Electric Power Systems Research 80 (2010) 14981505

Contents lists available at ScienceDirect

Electric Power Systems Research
j our nal homepage: www. el sevi er . com/ l ocat e/ epsr
A control system for dc arc furnaces for power quality improvements
G.C. Lazaroiu
a,
, D. Zaninelli
b
a
Department of Power Plants, University Polytechnic of Bucharest, Romania
b
Department of Energy, Politecnico di Milano, Italy
a r t i c l e i n f o
Article history:
Received 2 July 2008
Received in revised form 17 February 2010
Accepted 10 June 2010
Available online 7 July 2010
Keywords:
dc arc furnace
Light voltage icker
Harmonics
a b s t r a c t
The dc arc furnaces are highly disturbing loads for transmission and distribution networks. The present
paper proposes and implements an acdc converter control system for power quality improvements
taking into account both icker levels and harmonics at the point of common coupling. Different set of
parameters for the proportionalintegral controller are investigated for optimizing the dc arc furnace
power quality impact on the ac network. In this study both deterministic and random models of the
arc furnace are considered together with different structures of the acdc converter unit. The proposed
control system is tested on a real case application.
2010 Elsevier B.V. All rights reserved.
1. Introduction
High capacity dc arc furnaces, connected at the high voltage
level, cause light icker in the entire network [1,2]. The mea-
surement of this perturbation value at each bus of the network
is required for verifying the power quality with regards to other
connected customers and depending on standardized limits [3].
The dc arc furnaces are suppliedfromthe power systemthrough
various acdc conversion topologies. The main advantage of the dc
arc furnaces with respect to the ac arc furnaces is the possibility to
regulate the output voltage of the converter. Inthis way, the control
of the current suppliedtothe installationis allowedanda reduction
of the light icker can be determined. In addition, a fast opera-
tion control on the dc arc furnace, obtained with electronic devices,
permits the introductionof sophisticatedandefcient control tech-
niques. The operation with nonlinear power electronic devices also
causes harmonics. The identication of the harmonics presence in
the voltage waveform is possible by observing the prole of the
measured parameters during the operation process.
The ickermeter model, presented in detail in Ref. [4], is a con-
ventional analog one, in agreement with the models proposed
in current standards that of the necessity to comparative analy-
ses betweeninnovative equipment andconventional architectures,
whose disturbances were measured using analog instruments.
The present paper deals with the study and detailed analysis of
the controller structure for the acdc converter, highlighting the
advantages obtained regarding power quality improvement. Var-

Corresponding author.
E-mail address: [email protected] (G.C. Lazaroiu).
ious strategy of controls for the acdc converters in dc furnace
applications have been presented in the past [58]. However, they
do not focus on the detailed investigation of the acdc converter
controller.
2. dc arc furnace model
In this paper, the operation of a dc arc furnace in the
MatlabSimulink environment is studied and implemented in Sec-
tion 3 [9].
Compared to the arc furnace model, already dened and
described in Refs. [1013], the dc arc furnace simulation introduces
the acdc conversion modeling and the control systemrepresenta-
tion with the relevant regulation logics.
The acdc static conversion system is realized with a full con-
trolled Graetz bridge that supplies the operation power of the dc
furnace and stabilizes the current in the arc furnace.
The control system guarantees the correct operation of the
installation and a fast answer to the electric arc variations, in addi-
tion to the reduction, as much as possible, of the perturbations
injected in the power system. The control and regulation system
has to be a circuit that, with the help of feedback, allows stabilizing
the operation of the arc furnace with the maintenance of a certain
quantity (for example the dc current) at a dened value.
The acdc conversion and the controlregulation systems have
to reach the following objectives:

the correct operation of the arc furnace: maintaining scheduled

prole of energy, possibility of regulation, operation stability
(static and dynamic), high speed answer of the system to vari-
ations of the reference value;
0378-7796/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsr.2010.06.007
G.C. Lazaroiu, D. Zaninelli / Electric Power Systems Research 80 (2010) 14981505 1499
Fig. 1. Schematic diagram of the conversion system model without freewheel diodes.

the energy saving in the arc furnace process;

the proper acdc conversion of the voltage, with small dc voltage

ripple;

commutations with minimum perturbation injections in the

power system (compatible with the used devices and the opera-
tion conditions of the arc furnace);

the stabilization of the controlled quantity (guaranteeing the

reductionof icker andinterharmonics, without theuseof expen-
sive ltering systems and reactive power compensation);

the maximumconstructive simplicity (reducing the construction

investments of the arc furnace and, once realized, simplifying
maintenance and increasing the reliability).
In the following sections the acdc conversion and the control
systems are analyzed in detail. The possible congurations and
the motivations justifying the adopted solutions are highlighted.
In addition, where it was considered necessary, the sizing parame-
ters and implemented schemes of the circuit within the simulator
are reported for clarication.
3. acdc conversion system
The common used systems for dc arc furnaces applications [14]
are:

for 6-pulse acdc converter: two-star conguration with cou-

pling inductor or three phase Graetz bridge;

for 12-pulse acdc converter: two series connected three phase

bridge or two parallel connected three phase bridge.
Based on the objectives of constructive simplicity, regulation
possibility, reliability and maintenance facility, the Graetz bridge
solution is preferred with respect to the two-star conguration
(even if this one allows Joule losses reduction). This choice reduces
the complexity of the converter, allows an optimal control and
reduces the investment costs. In fact, it requires a two winding
transformer with respect to the three winding one used for the
two-star conguration, and it does not need the interposed induc-
tance.
In the present paper a 6-pulse converter is used in order to
compare the results with the ones reported in previous stud-
ies, conducted on arc furnaces with 6-pulse converter [10,15,16].
In this way, it is possible to evaluate the efciency of the
proposed control systemonthe dc arc furnace operation, for reduc-
ing the disturbances injected in the power system. If it is not
possible to respect the limits imposed by [3], the economic con-
venience of the 12-pulse converter can be evaluated instead of
using lters and/or static compensators for disturbance reduc-
tion.
In Ref. [17] different conversion systems for the dc arc furnaces
are tested and developed. They imply the combination of two tech-
niques:

freewheel diodes;

phase angle control.

These techniques have the purpose to reduce the variation of
the reactive power exchanged with the power system, reducing in
this way the icker phenomenon. With these methods also voltage
harmonics are reduced. The possibility of parallel resonance due to
lters and upstream network becomes less probable.
As a consequence of the current control, the variations of the
arc length lead to uctuations of the ring angle control of the
converter and to variations of the reactive power.
In the present work the 6-pulse acdc converter model with
ring angle control, with and without freewheel diodes, is imple-
mented and simulated.
Fig. 1 shows the converter model without freewheel diodes.
The inputs of the power conversion system are: the three phase
low-voltage values (v
a
, v
b
, and v
c
), and the instantaneous icker
sensation. The outputs of the power conversion system are: the
positive (+) andnegative () terminals onthe dc side; the dc current
absorbed by the load (I
dc
) and the reference current (I
ref
).
The model is composed by the following sub-blocks:
Pulse generator, PI controller and 6-pulse converter.
The Pulse generator sub-block automatically elaborates the
ring pulses for each thyristor of the converter bridge, using the
three phase line-to-line voltages and the signal from the PI con-
troller.
The PI controller sub-block establishes the operation of the
arc furnace. The control logic and the structure of this sub-block
are explained in detail in the following section.
The 6-pulse converter sub-block is the controlled 6-pulse
Graetz bridge with: thyristor resistance and inductance of 2
1500 G.C. Lazaroiu, D. Zaninelli / Electric Power Systems Research 80 (2010) 14981505
Fig. 2. Schematic diagram of the current controlled converter for supplying dc arc
furnace.
and 200.5mH respectively; snubber resistance and capacitance of
4 and 300F respectively.
The parameters of each freewheel diode are: resistance and
inductance of 0.1m and 1.6H respectively; voltage drop of
0.8V; snubber resistance andcapacitance of 3and200F respec-
tively.
These data refer to the existing plant taken as reference for the
case studied in Section 7.
4. Control system
The control of the thyristor bridge ring angles represents the
main goal of the present paper. Reference is made to [1820] and
the adopted control logic is based on proportionalintegral (PI)
regulation. The regulation, done at constant current in accordance
withthe existing industrial installations [21], is implementedinthe
simulator with reference to Fig. 2, [18].
The circuit shown in Fig. 2 is also described by the diagram
blocks reported in Fig. 3, where the converter and the ring circuit
are approximated by a linear delay that, for a 6-pulse converter
supplied at 50Hz, is set as T
T
=3.3ms.
In order to smooth the current signal in the feedback for the 6-
pulse converter, a lter characterized by a time constant T
F
=2ms
is chosen. This value is relevant to current sensors on the dc side.
Sensors on the ac side would have been less expensive, but much
slower. The delays corresponding to the converter and to the lter,
both small with respect to the time constant of the load, can be
resumed by an equivalent time T
E
=5.3ms.
For designing the control device it is necessary to know the
transfer function of the load in the frequency domain. For this rea-
son it is assumed that the electrical arc is modeled by an equivalent
variable resistance [15].
Due to the time-varying nature of the load, in order to precisely
size the PI controller a study of nonlinear systems, time-varying
and periodic would be necessary. Such study introduces many dif-
culties, so it is preferred to realize an approximate sizing of the
controller and to proceed, iteratively, to an experimental deter-
mination of its characterizing parameters. This can be done also
Fig. 3. Diagram of the current control loop through simplied blocks.
Fig. 4. Schematic diagram of the PI controller.
providing to the feedback system a sufciently large phase margin
for giving the capability to absorb load uctuations.
On the basis of the voltage and current measured on the arc fur-
nace real case applicationduringoperation, the average value of the
equivalent resistance for the electrical arc is deduced: R=0.004.
When the smoothing inductance, L =90H, is connected in series
with the electrical arc, the acdc system sees globally a load,
that in Laplace transformation, is characterized by the following
impedance Z(s) and admittance G(s):
Z(s) = 0.004 +9 10
5
s (1)
G(s) =
1
Z(s)
=
1
0.004 +9 10
5
s
(2)
The parameters of the regulator are
PI
=0.027, G
PI
=0.021 and
the transfer function of the PI controller is:
F
PI
(s) = G
PI

PI
s +1

PI
s
= 0.78
27 10
3
s +1
s
(3)
5. The proposed auto-adaptive control system
The above results have to be considered as an initial point for
analyses more elaborated, in order to establish the parameters of
the regulator that best guarantees the correct operation of the arc
furnace.
The structure of the used PI controller is reported in Fig. 4.
There are three input parameters: the coefcient for the
proportional part, the coefcient for the integral part and the
coefcient that is a constant of the output signal translation. The
saturation block does not allow the amplitude of the output signal
of the PI controller to exceed the acceptable limits.
Unless the coefcient , introduced in the controller so that the
output signal becomes positive andthe pulse generator canoperate
correctly, the transfer function of the control block implemented
in Simulink has an expression similar to (3). The architecture pre-
sented in Fig. 4 is perfectly compatible with the structure described
in the above section, by using block diagrams.
The values of these coefcients are set by an iterative simula-
tion process on the dc arc furnace operation. The iterative process
is an optimization process that establishes the parameters of the
controller. The objectives of the optimization process are:

the guaranteeing the stabilization of the absorbed current,

a short response time to the variations of the arc furnace mea-

sures,

ensuring the stability of the facility in case of load changes or

reference current variation.
The goal is to limit the icker effects and harmonic disturbances
onthe ac point of commoncoupling (PCC) and thus inthe transmis-
sion and distribution networks. As result of the simulation activity,
two values of parameters are obtained as listed in Table 1.
Both sets of parameters satisfy the operation requirements of
the dc arc furnace: guarantee a good stabilization of the absorbed
G.C. Lazaroiu, D. Zaninelli / Electric Power Systems Research 80 (2010) 14981505 1501
Table 1
Set of parameters used for the PI controller.
Parameters
Good compromise between the harmonics and the icker on ac network 0.1 0.05 2
Important reduction of the icker on ac network, to the detriment of harmonic perturbations 0.15 0.2 2
current, have a short time of response to the variations of the func-
tioning parameters, and are stable to the variations of load and
reference current. The purpose of the paper is to set a control strat-
egy that improves power quality.
The difference between the two sets of values consists
in the different disturbances on the ac supply system. The
rst set guarantees a good compromise between the harmon-
ics and the icker values in the ac network feeding the arc
furnace. With these parameters, the output of the block is
Output 1. The second set of values considerably reduces the
icker perturbation, but produces major harmonic disturbances.
With these parameters, the output of the block is Output
2.
Because both sets of values present advantages and disadvan-
tages, it has been decided to realize an auto-adaptive logic of
control: two controllers in parallel are implemented, where one is
sized with the rst set of parameters and the other with the second
one. The system, with the help of the feedback of the output signal
of the ickermeter, measures the instantaneous icker sensation in
the power system and chooses the more convenient control logic.
When it is possible to respect the icker emission limits using the
parameters that not produce an excessive harmonic disturbance,
the system chooses the output signal of the controller related to
the rst set of parameters. Otherwise the control system chooses
the set of parameters that guarantees the maximum reduction of
the icker, to the detriment of the harmonic disturbance.
This typeof implementationallows tosimultaneouslyobtainthe
advantages of both solutions, with a consequently optimization of
the dc arc furnace operation. Moreover, as both control systems
are always active during the furnace operation, the commutation
between the two sets of parameters is instantaneous. This commu-
tation can occur an innite number of times during an operation
process of the arc furnace without producing damage and does not
lead to anomalies neither inthe operationprole of the arc furnace,
nor in the current and voltage waveforms at PCC.
The proposed adaptive PI controller is shown in Fig. 5.
The output signal Pulses in Fig. 5 governs the ring pulses of
the thyristor generated by the block Pulse generator in Fig. 1. In
addition, it is necessary to implement an automatic system that
furnishes to the pulse generator the optimum control signal for
the instantaneous requests of the load, choosing between the two
possible outputs of the PI controllers (Output 1 or Output 2).
This is obtainedthrougha control that has as input theoutput signal
of the ickermeter. On the base of the disturbance amplitude, the
control logic selects at each instant of time the output signal from
the block with the most convenient parameter set.
Fig. 6. Block diagram of the ickermeter model.
The above described adaptive control system takes also in
account possible variation of the reference current with a delay
corresponding to the inertia of the icker measuring system.
6. UIE ickermeter model
The proposed auto-adaptive control uses the instantaneous
icker sensation. In order to obtain this measure, a ickermeter
is required. The Simulink block diagram of the International Union
for Electroheat UIE ickermeter model is shown in Fig. 6 [4], where
the absolute value of the output U
E
represents the instantaneous
icker sensation in accordance with the information supplied by
the current standard [22]. The functions of each block in Fig. 6 are
explained in detail in Ref. [4].
Once the trend in respect of the instantaneous icker sensation
was obtained, from the latter, the calculation of the indices, that
make it possible to evaluate the severity of the disturbance injected
intothe network, is conducted. This may be done by memorizing on
disk the sampling of the signal measuring the instantaneous icker
trend during simulation and the subsequent deferred analysis of
the results by means of the mathematical functionalities of Mat-
lab. In this way it is possible to enable the program to distribute
the instantaneous icker perception levels, based on frequency, to
weight the disturbance generated based on their duration, and to
return the accumulated icker probability curve and the index P
st
(short term icker severity).
7. Case studies
The goal of the case studied is mainly devoted to analyze the
system control of the above described dc arc furnace system. The
option of the presence of a freewheel diode in the converter struc-
Fig. 5. Proposed adaptive PI controller block diagram.
1502 G.C. Lazaroiu, D. Zaninelli / Electric Power Systems Research 80 (2010) 14981505
Fig. 7. Simplied diagram of the dc arc furnace power supply.
ture is investigated, too. Inorder toallowa better evaluationof each
control operation, the controller works with one of the two sets of
parameters described in Section 5, respectively. Obviously, during
normal operation, the controller chooses, at each time, the more
suitable set of parameter values , and as reported in Table 1.
Two different models of dc electric arc are considered: the
deterministic model with a sinusoidal time-varying disturbance
at 10Hz, and the random model with a white noise behav-
ior of the disturbance representation. The operation of each
control architecture for the two arc models is evaluated and
tested.
The system response at current reference value variation dur-
ing arc furnace functioning is considered in order to evaluate the
stability of the dc arc furnace at the furnace operation power varia-
tions. In this way the dynamic of the system is highlighted and the
answer to usual operations conducted by an operator during real
functioning is tested.
The case studied is referred to a real plant sited close to Brescia,
in North Italy. The measurements were done using the device ION
7600. At PCC the line-to-line voltage is 220kV, the short-circuit
power is 3500MVA and the rated power of the dc arc furnace
installation is 60MVA (see Fig. 7). The ac power supply system
(the ideal voltage source, the transmission line and the HV/MV and
MV/LV transformers), the rectifying circuit, constituted by a three
phase current controlled converter, and the dc arc furnace, behind
a smoothing reactance, are illustrated in Fig. 7 [23].
For a dc electric arc with sinusoidal disturbance at 10Hz, with
voltage variation between 225V and 600V, the waveform of the
dc current is shown in Fig. 8, where a variation on the reference
current (I
ref
) gives rise to a variation on the dc furnace current (I
dc
)
following the control dynamics. After the ring transient, the dc
current varies between 95kA and 105kA, with an average value
Fig. 8. Waveforms of the dc current and of the reference current during the power
furnace regulation.
Fig. 9. Waveform of the ac voltage at PCC (HV level).
of 100kA. The acdc conversion and the control systems contain
current uctuations inthe range of 5%of the imposedvalue, while
the error between the average value of dc current and the reference
value is 0.68%.
The variations of the RMS values of the voltage and current at
the PCC over the period of monitoring are shown in Figs. 9 and 10,
respectively. The power quality metering was conducted on a
period of two weeks. The arc furnace down period is clearly evi-
dent in all graphs. The spectrum analysis of the measured voltage
at PCC is reported in Fig. 11, where the fundamental component
is truncated. The observation duration corresponds to the period
when the arc is constantly burning during the rening stage.
For the dc arc furnaces two stages were considered:

the melting stage: non-stationary and is characterized by icker;

the rening stage: stationary and can be characterized by the

harmonics. The process was considered periodical.
Simulation studies are performed on the above described sys-
temwhere every combination of the following factors is taken into
account:

arc model: deterministic or random one as reported in the tech-

nical literature [12,2426];

control parameters: proposed values optimized for a good com-

promise between harmonic and icker on ac network and
proposed values optimized for icker reduction on ac network
in detriment of harmonic perturbations;
Fig. 10. Waveform of the ac current at PCC (HV level).
G
.
C
.
L
a
z
a
r
o
i
u
,
D
.
Z
a
n
i
n
e
l
l
i
/
E
l
e
c
t
r
i
c
P
o
w
e
r
S
y
s
t
e
m
s
R
e
s
e
a
r
c
h
8
0

(
2
0
1
0
)

1
4
9
8

1
5
0
5
1
5
0
3
Table 2
Power quality indices.
Percentage waveform
distortion indices and Pst
Control parameters Arc model THDv TIHDV TDV THDI TIHDI TDI P
ST
The model of the arc
furnace without
freewheel diodes
Optimized control for a
good compromise
between icker and
harmonics injected on
ac network
Deterministic arc
model
LV 7.28 7.77 10.65 9.55 1.94 9.75 0.64
MV 3.72 3.99 5.46 9.57 2 9.78
HV 1.36 1.47 2 9.57 2 9.78
Random arc model LV 7.11 7.81 10.56 5.55 3.03 6.33 0.55
MV 3.65 4.01 5.42 5.64 3.05 6.41
HV 1.34 1.47 1.99 5.63 3.05 6.4
Optimized control for
an important reduction
of the icker on ac
network, to the
detriment of harmonic
perturbations
Deterministic arc
model
LV 9.39 5.11 10.69 9.54 1.94 9.73 0.62
MV 4.83 2.55 5.46 9.48 1.98 9.68
HV 1.76 0.93 1.99 9.54 1.94 9.73
Random arc model LV 8.76 6.09 10.67 5.57 3.02 6.33 0.54
MV 4.5 3.05 5.44 5.62 3.02 6.38
HV 1.64 1.11 1.98 5.61 3.02 6.37
The model of the arc
furnace with freewheel
diodes
Optimized control for a
good compromise
between icker and
harmonics injected on
ac network
Deterministic arc
model
LV 9.77 2.98 10.21 5.85 4.86 7.6 0.53
MV 6.21 0.69 6.25 8.87 3.8 9.65
HV 2.26 0.49 2.31 8.97 3.79 9.64
Random arc model LV 14.39 1.06 14.43 8.42 2.76 8.86 0.53
MV 6.21 0.69 6.25 8.87 3.8 9.65
HV 2.26 0.49 2.31 8.97 3.79 9.64
Optimized control for
an important reduction
of the icker on ac
network, to the
detriment of harmonic
perturbations
Deterministic arc
model
LV 13.66 6.7 15.22 8.38 2.97 8.89 0.46
MV 6.12 3.46 7.03 8.5 3.83 9.32
HV 2.26 1.3 2.62 8.55 3.8 9.35
Random arc model LV 13.48 7.07 15.22 9.1 1.97 9.31 0.44
MV 5.85 3.73 6.94 9.32 2.93 9.76
HV 2.18 1.4 2.59 9.54 2.91 9.97
1504 G.C. Lazaroiu, D. Zaninelli / Electric Power Systems Research 80 (2010) 14981505
Fig. 11. Spectrum analysis of the voltage waveform at PCC.

acdc converter structure: with or without freewheel diode.

Results are summarized in Table 2, where the main power
quality indices are evaluated in each case on the basis of recom-
mendation reported in Ref. [27].
The indices THD, TIHD, TD (for voltage and current) measured
at all voltage levels and the short term icker severity value at PCC
appear in Table 2 are given by:
THD
I
=

40

k=2

I
k
I
1

2
100, THD
V
=

40

k=2

V
k
V
1

2
100 (4)
TIHD
I
=

40

k=2

I
i
I
1

2
100, TIHD
V
=

40

k=2

V
i
V
1

2
100
TD
I
=

THD
2
I
+TIHD
2
I
, TD
V
=

THD
2
V
+TIHD
2
V
where V
1
, I
1
are the rms values for voltage and current on the fun-
damental, V
k
, I
k
the rms values for current and voltage on the k-th
harmonic order, and V
i
, I
i
the rms values for current and voltage on
the i-th interharmonic order.
Without the optimized control systemthe P
st
(50%) value is 3.72,
THD
V
value is 8.28% and the THD
I
value is 41.06%.
In Fig. 12 the variation of the P
st
values is illustrated, while
Figs. 13 and 14 shows the waveforms of the power quality indices
THD
V
and THD
I
. The 95% value of the short term icker severity
Fig. 13. Variation of THDV values.
value is P
st
(95%) =0.9. The standard EN 50160 [28] stipulates that
the icker severity index at the high voltage level, such that at the
lowvoltage level tobe 1, canexceed1considering the transfer coef-
cients fromthe lowvoltage tothe highvoltage level. The P
st
values
were computed on each ten minutes interval fromthe observation
period.
Some interesting comparisons on these results are made in
Figs. 15 and 16 regarding the indices THD, TIHD, TD.
The 50% values of the P
st
are within the maximum value limit
of 0.85 for a single customer imposed by the Italian standard [3]
by more than 15%. This allows the connection of the plant with-
out the necessity to introduce dynamic compensating systems for
the reactive power control. Moreover, the distance between the P
st
obtainedvalues andtheimposedlimit permits totheproposedcon-
trol systemto operate with the parameter values devoted to a good
compromise between the harmonics and icker on ac networks. In
this way, the harmonic distortion indices of the voltage at PCC are
minimized.
In the present study ltering systems are not included in order
to evaluate the effective installation disturbances. Introducing l-
ters tuned at low order harmonics would reduce harmonic voltage
distortions at PCC.
Fig. 14. Variation of THDI values.
Fig. 12. Variation of Pst values.
G.C. Lazaroiu, D. Zaninelli / Electric Power Systems Research 80 (2010) 14981505 1505
Fig. 15. Percentage waveform distortion indices at PCC for deterministic model of
electric arc.
Fig. 16. Percentage waveformdistortionindices at PCCfor randommodel of electric
arc.
The harmonic distortion level increases with the introduction
of the freewheel diodes, due to two factors that characterize this
installation typology:

the introduction of an additional nonlinear element;

a not ideal sizing of the PI controller parameters for this particular

change on the converter architecture.
In the present study intentionally the controller parameters are
not varied with the introduction of the freewheel diodes for the
following two reasons:

the willingness to test the reductions of the icker disturbance

by introducing freewheel diodes not modifying the other charac-
teristics of the installation;

the intention to verify the reliability and stability of the proposed

control system in case of modications of the plant structure.
8. Conclusions
The high capacity dc arc furnaces connected to the HV level can
determine power quality perturbations in transmission and distri-
bution networks. dc arc furnaces have the important advantage to
control the output voltage of the acdc converter.
Control system assumes a key role in the furnace operation and
regulation strategies can be studied for improving power quality at
the point of common coupling. Using a set of iterative simulations
the parameters of the PI controller have been dened for consider-
ing both the industrial necessities of a correct functioning of the dc
arc furnaceandtherespect of power qualityrequirements. Twosets
of optimized parameters for the acdc converter controller have
been obtained: the rst one offers a good compromise between
the harmonics and the icker reduction on ac network; the sec-
ond one strongly limits the icker disturbance on ac network with
detriment of harmonic perturbations.
The present paper proposes and implements a control system,
which is based on the icker level on the ac network and chooses
in real time the most convenient set of parameters for the PI con-
troller.
This active control systemis testedonthe existing dc arc furnace
facility considering two types of arc modeling (deterministic and
random one) and two acdc converter structures (with or without
freewheel diodes). For all the cases investigated, the main power
qualityindices regardingharmonics andicker levels arecomputed
and reported in the paper with reference to IEEE standards.
References
[1] L. di Stasi, Electric Furnaces (in Italian), Patron Editore, Bologna, Italy, 1976.
[2] E. Tironi, Electro-Thermal Applications (in Italian), CUSL, Milan, Italy, 2000.
[3] Connection criteria for user electrical plants with ac arc furnaces, Standard
ENEL DK5550, Rome, Italy, March 1991.
[4] A. Bertola, G.C. Lazaroiu, M. Roscia, D. Zaninelli, AMatlabSimulinkickermeter
model for power quality studies, in: Proceedings 11thInternational Conference
on Harmonics and Quality of Power, Lake Placid, USA, 2004.
[5] L. Tang, D. Mueller, et al., Analysis of DCarc furnace operationandicker caused
by187Hzvoltagedistortion, IEEETrans. Power Deliv. 9(April 1994) 10981107.
[6] V. Scaini, T. Ma, High current DC choppers in the metals industry, in: Proceed-
ings IEEE Industry Applications Conference, Rome, October 812, 2000.
[7] P. Ladoux, G. Postiglione, H. Foch, J. Nuns, A comparative study of AC/DC con-
verters for high-power DC arc furnace, IEEE Trans. Ind. Electron. 52 (June 2005)
747757.
[8] G. Carpinelli, A. Russo, Comparison of some active devices for the compensa-
tion of DC arc furnaces, in: Proceedings Power Tech Conference, Bologna, June
2326, 2003.
[9] A. Cavallo, R. Setola, F. Vasca, Matlab, Simulink and Control Toolbox Operative
Guide (in Italian), Liguori Editore, Italy, 1994.
[10] A. Cavallini, G.C. Montanari, L. Pitti, D. Zaninelli, ATP simulation for arc furnace
icker investigation, Eur. Trans. Power Eng. 5 (MayJune 1995).
[11] G.C. Montanari, M. Loggini, A. Cavallini, L. Pitti, D. Zaninelli, Arc-furnace model
for the study of icker compensation in electrical networks, IEEE Trans. Power
Deliv. 9 (October 1994) 20262036.
[12] E. ONeill-Carrillo, G.T. Heydt, E.J. Kostelich, S.S. Venkata, A. Sundaram, Nonlin-
ear deterministic modeling of highly varying loads, IEEE Trans. Power Deliv. 14
(April 1999) 537542.
[13] F. Chen, K.B. Athreya, V.V. Sastry, S.S. Venkata, Function space valued Markov
model for electric arc furnace, IEEE Trans. Power Syst. 19 (May 2004) 826833.
[14] S. Crepaz, Static Conversion of the Electrical Energy (in Italian), Citt Studi
Edizioni, Italy, 1995.
[15] G. Carpinelli, M. di Manno, P. Verde, E. Tironi, D. Zaninelli, AC and DC arc fur-
naces: acomparisononsomepower qualityaspects, in: Proceedings IEEEPower
EngineeringSocietySummer MeetingConference, Edmonton, July1822, 1999.
[16] G.C. Montanari, M. Loggini, L. Pitti, E. Tironi, D. Zaninelli, The effects of series
inductors for icker reductioninelectric power systems supplyingarc furnaces,
in: Proceedings Industry Applications Society Annual Meeting Conference,
Toronto, October 28, 1993.
[17] M. Wursteisen, J. du Parc, C. Glinski, Converters with low disturbances for the
electric power supply of DC furnaces, in: Proceedings Fifth European Electric
Steal Congress, Paris, June 1995.
[18] W. Leonhard, Control of Electrical Drives, Springer, 1996.
[19] S. Bittanti, N. Schiavoni, Modeling and Control (in Italian), Citt Studi Edizioni,
Italy, 1996.
[20] W.J. Palm III, Matlab for Engineering (in Italian), McGraw Hill, Italy, 1999.
[21] G. Ferli, DCarc furnace: comparisonwiththeac arc furnaceandrst experimen-
tal results (in Italian), in: Proceedings ASSOFOND XIX Congresso di Fonderia
Santa Margherita Ligure, Genova, 30 September 1988.
[22] IEC Standard 868, Flickermeter. Functional and design specications, 1996.
[23] A. Bertola, DC electric furnaces: modeling, control and disturbance analysis (in
Italian), M.Sc. Thesis, Tutor D. Zaninelli, G. Carpinelli, Politecnico di Milano,
Milan, Italy, Academic Year 1999/2000.
[24] G. Carpinelli, F. Iacovone, A. Russo, P. Verde, D. Zaninelli, DC arc furnaces: com-
parisonof arc models toevaluate waveformdistortionandvoltage uctuations,
in: Proceedings, College Station, October 2001.
[25] G. Carpinelli, F. Iacovone, A. Russo, P. Varilone, Chaos-based modeling of DC
arc furnaces for power quality issues, IEEE Trans. Power Deliv. 19 (Oct 2004)
18691876.
[26] K.T. Alligood, T.D. Sauer, J.A. Yorke, ChaosAn Introduction to Dynamical Sys-
tems, Springer-Verlag, New York, 1996.
[27] IEEE 519-1992, IEEE recommended practices and requirements for harmonic
control in electric power systems.
[28] EN 50160, Voltage characteristics of electricity supplied by public distribution
networks, 2008.