
Power Systems
Drag the conductors to change a line's inductance, watch the Ferranti rise at no load, and clear a fault before the rotor slips a pole
24 chapters, 24 interactives, few words. Every chapter is mapped to the Anna University, VTU, SPPU, Mumbai University and KTU units it covers: load curves, line resistance, inductance and capacitance, short, medium and long lines, power flow over a line, corona, sag, insulators, cables, distributors and power factor improvement, the per-unit system, the bus admittance matrix, Gauss-Seidel and Newton-Raphson load flow, economic dispatch, symmetrical and unsymmetrical faults, symmetrical components, travelling waves, and the swing equation and equal area criterion.
- AUEE3401Transmission and Distribution
- AUEE3501Power System Analysis
- VTUBEE402Transmission and Distribution
- VTUBEE601Power System Analysis I
- VTUBEE703Power System Analysis II
- SPPU203145Power System-I
- SPPU303148Power System-II
- MUEEC304Electrical Power System I
- MUEEC702Electrical Power System III
- KTUPCEET501Power Generation, Transmission and Protection
- KTUPBEET604Power System Analysis
- MUEEC502Electrical Power System II
By Skill-Lync · Updated 19 Nov 2026 · 24 chapters · 24 interactives · free, no sign-up
Structure of the power system, load curves and load factors
Which syllabus covers thisAU EE3401 Unit IVTU BEE402 Module 1SPPU 203145 Unit 1MU EEC304 Module 1KTU PCEET501 Module 1
What stages does power pass through between the power station and a consumer?
Generators feed step-up transformers, which raise the voltage for long-distance transmission: in India 400 kV and 765 kV AC, and ±500 kV or ±800 kV HVDC. Substations then step it down through 220, 132 and 66 kV (sub-transmission) to 33 and 11 kV distribution feeders, and distribution transformers give consumers 230 V single-phase or 400 V three-phase at 50 Hz. The reason for high voltage: for the same power the current falls as 1/V, so the I²R loss falls as 1/V². Sending 100 MW over the same 100 km line at 220 kV instead of 132 kV cuts the loss from 4.37 MW to 1.57 MW.
What do a load curve and a load duration curve show?
A load curve plots demand against the time of day. Its highest point is the maximum demand and the area under it is the energy delivered. Rearrange the same 24 hourly loads from largest to smallest and you get the load duration curve: it shows for how many hours the load stays above any level, and its area is the same energy. In the interactive, an illustrative feeder with domestic (12 MW maximum), commercial (6 MW) and industrial (8 MW) consumers peaks at 21.84 MW at 19:00 and delivers 379.1 MWh a day. Its load is above the average for 15 of the 24 hours.
How are load factor, demand factor, diversity factor and capacity factor defined?
Load factor = average load / maximum demand, over a stated period. Demand factor = maximum demand / connected load: a workshop with 50 kW of machines that never draws more than 30 kW has a demand factor of 0.6. Diversity factor = sum of the individual maximum demands / maximum demand of the group. It is at least 1, because consumers do not all peak at the same moment. Plant capacity factor = average load / installed capacity, and utilisation factor = maximum demand / installed capacity. The illustrative feeder: load factor 0.723, diversity factor 1.19, and on 30 MW of capacity a capacity factor of 0.526.
Why do these factors matter to the utility?
A plant's fixed cost (capital, interest, staff) is paid whether it runs or not, so the cost per kWh falls as the load factor rises. Diversity lets one station and one feeder serve a group whose individual maxima add up to more than its capacity. The load duration curve also splits the work. The bottom slice, needed all 24 hours, suits base-load plant with low running cost. The thin top slice, needed only a few hours a day, suits peaking plant such as hydro or gas turbines, which are cheap to keep idle and quick to start.
A purely domestic feeder peaks at 19:00. You add a commercial load that peaks at noon. The feeder's diversity factor
With one class the group maximum equals the sum of individual maxima, so the diversity factor is 1. Once the classes peak at different hours, the combined maximum is less than the sum of the separate maxima, so the ratio rises above 1. It can never fall below 1.
Worked examples
Factors of a mixed feeder · An illustrative 11 kV feeder: domestic maximum 12 MW (peak 19:00), commercial 6 MW (peak 12:00), industrial 8 MW (peak 09:00), hourly profiles as in the interactive; installed capacity 30 MW
- Adding the three classes hour by hour, the combined load peaks at 19:00: maximum demand = 21.84 MW
- Energy = sum of the 24 hourly loads × 1 h = 379.08 MWh; average load = 379.08/24 = 15.79 MW
- Load factor = 15.79/21.84 = 0.723
- Diversity factor = (12 + 6 + 8)/21.84 = 1.19
- Capacity factor = 15.79/30 = 0.526; utilisation factor = 21.84/30 = 0.728
Answer: Maximum demand 21.84 MW at 19:00, load factor 0.723, diversity factor 1.19, capacity factor 0.526
Splitting a load duration curve between base and peak plant · An illustrative station's daily load duration curve: 90 MW for 4 h, 70 MW for 8 h, 45 MW for the remaining 12 h. A 45 MW base unit runs all day; a 45 MW peaking unit supplies the rest
- Energy = 90 × 4 + 70 × 8 + 45 × 12 = 1460 MWh; load factor = 1460/(24 × 90) = 0.676
- Base unit: 45 MW for 24 h = 1080 MWh, capacity factor 1
- Peaking unit: the area above 45 MW = (90 − 45) × 4 + (70 − 45) × 8 = 380 MWh
- Its capacity factor = 380/(45 × 24) = 0.352
- While it runs (12 h) its use factor = 380/(45 × 12) = 0.704
Answer: Station load factor 0.676; base unit 1080 MWh, peaking unit 380 MWh at capacity factor 0.352
Why transmission voltage is high · 100 MW at 0.95 power factor sent over 100 km of three-phase line with one ACSR Zebra conductor per phase (DC resistance 0.0687 Ω/km at 20 °C, IS 398 table), at 132, 220 or 400 kV
- R per phase = 0.0687 × 100 = 6.87 Ω
- 132 kV: I = 100 × 10⁶/(√3 × 132 × 10³ × 0.95) = 460.4 A; loss = 3I²R = 4.37 MW
- 220 kV: I = 276.2 A; loss = 1.57 MW
- 400 kV: I = 151.9 A; loss = 0.476 MW
- The loss scales as 1/V²: (132/220)² = 0.36
Answer: Loss 4.37 MW at 132 kV, 1.57 MW at 220 kV, 0.476 MW at 400 kV (DC resistance at 20 °C, so real losses are a little higher)
Line resistance and inductance: skin effect, GMR, GMD, transposition and bundled conductors
Which syllabus covers thisAU EE3401 Unit IVTU BEE402 Module 2SPPU 203145 Unit 4MU EEC304 Modules 2-3KTU PCEET501 Module 2
What sets the resistance of an overhead conductor, and why is AC resistance higher than DC?
DC resistance is ρl/A. Stranding adds a little because each spiralled strand is longer than the conductor, and resistance rises with temperature. ACSR Moose (54 aluminium strands of 3.53 mm round a 7-strand steel core) is listed at 0.0560 Ω/km at 20 °C, so 100 km has 5.6 Ω. With AC the current crowds towards the surface (skin effect), and nearby conductors push it to one side (proximity effect). Both shrink the useful area. Using the resistivity implied by that table, the skin depth at 50 Hz is about 12.2 mm, more than the 10.6 mm thickness of Moose's aluminium layers, so the current still uses nearly all of them and the rise is small.
Why does a conductor's inductance use 0.7788 r instead of its radius?
Current flows through the whole cross-section, so some flux links only part of it. Adding up the flux linkage inside a solid round conductor gives the same result as a thin tube with no internal flux of radius r' = r·e^(−1/4) = 0.7788 r. This is the geometric mean radius (GMR). Each conductor of a single-phase pair then has L = 0.2 ln(D/r') mH/km. Two Panther conductors (21.00 mm diameter) 1.5 m apart give a loop inductance of 2.085 mH/km, or 0.655 Ω/km at 50 Hz.
What is GMD, and why are three-phase lines transposed?
With unequal spacing each phase links a different flux, so the three phase inductances differ and the line looks unbalanced. Transposition swaps the conductor positions at intervals so that each phase spends a third of the length in each position. The average inductance per phase is then 2×10⁻⁷ ln(GMD/GMR) H/m, with GMD = ∛(Dab·Dbc·Dca). For a flat line, GMD = 1.26 D. Zebra (28.62 mm diameter) at 7 m flat spacing: GMD 8.82 m, L = 1.335 mH/km, X = 0.419 Ω/km.
How does a bundle of conductors lower the inductance?
Replace one conductor per phase by two to four sub-conductors a few tens of centimetres apart. The bundle behaves like one conductor of much larger GMR: √(r'·d) for two, ∛(r'·d²) for three, 1.09·⁴√(r'·d³) for four. With Moose at 11 m flat spacing and d = 45 cm, the GMR grows from 12.4 mm (single) to 74.6 mm (twin) and 199.8 mm (quad), and X falls from 0.441 to 0.328 and 0.266 Ω/km. Bundles also lower the surface field, which reduces corona (chapter 7).
A flat line with 8 m spacing changes from one Moose conductor per phase to a twin-Moose bundle 45 cm apart. Its inductance per phase
Inductance depends on ln(GMD/GMR). Twinning raises the GMR from 12.4 mm to √(r'·d) = 74.6 mm. The logarithm falls from ln(815) to ln(135), so L drops from 1.341 to 0.981 mH/km, about 27 %. The logarithm stops it from halving.
Worked examples
Loop inductance of a single-phase line · Two ACSR Panther conductors (overall diameter 21.00 mm, IS 398 table) 1.5 m apart, length 10 km (illustrative)
- r = 10.5 mm; r' = 0.7788 × 10.5 = 8.18 mm
- Per conductor: L = 0.2 ln(1500/8.18) = 1.042 mH/km
- Loop (go and return): 2.085 mH/km; for 10 km 20.85 mH
- X at 50 Hz = 2π × 50 × 20.85 × 10⁻³ = 6.549 Ω
Answer: Loop inductance 20.85 mH, reactance 6.549 Ω for 10 km
A transposed 220 kV flat line · Three-phase, one ACSR Zebra (28.62 mm diameter) per phase, flat horizontal with 7 m between neighbours (illustrative), fully transposed
- Distances 7, 7 and 14 m: GMD = ∛(7 × 7 × 14) = 8.819 m
- r' = 0.7788 × 14.31 = 11.14 mm
- L = 0.2 ln(8.819/0.01114) = 1.335 mH/km per phase
- X = 2π × 50 × L = 0.4193 Ω/km, so 41.9 Ω for 100 km
- Without transposition the middle phase alone would see 0.2 ln(7/0.01114) = 1.289 mH/km
Answer: GMD 8.82 m, L = 1.335 mH/km, X = 0.419 Ω/km
Single, twin and quad Moose compared · ACSR Moose (31.77 mm diameter, 0.0560 Ω/km DC at 20 °C), flat horizontal at 11 m, bundle spacing d = 45 cm (illustrative), transposed
- GMD = 1.26 × 11 = 13.86 m; r' = 0.7788 × 15.885 = 12.37 mm
- Twin: GMR = √(0.01237 × 0.45) = 74.6 mm; quad: 1.09 × ⁴√(0.01237 × 0.45³) = 199.8 mm
- L = 0.2 ln(GMD/GMR): 1.404, 1.045 and 0.848 mH/km
- X at 50 Hz: 0.441, 0.328 and 0.266 Ω/km
- DC resistance per phase per km: 0.0560, 0.028 and 0.014 Ω
Answer: X falls from 0.441 (single) to 0.328 (twin) and 0.266 Ω/km (quad)
Line capacitance, charging current and the effect of earth
Which syllabus covers thisAU EE3401 Unit IVTU BEE402 Module 2SPPU 203145 Unit 5MU EEC304 Module 3KTU PCEET501 Module 2
Where does the capacitance of a line come from?
A charged conductor carries charge q per metre and sets up a radial field E = q/(2πε0x) that falls with distance x. The voltage between two conductors is the integral of that field, and C is the charge per volt. For a single-phase pair, C between the wires = πε0/ln(D/r), and the capacitance to neutral (the mid-plane) is twice that, 2πε0/ln(D/r) = 55.63/ln(D/r) nF/km. Charge sits on the surface, so the actual radius r is used here, not 0.7788 r. Two Dog conductors (14.15 mm diameter) 1.2 m apart: 10.84 nF/km to neutral, ignoring the earth.
What is the capacitance to neutral of a three-phase line?
For a transposed line the same steps give Cn = 2πε0/ln(Deq/r), with Deq the GMD of the phase spacings. Zebra (28.62 mm) at 7 m flat spacing: Deq = 8.82 m and Cn = 8.66 nF/km. Bundles raise C just as they lower L, with the bundle radius √(r·d) for two sub-conductors. Twin Moose 45 cm apart at 11 m flat spacing gives 10.91 nF/km, against 8.22 for one Moose. A line with low L and high C has a lower surge impedance and can carry more power (chapter 5).
How does the earth change a line's capacitance?
The earth is a conducting plane at zero potential, so it bends the field lines. Replace it by an image conductor, carrying the opposite charge, the same height below the surface. The images add a term −ln(Hm/Hs) to the denominator, where Hs = 2h and Hm is the mean distance to the other conductors' images, so C rises. For the Zebra line at 15 m the rise is 0.79 %; at 40 m it is 0.12 %; as h → ∞ it vanishes. The effect is small but grows when conductors are low or far apart.
What is the charging current, and why does it matter?
Even with the far end open, a line draws a leading current through its capacitance: I = (V/√3)·ω·Cn per phase. For the 220 kV Zebra line it is 0.348 A per km, about 34.8 A for 100 km. That is 13.3 MVAr of reactive power generated by every 100 km. At 400 kV with twin Moose the figure is 54.8 MVAr per 100 km. This charging current flows even at no load, causes the Ferranti rise of receiving-end voltage (chapter 5), and is why long EHV lines need shunt reactors.
The conductors of a line are raised from 15 m to 40 m above the earth, all else unchanged. The capacitance to neutral
The images move further away, so ln(Hm/Hs) shrinks towards 0 and the earth correction fades. For the 220 kV Zebra example the rise over the no-earth value drops from 0.79 % to 0.12 %. The capacitance always stays at or above the no-earth value.
Worked examples
A single-phase line with and without earth · Two ACSR Dog conductors (14.15 mm diameter) 1.2 m apart, both 8 m above earth, 11 kV between them, 20 km long (illustrative)
- Without earth: Cn = 55.63/ln(1.2/0.007075) = 10.837 nF/km
- Images: Hs = 16 m, Hm = √(16² + 1.2²) = 16.045 m; ln(Hm/Hs) = 0.0028
- With earth: Cn = 55.63/(5.1335 − 0.0028) = 10.843 nF/km; between wires Cab = Cn/2 = 5.422 nF/km
- Charging current = 11 000 × 2π × 50 × 5.422 × 10⁻⁹ × 20 = 0.375 A
Answer: Cn = 10.843 nF/km with earth (10.837 without); charging current 0.375 A for 20 km
Earth effect and charging of a 220 kV line · Three-phase, one Zebra (28.62 mm diameter) per phase, flat 7 m spacing, 15 m above earth (illustrative), transposed, 220 kV, 50 Hz
- Deq = ∛(7 × 7 × 14) = 8.819 m; without earth Cn = 55.63/ln(8.819/0.01431) = 8.66 nF/km
- Image distances: Hs = 30 m; H12 = H23 = √(30² + 7²) = 30.81 m, H31 = √(30² + 14²) = 33.11 m; Hm = 31.55 m
- With earth: Cn = 8.729 nF/km, 0.79 % higher
- Charging current = (220 000/√3) × 2π × 50 × 8.729 × 10⁻⁹ = 0.3483 A/km
- Charging reactive power = 3 × 127.02 kV × 34.83 A = 13.27 MVAr per 100 km
Answer: Cn = 8.729 nF/km, 0.348 A/km, 13.3 MVAr per 100 km
Twin-Moose bundle at 400 kV · Three-phase, flat 11 m spacing, ACSR Moose (31.77 mm diameter), twin bundle d = 45 cm (illustrative), earth neglected, 400 kV
- Deq = 1.26 × 11 = 13.86 m
- Single Moose: Cn = 55.63/ln(13.86/0.015885) = 8.216 nF/km
- Twin: bundle radius √(0.015885 × 0.45) = 84.5 mm, Cn = 10.91 nF/km
- Charging current per phase = (400 000/√3) × 2π × 50 × Cn: 0.596 and 0.792 A/km
- Per 100 km: 41.3 and 54.8 MVAr
Answer: Twin Moose: Cn = 10.91 nF/km and 54.8 MVAr per 100 km (41.3 with one Moose)
Short and medium lines: phasor diagrams, ABCD constants, regulation and efficiency
Which syllabus covers thisAU EE3401 Unit IIVTU BEE402 Module 3SPPU 203145 Unit 6MU EEC304 Module 5KTU PCEET501 Module 2
When can a line be modelled as just R + jX, and what does its phasor diagram look like?
When it is short, up to about 80 km by the usual classroom rule: its charging current is then tiny next to the load current. Take the receiving-end phase voltage Vr as the reference. The load current I lags it by φ. The drop I·R is drawn parallel to I and I·X at 90° ahead of it; Vs closes the triangle. CEA's standard data for a 132 kV Panther circuit work out to 0.162 Ω and 0.386 Ω per km, so a 60 km line has R = 9.73 Ω and X = 23.2 Ω.
What are regulation and efficiency, and how can regulation go negative?
Regulation is how much the receiving voltage would rise if the full load were thrown off, as a fraction of the full-load value: (|Vs|/|A| − |Vr|)/|Vr|, with A = 1 for a short line. Efficiency is Pr/(Pr + 3I²R). For a lagging load both drops add to Vr and regulation is positive. A leading current swings I·X backwards, so past a certain lead (here about cos φ = 0.92) Vs is smaller than Vr: on our 60 km line, 25 MW at 0.8 leading gives −1 %.
What are the ABCD constants of a line?
Any line is a two-port: two equations tie the sending-end voltage and current to the receiving-end ones. A and D are ratios (no units), B is an impedance in ohms and C an admittance in siemens. Because the line is passive and the same from both ends, A = D and AD − BC = 1, a useful check on any hand calculation. A short line has A = D = 1, B = Z, C = 0. For our 160 km Panther circuit as a nominal π, A = 0.9856∠0.35°, B = Z = 67 Ω and C = 0.465 mS.
How do the nominal-T and nominal-π models differ, and when do they matter?
Between about 80 and 250 km the charging current is no longer negligible: our 160 km line draws 35.7 A of charging current at 132 kV against a 35 MW load current of 161 A. Nominal π puts half the shunt admittance at each end with the whole Z between them; nominal T splits Z in two halves with all of Y in the middle. Both give A = 1 + ZY/2 but slightly different B and C. They are different approximations of the same line, and at this length their sending voltages differ by only 138 V.
A 60 km, 132 kV line delivers 25 MW at 0.8 leading power factor. Compared with the receiving-end voltage, the sending-end voltage is
With a leading current the reactive drop I·X points backwards and outweighs the resistive drop: |Vs| = 130.68 kV against 132 kV, a regulation of −1 %.
Worked examples
A short 132 kV line at lagging power factor · 132 kV Panther line, 60 km, R = 0.1622 Ω/km, X = 0.3861 Ω/km (CEA standard data); receiving end 132 kV, 40 MW at 0.9 lagging
- R = 9.733 Ω, X = 23.167 Ω per phase
- Vr = 132/√3 = 76.21 kV per phase; I = 40 MW/(√3 × 132 kV × 0.9) = 194.39 A at −25.84°
- I·R = 1.892 kV along I, I·X = 4.504 kV at 90° ahead of I
- Vs = Vr + I·Z = 79.941 kV∠2.31° per phase, 138.46 kV line
- Regulation = (79.941 − 76.21)/76.21 = 4.9 % (the approximate I(R cos φ + X sin φ)/Vr gives 4.81 %)
- Loss 3I²R = 1.103 MW, efficiency = 40/41.103 = 97.32 %
Answer: Vs = 138.5 kV, regulation 4.9 %, efficiency 97.32 %
The same line at leading power factor · Same 60 km line, receiving end 132 kV, 25 MW at 0.8 leading
- I = 25 MW/(√3 × 132 kV × 0.8) = 136.68 A at +36.87°
- Vs = Vr + I·Z = 76.21 + (136.68∠36.87°)(9.733 + j23.167) = 75.448 kV per phase
- Line value 130.68 kV, below the 132 kV at the load
- Regulation = −1 %; the approximate formula I(R cos φ − X sin φ)/Vr gives −1.1 %
- Regulation is zero near cos φ = X/|Z| = 0.922 leading; any more lead makes it negative
Answer: Vs = 130.68 kV, regulation −1 % (negative)
A 160 km line as a nominal π · Same Panther constants, 160 km, B = 2.927 µS/km; receiving end 132 kV, 35 MW at 0.95 lagging
- Z = 25.95 + j61.78 Ω, Y = j0.4683 mS
- A = D = 1 + ZY/2 = 0.98555∠0.353°, B = Z = 67.01∠67.21° Ω, C = Y(1 + ZY/4) = 0.46493∠90.18° mS
- Check: AD − BC = 1 (to 1e−12)
- Ir = 161.14 A at −18.19°; Vs = A·Vr + B·Ir = 82.64 kV per phase = 143.14 kV line
- Is = C·Vr + D·Ir = 151.65∠−5° A
- Regulation = (82.64/0.9856 − 76.21)/76.21 = 10.03 %
- Ps = 3 Re(Vs·Is*) = 36.907 MW, so efficiency = 35/36.907 = 94.83 %; nominal T gives Vs = 143.00 kV
Answer: Vs = 143.14 kV, regulation 10.03 %, efficiency 94.83 %
Long lines: propagation constant, surge impedance, SIL and the Ferranti effect
Which syllabus covers thisAU EE3401 Unit IIVTU BEE402 Module 3SPPU 303148 Unit 1MU EEC304 Module 5KTU PCEET501 Module 3
Why does a long line need distributed parameters, and what is the propagation constant?
Beyond about 250 km a single lumped π starts to drift from the truth: every kilometre has its own series impedance z and shunt admittance y, and voltage and current change along the line as waves. γ = √(zy) = α + jβ per km. For a 400 kV twin Moose line (CEA data: z = 0.0298 + j0.332 Ω/km, y = j3.469 µS/km), α = 4.81×10⁻⁵ Np/km and β = 1.0742×10⁻³ rad/km: one wavelength is 5,849 km and the wave moves at about 292,000 km/s.
What are surge impedance and surge impedance loading?
Zc = √(z/y) is the ratio of voltage to current in a wave travelling one way. Ignoring resistance, Zc = √(x/b) = 309.4 Ω for twin Moose. Terminate the line in a resistance equal to Zc and it delivers its surge impedance loading, SIL = V²/Zc = 400²/309.4 = 517 MW. At SIL the reactive power the line generates (V²b = 0.555 MVAr per km) exactly equals what its inductance absorbs, so the voltage stays flat end to end. Four Moose sub-conductors per phase lower Zc to 235.1 Ω and lift SIL to 681 MW.
What are the exact ABCD constants of a long line?
Solving the line's wave equations gives A = D = cosh γl, B = Zc sinh γl and C = sinh(γl)/Zc, with AD − BC = 1 still. The same answer comes from chaining thousands of short nominal-π sections, which is a good numerical check. The equivalent π uses Z' = Z·sinh(γl)/γl and Y'/2 = (Y/2)·tanh(γl/2)/(γl/2). For 350 km of twin Moose, A = 0.9303∠0.38° and B = 113.9∠84.99° Ω, and the nominal π misses the sending voltage by only 0.96 kV. At 800 km on open circuit it puts the far end at 633 kV instead of 612 kV.
What is the Ferranti effect?
On an open or lightly loaded line the charging current flows through the series inductance and raises the voltage along the way, so the far end sits above the sending end: Vr = Vs/|A|, about Vs/cos βl. Energise 400 km of twin Moose at 400 kV with the far end open and it rises to 439.9 kV. The CEA Grid Standards cap a 400 kV system at 420 kV, which this line reaches at about 289 km. That is why long EHV lines carry shunt reactors at their ends.
A 400 km, 400 kV line with no resistance is terminated in a resistance equal to its surge impedance. Along the line, the voltage magnitude
At surge impedance loading the line's charging (V²b) exactly balances its series reactive loss (I²x) in every kilometre, so there is nothing to push the voltage up or down: |V| is flat. Below SIL it rises towards the load (the Ferranti side), above SIL it sags.
Worked examples
From CEA per-unit data to γ, Zc and SIL · 400 kV twin Moose single circuit: R = 1.862×10⁻⁵, X = 2.075×10⁻⁴, B = 5.55×10⁻³ pu/km on 100 MVA (CEA Table I(a))
- Zbase = 400²/100 = 1600 Ω
- r = 0.02979 Ω/km, x = 0.332 Ω/km, b = 5.55×10⁻³/1600 = 3.4688 µS/km
- γ = √((r + jx)(jb)) = 4.81×10⁻⁵ + j1.0742×10⁻³ per km
- Zc = √((r + jx)/(jb)) = 310∠−2.56° Ω; ignoring r, Zc = √(x/b) = 309.37 Ω
- SIL = 400²/309.37 = 517.2 MW
- Wavelength 2π/β = 5,849 km; charging V²b = 0.555 MVAr per km
Answer: Zc ≈ 309 Ω, SIL ≈ 517 MW
Sending end of a 350 km line by the exact equations · Same line, 350 km; receiving end held at 400 kV delivering 400 MW at unity power factor
- γl = 0.0168 + j0.376
- A = D = cosh γl = 0.9303∠0.381°, B = Zc sinh γl = 113.94∠84.99° Ω, C = sinh γl / Zc = 1.1857∠90.12° mS
- Vr = 230.94 kV per phase, Ir = 400 MW/(√3 × 400 kV) = 577.35 A
- Vs = A·Vr + B·Ir = 230.519 kV∠16.89°, i.e. 399.27 kV line
- Is = C·Vr + D·Ir = 604∠27.34° A (it leads: the line is below SIL and has surplus charging)
- Ss = 3·Vs·Is* = 410.76 MW and −75.79 MVAr; efficiency 400/410.76 = 97.38 %
- Nominal π gives Vs = 400.23 kV
Answer: Vs = 399.3 kV at δ = 16.9°, Ps = 410.8 MW, efficiency 97.4 %
How long can an open 400 kV line be? · Twin Moose line energised at 400 kV, receiving end open; limit 420 kV (CEA Grid Standards)
- Open end: Ir = 0, so Vr = Vs/|A| = 400/|cosh γl|
- 200 km: Vr = 409.4 kV; 400 km: Vr = 439.9 kV (lossless 400/cos βl = 439.9 kV)
- Limit: cos βl = 400/420 gives l = acos(0.9524)/β = 288.7 km ignoring resistance
- With resistance, bisection on |cosh γl| = 400/420 gives 288.7 km
- At 400 km the line still draws Is = |C|·Vr = 341.7 A of charging current, about 237 MVAr from the source
Answer: About 289 km before the open end passes 420 kV; at 400 km it reaches 440 kV
Real and reactive power flow over a line and power circle diagrams
Which syllabus covers thisAU EE3401 Unit IIVTU not in syllabusSPPU 303148 Unit 1MU not in syllabusKTU not in syllabus
What decides how much real power a line carries?
Mainly the angle δ by which the sending voltage leads the receiving voltage. For a line treated as a pure reactance X, P = |Vs||Vr| sin δ / X: zero at δ = 0, largest at 90°. A 250 km twin Moose line has X = 83 Ω, so with 410 kV and 395 kV at its ends it carries 505 MW at δ = 15° and could carry at most 1,951 MW. Real power flows from the leading-angle end to the lagging one.
What decides the reactive power flow?
Mainly the difference in voltage magnitudes. In the same lossless picture the reactive power arriving at the receiving end is Q = |Vr|(|Vs| cos δ − |Vr|)/X: reactive power flows from the higher-voltage end to the lower. In the 250 km example, 410 kV against 395 kV at 15° delivers only 4.9 MVAr even though the sending end is 15 kV higher, because cos δ eats most of the difference. Reactive power does not travel far, which is why it is supplied close to where it is used.
What is the receiving-end power circle?
With the exact ABCD constants, A = |A|∠α and B = |B|∠β, the receiving-end power Pr + jQr = (|Vs||Vr|/|B|)∠(β − δ) − (|A||Vr|²/|B|)∠(β − α). For fixed voltage magnitudes the first term only turns as δ changes, so the operating point moves round a circle: centre at −|A||Vr|²/|B| ∠(β − α), radius |Vs||Vr|/|B|. Real power peaks where the radius is horizontal, at δ = β. For 300 km of twin Moose at 400 kV both ends, β = 84.96° and Pmax = 1,485 MW.
How is the circle used to size compensation, and how does series compensation help?
Draw the load line from the origin at the load's power-factor angle. If the line alone cannot deliver the load's reactive power at the chosen voltages, the vertical gap between the circle and the load line is what a capacitor bank or compensator at the receiving end must supply. For 500 MW at 0.95 lagging over 300 km with both ends at 400 kV, that gap is 206 MVAr. Series capacitors cut the net X instead: 40 % compensation (illustrative) lifts the lossless limit from 1,606 to 2,677 MW.
Two 400 kV buses are both held at 400 kV. To send more real power from one to the other, the operator must mainly
Real power follows the angle: P = |Vs||Vr| sin δ / X. Changing the voltage magnitudes mainly moves reactive power, and resistance is fixed by the conductor.
Worked examples
Power over a line treated as a reactance · 250 km of 400 kV twin Moose (x = 0.332 Ω/km, resistance and charging ignored); |Vs| = 410 kV, |Vr| = 395 kV, δ = 15°
- X = 0.332 × 250 = 83 Ω
- P = 410 × 395 × sin 15° / 83 = 505 MW
- Qr = 395 × (410 cos 15° − 395)/83 = 4.9 MVAr
- Pmax = 410 × 395/83 = 1,951.2 MW, at δ = 90°
Answer: P = 505 MW, Qr = 4.9 MVAr, Pmax = 1,951 MW
The receiving-end circle from exact constants · 300 km twin Moose (exact long-line ABCD), |Vs| = |Vr| = 400 kV, δ = 20°
- A = 0.9486∠0.276°, B = 98.28∠84.96° Ω (so α = 0.276°, β = 84.96°)
- Centre: −|A|Vr²/|B| = 1,544.3 MVA at angle (β − α) + 180°, i.e. (−143, −1,537.7)
- Radius |Vs||Vr|/|B| = 1,628.0 MVA
- Operating point: centre + 1,628.0∠(β − 20°) = 546 MW + j(−62.7) MVAr
- Direct check: Ir = (Vs − A·Vr)/B and Sr = Vr·Ir* give the same 546 MW, −62.7 MVAr
- Pmax = radius + centre's P = 1,484.9 MW at δ = β = 84.96°
Answer: Pr = 546 MW, Qr = −62.7 MVAr; Pmax = 1,485 MW
Compensation needed at the receiving end · Same 300 km line, both ends held at 400 kV, load 500 MW at 0.95 lagging
- P = 500 on the circle: cos(β − δ) = (500 − (−143))/1,628.0, so δ = 18.23°
- The line then delivers Qr = −42.1 MVAr: with resistance in the line, holding |Vr| = |Vs| needs reactive power fed into the line at the receiving end
- The load needs Q = 500 × tan(acos 0.95) = 164.3 MVAr
- Compensator at the receiving end: 164.3 − (−42.1) = 206.5 MVAr (capacitive)
Answer: δ = 18.2°, about 206 MVAr of capacitive support
Corona: disruptive and visual critical voltages, corona loss and how to reduce it
Which syllabus covers thisAU EE3401 Unit IIVTU BEE402 Module 4SPPU 303148 Unit 2MU EEC502 Module 6KTU PCEET501 Module 3
What is corona?
The field at a conductor's surface is E = V/(r ln(D/r)) for a phase voltage V. When it exceeds the breakdown strength of air, about 30 kV/cm peak or 21.1 kV/cm rms at 25 °C and 76 cm of mercury (Peek), the air next to the wire ionises. The field falls as 1/x away from the wire, so the breakdown stays in a thin sheath instead of bridging the gap. The result is a hissing noise, a violet glow, ozone, radio interference and a power loss. A 220 kV Zebra line at 7 m spacing has 14.3 kV/cm rms at the surface, below the limit.
What is the disruptive critical voltage, and what changes it?
It is the phase voltage at which the surface field reaches the strength of air: Vd = 21.1·m0·δ·r·ln(D/r) kV, r and D in cm. δ = 3.92b/(273 + t) is the air density factor, with b in cm of mercury and t in °C. The surface factor m0 is 1 for a polished wire, 0.98-0.93 if weathered and 0.87-0.83 if stranded. In rain, fog or snow, Peek takes Vd ≈ 0.8 of its fair-weather value. Thin air lowers Vd: for Zebra at 7 m, Vd falls from 158.9 kV at sea-level pressure to 125.4 kV at 60 cm.
Why does the glow start above the disruptive voltage?
Ionisation needs the field to stay above the limit over a finite distance, about 0.301√(r/δ) cm from the surface, so a thin wire needs a higher surface field. Peek's visual critical voltage is Vv = 21.1·mv·δ·r·(1 + 0.301/√(δr))·ln(D/r), with mv = 0.72 for local glow at a few spots and 0.82 for glow all along a stranded conductor. For the 220 kV Zebra line on a 35 °C day at 74 cm, Vd = 149.7 kV, local Vv = 159.7 kV and general Vv = 181.8 kV, all phase values.
How large is the corona loss, and how is corona reduced?
Peek's loss is P = (241/δ)(f + 25)√(r/D)(V − Vd)² × 10⁻⁵ kW per km per phase, so it grows with the square of the excess voltage. Peek notes that the loss between Vd and Vv is erratic, depending on dirt and surface condition. One Moose conductor at 400 kV and 11 m spacing would lose 41.1 kW per km of line in fair weather. A single conductor would need a radius of 2.05 cm to stay below Vd. The practical cures are larger conductors, bundles, which lower the surface field, wider spacing (a weak, logarithmic effect) and smooth, clean conductor surfaces.
The same line is built at a hill site where the barometer reads 60 cm of mercury instead of 76 cm, at the same temperature. Its disruptive critical voltage
Vd is proportional to the air density factor δ = 3.92b/(273 + t). At the same temperature δ scales with pressure, 60/76 = 0.789. Thinner air breaks down at a lower field, which is why high-altitude lines need larger conductors.
Worked examples
Does a 220 kV line go into corona? · 220 kV three-phase, one Zebra (28.62 mm diameter, r = 1.431 cm) per phase, equilateral spacing 7 m (illustrative), 35 °C, 74 cm of mercury, stranded m0 = 0.85
- δ = 3.92 × 74/(273 + 35) = 0.9418
- Vd = 21.1 × 0.85 × 0.9418 × 1.431 × ln(700/1.431) = 149.7 kV per phase (259.3 kV line)
- Operating phase voltage = 220/√3 = 127 kV: below Vd, no corona in fair weather
- In foul weather Vd ≈ 0.8 × 149.7 = 119.8 kV, below 127 kV, so corona starts
- Storm loss = (241/0.9418) × 75 × √(1.431/700) × (127 − 119.8)² × 10⁻⁵ = 0.458 kW/km per phase
Answer: No corona in fair weather (Vd = 149.7 kV per phase); in storms about 1.37 kW per km of line
Visual critical voltages of the same line · The 220 kV Zebra line above: δ = 0.9418, r = 1.431 cm, D = 700 cm
- √(δr) = √(0.9418 × 1.431) = 1.1609; 1 + 0.301/1.1609 = 1.2593
- Local glow (mv = 0.72): Vv = 21.1 × 0.72 × 0.9418 × 1.431 × 1.2593 × 6.1927 = 159.7 kV
- General glow (mv = 0.82): Vv = 181.8 kV
- Line values: 277 kV and 315 kV, well above 220 kV
Answer: Vv = 159.7 kV (local) and 181.8 kV (general) per phase
Why 400 kV lines use bundles · One ACSR Moose (r = 1.5885 cm) per phase, equivalent spacing 11 m (illustrative), 25 °C and 76 cm (δ = 1), m0 = 0.85, 400 kV, 50 Hz
- δ = 3.92 × 76/298 = 0.9997
- Vd = 21.1 × 0.85 × 0.9997 × 1.5885 × ln(1100/1.5885) = 186.3 kV per phase
- Phase voltage = 400/√3 = 230.9 kV: above Vd, so the line is in corona even in fair weather (surface field 22.2 kV/cm rms)
- Loss = (241/0.9997) × 75 × √(1.5885/1100) × (230.9 − 186.3)² × 10⁻⁵ = 13.7 kW/km per phase
- A single conductor would need r ≈ 2.05 cm for Vd to reach 230.9 kV
Answer: Vd = 186.3 kV < 230.9 kV: about 41.1 kW per km of line, so a bundle is used
Sag and tension: supports at equal and unequal levels, wind and ice loading
Which syllabus covers thisAU EE3401 Unit IIIVTU BEE402 Module 1SPPU 203145 Unit 3MU not in syllabusKTU not in syllabus
Why does a line sag, and by how much?
A conductor between two towers settles where its tension balances its weight. When the sag is small next to the span, the curve is a parabola and the mid-span sag is S = wL²/8T. Panther ACSR (IS 398) weighs 974 kg/km, which is 9.55 N/m, and is rated at 89.67 kN. Strung at a factor of safety of 2 (T = 44.8 kN) across 300 m, it sags 2.4 m. Double the span and the sag goes up four times; double the tension and it halves.
Is the parabola the true shape?
No: a uniform cable hangs as a catenary, y = c(cosh(x/c) − 1) with c = T/w. For the 300 m Panther span c = 4692 m and the exact sag is 2.3977 m against the parabola's 2.3975 m, a difference of 0.2 mm. The error grows roughly as (S/L)², so the catenary matters only for very long or very slack spans, such as river crossings. The tension is lowest at the lowest point and rises to T + wS at the supports.
What changes when one tower stands higher than the other?
The lowest point slides towards the lower tower: x1 = L/2 − Th/(wL) from the lower support, x2 = L − x1 from the higher one, and each side is half a parabola, S1 = wx1²/2T and S2 = wx2²/2T, with S2 − S1 = h. For the Panther span with the far tower 5 m higher, x1 = 71.8 m, S1 = 0.55 m and S2 = 5.55 m. Once h exceeds wL²/2T (9.59 m here), x1 turns negative: the conductor climbs all the way from the lower tower, which then carries little of the span's weight.
How do wind and ice change the sag?
Ice adds weight: a radial layer t on a conductor of diameter d weighs 917 × 9.81 × πt(d + t) N/m. Wind pushes sideways with pressure × (d + 2t) per metre. The conductor swings into the plane of the resultant w = √((wc + wi)² + ww²), sags by wL²/8T in that plane, and its vertical sag is S cos θ. A Dog conductor with 10 mm of ice and 400 N/m² of wind carries 17.3 N/m instead of 3.87 N/m, so its sag is 4.5 times the bare value.
Why not pull the conductor tight and make the sag tiny?
Tension and sag trade against each other, and the conductor must survive its worst day: ice, high wind and the lowest temperature, when the wire contracts and its tension rises. The working tension is therefore a fraction of the rated strength, set by a factor of safety. Panther at a factor of 2 runs at 44.8 kN and sags 2.4 m on 300 m; at a factor of 4 it runs at 22.4 kN and sags 4.8 m, which needs taller towers for the same ground clearance.
A conductor sags 0.6 m across a 150 m span. With the same conductor and the same tension across a 300 m span, the sag is about
Sag goes as the square of the span (S = wL²/8T): twice the span gives four times the sag, 2.4 m. This is why tower height, not just tower count, drives the cost of a longer span.
Worked examples
A level Panther span, parabola and catenary · Panther ACSR (IS 398: 974 kg/km, rated strength 89.67 kN), span 300 m, supports level, factor of safety 2
- Weight w = 0.974 × 9.81 = 9.555 N/m; working tension T = 89.67/2 = 44.835 kN
- Parabola: S = wL²/8T = 9.555 × 300²/(8 × 44835) = 2.398 m
- Conductor length = L + 8S²/3L = 300.051 m, only 5.1 cm more than the span
- Catenary: c = T/w = 4692.3 m, S = c(cosh(L/2c) − 1) = 2.3977 m
- Tension at the supports = T + wS = 44.86 kN
Answer: Sag 2.4 m; the exact catenary adds only 0.2 mm
Dog conductor under ice and wind · Dog ACSR (IS 398: 394 kg/km, diameter 14.15 mm, rated strength 32.41 kN), span 200 m, factor of safety 2.5, radial ice 10 mm, wind 400 N/m² on the iced conductor (ice and wind illustrative)
- Own weight wc = 0.394 × 9.81 = 3.865 N/m
- Ice wi = 917 × 9.81 × π × 0.010 × (0.01415 + 0.010) = 6.825 N/m
- Wind ww = 400 × (0.01415 + 2 × 0.010) = 13.66 N/m
- Resultant w = √(10.69² + 13.66²) = 17.35 N/m, at θ = 52° from the vertical
- T = 32.41/2.5 = 12.964 kN; slant sag S = 17.35 × 200²/(8 × 12964) = 6.69 m
- Vertical sag = S cos θ = 4.12 m (bare and still air: 1.49 m)
Answer: Slant sag 6.69 m at 52°, vertical sag 4.12 m
Towers at different heights · The Panther span of the first example (300 m, T = 44.835 kN, w = 9.555 N/m) with the far tower 5 m higher
- x1 = L/2 − Th/(wL) = 150 − 44835 × 5/(9.555 × 300) = 71.79 m from the lower tower
- x2 = 300 − 71.79 = 228.21 m
- S1 = wx1²/2T = 0.549 m; S2 = wx2²/2T = 5.549 m; S2 − S1 = 5 m = h, as it must
- Exact catenary through both supports (solved numerically): lowest point at 71.81 m
- With a 10 m difference x1 = −6.4 m: the lowest point falls outside the span, because 10 m > wL²/2T = 9.59 m
Answer: Lowest point 71.8 m from the lower tower; sags 0.55 m and 5.55 m
Overhead line insulators: voltage across a string, string efficiency and grading
Which syllabus covers thisAU EE3401 Unit IIIVTU BEE402 Module 1SPPU 203145 Unit 3MU EEC304 Module 2KTU PCEET501 Module 3
Which insulators hold up an overhead line?
Pin insulators sit on the cross-arm and suit lower voltages, because one large porcelain piece becomes heavy and costly as the voltage rises. Higher voltages hang the conductor from a suspension string of identical discs linked by metal caps and pins, so the voltage rating grows by adding discs. OPTCL, Odisha's transmission utility, specifies 9 discs of 255 × 145 mm for a 132 kV suspension string and 14 for 220 kV. Strain (tension) strings take the full pull of the conductor at dead ends and sharp angles.
Why don't the discs of a string share the voltage equally?
Each disc is a capacitor C between its cap and pin, but every metal link also has a small capacitance mC to the earthed tower. Charging current leaks to the tower at each link, so the discs nearer the line carry more current and so more voltage. On a 132 kV line (76.2 kV to earth) with 9 discs and m = 0.1, the top disc carries 2.88 kV and the line-end disc 20.8 kV, against an equal share of 8.47 kV. The line-end disc fails first.
What is string efficiency?
It compares the string with an ideal one: η = V/(n × voltage across the line-end disc), where V is the voltage to earth and n the number of discs. Equal sharing gives 100%. The 9-disc string above, with m = 0.1, reaches only 40.8%. Adding discs helps less and less, because each extra disc lowers the efficiency: with m = 0.1, three discs give 86.8% and nine give 40.8%. A smaller m (longer cross-arm) raises it: 53% at m = 0.05.
How can the voltage be shared more evenly?
Three ways. A longer cross-arm moves the string away from the tower and lowers m. Capacitance grading uses discs of larger capacitance towards the line, C_k = C(1 + m k(k − 1)/2), so each disc's extra current is matched by its extra capacitance; with m = 0.15 four discs need 1, 1.15, 1.45 and 1.9 times C. A guard ring around the line end adds capacitance from the line to each link and cancels part of the leakage: a ring giving 60% of the ideal values lifts the 9-disc string from 41% to 74%.
A three-disc string (m = 0.1) is lengthened to six discs for a higher voltage. Its string efficiency
More links leak more charging current to the tower, so the line-end disc's share falls more slowly than 1/n. The efficiency drops from 86.8% with three discs to 58.5% with six.
Worked examples
A three-disc string · A 33 kV line on a string of three discs; capacitance of each link to the tower = 0.2 × the disc capacitance (illustrative)
- Voltage to earth V = 33/√3 = 19.053 kV
- Take V1 = 1 unit at the top. Link 1: V2 = V1 + 0.2V1 = 1.2
- Link 2: V3 = V2 + 0.2(V1 + V2) = 1.2 + 0.2 × 2.2 = 1.64
- Sum 3.84 units = 19.053 kV, so one unit = 4.962 kV
- V1 = 4.96 kV, V2 = 5.95 kV, V3 = 8.14 kV
- η = 19.05/(3 × 8.137) = 78%
Answer: Line-end disc 8.14 kV; string efficiency 78%
A 132 kV string of nine discs · 132 kV line, 9 discs (as OPTCL specifies for 132 kV suspension), m = 0.1 (illustrative)
- V = 132/√3 = 76.21 kV
- Relative voltages from the top, Vk+1 = Vk + 0.1 × (sum so far): 1, 1.1, 1.31, 1.651, 2.157, 2.879, 3.889, 5.287, 7.214
- They add to 26.487 units, so one unit = 2.877 kV
- Top disc 2.88 kV; line-end disc 20.76 kV (27.2% of V)
- η = 76.21/(9 × 20.76) = 40.8%
Answer: The line-end disc carries 20.8 kV, 2.5 times an equal share; η = 40.8%
Grading a four-disc string · Four discs, m = 0.15, no disc may see more than 15 kV (both illustrative). Compare plain discs with capacitance-graded discs
- Plain: relative voltages 1, 1.15, 1.15 + 0.15 × 2.15 = 1.4725, 1.4725 + 0.15 × 3.6225 = 2.0159
- Sum 5.6384; the line-end disc reaches 15 kV when V = 15 × 5.6384/2.0159 = 41.95 kV to earth, a 72.7 kV line
- Plain string efficiency = 5.6384/(4 × 2.0159) = 69.9%
- Graded: C_k = C(1 + 0.15 × k(k − 1)/2) gives 1, 1.15, 1.45 and 1.9 C, and every disc then carries V/4
- So V can rise to 4 × 15 = 60 kV to earth, a 103.9 kV line
Answer: Grading raises the line voltage the string can hold from 72.7 kV to 103.9 kV
Underground cables: capacitance, dielectric stress, economical size and grading
Which syllabus covers thisAU EE3401 Unit IVVTU BEE402 Module 4SPPU 203145 Unit 2MU EEC304 Module 6KTU PCEET501 Module 3
What sets the capacitance of a single-core cable?
Core, insulation and earthed sheath form a cylindrical capacitor: C = 2πε0εr/ln(R/r), with r the radius over the conductor screen and R over the insulation. LS Cable's 36/66 kV, 400 mm² XLPE cable has r = 12.6 mm and R = 23.6 mm and lists 0.23 µF/km, which implies εr ≈ 2.6 for XLPE. At 36 kV and 50 Hz that capacitance draws 2.61 A of charging current per km, so 20 km of cable needs about 52 A just to charge it, before any load.
Where is the insulation stressed most?
The electric field in the insulation, g = V/(x ln(R/r)), falls as 1/x, so the layer touching the core works hardest and the outer layers are lightly used. In the 66 kV cable above (36 kV core to sheath) the stress is 4.55 kV/mm rms at the core and 2.43 kV/mm at the outside, 6.44 kV/mm at the peak of the wave. Breakdown starts where the stress is highest, so g at the core sets the insulation thickness.
What is the most economical core size?
For a given voltage V and permissible stress, make R as small as possible. g at the core is V/(r ln(R/r)); for a fixed R this is least when ln(R/r) = 1, that is r = R/e and R = eV/g. A 33 kV cable (19.05 kV to earth) stressed to 4 kV/mm gives r = 4.76 mm and an insulation of 8.18 mm, close to the 8.8 mm IS 7098-2 sets for 19/33 kV. In practice the current rating often demands a fatter core than this.
How does grading even out the stress?
Intersheath grading puts thin metal sheaths inside the insulation, each held at a chosen potential, so every layer starts again at a high stress and the core layer is relieved. One intersheath at the best radius cuts the LS cable's peak stress from 4.55 to 3.84 kV/mm. Capacitance grading instead layers materials with εr falling outwards, ideally εr ∝ 1/x for a uniform stress. Both are hard to make, and modern extruded cables rely on smooth semiconducting screens and thicker insulation.
A cable has R = 20 mm. At the same voltage, the core radius is raised from 5 mm to 10 mm. The maximum stress
g max = V/(r ln(R/r)) and 5 × ln 4 = 6.93 = 10 × ln 2, so the stress is the same. Between them, at r = R/e = 7.36 mm, it is lowest: the most economical core size.
Worked examples
Reading a 66 kV cable datasheet · LS Cable 36/66 kV, 400 mm² XLPE: conductor 23.2 mm, conductor screen 1.0 mm, insulation 11.0 mm, capacitance 0.23 µF/km
- r = 23.2/2 + 1.0 = 12.6 mm; R = 12.6 + 11.0 = 23.6 mm; ln(R/r) = 0.6275
- εr = C ln(R/r)/(2πε0) = 2.59 (the rounded 0.23 allows 2.54 to 2.65); take 2.6
- Charging current per km = ωCV = 314.16 × 0.2305 µF × 36 kV = 2.61 A
- Stress at the core = 36/(12.6 × 0.6275) = 4.55 kV/mm; at the sheath 2.43 kV/mm
- Best core radius for this R would be R/e = 8.68 mm: the 400 mm² core is larger, because it is sized for current
Answer: εr ≈ 2.6, 2.61 A/km charging current, 4.55 kV/mm at the core
Most economical size for a 33 kV cable · Single-core cable for a 33 kV system, permissible stress 4 kV/mm rms (illustrative)
- Core to sheath V = 33/√3 = 19.05 kV
- r = V/g = 19.05/4 = 4.76 mm, so that ln(R/r) = 1
- R = e × r = 12.95 mm; insulation = R − r = 8.18 mm
- Check: g at the core = 19.05/(4.76 × 1) = 4 kV/mm
- With the 8.8 mm IS 7098-2 thickness on the same core the stress falls to 3.82 kV/mm
Answer: r = 4.76 mm, R = 12.95 mm, insulation 8.18 mm
One intersheath · Core to sheath 36 kV, r = 8 mm, R = 24 mm, an intersheath at 14 mm (radii illustrative)
- Ungraded: g max = 36/(8 ln 3) = 4.096 kV/mm
- Graded, equal peak stress in both layers: G = V/(r ln(r1/r) + r1 ln(R/r1)) = 36/12.023 = 2.994 kV/mm
- Inner layer voltage = G r ln(r1/r) = 13.41 kV; outer layer = 22.59 kV
- So the intersheath is held at 22.59 kV above the sheath
- The best radius solves r/r1 + ln(R/r1) = 1: r1 = 15.03 mm, giving 2.98 kV/mm
Answer: Peak stress falls from 4.1 to 2.99 kV/mm (27% less); intersheath at 22.6 kV
Distribution systems: feeders, distributors with concentrated loads and Kelvin's law
Which syllabus covers thisAU EE3401 Unit VVTU BEE402 Module 5SPPU not in syllabusMU not in syllabusKTU PCEET501 Module 3
What are feeders, distributors and service mains?
A feeder carries power from a substation to a point in the area it serves and has no loads tapped along it, so it is sized for current. A distributor runs past the consumers, with loads tapped all along it, so it is sized for voltage drop: India's supply is nominally 230 V single-phase and 400 V three-phase, and every consumer must get close to that. A service main is the short connection from the distributor to one consumer. Radial systems are cheap; ring mains and distributors fed from both ends keep supply and voltage up.
How do you find the voltages along a distributor fed at one end?
Work back from the far end: each section carries the sum of all the loads beyond it, and its drop is that current × the loop resistance of the section (both wires). The drops add up from the feeding point. A 500 m Rabbit ACSR distributor (loop resistance 2 × 0.5524 = 1.105 Ω/km) with 30, 40 and 25 A at 100, 250 and 400 m falls from 230 V to 204.6 V at the last load, 11% low, because the first section carries all 95 A.
What changes when the distributor is fed from both ends?
Each end supplies part of every load. With both ends at the same voltage, end A supplies each load in proportion to its distance from B; any difference VA − VB adds a circulating current (VA − VB)/(rL). Somewhere the current changes direction, and that load is fed from both sides and has the lowest voltage. The same 500 m distributor fed at 230 V from both ends draws 49 A from A and 46 A from B, and its lowest voltage is 221.4 V at 250 m.
What if the load is spread evenly along the distributor?
With i amperes per metre, the current at distance x from a single feed is i(L − x), and the drops add to irL²/2 at the far end, the same as the whole load iL taken at the middle. Fed from both ends at equal voltage, each half acts as a separate distributor of length L/2 and the largest drop, at the centre, is irL²/8, one quarter. For 0.1 A/m over 500 m of Rabbit: 13.8 V from one end, 3.45 V from both.
What is Kelvin's law?
A thicker conductor costs more to buy (interest and depreciation grow with its cross-section a) but wastes less energy (loss falls as 1/a). The total annual cost A/a + Ba is least where the two parts are equal: the most economical section is the one whose annual cost of energy lost equals the annual cost of the capital that depends on its size. The law ignores voltage drop, current rating and standard sizes, so the answer is rounded up to the nearest conductor that passes those checks.
A uniformly loaded distributor fed at one end has a largest drop of 12 V. Fed from both ends at the same voltage, the largest drop becomes
From one end the drop is irL²/2. From both ends each half is a distributor of length L/2 fed at one end, so the drop is ir(L/2)²/2 = irL²/8, a quarter: 3 V, at the centre.
Worked examples
Concentrated loads, fed at one end · Two-wire distributor, 500 m of Rabbit ACSR (IS 398: 0.5524 Ω/km at 20 °C per wire), fed at 230 V at A; loads of 30 A, 40 A and 25 A at 100 m, 250 m and 400 m (illustrative), unity power factor
- Loop resistance r = 2 × 0.5524 = 1.1048 Ω/km
- Section currents from A: 30 + 40 + 25 = 95 A, then 65 A, then 25 A; the last 100 m carries nothing
- Drops: 95 × 0.11048 = 10.5 V, 65 × 0.16572 = 10.77 V, 25 × 0.16572 = 4.14 V
- Voltages: 219.5 V at 100 m, 208.73 V at 250 m, 204.59 V at 400 m and at B
Answer: Lowest voltage 204.6 V, 11% below 230 V
The same distributor fed from both ends · The distributor of the first example, now also fed at B: (a) at 230 V, (b) at 226 V
- (a) IA = Σ I(L − x)/L = 30 × 0.8 + 40 × 0.5 + 25 × 0.2 = 49 A; IB = 95 − 49 = 46 A
- Section currents from A: 49, 19, −21, −46 A: the current reverses at the 40 A load at 250 m
- V at 100 m = 230 − 49 × 0.11048 = 224.59 V; at 250 m = 224.59 − 19 × 0.16572 = 221.44 V
- (b) IA = (230 − 226)/(1.1048e−3 × 500) + 49 = 56.24 A; the lowest point stays at 250 m, now 219.44 V
Answer: (a) 49 A from A, 46 A from B, lowest 221.44 V at 250 m; (b) 56.2 A from A, lowest 219.44 V
Kelvin's law for a two-wire line · A 1 km two-wire aluminium line carrying 40 A for an equivalent 3,000 hours a year at full loss, energy ₹6/kWh, interest and depreciation ₹120 per mm² per km of conductor per year (all costs illustrative); resistivity from Rabbit's datasheet
- Rabbit has 6 aluminium wires of 3.35 mm = 52.88 mm² and 0.5524 Ω/km, so ρ = 0.5524 × 52.88/1000 = 0.02921 Ω·mm²/m
- Energy cost = I²ρ(2000 m)/a × 3000 h × ₹6/1000 = 1,682,701/a rupees a year
- Capital cost = 120 × 2 km × a = 240a rupees a year
- Least total where they are equal: a = √(1,682,701/240) = 83.7 mm²
- Each part is then ₹20,096 a year; the next ACSR size up, Dog (aluminium area 105 mm²), costs only 2.6% more in total
Answer: Most economical section ≈ 84 mm² of aluminium; choose Dog (105 mm²) after checking voltage drop
Power factor improvement and voltage control
Which syllabus covers thisAU EE3401 Unit VVTU not in syllabusSPPU 303148 Unit 1MU not in syllabusKTU PCEET501 Module 3
Why does a low power factor cost money?
For the same kW the current is inversely proportional to the power factor: I = P/(√3·V·cos φ). On a 415 V supply, 200 kW at 0.7 lagging draws 397 A; at 0.95 it draws 293 A. Everything upstream (cables, transformer, switchgear) is sized for current, and the I²R loss in it falls by 1 − (0.7/0.95)² = 46 %. The lagging kVAR mostly comes from induction motors and transformers drawing magnetising current.
How big a capacitor bank is needed, and should it be in star or delta?
The kW does not change; the bank only cancels part of the lagging kVAR, so Qc = P(tan φ1 − tan φ2). For 200 kW from 0.7 to 0.95 that is 138.3 kVAR. The capacitance per phase follows from Q = 3V²ωC with the voltage each capacitor sees: a star bank on 415/√3 V needs 2,556 µF per phase, a delta bank on the full 415 V needs only a third of that, 852 µF, though each unit must be rated for line voltage.
Why not correct all the way to unity, and what is the most economical power factor?
Each extra kVAR of capacitors saves less kVA than the one before, because kVA flattens out near unity. If maximum demand is billed at x rupees per kVA per year and capacitors cost y rupees per kVAR per year, the total is cheapest at cos φ2 = √(1 − (y/x)²), whatever the starting power factor. With x = 4,800 and y = 900 (illustrative) that is 0.982: going on to unity would cost more than it saves.
How is the voltage at a load held within limits?
The drop along a line is about (PR + QX)/V, and on lines X is larger than R, so reactive power moves voltage most. Utilities use on-load tap changers on transformers; shunt capacitors at heavy load and shunt reactors at light load; synchronous condensers; and static compensators (SVC, STATCOM) that adjust continuously. On a 33 kV line with R = 3 Ω and X = 6 Ω (illustrative), 9 MW at 0.8 lagging pulls the far end down to 30.79 kV; about 6.1 MVAr of capacitors there brings it back to 32 kV.
A plant's power factor is raised from 0.7 to 0.95 lagging at the same kW. The I²R loss in its supply cable falls by about
Current scales as 1/pf, so the loss scales as (0.7/0.95)² = 0.543: a saving of 45.7 %.
Worked examples
Sizing a capacitor bank at 415 V · A 415 V, 50 Hz three-phase workshop load of 180 kW at 0.68 lagging, to be raised to 0.96 lagging
- tan φ1 = tan(acos 0.68) = 1.0783, so Q1 = 194.09 kVAR
- tan φ2 = tan(acos 0.96) = 0.2917, so Q2 = 52.5 kVAR
- Qc = 194.09 − 52.5 = 141.59 kVAR
- Line current: 368.3 A before, 260.9 A after; I²R loss down 49.8 %
- Star bank: C = Qc/(ωV²) = 141586/(314.16 × 415²) = 2,617 µF per phase
- Delta bank: C = Qc/(3ωV²) = 872 µF per phase
Answer: 141.6 kVAR: 2,617 µF per phase in star or 872 µF in delta
The most economical power factor · Illustrative: 400 kW at 0.75 lagging; maximum demand charge ₹4,800 per kVA per year; capacitors cost ₹900 per kVAR per year (interest and depreciation)
- sin φ2 = y/x = 900/4,800 = 0.1875, so cos φ2 = √(1 − 0.1875²) = 0.9823
- Q1 = 400 × tan(acos 0.75) = 352.77 kVAR; Q2 = 400 × tan(acos 0.9823) = 76.35 kVAR; bank = 276.4 kVAR
- Annual cost with no bank: 4,800 × 533.33 kVA = ₹25,60,000
- At the optimum: 4,800 × 407.22 + 900 × 276.4 = ₹22,03,438
- Going on to unity: ₹22,37,490, more than at 0.982; a numerical search of the cost curve lands on the same 276.4 kVAR
Answer: Correct to about 0.982 lagging with a 276 kVAR bank, saving ₹3,56,562 a year
Holding the far-end voltage of a 33 kV line · Illustrative 33 kV line, R = 3 Ω, X = 6 Ω per phase, sending end held at 33 kV; load 9 MW at 0.8 lagging. Keep the receiving end at 32 kV with a shunt capacitor there.
- Q = 9 × tan(acos 0.8) = 6.75 MVAr; quick estimate of the drop (PR + QX)/V = (27 + 40.5)/33 = 2.05 kV
- Exact (Vr on the reference): |Vs|² = (Vr + (PR + QX)/Vr)² + ((PX − QR)/Vr)², solved for Vr = 30.789 kV
- For Vr = 32 kV the line may carry only 0.62 MVAr of the load's 6.75 MVAr (found by bisection)
- Capacitor at the load: 6.75 − 0.62 = 6.13 MVAr
- Quick check with ΔV ≈ X·ΔQ/V: ΔQ ≈ (1.211 × 32)/6 = 6.46 MVAr
Answer: Vr = 30.79 kV without support; about 6.1 MVAr of capacitors holds 32 kV
The per-unit system, one-line diagrams and reactance diagrams
Which syllabus covers thisAU EE3501 Unit IVTU BEE601 Module 1SPPU 303148 Unit 3MU EEC304 Module 4KTU PBEET604
What is a per-unit value, and why do power engineers use it?
A per-unit value is the actual value divided by a chosen base. You pick two bases, power (MVA, the same everywhere) and voltage (kV, line to line); base current and base impedance follow. At 100 MVA and 132 kV the base impedance is 132²/100 = 174.24 Ω and the base current is 437.4 A, so a 50 Ω line is 0.287 pu. The payoff: machines of very different sizes have pu reactances in a narrow band, the √3 factors drop out, and ideal transformers disappear from the circuit.
How does the base voltage change across a transformer?
It follows the turns ratio. Choose 11 kV in the generator zone and an 11/132 kV transformer forces 132 kV in the line zone; a 132/33 kV transformer then forces 33 kV at the motor. With bases chosen this way a transformer's pu reactance is the same from either winding: T1's 0.1 pu on 50 MVA is 34.85 Ω seen from 132 kV and 0.242 Ω from 11 kV, both 0.1 pu. Change the base kV in any one zone and watch the other two follow.
How do you move a pu value from the nameplate base to the system base?
A nameplate pu value is on the machine's own rating. Keep the ohms the same and change the base: multiply by (old kV / new kV)² and by (new MVA / old MVA). A 40 MVA, 11 kV generator with X'' = 0.2 pu becomes 0.2 × 100/40 = 0.5 pu on 100 MVA. A 25 MVA motor rated 30 kV sitting in a 33 kV zone needs both factors: 0.15 × (30/33)² × 100/25 = 0.496 pu. Forgetting the voltage factor is the usual exam mistake.
What are the one-line, impedance and reactance diagrams?
A balanced three-phase system is drawn as one line with standard symbols: generators, transformers, lines, loads and breakers. For analysis it becomes an impedance diagram, one phase to neutral with every element in pu. For fault studies it is trimmed to a reactance diagram: resistances, transformer magnetising branches, line charging and static loads are left out because they barely change the fault current, and each machine is an EMF behind its reactance.
A transformer has X = 0.1 pu on its own 50 MVA rating. On a 100 MVA base at the same kV its per-unit reactance is
Its ohms do not change, but the base impedance kV²/MVA halves when the base MVA doubles, so the pu value doubles to 0.2. On a bigger base every element looks bigger.
Worked examples
Reactance diagram on a common base · G: 40 MVA, 11 kV, X'' 0.2 pu; T1: 50 MVA, 11/132 kV, X 0.1 pu; line X 50 Ω; T2: 40 MVA, 132/33 kV, X 0.08 pu; motor M: 25 MVA, 30 kV, X'' 0.15 pu. Base 100 MVA, 11 kV in the generator zone
- Zone bases: 11 kV, 11 × 132/11 = 132 kV, 132 × 33/132 = 33 kV
- G: 0.2 × 100/40 = 0.5 pu; T1: 0.1 × 100/50 = 0.2 pu
- Line: base impedance 132²/100 = 174.24 Ω, so 50/174.24 = 0.287 pu
- T2: 0.08 × 100/40 = 0.2 pu; M: 0.15 × (30/33)² × 100/25 = 0.4959 pu
- Check by ohms on the 132 kV side: total 293.2 Ω ÷ 174.24 Ω = 1.6828 pu
Answer: j0.5, j0.2, j0.287, j0.2, j0.496 pu; 1.683 pu in total from G to M
Generator voltage needed for a motor load · Same system. The motor draws 20 MW at 0.9 pf lagging with 30 kV at its terminals. Find the generator terminal voltage
- Motor voltage: 30/33 = 0.9091 pu; power: 20/100 = 0.2 pu, Q = 0.2 × tan(cos⁻¹0.9) = 0.0969 pu
- Current I = (S/V)* = 0.22 − j0.1066 pu, magnitude 0.2444 pu = 427.7 A in the 33 kV zone
- Series reactance T1 + line + T2 = 0.687 pu
- Vg = V + jX·I = 0.9823 + j0.1511 = 0.9938∠8.75° pu
- In kV: 0.9938 × 11 = 10.93 kV (the same answer by ohms on the 132 kV side)
Answer: About 10.93 kV at the generator, leading the motor voltage by 8.7°
Base current and base impedance in each zone · 100 MVA base with 11, 132 and 33 kV zones. Find base current and base impedance in each, and what 1 pu current means
- Base current = 100 × 10⁶ / (√3 × kV × 10³): 11 kV 5249 A, 132 kV 437.4 A, 33 kV 1750 A
- Base impedance = kV²/MVA: 1.21 Ω, 174.24 Ω, 10.89 Ω
- The ratios are the turns ratios (12 and 4) and their squares (144 and 16)
- So 1 pu current is the same per-unit current everywhere along the chain, though its amperes differ by the turns ratio
Answer: 5249 A / 1.21 Ω, 437.4 A / 174.24 Ω, 1750 A / 10.89 Ω
Network graphs and the bus admittance matrix
Which syllabus covers thisAU EE3501 Unit IVTU BEE703 Module 1SPPU 303148 Unit 3MU EEC702 Module 4KTU PBEET604
What do tree, link, cut-set and incidence matrix mean for a network?
Draw each element (line, transformer) as a segment between its two nodes: that is the graph. A tree joins every node with no closed loop; with n nodes it has n − 1 branches, and each element left over is a link that closes one loop. Each tree branch also defines a fundamental cut-set, the elements you must cut to split the graph there. The test network has 4 buses and 5 lines: a tree of 3 lines, 2 links, 2 loops, 3 cut-sets. The bus incidence matrix A has a row per element and a column per bus: +1 at the element's start, −1 at its end.
How is Ybus written down by inspection?
Turn each line's impedance into an admittance y = 1/z. A diagonal entry Yii is the sum of the admittances of every element touching bus i (including shunts such as half the line charging); an off-diagonal Yij is minus the admittance joining i and j, and 0 if no line joins them. In the test network line 1-2 has z = 0.02 + j0.1 pu, so y = 1.923 − j9.615 pu, and Y11 = y12 + y13 = 3.884 − j17.459 pu. Switch lines off and on to see which entries move.
What is the singular transformation method?
List the elements' own (primitive) admittances on the diagonal of a matrix y; the elements are not yet connected. The incidence matrix A says how they connect to the buses. Then Ybus = Aᵀ·y·A. It gives exactly the matrix you get by inspection, but it is systematic, so programs use it, and it extends to mutually coupled elements, where y gains off-diagonal terms and inspection no longer works.
What properties of Ybus matter in practice?
It is symmetric (for lines and transformers at nominal tap), complex, and sparse: bus i only couples to its neighbours. With no shunts each row sums to zero, so Ybus alone is singular and a reference voltage (the slack bus) is needed. With line charging each row sums to the shunt admittance at that bus. A network of 1,000 buses and 1,500 lines has only 1,000 + 2 × 1,500 = 4,000 non-zero entries out of a million (0.4%), which is why load-flow programs store only the non-zeros.
Line 2-4 of a network is switched out. Which entries of Ybus change?
A line only ever appears in four places: added to the two diagonals it touches and subtracted at the two off-diagonal positions that join its ends. Removing it subtracts the same stamp.
Worked examples
Ybus of the test network by inspection · Lines (pu on 100 MVA): 1-2 0.02 + j0.10, 1-3 0.03 + j0.12, 2-3 0.02 + j0.09, 2-4 0.025 + j0.11, 3-4 0.015 + j0.07; line charging ignored
- Admittances: y12 = 1.923 − j9.615, y13 = 1.961 − j7.843, y23 = 2.353 − j10.588, y24 = 1.965 − j8.644, y34 = 2.927 − j13.659
- Y11 = y12 + y13 = 3.884 − j17.459; Y22 = y12 + y23 + y24 = 6.241 − j28.848
- Y33 = y13 + y23 + y34 = 7.241 − j32.09; Y44 = y24 + y34 = 4.891 − j22.303
- Off-diagonals are −y: Y12 = −1.923 + j9.615, Y34 = −2.927 + j13.659; Y14 = Y41 = 0 (no line)
- Check: every row sums to zero (no shunts), and 14 of the 16 entries are non-zero
Answer: Diagonal 3.88 − j17.46, 6.24 − j28.85, 7.24 − j32.09, 4.89 − j22.3 pu; off-diagonals −y; Y14 = 0
Singular transformation for a 3-bus reactance network · Elements a: bus 1 to 2, j0.2 pu; b: bus 2 to 3, j0.25 pu; c: bus 1 to 3, j0.4 pu. Build Ybus = Aᵀ·y·A
- Incidence matrix A (rows a, b, c; columns buses 1, 2, 3): [1 −1 0; 0 1 −1; 1 0 −1]
- Primitive admittances y = diag(−j5, −j4, −j2.5)
- Aᵀ·y·A: Y11 = −j5 − j2.5 = −j7.5, Y22 = −j5 − j4 = −j9, Y33 = −j4 − j2.5 = −j6.5
- Y12 = +j5, Y23 = +j4, Y13 = +j2.5: the same as inspection
- Each row sums to zero: with no path to ground this Ybus is singular, so one bus must be the voltage reference
Answer: Ybus = j[−7.5 5 2.5; 5 −9 4; 2.5 4 −6.5] pu
Using Ybus once: loads as fixed admittances · Test network (no charging). Bus 1 held at 1.03 pu; loads 1.5 + j0.6 pu at bus 2 and 0.9 + j0.35 pu at bus 3 modelled as admittances (P − jQ at 1 pu); bus 4 has no generation
- Load admittances 1.5 − j0.6 and 0.9 − j0.35 pu are added to Y22 and Y33
- Rows 2-4 of I = Y·V with zero injections give a 3 × 3 complex system: Y' [V2 V3 V4] = −[Y21 Y31 Y41] × 1.03
- Solved by complex Gauss elimination: V2 = 0.9452∠-6.52°, V3 = 0.9495∠-6.04°, V4 = 0.9478∠-6.23° pu
- The loads now take only 134 MW and 81.1 MW, because an admittance load falls with V²
- Real loads draw nearly constant power, which makes the equations non-linear: that is why load flow needs iteration
Answer: V2 ≈ 0.945 pu, V3 ≈ 0.95 pu, V4 ≈ 0.948 pu in one linear solve
Load flow by the Gauss-Seidel method: bus types, iterations and PV buses
Which syllabus covers thisAU EE3501 Unit IIVTU BEE703 Module 2SPPU not in syllabusMU EEC702 Module 4KTU PBEET604
What are slack, PV and PQ buses?
Each bus has four quantities, P, Q, |V| and angle δ, and the load flow fixes two. A load (PQ) bus knows P and Q and solves for |V| and δ. A generator (PV) bus knows P and the voltage its regulator holds, and solves for Q and δ. One slack bus fixes |V| and δ = 0 and supplies whatever the others do not, because the line losses are unknown until the solution is found. This book's test system: bus 1 slack at 1.03 pu, loads of 150 MW + 60 MVAr at bus 2 and 90 MW + 35 MVAr at bus 3, and a 100 MW generator at bus 4 holding 1.02 pu.
How does one Gauss-Seidel update work?
Bus i's power equation Pi − jQi = Vi*·Σ Yij Vj is solved for Vi, giving the update below. Go round buses 2, 3, 4 in turn, always using the newest values (that is the Seidel part); one round is a sweep. From a flat start (every unknown voltage 1∠0) the test system needs 20 sweeps to bring the largest power mismatch under 10⁻⁴ pu (0.01 MW). Each sweep is cheap, but the error only shrinks by a roughly constant factor per sweep.
How is a voltage-controlled (PV) bus handled, and what are Q limits?
Q is not known at a PV bus, so each sweep first computes it from the present voltages, Qi = −Im(Vi*·Σ Yij Vj), uses it in the update, then scales the new voltage back to the set magnitude, keeping only its angle. If the Q needed exceeds the generator's limit, the bus becomes a PQ bus with Q held at that limit and its voltage is allowed to sag. In the test system bus 4 needs 47.9 MVAr to hold 1.02 pu; with a 25 MVAr limit it turns PQ and its voltage falls to 0.997 pu.
What does the acceleration factor do?
Each correction is stretched: V ← V + α(Vnew − V). Because Gauss-Seidel creeps towards the answer, a modest overshoot helps. On the test system α = 1 needs 20 sweeps, α = 1.3 only 9, but α = 1.9 needs 84: too much stretch makes the iteration oscillate. The best α depends on the network and the loading, so it is found by trial.
On the test system, raising the acceleration factor from 1.0 to 1.9 makes Gauss-Seidel
Over-relaxing that far overshoots each time: it needs 84 sweeps instead of 20. A moderate α (1.3) is best here, and every α that converges reaches the same voltages.
Worked examples
The first Gauss-Seidel sweep by hand · Test system with line charging (Y22 = 6.241 − j28.793, Y33 = 7.241 − j32.04, Y44 = 4.891 − j22.273 pu); flat start V2 = V3 = 1∠0, V4 = 1.02∠0, bus 4 PV; α = 1
- V2 = [(−1.5 + j0.6)/1 − Y21·1.03 − Y23·1 − Y24·1.02]/Y22 = 0.9871 − j0.0459 = 0.9882∠-2.66° pu
- V3 uses the new V2: V3 = [(−0.9 + j0.35)/1 − Y31·1.03 − Y32·V2 − Y34·1.02]/Y33 = 0.9974∠-2.29° pu
- Bus 4 is PV: Q4 = −Im(V4*·Σ Y4j Vj) = 0.3732 pu with the newest V2, V3
- V4 from the update = 1.0191∠-0.22°, scaled back to 1.02: V4 = 1.02∠-0.22° pu
- Largest mismatch after this sweep: 0.4465 pu, down from 1.403 pu at the flat start
Answer: After sweep 1: V2 = 0.9882∠-2.66°, V3 = 0.9974∠-2.29°, V4 = 1.02∠-0.22° pu
Slack power, line flow and losses at convergence · The converged solution of the test system with bus 4 as a PV bus
- Voltages: V2 = 0.9834∠-4.48°, V3 = 0.9944∠-3.6°, V4 = 1.02∠-1.79° pu
- Slack: S1 = V1·(Σ Y1j Vj)* = 1.4392 + j0.4691 pu = 143.9 MW + 46.9 MVAr
- Line 1-2 (with half charging at each end): S12 = 0.8582 + j0.3182, S21 = −0.8422 − j0.2785 pu; loss 1.61 MW
- Total losses = generation − load = (143.9 + 100) − (150 + 90) = 3.92 MW, equal to Σ r|I|² over the five lines
Answer: Slack 143.9 MW, losses 3.92 MW, bus 4 supplies 47.9 MVAr
A PV bus hits its reactive limit · Bus 4's generator can supply at most 25 MVAr. Check the PV solution and redo the load flow if needed
- PV solution: bus 4 must supply Q4 = 47.9 MVAr to hold 1.02 pu: over the 25 MVAr limit
- Make bus 4 a PQ bus: P = 1.0 pu, Q = 0.25 pu fixed, |V4| unknown
- Re-solved: V4 = 0.9974∠-1.46°, V2 = 0.971∠-4.39°, V3 = 0.9805∠-3.47° pu
- The slack now supplies 71.2 MVAr instead of 46.9 MVAr, and every voltage is lower
Answer: Bus 4 cannot hold 1.02 pu; at its limit it settles at 0.997 pu
Load flow by the Newton-Raphson and fast decoupled methods
Which syllabus covers thisAU EE3501 Unit IIVTU BEE703 Module 3SPPU 303148 Unit 3MU EEC702 Module 4KTU PBEET604
What equations does Newton-Raphson load flow solve?
In polar form every bus has P and Q written in terms of voltage magnitudes and angle differences (below, V are magnitudes, θij = θi − θj and Yij = Gij + jBij). The unknowns are the angles of all buses except the slack and the magnitudes of the PQ buses. The mismatches are specified minus calculated: ΔP at every non-slack bus and ΔQ at PQ buses. With bus 4 as a PV bus the test system has 3 angle unknowns and 2 magnitude unknowns, so 5 equations.
What is the Jacobian and what does one iteration do?
The Jacobian holds the sensitivities of the mismatches: H = ∂P/∂θ, N = ∂P/∂|V|, M = ∂Q/∂θ, L = ∂Q/∂|V|. Each iteration builds J at the present voltages, solves J·Δx = mismatch and adds Δx. At the flat start of the test system H22 = 29.31 and H23 = −10.59: P at bus 2 is far more sensitive to its own angle than to a neighbour's. J changes every iteration, so it is rebuilt and refactorised each time.
Why does Newton-Raphson need so few iterations?
Close to the solution the error is roughly squared each iteration. On the test system the largest mismatch goes 1.4, 4.0e-02, 1.1e-04, 8.3e-10 pu: 3 iterations to get under 10⁻⁴ pu, where Gauss-Seidel needs 20 sweeps. At 160% load NR still needs 3, Gauss-Seidel 24. Each NR iteration costs more (a sparse linear solve), but the count of iterations barely grows with network size, which is why NR scales to large grids.
What does the fast decoupled method change?
On transmission lines resistance is much smaller than reactance and angle differences are small, so P depends mainly on angles and Q mainly on voltage magnitudes. Dropping the N and M blocks and freezing the rest gives two constant matrices: B' (built from 1/x of the lines) for ΔP/|V| = B'·Δθ and B'' (−Im Ybus over the PQ buses) for ΔQ/|V| = B''·Δ|V|. They are factorised once. Convergence becomes linear: on the test system both methods reach 10⁻⁴ pu in 3 iterations, but to reach 10⁻⁸ pu fast decoupled needs 6 against 3. Each of its iterations is far cheaper.
Near the solution a Newton-Raphson load flow has a mismatch of about 10⁻² pu. After one more iteration expect roughly
Quadratic convergence roughly squares the error each step (times a constant, here about 0.07 or less): 10⁻² becomes about 10⁻⁴, then 10⁻⁸. Gauss-Seidel and fast decoupled only cut the error by a fixed fraction per step.
Worked examples
Jacobian entries at the flat start · Test system, bus 4 PV, flat start (V2 = V3 = 1∠0, V4 = 1.02∠0, V1 = 1.03∠0); Y22 = 6.241 − j28.793 pu
- Calculated injections at the flat start: P2 = −0.097, Q2 = −0.5163 pu
- H22 = −Q2 − B22·|V2|² = 0.5163 + 28.793 = 29.309
- H23 = |V2||V3|(G23 sin θ23 − B23 cos θ23) = −B23 = −10.588; H24 = −8.817
- N22 = P2/|V2| + G22|V2| = 6.144; M22 = P2 − G22|V2|² = −6.338; L22 = Q2/|V2| − B22|V2| = 28.277
- Each entry matches a finite-difference derivative of the power equations
Answer: H22 = 29.31, H23 = −10.59, N22 = 6.14, M22 = −6.34, L22 = 28.28
The first Newton-Raphson iteration · Same flat start. Unknowns θ2, θ3, θ4, |V2|, |V3|
- Mismatches: ΔP2 = −1.403, ΔP3 = −0.7826, ΔP4 = 0.9002, ΔQ2 = −0.0837, ΔQ3 = 0.2085 pu
- Solve the 5 × 5 system J·Δx = mismatch by Gauss elimination
- Δθ2 = −4.37°, Δθ3 = −3.52°, Δθ4 = −1.69°; Δ|V2| = −0.0144, Δ|V3| = −0.0041 pu
- New estimate: V2 = 0.9856∠-4.37°, V3 = 0.9959∠-3.52° pu
- Largest mismatch falls from 1.403 to 4.0e-02 pu; two more iterations reach 8e-10
Answer: After one iteration V2 = 0.9856∠-4.37° pu (converged value 0.9834∠-4.48°)
The fast decoupled P-θ step · Same flat start. B' from the line reactances only: x12 0.10, x13 0.12, x23 0.09, x24 0.11, x34 0.07 pu
- B'22 = 1/0.10 + 1/0.09 + 1/0.11 = 30.202; B'23 = −1/0.09 = −11.111; B'33 = 33.73; B'44 = 23.377
- Right-hand side ΔP/|V|: −1.403, −0.7826, 0.8826
- Solve B'·Δθ = ΔP/|V|: Δθ2 = −4.5°, Δθ3 = −3.56°, Δθ4 = −1.76°
- Newton-Raphson's first step gave −4.37°, −3.52°, −1.69°: close, from a much simpler fixed matrix
- Then a Q-V half step with B''; it reaches 10⁻⁴ pu in 3 iterations like NR, but needs 6 (NR 3) for 10⁻⁸ pu
Answer: Δθ ≈ −4.5°, −3.56°, −1.76° from B' alone
Economic dispatch: incremental cost, generator limits and transmission losses
Which syllabus covers thisAU not in syllabusVTU BEE703 Module 4SPPU not in syllabusMU EEC702 Module 1KTU not in syllabus
What is a unit's cost curve, and what is its incremental cost?
A thermal unit's fuel cost per hour rises with its output, and faster than in proportion, so it is fitted with a quadratic C(P) = a + bP + cP² (₹/h, P in MW). The incremental cost dC/dP = b + 2cP is the cost of one more MWh at that output: a straight line that climbs with P. For the book's illustrative unit U1 (b = 2000, c = 1.0) it is ₹2,000/MWh near zero output and ₹3,000/MWh at its 500 MW maximum. The constant a is the cost of keeping the unit running at all, and never affects how load is shared.
Why should every unit run at the same incremental cost?
If unit A's next MWh costs less than unit B's, moving 1 MW from B to A saves money, so the cheapest split is the one where all incremental costs are equal to a common λ. Each unit then runs at P = (λ − b)/2c, and λ is raised until the outputs add up to the demand: the λ-iteration. With the three illustrative units at 700 MW, λ = ₹2,829/MWh, outputs 414.3, 196.4 and 89.3 MW, total ₹17,54,500/h, which is ₹84,722/h cheaper than an equal three-way split.
What happens when a unit reaches its limit?
It is held at the limit and the rest of the demand is shared at a new, common λ by the units that are still free. At 250 MW the equal-λ rule alone would ask U3 for a negative output, so U2 and U3 sit at their minimums (80 and 50 MW) and U1 takes the remaining 120 MW at λ = ₹2,240/MWh. At 1050 MW it is the other way: U1 is at its 500 MW maximum and its incremental cost (₹3,000/MWh) is below λ. A unit at its maximum is cheaper than λ; one at its minimum is dearer.
How do transmission losses change the rule?
A unit far from the load loses part of every extra MW in the lines, so its power is worth less at the load. Losses are written with B-coefficients, PL = ΣΣ Pi·Bij·Pj, and each unit's incremental cost is multiplied by a penalty factor Li = 1/(1 − ∂PL/∂Pi) before it is compared with λ. A unit with L = 1.079 must be about 7.3 % cheaper at the margin to be loaded the same, so the remote unit is backed off and the generation now covers the demand plus the loss.
Which units should be running at all? (unit commitment)
Dispatch shares load among units that are already on; unit commitment decides which to switch on, hour by hour, taking start-up costs and minimum up and down times into account. The simplest method is a priority list by full-load average cost: for the illustrative units, U1 costs ₹2,524/MWh at full output, U2 ₹2,862 and U3 ₹3,024, so U1 is committed first and U3 last. Dynamic programming or Lagrangian relaxation improve on the list when start-up costs matter.
Demand is 1050 MW and U1 is already at its 500 MW maximum. Demand rises by 10 MW. Where does the extra 10 MW come from, and what happens to λ?
A unit at its limit is held there. The free units, U2 and U3, take the extra 10 MW in the ratio 1/2c (3 : 2), so λ rises by 10/(1/3.2 + 1/4.8) = ₹19.2/MWh.
Worked examples
Two units, no limits reached · Illustrative units: CA = 8000 + 1800P + 1.2P² and CB = 7000 + 2000P + 0.8P² (₹/h, P in MW); demand 600 MW; losses neglected
- Incremental costs: ICA = 1800 + 2.4PA, ICB = 2000 + 1.6PB (₹/MWh)
- Equal λ: PA = (λ − 1800)/2.4 and PB = (λ − 2000)/1.6, with PA + PB = 600
- λ(1/2.4 + 1/1.6) = 600 + 750 + 1250, so λ = 2496 ₹/MWh
- PA = 290 MW, PB = 310 MW
- Cost ₹13,34,800/h, against ₹13,35,000/h for 300 MW each: ₹200/h saved
Answer: λ = ₹2,496/MWh; PA = 290 MW, PB = 310 MW
A unit reaches its maximum · The interactive's three illustrative units (U1: b 2000, c 1.0, max 500 MW; U2: b 2200, c 1.6; U3: b 2400, c 2.4), demand 1050 MW, losses neglected
- Ignoring limits: λ = (1050 + 1000 + 687.5 + 500)/(0.5 + 0.3125 + 0.2083) = 3171.4 ₹/MWh, giving P1 = 585.7 MW
- That is above U1's 500 MW limit, so fix P1 = 500 MW
- U2 and U3 share 550 MW: λ(1/3.2 + 1/4.8) = 550 + 687.5 + 500, λ = 3336 ₹/MWh
- P2 = (3336 − 2200)/3.2 = 355 MW, P3 = (3336 − 2400)/4.8 = 195 MW
- Check: U1's incremental cost at 500 MW is ₹3000/MWh, below λ, so holding it at its maximum is right; total cost ₹28,18,900/h
Answer: P1 = 500 MW (at limit), P2 = 355 MW, P3 = 195 MW, λ = ₹3,336/MWh
Penalty factor for a remote unit · Illustrative: C1 = 2000P1 + P1², C2 = 1900P2 + P2² (₹/h); unit 2 is remote, losses PL = 0.0002·P2² MW; load 400 MW
- ∂PL/∂P2 = 0.0004·P2, so unit 2's rule is (1900 + 2P2) = λ(1 − 0.0004·P2); unit 1's is 2000 + 2P1 = λ
- Balance: P1 + P2 = 400 + 0.0002·P2²; solve for λ by bisection
- λ = 2446.5 ₹/MWh, P1 = 223.3 MW, P2 = 183.5 MW
- Loss = 0.0002 × 183.5² = 6.73 MW; penalty factor L2 = 1/(1 − 0.0004 × 183.5) = 1.079
- Ignoring losses would give P1 = 175 MW, P2 = 225 MW and a 10.12 MW loss left uncovered
Answer: P1 = 223.3 MW, P2 = 183.5 MW, loss 6.73 MW, λ = ₹2,447/MWh
Three-phase short circuit of an alternator: subtransient, transient and steady currents
Which syllabus covers thisAU EE3501 Unit IIIVTU BEE601 Module 2SPPU 303148 Unit 4MU EEC502 Module 1KTU PBEET604
Why does a short-circuited alternator's current fall in three stages?
The fault flux tries to enter the rotor, and every closed rotor circuit fights it by inducing a current. For the first few cycles the damper windings keep the flux out, the machine looks like a small reactance Xd'' and the current is largest. Once the dampers' currents die away (time constant Td'', tens of ms) the field winding still resists: Xd', transient, lasting about a second (Td'). Finally the flux settles and only the synchronous reactance Xd is left. A real 192 MVA, 18 kV machine has Xd'' = 0.25, Xd' = 0.32 and Xd = 1.57 pu on its own base.
How is the AC current's envelope built?
The peak of the symmetrical current is √2E times a sum of three parts: (1/Xd'' − 1/Xd') decaying with Td'', (1/Xd' − 1/Xd) decaying with Td', and the constant 1/Xd. At t = 0 the sum is 1/Xd''; once the first term has died, the envelope traced back to t = 0 lands on √2E/Xd'; far out it is √2E/Xd. Drag Xd'' and the first spike changes; drag Td' and the long middle stretch shrinks or grows. With the default machine the current falls from 4 pu to 0.64 pu RMS.
Where does the DC offset come from, and when is it worst?
Current in an inductive circuit cannot jump, so at the instant of the fault the AC component is cancelled by a DC component of equal and opposite size, which then decays with the armature time constant Ta. If the fault strikes at a voltage zero the AC current would start at its peak, so the offset is largest and the first peak, about half a cycle later, nearly doubles: 1.88 times √2I'' for the default machine. At a voltage maximum there is no offset. The three phases are 120° apart, so at least one always carries a large offset.
Which current does a circuit breaker have to handle?
Two, at different times. The breaker must close onto a fault and survive the first, offset peak: its making current. It must then open a few cycles later and interrupt the current flowing at that instant, mostly the subtransient-to-transient AC current plus what remains of the DC: its breaking current. A 400 kV breaker specified for an Indian substation in 2025 is rated to interrupt 63 kA RMS symmetrical and to make 157.5 kA peak, 2.5 = √2 × 1.77 times as much: room for a first peak 1.77 times the symmetrical peak, the near-doubling that the DC offset causes.
The same alternator is shorted twice: once at the instant phase a's voltage passes zero, once when it is at its peak. Which gives phase a the larger first current peak?
The AC current lags the voltage by 90°, so at a voltage zero it would already be at its peak; a DC offset of the same size cancels it at t = 0 and adds to the next peak. For the default machine the first peak is 1.88 × √2I'' at voltage zero but only 0.95 × √2I'' at voltage peak, the AC envelope having decayed for a quarter cycle.
Worked examples
Fault currents of a real 192 MVA machine · Generator data (Margitová et al., 2019): 18 kV, 192 MVA, Xd = 2.648 Ω, Xd' = 0.547 Ω, Xd'' = 0.42 Ω; three-phase terminal fault from no load at rated voltage
- Base impedance 18²/192 = 1.6875 Ω, so Xd'' = 0.249, Xd' = 0.324, Xd = 1.569 pu
- Phase voltage 18/√3 = 10.392 kV; base current 192/(√3 × 18) = 6.158 kA
- I'' = 10.392/0.42 = 24.74 kA RMS (= 4.02 pu)
- I' = 10.392/0.547 = 19 kA, steady I = 10.392/2.648 = 3.92 kA
- The steady fault current is only 0.64 × rated current (6.16 kA): the synchronous reactance limits it
Answer: I'' = 24.7 kA, I' = 19 kA, I = 3.92 kA
Reactances from an oscillogram · Illustrative 11 kV, 50 MVA alternator shorted from no load at rated voltage. From the trace of one phase: steady envelope 2.3 kA peak; the transient envelope extended back to t = 0 reads 12.0 kA peak; the initial envelope 16.5 kA peak
- Phase voltage 11/√3 = 6.3509 kV; base impedance 11²/50 = 2.42 Ω
- Xd = 6.3509 kV ÷ (2.3/√2 kA) = 3.905 Ω = 1.614 pu
- Xd' = 6.3509 ÷ (12.0/√2) = 0.7485 Ω = 0.309 pu
- Xd'' = 6.3509 ÷ (16.5/√2) = 0.5443 Ω = 0.225 pu
- Peaks are divided by √2 to get RMS before dividing the phase voltage by them
Answer: Xd = 1.61 pu, Xd' = 0.31 pu, Xd'' = 0.225 pu
First peak and the current a breaker sees · The interactive's default machine: Xd'' 0.25, Xd' 0.32, Xd 1.57 pu, Td'' 0.023 s, Td' 0.83 s, Ta 0.25 s, 50 Hz; fault at a voltage zero from 1 pu
- AC envelope at 10 ms: √2[(4 − 3.125)e^(−0.01/0.023) + (3.125 − 0.637)e^(−0.01/0.83) + 0.637] = 5.178 pu
- DC offset at 10 ms: √2 × 4 × e^(−0.01/0.25) = 5.435 pu
- Dense time scan: first peak 10.62 pu at 9.9 ms, 1.88 × √2I''
- At 60 ms (a breaker opening after 3 cycles): AC 3.02 pu RMS, DC 4.45 pu
- Total RMS √(AC² + DC²) = 5.38 pu, 1.78 times the symmetrical value
Answer: First peak 10.62 pu (1.88 × √2I''); at 60 ms 3.02 pu symmetrical, 5.38 pu with DC
Symmetrical faults in a network: fault level, Zbus building and current-limiting reactors
Which syllabus covers thisAU EE3501 Unit IIIVTU BEE703 Module 5SPPU 303148 Unit 4MU EEC502 Module 1KTU PBEET604
What is a bus's fault level, and why is it quoted in MVA?
Fault level = √3 × line voltage × three-phase fault current. Quoting it in MVA folds voltage and current into one number that tells how strong the network is at that point: in per unit it is simply base MVA divided by the Thevenin impedance at the bus. Breakers and busbars are bought to survive it. Equipment for a 400 kV substation in India was specified in 2025 for 63 kA, which at 400 kV is 43,648 MVA; its 220 kV side for 50 kA, 19,053 MVA.
How is Zbus built one branch at a time?
Start empty and add branches. A branch x from the reference to a new bus adds a row and column with x on the diagonal. From an existing bus j to a new bus, copy row and column j and put Zjj + x on the diagonal. From an existing bus to the reference, the size stays the same and every entry is reduced: Z − Zcol·Zrow/(Zjj + x). A link between existing buses i and j is the same with column i minus column j. The matrix that results equals the inverse of Ybus, without ever inverting.
How does Zbus give the fault current and every bus voltage at once?
Before the fault every bus is at Vf (1 pu at no load). A fault at bus k through Zf draws If = Vf/(Zkk + Zf), and each bus voltage falls by Zik·If, so column k of Zbus is all that is needed. In the interactive's network on 100 MVA, a bolted fault at bus 2 draws 4.02 pu, 402 MVA or 1.76 kA at 132 kV, and buses 1 and 3 sag to 0.58 and 0.52 pu. With only G1 connected the fault at bus 1 was 5 pu; the full network makes it 7.08 pu.
What do current-limiting reactors do, and where are they put?
A reactor is an air-cored coil whose reactance adds to the Thevenin impedance, cutting the fault current to what the switchgear can break. It sits in series with a generator, in a feeder, or between bus sections, where it lets two sections share load normally but keeps each from feeding the other's fault in full. The price is a voltage drop and some reactive loss at full load, so reactors are sized to the smallest value that does the job, typically a few per cent on the circuit's rating.
A link (a new line between two buses already in Zbus) is added. What happens to the matrix?
No new bus is created, so the size stays the same, but the link opens a new path between the two buses and the Kron reduction subtracts a term from every entry that involves either bus's column. In the interactive, step 5 changes all nine entries.
Worked examples
Fault level at a 33 kV bus · Illustrative substation: 132 kV grid with 2000 MVA fault level; two 20 MVA, 132/33 kV transformers in parallel, each 10 % reactance; fault on the 33 kV bus; 100 MVA base
- Grid: 100/2000 = 0.05 pu
- Each transformer: 0.10 × 100/20 = 0.5 pu; two in parallel 0.25 pu
- Total 0.3 pu, so fault level = 100/0.3 = 333.3 MVA
- Fault current = 333.3 MVA/(√3 × 33 kV) = 5.83 kA, well inside a 31.5 kA rating
- With one transformer out: 100/0.55 = 181.8 MVA
Answer: 333 MVA, 5.83 kA
Zbus by all four modifications · Illustrative 2-bus network (pu): generator j0.25 at bus 1, line j0.15 from bus 1 to new bus 2, generator j0.3 at bus 2, then a second line j0.2 between buses 1 and 2; bolted fault at bus 2
- Type 1: Z = [0.25]
- Type 2 (bus 2 from bus 1): Z = [[0.25, 0.25], [0.25, 0.40]]
- Type 3 (bus 2 to reference, 0.3): subtract column 2 × row 2 / (0.40 + 0.3): Z11 = 0.1607, Z12 = 0.1071, Z22 = 0.1714
- Type 4 (link 1-2, 0.2): divisor Z11 + Z22 − 2Z12 + 0.2 = 0.3179; Z11 = 0.1517, Z12 = 0.118, Z22 = 0.1584
- Check: the same matrix comes from inverting Ybus = [[1/0.25 + 1/0.15 + 1/0.2, −(1/0.15 + 1/0.2)], [−(1/0.15 + 1/0.2), 1/0.3 + 1/0.15 + 1/0.2]]
- Fault at bus 2: If = 1/0.1584 = 6.312 pu; V1 = 1 − 0.118 × 6.312 = 0.255 pu
Answer: Z11 = j0.1517, Z12 = j0.118, Z22 = j0.1584 pu; If = 6.31 pu, V1 = 0.26 pu
Sizing a feeder reactor · Illustrative 11 kV bus with a fault level of 500 MVA; a 10 MVA outgoing feeder's switchgear must not see more than 15 kA for a fault just beyond its reactor
- Source impedance 11²/500 = 0.242 Ω per phase (alone it would allow 26.2 kA)
- Allowed total: (11/√3 kV)/15 kA = 0.4234 Ω
- Reactor = 0.4234 − 0.242 = 0.1814 Ω
- On the feeder's 10 MVA, 11 kV base (12.1 Ω): 1.5 %, so at most about 1.5 % voltage drop at full load
Answer: About 0.181 Ω (1.5 % on 10 MVA)
Symmetrical components, sequence impedances and sequence networks
Which syllabus covers thisAU EE3501 Unit IVVTU BEE601 Module 3SPPU 303148 Unit 5MU EEC502 Module 2KTU PBEET604
What are symmetrical components?
Any three phasors, however unbalanced, can be written as the sum of three balanced sets. The positive-sequence set has equal phasors 120° apart in the normal order a, b, c. The negative-sequence set is the same but in the order a, c, b. The zero-sequence set is three equal phasors all in phase. The operator a = 1∠120° does the rotating: 1 + a + a² = 0 and a³ = 1. Unbalanced faults and loads become three separate balanced problems, one per sequence.
How do you find the three components of a given set?
Add the phasors after rotating them. V0 is the plain average. V1 rotates Vb by +120° and Vc by +240°, which lines up the positive set and cancels the others; V2 uses the opposite rotations. Open phase a of a balanced 1 pu set: V1 = 2/3, V2 and V0 are both 1/3 in size, so the unbalance factor V2/V1 is 50 %. The interactive shows the three sets and rebuilds each phase tip to tail as V0 + V1 + V2.
What are sequence impedances, and why do they differ?
A sequence impedance is the voltage-to-current ratio when only that sequence flows. For a line or a transformer the positive and negative impedances are equal, because the order of the phases does not matter to static coils. The zero-sequence impedance is usually much larger for a line, since the three in-phase currents must return through the earth and ground wires. A rotating machine has a smaller negative-sequence reactance, because that field spins backwards against the rotor. Power splits cleanly too: total power = 3(V0I0* + V1I1* + V2I2*).
How do transformer connections and neutral grounding change the zero-sequence network?
Zero-sequence current needs a path to ground on both sides. A grounded-star to grounded-star transformer passes it through. A delta winding lets it circulate inside the delta but blocks it from the line beyond. An ungrounded star is an open circuit. A neutral grounded through Xn appears as 3Xn, because all three phase currents add up in it. Example: generator X0 = 0.06 pu grounded through Xn = 0.04 pu gives 0.18 pu at its terminals. In a Yd unit the positive set also shifts by 30° one way and the negative set the other way.
One phase of a balanced 1 pu set is disconnected, so Va = 0 while Vb and Vc stay as they were. The sequence voltages are
V0 = (0 + a² + a)/3 = −1/3, V1 = (0 + a·a² + a²·a)/3 = 2/3 and V2 = (0 + a²·a² + a·a)/3 = (a + a²)/3 = −1/3. Check: V0 + V1 + V2 = 0 = Va. Losing one phase leaves two thirds of the positive sequence and creates equal negative and zero sequence.
Worked examples
Sequence voltages of an unbalanced set · Va = 1.0∠0°, Vb = 0.8∠−125°, Vc = 0.9∠115° pu
- V0 = (Va + Vb + Vc)/3 = 0.076∠44.9° pu
- aVb = 0.8∠−5° and a²Vc = 0.9∠355°, so V1 = (Va + aVb + a²Vc)/3 = 0.899∠−3.1° pu
- a²Vb = 0.8∠115° and aVc = 0.9∠235°, so V2 = (Va + a²Vb + aVc)/3 = 0.049∠−4.8° pu
- Check: V0 + V1 + V2 = 1∠0° = Va
- Unbalance factor V2/V1 = 0.0487/0.8992 = 5.4 %
Answer: V1 = 0.899∠−3.1°, V2 = 0.049∠−4.8°, V0 = 0.076∠44.9° pu; unbalance 5.4 %
Neutral current from the zero sequence · A four-wire feeder carries Ia = 120∠0° A, Ib = 80∠−110° A, Ic = 100∠135° A
- Ia + Ib + Ic = 22.4∠−11.5° A
- I0 = one third of that = 7.46∠−11.5° A, so the neutral carries In = 3I0 = 22.4 A
- I1 = (Ia + aIb + a²Ic)/3 = 99.35∠7.7° A
- I2 = (Ia + a²Ib + aIc)/3 = 18.47∠−39.6° A
- Rebuild Ib = I0 + a²I1 + aI2 = 80∠−110° A, as given
Answer: I1 = 99.3 A, I2 = 18.5 A, I0 = 7.5 A; neutral current 22.4 A
Sequence reactances seen from a 132 kV bus · Generator X1 = 0.20, X2 = 0.15, X0 = 0.06 pu, neutral grounded through Xn = 0.04 pu; delta-star step-up transformer (delta on the generator, star solidly grounded on the 132 kV side), X = 0.12 pu; all on 100 MVA
- Positive: generator and transformer in series, Z1 = 0.20 + 0.12 = j0.32 pu
- Negative: Z2 = 0.15 + 0.12 = j0.27 pu (the source is short-circuited in the negative network)
- Zero at the generator terminals: X0 + 3Xn = 0.06 + 0.12 = j0.18 pu
- Zero at the 132 kV bus: the delta blocks the generator, so only the transformer is seen to ground: Z0 = j0.12 pu
Answer: At the 132 kV bus Z1 = j0.32, Z2 = j0.27, Z0 = j0.12 pu
Unsymmetrical faults: LG, LL and LLG with and without fault impedance
Which syllabus covers thisAU EE3501 Unit IVVTU BEE601 Module 4SPPU 303148 Unit 5MU EEC502 Module 3KTU PBEET604
How are the sequence networks connected for a single line-to-ground fault?
Phase a touches ground through Zf, and phases b and c carry no fault current. Put Ib = Ic = 0 into the transform and all three sequence currents come out equal, Ia1 = Ia2 = Ia0. Since Va = Zf·Ia, the three networks must be in series, with 3Zf. The fault current is Ia = 3Ia1. With Z1 = Z2 = j0.25 and Z0 = j0.15 pu, a solid fault draws 3/0.65 = 4.62 pu.
What changes for a line-to-line fault?
Phases b and c touch each other but not ground, so Ia = 0 and Ib = −Ic. No current returns through the earth, so there is no zero sequence at all. The positive and negative networks face each other in parallel (with Zf in between): Ia1 = −Ia2 = E/(Z1 + Z2 + Zf), and Ib = −j√3·Ia1. When Z2 = Z1 the current is √3/2, or 86.6 %, of the three-phase value.
And a double line-to-ground fault?
Phases b and c are shorted together and to ground through Zf. Now Ia = 0 and Vb = Vc, which makes Va1 = Va2. The positive network feeds the negative network in parallel with the zero network plus 3Zf. The ground current is In = 3Ia0. A small Z0 pulls more of the current into the zero path and into the earth, which is what earth-fault relays measure.
Which fault gives the largest current?
Usually the three-phase fault, which is why fault-level studies start with it. Not always: at the terminals of a solidly grounded generator Z0 is smaller than Z1, and a line-to-ground fault can draw more. With X1 = X2 = 0.25 and X0 = 0.15 pu the three-phase fault gives 4 pu and the LG fault 4.62 pu. Grounding the neutral through a reactor raises Z0 by 3Xn and brings the LG current down.
At a bus with X1 = X2 = 0.25 pu and X0 = 0.15 pu, which solid fault draws the largest current?
Three-phase: 1/0.25 = 4 pu. LG: 3/(0.25 + 0.25 + 0.15) = 4.62 pu. LL: √3/0.5 = 3.46 pu. Because Z0 is smaller than Z1, the LG fault exceeds the three-phase fault.
Worked examples
A solid LG fault on a 132 kV bus · Z1 = j0.32, Z2 = j0.27, Z0 = j0.12 pu on 100 MVA, 132 kV (the bus of chapter 20's example); prefault 1 pu
- Base current = 100 MVA / (√3 × 132 kV) = 437.4 A
- Ia1 = Ia2 = Ia0 = 1/j(0.32 + 0.27 + 0.12) = −j1.408 pu
- Ia = 3Ia1 = 4.225 pu = 1.848 kA
- Healthy phases: |Vb| = |Vc| = 0.844 pu
- A three-phase fault here gives 1/0.32 = 3.125 pu, so the LG fault is 1.35 times larger
Answer: Ia = 4.23 pu = 1.85 kA; Vb and Vc sag to 0.84 pu
LL fault through a resistance · The same bus; phases b and c short through Rf = 0.05 pu
- Ia1 = 1/(0.05 + j(0.32 + 0.27)) = 1.689∠−85.2° pu, and Ia2 = −Ia1
- Ib = −j√3·Ia1, so |Ib| = |Ic| = √3 × 1.6889 = 2.925 pu = 1.279 kA
- Voltages: |Va| = 0.916, |Vb| = 0.531, |Vc| = 0.385 pu
- Solid (Rf = 0) it would be √3/0.59 = 2.936 pu
Answer: |Ib| = 2.93 pu = 1.28 kA
LLG fault and the ground current · The same bus; phases b and c short to ground solidly
- Z2 in parallel with Z0 = j(0.27 × 0.12)/(0.27 + 0.12) = j0.0831 pu
- Ia1 = 1/j(0.32 + 0.0831) = −j2.481 pu
- Ia2 = −Ia1 × 0.12/0.39 = j0.763 pu; Ia0 = −Ia1 × 0.27/0.39 = j1.718 pu
- Ground current In = 3Ia0 = 5.153 pu = 2.254 kA; |Ib| = |Ic| = 3.812 pu
- Unfaulted phase: |Va| = 3Va1 = 0.618 pu
Answer: |Ib| = |Ic| = 3.81 pu; ground current 5.15 pu = 2.25 kA
Travelling waves on lines: surge impedance, reflection, refraction and the Bewley lattice
Which syllabus covers thisAU not in syllabusVTU not in syllabusSPPU not in syllabusMU EEC502 Module 4KTU not in syllabus
What is a line's surge impedance, and how fast does a surge travel?
A surge (lightning, switching) is far too fast for the line to act as one lumped circuit: it charges the line's capacitance section by section, travelling at v = 1/√(LC), and while it travels voltage and current are tied by Zc = √(L/C), whatever is at the far end. Typical values are about 400 Ω for an overhead line and about 40 Ω for an underground cable. Overhead, v is close to the speed of light: a 30 km line takes about 100 μs. In a polyethylene-insulated cable waves travel at about 0.66 of that.
What happens when a wave reaches the end of a line?
Whatever the termination ZL, voltage and current there must obey both ZL and the line's own Zc, and a reflected wave makes up the difference. A fraction Γ = (ZL − Z1)/(ZL + Z1) of the voltage is sent back and 1 + Γ appears across the termination. An open end (Γ = +1) doubles the voltage and sends back a wave that cancels the current; a short (Γ = −1) sends back an inverted wave and the voltage there stays zero; a matched end (ZL = Z1) absorbs everything, as if the line went on for ever.
What happens where an overhead line meets a cable?
The cable, with its lower surge impedance, looks like a near-short to a wave arriving on the line: only 2 × 40/(40 + 400) = 0.182 of the voltage passes into it and 0.818 is reflected back inverted. A short cable section therefore softens a surge entering a substation (though its far-end reflections then build the voltage up again). The reverse is dangerous: a wave leaving a cable for an overhead line is multiplied by 2 × 400/440 = 1.82, close to doubling.
How does a Bewley lattice diagram keep track of all the reflections?
Draw distance across and time downward. Each wave is a sloping line from one end to the other, taking one travel time τ. At every end its height is multiplied by that end's Γ to give the next wave, and the voltage at a point is the sum of all the waves that have passed it so far. With a source resistance at one end and a resistor at the other, the receiving-end voltage steps up and down every 2τ and settles at the resistive divider value E·ZL/(ZL + Zs).
A step from a source with zero internal impedance reaches the open far end of a lossless line. What is the voltage at the open end just after the wave arrives?
At an open end Γ = +1, so the reflected wave equals the incident one and the voltage there is 1 + Γ = 2 times the arriving wave. With Zs = 0 the arriving wave is the full source voltage E, so the open end jumps to 2E. On a lossless line with Zs = 0 it then swings between 0 and 2E for ever.
Worked examples
Energising an open-ended line · A 100 kV step from a source with 100 Ω internal resistance is switched onto a 400 Ω lossless line whose far end is open
- Launched wave: 100 × 400/(400 + 100) = 80 kV; Γ at the source = (100 − 400)/(100 + 400) = −0.6; Γ at the open end = +1
- First arrival (t = τ): far-end voltage 2 × 80 = 160 kV
- At 3τ the wave has gone back and forth: + 80 × 1 × (−0.6) × 2 = −96 kV, so 64 kV
- At 5τ: 121.6 kV; at 7τ: 87.04 kV
- The swings shrink by 0.6 each round trip and settle at 100 kV, as the open circuit draws no current
Answer: 160, 64, 121.6, 87.04 kV … → 100 kV
A surge meets a cable · An illustrative 600 kV surge on a 400 Ω overhead line reaches a long 40 Ω cable
- Transmitted fraction 2 × 40/440 = 0.1818: 109.1 kV enters the cable
- Reflected fraction (40 − 400)/440 = −0.8182: −490.9 kV travels back along the line
- Currents: incident 600/400 = 1.5 kA, transmitted 109.09/40 = 2.727 kA, reflected +1.227 kA (incident + reflected = transmitted)
- Power check: 600²/400 = 900 MW = 602.5 reflected + 297.5 transmitted
Answer: 109.1 kV into the cable, −490.9 kV reflected
Lattice with resistive ends · Step E from a 200 Ω source onto a 400 Ω line ending in a 1200 Ω resistor
- Launched wave 400/600 E = 0.6667 E; Γs = (200 − 400)/600 = −0.3333; Γr = (1200 − 400)/1600 = 0.5
- At τ: Vr = 0.6667 × 1.5 = 1 E
- At 3τ: add 1 × (−1/3)(0.5) = −1/6 of the first step: 0.8333 E; at 5τ: 0.8611 E; at 7τ: 0.8565 E
- Sum of the geometric series: 1/(1 + 1/6) = 0.8571 E = 1200/(200 + 1200), the resistive divider
Answer: Vr: 1, 0.833, 0.861, 0.856 E … → 0.857 E
Rotor angle stability: the swing equation, inertia constant and step-by-step solution
Which syllabus covers thisAU EE3501 Unit VVTU BEE601 Module 5SPPU not in syllabusMU EEC702 Modules 2-3KTU PBEET604
What is rotor angle stability?
A synchronous machine sends out Pe = Pmax·sin δ, where δ is the angle between its internal voltage and the grid. In steady state the turbine's Pm equals Pe. A fault suddenly cuts Pe while Pm stays put, so the surplus speeds the rotor up and δ grows. After the fault is cleared, Pe exceeds Pm and pulls the rotor back. If it is pulled back before δ passes the unstable angle, the machine stays in step; if not, it slips a pole and must be tripped.
What is the inertia constant H, and what values are typical?
H is the kinetic energy stored in the spinning rotor at synchronous speed, divided by the machine's MVA rating, so it is in seconds. Published typical ranges are 2.5 to 6 s for two-pole thermal units, 4 to 10 s for four-pole units and 2 to 4 s for hydro. On a common base H scales with the rating: a 500 MW unit at 0.85 pf with H = 4 s stores 2353 MJ, which is H = 23.53 s on a 100 MVA base. Grid inertia also sets how fast frequency falls when a unit trips.
What is the swing equation?
It is Newton's law for the rotor, written in per unit: (2H/ωs)·d²δ/dt² = Pm − Pe, with ωs = 2π × 50 = 314.16 rad/s. With Pe = 0 during a bolted fault, a machine at Pm = 0.8 pu with H = 5 s accelerates at 25.1 rad/s², so δ climbs from 23.6° to 30.8° in the first 0.1 s. The equation is nonlinear because Pe depends on sin δ, so it is solved numerically.
How does the step-by-step method work?
Time is cut into steps, typically Δt = 0.05 s, and the accelerating power Pa = Pm − Pe is assumed constant over each step. With δ in degrees and M = H/(180f), the change in angle grows by (Δt²/M)·Pa every step: Δδn = Δδn−1 + (Δt²/M)·Pa,n−1. When the network switches at a step (the fault at t = 0, the clearing), the average of the Pa values just before and just after is used. For H = 5 s, Δt²/M = 4.5 degrees per pu. RK4 with a small step gives a reference to compare.
A bolted fault (Pe = 0 while it lasts) is cleared 0.05 s later than its critical clearing time. The rotor
The critical clearing time is exactly the case where the rotor just reaches the unstable equilibrium angle. Clearing later adds accelerating energy and takes away decelerating energy, so the rotor passes that angle, Pe then falls below Pm again, and it keeps speeding up until it is tripped.
Worked examples
Inertia constant on a common base · A 500 MW unit at 0.85 power factor, H = 4 s on its own rating (illustrative, inside the typical 2.5-6 s range for two-pole units)
- Rating = 500/0.85 = 588.2 MVA
- Stored energy = H × S = 4 × 588.2 = 2353 MJ
- On a 100 MVA base: H = 2353/100 = 23.53 s
- Two identical units swinging together: H = 47.06 s on 100 MVA
- Inertia coefficient on its own base for the step-by-step method: M = H/(180f) = 4/9000 = 4.444 × 10⁻⁴ pu·s²/degree
Answer: 2353 MJ stored; H = 23.53 s on 100 MVA
How fast frequency falls · A grid with H = 5 s on its total rating suddenly loses 3 % of its generation (illustrative). India's allowable band is 49.90-50.05 Hz
- Summed over all machines, the swing equation in frequency is (2H/f0)·df/dt = −ΔP
- df/dt = −f0·ΔP/(2H) = −50 × 0.03/(2 × 5) = −0.15 Hz/s
- From 50.00 Hz to the lower band edge 49.90 Hz takes 0.10/0.15 = 0.67 s
- This is only the first moments: governors and frequency response then halt the fall, but they need that time to act, which is why falling inertia matters
Answer: Frequency falls at 0.15 Hz/s and leaves the IEGC band in about 0.67 s
Step-by-step solution through a clearing · Pm = 0.8 pu, Pmax 2.0 before, 0.5 during and 1.5 after the fault, H = 5 s, 50 Hz; fault at t = 0, cleared at 0.1 s; Δt = 0.05 s
- δ0 = asin(0.8/2.0) = 23.58°; Δt²/M = 0.0025 × 9000/5 = 4.5
- t = 0: Pa = (0 + 0.6)/2 = 0.3; Δδ = 1.35°, δ(0.05) = 24.93°
- t = 0.05: Pa = 0.8 − 0.5 sin 24.93° = 0.589; δ(0.10) = 28.93°
- t = 0.10 (clearing): Pa = average of 0.558 and 0.074 = 0.316; δ(0.15) = 34.35°
- t = 0.15: Pa = −0.046; δ(0.20) = 39.57°. t = 0.20: Pa = −0.156; δ(0.25) = 44.09°
- t = 0.25: Pa = −0.244; δ(0.30) = 47.51° (RK4 with a 0.1 ms step: 47.3°)
Answer: δ rises to 47.5° at 0.30 s and Pa is already negative, so the rotor is being pulled back
The equal area criterion and the critical clearing angle and time
Which syllabus covers thisAU EE3501 Unit VVTU BEE601 Module 5SPPU not in syllabusMU EEC702 Module 3KTU PBEET604
What does the equal area criterion say?
Multiply the swing equation by dδ/dt and integrate: the rotor's speed above synchronous returns to zero when the area between Pm and the Pe curve, counted with sign, adds up to zero. While Pm is above Pe the rotor gains energy, the accelerating area A1. Once Pe is above Pm it gives that energy back, the decelerating area A2. The machine is stable on its first swing if A2 can equal A1 before δ reaches the unstable angle δu. No time integration is needed for the verdict.
How does it handle a sudden increase in mechanical input?
On a single curve Pmax·sin δ, raising Pm from Pm0 to Pm1 moves the equilibrium from δ0 to δ1, but the rotor overshoots. A1 lies between the new Pm line and the curve from δ0 to δ1; the rotor stops at δmax where A2 = A1. With Pmax = 2 pu and a step from 0.5 to 1.0 pu, δ swings from 14.5° past 30° to 46.4°. The largest step from 0.5 pu that stays stable is to 1.6 pu, well short of Pmax.
What is the critical clearing angle?
It is the clearing angle at which the decelerating area available up to δu exactly equals A1. Setting the two areas equal gives a closed form in which Pm, Pmax during the fault (P2) and Pmax after clearing (P3) appear. For Pm = 0.8, P1 = 2.0, P2 = 0 and P3 = 1.5 pu, δ0 = 23.6°, δu = 147.8° and δcr = 71.9°. Clear before the rotor gets there and it survives.
How is the critical clearing angle turned into a time?
Only through the swing equation. If no power flows during the fault (P2 = 0), the rotor's acceleration is constant and δ grows as a parabola, so tcr = √(4H(δcr − δ0)/(ωs·Pm)): 0.259 s, or about 13 cycles at 50 Hz, for H = 5 s. When some power still flows, the time comes from a numerical solution of the swing equation. The criterion assumes one machine against an infinite bus, no damping and constant internal voltage, which is why grid studies use full time simulations.
The clearing angle is moved from 5° below δcr to 5° above it. What happens to the areas?
Every extra degree under the fault adds a strip to A1 (Pm is above the during-fault curve) and the same degree is no longer available for deceleration after clearing. Below δcr, A2 available exceeds A1; above it, A1 wins and the rotor passes δu.
Worked examples
Critical clearing angle and time for a bolted fault · Pm = 0.8 pu; Pmax 2.0 pu before, 0 during (three-phase fault at the machine's bus) and 1.5 pu after (one line out); H = 5 s, 50 Hz
- δ0 = asin(0.8/2.0) = 23.58° = 0.4115 rad; δu = 180° − asin(0.8/1.5) = 147.77° = 2.5791 rad
- cos δcr = [0.8 × (2.5791 − 0.4115) + 1.5 cos 147.77° − 0]/(1.5 − 0) = 0.3101
- δcr = 71.93° = 1.2555 rad
- Check: A1 = 0.8 × (1.2555 − 0.4115) = 0.6752; A2 = ∫(1.5 sin δ − 0.8)dδ from δcr to δu = 0.6752
- tcr = √(4 × 5 × 0.844/(314.16 × 0.8)) = 0.259 s, about 13 cycles
Answer: δcr = 71.9°, tcr = 0.259 s; cleared at 0.2 s the rotor peaks at 91° and survives
A sudden step in turbine power · A machine on an infinite bus with Pmax = 2.0 pu runs at Pm = 0.5 pu; the input is suddenly raised to 1.0 pu
- δ0 = asin(0.5/2) = 14.48°; new equilibrium δ1 = asin(1.0/2) = 30°
- A1 = 1.0 × (0.5236 − 0.2527) − 2(cos 14.48° − cos 30°) = 0.0665
- Decelerating area available up to δu = 150°: 1.3697, far more than A1, so stable
- δmax from 2(cos δ0 − cos δmax) = 1.0 × (δmax − δ0): δmax = 46.39°
- The largest stable step from 0.5 pu (A1 = A2 available, by bisection): Pm1 = 1.599 pu
Answer: Stable; δ swings to 46.4° and oscillates about 30°; steps beyond 1.6 pu would lose synchronism
When some power still flows during the fault · The same machine as the first example, but the fault is part-way along a line, so Pmax during the fault is 0.5 pu
- δ0 = 23.58°, δu = 147.77° as before
- cos δcr = [0.8 × (2.5791 − 0.4115) + 1.5 cos 147.77° − 0.5 cos 23.58°]/(1.5 − 0.5) = 0.0069
- δcr = 89.6°: later than with a bolted fault, because the rotor accelerates less
- Areas check: A1 = 0.4671 = A2 = 0.4671
- No parabola now: solving the swing equation numerically, the rotor reaches δcr at 0.376 s
Answer: δcr = 89.6°, tcr = 0.38 s (against 0.26 s for the bolted fault)
Keep going
- GATE Electrical Engineering (EE) preparation
- Electrical Machines interactive book
- Power Electronics interactive book
- Control Systems interactive book
- Signals and Systems interactive book
- Circuit Theory and Network Analysis interactive book
- All Skill-Lync interactive books
Recommended reading
- Power System Analysis, J. J. Grainger and W. D. Stevenson
- Modern Power System Analysis, D. P. Kothari and I. J. Nagrath
- Power System Analysis, Hadi Saadat
- Power System Analysis and Design, J. D. Glover, M. S. Sarma and T. J. Overbye
- Electrical Power Systems, C. L. Wadhwa
- Principles of Power System, V. K. Mehta and R. Mehta
- Power System Stability and Control, P. Kundur
How this book was made
Chapters follow the official syllabi listed above. Every interactive's formulas and every worked-example number were checked by script. Explanations, animations and examples are original to Skill-Lync. Topics informed by MIT OpenCourseWare 6.061 (Introduction to Electric Power Systems), CC BY-NC-SA. All explanations, animations and examples are original to Skill-Lync; MIT does not endorse this page.