OVERVOLTAGES INSULATION COORDINATION Lecture 6 ELECE 8409 High

OVERVOLTAGES & INSULATION COORDINATION Lecture 6 ELEC-E 8409 High Voltage Engineering

OVERVOLTAGES Overvoltage Shape Temporary (Sustained) Overvoltage Test Voltage 0. 1 s - earth fault - load disconnection - resonance and ferro-resonance - open phase connection Slow-Front (Switching) Overvoltage 10 ms - connecting load to network - faults and re-closure - disconnection of load current Fast-Front (Lightning) Overvoltage 250 µs 100 µs - lightning (induced, back surge, straight hit) Very-Fast-Front Overvoltage - arc interruption and restriking 2500 µs 50 µs 1. 2 µs 0. 1 µs No standard yet (equipment type specific)

OVERVOLTAGES U [p. u. ] Lightning Surges (Fast-Front Overvoltage) 6 5 Switching Surges (Slow-Front Overvoltage) 4 3 Temporary Overvoltages 2 (Sustained Overvoltage) 1 t [s] 10 -6 10 -4 10 -2 100 102 104

Temporary Overvoltages a. k. a. Sustained Overvoltages Earth Fault Load Disconnection Resonance Open-Phase (Asymmetric) Connection

EARTH FAULT Most common cause for temporary overvoltages • Overvoltage caused between healthy phase and earth a. Magnitude depends on earthing type: • a. ) Isolated neutral • b. ) Resonant earthed (Peterson coil) • c. ) Direct earthing (magnitude ≈ 1. 8 p. u. ) b. Magnitude given as earth fault coefficient k • Ratio between normal operation voltage U and peak phase-earth voltage Up • Network is “effectively earthed” when k ≤ 1. 4 c.

LOAD DISCONNECTION Zk Zj (network represented as short circuit impedance Z k and single phase emf) ~ U 1 network Yj 1 Yj 2 line U 2 ZL The line is represented by a π-equivalent circuit U 2 is the voltage at the end of the line load Increase in voltage at beginning of line: U [p. u. ] 3 c 2 (Sk = 5) Increase in voltage at end of line: c 2 (Sk = 10) c 2 (Sk = 500) 2 c 1 (Sk = 5) c 1 (Sk = 10) c 1 (Sk = 500) 1 pacitive 0 0 The network supplies voltage U 1 to the beginning of the line Voltage increase greatest when load is inductive and the grid’s short circuit power is small (Zk is large – weak grid) Difference between c 1 and c 2 is the Ferranti phenomenon: 500 s [km] 1000 • approaches the open end of the line

RESONANCE Oscillations of higher amplitude at certain frequency (determined by R, C, and L components in the circuit) Reactance X = opposition to alternating current (caused by the build up of electric or magnetic fields in an element). Total reactance is the sum of capacitive and inductive reactance, X = XC + XL X>0 X=0 X<0 Reactance is inductive Impedance is purely resistive Reactance is capacitive (opposes change in current) (Z = R + j. X) (opposes change in voltage) DC (low f): Capacitor is open circuit (charges balanced with applied voltage = no current) AC: Capacitor only accumulates a certain amount of charge before polarity changes and charges dissipate. Increasing frequency (less charges) decreases opposition to current - As frequency increases, inductive reactance XL increases while capacitive reactance XC decreases - At a particular frequency these two reactances are equal in magnitude but opposite in sign Different resonance states and nonlinear components can cause temporary overvoltages

FERRO-RESONANCE Oscillations caused by series connection of capacitance and nonlinear inductance - causes temporary overvoltages concurrent with harmonic distortion Voltage transformer connected to an unearthed system U L Resonant circuit formed by the earth capacitance of the system and the nonlinear inductance of the voltage transformer connected between phase and earth UV UC Nonlinearity effect – inductance decreases as current increases when saturation occurs giving resonant conditions at basic frequency (or multiples). Resonance may cause the voltages and currents to jump from one state to the other and current overloading of the transformer may lead to thermal damage.

ue ical d. Iif P 2: Operation should be where the difference in UL and UC is equal and in the same direction to UV • P 1: stable operation when d. UL/d. I > d. U/d. I. Circuit returns to P 1 after short current surges (e. g. switching operations). • increase in current above I 2 results in a larger voltage UV than required to maintain increasing current. Thus, current increases further. - Jump-resonance – transition from P 1 to P 3 is called jumping (FI: kippaus) UV UL U P 1 P 2 UC U = |UV – UC| –|UC| |UL| U U Inductive (UL > UC) XL(lin) > XC UV = UL – UC Capacitive (UC > UL) XL(sat) < XC UV = U C – UL UL UC UV UL - U C I 3 UL I 2 I UC UV P 3 I I 1 P 2 UV P 3 I

For continuous operation, jumping occurs during each half cycle: 1. P 2 P 3 P 1 UL UC Induction law: 2. Flux density reaches +BS. Inductance decreases. A rapid increase in current follows. 2 – 3. Inductor is saturated and large current flows in the circuit. As a consequence of the current surge, the capacitor charge changes from –UC to +UC. t UV +Bsat t i –Bsat 1. The capacitor has just become charged. Due to the voltage difference between UV and UC, voltage UL increases forming a voltage-time area (displayed in grey). As this area becomes larger, the magnetic flux of the iron core decreases. 2. 3. 4. 5. 3. The inductor’s voltage becomes negative. Since there is no current flowing in the circuit, the capacitor’s voltage UC is constant. 4. Saturation limit is onec again achieved 4 – 5. 5. The increasing U, t-area changes the flux from +Φ to –Φ. Inductor is saturated during which currents are large. Process repeats itself from point 1.

ASYMMETRIC CONNECTION (OPEN PHASE) Burnt Fuse Broken Conductor Switchgear Malfunction One or two phases are disconnected

Slow-Front Overvoltages a. k. a. Switching Overvoltages Load Connection Applied Voltage and Re-closure Faults Disconnection of Load Current

Previously called Switching Overvoltage Caused in most cases by changes in the grid • Faults – short circuit, earth fault, load disconnection, asynchronous operation • Switching operation – opening or closing of circuit Switching operations can cause significant stress between switchgear terminals (contact gap) Magnitude and shape of overvoltage depends on switchgear used for current disruption and network properties (L, C, load) instantaneous configuration network and value of voltage and current at the moment when switchgear is opened or closed

a. Busbar Short Circuit b. Line Fault c. Asynchronous network d. Disruption of Small Inductive Load e. Disruption of Capacitive Current f. Voltage Applied to No-Load Line

LOAD CONNECTION Standard procedure in network (causes a slow-front overvoltage) CONNECTING A CAPACITOR Peak value of overvoltage depends on instantaneous voltage at moment of switch closure • Maximum peak value is 2. 0 p. u. • Angular frequency ω caused by connection is typically ~ 100 Hz Asynchronous closure Earthing technique Resonance Superpositioned oscillations over 2, 0 p. u. CONNECTING A MOTOR Similar peak value as capacitor (2. 0 p. u. ) • Steeper overvoltage • Voltage stress concentrated at the beginning of the winding (not evenly distributed)

NO-LOAD LINE VOLTAGE APPLICATION U Applying voltage to a no -load line is one of the major causes for overvoltages at high operating voltages (≥ 245 k. V). • Applied voltage creates a travelling wave which doubles the voltage once it reflects back from the end of the line 2 U U -U 3 U

NO-LOAD LINE VOLTAGE APPLICATION Voltage at end of line after closure: no trapped charge (disconnected from network for some minutes) 1. with trapped charge -1, 0 p. u. (when re-closed, trapped charge is seen as an opposite sign voltage) 2. Ideal Zk = 0, no losses 22. 2 1 1 1. 0 -1 -2 -3 0 4 8 t [ms] 12 ~ s = 430 km t = s/c ≈ 1. 43 ms u(t) = cos ωt Real Zk ≠ 0, losses included 3 ur [p. u. ] Zk = 0 16 20 u. R

FAULTS Different network failures can cause overvoltages (earth fault is most common) • The onset of a fault and its removal both cause transients Tripping action results in overvoltage Fault causes voltage drop Circuit breaker is tripped (opened) to remove voltage drop Network side voltage oscillates and settles eventually at value determined by the network supply voltage Typically, slow-front overvoltages related to faults do not exceed: Onset of fault: umax < 2 k – 1 Removal of fault: umax = 2 k = earth fault coefficient

DISCONNECTION OF LOAD CURRENT When a switch is opened, arcing may occur over the gap between terminals Basic Situation: • Arc is permanently extinguished at zero current, followed by a voltage transient. Voltage drop in the network caused by current (arcing) oscillates and attenuates and voltage settles at the continuous operation level. It takes some time for ionization in the contact gap to disappear and for the switch to regain its insulating properties Immediately after arcing is extinguished (zero current), voltage over the contact gap is formed by the supply network and the load side potential difference (recovery voltage ur) i(ωt) u(ωt) ~ u. R Z

DISCONNECTION OF LOAD CURRENT i(ωt) u, i [p. u. ] uload = 0, ωt > ωt 1 1. 0 ωt 1 i(ωt) usupply(ωt) RECOVERY VOLTAGE uload = 0, ωt > ωt 1 ωt usupply(ωt) RECOVERY VOLTAGE -1. 0 ωt 1 ωt ωt 1 -1. 0 ul 1. 0 i(ωt) usupply(ωt) i(ωt) u, i [p. u. ] ur ~ RECOVERY VOLTAGE uload = -1, ωt > ωt 1 Load Z = only RESISTANCE Load Z = only INDUCTANCE Load Z = only CAPACITANCE R jωL 1/ jωC • No phase difference between supply voltage and load current • If current is disrupted at first zero point, load side voltage remains at zero • Recovery voltage increases along normal sinusoidal fluctuation of supply voltage • 90° phase shift between voltage and current • 90° phase shift between current and voltage • When current is disrupted at zero level, voltage is at peak value (= onset recovery voltage) • When current is disrupted at zero level, charge equal to voltage at moment of disconnection remains in capacitor (load) • The initial steepness of the recovery voltage is large but the peak value is still the same as the supply voltage • Recovery voltage increases from zero to twice the peak value of the supply voltage Z

DISCONNECTION OF LOAD CURRENT If the recovery voltage exceeds the voltage withstand strength of the contact gap Restrike Magnitude of overvoltage depends on moment of occurrence Reignition Restrike • Voltage at both terminals have same polarity small overvoltage • Voltage at terminals have opposite polarity large overvoltage Most common cases of restriking occur when: Disconnecting CAPACITIVE current Disconnecting small INDUCTIVE current

• Disconnecting Capacitive Current Once arcing is extinguished, a trapped charge (value = peak voltage) remains in the capacitor (a. ) • If switch cannot regain insulating properties fast enough (recovery voltage higher than withstand voltage), restriking occurs at peak of supplied voltage (b. ) • No-load line or cable disconnection from grid • Restriking creates another transient which can provide an even greater trapped charge to the capacitor • Disconnection of storage capacitor ~ a. L û sin (ωt) C CS û sin (ωt) u. R uc uc u. R L, C and Cs define oscillation frequency of transient. If restriking continues repetitively at the peak value of the supplied voltage, an extremely large overvoltage is created. uc b. û sin (ωt) uc t t u. R i u. R t uc i t

LS Disconnecting Small Inductive Current • Disconnecting transformer no-load current • Disconnecting HV motor starting current • Disconnecting reactor current ~ u(t) = √ 2 U cos (ωt) C 1 u. L C 2 (small inductive current = cos φ < 0. 5 and significantly smaller than the breaker’s breaking capacity) Disconnection may occur before current has reached zero level because the breaker’s breaking capacity significantly exceeds the magnitude of the current to be interrupted. Energy remains in the load inductance L 2 and causes the LC circuit (formed by L 2 and C 2) to oscillate. Peak voltage at C 2: L 2

Disconnecting Small Inductive Current: Restriking occurs several times until the withstand strength of the contact gap exceeds the recovery voltage stress Transient voltages increase rapidly and restriking often occurs considerably earlier before voltage has reached its maximum value WITHOUT restriking WITH restriking u. L i i i 0 u. L i 0 t 0 t u 0 u 1 u 2 In this case, restriking limits overvoltages i 0 u 1 u 2 1. = = = current at moment of interruption (chopping current) voltage at moment of interruption peak value of oscillations with restriking peak value of oscillations without restriking increase in dielectric strength of contact gap 1.

Fast-Front Overvoltages a. k. a. Lightning Overvoltages Direct Strike to Conductor Back Flashover Induced Overvoltages

Previously called Lightning Overvoltage Typically caused by lightning: • Direct strike to conductor • Back flashover via grounded components • Induced by nearby stroke Lightning is a very large leader discharge starting from clouds or ground Requires strong up-flow of air mass and high humidity of the rising air Cold and warm air masses meet Air heated by the sun rises up Not all factors and mechanisms for the formation of thunder clouds fully understood

LIGHTNING Lightning discharge begins where the charges increase the electric field above the breakdown strength of air (~1 MV/m for air inside a water droplet) • the lightning flash can consist of numerous subsequent strokes traveling along the same channel and also branching discharges which terminate in air km 14 °C Direction of Motion -48 ICE 10 -32 + + + + Particles (ice, snow) inside cloud collide due to the strong up-flow of air and become charged. + + + Negative charges – heavy particles (snow/hail) accumulate at mid section and lower area of cloud + – – ––– –– –––––– –– –––– – – – + + + – – + + + + – – 4 Warm Air 0 WATER, VAPOR 0 Cold Air SNOW +24 – Positive charges – small ice crystals at the top of the cloud. Commonly also a small area of positive charges are the bottom – + + + + – – Potential difference inside the cloud can reach ~GV – + + Induced Charges at Ground

Negative Lightning Discharge (cloud to ground) 1. ) Leader discharge begins from cloud towards ground 2. ) Destination of lightning stroke determined ~ 100 - 150 m from ground 1. 3. ) Breakdown strength of air is exceeded and a leader channel starts from the ground towards the opposite charged discharge ++++ ––––––– – – – – ––– – – +++++ ––––––– + + + + + ––––– – – –––– – + ++++ + 5. 5 - 6. ) Subsequent strokes can form from other negatively charged areas in the cloud using the same discharge channel 3. 2. ––––––– – ––––– 4. ) When the two discharges meet, a main stroke (return stroke) travels from ground to cloud discharging the negative charge area (where the leader started) in the cloud + + + ––––––– – – – –– – – – ++++ – – – – + ++++ ––––– – – –––– – 6. + + – – – + –– – +– – – ++ + – – + + +++ – – – – + + + + +

Return Stroke Negative Main Stroke i [p. u. ] Milder slope 0 Larger current Steeper 0 0. 2 0. 4 5% 0. 6 0. 8 – 10 0 5 t [µs] 95% 1. 0 95% – 5 50% 0. 8 1. 0 1. 2 5% 0. 6 50% Smaller current 1. 2 10 – 5 0 5 t [µs] 10

DIRECT STRIKE TO CONDUCTOR Typically Zw = 250 – 500 Ω • Overvoltage ≈ MV • Corona causes losses which attenuate and flatten overvoltage U [k. V] 0 m 1600 620 m 1300 m 2200 m 1200 800 400 0 0 1 2 3 4 t [µs]

BACK FLASHOVER Lightning strike to grounded line components (pole or lightning shield wire) Reflections from neighbouring poles and the pole itself significantly alter the voltage waveform ZT 1 = 0 τ1 τ1 Z 1 τT ZT Rf u u 0 τT = propagation time in pole τ1 = propagation time to adjecent pole ZT 2 = 0 (lightning impulse assumed as step) 2 t. T If voltage u exceeds the voltage withstand between the grounded component and the live phase conductor: Flashover from grounded component to phase conductor • backward flashover called back flashover Highest probability when lightning current is high or poor earthing conditions (large earth impedance) 10 t. T = 2 t 1 20 t. T = 4 t 1 Flashover occurs between the phase with the largest opposite voltage relative to the lightning overvoltage t

INDUCED OVERVOLTAGE Overvoltage induced by a lightning stroke in the vicinity of a conductor or equipment not channel) is perpendicular to the conductor Lightning current cause a rapidly changing magnetic field into the lc loops of the line inducing a voltage: i k = propagation speed of discharge current (constant ≈ 1. 2 – 1. 3) Z 0 = 1/4π √(µ 0/ɛ 0) = 30 Ω (constant) i = peak lightning current h = height of conductor d = distance of stroke from conductor Induced voltages are typically smaller (200 – 300 k. V) and slower (front time c. 10 µs) d h

Very-Fast-Front Overvoltages

VERY-FAST-FRONT OVERVOLTAGES a. k. a. Very Fast Transients (VFT): in practice restricted to transients with frequency above 1 MHz Typical for (HV) disconnector operation 1500 1000 i [A] 500 0 – 500 1 2 3 t [µs] 4 5 Overvoltage caused by arc interruption and restriking when opening disconnector (GIS faults, switching of motors and transformers with short connections to switchgear, certain lightning conditions) E. g. air insulated switching station: • Steep transients attenuate quickly • Only dangerous to equipment located close to the disconnector (heats wiring, causes internal resonance) Tens of restrikings during opening • Each restrike generates high frequency oscillations • Oscillation frequency typically 100 k. Hz – 50 MHz • Discharge currents can reach 2 – 3 k. A. Shape of VFT (IEC 71 -1): time-topeak < 0. 1 µs, total duration < 3 ms, superimposed oscillations with f ranging 30 MHz – 100 MHz.

INSULATION COORDINATION

OVERVOLTAGE PROTECTION Protection levels: 1. Avoid direct impact of overvoltage by directing it towards designated routes (lightning conductors, shield wires, and Faraday cages) 2. Ensure basic impulse level BIL (withstand level) is not exceeded using HV protection elements: Surge Arrester Spark Gap (FI: venttiilisuoja) 3. Extra protection for sensitive equipment telecommunication) (LV filters for computers and

Surge Arresters Spark Gap with Non-linear Resistor Magnetic Blow-out Arrester Metal-Oxide Varistor

OVERVOLTAGE PROTECTION Surge Arrester Decrease magnitude of overvoltage in network Traditionally located at substation • Protects only most important equipment – transformers, GIS - used in areas (FIN) where lightning density is low (intensified protection not necessary) • Placed at all incoming lines to substation on line-side of feeder circuit breaker - all equipment has some level of protection - protection level decreases with distance between surge arrestor and protected device Also located at poles • Decrease back flashover in areas of high lightning density and poor earthing conditions (not economically feasible in Finland)

NONLINEAR RESISTANCE TYPE ARRESTER Ideal • When voltage exceeds peak operating voltage, the arrester becomes conductive (weak resistor) allowing the surge energy to be discharged without increasing voltage over the protected device. • Immediately after excess energy is discharged, the arrestor regains its insulting state Reality • Limited capacity energy discharge (only applicable to relatively short duration overvoltages) • Discharge of overvoltage is not immediate • Leakage current is present even in insulating mode

NONLINEAR RESISTANCE TYPE ARRESTER 1. Nonlinear resistor, 2. Disc spark gap, 3. Active spark gap, 1 4. Blow-out coil, 5. Shunting resistor 1 1 5 4 2 3 5 4 5 1 1 Disk Spark Gap with Nonlinear Resistor (silicone-carbide gap type) 3 4 1 Magnetic Blow-Out Arrester (active gap surge arrestor, expulsion type) Metal-Oxide Varistor

NONLINEAR RESISTOR TYPE WITH GAPS Disk spark gap (2) in series with Si. C resistor (1) encased in a porcelain shell 2 Dividing the spark gap into sections decreases breakdown voltage scatter and flattens the steep transient resulting from flashover. The nonlinear resistor limits the earth fault current so that arcing is extinguished by itself: high currents low resistance 1 low current high resistance u 1 us ip As voltage over the arrestor exceeds sparkover (striking) voltage us, the spark gap is ignited. u u s ur ur ip u ij t i u 1 = overvoltage peak (without arrestor) u = normal operating voltage ur = residual voltage us = sparkover (striking) voltage ij = follow-through current ip = surge current peak Surge current ip grows to a value determined by the overvoltage magnitude Residual voltage ur (maximum voltage over arrestor during operation) is determined by the discharge current and nonlinear resistor magnitude After the overvoltage has passed, the arrestor remains conductive and follow-through current ij (fed by the power frequency voltage) is present until the spark gap is extinguished (voltage becomes zero)

MAGNETIC BLOW-OUT ARRESTER parallel resistance coil metal electrode spark gap formed when piling elements (rings) together ring parallel resistance & coil

MAGNETIC BLOW-OUT ARRESTER u t i

METAL-OXIDE VARISTOR u covered by a metal oxide surface layer parallel inside porcelain/polymer shell tive gaps can be left out (e. g. R(normal operation) = 1. 5 MΩ, R(discharge) = 15 Ω) No rapid voltage changes E [V/mm] 500 No breakdown voltage scatter Nonlinearity of Zn. O vs. Si. C 200°C 200 25°C 100 Zn. O 150°C Si. C 60 30 10 -9 Area 1 Area 2 10 -6 t 10 -3 J [A/mm 2] Area 3 100 10 2 Insignificant back current Area 1: Zn. O penetrating current decreases radically under voltage threshold value (high resistivity). Poorly conductive surface layer determines magnitude of current. • i

METAL-OXIDE VARISTOR

ARRESTOR SELECTION The arrestor must be selected so that the margin between protection level of arrestor and the device’s withstand level is large enough. Safety Margin Protection level Withstand level U Urp = Ucw = kc = representative overvoltage withstand level of device protection factor The protection level must be set high enough to avoid arrestor operation under normal continuous operating voltage but also low enough to avoid overvoltages above the withstand level Margin exists only if arrestor is infinitely close to the protected apparatus Otherwise, must consider: • • Voltage increase in line caused by propagating overvoltage (superposition of traveling waves) Voltage drop caused by surge current at earthing conductor and arrestor connection (coupling)

ARRESTOR PLACEMENT Protected device (T) is at a distance D from the arrestor (A) • The front of the voltage pulse is linear • Inductance of earthing circuit assumed insignificantly small u Effective Protection Level: up(eff) ∆u 2 2∆u 2 SD/v ∆t∙v ∆u 1 up (d 1 + d 2)l ∆i ∆t distance D d 1 A d 2 T up Δu 1 Δu 2 d 1 d 2 l D S v = = = = = rated protection level of arrestor inductive voltage loss at earth and joint coupling voltage increase between arrestor and protected device length of arrestor connection length of arrestor earthing inductance of joint and earthing conductor (~1 μH/m) distance between arrestor and protected device steepness of linear impulse voltage propagation speed of impulse voltage

ARRESTOR PLACEMENT E. g. A 1500 k. V/ μs steep propagating wave is approaching a transformer along a 123 k. V line. The voltage withstand level of the transformer is 550 k. V. The arrestor is located 10 m away from the transformer and has a protection level of 380 k. V. Voltage drop Δu 1 caused by joint and earthing coupling (d 1, d 2) is assumed to be 20 k. V. Distance and junction results in a 32% increase in protection level • Safety margin reduced from 170 k. V to 50 k. V • Protection factor reduced to kc = Ucw/Urp = 550/500 = 1. 1 Effective protection level less than withstand level of transformer OK If S = 2250 k. V/μs, withstand level is exceeded. To protect against steep impulses • bring arrestor closer • select arrestor with lower protection level up

ARRESTOR PLACEMENT a) Transformer Protection b – e) Cable Protection Short cables (30 – 50 m): Arrestors at end of cable (c) Longer cables: Riskofbackflashover. Arrestorsatbothendsofcableoruselightning shield wire and minimize earthing resistance. Important to ground arrestor and cable sheath to same point (b) f) Protection of important line-side measuring equipment g) GIS, RMU protection - arrestors at all line outputs

GENERATORS AND MOTORS 500 – 600 m a) Straight connection to overhead line: 0. 1µF d > 500 m Typically 500 m distance between arrestors with protective capacitor (reflections) b) Connection to overhead line via cable: Capacitor not needed when distance is over 500 m Phase-earth and phase protection: a) 6 separate arrestors b) 4 arrestor group a b

Spark Gap

SPARK GAP Simple device consisting of two electrodes – one connected to the conductor to be protected and the other to ground. d Spark gaps form a weak point enabling overvoltages to flow to earth instead of to the protected device. Breakdown voltage can be adjusted Surge arresters are more expensive and require monitoring (arrester can fail) Cheaper and simpler solution for protecting smaller pole transformers is to use a spark gap • at most 240 k. VA, 24 k. V transformer (FIN) • transformer must withstand spark gap overvoltage and steep voltage transient d/2

SPARK GAP Voltage-Time Curve: 300 0 k V/ µs k. V/µ s 1000 u k. V 50 400 /µs V/ µs 0 k Double gap 2000 k. V 300 50 2000 k. V u k. V 1000 /µs 400 k. V/µ s Voltage-Time Curve: 120 mm 90 mm 200 120 mm 100 mm 200 80 mm 60 mm 100 0 0 Single gap 0. 2 0. 4 0. 6 0. 8 t / µs 1 0 0 0. 2 0. 4 500 k. V/µs: Direct lightning stroke to conductor 1000 – 2000 k. V/µs: Back flashover (rare) 0. 6 0. 8 t / µs 1

SPARK GAP Inter-electrode distance d of spark gap: Wet Test 100 80 U k. V 60 double gap 40 single gap 99 % protection level (U 50 + 2. 3 s) 20 1 % ignition level (U 50 – 2. 3 s) 10 0 20 40 60 80 100 120 • Large enough to avoid breakdown by temporary overvoltages and small transients • Small enough to protect against fast-front transient voltages (lightning) Problems with spark gaps: Gap operation causes an earth fault 140 160 d / mm Dry Test Short zero voltage period needed to remove fault (requires fast reclosing system) Polarity dependence 100 80 99 % protection level (U 50 + 2. 3 s) U k. V 60 Weather conditions Temperature, humidity, and pressure affect ionization Large operating voltage spread 40 Up to 40%, also dependent on overvoltage shape, i. e. steepness 1 % ignition level (U 50 – 2. 3 s) 20 Spark gap implementation: 10 0 20 40 60 80 100 120 140 160 d / mm • Reasonable number of atmospheric overvoltages • Short outages allowed

SUMMARY Overvoltages: • • Temporary (sustained) Slow-front (switching) Fast-front (lightning) Very-fast-front Insulation coordination: • Surge arrestors • placement • Spark gaps