Why this process drifts more than other cleaning tasks
A cleaning process that lives in an experienced operator's head works well right up to the day that operator is on leave. Machining fluid buildup is a harder target than most shop cleaning because the deposit itself changes: the fluid type, the time it has been sitting, and the surface it landed on all decide what the deposit is and what will remove it. A documented process makes results repeatable across shifts by writing down the three things most shops leave unwritten - how deposits are observed, which method each observed state triggers, and the settings that actually worked on your own machines.
Three causes of drift are worth naming before you write anything:
- The deposit changes between runs. A water-diluted cutting fluid loses water and leaves a dry additive film that re-softens with heat. Oil and grease oxidise slowly into a tougher film. Fluid that carries fines stacks into a hard crust layer by layer. The same machine cleaned in its first week and cleaned two months later is not the same job.
- "Clean" is never defined. One shift wipes until the surface looks fine; another wipes until a cloth comes away clean. Both consider themselves finished, and the difference only appears at the next audit.
- Settings are guessed. Distance, dwell and number of passes vary from person to person, so results vary too - and nobody can say which variable caused the difference.
A written process removes all three causes by documenting observation, decision rules and the recipe. Everything below belongs in one short document per machine type, not in a manual nobody reads.
Build the document around a fixed list of hotspots
Do not start with the whole machine. Walk each machine type and mark where fluid actually lands and collects: drip points under seams and guard edges, the area beneath way covers, the chip conveyor and its discharge end, coolant tank lids and their edges, splash panels and wiper blades, cable entries and nameplates, and the floor strip between the machine and the chip cart.
Photograph each hotspot from the same position and height every time - fixed reference points. Keep the list identical from one cleaning to the next; a fixed list is the only way to tell whether deposits are growing or shrinking. Each hotspot then gets two written lines: what an acceptable surface looks like at that spot, and what finding triggers action. This list is the skeleton of the process document, and it is also what makes machining line oil cleaning auditable instead of a matter of opinion.
Write the observation step as precisely as the cleaning step
Repeatability usually fails at classification before it fails at cleaning. If two people can call the same deposit different things, every later step drifts. So the document should describe the observation itself:
- Shine a work light at a low angle across the surface and note where a sheen or a matte patch appears.
- Wipe an agreed white cloth over an agreed area and look at what it picks up.
- Scratch a plastic scraper through the deposit and rub a small amount between gloved fingers to check for embedded fines.
Then define the three states in observable words, not adjectives. State one: a wet or fresh film - the cloth lifts it easily and the scraper finds no hardness. State two: a dry or polymerised film - matte or varnish-like, the scraper lifts flakes, and heat with dwell time is needed. State three: a crust with embedded solids - a hard layer that must be removed mechanically before any heat is worth applying. Note beside each hotspot which fluid is in use there, water-diluted or oil-based, because the two leave different residues, and the note tells the operator which description to expect.
Write the procedure as decision rules, not one path
A single linear procedure forces people to improvise the moment the deposit differs from what the writer saw. Decision rules do the opposite: they tell the operator what to do when the observation differs.
- If state one: wipe while still wet with warm detergent and rinse the cloth often. No special equipment is needed, and this is a task for the end of the shift.
- If state two: heat and dwell do the work. Apply a heated low-moisture method, let it sit, wipe, and repeat short cycles rather than one long attempt. Repeated short cycles beat one long soak because the film releases in stages.
- If state three: mechanically remove most of the crust first - scrape, brush, pick - then treat the remaining film as state two. Heat applied on top of a thick crust mostly heats the crust.
Where a steam or hot-water machine is used, the document states which surfaces take the drier mode - near electrical parts, labels, cable entries - and which take the wetter mode - stubborn films on open panels - and where the run-off is expected to go. It also states plainly what the method does not do: it does not reach behind panels or inside a coolant tank, it does not remove a crust with embedded fines without the mechanical step, and it does not fix the condition of the coolant system itself. A written cutting fluid residue removal process is a surface process, and saying so keeps expectations honest.
Capture the recipe from your own trial
The settings that make results repeatable - nozzle distance, movement speed, dwell, number of passes, mode selection - come from a trial on your own equipment, not from a catalogue. Run a half-panel test: treat one half of a panel and leave the other half untouched as a reference, change one variable at a time, and write down the minutes each pass takes and what visibly changed. When one combination clears the agreed hotspot reliably, that combination is the recipe, and it goes into the document as a single line: mode, distance, dwell, passes, and who is allowed to change them and where a change is recorded.
The machine's data sheet supports the recipe but does not replace it. Reading a PowerDe HW02 fixed steam cleaning machine sheet as a worked example: 160 °C and 0.8 MPa are two readings of the same saturated-steam state - they describe softening capability, not completion speed, so dwell time still belongs in your recipe; 6 kW heating power and the 80 L stainless tank set the work cycle and the refill rhythm; the 220 V supply needs a verified circuit from your own electrician, and the listed protections - leakage, short-circuit, high temperature, high voltage and water shortage - protect the machine, not your process or your schedule. The HW02 product page lists vehicle exterior cleaning, interior vacuuming and disinfection, and engine-compartment cleaning, so machine-tool use should be confirmed with the manufacturer in writing before it enters a published process. Ask for the items the page does not publish: hose and lance lengths, warm-up time, run time per tank, and whether the figures are measured at the boiler or at the nozzle.
Name what stays dry inside the steps
Every machine has parts that must stay dry and parts that can be cleaned directly, and the document should name both per machine type. Guideways and ball screws, spindle tapers, electrical cabinet doors and cable entries, hanging operator buttons and printed labels normally need protection or a dry cloth only - and where protection is removed for cleaning, re-oiling the guideways per the maintenance schedule belongs in the same step. Painted sheet metal, cast housings, guards and a chip conveyor with a defined drainage path can normally be cleaned directly. Writing the protective action into the step that precedes it - "mask the cable entry, then start from the top panel" - is what stops protection from being skipped on a busy night.
Verify with checks that everyone reads the same way
Completion criteria must be observable and identical for every person who checks:
- Angled light shows no remaining sheen on the agreed panel areas.
- The agreed white cloth wipes within an agreed limit on the listed hotspots.
- Bare metal passes a water-break test - water sheets across the surface instead of beading.
- Flash rust is checked several hours later on bare steel, and rust-preventive oil is reapplied where a downstream process needs it.
Record one line per machine per cleaning: date, machine, hotspot, observed state, method used, minutes taken, findings, and initials. Compare records between shifts, not only within one shift - when the same finding appears in two consecutive cleanings, the document needs revision, not the operator. Keep the boundaries in the same section: coolant concentration and filter condition, cleanliness grades required by downstream processes, and hygiene validation are separate programs, and a cleaning record supports them without claiming them.
Prove it with a second person before publishing
Write version 0.1 from your trial on one machine. Then hand it to a second operator, without coaching, on another machine of the same type. Every place they stop to ask, guess, or quietly do something different is a defect in the document - fix the wording, not the person. When a second person completes the job to the written criteria unaided, publish version 1.0 with a version number, a date, and a training record for everyone who will run it.
Define revision triggers in the document itself: a new fluid or fluid supplier, a new machine or a changed guard layout, the same finding recurring in the records, and a seasonal change in shop temperature that alters how fast films dry. A process document without revision triggers is a photograph of last year's method.
When you send the document to an equipment supplier, send the records rather than a summary: the hotspot list, the deposit descriptions, the recipe line and the trial timings. That is what lets them confirm in writing which parts of your process their machine supports and which need a different solution. You can start that conversation through PowerDe's contact page.
A repeatable process is one a stranger can follow and produce the same result. Start with one machine, one hotspot list, and one half-panel trial - the document writes itself once the observations and decisions are on paper.
