Metal Oxide Varistor Model With Hysteresis Loop For ATP/EMTP
Metal Oxide Varistor Model With Hysteresis Loop For ATP/EMTP
Manuscript received January 25, 2004. The author is with the Technical
University of Bialystok, Bialystok, Wiejska 45D, Poland (e-mail:
[email protected]).
2
b)
Measurements were made by digital oscilloscope and
save to spreadsheet file format. Digital measurement
technique allows decreasing measurement error to half
AC/DC quantization step. 12bit AC/DC converter in digital
oscilloscope allows gathering enough points to draw V-I
characteristic. Normally ATP/EMTP allows only 16 points in
V-I characteristic. When we have got much more of it, why
we shouldn’t used it to decrease simulation error. There is no
necessary to make any assumptions to approximate MOAs
parameters in new model presented below. In new model only
8 point were used to clearly present idea of it. It can be
increased to 32768 points (limited by MODELS) [2].
There is no necessary to use any inductances or
capacitances in new model and most important, parameter
c) approximation disappears. Necessary phase shift ensure non-
linear V-I characteristic and hysteresis loop.
Main improvement in new algorithm is two calculations
ways. First, when MOA voltage derivative is greater than zero
and the second one, when derivative is less then zero. Off
course all calculations are made for discrete time values.
When algorithm splits up voltage range in V-I array is
calculated. Defined array at the beginning have got limited
number of points. Linear approximation is used to calculate
rest of it. In last step of algorithm resistance of MOA is
calculated. In this point oscillations can appear because step
MOA resistance change. Small trick is used in algorithm
which is not clearly visible. There is no direct resistance
determine from V-I array. And there is no use of previous
value of resistance. If one of it will be used, oscillations will
d) appear. Resistance calculation based at vector deviation
change. Angle of resistance change is calculated. This angle
also makes linear approximation rest of V-I points.
Major disadvantage of new model are measurements.
Digital oscilloscope is necessary. Major advantage of new
model is possibility to use it with any non-linear element. Of
course surge arrester characteristic measurements have got
specific resolution - error. It is equal half of AC/DC converter
quantization step. These kinds of errors are commonly
accepted.
MOA and new model are small but must be taken into
consideration. These differences reduce a little bit surge
arrester voltage protection level. In special types of calculation
it may be crucial for computation results. Major differences
occur when computations of complex building structure are
made.
Fig. 3.5 Surge arrester V-I characteristic - new model with hysteresis loop.
Fig. 3.6 Surge arrester current in time domain - new model with hysteresis
loop (red line), MOV (blue line).
Fig. 3.7 Surge arrester voltage in time domain - new model with hysteresis
Fig. 3.3 Circuit diagram with standard MOV element. loop (red line), MOV (blue line).
4
Fig. 3.8 Surge arrester new model resistance change in time domain.
IV. CONCLUSIONS
Presented model more precisely approximates surge arrester
characteristic. Real MOA V-I characteristic can be directly
transferred from digital measurement to computer simulation
program. There is no need to approximate any MOA analog
parameters. New model can be also use for any non-linear
element with hysteresis loop in V-I characteristic or not too.
Tricky algorithm eliminates oscillation during step parameter
change. It’s easy to improve in future by simple ATP/EMTP
user.
V. REFERENCES