Key points
- Routine maintenance is minutes of checking and care per asset, done often; preventive maintenance is the larger planned job on an interval, and the two feed each other.
- Build rounds, not one task per asset: one person, one frequency, one area, one walk in the order people actually move.
- Write checks that cannot be answered from the doorway: readings with limits, three-way choices, photos on findings, and an honest 'not checked' option.
- A round only earns its time if findings become work requests the same shift, and if someone watches for rounds ticked without being walked.
What counts as routine maintenance
Routine maintenance is the small, repetitive work that keeps a machine, a building system or a vehicle in its normal state: wiping down, draining condensate, topping up lubricant, reading pressures and temperatures, checking that guards are in place, listening for a bearing that has started to sing. It takes minutes per asset, rarely needs spare parts or isolation, and is repeated per shift, daily or weekly. Preventive maintenance is the bigger planned job on an interval, such as a filter change by running hours or a yearly belt replacement. The two are not competitors. Routine checks are what tell you whether the preventive intervals are right, and they catch the slow drifts (a pressure creeping up, a level dropping a little faster each week) that turn into breakdowns if nobody is looking.
| Routine maintenance | Preventive maintenance | |
|---|---|---|
| Typical frequency | Per shift, daily, weekly | Monthly, by hours or kilometres, yearly |
| Who does it | Operator or technician on a round | Technician or contractor |
| Time per asset | Minutes | Often hours |
| Parts and isolation | Rarely | Often |
| What it produces | Readings and findings | A completed job with parts and hours |
| When something is wrong | Raises a work request | Work is done inside the job or a follow-up work order is raised |
Build rounds, not one task per asset
The most common setup mistake is to create one daily routine task for every asset. A site with sixty assets then generates sixty small jobs every day, the list is never empty, and people stop opening it. Instead, group the checks into rounds. A round is one person, one frequency, one area and one walk. The operator on the packing line does the packing line round at the start of each shift; the facilities technician does the plant room round every morning; the workshop does the weekly vehicle walkaround on Monday. Each round is one thing to do, with one completion, and missing it is visible.
- One frequency per round. Mixing daily and weekly items on one sheet means the weekly ones are either done daily (wasted time) or skipped silently. Make a separate weekly round, even if it covers the same room.
- Order items by the walking path, not by asset number. If the round makes someone walk back across the hall for the last item, that item will be the one that gets ticked from memory.
- Keep it short enough to fit the real day. Time the round once by walking it yourself. If it does not fit in the time the person actually has at that point in the shift, split it rather than hope.
- Name an owner per shift, not a team. 'Production' does not walk rounds; the operator on line 2, early shift, does.
- State the machine condition for each check: running or stopped. A temperature taken on a cold machine and one taken after two hours of running are not comparable, and a mixed history is useless for trends.
Write checks that cannot be answered from the doorway
A check that reads 'Compressor OK? Yes/No' will be answered yes, often without going near the compressor. That is not bad faith; it is what a yes/no question invites when the answer was yes yesterday and the day before. The useful trick in routine maintenance is to design each item so that answering it requires looking. The table shows typical rewrites. The limits in it are only an example: take yours from the manufacturer's manual or from a few weeks of normal readings on your own equipment.
| Weak check | Better check | Why it works |
|---|---|---|
| Compressor OK? | Discharge pressure in bar (example limit: 6.5 to 7.5), machine running | A number has to be read off the gauge |
| Oil level OK? | Oil in sight glass: above the mark, between the marks, below the mark | Three choices break the automatic yes |
| Check for leaks | Oil or water on the floor under the unit? If yes, photo | Specific place to look, proof when found |
| Belts OK? | Belt guard fitted, and no squeal at start-up | Observable without opening the guard |
| Clean the filter if dirty | Filter differential pressure reading | The trend shows when to change it, instead of guessing |
Two more details matter more than they look. First, give every item an honest 'not checked' answer with a reason: machine stopped, area locked, no access during production. Without it, people tick OK on things they could not see, and your history quietly fills with false normals. A 'not checked' is useful data; a fake OK is worse than no record. Second, decide in advance which findings stop the machine and write that on the item itself, for example 'If below the minimum mark, stop the unit and call the shift supervisor'. An operator on night shift should not have to make that call alone.
What happens when a check finds something
A round is only worth the time it takes if its findings leave the checklist. The rule that makes this work is simple: every out-of-limit reading or abnormal answer becomes a work request before the round is closed, linked to the asset, with the reading and a photo. The person on the round does not diagnose or fix; they report. The supervisor or planner triages it the same shift and decides whether it is a work order now, a job for the next planned stop, or a 'watch'. A watch is for readings that are drifting but still inside the limit: instead of raising a job, add a temporary extra check on that item for a week or two, then decide.
How to tell a round was ticked, not walked
Every maintenance manager eventually meets the round that was completed perfectly every day while the machine ran out of oil. You will not catch it by asking people whether they did it. You catch it in the records, which is one more reason to collect readings and timestamps rather than ticks. These are the signals worth checking every month:
- Readings that never move. A gauge that reads exactly the same value for weeks on a machine whose load changes is usually being copied from yesterday, not read.
- Impossible completion times. A round you timed at a quarter of an hour, closed in two minutes, or all items stamped within the same minute.
- Batch completion. Several days of rounds closed together, or every round closed in the last minutes of the shift.
- No findings for months on old equipment that still breaks down. Ageing assets produce findings; a round that never finds anything is either on the wrong items or not being walked.
- A breakdown the round should have caught, with 'normal' recorded the day before. Low oil, a missing guard, a blocked filter: check the last few rounds whenever one of these appears in a breakdown.
When you find it, look at the design before the person. The usual causes are a round that is too long for the time available, items that are unclear, a round placed at the busiest moment of the shift, or findings that were never acted on. Fix those first. Then walk the round with the operator now and then, without warning and without making it an inspection of them: you will learn more about which items are unworkable in ten minutes on the floor than in a month of reports.
Keep rounds alive with a quarterly review
Rounds grow. Every incident adds an item, nobody removes one, and after a year the round takes twice as long and gets walked half as carefully. Once a quarter, sit down with the round owners and go through each round with its history in front of you:
- Breakdowns since the last review: could a check have seen any of them coming? If yes, add that one specific check.
- Items that always pass on non-critical points: move them to a weekly round or merge them. Safety items (guards, emergency stops, pressure relief) stay at their frequency regardless of history.
- Items that keep finding the same problem: that is not a round problem. It is a preventive maintenance interval that is too long or a root cause nobody has looked at.
- Actual round time against the target: if it has grown, trim or split before people start skipping.
- Assets added, moved or retired: update the round and its walking order.
Whether you run rounds on paper, in a spreadsheet or in a CMMS, the principles are the same: short rounds, real readings, findings that become work requests, and a regular look at the records. If you use Solmex, the Routine Maintenance module is where rounds like these live, alongside the work requests they raise.