Most organisations clean the outside of their numerically controlled machine tools, but they cannot say how it was done last month, who decided the method, or what the next shift is supposed to reproduce. The work exists; the process around it does not. Building that process is not a paperwork exercise. It is a short sequence of decisions - what is on the machine, what has to come off, how the surface list is grouped, who owns each group, and what closes the job - written down once and then run the same way every time. The steps below are the smallest version that a plant can actually keep using.
What the business is actually trying to repeat
Before the procedure is written, the organisation has to decide what outcome the procedure exists to produce. In a shop with several numerically controlled machine tools, exterior cleaning is repeated because the coating and the surrounding area collect machining soil continuously, and because the state of the machine is part of how the shop presents itself, how it maintains it, and how safely it is operated. Three reasons tend to dominate, and each one sets a different standard for the same surface list.
Protection of the asset is the first. Coolant mist, cutting fluid residue and airborne oil drift do not stay in the work zone; they settle on the enclosure, creep down the frame and collect at ledges, seams and the conveyor housing. Left in place they hold abrasive chips against painted and bare metal surfaces, and they travel into the areas that later have to be opened for maintenance.
Serviceability and safety is the second. Grime on the control housing, the operator switches, the safety switch, the worklight and the emergency stop is an operating problem before it is an appearance problem. A label, an engraving or a switch marking that a film of oil has covered cannot be read reliably, and a handle that is slippery is a hazard that nobody records.
Shop discipline is the third, and it is the reason the process has to be repeatable rather than excellent once. A machine that is cleaned to the same condition each week tells the crew whether the last week was normal, and an inconsistent result destroys that signal.
Write those reasons once, at the head of the standard, because they decide which lines on the surface list are mandatory and which are optional. A business that never states its reasons ends up arguing about the same panel every month.
Start with a written surface list, not a cleaning task
A process cannot be built on the words "clean the machine", because two people read that instruction differently. It is built on a list of surfaces, written once per machine family and kept where the crew can see it. Each line of the list carries three columns: the deposit usually found on that surface, the reason it is worth removing, and how the surface is reached.
For the exterior of a numerically controlled machine tool, the list normally runs in the same order every time. Castors and floor tracks. Way covers and the bellows or telescopic guards. The enclosure panels and the frame below them. The control housing, its screen surround and the operator switches. The chip conveyor lid and the conveyor housing. The coolant tank lid, the sump edges and the pump cover. The worklight housing, the safety switch and the pneumatic fitting. The painted cabinet that holds the drive electronics, which is handled last and separately.
Writing the list is what makes the process repeatable across people and across shifts. Without it, every operator is rebuilding the route from memory, and the only quality check is whether somebody noticed a missed panel afterwards.
- Write one list per machine family, not one per machine. Two similar machines with two different lists are the beginning of an unrepeatable process.
- Add a column for anything on that machine that must not be cleaned by this route. That column is as important as the list itself, and it is written by the maintenance lead, not the cleaner.
- Keep the list short enough to be read in one minute on the shop floor.
Name the deposit before choosing the method
Repeatable cleaning fails when the method is chosen by habit instead of by the soil. On machine exteriors there are a few recurring deposit types, and each one answers a different question.
Loose, dry chips and dust sit on ledges, in the folds of way covers and around the conveyor lid. They are a reachability and containment problem, not a chemistry problem, and they have to come off before anything damp touches the surface.
Coolant mist and oil film coat the enclosure, the frame below the work zone and the conveyor housing. This is the deposit that turns a grey machine black, and it is the one that a heated cleaning medium changes the most, because warmth and water soften a rolled-on oil film so that the following wipe is shorter and is repeated fewer times.
Coolant residue and dried solution sit around the tank lid and sump edges, where evaporation leaves a crust. These are usually the heaviest lines on the list and the ones an operator is most likely to postpone.
Skin oil, floor dust and grime collect on handles, switch surrounds and the push buttons, where the touch pattern, not the machining, creates the soil.
Once the deposit is named on the line, the method follows from it: dry removal first, then a heated low-moisture application on the film, then a wipe and a dry-down, with the sensitive assemblies handled separately or excluded. A process that names deposits is transferable to a new employee; a process that names products is not.
Where the CNC-specific exclusions belong
A numerically controlled machine tool is not a single open surface, and a repeatable process for its exterior has to say plainly where the process stops. The exterior of a machining centre ends at the enclosure, the guards, the panels, the control housing and the outer surfaces of the conveyor and coolant systems. The work zone itself - the tables, the ways, the tool changer, the spindle area - is governed by the machine builder's own maintenance instructions, and a cleaning route that wanders into it has left the scope of this process.
That boundary is why the exclusion column is written by the maintenance lead. On one machine family the drive cabinet and its cooling may be excluded entirely; on another they may be cleaned by a dry method with the supply isolated. Both decisions are legitimate; the only unacceptable answer is an unwritten one.
The same discipline applies to what the plant assumes about equipment. Published catalogue pages describe the settings each machine is built for, and those settings are not portable from one plant to another. Where a business wants a route that is fixed to a location, contained, and matched to its own services and floor layout rather than carried between machines, that is built with the supplier as customized steam cleaning equipment, and the machine shop's surface list and exclusion column become the starting point of that conversation.
Group the list into routes the plant can actually run
A list of surfaces is not a schedule. To make the work repeatable, group the lines into routes that share an access condition and a shutdown requirement.
The upper route is the enclosure, the control housing, the panels and the handles. It can usually run while the machine is idle and does not need the utility area touched.
The lower route is the frame, the conveyor housing, the tank lid, the sump edges and the castors. It needs the floor clear, the run-off controlled, and usually a longer window.
The separate route is the electrical cabinet, the drive cooling, the worklight, the safety switch and any connector. It either has its own dry method or it is written onto the exclusion list, and that decision is recorded rather than remembered. Grouping this way is what makes the work schedulable, because each route carries its own duration, access window and owner.
Put utilities and consumables into the cycle time, not in a footnote
A process that ignores the plant services it needs is never repeatable, because the crew improvises around them. Before any trial, write down what is available at the machine.
Electrical supply at the location decides whether equipment runs at all, and the published product pages list different supply arrangements for their own settings, so this is checked against the plant's own socket, not against a brochure.
Water availability and refill distance decide how many walking minutes are added per machine. A cleaning method that halves scrubbing time and adds a refill walk every machine has not been improved; it has been moved.
Drain and run-off arrangements decide whether the lower route can be done at all. Where the floor-side method cannot hold the run-off, that line moves to a contained bay or it is written out of the standard, and the choice is stated rather than discovered later.
Consumables and the ability to replenish them mid-job are the most common reason a documented process quietly stops being followed. Count them in the cycle time alongside the cleaning itself.
Assign named owners, not job titles
Repeatable processes have owners you can name. For each route decide four things, on paper.
- Who decides that a machine is ready to be cleaned, and what "ready" means on the shop floor: idle, cooled, guards closed, run-off prepared.
- Who performs each route, and whether the same person can do both the upper and the lower route in one window.
- Who checks the result, and whether that check is a second person or a photograph reviewed later by somebody who did not do the work.
- Who closes the job - who writes down that the machine was returned to service, and who is told when it was not.
A cleaning routine with no named closer produces silent exceptions, and silent exceptions are how a documented process turns back into a habit.
Define what "finished" means before the trial
Most in-plant cleaning attempts are abandoned because nobody agreed in advance what success looked like, so the result is argued about instead of measured. Write a closing condition per route.
- A visible standard for each surface group: free of loose chips, free of rolled-on film at the touch points, no standing water left at joints, ledges or the conveyor lid.
- A time standard for the route, measured in the plant rather than quoted, and written down as the figure to beat.
- A stop condition: if the route needs an unplanned repeat on the same surface, or if run-off appears where it was not planned, the trial stops and the standard is revised.
- A second reviewer who grades the outcome from the photographs and a walk-around, not from the memory of the person holding the tool.
Run the trial on one machine of the family, then repeat it on a second machine with a different operator on a different shift. A routine that only works with one confident operator is a training gap, not a process.
Write one page per family and set the review date
The output of the whole exercise is a single page per machine family, and it should be short enough that it is used instead of filed. It carries the surface list with its deposit column, the exclusion column, the route grouping, the named owners, the closing condition and the measured cycle time.
Train against that page in place, on the machine, with the person who will be doing the work. A new employee who can follow the page and produce the same closing condition has inherited the process; one who needs to be shown a trick has not.
Put a review date on the page. Machine exteriors change - guards are replaced, controls are retrofitted, a conveyor is swapped - and a standard that nobody re-reads drifts away from the asset it describes. Comparing cycle-to-cycle results is what tells you whether the drift has already started.
Where a plant is standardising on heated cleaning for the film-heavy lines, the equipment options are grouped on the machine tool steam cleaning page, and what each published machine is set up for is read from that page rather than assumed for an in-plant route.
Where a repeatable process breaks down
Three failure patterns are worth naming, because they are the ones that kill the process after a quarter.
- The standard describes a cleaning product rather than a closing condition. Operators then optimize the product use, not the result.
- The exclusion column is empty because nobody owned it. The first damaged connector ends the credibility of the whole page.
- The cycle time is assumed rather than measured, so nobody can tell whether the process is getting slower.
None of these are technical problems. They are the price of writing a process without an owner, a closing condition, or a measured baseline, and all three are fixed on paper before a single machine is cleaned.
