Stray Flux and its Influence on Protection Relays

Authors: Z. Gajic, S. Holst, ABB AB, SA Products, Sweden, D. Bonmann, ABB AG, Transformers, Germany, and D. Baars, ELEQ bv. Netherlands

The bar-primary (i.e. ring-core; window type) current transformers are typically designed assuming that the flux in the CT magnetic core is homogeneous and only caused by the current flowing in the CT primary conductor. Thus, this means that:

  • The primary CT conductor is ideally centered in the middle of the CT toroidal magnetic core
  • The primary CT conductor is straight and infinitely long; and
  • There are not external magnetic fields which can cause additional flux in any part of the CT core

However, in practice the primary conductor is never straight and infinitely long and the CTs are commonly installed in a three-phase system. Thus, at least the magnetic fields from the other two phases are present in the vicinity of the CT. These “external magnetic fields” may under certain circumstances produce significant stray flux in the CT magnetic core, which can cause problems for protection systems connected to that CT.

As shown in Figure 1, the stray flux will split into two paths through the CT core. Thus, at one side of the CT core the resultant flux will be equal to the sum of the “usual flux” caused by the CT primary current and the stray flux, while at the other side of the CT core the resultant flux will be equal to the difference between the “usual flux” and the stray flux. Obviously the resultant flux will have different values in different parts of the CT core and a partial CT saturation may occur.

There are quite a number of papers published regarding CT accuracy under such operating conditions. Surprisingly very few papers discuss the influence of the stray flux on the relay protection systems. Even in some of the above mentioned papers it is stated that stray flux should not produce big impact on the relay protection. This might be true for the relays with time delayed operation such as phase or ground overcurrent relays. However, stray flux can easily cause unwanted operation of the instantaneous and sensitive relays like differential protection. Note that both high impedance and low impedance differential protection relays can be affected by this phenomenon.

Testing in the Laboratory
The laboratory testing was performed on the two CT cores designated CT #1 and CT #2 as shown in Figure 2.  Both CT cores have the ratio 800/1A with a relative small core cross section. The only important difference between the two CT cores is rating data and consequently the magnetic core cross section area. These two CT cores had the following ratings:

  • CT #1: 800/1A; 5P30; 10VA with  core cross section area of 17.1cm2
  • CT #2: 800/1; 5P5; 10VA with core cross section area of 1.9cm2

As shown in Figure 3, the stray flux influence is tested by positioning the CT core close to an adjacent primary conductor. Figure 3a was taken by digital camera during laboratory testing, while Figure 3b represents the simplified geometrical view of the test setup. The distance X is 6cm during these tests.  The test was done by applying the 50Hz, AC current with the RMS value of 6.5kA. The primary current could be injected with or without a DC offset. The applied current through the primary conductor and the CT secondary current were recorded by an oscilloscope as Channel 1 and Channel 2 respectively. These two waveforms are given in Figure 4. In the Figure 4a, the two waveforms are given when the DC offset is present in the primary current. The CT secondary current with maximum peak of 1.5A was recorded during this test. In Figure 4b, the two waveforms are given for the symmetrical primary current with AC RMS magnitude of 6.5kA. The recorded peak of the CT secondary current during this test reaches 0.2A. Note that secondary current spikes are only observed during testing of CT #2 and not during testing of CT #1. Consequently it can be concluded that CT core with bigger cross section area of the core has potentially less problems in the actual installation.

Field Recordings

The authors have observed and recorded this phenomenon in the field mostly in installations of phase shifting transformers (PST), power transformers and shunt reactors. As described in reference differential protection schemes for such equipment often require buried CTs within the tank.

First Field Recording:  Within a symmetrical, single-core PST having rating data 450MVA, 138/138kV, ±58o, 60Hz, six buried CTs, with ratio 3000/5 and class C800, are installed. Two CTs, one at each side of every phase of the delta winding are used for the low impedance differential protection scheme, as shown in Figure 5a.
The two recorded currents should have the same waveforms with opposite polarity in phase A (e.g. their sum shall be zero). However, it is clear that one of the two currents (i.e. red trace in Figure 6) is distorted for a part of the power system cycle. Actually, its peak value is almost completely reversed. Later it was concluded that this waveform distortion was caused by stray flux from the primary current in the neighboring phase (see Figure 5b).

Second Field Recording: Within a symmetrical, dual-core PST having rating data 600MVA, 232/232kV, ±35o, 50Hz, buried CTs (with rated data 1200/1A; 10P20; 60VA) are installed. CTs are located at the neutral point of the PST primary exciter winding in each phase, in accordance with primary differential protection scheme recommended by IEEE Guide which is shown on page 38 1a.  Due to space limitations within the tank, neutral point CTs used for this differential protection are located next to the yoke of the magnetic core of the excitation transformer (see page 38 1b). Unwanted operation of the differential protection has spuriously occurred during PST energizing.

The current waveforms in phase L1, captured by the built-in disturbance recorder within the differential relay, during one PST inrush case, are shown on page 38 2a. Because the Load-side circuit breaker was open during PST energizing, the two recorded currents should have the same waveforms with opposite polarity (e.g. their sum shall be zero). However, it is clear that the neutral point current is distorted for a part of the power system cycle. Actually, its peak value is reversed. Later it was concluded that this waveform distortion was caused by stray flux coming from the PST excitation transformer core.

Several unwanted operations of the differential relay were recorded. During another inrush case the distorted CT current reached more than 25A secondary in peak current value in phase L3, as shown on page 38 2b. This large current caused unwanted operation even of the unrestrained differential protection element.
Finite element magnetic field calculations were performed for this PST in order to check the stray magnetic field during inrush and during external fault. During external faults there is ampere-turn balance between the excitation transformer primary and secondary windings and the stray flux is essentially contained in the main ducts between the two windings. During inrush the secondary winding is open circuited and the magnetic field caused by the inrush current spreads throughout the entire transformer tank. From these calculations it can be concluded that a much greater stray flux is present at the CT location during inrush condition than during an external fault. Suchtheoretical calculations were confirmed by primary full scale testing. Different types of external faults were applied on the PST Load-side and stability of the differential protection was verified.

Third Field Recording:  Within an auto-transformer having rating data 500MVA, 400/132kV, 50Hz a common neutral point CT (with ratio 2400/1A) is installed within the tank. Several unwanted operations of the low impedance restricted earth-fault (REF) relay were recorded. Figure 7 shows one of captured inrushes when low impedance REF relay operated. From the recording it is clear that auto-transformer neutral point current is distorted. Its peaks are much bigger then corresponding 3Io current at 400kV side. This neutral point current distortion is caused by the stray flux from the auto-transformer main magnetic core.

Fourth Field Recording: Within a SVC transformer having rating data 100MVA, 220/19kV, YNd11, 50Hz; LV side CTs (with ratio 6000/1A) are installed as bushing CTs. However due to space limitations they were physically installed just under the LV bushings within the transformer tank. Several unwanted operations of the high impedance bus differential protection relay for the 19kV SVC bus were recorded during SVC energizing (i.e. SVC transformer together with SVC filter branches on the LV side were energized) from the 220kV side. Figure 8 shows recorded transformer LV side currents while the Figure 9 shows recorded resultant false differential currents seen by the numerical, high impedance bus differential relay during one such incident. Note that the high impedance differential relay stabilizing resistor had value of 1500Ω and that the relay pickup was set at 130V.
Investigation has shown that these unwanted operations were caused by the stray flux from the power transformer core during inrush conditions.

Fifth Field Recording: Within a reactor having rating data 15MVA, 132kV, 50Hz three neutral point CTs (i.e. one per phase) with ratio 300/1A are installed within the tank. Several unwanted operations of the reactor’s low impedance differential protection relay were recorded. Figure 10 shows one of captured inrushes, for phase L1, when the low impedance differential relay has operated. The two recorded currents should have the same waveforms with opposite polarity (e.g. their sum shall be zero). From the recording it is clear that reactor neutral point current (blue trace in Figure 10) is distorted for some parts of the power system cycle. Its peaks are reversed and have a much bigger value than the corresponding L1 phase current at the reactor HV side. The neutral point current distortion is caused by stray flux from the reactor’s main magnetic core during inrush.

Sixth Field Recording: A 380kV substation having a one-and-a-half CB switchgear arrangement has been used. Three auto-transformers AT1, AT2 and AT3 having rating data 500MVA, 380/132kV, 60Hz are installed in order to interconnect 380kV and 132kV networks as shown in Figure 11.  Two 380kV busbars (i.e. BB1 and BB2 as shown in Figure 11) are protected with high impedance differential relays. The serial and common windings of each auto-transformer are also protected by phase-segregated, high-impedance differential protection. Consequently three, neutral point CTs have been installed within each auto-transformer tank in the neutral point (i.e. one per phase). All CTs involved in these five high impedance differential protection schemes have ratings of 3000/1A, Class X and knee point voltage of 1000V. All five high impedance differential relays have identical settings: series resistor of 1800Ω and set pickup value of 180V. Thus pickup current for all high impedance differential relays is 100mA on CT secondary side. It shall be noted that all five high impedance differential relays are numerical with built-in disturbance recording capability.

Phase L1 to ground fault occurred in a 380kV CB associated with the 380kV busbar one, as shown in Figure 11. The high impedance differential relay for BB1 cleared the fault instantaneously. Operation was recorded in phase L1 and verified by visual switchgear inspection, however at the same time the high impedance differential relays for AT1 and AT3 operated, as well, disconnecting the two auto-transformers from the network.
Figure 12 shows recorded differential currents and associated trip signals from the numerical high impedance differential relay for AT1 during this event. It shall be noted that high-impedance AT1 differential relay operated in phases L2 and L3, while the external fault was actually in phase L1. Recorded peak value of the instantaneous differential current in phase L2 and L3 were 435mA. Thus relay operated in accordance with its settings.

Investigation has shown that these unwanted operations of two auto-transformer high impedance differential relays were caused by the stray flux from the auto-transformer core during the external fault.

Stray Flux Influence

As shown in the previous sections, the consequences of the magnetic stray flux existence at the CT location appear as current pulses on the CT secondary side. Such pulses have varying magnitude and they are only present for a part of a power system cycle. These current pulses will be measured by the protection relay(s) connected to that CT. As already mentioned, their main influence will be the possible unwanted operation of the protection relays. Typically, the differential relays are mostly affected. Such unwanted relay operations often cause confusion and require special investigations in order to understand and rectify the problem.

Thus, the protective relaying community shall be aware of possible problems caused by stray flux, because unwanted operations of this nature can be quite costly and may require lengthy investigations.
Note that problems with stray flux can occur in applications where strong external magnetic fields are present at the CT location. Such installations are typically characterized by high phase current magnitudes and small distances between phases or sharp bends of the primary conductor close to the CT location.

Installations where this problem may occur more frequently are:

  • CTs within LV and MV metal-clad switchgear
  • CTs within HV GIS switchgear;
  • CTs at the generator HV or neutral point terminals;
  • CTs in the generator bus ducts
  • CTs buried within power transformer, phase shifting transformer or shunt reactor tanks

When such problems are encountered the following possible actions may be taken:

  • Increase relay pickup settings
  • Add intentional time delay for relay operation
  • Use second harmonic blocking for low-impedance differential protection
  • Use differential protection principle not requiring buried CTs
  • Change CT location (if possible)
  • Replace CT with a CT equipped with a flux equalizing winding

Which action shall be taken in a particular installation depends heavily on the extent of the problem and cost involved. Where second harmonic blocking is utilized, in order to stabilize the bus-type differential relays, it is advisable to check whether that decreases the dependability of the differential relay for internal faults. Alternatively, a differential relay capable of bypassing the second harmonic blocking criterion for internal faults, can be used.

CT with a Flux Equalizing Winding
Current transformers can be protected from the stray flux by shielding. The first solution was to utilize separate solid copper shields. Later on the use of a flux equalizing winding was suggested. When the flux equalizing winding is not combined with the usual secondary CT winding it can be represented as a separate winding divided for example into four segments, equally distributed around the CT core circumference as shown in Figure 14. Each winding segment is marked with a number in this figure which corresponds to the winding marking visible for CT #2 in Figure 2. Sometimes this winding might have a higher, but most often even number of segments. All segments should have an equal numbers of turns.

In the first solution, by cross-connecting the two segments located on diametrically opposite parts of the CT core (1&3 and 2&4 in Figure 14a), a path for flow of a circulating current between each pair of the flux equalizing winding segments, during stray flux condition, is achieved. These circulating currents will produce a magnetic field in the CT core with an opposite direction to the stray flux diminishing its influence on the CT. These circulating currents will not exist during normal operation when the stray flux is not present.

In the second solution, (Figure 14b), all flux equalizing winding segments are connected in parallel. This connection enables path for flow of circulating currents between all segments of the winding, during stray flux conditions. These circulating currents will produce a magnetic field in the CT core with an opposite direction to the stray flux diminishing its influence on the CT.  Note that these circulating currents will not exist during normal operation of the CT when the stray flux is not present.

On both CT cores (Figure 2,) an additional secondary flux equalizing winding is already wound. This flux equalizing winding is divided in four equal sections, evenly distributed around the core circumference. Its influence can be enabled or disabled by changing externally available connections of the four segment ends. Results from the laboratory testing were shown in Figure 4 with disabled flux equalizing winding influence.

When the laboratory testing of CT #2 was repeated with the flux compensation winding enabled by connecting its 4 segments in accordance with the second solution shown in Figure 14b, no secondary current spikes were observed when current was injected through the primary conductor adjacent to the CT core (see Figure 3).  The flux equalizing winding can be either a standalone winding or combined with the main CT secondary winding (Figure 13.)
To mitigate the problem of unwanted operation of protection relays for the new installations where there is a risk of the stray flux, it may be advisable to always order and use specially made CTs having flux equalizing winding. This relatively small investment can be worthwhile.  


Zoran Gajic is Technical Marketing Manager for Nordic Countries at ABB AB, Substation Automation Products in Vasteras, Sweden. He received his MSEE degree with honors from the University of Belgrade, Serbia in 1990 and PhD degree in electrical engineering from Lund University, Sweden in 2008. Since 1993 he has had various engineering positions within ABB Group of companies. From 2008 to 2014 he was ABB Global Product Manager for Generator and Transformer Protection. He has published many technical papers and holds more than fifteen patents. Zoran has participated in different CIGRE, IEC and PSRC/IEEE WGs and was the Convenor for CIGRE Working Group “Modern techniques for Protecting Busbars in HV Networks”. In 2014 he received Technical Committee Award from Cigré Study Committee B5 (Protection and Automation). 

Douwe Baars, BSc is Senior Application Engineer in the product development center of ELEQ Steenwijk b.v., the Netherlands. He is a specialist regarding current transformer design for applications in HV and LV installations. He is a member of several WGs (MT48, WG49, WG47) of IEC Technical Committee 38, Instrument Transformers.

Stig Holst is Senior Specialist at ABB AB in Vasteras, Sweden. He has a MSEE degree from Chalmers University of Technology, Gothenburg, Sweden. He joined ABB in 1996 and his main working area has been application of protection relays. He is also a specialist in transient performance of CTs and CT requirements. Between 1973 and 1996 he held different leading positions at the Swedish utility Sydkraft (now a part of E.ON). He has also been project manager for international consultant services. Stig has participated in different international WGs and was the convenor of a CIGRE WG, coordinating protection relays and conventional CTs. He has published several technical papers. Stig is a member of Cigré and was the Swedish member of SC B5 (Protection and Automation). He is also a member of IEC TC95/MT4 and received the IEC 1906 Award in 2015 which honors the IEC experts in recognizing their exceptional recent achievements and contributions to the IEC committees.


Power. Flexible. Easergy.
Let?s start with organization in protection testing