Advanced Power System Analysis

Unlike ANSI/IEEE (prefault voltage = 1.0 pu), IEC 60909 uses voltage factors to account for voltage variations, tap positions, and subtransient behaviour.

• $c_{max}$: equipment rating (maximum fault current)
• $c_{min}$: protection coordination (minimum fault current)

Image: Three-phase short circuit — Wikimedia Commons, CC BY-SA 3.0

The RMS value at fault inception:

$$I''_k = \frac{c \times U_n}{\sqrt{3} \times Z_k}$$

Where:

• $c$ = voltage factor
• $U_n$ = nominal system voltage
• $Z_k$ = positive-sequence impedance at fault location

Accounts for DC offset (worst-case asymmetry):

$$i_p = \kappa \times \sqrt{2} \times I''_k$$

The kappa factor:

$$\kappa = 1.02 + 0.98 \times e^{-3R/X}$$

Ranges from 1.1 (high R/X) to 2.0 (low R/X, near generators).

Image: Short-circuit current waveform with DC offset — Wikimedia Commons, Public Domain

Key difference from ANSI: IEC applies correction factors to component impedances.

Transformer correction:

$$K_T = \frac{0.95 \times c_{max}}{1 + 0.6 \times x_T}$$

Generator correction:

$$K_G = \frac{U_n}{U_{rG}} \times \frac{c_{max}}{1 + x''_d \times \sin\varphi_{rG}}$$

These corrections make IEC results 5–10% higher than ANSI — critical when mixing equipment rated to different standards.

Beyond $I''_k$, IEC 60909 defines:

• Symmetrical breaking current $I_b$: accounts for AC decay near generators (relevant for breaker interrupting rating)
• Thermal equivalent current $I_{th}$: for equipment thermal withstand over fault duration

$$I_{th} = I''_k \times \sqrt{m + n}$$

Where $m$ accounts for DC component decay and $n$ for AC component decay.

Unbalanced faults are solved using symmetrical components:

Line-to-earth fault (most common, ~70% of faults):

$$I''_{k1} = \frac{\sqrt{3} \times c \times U_n}{Z_1 + Z_2 + Z_0}$$

Series connection of all three sequence networks.

Line-to-line fault:

$$I''_{k2} = \frac{c \times U_n}{Z_1 + Z_2}$$

No zero-sequence involvement.

Image: Symmetrical components — Wikimedia Commons, CC BY-SA 4.0

Zero-sequence behaviour depends critically on transformer winding connections:

Delta windings act as zero-sequence current traps — they circulate $I_0$ within the winding but don't pass it through.

Image: Wye-Delta transformer connection — Wikimedia Commons, Public Domain

The 2018 revision introduced five electrode configurations based on over 1,800 arc flash tests (vs ~300 in 2002):

HCB is worst-case — arc plasma is directed toward the worker.

Image: Electric arc diagram — Wikimedia Commons, CC BY-SA 3.0

Arcing current is derived from bolted fault current using empirically fitted polynomials:

$$\lg(I_{arc}) = k_1 + k_2 \lg(I_{bf}) + k_3 [\lg(I_{bf})]^2 + \cdots$$

Where coefficients vary by electrode configuration and voltage level.

A critical improvement: the variable arc current variation factor replaces the fixed 15% reduction from the 2002 edition. Lower arcing currents at reduced voltage can cause:

1. Longer protective device clearing times
2. Potentially higher incident energy than full arcing current

Software must calculate at both normal and minimum arcing current, reporting the higher value.

The fundamental relationship:

Incident energy is directly proportional to arc duration.

Arc duration = protective device clearing time at the calculated arcing current.

$$E \propto t_{arc}$$

Reducing clearing time from 500 ms to 50 ms reduces incident energy by ~90%.

Worst-case scenario: If arcing current falls on an inverse-time curve, reduced current → longer clearing time → potentially higher incident energy. This is why IEEE 1584 requires dual calculations.

IEEE 1584-2018 is valid for:

Outside these limits, alternative methods (e.g., Lee method for >15 kV) must be used.

The distance at which incident energy equals 1.2 cal/cm² — the threshold for onset of second-degree burns on unprotected skin.

AFB is calculated using the same IEEE 1584 equations but solving for distance rather than energy at a fixed distance.

Typical AFB values:

• Low voltage switchgear: 0.5–3 m
• Medium voltage switchgear: 3–10+ m
• Depends heavily on clearing time and fault current

Image: Arc flash hazard warning sign — Wikimedia Commons, Public Domain

NFPA 70E translates IEEE 1584 incident energy into PPE categories:

Above 40 cal/cm², engineering controls are mandatory: de-energisation, remote operation, or arc energy reduction.

Image: Electrical substation — Wikimedia Commons, CC BY 2.0

NEC 240.87 mandates arc energy reduction for circuit breakers ≥1200 A. Approved methods:

1. Zone-selective interlocking (ZSI): Upstream device reduces delay when downstream device detects fault
2. Differential relaying: Compares current in/out of protected zone
3. Energy-reducing maintenance switch: Temporarily reduces trip delay for maintenance
4. Arc flash relays: Detect arc light + overcurrent, initiate rapid trip (<50 ms)

Arc flash relays are increasingly popular — they detect visible light from the arc and can reduce clearing time to <35 ms, dramatically cutting incident energy.

Image: Electromechanical protective relays — Wtshymanski, Wikimedia Commons, CC BY-SA 3.0

The general IEC 60255 inverse-time equation:

$$t = TMS \times \frac{k}{\left(\frac{I}{I_s}\right)^\alpha - 1}$$

where $t$ is operating time, $TMS$ is the time multiplier setting, $I$ is fault current, $I_s$ is pickup current, and $k$, $\alpha$ are curve constants.

Standard curve types and their constants:

The IEEE C37.112 formulation uses a slightly different structure:

$$t = TD \times \left(\frac{A}{\left(\frac{I}{I_s}\right)^p - 1} + B\right)$$

Curve selection guidance:

• SI: Fault current does not vary much across the network
• VI: Significant impedance between relay locations changes fault magnitude
• EI: Need to coordinate with fuse curves or ride through motor starting current

The grading margin (CTI) between series devices must account for:

1. Relay timing error — ±5% for numerical, ±7.5% for electromechanical (at 2–5× pickup)
2. Overshoot time — negligible for numerical, 0.05–0.1s for electromechanical
3. Circuit breaker opening time — 0.04–0.08s modern, 0.1–0.15s older
4. CT errors — ±3% typical, affects apparent current seen by relay

Resulting CTI values:

Grading procedure (load → source):

1. Set downstream device below equipment damage curve
2. Calculate downstream operating time at maximum through-fault
3. Add CTI to get required upstream time
4. Solve for upstream TMS:

$$TMS_{upstream} = \frac{(t_{downstream} + CTI) \times \left[\left(\frac{I}{I_s}\right)^\alpha - 1\right]}{k}$$

5. Verify coordination holds at all intermediate fault levels
6. Check for crossover points where curves intersect (loss of coordination)

Four mechanisms achieve selective fault isolation:

1. Current grading (instantaneous elements, ANSI 50)

• Downstream devices see higher fault current due to source impedance
• Set pickup above maximum downstream fault: $I_{pickup} > I_{fault,downstream}$
• Limitation: ineffective on short feeders where fault current varies little

2. Time grading (IDMT elements, ANSI 51)

• Upstream devices have progressively longer delays
• Works universally but accumulates delay toward source

3. Energy-based selectivity (I²t)

• Compare let-through energy of downstream fuse/breaker
• Upstream device must not operate before downstream clears

4. Zone-selective interlocking (ZSI)

• Direct communication between devices via control wiring
• Downstream device signals upstream: "I see the fault, wait"
• If no restraint signal → upstream trips instantaneously (close-in fault)
• Dramatically reduces arc flash energy at upstream locations

Image: Relay circuit symbols — Wikimedia Commons, Public Domain

Differential protection applies Kirchhoff's Current Law: under healthy conditions, current entering a protected zone equals current leaving. Any difference indicates an internal fault.

$$I_{operate} = |I_{in} - I_{out}|$$

Key property: 100% selective — only operates for faults within the zone boundary defined by CT locations.

Applications by ANSI device number:

• 87T — Transformer differential
• 87B — Busbar differential
• 87G — Generator differential
• 87M — Motor differential

Transformer differential relays use percentage restraint characteristics to remain stable during through-faults where CT errors produce false differential current.

The restraint current is typically:

$$I_{restraint} = \frac{|I_1| + |I_2|}{2}$$

Dual-slope characteristic:

Harmonic restraint prevents false tripping during transformer energisation (magnetising inrush):

• Inrush produces characteristic 2nd harmonic content
• Typical restraint threshold: 15–20% second harmonic relative to fundamental
• Over-excitation produces 5th harmonic — separate restraint at ~25%

CT ratio mismatch compensation: Software in numerical relays applies vector group correction (e.g., Dyn11 → 30° phase shift) and ratio compensation automatically. For electromechanical relays, interposing CTs are required.

Image: Electrical substation — Wtshymanski, Wikimedia Commons, Public Domain

Distance relays measure impedance to determine fault location. The apparent impedance during a fault:

$$Z_{apparent} = \frac{V_{relay}}{I_{relay}}$$

If $Z_{apparent}$ falls within the relay's operating characteristic (mho circle or quadrilateral), a trip is initiated.

Multi-zone reach settings:

Why Zone 1 underreaches (80–85%):

• CT and VT measurement errors (±3% each)
• Line impedance data uncertainty
• Prevents unwanted instantaneous trip for faults beyond the remote bus

Zone 2 overlap coordination: Zone 2 of line A overlaps with Zone 1 of line B at the remote bus. The time delay ensures the local Zone 1 relay trips first. If Zone 1 fails, Zone 2 provides time-delayed backup.

Mho (circular) characteristic:

• Inherently directional
• Good reach along resistive and reactive axes
• Traditional choice for transmission lines
• Reach affected by fault resistance and load encroachment

Quadrilateral characteristic:

• Independent resistive and reactive reach settings
• Better coverage of resistive faults (e.g., arcing faults, tower footing resistance)
• Preferred for shorter lines and cable circuits
• Requires separate directional element

Practical considerations:

• Load encroachment: Heavy load produces low apparent impedance that may enter Zone 3 — use load blinders or shaped characteristics
• Infeed effect: Intermediate source current makes fault appear further away — Zone 2 reach must account for this
• Mutual coupling: Zero-sequence mutual impedance between parallel lines affects ground distance elements — requires compensation factor

Coordination study methodology (complete):

1. Start at load end, set downstream devices below equipment damage curves
2. Add grading margins progressively toward source
3. Set distance Zone 1 to 80–85% of each line
4. Set Zone 2 to cover full line with time delay
5. Verify Zone 2 does not overreach into Zone 1 operating time of adjacent lines
6. Set Zone 3 for remote backup with adequate delay
7. Verify all settings under maximum and minimum fault conditions
8. Check performance with and without infeed sources

Image: Steam turbine rotor — Wikimedia Commons, CC BY-SA 3.0

The swing equation governs rotor angle dynamics following a disturbance. It balances mechanical input against electrical output through the rotor's angular momentum:

$$M \frac{d^2\delta}{dt^2} = P_m - P_e = P_a$$

where $M$ is angular momentum, $\delta$ is rotor angle relative to synchronous reference, $P_m$ is mechanical power input, $P_e$ is electrical power output, and $P_a$ is accelerating power.

In per-unit form using the inertia constant $H$ (MJ/MVA):

$$\frac{2H}{\omega_s} \frac{d^2\delta}{dt^2} = P_m - P_e$$

Typical inertia constants:

Higher $H$ means more stored kinetic energy and longer time to respond to disturbances. This explains concern about reduced system inertia as synchronous generation is displaced by converter-interfaced renewables.

Physical interpretation:

• When $P_m > P_e$: rotor accelerates, $\delta$ increases
• When $P_e > P_m$: rotor decelerates, $\delta$ decreases
• Steady state: $P_m = P_e$, $\delta$ constant

The equal area criterion (EAC) provides graphical stability assessment for single-machine-infinite-bus (SMIB) systems without solving differential equations.

For a lossless system, electrical power output follows the power-angle curve:

$$P_e = P_{max} \sin\delta = \frac{|V_G||V_\infty|}{X} \sin\delta$$

When a fault occurs:

1. $P_e$ drops (possibly to zero for a bolted terminal fault) while $P_m$ continues → creates accelerating area $A_1$
2. Rotor angle $\delta$ advances as rotor speeds up
3. After fault clearing, $P_e > P_m$ → creates decelerating area $A_2$
4. Stability requires $A_2 \geq A_1$ before $\delta$ exceeds $\delta_{max}$ on the post-fault curve

$$A_1 = \int_{\delta_0}^{\delta_c} (P_m - P_{e,fault}) \, d\delta$$

$$A_2 = \int_{\delta_c}^{\delta_{max}} (P_{e,post} - P_m) \, d\delta$$

The critical clearing angle $\delta_{cr}$ is the maximum angle at which the fault can be cleared and still maintain $A_2 = A_1$. Beyond $\delta_{cr}$, synchronism is lost.

Key insight: The system can remain stable even when $\delta > 90°$ during the transient swing — the 90° limit applies only to steady-state stability, not transient stability.

Critical clearing time (CCT) is the maximum fault duration that maintains stability — protection must operate faster than CCT.

Typical CCT values:

Factors that increase CCT (improve stability):

• Higher machine inertia ($H$)
• Lower pre-fault loading (larger stability margin)
• Greater electrical distance to fault
• Faster excitation systems
• Higher post-fault transmission capacity (redundant lines)

Relationship to protection coordination: The total fault clearing time comprises:

• Relay operating time (20–40 ms for numerical relays)
• Circuit breaker opening time (40–80 ms modern SF₆)
• Total: typically 60–120 ms

For a CCT of 150 ms, the stability margin is only 30–90 ms — demonstrating why fast protection is critical for generator stability.

Study timeframe:

• First-swing stability: 3–5 seconds
• Multi-swing (large interconnections): up to 20 seconds
• Governor response (too slow for first swing): 0.5–2 seconds

Image: Electrical substation — Wtshymanski, Wikimedia Commons, Public Domain

IEEE Std 421.5-2016 defines standard models for excitation systems used in stability studies. Key model types:

Static excitation (ST1A) advantages:

• Very fast forcing capability (ceiling voltage in <0.1s)
• High initial response ratio
• Improves first-swing transient stability
• Can provide both positive and negative field forcing

Limitation: Static exciters depend on generator terminal voltage — during close-in faults, terminal voltage collapses and exciter output is reduced precisely when it's needed most.

Automatic Voltage Regulator (AVR) impact on stability:

• Fast AVR improves first-swing stability by maintaining field flux and thus electrical power output during faults
• However, high-gain AVR can reduce damping of subsequent oscillations — this is where PSS is essential

A Power System Stabilizer (PSS) adds supplementary damping to electromechanical oscillations by modulating the excitation system output.

Operating principle: The PSS detects rotor speed deviations and applies a compensating signal to the AVR voltage reference. When the rotor accelerates, PSS reduces excitation (reducing electrical torque), and vice versa — providing a damping torque component in phase with speed deviation.

PSS2B (dual-input stabilizer): Uses both electrical power and shaft speed as inputs:

• Speed signal: direct measure of rotor oscillation
• Power signal: provides better transient response
• Combination gives robust performance across operating conditions

Tuning methodology:

1. Identify electromechanical oscillation modes (typically 0.2–2.0 Hz)
   • Local modes: 1.0–2.0 Hz (single machine against system)
   • Inter-area modes: 0.2–0.8 Hz (groups of machines oscillating)
2. Set PSS phase compensation to provide damping torque at modal frequencies
3. Adjust gain for adequate damping without excessive AVR interaction
4. Verify with eigenvalue analysis (all modes positively damped)

Design trade-off:

• Too much PSS gain → excessive voltage variation during disturbances
• Too little PSS gain → insufficient damping, sustained oscillations
• Typical target: damping ratio $\zeta > 5\%$ for all electromechanical modes

Complete transient stability study workflow:

1. Build network model — generators with IEEE 421.5 excitation, governor models, load models (constant impedance / constant current / constant power mix)

2. Establish base case power flow — verify steady-state operating point, generator MW/MVAr dispatch, voltage profile

3. Define contingencies — N-1 and critical N-2 events: three-phase faults, line trips, generator trips, load rejection

4. Run time-domain simulation — typically 5–20 seconds, monitoring rotor angles, bus voltages, line flows

5. Assess stability criteria:

   • Rotor angle separation < 180° (or returns toward equilibrium)
   • Voltage recovery > 0.8 pu within 1 second of fault clearing
   • Frequency within acceptable band (49.5–50.5 Hz for UK)
   • Damping ratio of oscillations > 5%

6. Determine CCT for each contingency — binary search on fault duration until stability boundary found

7. Verify protection operates within CCT — if not, protection settings or system reinforcement required

Software tools: PSS/E, PowerWorld, DIgSILENT PowerFactory, ETAP, PSCAD (for electromagnetic transients)

Image: Stator and rotor — Wikimedia Commons, CC BY-SA 3.0

At standstill, an induction motor has no back-EMF, producing locked-rotor currents of 5–8 times full-load amperes (FLA) that persist for 5–30 seconds during acceleration.

This current flows at a severely lagging power factor (10–25%), imposing both real and reactive power demands.

NEMA code letters specify starting kVA per horsepower. Code G (most common) indicates 5.6–6.29 kVA/HP:

$$kVA_{start} = \text{Code Letter} \times HP$$

A 500 HP motor at Code G may draw over 3000 kVA during starting — potentially exceeding the transformer or generator rating at the supply bus.

Torque-speed relationship during acceleration:

• At standstill: high current, low power factor, moderate torque
• During acceleration: current decreases as back-EMF develops
• Near full speed: current drops to normal FLA, slip typically 2–5%

The percentage voltage dip at a motor bus:

$$\%V_{dip} \approx \frac{kVA_{start}}{kVA_{SC}} \times 100\%$$

where $kVA_{start}$ is motor starting kVA and $kVA_{SC}$ is the system short-circuit capacity at the motor bus.

Acceptance criteria for different equipment:

Generator-supplied systems are more severe: Generator transient reactance $X'_d$ (typically 15–25%) limits fault contribution, making voltage dips deeper compared to utility-supplied systems.

Generator sizing rules of thumb:

IEEE 1453 addresses voltage flicker using Pst (short-term) and Plt (long-term) indices, with planning level of $P_{st} \leq 1.0$ for low-voltage systems.

When multiple large motors must start on the same system, sequential starting with 30–60 second delays between starts allows voltage recovery before each subsequent start.

Study approach:

1. Static analysis (snapshot) — calculates voltage at the instant of starting, assumes worst-case locked-rotor impedance. Quick screening tool.

2. Dynamic analysis (time-domain) — simulates the complete acceleration transient including:

   • Motor torque-speed curve vs load torque curve
   • Generator excitation response and voltage recovery
   • Governor response for frequency dip
   • Interaction between motors already running and the starting motor

Key checks in dynamic study:

• Motor accelerates to full speed (torque exceeds load at all speeds)
• Voltage recovers above 90% within acceptable time
• Running motors do not stall during the voltage dip
• Frequency does not drop below 95% on island systems

Image: Induction motor internals — Wikimedia Commons, Public Domain

Star-delta starting (Y-Δ):

• Motor windings connected in star during start, then switched to delta
• Both current and torque reduced to 33% of DOL values
• Suitable only for unloaded or light-load starts
• Transition causes a current transient (momentary disconnection)
• Motor must have both ends of all three phase windings accessible (6 terminals)

Autotransformer starting: Common taps at 50%, 65%, 80% of line voltage.

At the 65% tap:

The autotransformer provides better torque-per-line-ampere ratio than star-delta because the transformer action reduces line current by the square of the voltage ratio.

Key advantage: Line current = motor current × tap², giving more starting torque for the same line current impact.

Soft starters (thyristor phase-angle control):

• Adjustable voltage ramp from reduced voltage to full voltage
• Current limiting typically settable from 150–600% FLA
• Provides smooth, controlled acceleration without current transients
• Can include soft-stop (voltage ramp-down) for pump applications
• Limitation: reduced voltage means reduced torque — not suitable for high-inertia loads requiring full torque from standstill

Variable Frequency Drives (VFDs):

• Starting current typically only 100–150% FLA
• Maintains full torque capability from zero speed
• Controlled V/Hz operation keeps flux constant during acceleration
• Most sophisticated and expensive solution
• Additional benefits: energy savings at partial load, precise speed control

Comparative summary:

Selection criteria:

• Required starting torque vs load torque curve
• Acceptable voltage dip at the motor bus
• Number of starts per hour (thermal duty)
• Need for speed control during normal operation
• Budget and maintenance capability

Image: Three-phase AC waveform (fundamental) — Wikimedia Commons

Harmonics are sinusoidal components at integer multiples of the fundamental frequency (50/60 Hz), originating from nonlinear loads that draw non-sinusoidal current.

Six-pulse rectifier characteristic harmonics at orders $h = 6k \$pm 1 (where $k = 1, 2, 3, \ldots$):

Without mitigation, six-pulse drives produce THDi of 80–100%.

Higher pulse configurations:

Total Harmonic Distortion:

$$THD = \frac{\sqrt{\sum_{h=2}^{H} V_h^2}}{V_1} \times 100\%$$

where $V_h$ is the RMS value of the $h$th harmonic and $V_1$ is the fundamental component. $H$ is typically taken as 50 (can be limited to 25 in most practical cases).

IEEE 519-2022 voltage distortion limits at PCC:

IEEE 519-2022 current distortion limits depend on the ratio $I_{SC}/I_L$ (short-circuit current to maximum demand load current) at the PCC — higher ratios (stiffer systems) allow more distortion.

The limits apply at the point of common coupling (PCC) between utility and customer, not at individual equipment terminals.

Transformers:

• Increased eddy current losses: $P_{ec,h} = P_{ec,1} \times h^2$
• Increased stray losses in structural parts
• K-factor rating quantifies harmonic heating:

$$K = \sum_{h=1}^{H} \left(\frac{I_h}{I_1}\right)^2 \times h^2$$

K-factor rated transformers (K-4, K-13, K-20, K-30) are designed for harmonic-rich loads without derating.

Cables:

• Skin effect increases AC resistance at harmonic frequencies: $R_{ac,h} \approx R_{$dc} \times \sqrt{h}
• Proximity effect further increases losses
• Triplen harmonics (3rd, 9th, 15th) add in neutral conductor — neutral may carry up to 1.73× phase current with single-phase nonlinear loads

Rotating machines:

• Negative-sequence harmonics (5th, 11th) create reverse-rotating fields causing torque pulsations
• Rotor heating from induced harmonic currents
• Audible noise and vibration

Capacitors:

• Current increases with frequency: $I_h = h \times \omega C \times V_h$
• Risk of thermal overload and dielectric failure
• IEEE Std 18 limits: 135% of rated current, 110% of rated voltage, 135% of rated kvar

Image: Electrical substation with capacitor banks — Wikimedia Commons

The critical concern is resonance between system inductance and power factor correction capacitors.

Parallel resonance creates high impedance at the resonant frequency, amplifying even small harmonic currents into large voltage distortion.

The resonant harmonic order:

$$h_r \approx \sqrt{\frac{MVA_{SC}}{MVAr_{cap}}}$$

Example: 20 MVA short-circuit system with 2 MVAr capacitor bank: $h_r = \sqrt{20/2} = 3.2$ — dangerously close to the 3rd harmonic.

Series resonance creates low impedance paths that can sink excessive harmonic currents into filters or capacitors, potentially causing thermal overload.

Frequency scanning analysis:

• Plots driving-point impedance vs frequency
• Peaks indicate parallel resonance (high Z)
• Valleys indicate series resonance (low Z)
• Critical check: resonant peaks must not coincide with characteristic harmonic frequencies

Harmonic penetration uses current injection where nonlinear loads are modelled as ideal current sources injecting their characteristic spectrum:

$$[V]_h = [Z]_h \times [I]_h$$

Harmonic voltages from harmonic currents and network impedance at each frequency $h$.

Frequency-dependent component modelling:

Transformer winding connections affect harmonic propagation — delta windings block triplen harmonics (3rd, 9th, 15th) from passing between voltage levels.

This is why distribution transformers are commonly Dyn11 — the delta primary traps triplen currents.

Passive single-tuned filter design:

• Tuned slightly below target harmonic (e.g., 4.7th for 5th)
• Provides low impedance path to divert harmonic current
• Also supplies reactive power at fundamental frequency
• Risk: can attract harmonics from elsewhere in the system
• Must be designed considering system impedance variations

Filter tuning calculation:

$$f_{tune} = \frac{1}{2\$$pi\sqrt{LC}}

Quality factor $Q = X_L / R$ determines sharpness of tuning (typical Q = 30–60 for power filters).

Other mitigation approaches:

Design sequence for harmonic study:

1. Model system with all nonlinear loads as current sources
2. Perform frequency scan — identify resonance risks
3. Check IEEE 519 compliance at PCC without mitigation
4. If non-compliant: apply mitigation and re-check
5. Verify capacitor duty (IEEE Std 18) and transformer K-factor
6. Check sensitivity to system impedance variations (±10% typical)

Image: Transmission lines — Wikimedia Commons

The N-1 criterion requires that power systems withstand loss of any single major component — transmission line, transformer, or generator — without violating:

• Thermal limits
• Voltage limits
• Stability constraints

Extended criteria:

Regulatory frameworks:

• NERC TPL-001-5.1 (North America): Transmission planning requirements for Category B (N-1) and Category C (N-1-1) events
• ENTSO-E System Operation Guideline (EU 2017/1485): European transmission security standards

DC screening enables rapid evaluation of thousands of contingencies using linear sensitivity factors.

Power Transfer Distribution Factors (PTDF): Fraction of a transaction flowing on each line:

$$PTDF_{l,i \to j} = \frac{\Delta P_l}{\Delta P_{transaction}}$$

Line Outage Distribution Factors (LODF): Flow redistribution when line $k$ trips:

$$LODF_{l,k} = \frac{\Delta P_l}{P_k^{pre}}$$

Post-contingency flow calculation:

$$P_l^{post} = P_l^{pre} + LODF_{l,k} \times P_k^{pre}$$

This allows checking all N-1 cases without re-solving power flow for each contingency — extremely fast.

Performance index ranking:

$$PI=\sum {\left(\frac{P_l}{P_l^{max}}\right)}^{2n}$$PI = \sum \left(\frac{P_l}{P_l^{max}}\right)^{2n}

Ranks contingencies by severity. Only the top-ranked (most severe) cases proceed to full AC analysis.

DC power flow screening is fast but approximate:

• Assumes flat voltage profile (all buses 1.0 pu)
• Ignores reactive power and losses
• Good for thermal screening
• Cannot detect voltage violations

Full AC contingency analysis captures:

• Voltage depression at load buses
• Reactive power redistribution
• Generator reactive limits (Q limits)
• Tap changer response
• Voltage stability concerns

Practical workflow:

1. DC screening: rank all N-1 contingencies by PI
2. Select top 50–100 most severe cases
3. Run full AC power flow for selected cases
4. Check thermal, voltage, and stability criteria
5. Identify violations and corrective actions

Image: Electrical substation — Wikimedia Commons

Transmission line thermal ratings:

Line ratings depend on conductor type, ambient temperature, wind speed, and acceptable sag. IEEE 738 defines the calculation methodology.

Transformer loading (IEEE C57.91):

Voltage limits:

P-V (nose) curves plot bus voltage against system loading — the fundamental tool for voltage stability assessment.

Key features of the P-V curve:

• Operating point: current voltage and loading
• Nose point: maximum loadability (voltage collapse)
• Stability margin: distance from operating point to the nose point, typically expressed in MW

$$\text{Margin} = P_{nose} - P_{operating}$$

A system is voltage stable if the operating point is on the upper portion of the curve (above the nose).

Factors reducing voltage stability margin:

• Loss of reactive power sources (generators at Q limit)
• Long transmission distances
• Heavy loading conditions
• Loss of transmission lines (N-1 contingencies)

Post-contingency P-V analysis determines whether voltage collapse risk exists after credible outages.

Corrective actions for contingency violations:

1. Generation redispatch — shift power from constrained path to relieve overloads
2. Topology switching — open/close breakers to redistribute flow
3. Reactive compensation — switch capacitor banks, adjust SVCs/STATCOMs
4. Load shedding — last resort, shed non-critical load to maintain system integrity

Remedial Action Schemes (RAS): Automatic corrective actions triggered by specific contingencies:

• Generator tripping
• Controlled separation
• HVDC power order changes
• Automatic load shedding

NERC defines RAS as "automatic protection systems designed to detect abnormal conditions and take corrective actions other than isolation of faulted components."

Security-Constrained Optimal Power Flow (SCOPF): Optimises dispatch while respecting both pre-contingency and post-contingency constraints simultaneously:

$$\min \sum C_i(P_{G,i})$$

Subject to:

• Pre-contingency power flow equations
• Post-contingency flow limits for all credible N-1 cases
• Generator limits, voltage limits

Provides the most economically efficient dispatch that maintains N-1 security — the standard tool for real-time market operations and planning studies.

Image: Electrical substation model — Wikimedia Commons

PSS/E (Siemens PTI) provides the psspy module, the industry-standard power system simulation API.

Initialisation:

import psse35
import psspy
psspy.psseinit(150000)  # max buses

Core functions:

All functions return ierr (integer error code). Always check: ierr == 0 means success.

One documented case achieved 2,544 power flow simulations using PSS/E automation — weeks of manual effort completed in hours.

PowerWorld SimAuto / ESA:

The ESA (Easy SimAuto) library wraps the COM interface and returns results as pandas DataFrames:

from esa import SAW
saw = SAW('case.pwb')
bus_data = saw.GetParametersSingleElement(
    'bus', ['BusNum', 'BusNomVolt', 'BusPUVolt'], [1])

DIgSILENT PowerFactory:

import powerfactory as pf
app = pf.GetApplication()
project = app.ActivateProject('MyProject')
ldf = app.GetFromStudyCase('ComLdf')
ldf.Execute()  # run load flow

Supports both DPL (internal language) and comprehensive Python API.

ETAP: RESTful APIs through ETAP DataHub with etapPy client, enabling local and distributed execution.

Open-source alternative — pandapower:

import pandapower as pp
import pandapower.networks as pn
net = pn.case9()
pp.runpp(net)
print(net.res_bus)  # voltage results

Complete Python-native environment — excellent for learning before moving to commercial tools.

The fundamental pattern for batch studies:

Load base case
FOR each scenario:
    Apply modifications
    Solve power flow
    IF converged:
        Extract results
    ELSE:
        Log failure
    Restore base case
END FOR
Generate report

PSS/E example:

results = {}
for scenario in scenarios:
    psspy.case('base.sav')
    apply_modifications(scenario)
    ierr = psspy.fnsl()
    if ierr == 0:
        results[scenario] = extract_results()

Critical best practices:

• Always reload base case between scenarios (modifications are cumulative otherwise)
• Check ierr return values — silent failures produce meaningless results
• Log all actions for troubleshooting
• Validate against manual GUI calculations before trusting batch results

Automated N-1 screening workflow:

1. Load base case, solve power flow
2. Record pre-contingency state
3. FOR each contingency (line/transformer/generator):
       Remove element
       Solve power flow
       Check thermal limits, voltage limits
       Record violations
       Restore element
4. Rank contingencies by severity
5. Report violations and critical contingencies

Typical result structure:

For large systems (1000+ contingencies), use PSS/E's built-in accc functions or PowerWorld's contingency analysis — far faster than Python loops.

Sensitivity analysis — sweep one parameter while holding others constant:

FOR load_level in [80%, 90%, 100%, 110%, 120%]:
    Scale all loads to load_level
    Solve power flow
    Record: voltages, line loadings, losses

Produces curves showing how outputs vary with the swept parameter — essential for identifying critical thresholds and operating margins.

Monte Carlo simulation — sample from probability distributions:

FOR trial in range(N_trials):
    Sample: load (normal dist), generation
            (availability), line status (outage rate)
    Apply sampled values
    Solve power flow
    Record outcomes (violations, losses, etc.)

Produces probability distributions of outcomes:

• P(voltage violation) at each bus
• Expected annual overload hours per line
• Loss distribution under uncertainty

Typically requires 1,000–10,000 trials for stable statistics — only practical with automation.

Automated reporting with Python ecosystem:

Typical report outputs:

• Voltage profile plots (pre/post-contingency)
• Branch loading bar charts (sorted by severity)
• Summary tables of all violations
• Sensitivity curves (loading vs parameter)

Engineering best practices:

1. Version control — Git for all scripts, track changes alongside model versions
2. Error handling — wrap all API calls in try/except, log failures with context
3. Validation — compare automated results against manual GUI runs for a subset of cases
4. Documentation — docstrings explaining study purpose, assumptions, and limitations
5. Reproducibility — save input parameters, random seeds, and software version with results
6. Modular design — separate data input, simulation engine, and reporting into distinct modules for reuse across studies