10.1 Systematic Troubleshooting and Root Cause Analysis

Method beats intuition

Experienced mechanics diagnose quickly because they have seen the failure before. That works until they meet a failure they have not seen, at which point pattern matching becomes guessing — and guessing means replacing parts until the symptom stops, with no idea which replacement fixed it.

A systematic method produces a defensible answer on an unfamiliar machine and, just as importantly, produces a record of what was tested.

The troubleshooting sequence

1. Gather information before touching anything. What changed? When did it start? Is it continuous or intermittent? What did the operator hear, see, or smell? Were there recent repairs, product changes, or process upsets? The operator who runs the machine every day knows more about its normal behavior than anyone.
2. Verify the symptom yourself. Confirm the problem exists as reported and observe it directly. A large share of "equipment faults" turn out to be instrument faults, valve line-up errors, or a misunderstanding of normal behavior.
3. Review documentation. P&ID, equipment manual, prior work orders, and the previous failure history. A machine failing the same way for the third time is telling you something.
4. List the possible causes and rank them by probability and by how cheap they are to test. A five-minute check that eliminates a whole branch comes before a four-hour teardown.
5. Isolate the fault, testing one variable at a time. Changing two things at once destroys the information the test would have given you.
6. Repair or replace, following lockout and the correct procedure.
7. Verify the repair under real operating conditions, not just at idle.
8. Document what was found, what was done, the readings taken, and the parts used.

The half-split technique

When a fault could be anywhere along a chain — a long piping run, a lube oil circuit, a conveyor line, a control loop — testing components one at a time from one end is slow. The half-split method tests at the midpoint instead.

Each test tells you which half the fault is in, so each test eliminates half of the remaining possibilities. A system with 16 possible locations is narrowed to one in about four tests instead of up to sixteen.

Worked example. Lube oil is not reaching a bearing at the far end of a long distribution header. Rather than checking each branch in sequence, crack the connection at the midpoint of the header.

• Oil present at the midpoint: the fault is downstream of it. Split the downstream half.
• No oil at the midpoint: the fault is upstream — pump, filter, relief valve, or the first half of the header. Split the upstream half.

Two conditions make half-split work: the system must be roughly linear, and there must be an accessible, testable point near the middle. Where the system is a branching tree rather than a chain, split by branch instead.

Root cause analysis: the failure behind the failure

Replacing a failed bearing restores the machine. It does not explain why a bearing rated for years failed in four months. Root cause analysis (RCA) asks that question and keeps asking until the answer is something the organization can actually control.

RCA distinguishes three levels:

Stopping at the physical root means the next bearing fails the same way.

The 5-Why technique

Ask "why" repeatedly, following the chain of causation. The count of five is a guide, not a rule — stop when you reach a cause you can control.

Worked example.

• The conveyor gearbox seized. Why? The bearings ran dry.
• Why? The gearbox oil level was below the bearing.
• Why? Oil leaked past the output shaft seal over several weeks.
• Why? The seal was pushed out by internal pressure.
• Why? The gearbox breather was painted over during the last unit paint job, so the case could not vent as it heated.

The actionable root cause is a masking procedure that does not protect breathers, not "a bad seal." Replacing the seal without clearing the breather guarantees a repeat.

The discipline that makes 5-Why work: each answer must be verifiable evidence, not speculation. "Why? Probably because it was overloaded" is where 5-Why investigations go wrong.

The fishbone diagram

A cause-and-effect (Ishikawa, or fishbone) diagram puts the problem at the head and organizes candidate causes along branches, so the team considers whole categories rather than fixating on the first idea. Common branches for maintenance work:

Fill every branch before evaluating any of them. The value of the tool is that it forces the team past the obvious.

Failure evidence is the raw data

Everything in Chapters 6 through 9 is troubleshooting input. The wear pattern on a bearing, the appearance of a seal face, the drag lines on a cut, the ball path on a race, and the tooth contact pattern in a gearbox are all physical records of what the machine was subjected to. Keep failed parts until the investigation is complete, photograph them in place where possible, and note the orientation before anything is disassembled.