In short. Root-cause diagnostics is what a repeat generator fault deserves after ordinary repair has failed to end it. It separates the initiating failure from the damage that followed, lists every mechanism that could produce the evidence, tests each one against measurements and history, and closes only when the corrective action has been verified under the conditions that caused the fault. Replacing the same part a third time is a symptom of a cause nobody has found.
When a repair becomes an investigation
Breakdown work answers the question “what has failed”. Root-cause work answers “why did it fail, and why will it not fail again”. The first is enough for a battery at the end of its life. The second is needed when the evidence does not fit, when the same component fails repeatedly, or when a recurrence cannot be accepted. Typical triggers we see around Hyderabad are a third AVR in eighteen months, a set that overheats every April but passes every service, and an alternator bearing that fails within months of replacement.
Each of these has had parts fitted and none has had its cause removed. The move from breakdown diagnostics to a root-cause investigation is a decision to stop treating symptoms, taken deliberately rather than by drifting into a fourth repeat visit.
First failure versus consequential damage
A failed generator usually presents a pile of damage, and the most visible item is rarely the first one. An alternator with burnt windings may have been destroyed by a rotating diode that shorted a week earlier. A seized turbocharger may have been starved of oil by a blocked feed pipe, which in turn was blocked by sludge from extended oil change intervals. If the investigation starts at the most dramatic component, the sequence is read backwards and the wrong item is blamed.
The discipline is to reconstruct a timeline. The controller event log, the service file, the operator’s account and the condition of each part are placed in order, and each piece of damage is asked whether it caused the next one or was caused by it. Where the chain begins is the first failure, and that is the only point where prevention works.
Test competing explanations, not the favourite one
Every experienced technician has a favourite cause for a given symptom, and that bias is what a root-cause method is designed to control. For a recurring high coolant temperature shutdown, the credible mechanisms include a radiator core restricted by dust or scale, a slipping fan belt, hot air recirculating into the room inlet, a sticking thermostat, an overloaded set, degraded coolant or a sender reading high. Several can be present together.
| Candidate cause | Evidence that supports it | Test that distinguishes it |
|---|---|---|
| Radiator core restricted | Rising temperature trend over months, visible fouling, high top-to-bottom tank temperature difference small | Measure air pressure drop across core, inspect with light behind fins, compare inlet and outlet coolant temperature |
| Hot-air recirculation | Temperature rises only on hot afternoons and on load, room air at intake measurably above ambient | Log room inlet air temperature against outdoor ambient while on load; smoke test the airflow path |
| Overload | Shutdowns coincide with specific plant operations | Record kW and current per phase against time; compare with site rating after derating for ambient |
| Thermostat or water pump | Temperature climbs quickly even on light load, hoses show uneven heat | Check thermostat opening in hot water; measure coolant flow or pressure differential |
| Sender or wiring | Controller reading disagrees with hand instrument | Measure block temperature independently at the moment of alarm |
The table is illustrative; the method is general. Each candidate gets a test that would rule it in or out, and the conclusion follows the results rather than the opinion held before the visit. Where a cooling fault is confirmed, correction is scoped under cooling and radiator service, and the room side under generator room ventilation.
Using the controller as a witness
Modern controllers record far more than the operator sees: timestamped event logs, the engine and electrical parameters captured at each shutdown, trip counters and often trend data. On electronic engines the ECM holds its own fault history with freeze-frame values. Reading these before anything is cleared converts anecdotes into facts: the set was at 82 percent load, coolant had been rising for eleven minutes, and the previous three shutdowns all fell in the same hour of the afternoon. That kind of record separates a radiator problem from a thermostat problem in one reading. Interpretation of this data is the subject of controller diagnostics.
The eight steps, verification and prevention
A root-cause investigation follows a repeatable path: define the problem precisely, collect the history, gather measurements, list the possible causes, test each cause, confirm the root cause, correct it, and retest under the conditions that produced the fault. The written record keeps the reasoning and the verification together, so that a reviewer two years later can see what evidence supported the conclusion and whether the fault recurred.
Verification is the step most often skipped. A set that overheated every hot afternoon on load must be run on load on a hot afternoon after the correction, not idled on a cool morning and declared cured. Where the cause cannot be reached with the access or data available, the honest record says so and states what further test is needed.
The last output is a change, and it is often not to the component. It may be a radiator clean scheduled before every summer, a duct that stops recirculation, a load-shedding rule, or a parts policy that insists on the correct insulation class for an alternator in a hot room. If a part was the root cause, the replacement must be verified as the right part for the build under compatibility verification, and the finding feeds the preventive maintenance plan rather than being filed as a closed job.
Common mistakes
- Blaming the most damaged component, which treats the end of the failure chain as its beginning and leaves the initiating fault in service.
- Stopping at the first plausible explanation without testing the alternatives, which is how the same site gets three different diagnoses from three visits.
- Clearing the controller and ECM logs before they are read, which destroys the timeline that would have settled the question.
- Verifying the fix under easy conditions, so a fault that only appears on load in hot weather is declared solved in February and returns in May.
- Presenting an uncertain cause as established fact, which closes the file and guarantees surprise when the fault returns.
What to send us
For a root-cause review send the generator and engine identification, the full history of the recurring fault with dates, the service and repair reports for each visit, photographs of the failed parts if they were kept, the controller event log or screen photographs, and the operating conditions when the fault appears, including load, time of day and weather. Tell us what has already been replaced. Use the service and AMC enquiry form and mark it as a repeat fault.
Questions engineers ask
How is root-cause diagnosis different from a normal breakdown repair?
A breakdown repair identifies the failed component and restores operation. A root-cause investigation asks why that component failed, reconstructs the sequence of events, tests the competing explanations and verifies that the correction removes the initiating cause. It is justified when a fault repeats or when a recurrence would be costly.
Why does my generator keep failing the same component after replacement?
Because the replacement treated a consequence rather than a cause. Repeated AVR failures often trace to voltage spikes from a failing diode, loose sensing connections or an overheated panel. Repeated bearing failures often trace to misalignment, belt tension or vibration from an engine fault. Until the upstream condition is found and corrected, each new part inherits the same environment and fails the same way.
What information from the controller is most useful?
The timestamped event log, the parameter snapshot at each shutdown, and any trend data. On electronic engines the ECM fault codes with their freeze-frame values add engine-side detail. These should be downloaded or photographed before any reset, because once cleared they cannot be recovered.
Can a root cause be confirmed without stripping the engine?
Often, yes. Trend records, independent measurements, oil analysis, compression or blow-by tests, thermal imaging and a controlled load test narrow the cause considerably before any disassembly. Stripping is justified when the evidence points inside the engine or alternator and the decision between repair and replacement depends on what is found.
What if the investigation cannot find a definite cause?
The report says so, states which causes were eliminated, which remain possible and what measurement or test would settle it. Often the set is returned to service with added monitoring, such as logging temperatures or recording load at each alarm, so that the next occurrence is captured properly instead of lost. That is a legitimate and honest outcome.
Technical content reviewed by Srinivas Reddy, Managing Partner, Reddy Generators. Last updated 19 October 2026.