Setting Service Intervals From What You Observe

It is not wrong, and it is the right place to start. But it was set for a notional average installation, and the two most common outcomes of following it literally in the Gulf are servicing something that did not need it yet, or discovering something needed it a month ago. Both are avoidable, and replacing the manual with your own figure is a four-step exercise rather than an engineering project.

Step one: decide which assets deserve an interval at all

Before tuning anything, sort the estate into three groups, because only one of them is worth this effort.

Most estates have far fewer assets in the middle group than the maintenance schedule implies, and a large first pass of savings comes simply from moving things into the third. Fire systems and lifts are never candidates for tuning, and it is worth saying so explicitly so nobody optimises the wrong thing.

Step two: record condition at service, not just the date

This is the step that makes the rest possible, and it is the one usually missing. When a technician services an asset, the record has to capture how much life was left in what they replaced — not merely that they replaced it.

It does not need instrumentation. A three-point judgement recorded consistently is enough: the part was still serviceable, it was due, or it was overdue and had begun to affect performance. A technician can make that call in seconds and it is far more useful than a pass-or-fail tick. What matters is that it is recorded against the asset, with the date, so a pattern accumulates — which is a capture-design problem more than a maintenance one.

Step three: move the interval toward the observations

After a few cycles the record answers the question directly. If parts are consistently still serviceable at replacement, the interval is too short and you are buying components and downtime for nothing. If they are consistently overdue, it is too long and you are running degraded between visits.

Adjust in modest increments rather than halving or doubling, and change one interval at a time so you can tell what caused what. And adjust per location, not per asset type: two identical air handlers, one beside a road and one on a sheltered roof, will not foul at the same rate, and averaging them gives you an interval that is wrong in both places.

Step four: let the season set the shape

A flat annual interval is a compromise between conditions that are not flat. Cooling load is concentrated, dust arrives in seasons, and humidity varies through the year — so the right shape is often uneven: more frequent attention through the demanding months, less through the quiet ones, with the heavy work deliberately placed where plant can be taken offline.

This is more work to plan and it is where most of the benefit sits, because it aligns effort with when degradation actually happens rather than spreading it evenly across a year that is not itself even.

Two things that go wrong with this method

Lengthening intervals to save money rather than because of evidence. The method produces both directions, and a programme that only ever extends is not condition-based — it is cost-cutting wearing the language of it. If no interval ever shortens, the observations are not being read.

Losing the warranty argument. Deviating from a manufacturer's schedule can matter while equipment is under warranty or covered by a service agreement. Check before tuning, and where the schedule must be followed, follow it and record why — a documented reason is what stops the next person undoing the decision.