Oil analysis for gearboxes and hydraulics

The main drive gearbox on the mixer lets go on a Tuesday afternoon. Your fitter drains it to see how bad it is. What comes out looks like grey paint with glitter in it.

Nobody's surprised. It had been running warm since the spring. It had a new noise in the summer. And the oil had been that colour for months, if anyone had looked.

Now you need a gearbox that's eight weeks out, a crane to lift it, and someone who has changed one before. The line waits for all three.

Gearboxes and hydraulic systems rarely fail without warning. The warning is in the oil, and most sites never read it.

The oil knows first

Every gear tooth, bearing and pump vane that wears leaves metal in the oil. Every seal that lets in dirt or water leaves a trace too. A lab can measure both, long before the wear is bad enough to hear or feel.

A standard report tells you three things.

What's wearing. Iron points at gears and bearings. Copper or tin can point at bronze worm wheels, bushes or cages. A jump in one metal, with the others flat, tells you which part to worry about.

What's got in. Silicon usually means dirt, so look at the breather and the seals. Water means a failed seal, a cooler leak, or condensation in a box that runs hot then sits cold. On a food site, check for wash-down water first.

What state the oil is in. Viscosity, oxidation and acid number tell you whether the oil itself is worn out. A gearbox can be fine while the oil is finished.

For hydraulics, add a particle count. Dirty oil wears pumps, sticks spool valves and blocks pilot lines. Most hydraulic faults start with dirt, and a particle count is the cheapest way to see it coming. If your press or power pack keeps throwing odd faults, start with the oil before you start on the valves. Our guides to hydraulic pump noise and cavitation and hydraulic oil running hot pick up from there.

A bad sample is worse than none

A sample from the wrong place tells you the wrong story. And people act on it.

So take it the same way every time:

• with the machine at working temperature, or straight after it stops
• from the same point, ideally a fitted sample valve
• from the middle of the oil, not the sludge at the bottom of the sump
• after flushing a little oil through the valve or tube first
• into a clean bottle from the lab, labelled before you start

Never take the sample from the drain plug after a stop. The heavy wear settles there, and it'll read worse than it is. Never top up just before sampling either.

On hydraulics, sample from the return line before the filter, with the system running. That's where the dirt is heading. Fit a proper sample point with a minimess or similar fitting. Don't crack a fitting to catch the drips.

That last point matters for safety. HSE warns that hydraulic injection has happened at pressures over 100 bar*, and "may occur at pressures as low as 7 bar"* (HSE safety alert). Its list of operator errors includes "searching for leaks with bare hands" and failing to release pressure before maintenance (HSE, fluid injection hazards). A sample point made for the job takes the guesswork out of it.

Which machines get sampled

Don't sample everything. Start with the boxes and power packs that stop the site when they fail, and the ones with the longest lead time on a spare. If you haven't ranked your plant, our post on criticality ranking is the place to start.

A small gearmotor you can swap in an hour from stock isn't worth a lab fee. A big gearbox on the main drive, a press power pack or a big screw compressor usually is.

Set the sample interval with the lab and the maker's manual. Busy, hot or dirty kit gets sampled more often. The PUWER code of practice makes the same point about maintenance in general. How often you maintain should reflect how hard the kit works, where it works, and the risk if it fails (L22, paragraph 72).

Read the trend, then act on it

One report on its own tells you little. The first sample is your baseline. The second shows the direction. By the third you know what normal looks like for that box.

The lab will flag results as normal, watch or act. Treat "watch" as a job, not a note. Shorten the interval, check the breather and the seals, and look at the temperature. Our guide to a gearbox running hot covers what to check before you condemn it.

Here's our view. Every report needs a named person who reads it the week it lands. Too many sites pay for sampling, file the PDFs, and only open them after the failure. That's paying for a warning and ignoring it.

Lubrication is a safety matter as well as a reliability one. The code of practice notes that "failure to lubricate bearings or replace clogged filters might lead to danger because of seized parts or overheating" (L22, paragraph 70). Where a failure could create a risk, it calls for "a formal system of planned preventative or condition-based maintenance" (L22, paragraph 75). Oil analysis is one of the simplest ways to do the second.

When the report says change it

Here's where the money is made or lost.

A report that says a box is wearing gives you weeks, sometimes months. That's time to order the gearbox, book the crane, plan the stop for a weekend, and get the right people in. The job might take a shift.

Ignore it, and the same job happens on a Tuesday afternoon with the line stood. You're paying for the downtime, the scrap, the express freight on the part, and the overtime to catch up. Add up what an hour of downtime really costs you and the planned job looks cheap.

The other trap is people. A big gearbox change needs a fitter who's done one, with rigging and alignment skills. If your team is two fitters on a four-on, four-off, the planned job gets pushed back every time something else breaks. Book the extra hands when the report lands, not when the box seizes. A planned change with the right engineer beats a breakdown with whoever's on shift.

If the box does go, our post on gearbox failure on a critical line covers the first hour.

What to do on Monday

1. List your five most critical gearboxes and hydraulic power packs, with the lead time on each spare.
2. Fit proper sample points to the ones that don't have them.
3. Take a baseline sample from each and agree an interval with the lab.
4. Name one person who reads every report the week it lands and raises a job for every "watch" or "act".

* Hydraulic injection pressures correct as of 11 October 2026. Check HSE's hydraulic injection safety alert for the latest. Paragraph numbers refer to L22, fourth edition (2014, amended 2018).