Beruflich Dokumente
Kultur Dokumente
1 2 3 [ I i ] = [G j ][ xi ] + [ Bi ]
i
2V1c1 ( Gs + G ) − V2 c1 ( Gs + G )
V1c2 ( Gs + 2G ) − V2 c2 ( Gs + G ) V1c3 ( Gs + 2G ) − V2 c3 G −
−V3 c1 ( Gs + G ) + c1c2 x2 (V1 − V2 )
−V3 c2 G V3 c3 ( Gs + G )
+ c1c3 x3 (V1 − V3 )
−V1c2 ( Gs + G ) + 2V2 c2 ( Gs + G )
I1 x1 ( 2V1 − V2 − V3 )
x + ( s
I = 1 −V1c1 ( Gs + G ) + V2 c1 ( Gs + 2G ) −V1c3 G − V2 c3 ( Gs + 2G ) G 2 + 3GGs )
2 G −V c G −V3 c3 ( Gs + G ) + c2 c3 x3 (V2 − V3 ) 2 ( −V1 + 2V2 − V3 )
3 1 −V
3 3 sc ( G + G ) G
( −V1 − V2 + 2V3 )
+ c1c2 x1 (V2 − V1 )
I 3 TOT
x3
TOT
−V1c3 ( Gs + G ) − V2 c3 ( Gs + G )
−V1c1 ( Gs + G ) − V2 c1G −V1c2 G − V2 c2 ( Gs + G )
+2V3 c3 ( Gs + G ) + c1c3 x1 (V3 − V1 )
−V3 c1 ( Gs + 2G ) +V3 c2 ( Gs + 2G )
+c2 c3 x2 (V3 − V2 )
x3 M ATLAB
500 M ux Dem ux
4 Initial Model of EAF V1
Function m ag_i3
conductance1
Now that the EAF electrical circuit is solved, the next step is to 500
m ag_p1
V2
implement Equation 8 as an open-loop system to test different 500
cases for several electrode positions. For the output current V3 m ag_p2
matrix Ii, this conductance matrix Gij depends on the electrode 0.1
x1 delay m/sec/%
s
Saturation
Figure 6 shows the closed-loop system for the EAF.
1
0.1
s 1
x2 delay m/sec/% Saturation PID 0.0
s
PID Controller Saturation3 Transport Gain
500 Mux Delay2
1
1 V1 PID 0.0
s
0.1 Gain1
PID Controller2 Saturation4
s 500 Transport
Delay1 Mux
500
x3 delay m/sec/% Saturation V2 1 V1
PID 0.0 500 MATLAB
s conductance1
500 PID Controller1 Saturation5 Transport Gain2 V2
Function
Delay 500
V3 V3
0.1
0.1 G
err1 Mux
G
err2
Mux err3
Noise_I1 Noise_I2 Noise_I3 Noise_P1 Noise_P2 Noise_P3
I2_des 0.025s+1
actual time it takes for the valves to open/close as desired. The Demux
1
0.025s+1
mag_i3
0.025s+1
0.025s+1
mag_p3
Demux
10000
7 Conclusion
8000
Current (Amps)
3
and subtracted from the desired current to form an error signal.
The error signal was transformed to a percent error by a PID
2
controller. All simulations demonstrated throughout the project
1 Mag P1
Mag P2
implemented a P-controller with a 0.0045 gain. Adding the
0
Mag P3
differentiator and integrator gains will improve the step
1 2 3 4 5
Time (sec)
6 7 8 9 10
response. The coupling effects between electrodes were also
examined.
Figure 8: Closed-Loop Power Magnitudes (Desired Power =
5MW)
Future goals for the three-electrode arc furnace simulator study
Since the system model already contains an integrator (in order include further testing for all coupling currents, the development
to change electrode speed into electrode position), the effect of of optimal decoupling controllers, and linearizing the system in
adding an I-gain will be negligible. The next simulation involves order to implement a state-space controller. Analysis of many
testing for coupling of the electrode currents. Figure 9 represents parameters’ effects on the EAF model can also be included in
the current magnitude for the case when I1 = 10kA and I2 = I3 = future objectives. These parameters can include the effect of the
5kA. Having different desired currents when running a reactances on the system’s power factor and the effect of arcing
simulation leads to coupling between the electrodes. This as a function of temperature inside the furnace. Another
phenomenon is explained by the nonlinear terms found in the objective would be to see if the ideal PID-regulators for linear-
conductance matrix Gij . controlled electrodes would work in ‘bang-bang’ furnaces and if
not, find the relationship between the two types of furnaces.
Current vs Time with Delay and Noise
9000
8000
7000 8 References
6000
[1] B. Boulet, V. Vaculik, G. Wong, Control of High-Power
Current (Amps)
5000
4000
Non-Ferrous Smelting Furnaces, IEEE Canadian Review,
3000
summer 1997.
2000