The big idea: the fault-finding loop

At 02:10 the bridge calls the engine control room. The bow thruster start is blocked and the pilot boards in 40 minutes. The alarm list shows twenty new lines. The second engineer is already opening panel doors. Where do you start?

Good fault-finders are not faster with a screwdriver. They are faster at deciding where to look. They follow the same loop every time, whether the fault is a 4–20 mA signal, an earth fault on the 440 V board or a generator breaker that will not close.

The loop

1. Symptom: what exactly is wrong? “Thruster start blocked: PMS shows insufficient power available” is a symptom. “Thruster not working” is not.
2. Information: read the alarm and event list in time order, the drawings, the PLC status, the log book. Ask the watchkeeper: what changed? A filter cleaned, a breaker racked out, a card replaced, a program changed, weather, load.
3. Hypothesis: list two or three likely causes, most probable and easiest to check first.
4. Test: one measurement that proves or kills a hypothesis. Know the reading you expect before you take it.
5. Fix: repair the cause, not the symptom. Resetting a trip is not a fix.
6. Verify: run the system and prove it works under real conditions, with the alarm and protection functions back in service.
7. Record: log what you found, what you measured and what you changed, so the next watch and the superintendent can follow it.

Half-split: the fastest test plan

When a fault can be anywhere along a chain (sensor, cable, junction box, barrier, input card, PLC program, HMI), test in the middle first. If the signal is good there, the fault is downstream; if bad, upstream. Each test throws away half the chain.

With 12 possible places, testing one by one can take 12 tests. Halving needs at most

In this course’s simulation, an earth fault on feeder 9 of a 12-feeder board is found in 9 steps breaker by breaker, and in 4 steps by half-split.

A tiny worked example

A fuel oil pressure transmitter has a range of 0–10 bar on a 4–20 mA loop. The HMI shows 0.0 bar, but the pump is running and the local gauge reads 6 bar. Halfway along the chain is the input card. You read the card’s channel current: 0 mA.

• 6 bar should give \(4 + 16 \times 6/10 = 13.6\ \text{mA}\).
• 0 mA means no current is flowing at all: an open loop (broken wire, loose terminal, blown fuse, failed loop supply, reversed polarity) or a dead transmitter. It is not a program or HMI fault.
• Next test: measure the loop supply voltage at the transmitter terminals. That one reading splits the remaining chain again.

Two readings, and the fault is already in one cable run. That is the whole course in miniature: know what good looks like, measure in the middle, and let each reading cut the problem in half.

What you will be able to do

By the end of Part 3 you will have a symptom-to-first-checks table for the common shipboard faults, the formulas you need at the panel, worked cases with real numbers, step-by-step procedures, and a Python simulation of a 4–20 mA loop and an insulated 440 V board where you can inject faults and watch the detection rules and search methods at work.