Class I candidates are expected to reason like the engineer who signs the isolation permit, not just recite a higher voltage figure. This chapter works through the mechanics behind HV step-up, earthing, safe access and protection so the numbers hold up under exam pressure.
Every ship's electrical system obeys the same three-phase power relationship, and it is worth keeping in front of you for this whole topic: power is the product of voltage, current and power factor. For a fixed power demand, raising the voltage lowers the current in direct proportion — double the voltage and the current halves.
That single relationship is the whole economic case for HV aboard ship. Cable is sized on current, not power: a smaller current means a smaller conductor cross-section, less copper weight, less space in the cable trunking, and — because resistive loss is I²R — a much smaller power loss for the same length of run. Switchgear, too, gets physically smaller for a given power rating as the current it must interrupt falls. None of this is free: HV brings tighter clearances, more demanding insulation, and a heavier safety regime, so the change only pays for itself once the power involved is large enough. In practice that threshold sits somewhere around a few megawatts, which is why HV shows up on large passenger ships, diesel-electric and other electric-drive vessels, and rarely on a small dry-cargo ship running everything at 440 V.
By convention, low voltage (LV) means up to 1 kV and high voltage (HV) means above it. Shipboard HV systems are typically built around 3.3 kV, 6.6 kV or 11 kV, chosen to suit standard rotating machinery and switchgear ratings rather than picked arbitrarily.
HV is not a different kind of electricity — it is the same power delivered at a higher voltage and a proportionally lower current, and almost every other difference in this chapter follows from that one change in current.
Marine LV systems are deliberately run with the neutral insulated from earth — an IT system, in the standard terminology. With no intentional path to earth, a single fault from one phase to the hull does not by itself drive any significant current, so nothing trips and the vessel keeps running. An insulation-monitoring alarm flags the fault so it can be found and cleared at a convenient moment, because at sea there is usually no alternative supply to fall back on. The real danger with this arrangement is a second, independent fault on a different phase before the first is cleared — that turns two single faults into a full phase-to-phase short, which is exactly why the alarm is treated as an urgent job, not a maintenance note.
HV systems flip that logic. The star point of the generator or distribution transformer is earthed through a neutral earthing resistor, sized so that a line-to-earth fault produces a current that is large enough for protection to detect reliably, but limited enough to keep arcing damage, touch and step voltages, and fault energy within bounds. The resistor value follows directly from Ohm's law applied to the star point.
The trade-off is the opposite of the LV case: an HV earth fault is cleared automatically and quickly rather than alarmed and left running, because the fault energy available at HV — and the consequences of letting it develop into something worse — are much greater than at LV. Continuity of supply is deliberately sacrificed at HV in favour of speed and selectivity.
Do not carry LV insulated-neutral thinking into an HV answer. LV rides through a first fault and alarms; HV is earthed on purpose so that a fault is measurable and gets cleared, not tolerated.
Before anyone works on HV plant, a fixed sequence has to be followed, and every step in it exists because a shortcut at that specific point has killed someone on a real ship. It is worth understanding why each step is there, not just memorising the order.
Two failure modes account for most of the near-misses examiners ask about. The first is confusing switching off with isolation: an open breaker removes supply, but only a proven, locked-off and earthed circuit is isolated. The second is skipping the second proving check on the theory that the first one already showed the instrument working — the whole point of proving before and after is that the instrument's own state at the moment of the real test is what matters.
Prove, test, prove again — in that order, every time. This exact sequence is one of the most reliable oral-exam questions at Class I precisely because it is also one of the most reliable ways to stay alive.
Overcurrent protection on a distribution system is normally time graded: the device closest to a fault is set to trip fastest, and each device further upstream is given a slightly longer delay, typically stepped in increments of around 0.3 to 0.4 seconds. If the nearest device fails to clear the fault in its own time, the next one up trips as a backup — a little later, and taking a larger part of the system with it. Grading is what lets a fault on one feeder trip only that feeder instead of the whole switchboard; lose the margin between two consecutive devices, through a wrong setting or a device that will not open, and a local fault escalates into a full blackout.
Differential protection works on a different principle entirely, and it is worth being clear on why it is faster and more selective than grading. Current transformers measure the current flowing into a protected zone — a generator winding, a transformer, a section of busbar — and the current flowing out of it. In normal operation, and even during a fault outside the zone, those two currents are equal and the difference is zero.
A fault inside the protected zone breaks that balance: current appears from nowhere, or disappears into the fault, so the two sums no longer match. Because the scheme only compares what crosses its own boundary, it inherently cannot see, and cannot be fooled by, a fault anywhere else in the system — there is no time delay to coordinate with other devices, so it can trip instantaneously the moment the imbalance appears. That combination of speed and built-in selectivity is why differential protection is standard on the highest-value, highest-consequence items: generators, transformers and busbars.
Grading achieves selectivity by timing; differential protection achieves it by geometry — it only ever sees faults inside its own zone, so it needs no time delay to stay selective.
Diesel-electric and other electric-drive vessels run their propulsion motors from variable-frequency converters: a rectifier stage turns the fixed-frequency AC supply into DC, and an inverter stage turns that DC back into AC at whatever frequency and voltage the motor needs at a given speed. The rectifier is the part that causes trouble upstream — rather than drawing a smooth sinusoidal current, it draws current in pulses timed to the switching devices, and any non-sinusoidal current can be represented as a fundamental component plus a series of harmonics at multiples of the supply frequency.
Those harmonic currents circulate back through the supply, and generators and transformers were designed and rated around a clean fundamental waveform. Harmonic currents add extra eddy-current and skin-effect losses that rise with frequency, so a machine can run measurably hotter than its load current alone would suggest — overheating without overload is the tell-tale symptom, and it is easy to miss if only the fundamental current is being watched. Triplen harmonics (the third, ninth and so on) add together in the neutral conductor rather than cancelling the way balanced fundamental currents do, which is a particular risk for neutral conductors and earthing arrangements sized only for normal load.
The standard mitigations all work by either cancelling harmonics or keeping them out of the main system. Multi-pulse rectifier arrangements (12-pulse, 24-pulse) split the load across windings that are phase-shifted relative to each other by an interposing transformer, so that specific harmonic orders cancel between the windings before they ever reach the supply. Passive or active filters tuned to the troublesome orders can be added directly. Because harmonic content varies with drive design and loading, total harmonic distortion (THD) is treated as a design and survey matter, with a converter-fed installation commissioned and periodically checked against the distortion limit it was designed to.
A converter-fed machine can cook itself on a current reading that looks perfectly normal — harmonics are heat that never shows up as overload on an ordinary ammeter.
An insulation resistance (IR) test applies a DC test voltage — around 5000 V for HV plant, well above the working voltage but nowhere near a destructive level — and reads the resistance the insulation offers after a set time. It is a genuinely useful first check, but a single IR reading is coloured by winding temperature and by surface moisture or contamination, so a comfortable-looking megohm figure on its own does not prove the insulation is sound.
The polarisation index (PI) gets around that by looking at how the reading behaves over time rather than at a single point. Sound, dry insulation keeps absorbing charge for several minutes after the test voltage is applied, so its resistance reading continues to climb; wet or contaminated insulation is dominated by surface leakage current, which does not fall off the same way, so its reading is comparatively flat. Taking the ratio of the ten-minute reading to the one-minute reading turns that behaviour into a single comparable number.
A PI at or above roughly 2 is generally taken as acceptable for this class of insulation; a flatter curve than that is read as a warning to dry the machine out and re-test rather than to reconnect it. Because PI is a ratio of two readings taken on the same winding a few minutes apart, it is far less sensitive to temperature than either reading alone, which is exactly why it is trusted over a single IR figure.
Where IR and PI test the insulation at or near normal working voltage, an HV pressure test (a dielectric or hi-pot test) deliberately applies a voltage above the normal working voltage for a defined period, to prove the insulation can survive the transient overvoltages and switching surges it may see in service without breaking down. It is a more decisive test than IR, but it also stresses the insulation, so it is run after IR — typically following a repair, a cable re-lay, or as part of commissioning — and always under a written procedure by a competent, authorised person.
IR asks whether the insulation resists current now; PI asks whether it is dry; the pressure test asks whether it will hold when it matters most. They answer three different questions, not one question three times.
The three examples below carry the ideas above through to actual numbers: the cable saving that justifies going HV in the first place, the resistor that makes HV earthing work, and the test that decides whether a winding is fit to go back on the busbar.
A shipyard is re-engineering the feeder to a 4500 kW propulsion motor. The load draws 4500 kW at 0.85 lagging power factor. Compare the full-load line current if the feeder stays on the existing 440 V switchboard against running it from a new 6600 V HV switchboard, and state what the difference means for the cable.
Propulsion load, P = 4500 kW Power factor, cos φ = 0.85 (lagging) Option A — LV feeder: V_L = 440 V Option B — HV feeder: V_L = 6600 V Three-phase power: P = √3 × V_L × I_L × cos φ
Compare the full-load line current if the feeder stays on the existing 440 V switchboard against running it from a new 6600 V HV switchboard, and state what the difference means for the cable
Rearrange the three-phase power formula for line current.
Evaluate it at the present 440 V.
Repeat at the proposed 6600 V.
Nothing about the load has changed — only the distribution voltage.
Check it.
Because P and cos φ are unchanged, the ratio of the two currents should equal the ratio of the two voltages — a useful arithmetic check.
Why it matters.
Copper loss in a feeder is I²R, so a current cut to 1/15th of its former value cuts the resistive loss in the same cable far more than 15-fold, because the loss depends on the square of the current.
AnswerLine current falls from 6947 A at 440 V to 463 A at 6600 V — a 15:1 reduction that lets the HV feeder use a dramatically smaller, lighter cable for the same 4500 kW delivered.
The trap: quoting the 15:1 current reduction as if it were the loss reduction — loss falls with the square of current (225:1 for the same cable), and mixing the two ratios up in an exam answer costs marks even when the current figures are right.
A new 6600 V alternator is to be resistance-earthed. The design calls for a single line-to-earth fault to be limited to 400 A, so that protection can detect and discriminate it. Find the earthing resistor's value, the power it must dissipate during a fault, and the energy it absorbs if the protection clears the fault in 0.4 s — the bottom step of the time-grading scheme.
System line voltage, V_L = 6600 V Target earth-fault current, I_f = 400 A Fault clearance time (bottom grading step), t = 0.4 s
Find the earthing resistor's value, the power it must dissipate during a fault, and the energy it absorbs if the protection clears the fault in 0.4 s — the bottom step of the time-grading scheme
The resistor sits between the star point and earth.
So it only ever sees phase voltage, not line voltage — find the phase voltage first.
Size the resistor so that.
With the full phase voltage across it during a solid single-phase-to-earth fault, the current is limited to the target value.
Check the thermal duty.
For that fraction of a second the resistor carries the full fault current, so it has to survive the resulting I²R heating — a short-time rating, not a continuous one.
Multiply by the clearance time to get the energy the resistor element must absorb in that single event — this, not the megawatt figure on its own, is what its short-time rating (commonly quoted for 10 s or 30 s) is really certifying.
AnswerR ≈ 9.53 Ω, dissipating ≈1525 kW and absorbing ≈610 kJ during a 0.4 s fault — a resistor sized and rated for this duty, not for continuous service.
The trap: confusing the resistor's continuous rating (a few watts, for the small standing unbalance current in normal service) with its fault rating — the two figures differ by orders of magnitude, and specifying the wrong one leaves the resistor to fail during the very fault it exists to limit.
A ship's 6600 V shaft alternator has stood idle for three weeks in humid weather. Before reconnecting it to the HV busbar, the second engineer runs an insulation resistance test with a 5000 V megger, then repeats it after 48 hours of space-heater drying. Decide, from the polarisation index, whether the winding may be reconnected.
Test instrument: 5000 V megger (appropriate for a 6600 V stator winding) Before drying — IR at 1 min = 180 MΩ, IR at 10 min = 300 MΩ After drying — IR at 1 min = 400 MΩ, IR at 10 min = 920 MΩ Acceptance benchmark: polarisation index ≥ 2
Why look past the raw megohm reading.
IR alone is heavily affected by winding temperature and surface moisture, so a single figure — even a large one — does not by itself prove the insulation is dry. The polarisation index compares how the resistance changes as the test proceeds, which cancels out most of that dependence.
Calculate the polarisation index from the first test.
Judge it.
1.67 is below the ≥2 benchmark, meaning the resistance was not still climbing the way dry insulation does — the winding is holding moisture or surface contamination. Reconnecting on the strength of the raw 300 MΩ figure alone would be a mistake; dry it out and re-test instead.
Recalculate the polarisation index after the 48-hour dry-out.
AnswerPI rises from 1.67 (below benchmark) to 2.30 (above benchmark) after drying — the alternator may now be cleared for reconnection to the HV busbar.
The trap: reconnecting on the strength of the raw IR figure alone — 300 MΩ looks reassuring in isolation, but a polarisation index of 1.67 says the insulation is still wet, and a wet HV winding is exactly the condition this test exists to catch.
P = √3·V_L·I_L·cos φLine current halves each time line voltage doubles, for the same loadHV = above 1 kVTypically 3.3, 6.6 or 11 kV aboard; LV stays insulated-neutral, HV is resistance-earthedR = V_phase ÷ I_fault(limit)Sizes the neutral earthing resistor to a chosen, detectable fault currentIsolate → lock off → prove dead → discharge → earth → permitThe access sequence, always in that orderProve tester: before AND afterAn untested tester proves nothing — this is the step examiners probe hardestDifferential: ΣI_in = ΣI_out (healthy)Any imbalance across the protected zone is an internal fault — trips without waiting for gradingTime grading ≈ 0.3–0.4 s per stepEach upstream device waits one step longer than the device below itPI = IR(10 min) ÷ IR(1 min), accept ≥ 2Normalises for temperature and moisture better than a single IR readingTHD commonly held under roughly 5–8 % voltage distortionMulti-pulse rectifiers and phase-shifting transformers are the usual fixHV megger ≈ 5 kV; pressure/hi-pot test above working voltageIR proves insulation resistance; the pressure test proves it survives an overvoltage