Y.Q. Zhan, S.S. Choi, D.M. Vilathgamuwa, "A voltage-sag
compensation scheme based on the concept of power quality control
center," IEEE Trans. Power Delivery, vol. 21, pp. 296 – 304, Jan. 2006.
[2] D.M. Vilathgamuwa, H.M. Wijekoon, S.S. Choi, "A Novel Technique to
Compensate Voltage Sags in Multiline Distribution System - The
Fig. 7. Controlled current waveforms under harmonically polluted power grid
Interline Dynamic Voltage Restorer," IEEE Trans. Industrial
Electronics, vol. 53, pp. 1603 – 1611, Oct. 2006.
VI. CONCLUSIONS [3] M. Basu; S.P. Das, G.K. Dubey, "Investigation on the performance of
UPQC-Q for voltage sag mitigation and power quality improvement at a
This paper aimed at showing how the CFWT generation critical load point," IET Generation, Transmission & Distribution, vol.2,
system reacts under grid disturbances such as voltage sags and pp. 414 – 423, May 2008.
harmonically perturbed power grid. [4] Po-Tai Cheng, Jhao-Ming Chen, Chia-Lung Ni, "Design of a State-
Feedback Controller for Series Voltage-Sag Compensators," IEEE
Experimental tests on a hybrid real-time test rig (PHIL) Trans. Industry Applications, vol. 45, pp. 260 – 267, Jan.-feb. 2009.
validate the correct operation of this generation system, the [5] S. Antheaume, T. Maître, J.L. Achard, "Hydraulic Darrieus turbines
fault ride through and the rejection of grid harmonics pollution efficiency for free fluid flow conditions versus power farms conditions",
are both assured. These are important issues for the CFWT Renewable Energy 33 (2008), pp. 2186-2198.
[6] J. L. Achard and T. Maitre, "Hydraulic turbomachine," EP 1718863,
generation system as it is wished that this system experiences Aug. 18, 2005.
no disturbances. [7] J. L. Achard, D. Imbault and T. Maitre, "Device for maintaining a
Future work will focus on the connection of this generation hydraulic turbomachine," EP 1856406, Aug. 17 2008.
system to a microgrid, in order to experiment its capability of [8] S.E. Ben Elghali, R. Balme, K. Le Saux, M.E.H. Benbouzid, J.F.
Charpentier, F. Hauville, "A Simulation Model for the Evaluation of the
operating as an uninterruptible power supply. Electrical Power Potential Harnessed by a Marine Current Turbine,"
IEEE Journal of Oceanic Engineering, vol. 32, pp.786-797.
 6

[9] T. Burton, D. Sharpe, N. Jenkins, and E. Bossanyi, Wind energy Daniel Roye was born in France in 1946. He
handbook. New-York: John Wiley & Sons, 2001. received the Engineer degree in electrical
[10] M. Andreica, S. Bacha, D. Roye, I. Exteberria-Otadui, I. Munteanu, engineering in 1969and the Ph.D. degree in
”Micro-hydro water current turbine control for grid-connected or electrical engineering in 1974 from the Institut
islanding operation,” in Proc. 2008 39th IEEE Power Electronics National Polytechnique de Grenoble (INPG),
Specialists Conference, pp. 957-962. Grenoble, France. He is currently the Co-Manager
[11] I. Etxeberria-Otadui, A. López de Heredia, H. Gaztañaga, S. Bacha, R. of the Power Systems Group and a Professor at the
Reyero, “A Single Synchronous Frame Hybrid (SSFH) Multi-Frequency Grenoble Electrical Engineering Laboratory
Controller for Power Active Filters,” IEEE Trans. Industrial Electronics, (G2ELab), Institut National Polytechnique de
vol. 53, No 5, pp. 1640-1648, October 2006. Grenoble, Grenoble, France. His current research
[12] A. N. Gorban, A. M. Gorlov, and V. M. Silantyev, ”Limits of the turbine interests include renewable energy control and integration and building energy
efficiency for free fluid flow,” Journal of Energy Resources Technology, optimization.
vol. 123, pp. 311-317, 2001.
[13] X. Guillaud, P. Degobert, R. Teodorescu, "Use of Resonant Controller Iulian Munteanu (M’07) received a Master degree
for Grid-Connected Converters in Case of Large Frequency in Instrumentation and Control from Université du
Fluctuations", in Proc. 2007 European Conference on Power Havre (France) in 1997 and a Ph.D degree in
Electronics and Applications, pp. 1-8. Automatic Control from “Dunărea de Jos”
[14] H. Gaztañaga, I. Etxeberria-Otadui, D. Ocnasu, S. Bacha, "Real-Time University of Galaţi (Romania) in 2006. In 2002 he
Analysis of the Transient Response Improvement of Fixed-Speed Wind was a Fellow of the Marie Curie European
Farms by Using a Reduced-Scale STATCOM Prototype," IEEE Trans. Framework Programme at the Laboratoire
Power Systems, vol. 22, pp. 658-666, May 2007. d’Électrotechnique de Grenoble (LEG, France). At
the present he is a postdoctoral research fellow at the
Grenoble Electrical Engineering Laboratory
(G2ELab), France. His research work is oriented
IX. BIOGRAPHIES towards optimal control of renewable energy systems.

Maria Andreica (M’03) received an Engineer Antoneta Iuliana Bratcu (M’07) received a Master
degree from the Polithenica University of Bucharest degree in Electrical Engineering from “Dunărea de
in 2005 and a Master in Electrical Engineering at the Jos” University of Galaţi (Romania) in 1996 and a
Grenoble National Institut of Technology in 2006. Ph.D. degree in Automatic Control and Informatics
She joined the Grenoble Electrical Engineering from Université de Franche-Comté de Besançon
Laboratory (G2ELab) in 2006 and she presently (France) in 2001. From 1995 she is with the
works on her PhD. Hers main fields of interest are “Dunărea de Jos” University of Galaţi, where she
modeling and control of cross-flow water turbines, works as an Associate Professor with the Department
power electronics control and renewable energy grid of Energy Conversion Systems. She is presently a
integration. postdoctoral research fellow at the Grenoble
Electrical Engineering Laboratory (G2ELab), France.
Her research interests include discrete and continuous optimization and hybrid
Seddik Bacha (M’08) received his Engineer and dynamical systems with application to energy conversion systems and
Master from National Polytechnic Institute of management of industrial systems.
Algiers respectively in 1982 and 1990. He joined the
Grenoble Electrical Engineerign Laboratory
(G2ELab) and received his PhD and HDR Joël Guiraud received the Engineer degree in
respectively in 1993 and 1998. He is presently electrical engineering and industrial informatics from
manager of Power System Group of G2ELab and the Institute of Technology of Grenoble, Grenoble,
Professor at the University Joseph Fourier of France, and the second Engineer degree in industrial
control systems from the Conservatoire National
Grenoble. His main fields of interest are power
d’Arts et M´etiers (CNAM), Paris, France, in 1998.
electronics systems, modeling and control, power
Since 1997, he has been with the Power Systems
quality, renewable energy integration.
Group, Grenoble Electrical Engineering Laboratory
(G2ELab), Grenoble, France, as a Research
Engineer. His current research interests include
realtime control and rapid prototyping systems with
applications to power systems.