Control of Doubly-Fed Induction Machines as an Adjustable

Speed Motor/Generator
Christian R. Kelber Prof. Dr.-Ing. Walter Schumacher

Institute of Control Engineering Institute of Control Engineering

Technical University Braunschweig Technical University Braunschweig
Hans-Sommer-Str. 66 Hans-Sommer-Str. 66
38106 Braunschweig, Germany 38106 Braunschweig, Germany
http://www.ifr.ing.tu-bs.de

1. Introduction: The presented results were obtained in an experimental

system containing a 33kW doubly-fed induction machine
The use of adjustable constant frequency energy generation with the stator connected directly to the line and an IGBT
systems is becoming more common in hydroelectric power converter in the rotor circuit. The control was realized with a
plants [6], essentially in pumped storage plants where the VeCon digital signal processor system. The pump/turbine was
speed for most efficient operation differs in pump and turbine simulated by a 40kW DC-machine with a SIMOREG drive.
mode. Besides this ASCF energy generation in pumped
storage systems allows the realization of power control in
pump mode and efficiency optimization in turbine mode. 2. Stability analysis of doubly-fed induction machines:
A decoupled speed and power control with continuous speed
adjustment can be nowadays provideded by two technologies As shown in [1] the mathematical model of the doubly-fed
[7]: induction machine operated with impressed rotor currents can
be expressed in a stator-flux oriented reference frame by the
• back-to-back systems, where a conventional synchronous equations eq.1 till eq.4.
machine is connected to the line, trough a bidirectional
converter di ms R S 1 R
• or doubly-fed induction machines (DFIM) with a smaller + i ms = u Sd + S i Rd (eq.1)
converter in the rotor circuit (Figure 1)
dt LS L0 LS

When the mechanical speed is confined to a relative small 1 RS

band around synchronism (typically ± 10%) the use of ω ms = u Sq + i Rq (eq.2)
doubly-fed induction machines comes into consideration and
L 0 i ms L 0 i ms
the static converter is sized only for a portion of the rated
power of the plant. Due to reduced costs of the converter and dμ
many advantages for the power system stability [7] this = ω ms (eq.3)
dt
solution seems to be the most suitable one.
dω - 3L 0
Θ = m el − m w = i Rq i ms − m w (eq.4)
dt (1 + σ S )

As shown in [3] a linearized model for small perturbations

can be then obtained and expressed in state space. The
stability of the DFIM can be investigated by numerically
computing the eigenvalues with the rotor currents in direct
(iRd) and quadrature axis (iRq) as varied parameters. The
computed eigenvalues show a poorly damped system with
natural frequency near to the grid frequency. The quadrature
rotor current moves the eigenvalues quietly parallel to the
imaginary axis. The increase of direct rotor current otherwise
moves the eigenvalues parallel to the real axis towards
instability (Figure 2). When iRd reaches iRd0b , defined as the
rotor current value in the d-axis where the machine generates
Figure 1 – Doubly-fed induction machine system the same among of reactive power it would be otherwise
consuming if connected as an conventional induction motor to
In this article a resume of the stability analysis of DFIM and the line [3], the damping ratio (d) reaches 0 and the system
the control algorithms used in such a system will be become unstable.
presented.
Figure 3 shows experimental results of the stator flux mode. In generating mode the turbine wicket gates controls
response to a step change of the direct component of the rotor the speed while active and reactive power control is done in
currents. While in figure 3a were used fast current controllers, the electrical machine. In pumping mode speed and reactive
in figure 3b the controllers were restricted in such a way that power control are done in the electrical machine while the
the closed loop bandwidth became less than the resonance pump/turbine gates are used for power control.
frequency of the calculated eigenvalues.

Figure 4 – Overview of the system control scheme in

generating mode
Figure 2 – Eigenvalues at varying operation point
3.1. Control of the line side inverter:
Figure 3a clearly proves that a step change of the direct rotor
current excites oscillation in the stator flux ims , that do not
The line side converter is connected in series with inductors
decay even at steady state after the step change.
to the line as shown in figure 5. These inductors are used to
smooth the converter currents.

(a) (b)

Figure 3 – Stator flux response to a step change of iRd

a) fast current controllers
b) slow current controllers Figure 5 – Circuit of the line side inverter

In figure 3b the bandwidth of the rotor current controllers was Developing the mathematical equations tied to a network
restricted to not longer contain a significant 50Hz-component oriented reference frame, where the q-axis is the system
which could excite the resonance. As result of this the 50Hz voltage space vector uN direction, the line side converter
oscillations disappeared in the stator flux, but a poor transient model can be presented as shown in figure 6. The DC-link
results. voltage is obtained by the difference between input power PW
coming from the line and output power PR going to the
machine.
3. Control schemes for ASCF energy generation with
doubly-fed induction machines:

An overview of the control of doubly-fed induction machines

in generating mode can be seen in figure 4. The use of a
regenerative converter allows independent control of all
system variables. Thus the line side converter controls the
DC-link voltage and guarantees a converter operation with
unity power factor and the machine side converter controls
the rotor current components in the direct and quadrature axis.
A cascaded control scheme permits the independent control of
machine’s active and reactive power in generating mode and Figure 6 – Mathematical model of the line side inverter in
decoupled speed and reactive power control in pumping network coordinates.
The cross coupling terms in the converter currents can be 3.2. Control of doubly-fed induction machines:
canceled by a feed forward so that an independent current
control in d and q-axis can be achieved. The d-current The control of doubly-fed induction machines in stator flux
component is used to control the power factor while the orientation presumes impressed rotor currents. The
current component in the active power axis (q-axis) is used to differential equations of the rotor circuit are written in a d-q-
control the DC-link voltage. Figure 7 shows the current stator flux reference and shown below in equations eq.5 and
control scheme in quadrature direction. The same scheme is eq.6.
used for the direct current control. A PI-controller is used for
this procedure with the parameters chosen for a fast response di Rd R R
without overshoot. To operate the converter with unity power + i Rd =
factor the d-current reference is set to zero. dt LRσ
(eq.5)
1 (1 − σ) di ms
u Rd + (ω ms − ω) i Rq −
LR σ σ dt

di Rq RR
+ i Rq =
dt LR σ
(eq.6)
1 (1 − σ)
u Rq + (ω ms − ω) i Rd − (ω ms − ω) i ms
LR σ σ
Figure 7 – Line side converter current controller

The DC-link voltage can be controlled with the q-current The last terms in both equations represent the EMF
component by using a cascaded control with another PI- component coming from the stator. In the d-axis it can be
controller as shown in figure 8. For the cascaded controller neglected. Similar to the line side inverter current control the
the internal current control loop can be replaced by a first cross coupling terms can be canceled by a feed forward and
order lag with time constant Tei . so an independent d- and q-current control can be achieved.
The rotor current control loops are represented in figure 10.
As shown in the stability analysis the current are designed not
to excite the machine eigenvalues. Experimental results of
fast and slow current controllers were seen in figure 2.

Figure 8 – Cascaded DC-link voltage controller

Experimental results of responses from d-current reference

step and DC-link voltage reference step can be seen in figure
9.

Figure 10 – Rotor current control loops in d-q-frame

3.2.1. Control scheme for turbine mode

In turbine mode operation the speed is adjusted by the wicked

gates to have the most efficient operating point in the
hydraulic machine. The electrical machine is then responsible
a) b) for the power control. As mentioned before active and
reactive power can be controlled using the impressed rotor
Figure 9 – Step response from d-current (a) and currents in field orientation. Due to experimental verifications
DC-link voltage (b) [2,4] and consequently model simplification the active power
can be directly controlled by irq and the reactive power by ird
using a cascaded control scheme similar to the DC-link
voltage control. The measured step response of the control
loop for active power is seen in figure 11; it exhibits good
dynamic performance and little cross coupling with the
reactive power control.

Figure 13 – Speed control using the DFIM

4. Conclusions:

Figure 11 – Decoupled active/reactive power control The doubly-fed induction machine model contains poorly
damped eigenvalues with a corresponding natural frequency
Due to the independence of power and speed control the near to the line frequency. With an appropriate control
DFIM can be very useful to increase the power system algorithm for example reducing the bandwidth of the current
stability. This can be seen in figure 12 where the turbine controllers this problem can be overcome.
changes the speed from 90% to 110% nominal speed while With the use of a doubly-fed induction machine in ASCF
the DFIM maintains the electrical power output constant. energy generation plants there is a possibility of decoupled
speed and power control in all four quadrants.
In generating mode a decoupled active and reactive power
control is also possible while the turbine adjusts the speed.
In pumping mode the speed can be adjusted for maximal plant
efficiency using the available system power; reactive power
control is also possible in this mode.

References:

1. Leonhard, W.: Control of Electrical Drives; 2. Edition;

Springer 1996
2. Heller, M.: Die doppelt-gespeiste Drehtrommaschine für
Figure 12 – Decoupled speed and power control drehzahlvariable Pumpspeicherkraftwerke; Diss.; TU
Braunschweig 1998
3. Heller, M.; Schumacher, W.; Stability analysis of doubly-
3.2.2. Control scheme for pump mode fed induction machines in stator flux reference frame;
EPE 97.
In pumping mode there is a small change in the control 4. Arsudis, D.; Doppeltgespeister Drehstromgenerator im
scheme. In this case the pump is used for power control while Rotorkreis für Windkraftanlagen; Diss.; TU-
speed control is done in the electrical machine with the Braunschweig 1989
quadrature component of the rotor currents. To reach this the 5. Kuwabara, T.: Design and dynamic response
active power controller used in turbine mode cascaded with characteristics of 400MW adjustable speed pumped
the q-current controller is deactivated and a cascaded speed storage unit for Ohkawachi power station; IEEE Trans.
controller is inserted in its place. The mechanical speed step Energy Conversion; June 1996.
response can be seen in figure 13. The reactive power 6. Kudo, K.; Japanese experience with a converter-fed
controller is still active in this control scheme so that it is also variable speed pumped-storage system; Hydropower &
possible to control reactive power in pumping mode. Due to Dams; March 1994.
power adjustment done with the pump and reactive power 7. Kelber, C.; Schumacher, W.; Adjustable Speed Constant
control with the electrical machine the doubly-fed induction Frequency Energy Generation with Doubly-Fed
machine/converter system used for ASCF operation also Induction Machines; VSSHy 2000; European Conference
contributes in pumping mode to the power system stability. “Variable Speed in Small Hydro” in Grenoble, 2000.