86

4. PERFORMANCE ANALYSIS

Analysis of Problem

The problems identified in preceding chapter of system development are

analyzed by using following methods

1. Analysis using programs and softwares based on mathematical background

developed in System Development

2. Building simulation model and obtaining simulation results

3. Computer Analysis of Recorded Events

4. Developing a scientific hypothesis

5. Creation of new idea based on scientific hypothesis

6. Developing Logic based on new idea

4.1. Harmonic Analysis

The power quality disturbances mainly comprise harmonics which are

detrimental for equipment and plant life. The foremost step carried out was load

survey and determining plant powers and harmonics of different orders. Power quality

analyzer, softwares for analysis make the job of harmonic analysis easier.

4.2 Experimental Analysis

Findings in harmonic analysis-Surge Suppressor failure in EAF-2 and

Transformer failure in EAF-1.The EAF-1 has 42/36 KV, 50 kA surge arrestors (LA-

1), which never failed until now, while EAF-2 has 36/30 kV, 10 KA surge arrestors

(LA-2), which failed few times. Most surge arrestors fail when it cannot handle the

quantity of incoming energy. LA-1 with 50 KA capacities can handle much more

energy in Joules. In addition, its line-to-line voltage rating of 42 kV is away from

normal operating range of 33 kV ±10 %, avoiding unintended operation.

LA-2 with 10 kA capacity can handle much lesser energy in Joules. In

addition, its line to line voltage rating of 36 kV is within normal operating range of 33

kV± 10%, increasing possibilities of unattended operation during which huge energy

87

could flow through LA-2, placed between line to line and then between line to earth.

Such unintended operation if happens, continues until the LA itself fails clearing the

first fault, but could trigger another voltage related fault) or the upstream breaker trips

(breaker tripping takes long time typically 70- 100 ms or 500 ms (if secondary relay

acts) which is a long time for any LA, stressing and weakening it. Please also ensure

that the installed LA-2 is of station Class-3 grade and have the potential to handle

higher energy. It is better to replace LA-2 with 10 kA station class-3 (in India 50 kA

is generally not available, station class-3 with appropriate voltage grade can handle

more energy in Joules), 42 kV line-to-line grade (to avoid unintended operation) and

36 kV line to earth voltage grade.

Transformer and/or its component failure in EAF-1

Going by the nature of failures it is unlikely to have triggered by the passing

lightning surges. The more likely causes could 1) lie within transformer or its

components itself or 2) due to momentary high stress.

4.2.1 Harmonic Distortion

Non-Linear Load, the cause of harmonic distortion

Superimposition of harmonics to fundamental waveform causes distortion. Its

magnitude varies depending upon the superimposing harmonics (its magnitude,

frequency & phase angle) with reference to fundamental waveform. When a 3rd

harmonic coincides with fundamental, it distorts the fundamental like a “Square

Wave”. If the same 3rd

harmonic coincides in 900

phase shift, it distorts the

fundamental waveform into a series of “positive and negative spikes’. Odd harmonics

generate square wave, it clip sine wave peak and cause load saturation. A

phenomenon when an increase in voltage does not cause proportional increase in

current. This is why such “Non-Linear “Loads distort current sine wave.

Harmonic can cause both current and supply voltage distortion. Harmonic current

distortion is generally local to a particular facility, while harmonic voltage distortion

pervades through entire electrical system and affects many different localities. This is

88

because harmonically distorted load current from any facility flows through its source,

as it offers least impedance path.

Energy Loss and Harmonic Distortions

Harmonic distortion cause high-energy losses. This is because equipment such

as Transformers, Motors, generators, cables and other inductive loads are designed to

operate at fundamental (50 Hz) frequency. Losses increase since harmonic currents

are of higher frequencies.

Transformer lossesconsist of excitation i.e. no load losses and load losses. No

load losses are the sum of eddy current and hysteresis losses. Load losses include

D.C. winding loss; eddy-current stray loss in the winding (with skin effect&

proximity effects) and other stray losses due to flux linking the tank and structure).

Losses due to harmonics are more pronounced in eddy current losses since it increases

with square of harmonic frequency. Stray losses as electromagnetic flux in windings,

core and shields increase with harmonic distortion.

Motor lossesconsist of iron and copper losses increase with harmonic

distortions. Motor efficiency could reduce by 5 %, depending on the extent of

distortion of the wave shape.

Cable/Line losses increase as the square of harmonic current and its distance.

Additionally harmonic injected losses further compounded due to skin effect arising

due to its high frequency currents and voltages. Turbo-Alternator (TG) or DG loses

are more pronounced for the industry due to phenomenon of negative sequence

harmonics. Just imagine the damaging effect on your TG running at a high speed,

when subjected to this force from opposing direction. Depending upon distortions,

you lose out up to 30 % power burn excess fuel and end up in lowering TG life and

higher maintenance cost.

There are recommended limits, IEEE-519, on the magnitude of harmonics that

a facility could have for optimized operation.

89

4.2.2 Measurement on 33 kV Side Section II, EAF-1 Bus

MDH Germany make, 33 kV, AC furnace is fed from 33 kV bus section II of

switchyard. It was observed that at 9-55 AM, Fourth harmonic filter was switched off.

After 5 minutes, Third harmonics filter was switched off. Thus both filters were

switched off for 15 minutes. The effect of switching off capacitor banks is studied. It

was observed that the frequency increases from 50.1 Hz to 51.4 Hz and current

decreases from 0.7 kA to 0.5 kA.

Figure 4.2.2.1: Photograph of Set Up at 33 kV EAF Bus

The photograph 1 shows the set up used for measurement and data analysis at

33 kV bus bars at section 2 and section 3 for electric arc furnaces 1 and 2 respectively.

For EAF-1, it was noticed that the harmonics were maximum at the beginning of the

heat cycle, which approximately comes after one hour. At the initiation of melting and

during melting of cold scrap, it was found that the voltage and current waveforms

were severely distorted. The harmonics are generated. The current harmonics of third,

fourth and fifth order exceeds 3%. The current is resuming its sinusoidal nature after

5-8 minutes. The current refers to charging and meltdown of cold scrap. The flicker

level is also very high. Figure 4.2.2.1 shows the sample waveform with severe

distortion in each phase. The frequency recorded was 49.92 Hz. The values of RMS

voltages for three phases and currents were noted.

90

4.2.3 Harmonic Audit

Table 4.1: Plant Power at various Bus and Harmonic Levels

Date/Load Typica

l

kVA

PF-

Avg

+Lag

-Lead

Avg

Freq.

Typical

Amps

R-Y-B

Typical

Volts

R-Y-B

Major

harmonics

-

Currents

Major

harmonics

-

Voltage 12th Jan

220 kV

MSEB I/C

89,500 1.00 49.30 247

220 230

221.2

221.2 219.1

IO2=21.0%

IO3=09.0% IO4=02.0%

VO2=00.6%

VO3=00.5% VO7=00.4%

14th

Jan

33 KV Trs 1-

MILLS

27,800 1.00 49.70 495

476

482

32.9

33.2

32.9

IO5=03.2%

IO3=03.2%

IO7=02.7%

VO5=00.8%

VO3=00.7%

VO7=00.6%

14th

Jan

33 KVTrs.2

EAF-1

43,700 0.99 49.98 806

733

812

32.7

32.9

32.7

IO3=01.0%

IO7=00.4%

IO5=00.3%

VO5=00.4%

VO3=00.3%

VO7=00.2%

13th Jan

33 KV Trs-3

EAF-2

43,793 1.00 49.29 794

775

800

32.1

32.3

32.1

IO3=16.5%

IO2=05.0%

IO5=03.0%

VO5=03.5%

VO3=01.0%

VO7=00.5%

12th Jan

33 kV HRM

RM

6,380 1.00 49.59 110

112

104

32.9

33.1

33.0

I11=03.3%

I13=02.0%

I23=00.5%

VO5=00.5%

VO7=00.3%

VO3=00.2%

12th Jan

33 kV

HRM

FM

12,020 1.00 49.45 216

201

205

33.2

33.3

33.0

I11=08.0%

I13=06.0%

I23=02.5%

VO5=0.5%

VO7=0.3%

VO3=0.2%

12th Jan

33 KV CRM-1

5,910 0.90 49.94 104

100

102

33.2

33.5

33.2

I05=09.0%

I11=03.0%

I13=02.8%

VO5=0.8%

VO3=0.30%

VO7=0.20%

12th Jan

33 KV CRM-2

5,780 0.80 49.04 102

099

100

33.1

33.4

33.2

I11=06.%

IO5=05.5%

I13=03.8%

VO5=0.8%

VO3=0.3%

VO7=0.2%

14th Jan

11KV HRM

FM entry

1,352 0.021 49.87 70

67

68

11.3

11.4

11.3

IO5=21.0%

IO7=13.0%

I11=09.0%

VO5=1.8%

V13=0.62%

V11=0.55%

13th Jan

11 kVCRM-1

Reel

1,420 0.14 50.10 77

73

75

10.9

10.9

11.0

IO5=46.0%

IO7=12.0%

I11=09.0%

VO5=1.10%

V13=0.82%

V11=0.70%

13th Jan

11 kV CRM-1

Backup

2,370 0.62 49.34 127

121

123

11.1

11.0

11.0

I11=08.2%

IO5=04.5%

I13=02.0%

V11=2.00%

VO3=1.80%

VO5=1.00%

14th Jan

11 kV CRM-2

Entry

1,780 0.20 49.58 96

91

97

11.0

11.0

11.0

I11=08.0%

I13=0.3.0%

IO3=02.0%

V11=1.10%

V13=1.00%

VO3=0.80%

14th Jan

11 kV CRM-2

Top

1,210 0.63 50.11 62

61

62

11.3

11.4

11.3

I11=07.5%

I13=06.0%

I05=04.0%

V11=1.10%

VO5=0.60%

VO3=0.50%

91

4.3 Computational Analysis of Furnace

The furnace analysis is done by FFT analysis and the software used for isPowerflow.

All harmonic trends, flicker,T HD, Current distortion, individual current distortion,

waveforms of current, voltage ,and voltage sag are simulated. The details of all events

are recorded. The analysis of the same is done.

Figure 4.3.1: Voltage as a Function of Time for EAF-1

Voltage waveform is distorted for all three phases due to superimposition of Second

order, Third order, Twenty sixth and Twenty seventh order harmonics predominantly

on fundamental wave. This phenomenon is at the start of heat cycle. When cold scrap

is charged and arc strikes, the current waveform is severely affected.

Figure 4.3.2: Percent of Current Average Fundamental as a Function of Time

92

Figure 4.3.2 shows one heat duration of EAF-1 from 8-27 AM to 9-42 Am. At

8-29 AM, after arc striking the current distortions were noted. The values of Current

average fundamental for various harmonic orders were as follows. IH3= 13.8%; IH5=

6.5%; IH7= 3%; IH9= 1.6%; IH11= 1.8%. The similar results were obtained for next

charging. At the instant of striking of arc and beginning of melting, recorded values

were IH3= 11.18%; IH5= 7.0%; IH7= 3.4%; IH9= 3.4%; IH11=3.1%.

Figure 4.3.3: THD Voltage for EAF-1

At 1-59 AM, THD V1= 20.9%, THD V2=17.1%, THD V3=15.4%; Corresponding

I1=0.51 kA, I2=0.52 kA, I3=0.72kA. After one heat cycle, which is about 70 minutes

duration at 3-08 AM, THD V1= 20.4%, THD V2=17.8%, THD V3=17.1%;

Corresponding I1=0.54 kA, I2=0.45 kA, I3=0.53 kA. Total Harmonic distortion is

higher than power quality standards. Also there is unbalance in current for three

phases. The non-uniform peaks in voltage indicate erratic behavior of arc (Figure 4.4).

Figure 4.3.4: THD Voltage and THD Current

93

Figure 4.3.5: Percent of Voltage Fundamental as a Function of Harmonic

Voltage harmonics of Second order, Third, Twenty sixth and Twenty seventh

are more than 3%. Voltage harmonics are produced in this way through impedance of

the mains supply; the amplitude depends on the value of the harmonic currents

supplied and the effective impedances at the harmonic frequencies.

Because of the non-linear arc resistance, an arc furnace acts as a source of

current harmonics of the order Second, Third, Fourth, Fifth, Sixth, Ninth; their

individual magnitude is more than 3%. All mentioned harmonics exceed the limits

consisting of capacitor C with reactive power, and mains inductance from the mains

short circuit power. Another harmonic current producing source is the energization of

scrap metal and ladle furnace transformers.

Figure 4.3.6: Percent of Current Fundamental as a Function of Harmonic

94

The dynamic inrush current waveform associated with transformer energizing

operation includes both even and odd harmonics, which decay with time until the

transformer magnetizing current reaches steady state (Figure 4.3.7). The most

predominant harmonics during transformer energization are 2nd

, 3rd

, 4th

, and 5th

in

descending order.

Figure 4.3.7: Voltage When Voltage Sag Lasted For 6.5 Cycles

The figure 4.8 shows voltage sag phenomenon, which lasted for 6.5 cycles. It

is observed that oscillatory transients exist at the top and bottom peaks. The

oscillations are multi-frequency. Main circuit oscillations are induced by one or more

arc voltage discontinuities and having energy sources in capacitances of cables,

lumped inductances of the supply side and load side networks. They often cause

chopping of voltage waveform at peaks due to switching in of major capacitor banks,

unloaded line and cable transients, which lasted for several hundred milliseconds [3].

Current waveform during voltage sag is severely distorted. It contains double

frequency transients. Voltage sags are followed by high currents, shown in green

which in turn cause voltage drop at PCC. Asynchronous periodic current waveform is

observed.

Figure 4.3.8 shows that short term flicker severity index (Pst) is more than 1,

which exceeds IEC standards. Also long term Perception (Plt) is more than 0.74 [5].

Both these values are greater than standards. The voltage flickers are predominant at

the beginning of heat cycle when arc strikes and melting starts.

95

Figure 4.3.8: Flicker of EAF-1

From figure 4.3.9, it is noticed that supply frequency to arc furnace is not

constant but varying randomly because of nonlinear behavior of arc and load is not

constant.

Figure 4.3.9: Frequency Variation for EAF-1

The variations in frequency clearly indicate that load is continuously varying

and erratic behavior of arc current is responsible factor.

96

4.4 Comparison between Experimental Analysis and Computational Analysis for

Arc Furnace Harmonics

Arc Furnace 1 and Arc Furnace 2 are analyzed for Harmonic levels at 33 kV Bus.

When fourth Harmonics filter and third Harmonic filter were switched off for 15

minutes. The effect of switching capacitor banks was also studied.

Table No.4.2 : Effect of Switching Capacitor Bank on Harmonics Levels

Experimental Analysis Computational Analysis

Filter Switched

Off

Frequency

49.96 Hz

Current

Harmonics

I03 = 12%

I02 = 4%

I05 = 1.5%

I07 not

recorded

Frequency

49.97Hz

Current

Harmonics

R-Y-B

I03 = 14%

I02 = 6%

I05 = 2%

I07 = 3%

Filter Switched

On

50.13 Hz Current

Harmonics

I03 = 1%

I05 = 3%

I07 = 0.4%

Frequency

49.99Hz

Current

Harmonics

R-Y-B

I03 = 5%

I02 = 3%

I05 = 1.5 %

I07 = .5%

It is observed that, current is reduced and frequency increases as the filter is made on

after switching of for 15 minutes.

4.5 Justification of Error of Arc Furnace for Harmonic Analysis

Ø The instability and non-linearity are greatest during melting down of cold

scrap. The delay and erratic process of striking the arc and resulting gaps in

the current are conspicuous.

Ø As melting down progresses, the striking becomes more stable, but the current

can still contain low-frequency fluctuations. The temperature and heat of the

97

arc are high with a liquid steel bath, and the thermal conduction is low. So that

the arc characteristics begin to approach the linear behavior of an ordinary

resistance.

Ø The non-sinusoidal arc voltage waveform is not entirely adopted by the supply

current. Since inductance in the circuit gives rise to inductive reactance, which

increases with frequency and therefore resists harmonics. The current

waveform then becomes almost sinusoidal. The noticeable harmonic current

causes increase in reactive power and with a higher reactance.

Ø Flicker measured for short duration and long duration are more than the limits

specified in IEEE Standards. Voltage sags on one of the lines has produced

voltage and current unbalances.

4.6 Experimental Analysis of Harmonics in Rolling Mills

220 kV from Supply Company

220 kV/ 34.5 kV

EAF-1

220 kV/ 34.5 kV

EAF-2

EAF-1

FMRM

HR M

In House CRM

11 kV/750 V

4 Motors

11kV/440V

3 motors Thyristor

transforme

33 kV/ 11 kV

Figure 4.6.1 : Single Line Diagram of Lloyd Steel Plant, Bhugaon, Wardha

In this survey, power quality is monitored and the recorded waveforms give

information about many steady state characteristics, including harmonic distortion,

power factor, flicker, frequency, voltage and current harmonics. Because of the

periodic nature, harmonic distortion is analyzed by using Fourier analysis. The

harmonic levels are very high as compare to IEEE standards as shown in table.

98

Table 4.3: Harmonic Levels in Rolling Mills at Lloyd Steel Plant, Bhugaon

Date/Load Typica

l

kVA

PF-

Avg

+Lag

-Lead

Typical

Amps

R-Y-B

Typical

Volts

R-Y-B

Major

harmonics

-

Currents

Major

harmonics-

Voltage

14th Jan 33 KV Trs

1MILLS

27,800 1.00 495

476

482

32.9

33.2

32.9

IO5=03.2%

IO3=03.2%

IO7=02.7%

VO5=00.8%

VO3=00.7%

VO7=00.6%

12th Jan

33 KV HRM

RM

6,380 1.00 110

112

104

32.9

33.1

33.0

I11=03.3%

I13=02.0%

I23=00.5%

VO5=00.5%

VO7=00.3%

VO3=00.2%

12th Jan

33 KV HRM

FM

12,020 1.00 216

201

205

33.2

33.3

33.0

I11=08.0%

I13=06.0%

I23=02.5%

VO5=0.5%

VO7=0.3%

VO3=0.2%

12th Jan

33 KV CRM-

1

5,910 0.90 104

100

102

33.2

33.5

33.2

I05=09.0%

I11=03.0%

I13=02.8%

VO5=0.8%

VO3=0.30%

VO7=00.20%

12th Jan 33 KV CRM-2

5,780 0.80 102

099 100

33.1

33.4 33.2

I11=06.%

IO5=05.5% I13=03.8%

VO5=0.8%

VO3=0.3% VO7=0.2%

14th Jan

11KV HRM

FM entry

1,352 0.021 70

67

68

11.3

11.4

11.3

IO5=21.0%

IO7=13.0%

I11=09.0%

VO5=1.8%

V13=0.62%

V11=0.55%

13th Jan

11 KVCRM-

1

Reel

1,420 0.14 77

73

75

10.9

10.9

11.0

IO5=46.0%

IO7=12.0%

I11=09.0%

VO5=1.10%

V13=0.82%

V11=0.70%

13th Jan 11 kV CRM-1

Backup

2,370 0.62 127

121

123

11.1

11.0

11.0

I11=08.2%

IO5=04.5%

I13=02.0%

V11=2.00%

VO3=1.80%

VO5=1.00%

14th Jan 11 kV CRM-2

Entry

1,780 0.20 96

91

97

11.0

11.0

11.0

I11=08.0%

I13=0.3.0%

IO3=02.0%

V11=1.10%

V13=1.00%

VO3=0.80%

14th Jan 11 kV CRM-2

Top

1,210 0.63 62

61

62

11.3

11.4

11.3

I11=07.5%

I13=06.0%

I05=04.0%

V11=1.10%

VO5=0.60%

VO3=0.50%

The reason for all round harmonic distortion in many steel plants is often due to

numerous drives. It produces harmonic at load end, which flows upstream as Xd” of

utility grid or DG/TG offers least impedance path to the flow of harmonic current,

causing electrical stress to plant equipment in its path, which are usually dissipated as

energy losses.

In-addition use of capacitor further aggravates harmonic problem. Capacitors

are used for power factor improvement and voltage control. Unfortunately, with

addition of each new capacitor bank, electrical systems resonance frequency is

99

lowered, rendering it to be susceptible to resonance. Thus, though capacitor

themselves do not generate harmonics, but its use with non-linear loads magnify

harmonics to a level as determined by system parameters. PQ disturbance arises

mostly from industry itself. However, end-use equipment is more sensitive to PQ

disturbances. Thus industry not only suffers first but solution also lies within industry

itself. Both Current and voltage harmonic are high at LT level, i.e. AC input to

respective thyristor converter drives, providing scope for improvement and energy

savings by re-construction of the basic wave-shape. In addition to harmonic reduction,

power quality improves, thereby providing stabilization effect enhancing equipment

life. Higher voltage & current harmonic distortion causes stress on cables, bus bars,

alternators, motors, plant electronics and transformers. It reduces their efficiency and

life and increases maintenance and energy cost. This is because such inductive loads

are designed to operate only at fundamental, but get subjected to high frequency

harmonic environment.

The harmonic currents the steel sheet that feed into the rolling mill process

varies over the total period. The harmonics produced by the 12 pulse converters are

related to the pulse number of the drive and can be represented as harmonic current

sources as their characteristics frequencies. For the ideal case of instantaneous

commutation between conducting elements, the generated ac side harmonic currents

and the theoretical magnitudes for 12 pulse drives can be calculated as

n = 12k ± 1

where n-order of the harmonic, k-an integer 1,2,3,…, In - harmonic current magnitude,

I1- fundamental current magnitude. The large dc drives use thyristor converters to

control bus voltage or line current to maintain desired speed. The disadvantage of

thyristor converter is a greater ac current distortion during slower motor speeds.

4.7 Computational Analysis of Rolling Mills

Steel Plants with electric arc furnaces or Arc Plants are generally plagued with power

quality issues out of the harmonic distortions of fundamental wave shape. The

problem arises due to extensive use of non linear loads mainly EAF and

100

ThyristorizedDC driving and holding stage and decays rapidly through melts. The

harmonics generated are like transient phenomenon which rises sharply during initial

arcing stage and decays rapidly through melting and holding stages. Usually EAF

employ harmonic filter to mitigate harmonics and to control voltage flicker.

Rolling mill loads are fluctuating and the effect are observed. The dynamic

distortion current waveform caused by voltage sags includes both even and odd

harmonics which decay with time until the input current reaches steady state. The

characteristic of a input current is fundamental current with a percentage of THD. The

values recorded at 1.35 pm are THD I1 = 113.3%, THD I 2= 101.5%, THD I3 = 90.9%.

These values are very high as compared to standards. After a period of 9 minutes

approximately ( light load period) , load is increased with the introduction of new

sheet for rolling, the current shoots up and total harmonic current distortion now

shoots up to THD I1 = 111% , THD I 2=65.45%, THD I3 = 141.6%; the similar

readings are obtained.

Figure 4.7.1 CRM Backup Drive THD Current Variation with Time

Total harmonic distortion at 1.35 pm is THD I1= 113.3%, THD I2= 101.5%, THD

I3= 90.9%, the lowest value of THD at 1.42 pm are THD I1= 13.1%, THD I2= 12.8%,

THD I3= 11.9%. Both these values exceed the permissible limits of 5%. Third

harmonics coincides in 90 degree phase shift; therefore the fundamental waveform is

distorted into a series of positive and negative spikes.

Figure 4.7.2 shows the variation of power factor with time for back up drive in

cold rolling mill. The power factor is lowest as 0.025 and highest as unity power

101

factor. The value of power factor fairly steady for long time is he above 0.7. The wide

variation in power factor indicates the nonlinear and random load which is occurring

on the mill. The power factor is corrected by capacitor bank installed in the plant.

Figure 4.7.2 Power Factor Variation with Time – Back Up Drive

Figure 4.7.3 gives the variation of percent of current fundamental for phase 1

(I1) with time. The harmonics of order Three, fifth, seventh and ninth and eleventh are

predominant. Third order harmonics is within limits; around 3%. But fifth order

harmonics are 17.8%, 12.3%, and 11.4% at different hours of time. The seventh order

harmonics was observed to be maximum 5.8% which is higher value for individual

phase current. The higher values of 5.6 %, 7.4% and 8.1% are also recorded for

eleventh order harmonics at 1.46 pm, 2.25 pm and 2.01 pm respectively. This

indicates the need of filter for eleventh order harmonics. The three peaks correspond

to fifth order harmonics.

Figure 4.7.3:Percent of Current Fundamental Variation with Time

Figure 4.7.4 gives Fast Fourier Transform analysis gives the values of current

fundamental for phase 1,2 and 3 as follows: For predominant harmonics of 11th

order

102

the values are FFT I1 = 7.3% , FFT I2 = 7.3%, FFT I3 = 7.1%; for 13th

order the values

are FFT I1 = 6.9%, FFT I2 = 6.7%, FFT I3 = 6.3%; for 5th

order the values are FFT

I1 = 3.6% , FFT I2 = 3.5%, FFT I3 = 3.1%.The 25th

order harmonics are less than 3%.

Figure 4.7.4: Current Harmonics Variation with Time - 11 kV CRM Back Up Drive

The distortions in shape of the waveform are clearly observed. The waveform

for voltage contains notches nearer to zero crossing and near to peak of the voltage.

The commutation notches are caused by the short circuits caused in the system when

six PulseBridge or twelve PulseBridge is commutated. A six pulse gives six notches

per cycle and twelve pulse bridges gives twelve pulses per cycle. At the bus bar to

which the commutating bridge is connected the line to line voltage is reduced to zero

during the commutation. At other bus bars between the bridge and driving voltage,

the depth of notches is determined by ratio of inductances in the circuit. There are

three important factors related to notching. The first one is more than one notch nearer

to zero crossing. Unfortunately these crossovers are used either to detect frequency or

to give a reference for firing angle. There have been reports of loss of control of

voltage on the plant when there was significant voltage distortion. The solution to this

problem is to add a small filter, just a capacitor in shunt with low voltage input to the

control circuit. The second significant feature oscillations caused by Heaviside

excitation of the system inductances and capacitances. The capacitances are generally

those of either system cabling, power factor correction capacitors or filters. These

oscillations interfere with telecommunications either telephones or signal carrying.

The third significant factor is effect of firing angle of the bridge when it is near to the

103

peak of the supply voltage. The high rate of change of voltage dv/dt associated with

beginning and end of commutation have an effect on insulation lifetime .

Figure 4.7.5:Voltage Variation with Time – 11 kV CRM Entry Drive

33 kV Cold Rolling Mill Current and Voltage Wave Form

Figure 4.7.6: Current and Voltage Wave Form - Cold Rolling Mill

Figures 4.7.7 and 4.7.8 shows for back up drive, the harmonics of 3rd,

5th

, 11th

,

and 15th are more than 3%. The ripple content of harmonics distorts the current

waveform to the greatest extent. Odd harmonics generate square wave, it clips sine

wave peak and cause load saturation. If system is sharply resonant at one of the

predominant harmonic frequencies, the harmonic current will excite the system

causing voltage distortion and resonant, thus producing more harmonic current to

further excite the resonant system. This interaction between the resonant system and

current producing that effect from voltage sags can produce very high values of

104

rmsand peak voltage which can degrade and damage equipment and lead to

equipment failure.

Figure 4.7.7:Current Variation with Time – 11 kV CRM Back Up Drive

Figure 4.7.8:Frequency Variation with Time – 11 kV CRM Back Up Drive

It is observed that the real power drawn is not constant but depends on loading.

At the start of rolling the power increases typically to 6.392 MW, 6.74 MW from

0.001, 0.004 MW within 1 minute as can be seen from peaks in the Fig.4.6.10. The

power factor is lowest as 0.072 lag and highest 0.245 lag for reel drive (Fig. 4.5.11).

105

Figure 4.7.9Real Power Variation with Time – 11 kV CRM Entry Drive

Figure 4.7.10: Power Factor Variation with Time– 11 kV CRM Real Drive

The current and voltage harmonics are high at LT level. By reconstructing wave

shape there is scope for improvement and energy savings. Tuned harmonic filters are

required to be installed at locations where distortions are higher than limits prescribed

by 519-1992. In addition to harmonics reduction the power quality is improved and

provides stabilization effect and enhances equipment life. Filters restrict harmonics to

enter TG or DG, relieving the plant as a whole. They also restrict negative sequence

harmonics.

106

Very fast switching with suitable tuning reactor will be suitable in

consideration of fast varying load of the rolling mill. This will meet the dual purpose

of power factor improvement and harmonic limiting to a better extent. This will

achieve

• Reduced THDv

• Reduced voltage fluctuation

• Increased available driver power due to improved voltage

• Improved power factor

• KVA of the plant is reduced

• Transients free

• Rapid switching, fast response to meet the load pattern

Harmonic studies and analysis will increase system reliability and keep the

operations trouble free. The feasibility of tuned filters can be studied by calculations

of payback period. The TCR and DVR installation can benefit the plant.

4.8 Comparison between Experimental Analysis and Computational Analysis

The measurements of parameters related to arc furnace are recorded by

Candura Instruments. Electrical measurement is experienced on the high current side.

Measurements are carried out by Power Analyzer – Candura Instruments. The

PowerPro instrument is used for measuring and recording the following quantities.

Trends in voltage, current, power and frequency, Waveforms and harmonic trends to

64th

harmonic for both voltage and current, Voltages sags and swells Voltage

transients, Current Inrush, THD trend for voltage and current ,Frequency deviations

.It is noticed that harmonic performance at both furnaces almost matches with

computational performance, the reasons of accuracy of FFT method used is high and

for analysis Power flow software is very powerful. It generates harmonics of any

order and displayed the results for very small sampling time.

107

4.9 Justification of Error of Rolling Mill

There is slight difference in field measurements and software analysis. The variation

is due to inaccurate CTs used for connection, saturation of CTs, software analysis

accuracy will depend on input data.

4.10 Arc Furnace Performance Analysis

An AC arc furnace performance is determined by arc parameters study, arc stability.

The analytical performance can be estimated from circle diagram. The statistical

analysis can be done by using Histograms. Data from furnace can be analyzed to

arrive at conclusions. The arc parameters have influence on refractory wear. Wild

phase is issue in furnace operation. Effect of Refractive index on furnace operation

and regulation is studied.

4.11 Experimental Analysis of Arc Furnace Operation

The comparison of various taps is done for operating parameters of arc

furnaces 1 and 2

Table 4.4: Furnace Parameters for Arc Furnaces 1 and 2

Dec 06 Apr 03 Dec 06 Apr 03

Average Furnace Currents KAmp 39 41 41 41

Average Tap Voltage 491 504 505 505

Furnace Active Power (MW) 29.2. 30.7 32.3 33.6

Furnace Reactive Power (MVAR) 14.9 16.2 14.8 12.9

Furnace Apparent Power 33.0 35.4 35.7 36.0

Power Loss(MW) 2.34 2.66 2.52 2.57

Power factor 0.88 0.87 0.91 0.93

Operating Reactance-mohms 3.51 5.74 2.95 2.53

Current Balance % 96.9 92.6 81.0 97.1

108

Table 4.5: Arc Parameters for Furnaces 1 and 2

Arc Power (MW) 26.83 28.01 29.81 30.98

Arc Voltage in Volts 231 227 244 250

Arc Resistance in milliohms 5.96 5.52 6.05 6.07

Average Arc lengths (mm) 228 223 243 250

Electrical Efficiency of Arc (%) 92.0 91.3 92.2 92.3

Arc Stability Index 1.40 2.30 1.18 1.01

Arc Radiation Index (MWV/ sq.cm) 68 70 79 85

Arc Radiation Intensity (MW-

meter)

2.0 2.1 2.4 2.6

Table 4.6: Regulation Parameters for Furnaces 1 and 2

Current change rate – kA/Sec 5.8 4.3 1.3 1.7

Regulation deviation % (Constant

Current)

14.9 10.3 3.3 4.2

I/I0 of regulation System (Constant

current)

0.6 0.64 0.9 0.7

Average impedance (milliohms) 7.3 7.1 7.1 7.0

Impedance change rate 1.5 4.07 0.27 0.48

Regulation Deviation % -(Constant

Impedance)

20.6 57.7 3.8 6.9

Z/Z0 of Regulation System (Constant

Impedance)

1.7 1.60 1.6 1.6

109

Table 4.7: Analysis of Furnace Parameters for EAF 1

Furnace –EAF-1 Phase-wise Deviation Cumulative

Operating Voltages

Volts

U1 284.3 ∆ U1 1.3% Fur. Voltage

491 U2 284.2 ∆ U2 1.2%

U3 281.9 ∆ U3 1.4%

Operating Currents

kA

I1 39.2 ∆ I1 15.9%

Fur. currents

38.8 kA

I2 39.2 ∆ I2 13.2%

I3 38.0 ∆ I3 15.6%

Active Power Input

MW

P1 9.6 ∆ P1 18.6% Total 3 ph.

MVA

29.16

P2 10.0 ∆ P2 17.1%

P3 9.5 ∆ P3 18.7%

Transformer output

MVA

S1 11.14 ∆ S1 17.7% Total 3 ph-

MVA

32.98

S2 11.14 ∆ S2 16.1%

S3 10.70 ∆ S3 48.5%

Operating Reactance-

m ohms

X op-1 3.86 ∆ X op-1 25.9%

Xop3.51 mΩ X op-2 3.20 ∆ X op-2 22.0%

X op-3 3.46 ∆ X op-3 25.6%

Operating Impedance

m ohms

Z op-1 7.78 ∆ Z op-1 21.4%

Z op 7.77 Z op-2 7.63 ∆ Z op-2 16.5%

Z op-3 7.91 ∆ Z op-3 20.0%

Operating power

factor

Cos Φ1 0.86 ∆ Cos Φ1 8.8%

Avg Cos Φ

0.88

Cos Φ2 0.90 ∆ Cos Φ2 7.5%

Cos Φ3 0.89 ∆ Cos Φ3 7.3%

110

Table 4.8: Power Distribution Balance for EAF 1

Current Distribution

Ph-1 101.1% Voltage

Distribution

Ph-1 102.6%

Ph-2 101.0% Ph-2 102.3%

Ph-3 97.9% Ph-3 95.1%

Power Distribution

Ph-1 33.0% Overall

Power

balance

%

94.8 Ph-2 34.4%

Ph-3 32.6%

Overall

Current Balance

%

96.9 overall

voltage

Balance

%

99.1

Table 4.9: Arc Parameters for Furnace 1

Arc resistance

m ohms

R a-1 5.76 ∆ R a-1 25.4% R a 5.96

mΩ

R a-2 6.03 ∆ R a-2 21.4%

R a-3 6.09 ∆ R a-3 26.1%

Arc Power

MW

P a-1 8.83 ∆ P a-1 5.9% Total arc

power

26.83

P a-2 9.24 ∆ P a-2 5.6%

P a-3 8.76 ∆ P a-3 5.9%

Arc Voltage

Volts

V a-1 226 ∆ V a-1 42.60% Arc voltage

231 V a-2 236 ∆ V a-2 49.2%

V a-3 232 ∆ V a-3 11.4%

Arc Intensity MW-

meter

Ri-1 1.95 Arc

stability

index

STB-1 1.54

Ri-2 2.16 STB-2 1.28

Ri-3 2.00 STB-3 1.38

Relative refractive

index

RRi-1 33.1% Arc

radiation

index

MWV/

Rx-1

R x-2

Rx-3

65

RR i-2 34.5% 72

RR i-3 32.4% 67

111

sq.cm

Arc efficiency

(Electrical)

ηa-1 91.7 Power loss

MW

(inc.s/c

loss)

Lp-1

Lp-2

Lp-3

0.80

ηa-2 92.1 0.79

ηa-3 92.1 0.75

Arc length -mm La-1 221 Impedance

distribution

Ph-1

Ph-2

Ph-3

100%

La-2 234 98%

The performance of electric arc furnace is determined from circle diagrams. The

performance can be calculated provided the electrical parameters like arc length,

efficiency, refractive index, etc are measured. For two furnaces the measurements are

carried out and the various graphs are plotted.

4.12 Statistical Analysis of Arc Furnace

Current in kA

Phase Current 1

Phase Current 2

Phase Current 3

Figure 4.12.1: Histogram for Phase Current for Tap 19, 505 V, Scrap Melting

112

Current in kA

Phase Cur rent 1

Phase Cur rent 2

Phase Cur rent 3

Figure 4.12.2: Histogram for Phase Current for Tap 21, 545 V, Scrap Melting

Phase Cur rent 1

Phase Cur rent 2

Phase Cur rent 3

Current in kA

Figure 4.12.3: Histogram for Phase Impedance for Tap 19, 505 V Scrap Melting

113

Phase Current 1

Phase Current 2

Phase Current 3

Figure 4.12.4: Histogram for Phase Impedance for Tap 21, 545 V Scrap Melting

The Histograms are plotted to determine the quality of regulation and study

the pattern of current distribution and Percentage Phase Impedance at taps 19 and 21

during scrap melting. It is observed that at tap 19, the fast acting of regulation is

required. The sensitivity of all phases seem to be acceptable showing upward trend.

For tap 21, with higher currents, it was found that, there are few very high current

fluctuations showing similar upward trend.

The histograms for Phase Impedance for tap 19 shows almost equal

distribution of values for all phases. The impedance set point is well and level of

sensitivity is good. Significant upward trend is noticed. For higher tap 21, the

impedance set points are very well placed showing high level of sensitivity. Graph of

phase two is shifted due to high current in phase 2.

4.12.1 Analysis from Circle Diagram

The relation between the electrical properties of the arc furnace and the arc

resistance can be described by the circle diagram. Assuming the concept of balanced

operation, the figure illustrates the the equivalent circuit diagram of one phase.

The performance of electric arc furnace is determined from circle diagrams. The

114

circle diagrams are drawn for the commonly used tap 21, 19 and 14. the diagrams are

based on nominal secondary tap voltages which translates to nominal primary voltage

of 33 kV. The measured primary voltage was normally around 32 kV. That means the

diagrams show always little higher values for voltage and power than the furnace

actually performs.

The furnace is operating in general in the range of stable arc ignition

(particularly during DRI-melting and finishing, when the slag conditions are good.),

just in the beginning of melting the arc tends to be unstable. Because the current

during early stage of melting is slightly below the calculated limits. (e.g. for tap 19:

505 V the Theoretical limit is 42.2 kA

Figure 4.12.1.1 (a): Operating Points on Circle Diagram

A usual way of describe an electrical system is circle (or PQ) diagram, where active

power P is plotted versus reactive power Q. In this diagram the working points

corresponding to a certain secondary voltage are located along a semi-circular shaped

curve (that can be deformed if a variable operational reactance is considered). An

example of a typical circular diagram is shown in Fig. 1.The ideal circular diagram is

a useful design tool. It is obtained from equations Eq. 1, where U is the voltage, X is

the short-circuit reactance and cos Ф is the power factor. For a given secondary

voltage selected by the chosen transformer tap, and assuming a constant reactance

Using this diagram it is possible to define the right set points of operation in order to

achieve the highest active power within the arc stability limits.

115

The development of the flicker phenomenon in AC electric arc furnaces is also

explained by means of the circle diagram, The active power absorbed by the furnace

is

P = ( !" # )

( !" #)" $! (4.1)

whereVPCCis the voltage at the common point of coupling, Rtand Xtare respectively

the resistance and the inductance of the supply system and Ra is the arc resistance.

Thus the active power consumed by the load is maximum when Rt + Ra = Xtand is

equal to

Pmax = %

&$! (4.2)

P

R

A

A1

A2

Q X

X

Figure 4.12.1.1 (b): Operating Points on Circle Diagram

Operating diagram of the AC furnace can be drawn in the (X, R) plane as in

Fig.f.12.1.1 (a). From equations (1) and (2) it is deduced that the maximum active

power for an operating point corresponds to the same amount of reactive power

consumption. Arc furnaces consume consequently large quantities of reactive power

which should be compensated.

Around an operating point the resistance variations of the arc provoke

essentially variations of reactive power as shown in Fig.f.12.1.1 (b). If the operating

point moves from A1 to A2, the variation of the reactive power can be up to ten times

larger than the variation in active power. It is well known that the variation of the

reactive power has a direct impact on the voltage profile of the transmission system.

This explains why the variation of the reactive power consumption of an arc furnace

produces flicker phenomena.

The maximum active power for an operating point corresponds to the same

amount of reactive power consumption. Arc furnaces consume consequently large

116

quantities of reactive power which should be compensated. Around an operating point

the resistance variations of the arc provoke essentially variations of reactive power as

shown in Fig.1. If the operating point moves from A1 to A2, the variation of the

reactive power can be up to ten times larger than the variation in active power. It is

well known that the variation of the reactive power has a direct impact on the voltage

profile of the transmission system. This explains why the variation of the reactive

power consumption of an arc furnace produces flicker phenomena.

4.12.2 Effect of Electrode Regulation System on the Operating Point

It is concluded that the observed behavior is a consequence of the electrode

regulation system, whose effects are noticeable in the time scale (several seconds) that

is being considered in the measurements. In a short time scale (a cycle) the

characteristics that determine the arc behavior (voltage, reactance, arc length) can be

considered stable, and the relationships among them can be obtained theoretically or

from experimental measurements of wave forms, and expressions relating operational

reactance to intensities can be obtained. But in a longer time scale these relationships

are no longer valid; because they are superseded by a stronger (but slower) driving

force, that is, the electrode regulation system.

Figure 4.12.2.1: Setting of Operating Point by Electrode Control

117

As the electrical magnitudes change continuously during process in order to

keep a selected working point (for example a constant intensity), an active system that

is able to dynamically change some parameter is required. This is the task of the

regulation system, that monitors some system variable (intensities, impedance, power

and moves electrodes up or down until the right value is achieved. The movement of

the electrodes, that is the response of the regulation system, is much slower than the

fluctuations that are taking place in the arcs. So when we look at the system in a long

time scale the observed results are controlled by the regulation, and not by other short

time effects. In the long scale (several minutes), if operational settings (tap, intensity)

are not changed the secondary intensity is constant. Intensity will fluctuate a lot and it

can’t be considered constant in a short time scale, but in the long scale regulation

dominates; so on average intensity can be taken as constant. Thus, the fluctuations in

the arc behavior in the long scale cannot manifest as intensity variations, because it is

kept constant by the regulation, so it manifests as variations in tension and power

factor. But using the same reasoning in the long scale voltage is also constant

(transformer tap) so the only degree of freedom is the power factor.

4.12.3 Response to an Increase of Primary Voltage

Due to maintenance operations in the substation yard at Bhugaon Plant,

Wardha the second harmonic filter was down during some time. When it was working

again, an increase in the primary voltage of the EAF transformer was observed, .

Comparing power data before and after the primary voltage increase and separating

data collected during main melting phase and during slag foaming, different behavior

is observed. For data collected from slag foaming stage the averaged reactive power

remains constant after the change, while the apparent power is higher. This is not

observed for data collected during main melting phase, where reactive power and

apparent power both increase. Active power logically increases in both cases. This

different behavior can be explained using the constant apparent power diagram.

Comparing power data before and after the primary voltage increase and separating

data collected during main melting phase and during slag foaming, different behavior

118

is observed. For data collected from slag foaming stage the averaged reactive power

remains constant after the change, while the apparent power is higher. This is not

observed for data collected during main melting phase, where reactive power and

apparent power both increase. Active power logically increases in both cases. This

different behavior can be explained using the constant apparent power diagram. The

higher voltage is translated into a longer arc length. The reactive power during slag

foaming has the same value before and after the change because, first, secondary

intensity is the same (due to the electrode regulation set point) and, second, because

operating reactance has the same value: the minimum possible value. Slag foaming

stabilizes the arc in such way that despite of the arc length the measured reactance is

approximately the short-circuit reactance. In both situations the arc is fully covered by

the foamy slag. So reactive power, that can be expressed as Q = 3X (remains

constant. On the other hand, apparent power logically increases because voltage is

higher and intensity is constant. In a circle diagram, the new average working point

obtained after the primary voltage increase is obtained by a vertical displacement

from the point P1 (that lies on the S1 curve) to P2 (that lies on the S2 curve) . As active

power increases while reactive power doesn’t, power factor must also increase. This is

confirmed by the data collected: average power factor (computed as P/S) raises from

0.855 to 0.864 with the primary voltage increase.

P (MW)

COSϕ

P3

P2

P1

S1 (Vp<<)

S2 (Vp<<)

Q (Mvar)

Figure 4.12.3.1: Variation of Power with Increase in Primary Voltage

119

For the data collected during main melting phase the increase in primary

voltage results in increases both of apparent and reactive power (along with active

power). As before, the higher primary voltage is converted into a higher secondary

voltage, while intensity is the same due to the regulation set point that has elevated the

electrodes to a higher position in order to get a longer arc with higher resistance to

obtain the same intensity with higher voltage. On opposition to the slag foaming

situation an increase of arc length results in less stable arcs with a higher reactance.

So although intensity is the same reactive power increases due to the higher operating

reactance. In the circle diagram, the operating point after the change in primary

voltage will move from a point in the constant-S, S1 curve to a point in the S2 curve

with higher Q and P. As both powers change more or less in the same way, the

translation from one point to the other is done along a line of constant power factor,

so the new point can be computed as the cross of the line that contains P1 and the

origin and the S2 curve. The equal power factor before and after the voltage increase

is also confirmed experimentally, as the average values of power factor P/S are equal

to 0.84 in both situations.

Slag foaming quality and active power similar effect to the previously

described appears also in a different context. A foamy slag quality assessment

technique is used at Bhugaon plant, Wardha. If the slag foaming process is correctly

done, a low noise index is obtained, measured both using acoustic noise and electric

THD (Total Harmonic Distortion). If for some reason a good foaming is not achieved

(for example due to problems in oxygen or graphite injection) then a high noise index

is obtained. Relating the slag foaming noise index with the electrical variables (active

power and secondary voltages) measured in the furnace during the process an

interesting result is observed. When the noise index is low, meaning that a very good

foaming has been reached a strong correlation between measured active power and

averaged secondary voltage is observed. But if noise index is high, this dependence is

not observed and both variables are uncorrelated.

In the case of having a low noise index, with good quality foam, the slag

covers the arcs completely, so the operating reactance will be the lowest possible

120

value (close or equal to the short-circuit reactance), regardless of arc length and

secondary voltage. As secondary intensity is the same in all the process (due to the

regulation and in the long time scale), arc length is directly proportional to voltage,

and so is arc resistance. This results in an active power proportional to secondary

voltage through P=3=S cos Ψ while reactive power remains constant. Voltage

fluctuations during slag foaming process results in directly related active power

fluctuations (for low noise index).This can be written as:

= S

( !"#)

$ ≈ constant (4.3)

On the contrary, if the foaming achieved is not good enough (for example as

result of oxygen or graphite injection malfunction or non standard slag composition)

the noise index will be high. In this situation the variation in arc length caused by a

variation in secondary voltage will result in a variation of both resistance and

reactance, so the voltage fluctuations don’t result in power factor fluctuations. It can

be said that, approximately, reactance and resistance both changes in the same way so

power factor is the same. In this situation must be changed to

= S

( !"#)

$ ≈ 0 (4.4)

This explains why the active power and the averaged secondary voltages are

not correlated when a high noise index is detected.

Discussion

In this work power data coming from an AC EAF has been analyzed using a

PQ diagram. It is observed that for a given operation set points (secondary voltage)

the distribution of active and reactive power points in the diagram lies on a curve of

constant apparent power (S) rather than in a constant voltage curve, as it is used for

theoretical determination of set points. This is explained considering the long time

scale of the measurements, where the constant intensity electrode regulation system

response dominates over other effects. Making use of the interpretation of the diagram

as a constant-S diagram, we have explained several phenomena observed in the EAF

under study. The increase of primary voltage resulted in a higher apparent power, but

for data collected during foamy slag this is not translated in higher reactive power,

121

that remained the same. This is not observed for data coming from main melting

stage, where apparent and reactive power both increase. Also, the different

relationship observed between active power and secondary voltage during slag

foaming when the noise index is high or low is also explained.

4.12.4 Effect of Phase Rotation

The phase rotation of the furnace is measured clockwise, which is normally

not recommended, because of higher risk of unscrewing of electrode joints. In

addition to this the direction of the slag moving due to electromagnetic forces is

opposite and thus is anticlockwise. This means the slag moves to the front from the

area where it is mainly created. Less efficient melting together with higher energy

consumption is possible.

4.13 Computational Analysis of EAF

Test system under study is considered here for analysis. When furnace 1 is tested for

the input. The results for Iarc,Varc, and Rarc are obtained as shown in the figure. The

V-I characteristics for the arc furnace is similar and close to the theoretical model 3 as

mentioned earlier.

Figure 4.11.1: Waveforms of Iarc,Varc,Rarcand V-I Characteristics of Arc Furnace

122

Vig

R1

VaR3

R2

i3 i2i1

Current(A)-1 -0.5 0 0.5 x 105 1

400

300

200

100

0

-100

-200

-300

-400

Figure 4.13.1: Theoretical Model

It is noticed that, there is little difference in the two models as illustrated,

which confirms the validity of the model. The nonlinearity exists due to the fact that

assumptions are made mostly non electric factors which contribute.

4.14 Comparison of Experimental Analysis and Statistical Analysis

It has been observed that statistical analysis shows the trend variations for impedance

and current distribution which is slightly different than values of parameters as

mentioned in experimental analysis. There is no information regarding change in

operating band of impedance or tap setting in experimental analysis, so statistical tool

of histograms prove to be useful.

4.15 Justification of Error For Arc Furnace Model

There is slight difference in both these analysis, the main reasons behind the variation

in non electric factors like DRI feeding pattern, use of non precise method of

electrode control. Manual method of regulation may contribute to the error of results.

There is a need of exact and precise mode of operation leading to new control strategy

which will make current distribution to be more uniform and problem of thermal

balancing will be sorted out.

123

4.16 SVC Light Analysis

SVC is a power quality conditioning device used for limiting flicker in arc furnaces,

rolling mills. This proves to be beneficial when it was used at different sites in steel

plants. This discussion will incorporate the uses of SVC at different locations in arc

plants.

4.17 Experimental and Computational Analysis for Flicker Mitigation

The electric arc furnace is deriving its power from MSETCL grid. Due to the

large power of the new steel industries associated with EAF, above existing the

furnaces, and relatively weak grid, concerns were raised. The EAF is a generator of

several kinds of disturbances. Furthermore, the EAF is an unsymmetrical load on the

three phase feeding grid, giving rise to current and voltage unbalance in the grid. The

EAF is a generator of harmonics, odd and even as well as inter harmonics.

Considering this scenario, the Jalna MIDC area having 90% of the power in

the area is consumed by steel industries and more plants are to in waiting for

installations. The analysis of fluctuations and reactive power flow were carried out.

Due to a modest short circuit level at the PCC, unless properly remedied, the EAFs

would become a formidable source of disturbances, which would spread through the

grid to other consumers of electric power. The EAF is also a heavy consumer of

reactive power.

To perform the computational analysis, a detailed model was used with the

power system as well as the EAF with the SVC Light. The complete SVC Light

system was represented in the EMTDC model. The model included all main circuit

components such as the IGBT converter with its DC capacitors, the phase reactors and

the harmonic filters. The model also included a complete representation of the control

system and the valve firing system. The EAF was represented as a current source,

124

reproducing the current waveform recorded on the old Kalika Stainless EAF. The data

used in the simulation had a length of 10 minutes, starting at the beginning of a

meltdown process, the most troublesome part of the melt from a flicker point of view.

To evaluate the flicker level, the voltage fluctuations were computed at the 33 kV bus.

The flicker level Pst was then estimated according to IEC [1].

A model of the flicker meter was included in the digital model. The voltage

measured at the PCC was passed through different filters to obtain the flicker level

according to the desired criteria. The output of the flicker meter was based on the

instantaneous flicker value of the IEC meter (output 3). Using the model and the input

data, the flicker level at the 33 kV bus was estimated both with and without the SVC

Light in operation.

125

Load Active Power

Negative Sequence Current

33 kV Bus

220 kV Bus

Reactive Power

126

Figure 4.17.1: Simulation at Pre Melting Stage

a) EAF active power

b) EAF and step-down transformer reactive power

c) Flicker

d) Bus voltage

e) EAF and step-down transformer negative sequence current.

In Fig.4.17.1, the corresponding flicker levels at the PCC are shown with the

SVC Light in service. Based on the figures, it can be noted that the active, useful,

power of the EAF increases when the SVC Light compensates the reactive power

consumed by the EAF. This is beneficial for the steel process with shorter melting

times, as well as for the power supplier with less reactive power flow. The design

studies indicated a flicker improvement ratio of approximately 13/3.1=4.2.

To assess the accuracy of the SVC model for MSETCL, actual field

measurements are done. For that two sites were chosen to see the impact of SVC

Light on flickers. The analysis presented in the next section is based on the data and

waveforms collected on practical sites.

4.17.1 Experimental Analysis of SVC Light Commissioned at Bhilai Steel Plant

For analyzing the effect of SVC Light, practical measurement of flickers at the

EAF in the steel rolling plant of Bhilai Steel of Steel Authority of India Limited

(SAIL) situated in the Bhilai city of Chattisgarh, India was carried out. The Bhilai

Steel has commissioned 44 Mvar SVC Light for its electric arc furnace. The single

line diagram of the EAF and SVC Light is shown in (Figure 4.17.1). The EAF is rated

31.5 MVA with a 20% temporary overload capability, whereas the LF is rated 6 MVA

plus a 30% overload capability. Both furnaces are fed from a 132 kV grid via an

intermediate voltage of 11 kV (Fig.4.17.1). The feeding grid is relatively weak, with a

fault level at the P.C.C. of about 1000 MVA. It is obvious that this is quite insufficient

to enable operation of the two furnaces while up keeping reasonable power quality in

the grid. The SVC Light is rated at 0 - 44 Mvar of reactive power generation,

continuously variable. This dynamic range is attained by means of a VSC rated at 22

MVA in parallel with two harmonic filters, one rated at 14 Mvar existing in the plant

initially and one installed as part of the SVC Light undertaking, rated at 8 Mvar. Via

127

its phase reactors, the VSC is connected directly to the furnace bus voltage of 11 kV.

During the energisation of the VSC, the DC capacitors are charged via the charging

resistors. While the DC capacitors are charged, the by-pass switch is closed. The main

circuit diagram of the SVC Light is shown in the Figure 4.17.2.

Figure 4.17.2: Single-Line Diagram of EAF Feeding Network and Compensation

v

v

v

Figure 4.17.3: Main Circuit Diagram of Bhilai SVC Light

The possibility for the SVC Light to “follow” the stochastically varying

furnace current gives an enormous opportunity to reduce voltage flicker. However,

with its quick response, also other unpleasant disturbances can easily be reduced. The

recording below shows the initial operation of the SVC Light when paralleled to the

EAF in Bhilai steel Plant. Signals show the SVC Light ability to track the rapid

changing of the furnace current.

128

Figure 4.17.4: Tracking of Rapid Changes of EAF Current

With a switching frequency higher than 1 kHz, even fast transients can be

damped. In some installations switching transients from a furnace transformer could

for example lead to unexpected problems. With the SVC Light technology, the

furnace operation with all switching transients will be compensated.

Field measurements have been performed in order to evaluate the performance

of the SVC Light at Bhilai site. So far, the measurements have mostly been focusing

on flicker. The recordings and evaluation presented below show flicker mitigation of

3.4 times. The flicker mitigation is expected to increase to a factor around 4.5 to 5.0

times when new auxiliary transformers have been installed. The flicker measurements

have been performed according to the UIE/IEC method. The method is well known in

Europe, Canada, South America and South Africa and is also getting more common in

the United States.

Statistical evaluation

In order to evaluate flicker generation with and without SVC Light a statistical

evaluation has been performed. The figures below show histograms for the following

three cases:

Figure 4.17.5: Furnace in operation, without SVC Light. Pst(95%) = 3.90 p.u.

4.17.6: Furnace in operation, with SVC Light. Pst(95%) = 1.21 p.u.

4.17.7: Background flicker generation. Pst(95%) = 0.41 p.u

129

Figure 4.17.5: Flicker Generation without SVC Light

Figure 4.17.6: Flicker Generation with SVC Light in Operation.

Figure 4.17.7: Background Flicker Generation

130

According to the UIE (International Union For Electro heat), the recommend

practice to evaluate flicker severity from multiple sources is per the following

formula:

1/m

m

ST

i

PST = (P ) ∑ (4.5)

where m is the summation coefficient. The summation coefficient considers the risk

of coincident furnace operation. The factor is recommended to vary from 1 up to 4.

m=4 is used, when furnaces specifically run in order to avoid coincident melts. m=1 is

used when there is a very high occurrence of coincident voltage changes. The

background flicker is generated by many sources, and will therefore appear more or

less constant over the day. To deduct the background flicker, an “m” factor of 1

should be used. However to perform the evaluation in a very conservative way, a

factor of two is used.

In the case of flicker generated by the EAF only and without background

flicker, we get with no SVC Light in operation:

2 2 1/3 2 2 0.5

STA ST1 ST3P = [(P - P )] = [(3.90 - 0.41 )] = 3.88 (4.6)

p.u.

2 2 1/3 2 2 0.5

STB ST2 ST3P = [(P - P )] = [(1.21 -0.41 )] =1.14 (4.7)

p.u.

The residual flicker value with the SVC Light and EAF in operation and

without background flicker is:

The flicker mitigation by the SVC Light is:

STASVCLight

STB

P 3.88R = = = 3.4 p.u.

P 1.14 (4.8)

Increased furnace power

An arc furnace requires a stable voltage supply for optimum performance. An

SVC Light can instantaneously compensate the random reactive power variations, and

hence the voltage variations, of an EAF.

131

Reactive power compensation by an SVC Light or an SVC helps to obtain the

following production benefits:

A higher and stabilized voltage level at the furnace bus bar, giving:

- Shorter melt down times

- Reduced energy losses

- Reduced electrode consumption.

4.18 Experimental Analysis and Computational Analysis of SVC Light

Commissioned at Steel Plant

*

*

*

*

**

220 kV

33 kV

220/34.5 kV

50th

0.8 Mvar

25th

2 Mvar

3rd

25 Mvar

4th

31.3 Mvar+ 32 MvarEAF

Figure 4.18.1.: Single Line Diagram of the EAF and the SVC Light

In Fig. 4.18.1, the EAF is shown to the left. Then two high-pass filters for

reduction of harmonics follow together with the VSC and its switchgear. A resistor is

temporarily inserted in series with the VSC during start-up. The resistor is used to

limit the inrush current during the start sequence. Finally two filters to provide bulk

power to bias capacitive reactive power are found. The harmonic filters are designed

to fulfill the specified voltage and current distortion at PCC. To fulfill these

requirements the tuning and damping of the harmonic filters have been selected. The

total filter rating is equal to the VSC rating providing an operation range from zero to

132

twice the VSC rating. Fig. 4.18.1 shows of the installation of the SVC Light at the

Lloyd Steel plant.

Field measurements have been performed to validate the flicker improvement

performance. The phase currents of arc furnace and compensator should be digitally

recorded simultaneously during a typical melting period. The fluctuating voltages and

corresponding instantaneous flicker levels should then be calculated using the

impedance of the grid. The grid model was tuned to give optimal correlation between

flicker caused by the measured 33 kV voltages and flicker caused by the calculated

voltages, which were based on the measured currents. Using only the arc furnace

currents the reference Pst0 is found. This corresponds to what could be expected

without any compensation. Then, the reduced flicker level Pst1 is calculated using the

total current of the arc furnace and the compensator. This corresponds to the real

operation case. The improvement ratio Pst0/Pst1 is finally calculated over the whole

melting period and then evaluated.

Measurements in a dynamic process with high sampling frequency require

special attention when selecting the equipment. A sophisticated measurement

computer with high bandwidth and synchronous sampling was found suitable for the

task. For the measurements, totally four three-phase signals are of interest:

1. The 33 kV bus voltages

2. The EAF currents

3. The VSC currents

4. The filter currents

The raw data were sampled with a frequency of 5200 Hz for off-line analysis.

For verification purposes, two different standard flicker meters were connected

simultaneously with the measurement computer. In the following plots, data from

these flicker meters are shown together with the results from the specified method. A

good correlation between the signals will verify the evaluation method and the

correctness of the used model.

133

First, the off-line flicker meter was verified against the standard flicker meter

output. In this process, tuning of the grid model is one part. The results are shown in

Figure 4.18.2.

Figure 4.18.2: Verification of the Off-Line Flicker Meter

The deviations between the Pst value calculated from measured bus voltage

and the Pst values obtained from the flicker meter are ignorable. The error is generally

about 2-4 % except during large Pstchanges. Note that the Pst- meters are not

synchronized, which explains the large errors at large Pst value changes. This test

validates the quality of the input data and the accuracy of the flicker meter

implemented in the digital model.

In Figure 4.18.2, subplot 1 shows the EAF active and reactive power, the two

top curves. The bottom curve shows the reactive power taken from the grid. Subplot 2

shows a comparison between actual measured 33 kV voltage and the corresponding

calculated voltage. The ignorable error is a verification of the grid impedance model.

Subplot 3 shows four different Pst curves calculated during sliding 10 minute

intervals. The top curve with the highest Pst levels shows the case with only EAF

current and no compensation. The three bottom curves only have small deviations and

show flicker curves based on the measured voltage, on the calculated voltage and on

data from the standard Pst meter. The flicker level from the simulation using EAF,

VSC and filter currents gives the same flicker level as the external flicker meter and

the flicker level using the measured voltage. Subplot 4, finally, shows the flicker

improvement ratio calculated as the ratio between Pst values with only EAF currents

and Pst values with the SVC Light in operation. The flicker improvement at the

134

beginning of the melt-down process is ignored because it contains the time window

when the EAF was not in operation and non-applicable flicker values.

Operation of the EAF with full power requires stable voltage and an efficient

compensator. Operating the EAF without compensation is also possible, however, at a

reduced power. During the measurement campaign measurements at reduced EAF

power were performed both without and with the SVC Light in operation. Fig. 4.18.3

shows the results of the data processing.

135

EAF MW&Mvar, Network Mvar

33 kV bus voltage

P 10 minutesst

P 10 Improvementst

P 1

0 I

mp

rov

eme

nt

stP

st

p.u

.M

W M

var

Figure 4.18.3: Flicker with Full EAF Power and with SVC Light in Operation

• Subplot 1: EAF power and grid reactive power.

136

• Subplot 2: Voltage profile; measured (black) and simulated (red).

• Subplot 3: Sliding Pst 10-minute values from measured voltage (black), from

simulation with EAF current only (red), simulated with EAF+VSC+Filter

currents (blue) and from external flicker meter (green).

• Subplot 4: Simulated flicker improvement ratio.

Two melts with the same EAF transformer tap changer patterns are shown. At

approximately 14:40, the SVC Light was put on line. Subplot 2 showing the 33 kV

bus RMS voltages clearly indicates the difference. The large voltage peaks during the

melting without the SVC Light are due to load rejection. The voltage is then

controlled with the 132/33 kV transformer tap changer and finally reaches the set

point. The voltage set point was chosen below unity to reduce the voltage amplitudes

after load rejection.

The flicker improvement ratio is calculated during a time window were the EAF is in

operation. When the EAF is out of operation the flicker improvement will be unity in

case of no background flicker and below unity to zero if there is disturbing

background flicker existing. The flicker improvement ratio was calculated using a

time window beginning 10 minute after EAF melting start and ending at the end of

the same heat. The 10 minute delay corresponds to the 10-minute time window used

by the flicker meter to exclude the period the EAF was not in operation. The method

was applied to the data with full EAF power. This time window includes two EAF

transformer energizations and includes the time window with the highest flicker level

recorded during the measurements. The result was a flicker improvement ratio of 4.6

times for full EAF power. On the other hand, the flicker improvement ratio with

lower power in the EAF is between 5 and 6 times. If higher flicker improvement is

desired, it is hence essential to choose an adequate rating of the SVC Light.

137

Network MW & Mvar

P 10 minutesst

P i

mp

rov

emen

tst

Pst

p.u

.M

W M

var

P improvementst

33 kV bas voltage

Figure 4.18.4: Flicker Without and with the SVC Light, Reduced EAF Power

138

Apart from the flicker mitigation performance of the SVC Light, it also

contributes to the reduction of negative phase sequence currents in the grid. Fig. 4.32

shows the positive and negative phase sequences of the grid current without (top) and

with (bottom) the SVC Light in operation.

Net

work -Seq

.N

etw

ork +

Seq

.Network kA Currents

Figure 4.18.5: Positive & Negative Phase Sequence Currents without &with SVC Light

Comparison of field results with the simulated results shows consistency in the

operation of SVC Light. It clearly observed that SVC will be proved to be an efficient

flicker mitigate instrument to MSETCL network to the rolling mills in Jalna,

Maharashtra. Hence following benefits are assessed to MSETCL Network on basis of

analysis done above.

The installation of the SVC Light has brought benefits not only to the steel plant, but

also to the grid owner:

• Acceptably low flicker level at the Point of Common Coupling.

• Acceptably low amounts of harmonic distortion.

• Adequate load balancing between phases of the 132 kVgrid.

• A high and constant power factor at the feeding point of the plant, with low and

constant reactive power consumption from the grid.

139

• Keeping grid reinforcements at a minimum. Hence, this is a win-win situation where

the steel works as well as the utility can gain from the SVC Light installation.

4.19 Comparison

Comparison of Experimental analysis and Computational analysis shows the SVC

light is useful in mitigating flicker. The software used for analysis is EMTDC which

requires correctness of data to be fed.

4.20 Justification for Error:

There is difference in experimental results and computational analysis, the resons are

tuning of SVC to random variations in arc power.