A plant that says it has a heavy oil buildup problem is describing several different problems wearing one name. On one machine the buildup is a leak that has been running for a quarter. On the next machine it is mist from a blow-off station that has settled on everything within reach. On a third it is months of hands and rags leaving a film on every handle, button and panel operator. Each has a different source, returns at a different rate, and asks for a different first action. An assessment that treats them as one job produces a plant-wide cleaning project that never quite finishes. An assessment that separates them produces a short list, in order, that a small crew can actually complete.
The difficulty is that buildup is usually assessed by looking at it, and looking rewards the largest, darkest surface while the consequential problems are smaller and hidden: the sight glass nobody can read any more, the nameplate an auditor photographs, the new leak that cannot be traced through an old film. Severity is not how much oil there is. Severity is what the oil is preventing.
The sequence below starts with the sources, because a source predicts where the buildup will be next month as well as where it is today. It moves through what to record and how to tell an accumulation from an active problem, sorts the findings into classes and an order, and ends at the point where the record becomes an equipment question. It assumes no machine and quotes no specification, and it deliberately stops before any purchase.
Start from where the oil comes from, not from where it shows up
Every film on a machine surface arrived from somewhere, and each kind of source leaves a signature. Learn the signatures and the walk stops being a survey and becomes a confirmation.
Leaks and seepage build downward. The deposit sits below the origin and along the path gravity chose: drip edges, fastener heads, the base frame, the floor under a joint, the shelf beneath a breather vent. A pattern that thickens as it goes down is a leak until it is proven otherwise, and the oldest deposit sits directly under the seal or fitting that started it. Clear that patch, look up along the wet path, and the origin is usually visible.
Mist and aerosol build upward and outward. They settle on top surfaces, on the warm side of enclosures, on labels, on the tops of panels, and on every machine within reach of the station that makes the mist. A thin, even film across many machines is almost never many problems; it is one source that the machines are catching rather than producing. Cleaning twenty machines while the source runs buys a clean week and nothing after it.
Splash and workpiece transfer build around openings. The deposit is heaviest where material enters and leaves: load and unload zones, the area around dies and chucks, the mouth and discharge of a conveyor, the table where parts are presented by hand. It thins with distance from the working zone, and it carries whatever the workpiece carries, including abrasive fines.
Handling builds on the surfaces people touch. Handles, buttons, control panels, trolleys, railings, the staging area where parts wait. Oil on a control panel almost never came from the process; it came from hands, and it sits on the one surface everyone reads and nobody cleans. It is also the easiest finding in the plant to fix, which is why leaving it in the cleaning plan rather than in the way people work is a decision, not an oversight.
So the first pass of the assessment asks one question per finding: what put this here, and is that thing still running? The answer goes on the record line. A finding with no named source gets re-examined at the next walk, because it will be back.
Measure severity by what the buildup is hiding
Appearance ranks surfaces by size and darkness, and that ranking is wrong more often than not, because the cost of a buildup lies in what it prevents. Four questions per asset will do it, and each can be answered in minutes.
Can the required checks still be done? Sight glasses and oil levels, the inspection point an inspector is supposed to look at, the joint or weld where a leak would first show. An oily machine cannot be leak-hunted, because a new leak has no background to show against; the film that looks merely untidy is the reason the plant finds leaks late. This is the most expensive thing oil hides, and it never appears on a cosmetic complaint.
Can the required accesses be done? Latches, drain plugs, dipsticks, guard fasteners, the panel that has to be opened for the next service. A film turns an open-and-close task into a wipe-first task, and tasks that need a rag get postponed by the people who have to hold the handles afterwards.
Is it affecting function? Oil acts as insulation on finned surfaces and motor housings, so a coated fan cools worse than a clean one. Grit suspended in oil works into seals, guides and ways. A film on the floor under and around a machine is a slip question, not a tidiness question. And oil transferred from a handle to a control, or from a surface to a product or its packaging, is a contamination path with two ends worth naming.
Is it hiding an active source? The test is time. Clean a small patch, mark its edges, and look at it daily for a week. A film that returns within days is not history; it is current, and that asset belongs in a conversation about its source rather than in the cleaning plan. This is the single most useful measurement an assessment can make, and it costs one cloth and a week of glances.
Record the answers as what is hidden or prevented, in plain phrases: level unreadable, nameplate obscured, drain plug needs a rag, film returns in days. Those phrases rank themselves. Very dirty does not.
Record it in a form that survives comparison
An assessment is only useful if the next one can be compared with it, and comparison means the record has to be re-runnable: the same asset identifiers, the same marked points, the same conditions, the next quarter as well as today.
Use the asset numbers from the maintenance system, not room names or the press by the door. A buildup finding that lands on the asset's maintenance history gets acted on with everything else that history carries; a finding that lands in a cleaning folder does not. The two records should speak the same language, because the sources the assessment names are the same sources the maintenance system already tracks.
Write one line per surface, not one per machine. A line names the surface, the state of the deposit, the source you believe put it there, and what it is hiding. The states are few and can be told apart in a minute with a torch, a cloth and a plastic scraper: a fresh film is glossy and transfers fully onto a cloth with no resistance; a polymerised film is dark or dull, resists the cloth, and only breaks at the edge of a blade; a deposit bound with grit feels abrasive between the fingers and will not move until the solid is removed dry. The state decides the work, so it belongs on the line.
Mark reference points and photograph them from the same position and angle at every assessment, so the comparison does not depend on anyone's memory. Use raking light rather than overhead light, because a film is invisible under a ceiling lamp and obvious at a low angle. Agree on one white cloth and use it the same way on every asset. At drip edges, lift the deposit with the scraper and record its thickness honestly, because the difference between a surface film and a centimetre of bound sludge changes the class the finding belongs to.
Walk at the end of a shift, when surfaces are warm and the oil is soft and visible; the same machine at shift start, cold and congealed, reads differently for reasons that have nothing to do with the cleaning. Fix the time of day so the next walk is comparable, and note when each surface was last cleaned if the plant knows, because without that date the return rate is guesswork.
Sort the findings into treatment classes
The point of the record is not a to-do list. It is a sorting. Findings that look nothing alike belong in the same class, and findings that look alike belong in different ones, because what decides the class is the state of the deposit, what the work requires around it, and whether the source is still running.
The first class is short, repeatable work inside windows the schedule already gives: fresh films on reachable surfaces, small areas, nothing that needs protection or re-lubrication afterwards. These are industrial equipment oil removal tasks measured in minutes per face, and their value is that they keep the other classes from ever forming. They are also the class most often over-scoped: a task that needs masking, or that strips a protective oil film, does not belong here no matter how small it looks.
The second class is planned-stop work: polymerised films that need sustained heat and several short passes rather than one long one; anything near guides, ball screws or racks, where washing strips the film that was protecting them and re-oiling the same day is part of the job rather than a follow-up note; anything that needs masking, drying time, or a second person. This class belongs in downtime that is already on the plan, with the protection, drying and re-lubrication minutes counted inside the task, because a task estimated on its washing minutes alone will be cancelled the second time it overruns.
The third class is mechanical removal first. Thick deposits bound with grit do not wash off; heat moves them and water relocates the solids, which re-deposit wherever the run-off slows and cools. Scrape, bag what comes off, then treat what remains as first or second class. A plant that tries to skip this class with a stronger machine concludes that every method is unreliable, when the method was simply asked to do a step that had been left out.
The fourth class is not a cleaning task at all: the active leak, the blow-off station putting mist on a dozen machines, the hand-and-rag film on every control panel. These findings route to maintenance, to an engineering change, or to how the work is done, not to the cleaning plan. It is the class plants leave out, and it is the reason so many cleaning programmes look like failures: the work gets done, the film comes back, and the method gets blamed for a source that was never in scope.
A single machine will usually carry findings from several classes at once. That is normal, and it is why a task written as one line per machine, deep clean press two, never survives contact with a schedule. The classes are the unit of work; the machine is only where they happen to sit.
Rank by consequence and return, not by appearance
With the findings sorted, the order follows from two variables: what the buildup is preventing, and how fast it comes back. Appearance is neither of them, and a list ordered by appearance will spend the first crew on the most visible machine and leave the consequential one waiting.
First come the assets where a buildup blocks something the plant needs to be able to do: the unreadable level, the obscured label, the joint that cannot be leak-hunted, the floor that is slippery underfoot. These are rarely the worst-looking machines. They are the ones whose dirt is in the way of a check somebody is supposed to perform, and they are cheap to put right once the record has named them.
Second come the assets with active sources. Cleaning them before the source is addressed resets a clock rather than changing a rate; the film will be back within days, which is exactly when a plant concludes that cleaning does not hold. Addressing the source first turns the same cleaning task into a durable one, and it is the difference between a programme that improves and one that re-treads.
Third come the second-class deposits, before they become third-class ones, because the cost of the work rises with the age of the film, and a planned window is easier to obtain than an unplanned one. Last comes the cosmetic work, honestly labelled as such. There is nothing wrong with a plant wanting its machines presentable for a customer's visit; the mistake is letting that work take the first crew and the first window, so the hidden problems wait for another quarter.
Write the ranking down with the reason attached to each line. A ranked list without reasons gets re-argued by whoever walks past the dirtiest machine, and the re-argument, not the oil, is what ends up deciding the order.
When the record reaches the equipment conversation
An assessment done this way produces, before any equipment is discussed, a document most suppliers do not expect: an asset list with surface types, deposit states, named sources, what each buildup is hiding, the windows the plant can genuinely give, and the services at the points where the work would happen, meaning power, water, and where the run-off goes and who owns oily water. That last question belongs to the site and to the local drainage authority, and it is worth settling before anyone quotes.
That document changes the question you ask. Instead of what machine removes heavy grease, the question becomes what can do these surfaces, in these windows, by these operators, with this water and this drainage. Comparing heavy grease removal equipment against that question is a different exercise from comparing catalogues. The figures that matter describe a working cycle, how long the machine runs before it needs water or recovery, and how far its hose reaches from where it stands; the figures that decide the result on your surfaces are the ones catalogues usually do not carry.
Ask in writing for what a listing does not publish: the temperature and pressure at the end of the intended hose rather than at the boiler, the hose and wand lengths, the warm-up time, how long one tank of water supports continuous work, and the recovery time after a refill. Where a listing names its intended uses, put your surfaces to the manufacturer as a written question and keep the answer with the purchase record; a use list is a scope statement, not a suggestion, and numbers on a data sheet do not extend it.
And carry the boundaries the assessment already found into the same conversation. Surface cleaning does not open machines, does not replace lockout, does not fix a leak, and does not clean the inside of anything. The record should say what is out of scope as carefully as what is in, because a scope that only lists the inclusions will grow until it fails.
Run the first cycle small, and let it correct the method
The assessment is a hypothesis until one machine has been worked from it. Choose one asset from the first or second class, not the worst in the plant. Write the pass wording before anything is wet, in language a person can stand in front of: low-angle light shows no film across the tested area; the agreed cloth comes away clean from two marked panels; bare metal still looks bare when it is looked at again hours later. Mask half of one representative surface and treat only the other half, so the comparison happens on the same surface rather than between two opinions.
Time every phase, including cooling, protection, drying, re-oiling and the checks, because those minutes are what the ranking was estimated against, and a plan built on washing minutes alone will not survive its first week. Then check again the next morning, because machines answer late: flash rust appears on bare metal hours after the surface looked dry, water that migrated into a joint shows itself as a streak overnight, and a guide that was stripped of its protective film reports it on the next shift, in a machine that ran well the day before.
Correct the method with what that machine taught you. The classes will move. A source you guessed wrong will show itself. A surface you placed in the first class will turn out to need protection, and one you treated as urgent will turn out to be inert history. That is the assessment working, not failing. The output is still the same document, now with a week of records behind it: a ranked list, each line naming its source, each class with a window it fits into, and a first month of work small enough that a small crew finishes it.
A plant does not need to conquer its oil. It needs to know which buildup is hiding something, which one is being produced right now by something that is still running, and which one can honestly wait. Answer those three in that order, keep the record where the maintenance system can see it, and put the record rather than a summary in front of anyone you are comparing equipment with. The questions your own assessment left open can be put to the contact PowerDe team together with the surface list, the deposit states and the windows the plant can give, which is a shorter conversation than the one a wrong machine starts.
