Oil & fuel analysis
Taking an oil sample that means something
Taking an oil sample well is most of condition monitoring: a laboratory reports only what arrived in the bottle. If the sample is not representative, the report is accurate and useless at the same time, and the usual response is to distrust the laboratory rather than the sampling.
Six habits fix most of it.
Start by taking an oil sample from a live, turbulent line
Particles settle. Oil drawn from a still reservoir under-reports them; oil drawn from the bottom of a still reservoir over-reports them, and neither is the oil the machine is running on.
Take the sample with the machine at operating condition, from a line where the oil is moving and mixed. Where possible, sample downstream of the components you care about and upstream of the filter, that is the oil carrying the evidence.
Flush the port first
Whatever is sitting in the sampling valve and its dead leg is not your oil. It is a history of every previous sample, concentrated.
Flush five to ten times the dead volume before the bottle goes anywhere near it. On a minimess point this is a few seconds; on a longer tap it is more. Flushing is the single change that most improves repeatability.
Sample at operating temperature
Temperature governs how much water the oil holds and how well contaminants stay suspended. A sample taken cold, from a machine that has stood overnight, cannot be compared with one taken hot last month.
Run the machine for at least twenty to thirty minutes so the oil is at normal operating temperature and thoroughly mixed.
Use a clean bottle, and keep it closed
Sample bottles are supplied to a cleanliness level. An ordinary "clean" container from the stores can carry more particles than the oil you are testing, which shows up as a high count and sends everyone looking for a fault that does not exist.
Open the bottle at the sampling point, fill it, cap it immediately. Do not leave it open on a workbench, do not rinse it with the oil first unless the procedure says so, and do not decant between containers.
Fill it properly, and label it fully
Leave a small air gap for expansion, around 80% full is right. Then label it with:
- Machine and sampling point identification
- Date and running hours
- Oil type and grade, and whether it has been topped up or changed since the last sample
- Anything unusual: a recent flush, a leak, a filter change, a period at reduced load
The last line is the one most often left blank and most often needed when a result looks odd.
Sample the same point every time
More important than any of the above. Two valves on the same circuit will give you two different trends, and trend is what condition monitoring is for.
Fix the point, write it into the procedure, and mark it physically on the machine. If you must move it, note the change and treat the trend as restarting.
Malina supplies certified-clean bottles and sampling guidance with every oil and fuel analysis programme, and will fit permanent sampling points as part of an oil and lubrication engagement. If you would rather we took the first samples ourselves, book an audit. Sampling practice for in-service oils is set out in ISO 3722.
What good looks like
| Machine state | At operating temperature, running, oil circulating |
| Point | Fixed, live line, downstream of components, upstream of filter |
| Flush | 5–10× dead volume before filling |
| Bottle | Certified clean, opened only at the point, ~80% filled |
| Label | Machine, point, date, hours, oil, recent interventions |
| Repeat | Same point, same conditions, every time |
Get these right and a rising particle count means something changed in the machine. Get them wrong and it means somebody used a different valve.
Last reviewed 6 August 2026.