Introduction to differential protection

Transformers are important system components available in many different

constructions. The range of HV transformers reaches from small distribution
transformers (from 100 kVA) up to large transformers having several hundred
MVA.

The essential physics behind transformer differential protection you should

know (photo credit: Kam Abbott via Flickr)
Apart from the large number of simple two and three-winding transformers, a
range of complex constructions in the form of multi-winding and regulating
trans- formers also exist.

Differential protection provides fast and selective short-circuit protection on

its own, or as a supplement to Buchholz (gas pressure) protection.
It is usually applied on transformers above approx. 1 MVA. On larger units
above approx. 5 MVA it is standard.

The transformer differential protection contains a number of supplementary

functions (adaptation to transformation ratio and vector group, stabilisation
against in-rush and over-excitation) and therefore requires some fundamental
consideration for the configuration and setting calculation.

Essential physics
To better understand the protection response during short-circuits and switching
operations, the physical principles of the transformer are initially covered in
detail.

1. Equivalent circuit of a transformer

2. In-rush
3. Sympathetic in-rush
4. In-rush blocking
5. Cross-blocking
6. Transformer over-fluxing
<="" ins="" style="box-sizing: border-box; height: 280px; width: 336px !
important; display: block;">

1. Equivalent circuit of a transformer

The primary and secondary winding are linked via a magnetic core by means of
the main flux Φ Figure 1. To obtain the flux, the magnetising current
(excitation current) I m according to the magnetising curve is required.
Figure 1 – Equivalent circuit of a transformer
In the electrical equivalent circuit, this excitation requirement corresponds to
the main reactance Xm. The leakage flux Φσ1 and Φσ2 are only linked to their
respective own windings and make up the leakage reactances Xσ1 and Xσ2 .

R1 and R’2 are the respective winding resistances. All currents and impedances
are referred to the primary side.

Xm = U/Im corresponds to the slope of the magnetising curve. During load and
particularly in the event of short-circuits, the operating point is below the knee-
point in the steep portion of the curve.

The magnetising current at nominal voltage only amounts to approx. 0.2% In,
i.e. in the non-saturated segment of the curve X m is approx. 500 times larger
than the nominal impedance of the transformer and approx. 5000 times greater
than the leakage reactances.
During load and short-circuit conditions, a simplified equivalent circuit may
therefore be used for the calculations (Figure 2).
Figure 2 – Simplified
transformer equivalent circuit
The series reactance XT corresponds to the short-circuit voltage in %, relative
to the nominal impedance of the transformer:

The series resistance corresponds to the ohmic short-circuit voltage in %,

and is also based on the nominal impedance. For calculation of the short-
circuit current, the resistance may be neglected, it must only be considered
when calculating the DC time constant.

Table 1 lists typical transformer data.

Table 1 – Typical transformer data

Rating Transformation ratio Short-circuit Open circuit current
[MVA] [kV/kV] voltage ux-t [%] [% IN]
600 400/230 19 0.25
300 230/110 24 0.1
40 110/10 17 0.1
16 30/10 8.0 0.2
6.3 30/10 7.5 0.2
0.63 10/0.4 4.0 0.15
Go back to contents ↑
2. In-rush
When energizing a transformer, one-sided over-excitation results, due to
remanance causing large magnetising current flow (in-rush current).

The flux does not return to zero when the transformer is switched off, but
remains at the remanance point ΦRem, which may be above 80% of the
nominal induction. When the transformer is re-energized, the flux increase
starts at this point. Depending on the energizing instant on the sinusoidal
voltage (point on wave), an off-set course of the flux can result.

For the large flux values in the saturation range, a very large magnetising
current is required, and cyclic current peaks will result.
The curve form corresponds with the sinusoidal half-waves of a simple half-
wave rectified AC current that decays with a very large time constant (Figure 3
below).
Figure 3 – Origin of in-rush current
The rush current is particularly large when cores of cold rolled steel with
a nominal induction (1.6 to 1.8 Tesla) are operated close to the saturation
induction (approx. 2 Tesla).

On a three-phase transformer, a three-phase rush current will result, which

depends on the vector group and the method of star-point earthing on the
transformer.
In general, two phases will saturate and draw large magnetising currents. On star delta

transformers, these currents are coupled to the non-saturated phase via the delta winding.

This causes the typical rush currents as shown in Figure 5.

The rush currents in the three phases can be calculated from the required
magnetisation (ImA and ImC) of the two saturated core-limbs A and C with the
given equations. The current on phase B thereby corresponds to the current in
the delta winding ID. The shown oscillogram of an in-rush occurrence confirms
the calculated curves.

Amplitude and time constant depend on the transformer size (see figure 4).
Figure 4 – Typical rush current of a star delta transformer

It must be noted that a similar rush current also arises when a close-in external short–circuit

is switched off and the transformer is re-magnetised by the recovery of the voltage .

It however is substantially smaller than the in-rush following ener- gising of a

switched-off transformer.
Figure 5 – Typical in-rush current data
Large rush currents can also occur when asynchronous systems are switched
together via a transformer, as the large voltage difference can cause transient
saturation of the core.

Go back to contents ↑

<="" ins="" style="box-sizing: border-box; height: 280px; width: 336px !

important; display: block;">

3. Sympathetic in-rush
When transformers were connected in parallel, it was observed that the
differential protection of the transformer that was in service issued a trip.

The reason for this is sympathetic in-rush current, which results from the rush
current of the transformer that is being energized (Figure 6).
Figure 6 – Sympathetic inrush current
The voltage drop resulting from the initial rush current across the source
resistance of the in-feed affects the second transformer in parallel and causes
the sympathetic in-rush current (I 2).

The current from the system (IT) decays rapidly. However a current still
circulates between the two transformers due to the small damping (large time
constant τ = X/R of the windings).

Go back to contents ↑

4. In-rush blocking
The in-rush current flows into the protected object from a single side and
appears as an internal fault. The transformer differential protection must
therefore be stabilized against this phenomenon.

The large amount of second harmonic in the rush current was already used with
conventional protection for this purpose. The second harmonic is filtered out of
the differential current (operating current) by means of a filter, and is then used
as additional restraint current in the measuring bridge.

When it was above approximately 15% in relation to the 50(60) Hz

fundamental, a very large additional restraint was introduced to prevent
tripping.
Other manufacturers compared the 100(120) and 50(60) Hz components
directly with a separate bridge circuit, which then directly blocked the protection,
as it is now done in the software of numerical protection.

The 100(120) Hz component in the rush current depends on the base width of
the sinusoidal caps (as shown in Figure 7).

It decreases as the base width B increases.

Investigations have shown that a base width greater than 240° hardly ever
arises in practice, which implies a minimum second harmonic component of
17.5%. A setting of 15% therefore makes sense for in-rush blocking.
The 3rd harmonic may not be used for in-rush blocking, as it is strongly represented in the

short-circuit current when CT saturation takes place.

Figure 7 – Harmonic
content of in-rush current
A more sensitive setting than 15% of the second harmonic should normally not
be applied as the off-set short-circuit current will also have a second harmonic
component in case of CT saturation.

In rare cases, for example with weak in-feed, a soft energization with a very
small second harmonic component may occur. Under these conditions, a
reduced setting of e.g. 12% may be considered. Preference should however
be given to the cross-blocking function.
Go back to contents ↑

5. Cross-blocking
This function, which was already applied in conventional relays, is now
available in all numerical relays and may be activated, if desired.
It takes into account that the second harmonic component in the individual
phases is different and may not be sufficient, in the phase with the smallest
component, to activate the blocking.

The measuring system in all phases is therefore blocked when a single phase
detects the rush blocking condition.
Go back to contents ↑

6. Transformer over-fluxing
If the transformer is operated with excessively high voltage , then the required
magnetization is also increased. The magnetizing current rises sharply when
the operating point on the magnetizing curve is close to the point of saturation.

The wave form becomes more and more distorted with increasing odd harmonic content (as

shown in Figure 8).

Figure 8 – Magnetising current in the event of transformer over-fluxing
The increased magnetising current appears as a tripping current in the
differential protection with large overvoltage. This can cause tripping, depending
on the con- figuration of the transformer.

Overvoltages can occur in the system, due to the distribution of reactive power
flow in the event of tap changer problems, or following load shedding. This is
particularly true for geographically large systems with long lines.
One critical case is the switching off of a power station under full load
conditions which results in a severe overvoltage condition at the unit
transformer as a result of the large excitation of the generator.
The transformer can tolerate the over-excitation, which causes heating, for
a given time without sustaining damage. During this time the system regulation
must ensure that the voltage returns to the permissible range.

Only if this does not occur, must the transformer be isolated by a special over-
excitation protection having a U/f-dependent time delay. Tripping by the
differential protection with a fast measurement due to these conditions must be
avoided at all cost.

Modern numerical relays therefore provide an integrated blocking of the trip

in the event of over-excitation (over-fluxing). It is based on the large
5th harmonic component in the tripping current which clearly indicates over-
fluxing. Tripping is blocked when the ratio I150Hz/I50Hz exceeds a set value.
For this setting it must be noted that the 5th harmonic component again
decreases if the over-voltage is very large (Figure 8). A typical setting
is 30%.
If the over-voltage is very large, blocking is no longer sensible as the
transformer is at risk. The blocking can therefore again be re-set when the 5th
harmonic component is above a set ratio of the 50 Hz component, which
increases as the over-voltage increases.