Está en la página 1de 9

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/279206134

Numerical EAF oscillation model and


operational validation
Conference Paper May 2013

CITATIONS

READS

66

5 authors, including:
Hans-Jrgen Odenthal
SMS group GmbH
120 PUBLICATIONS 303 CITATIONS
SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Modelling of multiphase melt-slag-gas flows for metallurgical reactors, e.g BOF, AOD, EAF
etc. View project

All in-text references underlined in blue are linked to publications on ResearchGate,


letting you access and read them immediately.

Available from: Hans-Jrgen Odenthal


Retrieved on: 24 November 2016

AISTech 2013 The Iron & Steel Technology Conference and Exposition, 6-9 May 2013, Pittsburgh, PA., USA, not on conf. CD

Numerical EAF Oscillation Model and Operational Validation


Pavlo Grygorov, Mathias Lttenberg, Markus Reifferscheid, Frank Theobald, Hans-Jrgen Odenthal

SMS Siemag AG
Eduard-Schloemann-Strae 4, 40237 Dsseldorf, Germany
Tel.: +49 (0)211-881-5316
Fax: +49(0)211-881-4997
E-mail: pavlo.grygorov@sms-siemag.com

Keywords: Electric Arc Furnace (EAF), electric arc model, hydraulic model, electrode control optimization, experimental validation.
ABSTRACT
In this paper a dynamic Matlab/Simulink model of a three-phase 150 MVA AC electric arc furnace (EAF) is presented. The model
describes the main elements of the EAF, such as, the hydraulic system, the electric circuit including electric arcs, the mechanics of the
electrode arms and the electrode control. The latter comprises an impedance controller and control logics to detect respective states of
the process. The mechanical system, electrodes and horizontal current-conducting arms, is described by a two-beam model, in which
the measured eigenfrequencies of the electrode-arm system of the real furnace are reproduced. Special attention is paid to the
influence of electromagnetic forces on the movement of electrode arms. The simulation results are compared with operational data of
the EAF.
INTRODUCTION
Due to its complexity the mathematical modeling of EAF combining its electrical, mechanical and hydraulic components still remains
a challenge [1]. In the last decade, many models dealing with the certain EAF problems, e.g. the flicker mitigation or optimization of
the electrode control have been suggested [2], [3]. In this paper a mathematical model of the three-phase 135 t AC EAF is presented.
The model is validated by using the available operational data.
The paper is organized as follows: in the next paragraph the main elements of the EAF model are briefly reviewed. After that the
results of simulations are discussed and compared with the available operational data. After the validation, the EAF model is used for
simulation of a one phase short circuit as well as resonance oscillations of the electrode supporting arms. Finally, the main results are
summarized and an outlook for future investigations is given.
MATHEMATICAL EAF MODEL
The main elements of the EAF model are shown in Figure 1. The electrical inputs of the model are secondary transformer voltages
defined as
 =  sin2
,

1

 =  sin2
+ 2
/3 ,

2

 =  sin2
+ 4
/3 ,

3

where  is the phase voltage, f is the network frequency 50 Hz and t stands for the time. The equivalent electric circuit is described
by the system of differential equations

1
=
    

 

!, ",

4


 
=

,




5


1
=
    

 

6

!, ",

where % is the phase inductance, % is the ohmic phase resistance and  !,% stands for the arc voltage. Here only two currents are
independent, and the vector sum of all currents always adds to zero. The induced error voltage is given by [4]
 = &,



&,
,



7

where &, stands for the mutual inductance between the loop 1-2 and the loop 3-M.

Figure 1: Schematic representation of the EAF model


The electric arc is described as an impedance load, whose resistance is described by the phenomenological model [5]. In this model
the ohmic resistance Rarc is coupled with the deflection of the electric arc ( as


=  exp, ( ,

8

where , = 8 describes the deflection capability of the electric arc and  = 2m represents the ohmic resistance of long vertical arcs.
Finally, the arc reactance can be written as a quadratic variation of the arc resistance [6]
.

= 0.12  ! + 0.02 

9

The length of the electric arc is changing during the melting process due to its chaotic motion and movement of the electrode tip. In
this way the secondary system of the EAF (vertical mast, electrode supporting arm and graphite electrode) can influence the electric
circuit parameters. The position of the electrode tip can be found from the solution of equation of motion

23% + 245% 26% + 5%
2% = 7% ,

 = 8, 9, :

10

where 4 is the damping coefficient, 0 is the angular eigenfrequency of the secondary system and F is the net force acting on the
secondary system. For the outer phases the electromagnetic forces are directed to push the electrodes and supporting arms away from
the central phase in horizontal plane. For the outer phase the major component of the net force is in vertical plane. Here a simplified
two-beam model is used to connect 4 and 0 with the construction parameters and material properties of the secondary system [5].

The electrode length in the model is adjusted to reproduce measured damping coefficients and eigenfrequencies of the electrode-armsystem for each phase. The measurements were performed on the outer and central electrode supporting arms by using accelerometers
placed close to the electrode holders. The measured value of eigenfrequencies 0 and damping coefficients 4 are given in Table I.

Figure 2: Schematic representation of the EAF hydraulic electrode positioning system

The hydraulic system consists of hydraulic cylinders, pressure tank, pumps and the control valves (Figure 2). The tank and the pump
provide a constant supply pressure of the hydraulic fluid. The signal of electrode control drives the valve changing the flow from the
tank to the electrode control cylinders and thus determining the speed of the mast. The control valve 4WRTE (Bosch Rexroth)
& = 220 l/min of hydraulic liquid at pressure drop per edge p = 5 bar. The model of the valve also
provides the nominal flow Q
incorporates interior leakage and represents the dynamic behavior with respect to cut-off frequency and phase shift. The cylinders lift
the secondary system to achieve a desired vertical position of the electrodes. The total mass of the secondary system mtotal ~ 15000 kg
provides the hydraulic pressure p0 ~ 75 bar. The controller output voltage uc is in the interval from -10 V to 10 V, changing the relative
valve opening from -100 % to 100 %. At 100 % signal the respective electrode will be lifted with the maximum speed of 220 mm/s.
The model of the hydraulic cylinders considers the inner friction of the piston and the compressibility of the fluid. Finally, the real
system of hydraulic pipelines (each line from valves to cylinders) is replaced in the model by one physically equivalent pipeline,
reproducing the same kinetic energy and the compressibility modulus of the hydraulic fluid (Eliq = 2109 N/m2), propagating time of
the pressure wave and the pressure drop at a stationary flow. The pipeline model also takes into account the stiffness of the tube
material. Thus each equivalent pipeline is represented by a multi-element mass-spring-system to describe hydraulic pressure
oscillations.

Central phase
Outer phase

Horizontal oscillations
Vertical oscillations
Horizontal oscillations
Vertical oscillations

4
0.032
0.018
0.042
0.028

0 [ s-1]
10.6
17.8
9.5
14.5

Table I. Measured damping coefficients and angular eigenfrequencies.


The control valves are operated by proportional (P-) controllers. For each phase the P-controller monitors the deviation between the
reference and the averaged measured currents
; =  /<

=> ,

where the root mean square (rms) values of the measured currents and voltages are used

(11)

 = ?    ,
@

 = ?     .
@

(12)

The controller output voltage signal rises linear with the deviation as
! = A ;,

(13)

where kp is a gain factor (kp > 0). If > 0 the secondary current should be lowered by lifting the electrode and, thus, by increasing the
arc length. The main goal of the controller is to maintain the current difference as close to zero as possible.
The described EAF model is implemented in Matlab/Simulink with a discreet time-step T = 10-3 s. To reduce the computational time
the sub-models were programmed in C and converted to Matlab S-Functions.

Figure 3: Simulated arc initiation. From top to bottom: position of electrodes 1 and 2, current phase 1, current
phase 2, voltage phase 1, voltage phase 2.

RESULTS OD SIMULATION AND EXPERIMENTAL VALIDATION


In order to validate the model the results of simulations are compared with the available EAF operational data, including three groups
of signals. The electrical signals, e.g. the secondary currents and phase voltages were recorded with 20 kHz sampling rate. The second
group, the accelerations of electrode arms, measured on the outer and central arms close to the electrode holders, by using the state of
the art accelerometers for 3D measurements with a sampling rate of 500 Hz. The third group, recorded at 1 Hz, comprises the rms
values of the currents, voltages, electric power, power factor, pressure in hydraulic cylinders and other signals used to control the EAF
melting process.
In Figure 3 the simulated arc initiation is presented. In the first line the relative position of electrode tips is shown. In this model it is
assumed, that the electric arcs can be initiated after the electrode tips reaches the limit 0.6 m. The regulation system lowers the
electrodes with the maximal descent speed to reach the scrap (limit 0.6 m) at the shortest time. As soon as the electrode 1 (blue curve)
touches the cold scrap it is stopped (t = 1.5 s) to prevent the electrode damage. The line voltage of phase 1 drops to zero, since the
scrap and the EAF vessel are connected electrically to ground. At the same time the voltages of phase 2 and 3 are increased by 3 ,
that improves the initiation conditions. After the electrode 2 (red curve) reaches the scrap surface (t = 2.2 s) two electric arcs are
ignited. At time t = 2.8 s the third electrode (not shown here) reaches the scrap. The results of simulation can be compared with the
recorded operational data, presented in Figure 4. The results of simulation are in a good qualitative and quantitative agreement with

the measured currents and voltages. Here should be noticed that the stochastic behaviour of the electric arcs is left out of consideration
in the present work.

Figure 4: Measured arc initiation. From top to bottom: position of electrodes1 and 2; current phase 1,
current phase 2, voltage phase 1, voltage phase 2. Due to technical restrictions the voltage amplitudes
up to 1000 V are shown.

Figure 5: Measured process data: position of electrode 1, power factor, electric power, valve opening,
hydraulic pressure, horizontal acceleration of electrode 1.
In Figure 5 the operational data of phase 1 during the scrap boredown are shown. During this period, which lasts several minutes, the
electrodes sink to the bottom of the EAF vessel melting the scrap The first few seconds the regulation system lowers the electrode 1
with the descent speed about 150 mm/s (valve opening -70 %). This speed decreases to 20 mm/s after the ignition of the electric arcs.

The power factor cos is in the range from 0.7 to 0.8, thus the total power input about 100 MW is achieved. The pressure in hydraulic
cylinder fluctuates around its mean value p0 = 75 bar. The measured horizontal acceleration of the outer electrode arm is presented in
the last line of Figure 5. The stochastic fluctuations up to 0.5 g (g = 9.81 m/s2) are due to the electromagnetic forces acting on the
electrode and supporting arm, whereas the mean value is about 0.3 g. The operational data can now be compared with the results of
simulations presented in Figure 6. Three electrodes are lowering until at time t = 2.2 s electric arcs of phases 1 and 2 are ignited. This
time period is characterized by a step-like behavior of the total power consumption from 0 to 46 MW. The third electric arc is ignited
at t = 2.8 s. After that the regulation system lowers the electrodes to achieve a defined electric set point by changing the arc lengths.
The resulted total power is slightly below 100 MW and the power factors cos equal 0.75, being in a good agreement with the results
of measurements. Since the geometry and material properties of the secondary system in the model were adjusted to those of the real
EAF, the simulated total pressure in hydraulic cylinders reproduces the experimental values (75 bar) with a high accuracy. Finally, the
horizontal acceleration of the supporting arm 1 was calculated from the electromagnetic and elastic forces. It is obvious that the
knowledge of the secondary currents together with the material properties allows reproducing the net force (acceleration), acting on
the supporting arm. After the validation the EAF model can be used to investigate different regulation strategies and occurrence of
unstable state of EAF during the melting process.
During the scrap boredown period a short circuit can occur in one or more phases due to a scrap collapse. In this extreme case the
regulation system must reduce the secondary current by lifting the electrodes. Since there is no clear evidence of such event in the
given operational data, the short circuit was simulated by a sudden reduction of the arc length. An example of the dynamical modeling
of the short circuit is shown in Figure 7. At the time t = 5 s from the beginning of simulation the short circuit on the phase 3 (zero arc
length) takes place. Due to the coupling of all three currents a significant increase in secondary currents on the phase 3 as well as in
the unaffected phases 1 and 2 is observed. The regulation system detects the short circuit by comparing the measured phase currents
with a maximal critical value. To minimize the negative influence of the short circuit on the transformer all electrodes are raised with a
maximum speed. This simulated response of the electrode regulation system to the short circuit is in qualitative agreement with the
logic of real electrode controllers [7].
To demonstrate how the developed EAF model can be applied for the optimization of the electrode control, a particular case of
resonant oscillations of the supporting arms is considered. Such mechanical resonances of the secondary system occurring during the
scrap boredown are caused by periodic modulations of the arc voltages, if the frequency of modulation is close to the eigenfrequency
of the supporting arms (see, e.g. [5], [8], [9]).

Figure 6: Simulated process data: position of electrode 1, power factor, el. power total, valve reference, hydraulic
pressure, horizontal acceleration of electrode 1.

Figure 7: Simulation of short circuit: position of electrodes, secondary currents and valve opening.

Figure 8: Simulation of the resonance oscillations on phase 1. Damping of vertical oscillations by adjusting the
gain factor kp.
It is assumed, that a periodic horizontal displacement of the electrode tip changes the arc length and its voltage. In our model the
periodic modulation of the arc voltage on phase 1 is simulated by changing the arc voltage amplitude  !  as
 !  =  !  1 A cos ,

(14)

with A = 0.7 and the frequency of modulation  equals to the natural frequency of the secondary system in horizontal plane 5F . The
electrode control reacts on the periodic changes of the secondary current and voltage on phase 1 by lifting and lowering the respective
supporting arm. In Figure 8 the resulted position of the electrode 1 is shown by the blue dot-dashed curve. As shown in this figure, the
amplitude of vertical oscillations rises with time, reaching approximately 18 cm in 5 seconds after the start of simulation. In a real
furnace such vertical oscillations lead either to the electrode breakage or to arc extinction (unstable melting process). Other words, the
horizontal displacement of the electrode tip can lead to resonant vertical oscillations of the secondary system. There is a bunch of
approaches to damp the resonant oscillations of the EAF secondary system. In this study the so-called adaptive control is investigated
[10]. This method mainly consists in adjusting the parameters of the electrode control to get the stable EAF melting process. Since the
integral and derivative components are not considered in the model, only the proportional gain factor A can be changed in (13). The
electrode position with the adjusted gain factor is shown in Figure 8 by the red solid curve. By reducing the value of A from 210-5 to

0.610-5 the resonance can be significantly suppressed: the maximum amplitude of oscillations of about 5 cm (red curve) is reached in
10 s. In this way the EAF model can be used for a adjusting the parameters of electrode control and improving the control design.
Moreover, a dynamic gain factor A  , reacting on undesired mechanical oscillations of the secondary system, can be introduced.
SUMMARY
A mathematical model of EAF is presented describing the main elements of the furnace, such as the high-current system, electric arcs,
secondary system, electrode positioning system and the electrode control. The model implemented in Matlab/Simulink was validated
by using the operational data of a real 135 t 150 MVA AC EAF commissioned by SMS Siemag AG. The comparison between the
simulated and measured operational data shows that the developed model reproduces with the high accuracy the main EAF signals
during the arc initiation and the scrap boredown. The validated model is used for the investigation of the response of electrode control
to the short circuit. Furthermore, the model is used for the optimization of the electrode control with respect to the resonance
oscillations of secondary system, which can occur during the scrap boredown. By scaling the main elements, the presented model can
be used as optimization tool for construction of new furnaces. The model can also be extended to include the integral and derivative
components of the electrode control as well as the stochastic fluctuations of the arc length. In this way, the electrode control design
can be optimized during the commissioning phase. This leads to reduction of the commissioning time and operating costs.
The Matlab/Simulink model used together with the computational fluid dynamics (CFD) simulation of the melting process provides a
full self-consistent description of the EAF process [11]. By using a combined strategy of the mathematical modelling, numerical
simulation and the plant test, SMS Siemag continuously improves the design and the operating conditions of the EAF.
REFERENCES
[1] V. Logar et al., Mathematical Modeling and Experimental Validation of an Electric Arc Furnace, ISIJ International (2011), Vol.
51, p.382
[2] K. Anuradha et al., Electric Arc Furnace Modeling and Voltage Flicker Mitigation by DSTATCOM, Industrial and Information
Systems (2008), p. 1
[3] K. Krger, Anforderungen an eine modern Elektrodenregelung fr Drehstrom-Lichtbogenfen, Elektrowrme International
(2007), Vol. 4, p.245
[4] B. Bowman and K. Krger, Arc Furnace Physics, Verlag Stahleisen GmbH, Dsseldorf (2009), p. 246
[5] B. Remus, Analyse elektromechanischer Beanspruchungen von Elektroden in Lichtbogenfen, PhD Thesis, University HamburgHarburg (1984), p. 98
[6] S. Khle, Ersatzschaltbild und Modelle fr die elektrischen Gren von Drehstrom-Lichtbogenfen, Verlag Stahleisen,
Dsseldorf (1990), p. 276
[7] Electrode control DecNT, Ferrotron Technologies GmbH (2007)
[8] H.- J. Kunze, ber die elektrodynamische Beanspruchung von Elektroden in Lichtbogenfen, PhD Thesis, University HamburgHarburg (1981), p. 102
[9] H. Faber, Ein Beitrag zur Ursache periodischer Spannungschwankungen in Drehstrom-Lichtbogenofen, PhD Thesis, University
Hamburg (1979), p. 85
[10] B. Schwartz and K. Timm, Control of electrical quantities in three-phase a. c. arc furnaces, Elektrowrme International (1989), B
1, p. B 12
[11] P. Grygorov et al., Plant measurements and numerical simulation of EAF operational data, 30th JSI, Paris (2012), p. 34

También podría gustarte