Professional Documents
Culture Documents
Session 2000
Session 2000
Belgium
Application of Thevenin’s Theorem. In the course of Assessment of actual operating conditions. The above
the ongoing discussion, there will be a need for an analysis brings about two advantages : at first, to get rid
indisputable reference to the Thévenin’s Theorem. An of the hypothesis of the unloaded system ; second, to be
English reading thereof is taken from a world-wide able to make further calculations following the
accessible reference book : « With respect to any single theoretical definition of physical short-circuit power,
external circuit connected to any given pair of terminals based on the actual voltage contrarily to the IEC voltage
of a network, the network can be replaced by a single rating.
branch having an impedance, Z, equal to the impedance
measured at these terminals looking into the network
(when all the network emfs are made equal to zero) and
containing a single emf, Eo, equal to the open-circuit Iload
voltage of the network across the given pair of
Un/µ ~ ZSTo
Un
terminals » [2]. a
Effective short-circuit power in operating conditions. A numerical example follows, for a 120 MVA DC arc
The above analysis leads to propose the concept of ef- furnace installation, connected at 220 kV through a 150
fective short-circuit power in operating conditions, defi- MVA transformer. The numerical parameters are:
ning it as the physical short-circuit power on the loaded Apparent 220 kV short-circuit level : 3300 MVA (at 220
network and at the actual voltage at the substation. kV)
Step-down transformer : 239,8 kV / 33 kV, 150 MVA,
The approach yields information to be used ucc = 15 %
complementarily to the IEC Standard in special Zsc
discussions and measurements. That leads to = 0,2
Zsc + ZAB
3. SOME EXPERIMENTAL FINDINGS while measurements give an average of
Pst ( B ) P ( HV )
= st = 0,17
3.1. Simultaneous flicker measurements at both sides Pst ( A ) Pst ( MV )
of the transformer feeding an important
fluctuating load These values correspond to the slopes of the straight
lines in Figure 3.
Flicker measurements are systematically carried out The average experimental Pst ratio is 15 % lower than
when big industrial fluctuating loads are to be connected the theoretically expected one. It corresponds to a
to power grids [3]. In this context, it is necessary to rely virtual short-circuit level equal to 4070 MVA at 220 kV,
on suitable methods to assess the individual contribution instead of the theoretical 3300 MVA.
of a particular load, when multiple flicker sources are
operating simultaneously in the electrical vicinity. 2.5
0.5
0
0 2 4 6 8 10 12
Pst ( MV )
Figure 2 : Typical configuration for the assessment of Figure 3 : Simultaneous flicker measurement at the MV and
flicker emission level of consumer A HV sides of the transformer feeding an electrical arc furnace,
when the other disturbing loads are at still stand
The typical configuration of Figure 2 is considered, (upper line: ratio 0,2 - lower line: ratio 0,17)
where A and B are respectively the MV and HV sides,
ZAB being the transformer impedance.
3.2. Capacitor bank switching on and off: a classical where Xsc is the reactive part of Zsc, the short-circuit
way of assessing short-circuit power impedance in B.
The preceding equation can then be transformed into
Switching on and off capacitor banks produce voltage
changes in the network, which are often used for the ∆U
=
( )
U 2 I 2 − I1
≅
∆Q
(3)
assessment of the short-circuit power. However, it will 2
U2 U2 Ssc
be shown that the estimated value may be significantly
greater than theoretically expected. This is consistent X sc
with the results described in 3.1.
and the reactive part of the short-circuit power 2, Ssc, is
Figure 4 shows a typical example of the time evolution obtained by dividing the reactive power variation by the
of the r.m.s. voltage, recorded during successive 33 kV relative voltage variation.
capacitor bank switching operations. The tests were
carried out at the location of the above mentioned arc Repeating the test several times and averaging the
furnace installation, this one being out of service. results led to Ssc = 3900 MVA at 220 kV. This value is
consistent with the one deduced from the flicker
239
measurements given in 3.1, which is obviously greater
238
than the theoretical value (3300 MVA).
237
4. THE INFLUENCE OF THE LOADS
236
∆Qload ∆U 35
=2 (7)
Qload U 30
25
Alpha=0,5
that 0
-5
∆Qload ∆U
=α (8) -10
Qload U
0 1 2 3
Q load ( in % of Ssc )
X*sc S*sc
P st contract ( HV ) = P st emiss *
X tfo Ssc contract
The slope of this line is equal to k-1 and if the network In first approximation, we suggest to neglect the terms
impedances were perfectly voltage independent, it containing Qload and α:
should theoretically be equal to the ratio Xsc / Xtfo(where
Xsc is the short-circuit impedance of the HV network X*sc Ssc
and Xtfo, the transformer reactance). In fact, it may be P st contract ( HV ) = P st emiss (19)
significantly different, due to the voltage dependence of X tfo Ssc contract
the loads in the network. As described above, the
influence of the loads results in a virtual internal For practical purpose, equation (19) is generally
impedance of the network X*sc , different from the sufficient because of the uncertainty concerning the
parameters and also because the numerical difference
standardised short-circuit impedance. (the virtual
with equation (15) is negligible.
internal impedance X*sc is lower than Xsc when the
reactive power coefficient of the loads α is greater than 5.4. Difficult cases
2; it is greater than Xsc when α < 2).
Situations may be found where the correlation between
Knowing this, the flicker emission level at HV is Pst emiss and Pst (MV) is not as good as pictured in Figure
deduced from the emission level resulting from 6, leading to uncertainty in the determination of the
Sakulin’s method by: coefficient k and, consequently, the identification of
X*sc.
X*sc
P st ( HV ) = P st emiss ( 16 )
X tfo We experienced lately such a difficult case, see Figure 8
and Figure 9.
where X*sc is the empirical value of the virtual network
reactance, identified from Figure 7 (X*sc/Xtfo = k-1) or -
more easily - from Figure 6 (where k is given by the
slope of the least square straight line). This approach
relies upon the assumption that this impedance is not too
much varying (it is in fact depending on the loads) and
that it makes sense to use an average value.