UNIVERSITY OF CAPETOWN DEPARTMENT OF ELECTRICAL ENGINEERING TRANSFORMER RESPONSES TO GEOMAGNETICALLY INDUCED CURRENTS RESEARCH BY TALENT TAFADZWA MURWIRA MASTERS IN ENGINEERING BY DISSERTATION AT UNIVERSITY OF CAPETOWN SUPERVISOR

UNIVERSITY OF CAPETOWN

DEPARTMENT OF ELECTRICAL ENGINEERING
TRANSFORMER RESPONSES TO GEOMAGNETICALLY INDUCED CURRENTS
RESEARCH BY
TALENT TAFADZWA MURWIRA
MASTERS IN ENGINEERING BY DISSERTATION
AT
UNIVERSITY OF CAPETOWN
SUPERVISOR: DR. DAVID OYEDOKUN
CO-SUPERVISOR: PROF. KOMLA FOLLY
ABSTRACT
Geomagnetically induced currents (GIC) are quasi-dc currents that result from extra-terrestrial activities. Coronal mass ejections from the Sun’s surface are ejected from coronal holes and they travel towards the earth through space. The magnetosphere and ionosphere are regions above the Earth whose magnetic field is distorted by coronal mass ejections. The distortions in these regions results in variation of potential on the earth’s surface and distortions in the earth’s magnetic field. As a result, quasi-dc currents normally referred to as “geomagnetic induced currents” flow on the earth’s surface and some induced in transmission lines. Grounded neutrals of transformers are the entry points of GIC into the power system.

This project investigates the effects of GIC to three-phase five-limb transformers (3p-5L). Practical tests on 230/120V, 300VA, 3p-5L limb transformers have been carried out and the results are presented in this report. The results show fluctuating harmonics, increased surface temperature, increased reactive power and drop in load voltage as dc injected through the neutral of the transformer is increased. Further tests on the general power theory proved that there is an underestimation of reactive power when measured conventionally as opposed to the general power theory. Conventional methods of calculating power do not take into account the distortions that occur as a result dc bias and unbalanced system voltages.

The project is part of an ongoing research with the aim of estimating the thresholds of GIC initiating damage in power transformers and improving reactive power consumed by transformers conducting GIC. These preliminary studies show that the impact of GIC is dependent on the structure of the transformer. Rigorous testing is yet to be conducted on relatively larger transformers (15kVA, 400/400V, 3p-5L) that closely resembles the response of large power transformers to geomagnetically induced currents. Finite Element Modelling (FEM) is going to be used on the ANSYS Maxwell package to validate experimental results.

TABLE OF CONTENTS
CHAPTER 1: INTRODUCTION
INTRODUCTION
BACKGROUND
OBJECTIVES
HYPOTHESIS
RESEARCH QUESTIONS
CHAPTER 2: LITERATURE REVIEW
2.1 INTRODUCTION
2.2 HISTORICAL EVENTS
2.2.1 THE CARRINGTON EVENT (1859)
2.2.2 THE HYDRO-QUEBEC EVENT (1989)
2.2.3 THE HALLOWEEN STORM (2003)
2.3 GIC EVENTS IN LOW LATITUDES – INCLUDING AFRICA
2.3.1 GIC IMPACT ON REACTORS
2.4 SUSCEPTIBILITY TO GIC AMONG DIFFERENT STRUCTURES
2.5 TRANSFORMER RESPONSES
2.5.1 HALF-CYCLE SATURATION
2.5.2 THERMAL EFFECTS ON TRANSFORMER
2.5.3 HARMONICS GENERATION
2.5.4 REACTIVE POWER DEMAND
2.5.5 INCREASE IN TRANSFORMER NOISE
2.6 CONCLUSION
CHAPTER 3: LABORATORY TESTS AND RESULTS
3.1 INTRODUCTION
3.2 TESTS PERFOMED
3.3 LABORATORY SETUP
3.4 TEST EQUIPMENT
3.5 TRANSFORMER UNDER TEST SATURATION CHARACTERISTICS
3.6 REACTIVE POWER
3.7 SEARCH COIL VOLTAGE OR LEAKAGE FLUX MEASUREMENT
3.8 ANALYSIS OF HARMONICS
3.9 SURFACE TEMPERATURE MEASURENT
3.10 CONCLUSION

CHAPTER 4: REACTIVE POWER UNDER GEOMAGNETIC DISTURBANCE
4.1 INTRODUCTION
4.2 CONVENTIONAL METHOD OF CALCULATING POWER
4.2.1 REACTIVE POWER
4.2.2 ACTIVE POWER
4.2.3 COMPLEX POWER
4.2.4 POWER TRIANGLE
4.3 THREE PHASE POWER
4.4 DISTORTION AND UNBALANCE IN THREE PHASE SYSTEMS
4.5 GENERAL POWER THEORY (GPT)
4.6 REACTIVE POWER CONSUMPTION OF 3p-5L USING GPT
4.6.1 POWER FACTOR COMPARISON
4.7 CONCLUSION
CHAPTER 5: THRESHOLDS OF GIC INITIATING DAMAGE IN TRANSFORMERS
5.1 INTRODUCTION
5.2 DEFINITION OF DAMAGE
5.3 STANDARDS
5.4 CURRENT THRESHOLD STUDIES
5.5 TEMPERATURE THESHOLDS
5.5.1 HOTSPOTS
5.5.2 OIL AND TANK TEMPERATURE
5.5.3 WINDING HOTSPOTS
5.5.4 THERMAL EFFECTS ON FLITCH PLATE
5.6 REACTIVE POWER CONSUMPTION AND VOLTAGE VARIATIONS
5.7 HARMONICS
5.8 SATURATION
5.9 LOSSES
5.9.1 CORE LOSSES
5.10 CONCLUSION
LIST OF FIGURES
Figure 1.1: The space weather or GIC chain
Figure 1.2: Reconnection of the magnetosphere due to interference of CMEs
Figure 2.1: Transformer core types and GIC susceptibility variation between transformer types.

Figure 2.2: Effects of GIC on entire power system.

Figure 2.3: Excitation current of a transformer as a result of dc bias.

Figure 2.4: Observed Meadow Brook transformer hotspot temperature for a minor storm on May 10, 1992.

Figure 2.5: 14 500 kV simple demonstration circuit simulation results-transformer ac currents and distortion due to GIC.500 kV transformer ac current-normal and GIC-distorted.

Figure 2.6: Transformer increased reactive power demands (MVARs) due to GIC for a typical 500 kV transformer for single phase and three-phase three-legged core type.

Figure 2.7: Transformer magnetizing current characteristics for normal operation and for half-cycle saturation due to the presence of GIC.

Figure 3.1: Laboratory setup for testing transformer response to geomagnetically induced currents
Figure 3.2: Saturation characteristics of the 3p-5L transformer under test
Figure 3.3: Reactive power against injected dc trend within a 3p-5L transformer.

Figure 3.4: Leakage flux measurements on a 3p-5L transformer using search coils
Figure 3.5: Total harmonic distortion of 3p-5L transformer as measured by Yokogawa power meter
Figure 3.6: Increase in temperature with increasing dc current
Figure 3.7: Load voltage profile of 3p-5L transformer under dc bias
Figure 3.8: The voltage collapse and over-voltages as observed at the Jacques-Cartier substation..Figure 4.1: The power triangle
Figure 4.2: Line-to-line and line-to-ground voltages
Figure 4.3: Complete power triangle, in which Q = total non-active power, where Qa is the component that can be compensated without energy storage and QA the component that requires energy storage for compensation. S = apparent power without any compensation, Sa the apparent power after compensation without energy storage, and SA the apparent power after complete compensation so that SA=PFigure 4.4: m-wire system with resistances r1,r2,r3,…,rm, supplying load with voltages
e=e1,e2,e3,…,em, and currents, i=i1,i2,i3,…,im, with active supply current ia and local compensator current ic.

Figure 4.5: Reactive power comparison of 3p-5L, 300VA when measured conventionally (blue graph) and according to General Power Theory (brown graph).

Figure 4.6: Power factor comparison of 3p-5L, 300VA when measured conventionally (blue graph) and according to General Power Theory (brown graph).

Figure 5.1: Metallic hot spot temperature for a GIC wave shape derived from the March 1989 GMD event.

Figure 5.2: Observed Meadow Brook transformer hot-spot temperature for a minor storm on May 10, 1992.

Figure 5.3: GIC variation and calculated winding temperature from 1989 GMD event in Canada
Figure 5.4: Winding temperature rise 750 MVA, 765/345/35.5 KV, 1-phase, auto-transformer
Figure 5.5: Hotspot temperature on windings, tie plates and clamping plates.

Figure 5.6: Step thermal response of the Flitch plate of a 400 kV 400 MVA five-leg core-type transformer to a 10 A per phase dc step.

Figure 5.7: Asymptotic thermal response of the Flitch plate of a 400 kV 400 MVA five-leg core-type transformer.

Figure 5.8: Fundamental %VAR drawn by a 750MVA single phase autotransformer against GIC, A/Phase
Figure 5.9: Voltage total harmonic distortion (VTHD) as a function of GIC for GSU transformers operating at 500kV and above
Figure 5.10: The GIC causing transformer saturation is evident from the 6th harmonic measurement.

LIST OF TABLES
Table 2.1: Transformers and reactors that failed in South Africa (1989-1994)
Table 2.2: Substations most susceptible to GIC in South Africa
Table 2.3: Increase in core loss and core noise under varying GIC levels.

Table 3.1: Experimental results showing dc injected against reactive power consumed measured conventionally.

Table 3.2: Voltage distortion limits in IEEE Std. 519
Table 5.1: Maximum allowable temperatures on transformers
Table 5.2: Percentage Voltage THD, for low voltage transformers.

Table 5.3: Maximum temperatures on single phase transformer tested by NERC
CHAPTER 1: INTRODUCTION
1.1 INTRODUCTION
This chapter gives a brief overview of how geomagnetic currents arise, the processes that occur from Sun to earth and how they affect the power system. The propagation of coronal mass ejections through space affects the magnetic field of the earth, giving rise to variations in potential on the Earth’s surface. This potential difference results in geomagnetically induced currents (GIC) flowing and affecting transformers with grounded neutrals.

1.2 BACKGROUND
Geomagnetic induced currents (GIC) are quasi-dc currents with frequencies ranging from 0.001~0.1Hz 1 that flow through the transformer neutrals into the power system network. The peak values could be as high as 200A, lasting for several seconds to hours. The peak value ever calculated in Southern Africa is 108A, at Alpha substation in South Africa 2. In brief, geomagnetic induced currents are a result of solar storms. The Sun goes through 11-year solar cycles, with solar activity increasing towards the end of each cycle 3. The last solar cycle ended in the years 2008-2009 and it was the 23rd solar cycle since the first recorded cycle in 1755 4. Although the GIC activity peaks towards the end of the cycle, they are not limited to occurring at peak times only.

The chain of events that lead to geomagnetic disturbances begins from the Sun’s activities and ends when geomagnetic induced currents interfere with technological systems such as telecommunications and power systems at the Earth’s surface as shown in Figure 1.1.

Activity of the sun
Propagation of solar wind
GIC in technological systems
Geoelectric field at Earth’s surface
Ionosphere processes (auroral & electrojets release)
Earth’s surface
Magnetosphere processes (magnetic reconnection)
Network configuration
Activity of the sun
Propagation of solar wind
GIC in technological systems
Geoelectric field at Earth’s surface
Ionosphere processes (auroral & electrojets release)
Earth’s surface
Magnetosphere processes (magnetic reconnection)
Network configuration

Figure 1.1: The space weather or GIC chain 3
The hot, outer layer of the Sun is known as the corona 5. The corona is made of hot plasma, reaching temperatures between 1x106K to 6x106K. Periodically, the Sun loses mass in the form of coronal mass ejections (CMEs). The CMEs are ejected into the space towards the earth. The occurrence of a CME is solar cycle dependent. Each directed CME hit the earth’s magnetosphere and causes distortion of its magnetic field. CMEs stretch the magnetosphere on the night-side of earth causing it to release energy through magnetic reconnection (see Figure 1.2).

Figure 1.2: Reconnection of the magnetosphere due to interference of coronal mass ejections (CMEs)
The perturbations in the magnetosphere have an impact on the stability of the ionosphere 6. The dynamic changes in the magnetosphere link with the ionosphere through the ionosphere’s polar regions. During the magnetosphere-ionosphere interactions, the magnetosphere’s current system transfers energy to the ionospheric particles. These variations and couplings result in auroral and other electrojets in the ionosphere, which are horizontal electric currents flowing in the D and E layers of the ionosphere 7.
The variation of the magnetosphere-ionosphere electric fields result in temporary variation of the earth’s magnetic field at the earth’s surface. The potential difference, gives rise to the flow of geomagnetic induced currents. The conductivity profile of the earth’s surface determines the surface impedance which in turn determines the characteristics of the resultant geo-electric field. The magnitude of the induced electric field depends upon the rate of change of magnetic field, and the earth’s conductivity. The relationship between the changing magnetic and electric fields are given by the Maxwell-Faraday equations:
?xE=-?B?t where ?x is the curl operator (1.1)
?E E.dl=-??t B.ds (1.2)
V=-???t Faraday’s Law (1.3)
Countries close to the earth’s poles such as Canada, Norway, Sweden and Russia are more vulnerable to geomagnetic induced currents. The risk of geomagnetic induced currents is larger in networks located at highly resistive regions as on igneous rocks 3. A detailed mathematical model which confirms that a more resistive earth gives higher electric fields is given in 4.
1.3 OBJECTIVES
This project seeks to have an in depth understanding to the transformer responses to geomagnetic induced currents. The work will be centred on a 3p-5L core structure. In brief, the objectives of this thesis are:
To investigate through experiments and FEM simulations, the reactive power consumption of 3p-5L power transformers under the influence of geomagnetic induced currents.

To investigate the thresholds of geomagnetic induced currents that may cause noticeable degradation to power transformers.
1.4 HYPOTHESIS
Two hypothesis are to be tested:
These tests will be used to prove statistically two hypotheses;a) H2a Tests on model transformers and extension of the results to power transformerswith suitable transformer equivalent circuit and FEM simulations will improve theconventional models of the reactive power requirement in transformers conductingGICs.b) H2b Thresholds of GICs initiating damage in transformers, based on identifiablemechanisms of degradation, can be determined from the practical records oftransformer degradation leading to relatively early failure, and calculation of theassociated GICs.

1.5 RESEARCH QUESTIONS
The following research questions have been set up to assess the project hypothesis:
How does reactive power increase in transformers saturated by the flow of GIC affect power system stability?
What is the role of installing GIC monitoring devices in order to fully understand the phenomenon behind the risk of quasi-dc current to transformers?
How does different structure of transformers affect their response to GIC?
What are the different levels of GIC that cause noticeable degradation in power transformers?
How does the reactive power consumed by a power transformer vary with respect to GIC?
What is the implication of general power theory in determining reactive power absorbed by the transformer as opposed to conventional methods of calculating power?
CHAPTER 2: LITERATURE REVIEW
2.1 INTRODUCTION
This chapter introduces details of past GIC events and the extent of damage they caused to transformers. Records of currents that flow in transformers that were damaged by the Halloween Storm (2003) and the Hydro Quebec events are also given. In addition, a detailed explanation of the transformer electrical responses to geomagnetic induced currents, with particular reference to 3p-5L transformers that form the basis of this study is reviewed.
2.2 HISTORICAL EVENTS
There are three main events that occurred since the geomagnetic induced currents were discovered. The first event being, the Carrington event (1859) discovered by the British astronomer Richard Carrington which, for the very first time in the history, observed a solar flare. The second event in the history of GICs was the Hydro-Quebec event (1989). This had devastating effects on the entire Quebec power system, causing a blackout and an estimated loss amounting to $13.2 million 4. The most recent event was the Halloween Storm (2003) that ravaged a couple of transformers in Eskom’s network (South Africa), Sweden and England. This event cleared the popular belief that Africa was not prone to the geomagnetic induced currents.
2.2.1 THE CARRINGTON EVENT (1859)
This biggest solar flare known in man’s history took place on 1 September 1859 and it has been named Carrington event after the British astronomer Richard Carrington which, for the very first time in the history, observed a solar flare. On September 1, 1859, Richard Carrington observed a very intense white light flare on the surface of the sun from 11:18 to 11:23 a.m. GMT 8. The solar flare was followed by a magnetic storm on September 1–2, 1859 at the Earth. The time delay between the flare time and the magnetic storm was approximately 17 hours and 40 min. While a Coronal mass ejection (CME) normally spends two to four days on its journey from the Sun to the Earth, it took merely 17 hours before the Earth experienced a big geomagnetic storm, probably due to a large CME. It lasted for days and the effects where many and widespread. Colorful northern lights could were observed all over the world at latitudes as far south as Tahiti, and it is reported that light emitted from the northern lights was bright enough to read the newspaper without any further light sources. The geomagnetic storm also caused global telegraph lines to spark, setting fire to some telegraph offices and telegraph systems all over North America and Europe went down 9.

2.2.2 THE HYDRO-QUEBEC EVENT (1989)
On the night of March 13, 1989 a severe geomagnetic disturbance (GMD) caused a protective relay to trip due to GIC flow. The tripped static VAR compensators caused a cascade of failures throughout the Quebec power grid; most notably five transmission lines from James Bay were tripped causing a loss of 9,450 MW. The total load in the grid at the time was about 21,350 MW. This led to tripping of several protective relays, cascading failure, and resulted in the entire Quebec power grid to collapse. The whole sequence of failure events happened fast and within 75 seconds from the first capacitor tripped 10 six million people were left without power for up to nine hours during this wintry period. After the Hydro-Quebec the world realized the seriousness of the imposed GIC risk and several power companies in the Western world began to investigate GIC risk and do research on mitigation strategies.

2.3 REPORTED GIC EVENTS IN LOW LATITUDES – INCLUDING AFRICA
GICs were previously associated with areas of high latitudes (this excluded Africa in GIC studies) but recent researches have proved otherwise. Results from practical measurements have shown that GIC currents exists in low latitudes and several transformers were damaged in South Africa and Namibia during the Halloween storm 11. Measurements of GIC magnitudes during different geomagnetic storm phases show that values obtained in power networks located at low and middle latitudes can reach the same levels as those observed at high latitudes. For example, 13 present GIC recordings around 15 A in a Brazilian power network during 2004, which were also recently reproduced by calculations 14. Koen in 1999, 15 identified that the following substations are at high risk due to GIC; Alpha, Hydra, Beta, Grootvlei, Perseus, Grassridge. These substations were then equipped with GIC monitoring devices on their neutral. On 31 March 2001, GIC currents measured on transformer neutrals at Grassridge substation reached a peak of 5A for 1 minute. A 400/220/132 kV, 3p-3L transformer at this substation saturated. Sixth harmonic current in the neutral reached a maximum of 8A, this confirms the presences of GIC. Third order harmonics are more dominant in GIC saturated transformers. Further evidence of GIC effects in low latitudes was obtained after the Halloween storm of 2003 that left transformers damaged in South Africa and United Kingdom. On the 17th of November 2003, the transformer at Lethabo power station tripped on Buchholz protection on 17 November. This showed an accumulation of gas in the Buchholz relay. There was a further severe storm on 20 November. On 23 November the Matimba #3 transformer tripped and on 19 January 2004 one of the transformers at Tutuka was taken out of service. Two more transformers at Matimba power station (#5 and #6) had to be removed from service with high levels of dissolved gases in June 2004. A second transformer at Lethabo power station tripped on Buchholz protection in November 2004. In October 2003, a GSU at Matimba substation failed permanently, three weeks after the Halloween storm 12. Ruacana power station in Namibia is one of the substations identified as GIC prone by Koen in 1999. Again in Namibia during the Halloween storm had the same impacts as that noticed in other substations affected by GIC in South Africa. Geographically the Ruacana substation is separated 1400km from the substations that were affected by GIC in South Africa. The substations identified to be on high risk by Koen are in the Southern African Power Pool (SAPP) and are interconnected with Ruacana via an interconnector. This means that they are in the same synchronous network and the effects of GIC may have the same effect on both networks. On 11 December 2003 the protection tripped a two-year old 90 MVA 330 kV GSU at Nampower’s Ruacana power station in northern Namibia. One HV winding, of this GSU failed permanently and the mode of failure was similar to that of transformers in South African grid.

The Halloween storm had the same impacts in the United Kingdom that sit on the same latitude as South Africa. On 20th October, 1989, the transformer neutral current varied from +5 A to -2 A at Norwich Main in East Anglia, Pembroke in Wales, and Indian Queens in Cornwall for ten minutes. Two identical 400/132 kV, 240 MVA transformers at Norwich Main and Indian Queens failed, the voltage dips on the 400 and 275 kV systems were up to 5%; and very high levels of even harmonic currents were experienced due to transformer saturation by the geomagnetic storms 16
2.3.1 GIC IMPACT ON REACTORS
Literature has shown that not only transformers are at high risk of GIC. Static VAR compensators, insulators, surge arrestors and transformers were reported to have failed in the Hydro Quebec event 8. Well documented reports of transformer failures in South Africa has shown that a greater number of reactor also failed due to GIC exposure 17. The following transformers and reactors were affected by very severe storms during the period 1989 to 1994:
Table 2.1: Transformers and reactors that failed in South Africa (1989-1994) 17
Date K-index sequence Name Description
15 March 1989 6443 3332 Poseidon-Neptune reactor Permanent fault: inter-winding fault
28 July 1990 2456 6566 Beta reactor 4 Internal fault: reactor removed on 08/09/90
24 March 1991 2866 5378 25 March 1991 6643 4555 26 March 1991 6555 7443 Hydra transformer 21 Permanent fault: reason unknown
18 April 1991 Beta reactor 4 Neutral earthing reactor faulted
18 April 1991 Beta reactor 2 Internal fault
19 June 1991 Hydra reactor 2 Permanent fault, reactor was removed
14 August 1991 4375 4322 Beta reactor 4 Neutral earthing reactor faulted and was disconnected
19 August 1991 Hydra transformer 21 Permanent fault: transformer removed
25 May 1992 Hydra transformer 3 Transformer tripped on buchholz protection
06 May 1993 Hydra reactor 1 Internal fault
14 Dec 1993 Beta Alpha reactor 2 Red phase winding fault
21 March 1994 Hydra Poseidon reactor 1 Reactor faulty and replaced
Koen and Gaunt 18 reported reactor failures and elevated levels of dissolved gas closely associated with exposure to geomagnetic storms. A reactor at Poseidon-Neptune substation failed, following the two days of severe geomagnetic activity on 15 March 1989. Shunt reactors are similar in many respects to power transformers, except that they generally have gapped cores. Laboratory tests at University of Cape Town show that direct current can flow in model three-limb three-phase reactors with small gaps. The response of a reactor to GICs could be similar to the response of a three-limb transformer, with the core gap of a reactor having a similar effect to the core-tank gap of the transformer. Despite the relatively high reluctance of the magnetic path compared with a closed-core, some quasi-dc current will flow through a reactor and, as for a transformer.

Importance of installing GIC measuring devices.

As with any other threats to the power system, GIC currents need to be detected, measured and control measures should be taken to counter their effects. Many researchers have mentioned that, it is difficult to study the effects of GIC current on power systems due to the lack of measuring instruments in most parts of the world. In Africa, according to reviewed literature only South Africa and Namibia have taken the first step towards installing GIC measuring devices. The substations that are currently monitored are Ruacana power station (Namibia), Alpha, Hydra, Beta, Grootvlei, Perseus, and Grassridge substation. (South Africa) 12. These substations were identified to be at high risk, in event of high geomagnetic activity 15. Researchers are relying mainly on calculated values of GIC which may not give a true reflection of the real scenario. For instance, the highest the highest GIC calculated so far in South Africa is 108A at Alpha substation. The table below summarizes the calculated values of GIC at substations in South Africa.

Table 2.2: Substations most susceptible to GIC in South Africa 18.

Substation Maximum calculated GIC on 13 March 1989 averaged over 1 minute
Alpha 108
Hydra 67
Beta 64
Grootvlei59
Perseus 57
Grassridge41
On the other hand, the highest measured value of GIC is as low as +/-9A 12. The readings were obtained from Grassridge substation transformer neutral on 24 November 2001. The storm duration was very short. The 3p-3L transformer saturated and this appears to contradict the theory that three limb, core type transformers are not susceptible to GIC saturation. Takasu et al.’s model (1994) states that three phase three limb is not susceptible to GIC damage 19
2.4 SUSCEPTIBILITY TO GIC AMONG DIFFERENT STRUCTURES
The presence of low reluctance return paths (white arrows) increases the core’s tendency to saturate during GIC induced dc bias. 10. The susceptibility of a transformer core to GIC saturation is dependent of the presence of dc flux paths, represented by white arrows in Figure 2.1.

Figure 2.1: Transformer core types. GIC Susceptibility varies between transformer core types 10.

In the case of a three-phase three-leg transformer there is no complete dc flux path in the core. In these transformers, the dc flux must leak into the transformer tank. In fact, all transformers are subject to some degree of flux leakage into the tank. Because the transformer tank is not designed as a magnetic core, the tank can be very susceptible to damage due to heating.

2.5 TRANSFORMER RESPONSES TO GIC
GIC flow in a power system causes half-cycle saturation in transformers 19. Half-cycle saturation does not occur instantaneously and depends on the electrical characteristics of the transformer and GIC amplitude 20. The magnetization current in a GIC saturated transformer is rich in odd and even harmonics 12. The magnetization current is distorted because the transformer is operating in a non-linear region of the hysteresis curve. Harmonics cause mal-operation of protective relays, heating in transformers and other equipment on the power system 21.

GIC flow in transmission lines
Harmonics
Voltage ; angle stability
Transformer half cycle saturation
Transformer Heating
P ; C incorrect operation
Reactive power support tripping (SVC, Capacitors, re
Reactive power loss
Generator overheating and tripping
Voltage control
GIC flow in transmission lines
Harmonics
Voltage ; angle stability
Transformer half cycle saturation
Transformer Heating
P ; C incorrect operation
Reactive power support tripping (SVC, Capacitors, re
Reactive power loss
Generator overheating and tripping
Voltage control

Figure 2.2: Effects of GICs on entire power systems
In the transformer, the flow of harmonics increases the eddy currents in the transformer windings and core, causing additional heating and losses in the transformer. Beyond the knee point, the core’s inductance drops heavily and it approaches air-core inductance as the GIC continues to increase 22. At air-core inductance the core’s permeability reaches that of air. The drop in core inductance allows the easy flow of eddy currents exacerbating core heating.
Half-cycle saturation also causes the transformer to draw more reactive power 23. This poses stability issues on the power system. This may cause voltage drop at the load end and frequency increase on the power system. System stability can be severely increased by tripping of reactive power support such as lines, static VAR compensator and capacitors as a result of harmonic currents. At saturation, leakage flux flow in the metallic parts (tank, core bolts and clamps and tie plates). The leakage flux causes eddy currents to flow in metallic parts of the transformer, resulting in heating of these metallic parts. Excess temperature due to heating causes deterioration of transformer insulation and this reduces the life-span of the transformer 31.
In essence, when GIC flow through a transformer windings, ac voltages will superimpose with dc waveforms resulting in a dc-offset, driving the transformer into saturation in the positive or negative cycle depending on the polarity of the offset. To sum up, half-cycle saturation has the following repercussions (a) increased production of both even and odd harmonics, (b) increase in reactive power consumption and (c) increased heat production and an increase of transformer losses and (d) increase in transformer humming sound i.e. noise 24. The severity of these effects depends on the strength of the geomagnetic disturbance. These consequences will be discussed later in the sequence below:
Half-cycle saturation
Thermal effects on transformers
Harmonics generation
Reactive power demand
Increased noise
2.5.1 HALF CYCLE SATURATION
The flow of GIC causes asymmetrical saturation of transformers. This is normally termed half-cycle saturation. The superposition of the ac excitation current and the quasi-dc GIC current causes the transformer core to saturate for a portion of each half cycle 25. A transition from unsaturated to saturated core represents a change in inductance by several orders of magnitude. The magnetizing inductance of an unsaturated transformer is very large, and thus the rate of current increase is very slow until the transformer saturates. The effective core inductance variations reflect in the magnitudes of exciting current which account for the reactive power swings.

Figure 2.3: Excitation current of a transformer as a result of DC bias 26.

With ac excitation only, the transformer is designed to operate in the linear region of the characteristic curve – technically termed “hysteresis loop”. In this region, the transformer is capable of converting primary voltage induced into secondary voltage in a linear relationship determined by the equation: VpVs=NpNs. On the Y-axis the corresponding excitation current under normal operation is shown and on the X-axis the normal flux which causes this excitation current is drawn (Figure 2.3).

The introduction of quasi-dc currents results in a biased flux which offsets vertically in the positive direction, and this results in a sharp increase in exciting current – indicated as biased exciting current in (Figure 2.3). This shift reduces the effective core impedance and causes a corresponding increase in the reactive power absorbed by the transformer core 27
The biased exciting current lags the induced voltage by 90? 23. According to the power triangle, if the current lags voltage then reactive power is consumed by the transformer. Moreover, this current is so huge in comparison to normal excitation current, the resultant increase in reactive power is abnormal such that compensation equipment cannot supply this into the power system.

2.5.2 THERMAL EFFECTS ON TRANSFORMERS
The flow of GIC in transformers lead to a higher magnetizing current, which in-turn produce a higher leakage flux, which also contains a lot of harmonics 25. This leads to a significant increase of eddy and circulating current losses in both windings and structural parts of the transformer, causing heat generation and transformer losses 28, 29. During saturation most of the excess flux flows externally to the core into the transformer tank, generating currents and localized tank wall heating spots with temperatures reaching up to 175?C 30. The intensity of overheating depends on the level of GIC but is also a function of various design parameters of the transformer itself. These include the saturation flux paths, cooling flow and the thermal condition or loading of the transformer. When overheating occurs, it causes the breakdown of oil and paper insulation in the hot spot regions 31. The flux distribution under GIC is the most determinant factor to heating. The presence of a reluctance path in the 3p-5L allows it to be prone to saturation more than 3p-3L, hence GIC has more heating effect towards the 3p-5L 32. Repeated exposure to GIC results in long-term degradation of insulation, and because of lack of GIC monitoring equipment, most failures are not attributed to GICs.

Figure 2.4: Observed Meadow Brook transformer hotspot temperature for a minor storm on May 10, 1992 33.

Figure 2.4 illustrates thermal effects of GIC on transformers. This is a real scenario observed in Meadow Brook. The graph shows that geomagnetic induced currents of less than 60A have no effects on oil temperature. However, tank temperature seems to increase significantly under the same circumstances. None of the literature reviewed has stated that oil temperature increase under GIC. Nonetheless, temperature increase in the windings and structural parts have been widely reported 24. An unfamiliar explanation of increased heating in the windings was given in 25. R. Girgis et al, 1992 explains, “Transpositions that are provided in the windings are designed to minimize circulating currents due to flux present under normal operating conditions”. With the change in flux distribution pattern, these transpositions not only become ineffective but may also aggravate the situation and result in higher circulating currents.

2.5.3 HARMONICS GENERATION
The exciting currents of GIC-saturated transformers are highly distorted, and consist of harmonic components of both even and odd orders, as well as fundamental and dc components. Superimposed DC excitation will also cause the transformer to inject larger amounts of odd and even harmonics into the system thus affecting the normal operation of protective relays 34.

Figure 2.5: 14 500 kV simple demonstration circuit simulation results-transformer AC currents and distortion due to GIC.500 kV transformer AC current-normal and GIC-distorted 33.

Spot heating is a critical threat to power transformers as a result of GIC. Hotspots caused by spot heating due to harmonic currents can degrade the insulation in a transformer and reduce its service life; in extreme cases spot heating may cause acute failure of the transformer. The extent to which harmonic currents cause spot heating, and the impact of that heating on transformer life vary depending on various factors including transformer construction and core type. The precise harmonic current spectrums depend on transformer construction type but in general the harmonic magnitude tends to decrease with increasing harmonic order. The pattern in which harmonics vary in a 3p-5L shall be discussed in chapter 3, and simulations to validate the results shall also be carried out.

Harmonics generated will cause mal-operation of protection and control relaying. Moreover, compensation equipment will also fail to cope with the increased harmonics generated under GIC events. This has been the scenario in 1989 when GIC caused a blackout in Canada. Consequently, when compensation equipment fail, the system will eventually collapse due to the large reactive power requirement that occurs when half-cycle saturation takes place.

The harmonics injected into the system during GMD may, as a consequence of the physical and protection impacts resulting in critical line or equipment tripping, can potentially aggravate fundamental-frequency voltage stability issues 22. Thus, the critical concern regarding harmonics during GMD is not their impact on power quality in the conventional sense, but rather the potential impact of the harmonics on grid security. Some of the recorded effects of geomagnetic disturbances on power system are given in 21, 24:
Seven static VAR compensators (SVCs) tripped in rapid succession in the Hydro Quebec system, resulting in system instability and total blackout of that system. Post-event analysis revealed that harmonic distortion was the direct cause of the SVC trips.
Widespread capacitor banks trips, including 16 bank trips in the Virginia Power system, 12 in the New York Power Pool (predecessor to NYISO), four at Bonneville Power, seven in the Allegheny Power system, and at least three in the PJM system (including 500 kV capacitor banks).
Generator trips due to negative sequence or phase imbalance protection in the Manitoba Hydro and Ontario Hydro systems (including one major nuclear unit).
Static VAR compensator trip in the WAPA system.
HVDC system trip at the WAPA Miles City station and an HVDC filter trip at the Comerford converter station in the New England system.
Transmission line trips at Manitoba Hydro and WAPA.
2.5.4 REACTIVE POWER DEMAND
Figure 2.6 provides a comparison of reactive power loss for two core types of transformers as a function of the amount of GIC flow. It can easily be seen that single phase transformers are the worst affected by GIC flow as noticed by heavy consumption of reactive power of nearly 40 MVARs when 100A of GIC flows in the transformer, in comparison with slightly less than 10 MVARs in a three phase core structure.

Figure 2.6: Transformer increased reactive power demands (MVARs) due to GIC for a typical 500 kV transformer for single phase and three-phase three-legged core type 33.

In saturation region of the transformer B-H curve, a small increase in magnetic flux, causes a dramatic increase in the magnetizing current – typically 10-20 times the normal excitation current as seen from Figure 2.7. This increases the reactive power drawn by the transformer drastically. The large reactive power draws of GIC saturated transformer make proper operation of the power system difficult and tend to lead to power system instabilities.

Figure 2.7: Transformer magnetizing current characteristics for normal operation and for half-cycle saturation due to the presence of GIC 33.
The magnetizing current of a power transformer increases sharply when it is subjected to geomagnetic induced current. Since the magnetizing current lags the system voltage by 90°, it creates reactive power loss in the transformer and the impacted power system 23, 33. Under normal conditions, transformer reactive power loss is very small. However, the several orders of magnitude increase in exciting current under half-cycle saturation also results in extreme reactive-power losses in the transformer. For example, the three-phase reactive power loss associated with the abnormal magnetizing current of transformers (Figure 2.7), produces a reactive power loss of over 40 MVARs for this transformer alone. The same transformer would draw less than 1 MVAR under normal conditions.
The increased magnetizing current drawn by the GIC saturated transformer results in substantially greater core losses in the transformer. These core losses result in increased heating both in the transformer core and in other metallic components because of flux leakage. This heating can severely reduce the lifespan of a transformer. The tests conducted in 32, 35, show the order of increasing reactive power consumption as:
Three-phase three-limb
Three-phase five-limb
Single phase transformers
The presence of low reluctance return paths (white arrows) increases the core’s tendency to saturate during GIC induced DC bias, hence more reactive power absorbed 10
Significance of reactive power on a power system
When current is in phase with voltage on a power system, real power is transmitted and when the current is out-of-phase with voltage then reactive power is absorbed by the system. Reactive power supports voltage in a power system, therefore the amount of reactive power determines the value of voltage in a system. An increase in reactive power consumption by a system results in voltage collapse as seen in the case of GIC flowing in power transformers, and the opposite is true. Voltage control is done by regulating the amount of reactive power. Capacitors inject reactive power in a power system thereby boosting the voltage profile. Inductors on the other hand, consume reactive power thus resulting in a voltage drop. This is how voltage control is achieved.
On a transmission line, voltage is needed in order to transmit current to the load. Reactive power is used to build up the voltage levels necessary for active power to be transmitted. In essence, reactive power is essential to move active power through the transmission and distribution system to the customer. Reactive power is required to maintain the voltage to deliver active power through transmission lines. Finally, to understand the concept more clearly, reactive power is linked to power factor in the sense that they both measure losses in a power system. In the event that the phase angle is not zero i.e. voltage and current are out-of-phase, power factor is not equal to one, it shows that there are losses in the power in the power system. These losses, however, translate to reactive power drawn to support magnetic (either inductive or capacitive) needs of the system.

2.5.5 INCREASE IN TRANSFORMER NOISE
The noise in a transformer is a result of magnetostriction. Even under normal operation a transformer produces a humming sound as a result of magnetostriction. The degree of flux determines the amount of magnetostriction and hence, the noise level. The flow of GIC increases the flux in the core and causes the humming sound to increase 36. This may happen for a few minutes as GIC currents continuously fluctuate and have short duration peaks. This means the flux increase momentarily, hence a short duration of increased magnetostriction. The expansion and contraction of ferromagnetic material (magnetostriction) in saturated transformers causes noise and mechanical vibration, this may lead to mechanical failure 37
GIC related increase in transformer noise has been reported in China, in Guangxi province 38. Three heavy buzzing sound was heard three consecutive times for 1.5 min on October 31, 2003. This happened at 4:20am, 9:20am and 10:20am Universal Time (UT). Maintenance checks were conducted the circuit breakers, current transformers, surge arrestors and protection relays of the main transformer was made, and no abnormal condition were noticed. At 18:00 on November 5, 2003, the transformer had abnormal noise which was a little higher than usual and disappeared in about 2 min. At 20:38 on November 20, 2003, the same transformer had heavy buzz noise again. In December 2004, the transformer was repaired, and it was thought that the abnormal noise was caused by the loosening of the winding underlay and insulating brackets 39. However, the time this transformer experienced abnormal noise corresponds to high geomagnetic activities in China 40.

Another transformer at the Shanghe substation in the Jiangsu Province (China) was disturbed with unknown abnormal noise and severe vibration between March 2001 and October 2002 37. After joint analyses by specialists, it was concluded that the disturbance on the 750MVA transformer, which consists of three single-phase transformers, was caused by dc biasing resulting from GIC 42, 43.

Table 2.3: Increase in core loss and core noise under varying GIC levels (single phase transformer) 25.

GIC Amp/Phase % Core loss increase Core noise level in dB
15 31.1 27.5
20 34.2 30.0
30 38.9 33.8
40 42.6 36.9
50 45.5 39.3
Transformer noise increases significantly under GIC. This increases vibrations on the transformer, resulting in gradual loosening of bolts and other fastened areas within the transformer. Loose connections are known to cause hotspots in current carrying parts of electrical equipment. Table 2.1 shows experimental results validating that there is an increase in core loss and noise with increase in GIC.

2.6 CONCLUSION
As discussed, the impacts of geomagnetic induced currents on transformers are: half-cycle saturation, reactive power increase, harmonics generation, thermal effects, and erratic noise increase. As a result, insulation degradation due to heating, hotspots, transformer tank destruction, increase in core losses, and noise in transformers may occur in the affected transformer. Half-cycle saturation in transformers has ripple effects to the entire power system in general. And these are: protection and control mal-operation, voltage and load angle instability, reactive power loss, reactive power compensation equipment mal-functioning and harmonics generation. If proper planning is not available prior to GIC events, transformers can be totally destroyed and a possible blackout can happen depending on the magnitude of GIC.
CHAPTER 3: LABORATORY TESTS AND RESULTS
3.1 INTRODUCTION
This chapter describe the laboratory tests performed on 3p-5L transformer and present results from these tests. The main focus of the tests were to address the first hypothesis that needs to establish a clear relationship of varying geomagnetic induced currents and its respective reactive power consumption on a 3p-5L transformer. Some of the tests that would lead into the second hypothesis were done; that is to find the thresholds of GIC initiating damage in transformers. These tests are flux distribution measurement, harmonics and hotspot temperature variations in the transformer windings, core and extra limbs of the five legged transformer.

3.2 TESTS PERFORMED
The following tests were carried out to investigate the behavior of a 3P-5L transformer when subjected to geomagnetic storms:
Saturation characteristic tests
Reactive power consumption under dc-bias
Total harmonic distortion
Flux distribution tests
Load end voltage measurements
3.3 EXPERIMENTAL SETUP
In practice geomagnetic induced currents flow in the earth and enter the transformer neutral via grounded neutrals of transformers and they flow through transmission lines to the next substation and out again through grounded neutrals of star vector group transformers. The consequences of geomagnetic storms elaborated on the power system emanate from the transformer as discussed in the literature review. A complete replica of the real scenario would be to connect to transformers as given below:

Figure 3.1: Test system for transformer response to geomagnetically induced currents (GIC).
3.4 TEST EQUIPMENT
In order to carry out the tests effectively, without damaging the transformer and to extract reliable results, the following equipment was used for this project:
DC Source: The dc supply circuit consisted of a 12 V, 7.2Ah, rechargeable lead acid battery and a variable resistor to change dc current values.

Switch: 1? resistance switch
Power Meter: A high precision and wide bandwidth, IEC76-1(l976) compliant Yokogawa WT1600 digital power meter was used for reactive power, voltage, current, and harmonics measurements.

Load: A three-phase resistor bank load of 35 VA per phase and 100?.

Temperature Measurement: A temperature gun (Sentry ST642) that uses infrared technology was used, with an accuracy of ±2°C for temperatures ranging between -20 and 100°C.
Source transformer: 230/120V, 900VA, 3p-5L
Transformer under test: 120/230V, 300VA, 3p-5L
Variable load supply: A 3 phase (0-400V) variable three VARIAC.

3.5 TRANSFORMER UNDER TEST SATURATION CHARACTERISTICS
A transformer is designed to operate in the linear region below its knee point. If the transformer voltage is increased beyond the operating range which is normally within 10% of its operating range, it begins to saturate. In the laboratory experiment, I excited the transformer until there was no major changes in the input voltage but the current kept on increasing. This determines saturation and a plot of the input voltage versus input current in Figure 3.2 shows the saturation curve obtained. This was saturation using ac voltage. On the other hand, introducing a dc source into the transformer may lead to quick saturation.

6253931279212001156029103284600625393307959
Figure 3.2: Saturation characteristic of the 3p-5L transformer under test.

These ac injection tests established that the magnetization current magnitude of the 3p-5L transformer was 0.0408 A 74mA by hkc per phase, and this corresponds to transformer’s knee-point voltage. The knee point voltage is 74.828V. The current values were used to determine the dc values of injected of injected current, as advised under the laboratory protocol explained in 44. Introducing geomagnetic induced currents with a transformer may saturate the core, causing it to operate in the extremely non-linear portion of the core steel magnetization curve.
The excitation curve represents the core material characteristic and the core structure 22. At flux levels below the knee point, the curve is linear, with a very steep slope. This flux level is normally 1.7T for power transformers. A small exciting current flows when the flux is at or below the rated value. The slope in this range represents the magnetizing inductance. The magnetizing inductance (slope of the curve in the unsaturated region) is determined by both the characteristics of the core steel and the small gaps in the steel at core joints. As the flux level reaches the rated value, the slope or inductance decreases slightly. This is because the flux is concentrated at the joints, and localized saturation begins to occur. The saturation curve begins with slight saturation just above the knee point. As operating voltage continues to rise, this drives the transformer into deep saturation. Deep saturation causes the core’s inductance drops heavily and it approaches air-core inductance as the GIC continues to 45. As the transformer reaches air-core inductance, the core’s permeability reaches that of air. This inductance is commonly called the “air core inductance” because it is typically calculated based on the transformer winding configuration alone, as if that winding is suspended in air without any magnetic core. In reality, the influence of the tank, structural members, and flux shields will make the final slope of the saturation curve slightly greater than the true “air core” inductance. Despite this difference, the common industry usage applies the term “air core inductance” to the final slope of the saturation curve.

3.6 REACTIVE POWER
Reactive power measurements were taken using a high precision Yokogawa digital power meter. It has inbuilt voltage and ammeters and these were connected according to Figure 3.1. Automatically the values of reactive power are computed using the conventional power theory. DC current injections were done for a period approximately between 2 to 4 minutes before recording the final reactive power measurements. Table 3.1 shows the results obtained for varying dc current increasing from 0A to 1.3A per phase.

Table 3.1: Experimental results showing dc injected against reactive power consumed measured conventionally.

DC Input I_dc neutral/A Reactive power consumed/VA
DC0 0.000 3.02333
DC1 0.995 33.7967
DC2 1.155 40.1167
DC3 1.293 45.7233
DC4 1.516 54.8000
DC5 1.754 64.6800
DC6 2.056 77.1200
DC7 2.362 90.2300
DC8 2.637 101.503
DC9 2.883 111.567
DC10 144.443
Table 3.1: Test results for varying neutral current and reactive power consumption of the 3p-5L transformer.

The reactive power is the power required to maintain magnetic fields in ac circuits. The conventional methods of calculating power makes use of the power triangle and reactive power Q is obtained from the relationship, Q2=S2-P2. The results show a perfect linear relationship between the reactive power and geomagnetic induced currents within a 3p-5L transformer (Figure 3). Hence, power utilities may use this relationship to notice if there are any potential hazards that might result from geomagnetic storms. To ascertain this, the controller on duty must be alert and may contact the space weather department to know the conditions of geomagnetic activity at that moment. At present, most utilities are still unaware of the effects of geomagnetic induced currents on power system.

Figure 3.3: Reactive power against injected dc trend within a 3p-5L transformer.

3.7 SEARCH COIL VOLTAGE OR LEAKAGE FLUX MEASUREMENT
Coils made from 10 turns of Copper wire were made to capture the leakage flux on the center limb, outer limb and frame. The voltage induced by the leaked flux on the coil was measured as the value of DC injected to the transformer neutral was varied from zero to 1.2A/Phase.

Figure 3.4: Leakage flux measurements on 3p-5L using search coils.

Leakage flux on the centre limb increased slightly as GIC increased but on the outer limb the coils could not pick up any leakage flux. In actual sense, all coils should pick up stray flux, but in this case the no flux changes could be picked. The reason could be that the transformer did not reach deep saturation with the dc injected. Further dc could not be injected as the transformer was smelling, showing that the dc flow could not be sustained by the winding insulation. As a precaution, the experiment was stopped.
3.8 ANALYSIS OF HARMONICS
The principal threat to electrical infrastructure during a GIC event is spot heating of the transformer core due to harmonic currents. TDD is indicative of the ratio of aggregated harmonic currents to rated fundamental current.
TDD=h=2NI(h)Irated2 (2)
The definition of the current TDD is given in 46:
ITDD=I22+I32+I42+I52+…IL×100% (3.1)
Current TDD: Total demand distortion of the current waveform. The ratio of the root-sum-square value of the harmonic current to the maximum demand load current. A factor called the total harmonic distortion (THD) is used to quantify the harmonic content of a given voltage or current wave. The THD is given as a percentage figure. THD is a measure of the magnitude of all the harmonic components present in the wave as compared to the magnitude of the fundamental component 47. The THD is expressed as follows:
THD=h=2NI(h)I(1)2 (1)
A THD of 5% for a voltage wave means that the harmonic content is 5% of the fundamental component. Definitions of THD are given in 46:
VTHD=V22+V32+V42+V52+…V1×100% (3.1)
Voltage THD: Total harmonic distortion of the voltage waveform. The ratio of the root-sum-square value of the harmonic content of the voltage to the root-mean-square value of the fundamental voltage.

ITHD=I22+I32+I42+I52+…I1×100% (3.2)
Current THD: Total harmonic distortion of the current waveform. The ratio of the root-sum-square value of the harmonic content of the current to the root-mean-square value of the fundamental current.

The extent to which harmonic currents cause spot heating, and the impact of that heating on transformer life vary depending on various factors including transformer construction and core type. Figure 3.9 illustrates the THD as a percentage for current, voltage and power waveforms.

Figure 3.5: Total harmonic distortion as measured by Yokogawa power meter. (try use fluke power quality analyser to give harmonic content)
The results show a fluctuating trend in harmonic distortions of current and voltage waveforms as dc values injected increase. The results correspond to practical tests and MATLAB simulations performed in 48 also show fluctuating harmonics as dc values are increased. The bench-scale transformer (three-phase transformer, 440V/12V, 12VA) used in 48 was almost similar size to the one used in this dissertation.

Table 3.2: Voltage distortion limits in IEEE Std. 519 46.

Special applications General system Dedicated system
THD 3% 5% 10%
The limits are for low voltages ranging from 120V to 69000V. The definitions for special applications and dedicated systems are:
Special application: special applications include hospitals and airports.

Dedicated system: a dedicated system is exclusively dedicated to the converter load.

The choice of these limits is influenced by the cleanliness of the input power required by various loads that can be connected to the transformer load side. For example, section 6.6 of Std. 519 states: “Computers and allied equipment, such as programmable controllers, frequently require ac sources that have no more than a 5% harmonic voltage distortion factor, with the largest single harmonic being no more than 3% of the fundamental voltage. Higher levels of harmonics result in erratic, sometimes subtle, malfunctions of the equipment that can, in some cases, have serious consequences.

3.9 SURFACE TEMPERATURE MEASUREMENTS

Figure 3.6: Increase in temperature with increasing dc current.

Temperature measurements on the windings and core show an increasing trend with GIC increase. As explained in the literature review, the eddy currents flow in the windings and increase with GIC. The 3p-5L is no exception to this trend. In addition half-cycle saturation causes leakage flux to flow in the frame. The bench-scale transformers used had no tank, hence air cooled. Further experiments to determine the thresholds of GIC initiating damage in the 3p-5L are going to be carried out using a 15kVA, 400/208V transformer that resembles a larger power transformer. Extrapolation is going to be used to determine these thresholds in large power transformers. NERC has carried out tests on single phase transformers and set the threshold at 75A/Phase 49. This threshold assumed that the safe margin of hotspot temperature would be 150?C, although IEEE and IEC thresholds for hotspots are 200?C and 180?C.
3.9 LOAD END VOLTAGE MEASUREMENTS
The voltage profile of a power system under sever geomagnetic disturbance normally falls as shown in Figure 3.7. This is due to the increased reactive power consumption of the saturated transformer.

Figure 3.7: Load voltage profile of 3p-5L transformer under dc bias
The increased reactive power will add to the load of the generators at power station. A heavy load slows down the speed of the generators and as a result the frequency drops. The relationship below illustrates the how frequency depends on the speed of generators;
f=np120 (3.3)
Where;
f=frequency
n=speed in revolutions per minute
p=number of poles
The excitation system responds by increasing the excitation current and thus boosting the MVAR flow into the system. At the same, the SVC capacitors will also kick in in order to offset to supply the extra demand in reactive power. According to 50, the additional reactive power demand during a geomagnetic disturbance (GMD) event, if not offset by reactive power compensation equipment, can cause a reduction in the system voltage to the point of encroaching secure system limits. In actual sense, the quasi-dc caused by GIC is not unidirectional, therefore the reactive power is fluctuating resulting in power swings.

Figure 3.8: The voltage collapse and over-voltages as observed at the Jacques-Cartier substation 51
A closer look at the Hydro Quebec collapse (Figure 3.8) shows that voltage declines during a GMD, and as the voltage reached 0.7p.u the first line tripped. Lines have under-voltage and overvoltage protection, and in this case the under-voltage relay tripped the line. Also, if the line is protected by distance protection as the voltage drops drastically, the impedance of the line may fall within the tripping zone thus the line is taken out of service. The tripping of the line further worsened the reactive power demand, and the voltage further dropped. The tripping of the last line remaining line separated the La Grande network from the Hydro Quebec network. Complete isolation of La Grande network caused frequency to rapidly fall on the Quebec network. In response, automatic load shedding system tripped all load but could still not offset the loss of approximately 9400MW of generation from La Grande Complex. Complete separation of the network caused the voltage to rise dramatically. The reason for a sharp increase in voltage is that the generators were running fast and unexpectedly there was no load to supply hence an open circuit condition was created. The no load voltage is usually very high, the only load to the generator was the transformer and at nearly 1.5p.u one phase of the transformers permanently failed.

3.10 CONCLUSION
Transformers under dc bias experience an increase in reactive power consumption which causes power swings on the power system and this may lead to blackouts if the reactive power reserves do not offset this disturbance. Load end voltage was seen to decrease with increasing dc current injected in the neutral, this may result in power swings on the system. Temperature increase was also recorded with increase in dc bias. This may affect the insulation of the transformer and cause additional heating in structural parts of the transformer.

CHAPTER 4: REACTIVE POWER UNDER GEOMAGNETIC DISTURBANCE
4.1 INTRODUCTION
This chapter outlines the reactive power trends of a transformer under geomagnetic disturbance. In literature it was concluded that, a transformer under half-cycle saturation will absorb a considerable amount of reactive power. This poses voltage instability within a power system and may lead to a blackout if the reactive power reserves do not offset the imbalance. Methods of calculating reactive power are going to be examined. The conventional method of calculating power is widely used. Work done at University of Cape Town has proved that conventional methods may underestimate calculation of the reactive power under power disturbances. Hence this would be misleading to power system operators when planning their power system reserves and mitigation under severe geomagnetic disturbances. Action to be taken by system operators under severe GMDs will also be covered.
4.2 CONVENTIONAL METHOD OF CALCULATING POWER
In this principle there are three types of power: apparent power (complex power), reactive power and real power (active power). The active power is also called the real power in terms of its association with the real component in a mathematical expression of real and imaginary components.

4.2.1 REACTIVE POWER
Reactive power support the magnetic field and electric fields necessary to operate power system equipment 47. Reactive power is stored in the electrical and magnetic field that exists in the system. When electrical equipment is energized via ac voltage, an electrical field is created. When ac current flows through a conductor a magnetic field is created. The electric and magnetic field continually build and collapse with the changing magnitudes of ac voltage and current. When the electric and magnetic fields are building, they store reactive power. When these fields are collapsing, they return reactive power to the system. Reactive power is measured in volt-ampere-reactive (VAR).

4.2.2 ACTIVE POWER
Active power is often referred to as real power. Active power is the useful or working energy supplied by a power source. Active power is used to perform work such as lighting, heating and turning a motor shaft. Its unit is the watt (W).

4.2.3 COMPLEX POWER
Together, active power and reactive power make up complex power, which is the total power flow in a circuit. Utilities use generators to produce active and reactive power. Complex power is the total power transferred to customers through transmission lines. The unit of complex power is volt-ampere (VA).

4.2.4 POWER TRIANGLE
Complex power, reactive power and active power are represented by a power triangle Figure 4.1. The angle,? is known as the phase angle and cos?, is the power factor. Applying the Pythagoras theorem to the power triangle yields:
MVA2=MW2+MVAR2 (4.1)
OR
S2=P2+Q2 (4.2)

Figure 4.1: The power triangle 47
The cosine of the phase angle between the MVA and MW in the power triangle is the power factor. The power factor is also equal to the ratio of active power and complex power on the system.

Power factor=Active powerComplex power=cos? (4.3)
If a load has a power factor of unit, the load is purely resistive and requires no reactive power.
4.3 THREE PHASE POWER
In single phase systems, active power flow is the product of the voltage, current and the power factor. In a single phase system, the only voltage that can be specified is the voltage from the line to ground. In a three phase system, there are two ways of specifying the voltage. The first one, is the voltage from each phase conductor to ground, termed the phase voltage or the line-to-ground voltage. Second, there is the voltage between any two of the three phases. This voltage is called the line voltage or line-to-line voltage. In power systems, the voltage is usually specified as line voltage, for example 345kV. To calculate the phase voltage, simply divide the line voltage by 3 giving 199kV.

Figure 4.2: Line-to-line and Line-to-ground voltages 47.

Figure 4.2: illustrates the relationship between the phase voltage and the line voltage in a balanced 3-phase circuit. The active power in a three phase system is three times the product of line-to-ground voltage, the current, and the power factor. Since the voltage is usually given as line-to-line voltage, the three phase active power becomes the product of three times the line-to-line voltage, the current, and the power factor all divided by, 3. Mathematically this is represented by the following equations:
P3?=3×VL-G×I×p.f (4.4)
VL-G=VL-L3 (4.5)
P3?=3×VL-L3×I×p.f (4.6)
P3?=3×VL-L×I×p.f (4.7)
Where:
VL-L= Line-to-line voltage
VL-G= Line-to-ground voltage
The reactive power in a three phase circuit is calculated in the same manner. Recall that the power factor is equal to the cosine?. The active power is therefore equal to:
P3?=3×VL-L×I×cos? (4.8)
The formula for three phase reactive power is similar:
Q3?=3×VL-L×I×sin? (4.9)
4.4 DISTORTION AND UNBALANCE IN THREE PHASE SYSTEMS
In a balanced three phase network, the voltage is purely balanced and there is no current that flows in the neutral. This system is considered to be balanced and if the load is purely resistive, the power factor is unity and no losses are incurred. However, introducing inductive and capacitive load brings the concept of reactive power and the system begins to incur some losses as the power factor changes due to phase angle difference that exists in these reactive loads. Another cause of inefficiency is unbalanced load 52 across three phases and this results in an out-of-balance voltage drop, the resultant current returns to the source through the neutral, and there is an increase in the total losses in the supply cables. Non linearity in the voltage and current waveform is also known as a form of distortion and this is normally caused by transient in switching, harmonic generation by power systems equipment such as generators, transformers, motors and non-linear loads. In the conventional power theory, losses caused by distortion, harmonics and unbalance is represented by the power factor which is cos?. According to Gaunt and Malengret, the reduction in the efficiency of the transfer of real power caused by distortion and unbalance can still be described by an efficiency (or power) factor, but the term no longer refers to the displacement angle between the voltage and current vectors. The concept of distorted power and reactive power was first introduced in 1979 53 and this work was carried further developed by Gaunt and Malengret and they introduced the general power theory which extends the power triangle into a three dimensional tetrahedral pyramid introducing distorted power and formulae to calculate power in such conditions.
In power systems, when geomagnetic induced currents are flowing the transformer becomes a source of odd and triplen harmonics and this introduces distortion. Therefore, considering Gaunt and Malengret’s work the distorted power comes into play and there will be an underestimation of the reactive power computed using the conventional power theory. This might create problems to the utility when trying to calculate their reactive power reserves, and the reactive power compensation equipment may fail to deal with large distortions such as those caused by geomagnetic storms.

4.5 GENERAL POWER THEORY
The work performed by Malengret and Gaunt arises from more than twenty years of industrial experience and research. Based on observations and extensive research, they managed to formulate the general power theory. The work is based on the concepts of distortion and unbalance described earlier. The development of a general theory has been driven by needs for solutions to particular power systems problems 54. Most of the time, most power systems operate with sinusoidal, balanced supplies, for which existing definitions of power are adequate. However, at other times distortion, unbalance and dc or zero sequence current components do upset systems and the conventional definitions would give misleading results.

Figure 4.3: Complete power triangle, in which Q = total non-active power, where Qa is the component that can be compensated without energy storage and QA the component that requires energy storage for compensation. S = apparent power without any compensation, Sa the apparent power after compensation without energy storage, and SA the apparent power after complete compensation so that SA=P 52.

Overall the formula used to calculate the apparent power, active power and non-active power using the general power theory (GPT) is given below:
s2=P2+Q2+QA2 (4.10)
Figure 4.3 illustrates the general power theory, and the idea is that under distortion there is another component of reactive power QA that requires energy storage in case of reactive loads. This was not recognized in the conventional power theory and thus the subsequent calculations may be misleading.

Figure 4.4: m-wire system with resistances r1,r2,r3,…,rm, supplying load with voltages
e=e1,e2,e3,…,em, and currents, i=i1,i2,i3,…,im, with active supply current ia and local compensator current ic. 55.

Considering a three phase system with a neutral, where the resistances in each phase are equal r=r1=r2=r3 and the neutral resistance rm is not necessarily the same, shown in Figure 4.4. Applying Kirchhoff’s law to Figure 4.4 yields:
in=-(i1+i2+i3) (4.11)
Where i1,i2,i3, are the phase currents and in is the neutral current. The initial step in deriving the apparent power is to calculate the resistance-weighted square of the currents:
i’2=i12+i22+i32+in2rnr.r (4.12)
Where i’2 is the resistance weighted norm of the current. The resistance-weighted reference for the voltages for all the sample points is then calculated using equation 5.12 and from this, the weighted norm of the instantaneous voltages is calculated using equation 5.13:
eref=e1+e2+e33+rrm (4.13)
V2’2=e1-eref2+e2-eref2+e3-eref2.rrm/r (5.14)
The apparent power is then calculated using the resistance-weighted norms of the voltages and currents:
s=V2’I’ (5.15)
The delivered real power P does not change and may be calculated using the conventional approach or by taking the product of the instantaneous voltages and currents. The total non-active power Q may be calculated using the Pythagorean relationship obtained from the general power pyramid.

4.6 REACTIVE POWER CONSUMPTION OF A 3p-5L TRANSFORMER USING GPT
A comparative analysis of the reactive power consumption of GIC saturated transformers using two different methods of calculating reactive power is given in Figure 4.5. As can be seen reactive power calculated using the GPT is much higher than in conventional methods. As a precaution, utilities may also consider the GPT when planning their reactive power reserves ahead of a GMD event.

Figure 4.5: Reactive power comparison of 3p-5L, 300VA when measured conventionally (blue graph) and according to General Power Theory (brown graph).

Testing for reactive power involved the conventional IEEE method and a GPT. The GPT showed much higher increases in input reactive power as it takes into consideration the distortions in the line experienced under dc cases. At each instant, the GPT computed reactive power was bigger than the conventionally, with the margins widening with increasing dc injected. Increasing dc increases the amount of distortions hence the increase in the difference. The impact of distortions on reactive power is further noticed with the GPT curve being much steeper than the curve from conventional IEEE due to more distortions being present at higher dc injection
4.6.1 POWER FACTOR COMPARISON
Power factor is a good indicator of losses in the power system. The flow of dc in a power system reduces the power factor and hence the performance of the system.

Figure 4.6: Power factor comparison of 3p-5L, 300VA when measured conventionally (blue graph) and according to General Power Theory (brown graph).

The difference exhibited in the power factor in Figure 4.6 is largely due to increased reactive power losses when measured using the general power theory. In both cases the drop in power factor is caused by increased heating, reactive power losses, and increased flow of harmonics when a transformer is subjected to dc current flow. Thus the flow of dc results in poor performance of the system as indicated by power factor measurements.

4.7 CONCLUSION
Conventional methods of reactive power measurement underestimate the reactive power absorbed by the transformer. They neglect distortions and losses in the neutral are neglected, which may have implications of unexpected blackouts that could be avoided by more accurate calculations. Experiments on 3p-5L bench-scale transformers show that there is a wide underestimation of reactive power and power factor measurements when measured conventionally.
CHAPTER 5: THRESHOLDS OF GIC INITIATING DAMAGE IN POWER TRANSFORMERS
5.1 INTRODUCTION
Earlier chapters have described the effects of geomagnetic induced currents on transformers and power system. This chapter will investigate the thresholds of GIC that initiates degradation to power transformers with particular reference to the 3p-5L core structure and brief explanations of the consequences of operating above the stipulated references. Case studies will form the basis of this chapter and the laboratory experiments that I have carried out. A quick look at the standards formed by reputable organizations such as IEEE, IEC, ANSI, NERC and METATECH have guided the researcher to arrive at the conclusions. The chapter addresses the following hypothesis: Thresholds of GICs initiating damage in transformers, based on identifiablemechanisms of degradation, can be determined from the practical records oftransformer degradation leading to relatively early failure, and calculation of theassociated GICs.

5.2 DEFINITION OF DAMAGE
According to this research damage: refers to the noticeable degradation, any form of mal-operation that will cause harm to the operation of the transformer taking into account internationally recognized standards (IEEE, IEC).
5.3 STANDARDS
Temperature thresholds:
Table 5.1: Maximum allowable temperatures on transformers 56
Component GIC type
Base GIC Short duration GIC events
IEEE/IEC IEEE C57.91 IEC 60076
Cellulose insulation 140?C 180?C 160?C
Structural parts 160?C 200?C 180?C
Top oil 110?C 110?C 110?C
Table 5.1 identifies the temperature thresholds under base GIC, which assumes a continuous steady flow of a defined value of GIC. Sources of continuous dc, to power systems may be HVDC links, electric arc convertors. GIC data shows that GIC peak amplitudes may only last for periods between 1 to 5 minutes, therefore short duration GIC events thresholds would apply in this case. The purpose of these recommended temperature limits is to provide reasonable values for the rate of loss of life of the solid insulation used in the transformer and also prevent gas bubbles in the oil.

The ANSI guide specifies that the initial value of the strength of insulating paper is reduced by as much as 50% after 300 h (lifetime) of use at 170°C. At 115?C, the paper has a lifetime of 20,000 h 57
Voltage limits: the operating voltage should be within +/-10%. The transformer is designed to operate at knee point and if the voltages exceed the prescribed limits in the positive direction, the transformer may saturate leading to problems that arise due to saturation described in chapter 2. Both over-voltage and under-voltage damages the insulation of the transformer.

Harmonics: Total harmonic distortion measures
Table 5.2: Percentage Voltage THD, for low voltage transformers 46.

Special applications General system Dedicated system
Voltage THD 3% 5% 10%
For large power transformers the IEEE limit Voltage THD limit is 1.5% 66
5.4 CURRENT THRESHOLD STUDIES
The effects of GIC on power transformers has been researched for decades and it seems that most people seem to agree upon certain behaviors exhibited such as asymmetrical saturation, harmonics generation and their impacts on the entire power system. However, there is a huge variance on the thresholds of GIC that are being proposed by various research groups. For instance, NERC proposed a threshold of 75A/phase while Metatech consider a threshold of 90A/phase. At the same time, study by Q. Qui, R. Girgis et al on a 750 MVA, 765/345/35.5 KV, single phase transformer proposed a standard of 155A/phase after extensive practical experiments. Setting a single threshold value seems not quite a noble idea. Transformer parts responds differently under GIC, for example oil temperature, core and tank temperature vary differently with GIC. Rather specifying that this is a threshold for oil temperature, winding temperature etc. could separate the focus on the studies of GIC thresholds and the standards will speak to a certain kind of degradation. After taking into account all possible mechanisms that could lead to failure in a transformer, then an overall threshold can be set. This study tries to separate these aspects and come up with thresholds for safe levels of temperature, harmonics and reactive power absorption of the transformer. A preliminary study may focus on one structure of transformer and conclusive values may be obtained by considering the different levels of robustness of different core structures.

NERC
According to North American Reliability Council (NERC), voltage stability collapse of power systems exposed to extreme GMD events is of the greatest concern in North America, and the risk of transformer damage is negligible for GICs below 75A/Phase (225 A in the neutral). NERC hence proposed that transformers exposed to such levels of GICs should be assessed for damage, with the assumption that GIC magnitudes below this value offer some minimal effect 49. The later statement meant close examination such as performing dissolved gas analysis must be performed prior to a GMD that exposes to a transformer to such a quasi-dc current.

Table 5.3: Maximum temperatures on single phase transformer tested by NERC 58.

Effective GIC (A/Phase) Metallic Hotspot Temperature (?C) Effective GIC (A/Phase) Metallic Hotspot Temperature (?C)
0 80 140 172
10 106 150 180
20 116 160 187
30 125 170 194
40 132 180 200
50 138 190 208
60 143 200 214
70 147 210 221
75 150 220 224
80 152 230 228
90 156 240 233
100 159 250 239
110 163 260 245
120 165 270 251
130 168 280 257
NERC used single phase transformers to determine the 75A/Phase threshold, but for the purpose of TPL-007-1 is applicable to all types of transformer construction 58. The use of single phase transformers was justified because it is a well-known fact that the single phase transformer is the most susceptible transformer. The 75A/Phase threshold selected represents a 70?C incremental temperature rise from an initial of 80?C and it is well below the 180?C and 200?C short time duration thresholds set by IEC and IEEE respectively.
B. METATECH
The US National academy of sciences noticed inconsistences and uncertainty in GIC events. Citing such inconsistences, they could not give a single value of GIC as thresholds of GIC initiating damage. Instead, the provided two values and these values were 30 A/phase and 90 A/phase DC, where 90 A/phase was passed. The Metatech document 59 made assessments on the number of high voltage transformers that would be more at risk when the two different thresholds are considered, where it was found that the lower GIC level would increase the damage by a factor of two or more 59.

C. C. Q. QUI, R. GIRGIS et al
Qui and Girgis performed tests on a 750 MVA, 765/345/35.5 KV, 1-phase, auto-transformer 60. The tests were aimed at identifying thresholds of GIC that can damage transformers. They found that the test transformer could withstand 155A/Phase for a duration of 30 minutes without the need for reducing their load while limiting the rate of loss of life of insulation to less than 1% and at the same time reducing the risk of forming gas bubbles in the oil. However, transformer tests in this paper confirm that there are significant increases in core noise, core losses, and load losses below 155A/Phase and that the transformer would saturate for values of GIC greater than or equal to 30A/Phase. Similar to NERC, their capability was arrived at taking account of hotspots temperature capability, ignoring that the transformer would have saturated at GIC values slightly above 30A/Phase. As we know, this transformer would be generating harmonics that could cause mal-operation of protective relaying equipment, SVC tripping among other effects. In addition, a saturated transformer draws enormous amounts of reactive power which may lead to voltage instability issues in the network.

5.5 TEMPERATURE THRESHOLDS
In literature it was established that the flow of GIC, cause a considerable increase in transformer temperature. Experimental results presented in chapter 3, also has shown the same trends in the windings, core and the frame for the 3p-5L transformer. The flow of GIC causes half-cycle saturation and this result in increased flux flowing in the core. High flux concentration leads to eddy currents and the resulting increased core losses 61. Stray flux also riches the tank and some parts of the transformer such as flitch plates, tie plates, core bolts and joints leading to increased hotspots in these parts. Oil in the transformer acts as an insulator and coolant. Thus the heat produced in the core, tank and other metallic parts is transferred to the oil by conduction. This causes the oil temperature to rise slowly, and significantly high temperatures will cause partial discharging and gas bubbles to be produced in the oil. This degrades the transformer oil, causes carbonization and some lumps of carbon may be deposited in the oil. Carbon partially conducts electricity and the insulating properties of oil soon drop down. If oil maintenance by filtering or complete recycling is not done the transformer may be damaged permanently.

Practically the thermal GIC capability of a transformer has to consider the maximum allowed temperature in the windings, core and other structural parts. When carrying out these capability tests it is vital to consider the combined effect of AC and DC flowing in the transformer. Therefore the analysis used to reach at some threshold values in this section will consider the thresholds set by IEEE, IEC for temperature in the oil, windings, cellulose or paper insulation and other hotspots. These temperature thresholds are given in IEEE Std C57.91 and IEC 60076 standards for base GIC and short duration GIC events. The IEEE Std C57.91 clearly states, the purpose of these recommended temperature limits is to provide reasonable values for the rate of loss of life of the solid insulation used in the transformer and also prevent gas bubbles in the oil.

It is common practice in industry to set the protection system to operate slightly below these standards to allow for factory errors, deviations in required maintenance programs and to prolong the life of transformers as these are high cost devices.

5.5.1 HOTSPOTS

Figure 5.1: Metallic hot spot temperature for a GIC waveshape derived from the March 1989 GMD event. 62
Simulations performed in 62 show that, the IEEE Std. C57.91 emergency loading hot spot threshold of 200°C for metallic hot spot heating is not exceeded in this example. The peak temperature is 186°C. The manufacturers can provide guidance on individual transformer capability, which are lower than the IEEE threshold. The IEC standard however has set the temperature threshold to 180?C, this would mean that a GIC current of 210A/phase would cause the hotspot temperature to be exceeded for 5 minutes. This is too high a value compared to the threshold set by NERC on hotspots.

However, if a conservative threshold of 160°C were used to account for the age and condition of the transformer, then the full load limits would be exceeded for approximately 22 minutes.

5.5.2 OIL AND TANK TEMPERATURE
Figure 5.2 illustrates the temperature variations of top oil temperature and tank temperature that were taken at Meadow Brook substation after a minor storm in 1992.

Figure 5.2: Observed Meadow Brook transformer hot-spot temperature for a minor storm on May 10, 1992 33.

The results show an insignificant increase in top oil temperature while the tank temperature increased sharply as the GIC current peaks. From this short term duration, one can say GIC have little effect on the oil temperate. However, studies show that metallic hotspots may cause the oil to lose it integrity although the overall temperature of the oil may not rise significantly in a GMD event. Again, as the core saturates due to DC bias, some of the AC core flux will stray outside the core and into the tank creating localized additional losses and heating. However, because of the short duration of the high peak GIC pulses, the increase in temperature has less impact on the overall integrity of the transformer 50.

5.5.3 WINDING HOTSPOTS
An extract of a 60 minutes duration of GIC, from the 1989 geomagnetic disturbance is given in Figure 5.3. The transformers evaluated were the single phase, 250MVA GSU in the Hydro Quebec network. Analysis using real geomagnetic data gives a true representation of what could happen in a real GMD scenario.

Figure 5.3: GIC variation and calculated winding temperature from 1989 GMD event in Canada 50.

It is clearly noticeable that temperature rise of the winding hot-spot, due to the base GIC of 10 A, is negligible 50. As the GIC started to rise beyond 10A, the winding hot-spot temperature started to rise acutely till a temperature of 117?C was reached corresponding to a peak GIC current of 100A. The maximum temperature reached with 100A GIC is way below the temperature IEC and IEEE thresholds of 180?C and 200?C respectively. Despite the winding hot-spot temperature not reaching the threshold, it is interesting to note that the transformer may fail due to the voltage fluctuations that occur as a result of a GIC event of similar nature. In conclusion, such magnitudes of excursions of the winding hot-spot temperatures for short-duration GIC magnitudes should not cause any damage to the windings or any significant loss of life of the winding insulation.

Figure 5.4: Winding temperature rise 750 MVA, 765/345/35.5 KV, 1-phase, auto-transformer 60.

Windings prove to be very robust and unaffected by high levels of GIC even up to 200A/Phase. Figure 5.4 shows that even after 30 minutes of DC bias, the maximum temperature of the winding will reach 121?C. The results were extracted from experiments carried out by Qui and Girgis. Another set of results from Siemens confirm that the maximum temperature would rise to approximately 122?C with a DC bias of 200A/Phase (Figure 5.5)

Figure 5.5: Hotspot temperature on windings, tie plates and clamping plates 63
Figure 5.5 shows that the temperature of the tie plate on single phase passes the IEEE limit at 80A/Phase of GIC while that of 3p-5L passes the limit at 150A/Phase of GIC. Hence looking at these values, the NERC threshold looks reasonable on hotspots and applicable, since it correlates with most studies that takes place independently. For the purposes of general transformer thresholds the worst affected shall be used.

The high-peak magnetizing current pulse, associated with GIC asymmetrical saturation produces correspondingly high levels of leakage flux that is also rich in higher-order harmonics. This leakage flux impinges on the tie-plates causing high localized eddy losses. This component of losses increases approximately linearly with the level of dc current. The combination of these two loss components causes the higher temperatures in the tie-plates.

5.5.4 THERMAL EFFECTS ON FLITCH PLATE
Figure 5.6 illustrates the effect of a 10A per phase dc on a 3p-5L transformer. This causes a 30?C over a 35 minutes duration. Assuming the normal temperature of the flitch plate to be 65?C, the thermal increase would rise to 95?C which is far less than the IEEE and IEC thresholds. Therefore a 10A GIC has not fatal implications on the flitch plate temperature rise.

Figure 5.6: Step thermal response of the Flitch plate of a 400 kV 400 MVA five-leg core-type transformer to a 10 A per phase dc step 64.

Figure 5.7: Asymptotic thermal response of the Flitch plate of a 400 kV 400 MVA five-leg core-type transformer 64.

Figure 5.7 shows temperature gradient of a 3p-5L when GIC is increased, according to simulations performed in 64. 100A would give rise to a temperature rise of 140?C, implying that the temperature rise would reach 205?C, assuming an initial temperature of 65?C used earlier. This shows that 100A of GIC may cause the temperature rise of 3p-5L transformers to exceed the IEEE and IEC standards by 5?C and 25?C respectively. Using the 75A/Phase threshold set by NERC, the temperature would rise to approximately 182?C, thus exceeding the IEC standard by only 2?C and certainly the IEEE standard will not be exceeded. This example would agree with the NERC threshold, if the assumptions are valid to that particular situation.

5.6 REACTIVE POWER CONSUMPTION AND VOLTAGE VARIATIONS
The high-magnetizing currents, resulting from core saturation, will increase the effective reactive power absorbed by the transformer. Consequently, the bulk electric system sees a large increase in the reactive power (VAR) demand for the duration of the GIC flow. Also, the transformer magnetizing current pulse injects significant amounts of even and odd harmonics into the power system to which the transformer is connected. The additional reactive power demand during a GMD event, if not offset by available resources, can cause a reduction in system voltage to the point of encroaching secure system limits 50. In extreme cases, where a severe GMD is coupled with multiple contingencies occurring over a short period of time, the reactive power demand may result in voltage collapse.

Figure 5.8: Fundamental %VAR drawn by a 750MVA single phase autotransformer against GIC, A/Phase 60
Figure 5.8 presents experimental results 60 of reactive power consumption against GIC. The fundamental MVAR is 750MVA which is the transformer rating. The VAR consumption is linear as expected, in my experimental results this was the same for the 300VA, 3p-5L transformer. The additional reactive power consumption is calculated to be 4% of the rated MVA of the transformer for a 50 A/Phase GIC, which is equivalent to 30 MVAR for a 750 MVA transformer. The consumption rose to 120MVAR for a 4 times GIC current increase to 200 amps/phase.

5.6.1 MAGNETIZING CURRENT
In practice the magnetizing current of a power transformer is less than 5 % of full load current, in which case this increase and change in wave shape is relatively insignificant 61. Although the dc bias field itself cannot cause eddy currents the increased magnetizing current due to the dc bias excitation can cause high leakage flux in the transformer clamps and tank leading to high planar eddy currents and localized hot spots in the tank wall 65.

5.7 HARMONICS
A better indication of the severity of the disturbance is the THD level of the voltage 23. Hydro-Quebec utilizes a form of harmonic distortion level measurement by comparing successive voltage peaks to detect an unbalanced voltage. Typically, the THD for voltage is 2.5% on most power systems, but has been measured as high as 30% during severe GIC events. The THD was described in chapter 3 as a measure of the harmonic content of a given voltage or current waveform. IEEE set the THD level, after consulting equipment manufacturers on the level of harmonics that certain loads can tolerate. Apparently, the maximum allowable THD decreases as the voltage of a system becomes high. For systems operating above 500kV the IEEE 519 Std. 46 specifies the THD is 1.5% for 3seconds, while lower voltages may operate at a limit of 5% 58.

Figure 5.9: Voltage total harmonic distortion (VTHD) as a function of GIC for GSU transformers operating at 500kV and above 66.

Studies done by Dong 2001, 35 correlates with R. Walling 2014 22 and they have proved that currents below 10A in neutral, causes the voltage THD to go beyond the IEEE limit for large transformers which is set at 1.5% 66. This shows a major flaw in the NERC, METATECH and Qui and Girgis way of identifying the thresholds. It is clear that harmonics are also essential to consider when trying to come up with thresholds. Extending the graph to extrapolate the real margin of GIC corresponding to a value of 1.5% VTHD for 400MVA, 500kV GSU in Figure 5.9 shows that 2A GIC may cause a considerable amount of harmonics generation in large power transformers. This result correlates well with the result published in 12, 16, 18 that transformers failed in South Africa and Britain after they were exposed to GIC currents of similar magnitudes.
The study of harmonic effects to generators in 67 showed that the generator capability limit can be exceeded at moderate GIC levels, e.g. 50A/phase, and the rotor damage is likely during a severe GMD event. The studies show that negative sequence currents due to harmonics cause heating that may damage the generators. In addition, the study suggests that IEEE standards C50.12 and C50.13 require modifications to take into account the even harmonics of the generator current during a GMD event which is underestimated.

5.8 SATURATION
There is well documented evidence that power transformers may saturate at GIC levels as low as 1-2A 33. Therefore, it is not worthwhile to rely on the thresholds that have been set by NERC, METATECH and Qui and Girgis. The flaws identified earlier that their mechanism of identifying the thresholds did not identify all the mechanisms that would lead to transformer failure as a result of GIC. Saturation is undesirable in transformer operation. Therefore, thresholds that lead to transformer saturation are worthwhile investigating. Post event analysis of a transformer that failed at Grassridge substation in South Africa 18, show that saturation occurred with GIC currents that reached approximately 5A (1 min avg.). Figure 6.8 shows the GIC profile that was measured a three-phase, three-limb, 400/132 kV, 500 MVA at Grassridge substation on 31 March 2001. Sixth harmonic currents were also noticed and these are normally produced by transformers saturated with GIC. This result confirms that a very small GIC flow is adequate to saturate the core.

Figure 5.10: The GIC causing transformer saturation is evident from the 6th harmonic measurement 18.

Moreover, the transformer at Grassridge was a 3p-3L transformer that most researchers say are more robust and less affected by GIC. As is well documented, the presence of even a small amount of GIC (3-4 A per phase or less) will cause half-cycle saturation in a large transformer 33. On 20th October, 1989, the transformer neutral current varied from +5 A to -2 A at Norwich Main in East Anglia, Pembroke in Wales, and Indian Queens in Cornwall for ten minutes. Two identical 400/132 kV, 240 MVA transformers at Norwich Main and Indian Queens failed, the voltage dips on the 400 and 275 kV systems were up to 5%; and very high levels of even harmonic currents were experienced due to transformer saturation by the geomagnetic storms 16. The measurements of GICs and harmonics indicate that saturation occurred more than once in a three-limb three-phase transformer with GICs as low as 2 A/phase 12
5.9 LOSSES
Reference 57, 60, performed losses tests on single phase autotransformers, has shown that transformer losses increases with increase in GIC. The winding temperature increases, with increase in GIC, hence resulting in some losses. The temperature increase seem to affect the winding insulation more than losses would impact the system. Studies by Gaunt and Koen suggests that, “insulation failure is considered to be the main cause of damage to large power transformers in South Africa” 12. These failures occurred soon after GIC events, and harmonic analysis on these transformers showed an increase in even and odd harmonics that are present in GIC attacked transformers. Hence losses, are not worth considering when coming up with GIC thresholds initiating damage.

5.9.1 CORE LOSSES
Another consequence of the unidirectional flux density shift in the core is that significant increases in core losses are experienced for the duration of the GIC pulse. The increase in core losses results in an increase in the core hot-spot temperature. However, the core’s thermal time constant is typically much longer (30 min to 45 min) than the duration of the high peak GIC pulses and correspondingly only small increases in the core temperatures would be experienced 50. It has been estimated that around 5% of all electricity generated is dissipated as core losses, and the cost of no-load core losses in transformers was estimated to be £110 million in the UK alone during 1987/1988 68
5.10 CONCLUSION
The chapter examined case studies of the effects of geomagnetically induced currents to transformers with the aim of getting thresholds of GIC that initiates damage in transformers. Three separate researchers have identified different thresholds, although their assessment criteria was the same. The range of thresholds from these sources lies between 75A/Phase and 155A/Phase. In this research, it was identified that the assessment criteria used by these researches missed critical detail that leads to damage in transformers. Saturation and harmonic effects were not taken into account. Bringing in harmonics and saturation, suggests that the real safe margin could lie between (1 to 5)A/Phase GIC, since GIC around 3A/Phase causes saturation and the 1.5% Vthd to be exceeded, and transformers also saturate within low values of GIC (1 to 5)A/Phase. There is enough evidence from measurements and simulations that shows transformers saturating at low values of GIC. Among these is the saturation of the transformer at Grassridge substation and transformers in United Kingdom that saturated and harmonics measured show an increase in triplen harmonics that are present in GIC events. Empirical evidence from real GIC events, summarized in section
REFERENCES
P. R. Price, “Geomagnetically Induced Current Effects on Transformers,” IEEE Transactions on Power Delivery, 2002.

J. Koen, C. T. Gaunt, “Geomagnetically Induced Currents in the Southern African Electricity Transmission Network,” IEEE Bologna Power Tech Conference, Italy, 2003.

S. MacMillan, “Earth’s Magnetic Field in Geophysics and Geochemistry,” 2004.

N. Homeir, L. Wei, “Solar Storm Risk to the North American Grid,” Atmospheric and Environmental Research, 2013.

R. Pirjola, “Geomagnetically Induced Currents During Solar Storms,” IEEE Transaction on Plasma Science, December 2000
K. P. Arun Babu, “Coronal Mass Ejection from the Sun,” PhD Thesis, Indian Institute of Science and Research, Pune, 2014.

E. Matandirotya, “Measurement and Modelling of Geomagnetically Induced Currents in power lines, Doctor of Technology, Cape Peninsula University of Technology, 2015.

L. Bolduc, “GIC Observations and Studies in the Hydro-Quebec Power System,” Journal of Atmospheric and Solar-Terrestrial Physics 64, 2002.

R. Baker, R. Balstad, J. M. Bodeau, E. Cemron, J. F. Fennell, and G. M. Fisher, “Severe space weather events-understanding societal and economic impacts,” Report, 2008.

R. Thorberg, “Risk analysis of geomagnetically induced currents in power systems,” Report, 2012.

J. A. Marusek, “Solar Storm Threat Analysis,” 2007.

C. T. Gaunt, G. Coetzee, “Transformer failures in regions incorrectly considered to have low GIC-risk,” IEEE Powertech, 2007.

N. B. Trivedi, I. Vitorello, W. Kabata, S. L. G. Dutra, A. L. Padilha, et al, “Geomagnetically induced currents in an electric power transmission system at low latitudes in Brazil: a case study,” Space Weather, 2007.

C. S. Barbosa, G. A. Hartmann, and K. J. Pinheiro, “Numerical modeling of geomagnetically induced currents in a Brazilian transmission line,” Adv. Space Res., , 2015.

J. Koen, C. T. Gaunt, “Preliminary Investigation of GICs in the Eskom Network,” Report to Eskom, University of Cape Town, December 1999.

R. Zhang, Transformer modelling and influential parameters identification for geomagnetic events,” PhD Thesis, University of Manchestor, 2012.

J. Koen, “Geomagnetically Induced Currents and its Presence in the Eskom Transmission Network,” MSc Thesis, University of Capetown, 2000.

J. Koen & C. T. Gaunt, “Disturbances in the Southern African Power Network due to Geomagnetically Induced Currents”, Paris, 2002.

N. Takasu, F. Miyawaki, S. Saito, Y. Fujiwara, “An Experimental Analysis of DC Excitation of Transformers by Geomagnetically Induced Currents,” IEEE Transactions on Power Delivery, 1994.

L. Bolduc, A. Gaudreau, A. Dutil, “Saturation Time of Transformers under DC Excitation,” Electric Power Systems Research, 2000.

NERC Disturbance Analysis Working Group, “March 13, 1989 Geomagnetic Disturbance”.

R. Walling, “Analysis of Geomagnetic Disturbance (GMD) Related Harmonics”, EPRI, Palo Alto, CA: 2014. 3002002985
T. S. Molinski, “Why Utilities Respect Geomagnetically Induced Currents,” Journal of Atmospheric and Solar-Terrestrial Physics, 2002.

R. Girgis, K. Vedante, “Methodology for evaluating impacts of GIC and capability of transformer design,” IEEE Transaction on power delivery, 2013.

R. Girgis, K. O. Chung-Duck, “Calculation Techniques and Results of Effects of GIC Currents as Applied to Two Large Power Transformers,” IEEE Transactions on Power Delivery, April 1992.

P hurlet, F. Berthereau, “Impact of Geomagnetic Induced Currents on Power Transformer design,” JST Transformateurs, France, 2007.

J. Berge, R. K. Varma and L. Marti, “Laboratory Validation of the Relationship Induced Current,” in IEEE Electrical Power and Energy Conference, 2011.

T. Ngnegueu, F. Marketos, F. Devaux, T. Xu, R. Bardsley, S. Barker, J. Baldauf, J. Oliveira, “Behaviour of transformers under DC/GIC excitation: Phenomenon, Impact on design/design evaluation process and modelling aspects in support of Design,” CIGRE, 2012.
R. Girgis and K. Vedante, “Effects of GIC on Power Transformers and Power Systems,” Transmission and Distribution Conference and Exposition (T&D), 2012.

J. Kappenman, “Geomagnetic storms and their impact on power systems,” IEEE Power Engineering Review, May 1996.

J. Gilbert, J. Kappenman, W. Radasky, E. Savage, “The Late-Time (E3) High-Altitude Electromagnetic Pulse (HEMP) and Its Impact on the U.S. Power Grid,” Metatech Corporation, January 2010.

D. J. Fallon, P. M. Balma, W. J. McNutt, “The Destructive Effects of Geomagnetic Induced Currents in Power Transformers,” Doble Clients Conference, 1990.

L. L. Grisby, Electric Power Generation, Transmission, and Distribution,” 3rd Edition. 2016
J. G. Kappenman, V. D. Albertson, “Bracing for the Geomagnetic Storms,” IEEE Spectrum, March 1990.

X. Dong, Y. Liu, J. G. Kappenman, “Comparative Analysis of Exciting Current Harmonics and Reactive Power Consumption from GIC Saturated Transformers,” IEEE Power Engineering Society Winter Meeting, 2001.

K. DiZheng, Z. Yun, “Analysis and Processing of the Impact of DC Power transmission Grounding Current to the Network Equipment”, Automation of Electric Power Systems, 2005.

I. Y. Zois, “Solar Activity and Transformer Failures in the Greek National Electric Grid”, EDP Sciences, 2013.

X. N. Pan and X. L. Yu, “Discussion on Abnormal Noise of Transformer,” EDP Sciences, 2006.

C. Liu, R. Pirjola, “Geomagnetically Induced Currents in the High-Voltage Power Grid in China”, IEEE Transaction on Power Delivery, October 2009.

Q. Lin and Y. F. Gao, “On the October–November 2003 giant storms,” Seismol. Geomagn. Observation Res., November 2006.

B. Roen, Geomagnetic Induced Current Effects on Power Transformers, Master’s Thesis, Norwegian University of Science and Technology (NTNU), 2016.

L. Y. Liu and X. W. Xie, “Analysis of increase of noise of 500 kV transformer,” High Voltage Eng., 2005
D. Z. Kuai, C. M. Liu, and D. Wan, “Experiment and Research of the Influence of Direct-Current Magnetic Bias on Transformer,” Jiangsu Elect. Eng., 2004.

H. K. Chisepo, “The Response of Transformers to Geomagnetically Induced-like Currents,” MSc Dissertation, University of Cape Town, 2014.

J. Yao, M. Liu, C. Li and Q. Li, “Harmonics and Reactive Power of Power,” Power and Energy Engineering Conference (APPEEC), 2010, Asia-Pacific, IEEE, 2010.

IEEE Std. 519-1992, “IEEE Recommended Practices and Requirements for Harmonic Control in Electrical Power Systems, New York, NY: IEEE, 1992.

M. Terbrueggen, “EPRI Power Systems Dynamics Tutorial”, EPRI, Palo, 2002.

R. R. Jethani, H. Naidu, “A Novel Method to Analyse the Effects of Geomagnetic Induced Current on Transformer,” International Journal for Innovative Research in Science & Technology, 2016.

49. C. T. Gaunt, “Notice of Proposed Rulemaking on Reliability Standard for Transmission System for Transmission System Planned Performance for Geomagnetic Disturbance Events,” 2015.

J. Verner, G. Hoffman, W. Bartley, R. Ahuja, J. Arteaga et.al., “IEEE Guide for Establishing Transformer Capability while under Geomagnetic Disturbance”, September 2015.

51. S. Guillon, P. Toner, L. Gibson, D. Boteler, “The Havoc Caused by Auroral Electrojet Generated Magnetic Field Variations in 1989,” IEEE Power & Energy Magazine, October 2016.

C. T. Gaunt and M. Malengret, “Why we use the term non-active power, and how it can be measured under non-ideal power supply conditions,” IEEE PES PowerAfrica, Johannesburg, 2012.

International Electrotechnical Commission. Technical Committee No. 25, Working Group 7, Report: “Reactive power and distortion power”, No. 25, December, 1979
C. T. Gaunt, M. Malengret, “True power factor metering for m-wire systems with distortion, unbalance and direct current components,” Electric Power Systems Research 95, 2013.
M. Malengret, C. T. Gaunt, “General theory of instantaneous power for multi-phase systems with distortion, unbalance and direct current components,” Electric Power Systems Research 81, 2011.
IEEE Std. C57.91-1995, “IEEE Guide for loading mineral-oil-immersed transformers and step-voltage regulators”, 1995.

L. Bolduc, A. Dutil, V. Q. Pham, “Study of the acceptable DC current limit in core form power transformers,” IEEE Transactions on Power Delivery, January 1997.

L. Marti, “Assessment of the effects of GIC in power systems”, FERC Meeting, 2016
J. G. Kappenman, Geomagnetic Storms and Their Impacts on the U.S Power Grid, Goleta, Carlifonia: Metatech Corporation, 2010.

Q. Qiu, D. R. Ball and J. A. Fleeman, R. Girgis and K. Vedante, “Effect of GIC and GIC Capability of EHV Power Transformers – A Case Study on an AEP 765 kV Power Transformer Design,” CIGRE US National Committee: Grid of the Future Symposium, 2013.

P. Marketos, A. J. Moses, J. P. Hall, “Effect of DC Voltage on AC Magnetization of Transformer Core Steel,” Journal of Electrical engineering, 2010.

NERC, “Transformer Thermal Impact Assessment white paper,” Developed by the Project 2013-03 (Geomagnetic Disturbance) standard drafting team.

Siemens, “GIC Effects on Power Transformers,” NERC Taskforce Meeting”, November 2013.

L. Matti, J. Elovaara, “GIC Occurrences and GIC test for 400 kV System Transformer,” IEEE Transactions on Power Delivery, April 2002.

E. Mulasalihovic, “Effects of Geomagnetically Induced Currents on the Magnetic Performance of Transformer Cores, JMMM, 2008.

A. Vitols, F. Faxvog, “GIC Neutral Blocking System,” IEEE Meeting in Augusta, Maine, July 2015.

R. Zareand, L. Marti, “Generator Thermal Stress During a GMD”, IEEE PES, Vancouver, Canada, July 2013
M. Daniels, “Power from designer domains,” Physics World, 1988.

CHAPTER 5: LABORATORY PROTOCOL AND SIMULATION PROTOCOL
5.1 INTRODUCTION
This chapter is dedicated to developing a laboratory protocol to rigorously investigate the response of a 3p-5L transformer to geomagnetically induced currents. Preliminary tests will be conducted on 120/230V, 3p-5L, and 300VA bench-scale transformers. Further tests will be carried on 120/230V, 3p-5L, 15kVA transformers, a larger transformer may closely resemble the response of large power transformers. A relatively large power transformers may help in determining the thresholds of GIC initiating degradation in power transformers.
There are two laboratory environments that will be used in this investigation:
(a) A bench-scale test system involving transformers of differing core structures (900VA, 300VA – 120/230V). This is the platform on which the testing protocol is to be developed and implemented. Most aspects of the investigations in this study will be addressed in this environment.

(b) Based on (a), the developed methodology will then be implemented on a three phase medium scale test system (48kVA). The purpose of this procedure is to test the applicability of the developed protocol on higher capacity transformers. The transformers will be subjected to carefully selected levels of dc according to the protocol, and the results will be used to analyse their response.

5.2 PURPOSE OF TESTS
These tests are designed to determine the response of 3p-5L to geomagnetic induced currents. In particular the following aspects shall be examined, output reactive power , voltage profile at load end, harmonics generated by the transformer conducting gic, temperature measurements, leakage flux measurement, input and output active power and determining the saturation characteristics of transformers. All these tests shall be used to determine the thresholds of geomagnetic induced currents that may affect the transformer partially or permanently. Certain standards regarding have been set up by IEEE, IEC regarding the harmonic content and voltage operating limits of transformers. It was
5.3 EXPERIMENTAL SETUP
In practice geomagnetic induced currents flow in the earth and enter the transformer neutral via grounded neutrals of transformers and they flow through transmission lines to the next substation and out again through grounded neutrals of star vector group transformers. The consequences of geomagnetic storms elaborated on the power system emanate from the transformer as discussed in the literature review. A complete replica of the real scenario would be to connect to transformers as given below:

Figure 5.1: Test system for transformer response to geomagnetically induced currents (GIC).
5.4 TEST EQUIPMENT
In order to carry out the tests effectively, without damaging the transformer and to extract reliable results, the following equipment was used for this project:
DC Source: The dc supply circuit consisted of a 12 V, 7.2Ah, rechargeable lead acid battery and a variable resistor to change dc current values.

Switch: 1? resistance switch
Power Meter: A high precision and wide bandwidth, IEC76-1(l976) compliant Yokogawa WT1800 digital power meter was used for reactive power, voltage, current, and harmonics measurements.

Load: A three-phase resistor bank load of 35 VA per phase and 100?.

Temperature Measurement: A temperature gun (Sentry ST642) that uses infrared technology was used, with an accuracy of ±2°C for temperatures ranging between -20 and 100°C.
Source transformer: 120/230V, 900VA, 3p-5L
Transformer under test: 230/120V, 300VA, 3p-5L
Variable load supply: A 3 phase (0-400V) variable three VARIAC.

5.5 TEST PROTOCOL
This laboratory protocol is designed to determine the step by step procedure to be followed when carrying out the intended tests and to determine the safe magnitude of dc injections that are going to be used without damaging the transformer.
5.5.1 PROCEDURE
The test shall comprise of a source transformer and a transformer under test. The selection of the source transformer was such that the source transformer shall be significantly larger than the load transformer so that the dc injection levels based on the load transformer characteristics would have a negligible effect on its magnetization characteristics. The transformer nameplate ratings were as follows:
Source transformer: 120/400V, 900VA, 3p-3L or 3p-5L
Transformer under test (TuT): 120/230V, 300VA, 3p-5L.

Steps:
Initially the harmonics in the supply voltage shall be checked for compliance with IEEE Std. 579-19 which states that, harmonic content must not exceed a total harmonic distortion (THD) of 5%.
Check if the supply voltage is balanced, otherwise unbalanced sources results in distortions that may inherently cause the transformer reactive power to increase. This may distort the reactive power analysis of this research project.

Properly connect the transformer step-up as in Figure 3.1 and ensure that all connections are tight. Loose connections may cause heating of connections and burn the insulation.

Determine the short circuit and open circuit parameters of the transformer. These parameters were determined by Hilary Chisepo in his thesis for the 3p-5L bench-scale transformer.

Carry out open circuit tests on the load transformer and draw the magnetization curve.
Determine the magnetization current from the magnetization curve. It is the current that corresponds to the knee point voltage.
Use the magnetization current to calculate the per unit values of gic currents that must be injected.

5.5.2 DETERMINING DC VALUES TO INJECT
The following steps are taken to calculate the values of dc to be injected in the neutral as indicated in the experiment setup.

Determine the value of the magnetization current Imag, from the magnetization curve of the transformer.

Calculate the load current to DC current ratio KLD as given below:
KLD=IrIm p.u 5.1
where Ir , is the rated line current and Im is the magnetization current.

Calculate the DC current in per unit as:
Ipu=IdcImag 5.2
Idc coupled with AC current should not exceed the rated current.

Inject the DC current into the neutral such that satisfies the inequality:
1?Ipu?KLD 5.3
5.6 TRANSFORMER ACCEPTANCE TESTS
The preliminary tests were done with bench-scale transformers that have been used before and extensive research with larger transformers shall be carried out. Before carrying out any tests on these transformers, it is necessary to perform acceptance tests to determine any damages due to transportation, improper connections and compliance with specifications.

TABLE 5.1 3p-5L, 380/380V, and 15kVA design data
Transformer Rating 22.79A
Number of turns 140/140
Number of layers 4
Core length (mm) 810.0
Core width (mm) 69.4
Core weight/kg 90
5.6.1 WINDING RESISTANCE
Winding resistance is the resistance of a length of copper conductor from one end to the other, is related to the parameters of the conductor as:
R=?LA 5.4
where:
R is resistance in Ohms,
L is the effective length of the conductor and
A is the cross sectional area of the conductor.
The winding resistance are determined as part of the acceptance tests to determine the presence of short circuits, open circuits and poor connections. It is also important to determine the winding resistance, to check the balance of resistance in the phases. Unbalance in resistances create a potential difference between the phases and in such cases the conventional methods of calculating power deceiving readings. That is another justification of using two method of calculating power in this thesis.

Table: Results
HV (inner winding) LV (outer winding)
R 0.11? 0.15?
Y 0.11? 0.15?
B 0.11? 0.15?
5.6.2 EXCITATION CURRENT TEST
This test is conducted to determine the operating point or knee-point of the transformer. It is essential to know this point so as to limit the input voltage so that the transformer does not operate in saturation region as this will affect the validity of our results. Excitation current (no-load current) is the current that flows in any winding used to excite the transformer when all other windings are open-circuited. It is generally expressed in percent of the rated current of the winding in which it is measured
5.6.3 RATIO TEST
Power transformer turns ratio test is an AC low voltage test which determines the ratio of the high voltage winding to low voltage windings at no-load. The turn ratio is determined by using a VARIAC to excite the transformer and measure the input and output voltage using a power meter. The following formulae is used to calculate the theoretical turn ratio:
Theoretical turn ratio=H.V winding voltageL.V winding ratio 5.5
Calculate the measured turn ratio and find the deviation as follows:
Deviation=Measured turn ratio-Expected turn ratio*100Expected turn ratio 5.6
The turns ratio tolerance should be within 0.5% of the nameplate specifications for all windings, as indicated in IEEE Std. C57.12.00-1993.
HV (inner winding) LV (outer winding)
R 0.11? 0.15?
Y 0.11? 0.15?
B 0.11? 0.15?
5.7 TRANSFORMER EQUIVALENT CIRCUIT
The Steinmetz ‘exact’ transformer equivalent circuit shown in Figure 5.2 is often used to represent the transformer at supply frequencies C. R. Paul, S. A. Nasar, L. E. Unnewehr, ‘Introduction to Electrical Engineering’, McGraw-Hill, Inc., Singapore, 1986. Each component of the equivalent circuit can be determined by carrying out the open circuit and short circuit test.

Figure 5.2: The Steinmetz ‘exact’ transformer equivalent circuit, referred to the primary side C. B. Simon and P. S. Bodger, ‘Power transformer design using magnetic circuit theory and finite element analysis – a comparison of techniques’, AUPEC 2007, Perth, Western Australia, 9-12 December, 2007.

Where:
Xm is the magnetizing reactance,
Rc is the core resistances,
R1,R2 are series winding resistances
X1,X2 are the leakage reactance
5.7.1 SHORT CIRCUIT TEST
The load losses and transformer impedance are obtained from this test. The load losses are the losses caused by the phase currents in the transformer windings. Losses are proportional to the square of the current (or current density) and directly proportional to the winding resistance. The LV side of the transformer is short circuited and the wattmeter (W), the voltmeter (V) and the ammeter (A) are connected to the HV side of the transformer as shown in Figure 5.3.

Figure 5.3: Circuit diagram for the short circuit test of a three phase transformer.

Equipment required:
A variable 3 phase supply capable of delivering at least 20A.
A power analyser, or three watt-meters.

Voltage and current measuring devices
The test procedure:
Connect the circuit as in Figure 5.3
Increase the voltage of the VARIAC from zero until the ammeter reads full rated current.

At rated current record the values of voltage (Vsc), current Isc and power (Psc).

The computation of the transformer impedance is as follows. The short-circuit voltage is expressed by the formula:
Vsc=ZscIsc 5.7
Where Zsc represents the equivalent impedance of the transformer, referred to the input side. This impedance is composed of the equivalent resistance and the leakage inductance of the windings.

The short-circuit power and current can be related as:
Psc=Isc2Rsc, and therefore; Rsc=PscIsc2 5.8
Since the series impedance (equivalent impedance) can be given by:
Zsc=Req2+Xeq2, and Zsc=VscIsc 5.9
Since Z and R can be calculated then the reactance is obtained from:
Xeq=Zsc2-Req2 5.10
5.7.2 OPEN CIRCUIT TEST
No load losses can be measured from the L.V side using an adjustable 3-phase voltage source with neutral. It can be derived from mains or a D.G set. The voltage and frequency should steady and at rated values and as near as possible to 50Hz and it should be measured. This test can give a basic value near rated conditions if all precautions are taken.
The L.V side is energized at the rated voltage and power is measured by 3 single-phase wattmeters or 1 3-phase 4 wire single phase wattmeter/energy meter. Connections are made as given in the diagram below:

Figure 5.4: Circuit diagram for the short circuit test of a three phase transformer.

Equipment required:
A variable 3 phase supply capable of delivering at least 20A.
A power analyser, or three watt-meters.

Voltage and current measuring devices
The test procedure:
Connect the circuit as in Figure 5.4
Increase the voltage on the VARIAC from zero to full rated voltage.

At rated voltage, record the voltage (Voc) current (Ioc ) and power (Poc).
Mathematical computation of the core resistance Rc and the magnetizing reactance Xm is obtained from this test. The open circuit power (no load power) is given by:
Poc=VocIoccos?oc 5.11
The open-circuit power factor and power factor angle can be determined:
cos?oc=PocVocIoc or ?oc=cos-1PocVocIoc 5.12
Since the no load current Ioc will correspond to the magnetic branch current which constitutes two current components, Im and Ic:
Im=Ioccos?oc, and Ic=Iocsin?oc 5.13
The equivalent parameters Xm and Rc can then be determined as:
Xm=VocIm, and Rc=VocIc 5.14
Table 5.2 Equivalent circuit parameters
3p-5L, 380/380V, 15kVA Transformer
Imag Rc (?) 1976.01
Xm (?) 846.75
R 0.02833p.u
X 0.04213p.u
3p-5L, 120/230V, 300VA bench-scale
Imag 74mA
R 0.0154p.u
X 0.0049p.u
No load losses 0.0208p.u
5.8 TRANSFORMER RESPONSE INVESTIGATION TESTING
5.8.1 FLUX MAPPING
The magnetic field around the transformer core changes drastically when GIC currents are flowing within the transformer. In addition, leakage flux increases to the extent of flowing in the region where it is not supposed to such as the tank, tie plates and bolts, causing extensive hotspots in these regions. Thus, it is necessary to monitor that flux. The method employed in this thesis is the use of search coils. A search coil is a device that makes use of electromagnetic induction for measuring the strength of a varying magnetic field, for instance a wire made into a coil and tied around the core. Suppose a current I flows in a coil of N turns, a voltage E is induced in that coil, according to Faraday’s law. The transformer equation can be used to determine the relationship between the voltage measured and the flux.

E=4.44?mfN 5.15
Where: E is the induced voltage
N , is the number of turns
f , is the frequency
?m , is the peak flux in the core.

The flux in the coil related to the flux density, Bm and core area, Ac as:
?m=BmAc 5.16
Substituting 5.14 in 5.13 yields:
E=4.44Bm?mfN 5.17
When a search coil is placed inside a changing magnetic field perpendicular to the coil, a varying e.m.f. will be induced across the ends of the coil. From equation 5.17 it is evident that the maximum induced e.m.f. is proportional to the maximum field strength. Therefore, by measuring the amplitude of the induced e.m.f. using a multimeter, the magnetic field strength can be obtained. Search coils of two turns shall be used, on the 3p-5L transformer as shown in Figure below:

Figure 5.5 showing search coils employed around the limbs and yoke.

The search coil use 0.5mm diameter insulated wire conductor. The designed volts per turn e.m.f. of the cores is 1.5671V. So the search coils should read 3.134V at rated voltage (380V). This voltage corresponds to the designed flux of 1.7T of the transformer.

5.8.2 REACTIVE POWER AND HARMONICS MEASUREMENT
Voltages and currents were measured by an IEC76-1 (1976) compliant Yokogawa WT1800 Digital Power Meter. The meter is capable of performing online measurements and also has a facility whereby the instantaneous values of the voltage and current waveforms can be recorded and stored for post processing. Fast Fourier Transforms are done up to the 10th harmonic (500 Hz). Samples can be taken over two and a half cycles within a resolution of 1002 readings (20.04 kHz), more than satisfying the Nyquist criterion. The neutral current could easily be calculated by application of Kirchhoff’s laws during post processing. The image of the Yokogawa power meter is shown below.

Figure 5.6: IEC76-1(l976) compliant Yokogawa power meter for measuring power, harmonics, voltage and currents.
5.8.3 TEMPERATURE MEASUREMENT
Hotspot temperature increase is also a phenomenon to be investigated in this project. Temperature data acquisition will make use of the TC-08 temperature data logger. Thermocouples are connected to the data logger. The data logger is then connected to a computer via a USB port (no external power required).

Figure 5.7: TC-08 temperature data logger
The TC-08 thermocouple data logger is designed to measure a wide range of temperatures using any thermocouple that has a miniature thermocouple connector. Numerous thermocouples are supported, allowing an effective temperature range of –270 to +1820?C. It has eight channels, for measuring temperature and an additional built in Cold Junction Compensation (CJC) circuit can also be used as a 9th channel to measure room temperature.

SIMULATION PROTOCOL
Finite Element Modelling
Finite element modelling (FEM) is a powerful tool to model electromagnetic devices such as the transformer. It involves the creation of a geometry and the object or geometry is broken down into small elements i.e. finite elements. These elements are then represented by a set of equations, typically Maxwell’s equations. The practical application often known as finite element analysis (FEA), is a numerical technique for finding approximate solutions of partial differential equations (PDE) as well as of integral equations. The performance of electromagnetic equipment is governed by these Maxwell equations. A transformer is an electromagnetic device and in order to predict its electromagnetic field, FEM software compute the Maxwell’s equations and provide the solution as visual field lines. Maxwell equations can be given in differential form or alternatively in integral form. The software that is going to be used in this project is ANSYS Maxwell. It solves the electromagnetic field problems by solving Maxwell’s equations in a finite region of space with appropriate boundary conditions and user-specified initial conditions in order to obtain a solution with guaranteed uniqueness. The differential form of Maxwell’s equations are written as:
?×H=J+?D?t (4.1)
?×E=-?Bdt (4.2)
??B=0 (4.3)
??D=?
Where B and H are the magnetic flux density and the magnetic field intensity respectively. D and E are the electric flux density and electric field intensity respectively. J is the current density and ? is the resistivity of the material.
Electromagnetic analysis
Analytical techniques
Boundary elements
Numerical techniques
Differential equations
Integral equations
Closed form
Finite difference
Finite elements
Iterative
2D Magnetostatic 2D Eddy 2D Transient
Components of H-field
Scalar potentials
Vector potentials
2D Electrostatic 2D/3D Thermal 3DElectrostatic
3D Magnetostatic 3D Eddy 3D Transient
BEM
FDM
FEM
Electromagnetic analysis
Analytical techniques
Boundary elements
Numerical techniques
Differential equations
Integral equations
Closed form
Finite difference
Finite elements
Iterative
2D Magnetostatic 2D Eddy 2D Transient
Components of H-field
Scalar potentials
Vector potentials
2D Electrostatic 2D/3D Thermal 3DElectrostatic
3D Magnetostatic 3D Eddy 3D Transient
BEM
FDM
FEM

Figure 5.8: Flowchart of general FEM method 2, Ansys Manual, 2013
Figure 5.8 represents the methods of computing electromagnetic problems. Finite element method is a numerical technique for finding approximate solutions to boundary value problem. FEM can be used for solving differential equations in many disciplines like, electromagnetics, magneto statics, thermal conduction, structural mechanics, transient, fluid dynamics and acoustic D. Emad. “A simplified iron loss model for laminated magnetic cores”, IEEE Transactions on Magnetics 44.11 (2008):3169-3172.. The finite difference method (FDM) is another numerical technique frequently used to obtain approximate solutions of problems governed by differential equations. The finite difference method is based on the definition of the derivative of a function f(x) that is D. A Hutton, “Fundamentals of Finite Element Analysis”, McGraw-Hill Companies, 2004:
df(x)dx=lim?x?0fx+?x-f(x)?x 1.2
where: x is the independent variable. In the finite difference method, as its name implies, derivatives are calculated via Equation 1.2 using small, but finite, values of ?x to obtain:
df(x)dx?lim?x?0fx+?x-f(x)?x 1.3
From differential equation theory, we know that the solution of a first-order differential equation contains one constant of integration. The constant of integration must be determined such that one given condition (a boundary condition or initial condition) is satisfied. The reason why, these boundary conditions are set when analyzing electromagnetic problems using either FEM or FDM.

The most descriptive way to contrast the two methods is to note that the FDM models the differential equation(s) of the problem and uses numerical integration to obtain the solution at discrete points. On the other hand, FEM models the entire domain of the problem and uses known physical principles to develop algebraic equations describing the approximate solutions. Thus, the finite difference method models differential equations while the finite element method can be said to more closely model the physical problem at hand.

BASIC STEPS IN FINITE ELEMENT METHOD
In general, FEM uses specific formulae and mathematical models of the different problems under study and the basic steps for solving them are: creating the geometry, generating mesh, validation and retrieving the solution and post process. Finite element method can be divided into three steps: pre-processing, solution setup (processing) and post-processing. Pre-processing is building a finite element model and generating a mesh; processing uses the related equations and iterative algorithm to obtain results; post-processing is the collection and the processing of results.

Pre-processing (Material and boundary conditions application)
Solution setup (Evaluation of vector potential)
Post-processing (Evaluation of results)
Pre-processing (Material and boundary conditions application)
Solution setup (Evaluation of vector potential)
Post-processing (Evaluation of results)

Figure 5.9: A general representation of simulation steps in finite element method.

PREPROCESSING
This is the most critical step that defines the model. It includes:
Defining the geometric domain of the problem.

Defining the solution type.

Assigning the material properties of the elements.

Defining the element connectivity (mesh the model).

Define the physical constraints (boundary conditions).

Assigning other variables that defines the problem e.g. equipment rating.

SOLUTION
During the solution setup phase, finite element software assembles the governing algebraic equations in matrix form and computes the unknown values of the primary field variable(s). The computed values are then used by back substitution to compute additional, derived variables. Figure 5.10 illustrates the flowchart of the algorithm used in finite element analysis of a power transformer.

Generate initial mesh
Computes field
Perform error analysis
Start field solution
Refine mesh
Solution OK?
Stop field solution
No
Yes
Generate initial mesh
Computes field
Perform error analysis
Start field solution
Refine mesh
Solution OK?
Stop field solution
No
Yes

Figure 5.10: The flowchart of the algorithm for finite element analysis K. Mukesh, et al. “Study of stray losses reduction through Finite Element Method”, Annual IEEE India Conference (INDICON), 2013.

POSTPROCESSING
Analysis and evaluation of the solution results is referred to as post-processing. Postprocessor software contains sophisticated routines used for sorting, printing, and plotting selected results from a finite element solution. Examples of operations that can be accomplished include:
Core loss measurement
Flux analysis
Thermal analysis
Excitation characteristics, etcetera
While solution data can be manipulated by many ways in post-processing, the most important objective is to apply sound engineering judgment in determining whether the solution results are physically reasonable. Therefore, it is important to compare these results with some practically measured results.

DETAILED STEPS IN FEM MODELLING

Klaus-Jürgen Bathe, “Finite Element Procedures”, Prentice Hall, Pearson Education, Inc, 2nd edition 2016. Pp 2-4.

DEFINE THE SOLUTION TYPE
There are four main categories of problems that FEM can solve. Hence, the first step is to define the solver that you wish to use. A solver is simply a type of problem that you want to analyze and these are:
Magnetostatic – it solves static magnetic fields caused by DC currents and permanent magnets. Can solve both linear and non-linear materials.

Eddy current solver – it solves sinusoidally-varying magnetic fields in frequency domain. It is a full wave solver that considers displacement currents. Induced fields such as skin and proximity effects are also considered.

Transient magnetic – it solves transient magnetic fields caused by time-varying or moving electrical sources and permanent magnets in linear or non-linear materials. Induced fields such as skin and proximity effects are considered as well.

Electrostatic solver – it solves static electric fields in linear materials.

CREATING A GEOMETRY
Based on real transformer dimensions and geometry the FEM models have been constructed for 2D simulation of low frequency transient electromagnetic fields.

Figure: 3p-5L transformer 2D model in FEM
Specify Excitations – create windings and coils
Assigning coils

Adding a winding

Vpeak*(1-exp(-50*time))*cos(2*pi*60*time)
Vpeak*(1-exp(-50*time))*cos(2*pi*60*time+(2/3*pi))
Vpeak*(1-exp(-50*time))*cos(2*pi*60*time+(4/3*pi))
ASSIGN MATERIAL TO CORE AND WINDINGS
B-H curve of material
Magnetic core is characterized with B-H curve of magnetization and thin laminations. The designed magnetic flux density of the 3p-5L transformer is 1.703T and the B-H curve of the core material (M-5 grade steel) that has been used in FEM simulations is shown below. ANSYS Maxwell software allows the user to add their own material, and input the B-H curve and core loss data of the transformer.

Figure: M-5 Material B-H curve.

The winding are made of enamel covered copper wire. The enamel insulation is rated as temperature Class F, i.e. 180ºC. But for longevity sake, temperature rise is limited to 150ºC. The conductor used is 3.55mm diameter drawn ETP copper. The copper cross-sectional area is 9.8976mm². The design current density is 2.303W/kg.

ASSIGN MESH OPERATIONS
The process of representing a physical domain with finite elements is referred to as meshing, and the resulting set of elements is known as the finite element mesh D. A Hutton, “Fundamentals of Finite Element Analysis”, McGraw-Hill Companies, 2004. In the transient solvers, there is no automatic adaptive meshing. Therefore, the user must either link the mesh from an identical model solved using the magnetostatic and eddy current solvers, or alternatively a manual mesh must be created. In this project, a mesh is created manually using “inside selection” to create elements throughout the volume of the transformer.

Figure: A representation of meshing applied to a 3p-5L transformer.

CREATING THE BOUNDARY REGION
The boundary conditions are the specified values of the field variables (or related variables such as derivatives) on the boundaries of the field. Depending on the type of physical problem being analyzed, the field variables may include physical displacement, temperature, heat flux, and fluid velocity to name only a few. In the case of electromagnetic analysis, the boundary condition will be limiting the region in which the electric field extends to. This is done by first defining the region and assigning the vector potential of the boundary region to zero or assigning it a “balloon”- which is a non-conducting region.
MODEL VALIDATION

Run the software to compute results
CHAPTER 6: PRESENTATION OF RESULTS
Graphs
Transformers Used: Bench-scale
Source Transformer: 900VA, 120/230V (phase voltages)
Transformer under Test: 300VA, 230/120V (phase voltages)
Excitation curve

Knee point: 74mA
Phase transformation linearity (TuT)
From open circuit test,

Magnetizing Current Increase with DC bias

Reactive power consumption

Non active power

Reactive power using GPT and conventional comparison

S, P, Q input variation with DC bias

S, P, Q output variation with DC bias
Load end voltage with load

Load end voltage (No Load)
Power factor measured conventionally

Power factor when measured using GPT

Comparison of power factor using GPT and conventional

Larger transformer results

CHAPTER 7: CONCLUSION
This chapter summarises the findings of this research and highlights possible areas that can be improved to enhance understanding. Furthermore, the two major effects of GIC events; increased reactive power absorption and harmonic currents were thoroughly investigated. Two methods of measuring power were implemented and the findings were analysed in chapter 6 and chapter 3. Literature survey played a phenominal part in answering the research questions posed in chapter 1, and in validating some parts of the first and second hypothesis. An overview of all these aspects shall be given in the subsections of this chapter.

ACHIEVEMENT OF OBJECTIVES
This project has sought to investigate the effects of GIC on 3p-5L power transformers. A comprehensive review of literature was conducted providing a summary of the effect that direct current has on power transformers. The saturation phenomenon has been fully explored and the theory behind its occurrence elaborated upon. Experiments and simulations were rigorously done to investigate the response of 3p-5L to geomagnetic induced currents and the results were presented in chapter 6. Investigations on the thresholds of GIC initiating damage was done, and the research identified key areas that were neglected in similar previous researches. The results obtained in the key areas are summarized below.

REACTIVE POWER
Laboratory tests on reactive power trends show a linear relationship between Geomagnetically Induced Current flowing through a transformer’s windings and the reactive power absorbed by that transformer’s core. Many other researchers H. K. Chisepo, “The response of transformers to geomagnetically induced-like currents, “MSc. Eng Dissertation, University of Cape Town, 2014; R. Siti, S. Hassan, and M. Anuar, “Study the harmonic characteristics of DC bias on the single phase power transformer,” in Power Engineering and Optimization Conference (PEDCO) Melaka, Malaysia, 2012 Ieee International, 2012, pp. 501-504; A. Lotfi, H. K. Hoidalen, and N. Chiesa, “Effect of DC biasing in 3-legged 3-phase transformers taking detailed model of off-core. path into account”, Electric Power Systems Research, vol. 138, pp. 18-24, 2016; J. Yao, M. Liu, C. Li, and Q. Li, “Harmonics and reactive power of power transformers with DC bias”, in Power and Energy Engineering Conference (APPEEC), 2010 Asia-Pacific, 2010, pp. 1-4 achieved the same results experimentally. Further analysis on reactive power was conducted using the IEEE conventional way and the general power theory yielded different results. Conventional methods of reactive power measurement underestimate the reactive power absorbed by the transformer. These methods neglect distortions and losses in the neutral. This underestimation of consumed reactive power may have implications of unexpected blackouts, which can be avoided by using a more accurate method; the general power theory.
Harmonic analysis
A number of tests were conducted to determine the harmonic performance of a transformer operating under load conditions and exposed to a secondary direct current component. Tests showed increases in secondary voltage distortion beyond the 5 % THD limit and 3 % individual harmonic limit as defined by the IEEE519 standard. The level at which the standard was exceeded was in excess of 100 % of the transformer’s magnetising current.

Winding temperature
The winding temperature increases at a faster rate compared to the core limb due to increased line losses under dc, indicating that more heat is generated in windings due to line losses compared to heat from eddy current losses. Eddy currents are lower due to the very low frequency of GIC. PT100 will be used for further tests on larger transformers.

Voltage drop
Load end voltage was seen to decrease with increasing dc current injected in the neutral, this may result in power swings on the system.

THRESHOLDS OF GICs INITIATING DAMAGE IN TRANSFORMERS
Large organisations such as NERC, and METATECH are also looking into the area of thresholds. It has been highlighted in chapter 4 that, harmonics and reactive power increase in transformers conducting GICs are critical in finding these thresholds. These cause a lot of instability on the grid and pose threats to the transformer under GIC events. Literature has shown that, IEEE thresholds of harmonics have been surpassed with lower levels of GIC than threshold proposed by NERC and METATECH. Empirical evidence from real GIC events summarized in section 4.8 and 4.9 show that transformers have failed well below the proposed thresholds.

Work needs to be done to establish working limits for TDD in transformers. Once these limits are established they can be represented as reactive power absorption levels for those transformers. The limits can be used to inform system operation to protect transformers from damage in the case of a GIC event.

ANSWERS TO RESEARCH QUESTIONS
The following research questions have been set up to assess the project hypothesis:
How does reactive power increase in transformers saturated by the flow of GIC affect power system stability?
What is the role of installing GIC monitoring devices in order to fully understand the phenomenon behind the risk of quasi-dc current to transformers?
How does different structure of transformers affect their response to GIC?
What are the different levels of GIC that cause noticeable degradation in power transformers?
How does the reactive power consumed by a power transformer vary with respect to GIC?
What is the implication of general power theory in determining reactive power absorbed by the transformer as opposed to conventional methods of calculating power?
VALIDITY OF HYPOTHESIS
In the beginning of the thesis two hypothesis were formulated, and these were:
H2a Tests on model transformers and extension of the results to power transformerswith suitable transformer equivalent circuit and FEM simulations will improve theconventional models of the reactive power requirement in transformers conductingGICs.

H2b Thresholds of GICs initiating damage in transformers, based on identifiablemechanisms of degradation, can be determined from the practical records oftransformer degradation leading to relatively early failure, and calculation of theassociated GICs.

Thorough research carried out showed that hypothesis H2a was valid. Practical tests and FEM simulations on benchscale trasformers were conducted. Application of general power theory to measure reactive power on these transformers yielded different results from conventional methods of measuring reactive power. As elaborated earlier, GPT is a better method of measuring non-active power. Thus, conventional models of the reactive power requirement in transformers conducting GICs can be improved by applying GPT. However, tests on larger transformers that closely resembles the response of utility power transformers shall be done and the results extended to utility power transformers using transformer equivalent circuits. Extension of the results could not be done using bench-scale transformers due to the difference in construction in terms of joints, laminations and the quality of stacking.

Appendix
Experimental Results: 3p-5L, 300VA, 120/230V