Condition-Based & Predictive Maintenance

This domain measures whether equipment condition actually informs the maintenance decision — or whether the calendar decides everything.

Everything in it hangs on one concept, the single most useful idea in maintenance planning. Every failure has a point where it first becomes detectable — call that P, for potential failure — and a point where it becomes functional failure, F. The distance between them is the P-F interval, and it isn’t fixed. Depending on the failure mode it might be six months or six hours. A loosening connection heats up over months. Other failure modes give almost no warning at all. A calendar doesn’t know any of that; it inspects everything at the same rhythm regardless of what is actually degrading. Condition-based maintenance is the decision to find out where each asset sits between P and F, and let that set the schedule.

The standard is built this way now. Chapter 9 of NFPA 70B (2026) ties the maintenance interval directly to assessed condition — Condition 1, 2, or 3 — and Chapter 7 covers the test methods that feed the assessment: thermography, insulation resistance, ultrasonics, partial discharge. The 2026 edition goes further: it explicitly recognizes continuous monitoring — permanently installed infrared, IoT hot-spot sensors, remote diagnostics — as valid condition inputs, and it expects you to use the data you collect to adjust intervals. The standard isn’t tolerating condition-based maintenance anymore. It’s steering you into it.

Six criteria, each scored and weighted on its own — and this is the domain where honest scores drop:

1. Condition assessment is performed and scored. An actual assessment with an assigned condition level, not a walk-by with a flashlight.
2. Criticality is assigned to assets, in writing. Which equipment can hurt someone or stop the plant — ranked, not assumed.
3. Condition and criticality influence the maintenance interval. This is the hinge criterion. If every asset is on the same schedule regardless of its condition score, the assessment is decoration.
4. Predictive technologies are applied where the failure mode warrants. Where — not everywhere. Matching the technology to the failure mode is the skill; buying sensors is just shopping.
5. Condition trends are retained and reviewed. One reading is a data point. Three readings over time is a trajectory, and the trajectory tells you where you are on the P-F curve.
6. Degradation triggers action before failure. If a worsening trend has never once moved a work order earlier, the program is watching the failure happen in high resolution.

What good looks like fits in two sentences. Condition-based maintenance isn’t a fancier calendar. It’s the decision to stop guessing where you are on the P-F curve and start measuring it. In practice, critical assets carry condition scores, the scores move the intervals, and a degrading trend generates work before the failure generates downtime. The payoff runs both directions: healthy Condition 1 equipment earns longer intervals. Condition-based programs usually spend less than calendar programs — they just spend it in the right places.

What this domain is worth

The cost of the gap: a calendar program spends on healthy gear and still misses the gear that is actually degrading — you pay twice. The six-figure failure that “looked fine at the last inspection.” And capital spent early, replacing equipment nobody could prove was done, because nobody knew its condition. The return: spend less and spend it right — condition-based programs usually cost less than calendar programs. Failures become scheduled work; outages become shutdowns you planned. And equipment life gets extended, which defers replacement capital. That is exactly what the Circuit Yield ROI model in REALTIMEais puts a number on: net present value, payback, and benefit-cost ratio over five years, with your inputs — lost production per hour, outage hours, shutdown labor, contractor cost, replacement capital. Condition-based programs usually spend less than calendar programs. They just spend it in the right places.

Here is what the calendar really is: a bet. Calendar-based maintenance is a bet that your inspection frequency happens to be shorter than the P-F interval of whatever is currently degrading. When you lose, the postmortem always reads the same way: “it looked fine at the last inspection.” It did. You just looked before P. A plant I worked with lost a main breaker to a connection failure — unplanned outage, six figures before lunch. The inspection records were flawless: annual, on time, “no exceptions noted,” every word true. The failure mode had a P-F interval of maybe four months. The inspection interval was twelve. That connection could degrade from first-detectable to failed three times over between visits. The schedule guaranteed they’d miss it, and they had been unknowingly winning that bet for years until the day they didn’t. Nobody was negligent. The math was against them, and nobody had ever done the math. That’s this domain in one line: it scores whether you’re doing the math.

Three questions to ask your program this month:

• Has anything on our critical list been assessed for condition — or only inspected? Different verbs, and the standard now knows the difference.
• Do we know which of our assets are actually critical, in writing?
• Has a condition finding ever changed one of our intervals? If the answer is no, the data isn’t driving anything yet.

Send one to your maintenance lead today. The full REALTIMEais 70B READINESS SCORE assessment is free at REALTIMEais.com/readiness, scored in REALTIMEais, The Electrical Intelligence System™ — where sensor data lands in the asset record and moves the interval.

LinkedIn teaser

“It looked fine at the last inspection.”

It did. You just looked before P.

Every failure has a point where it first becomes detectable (P) and a point where it fails (F). The gap between them is the P-F interval — six months for one failure mode, six hours for another. A calendar doesn’t know the difference.

A plant lost a main breaker to a connection failure. Six figures before lunch. The records were flawless: annual, on time, no exceptions. The failure mode had a P-F interval of about four months. The inspection interval was twelve. The schedule guaranteed they’d miss it.

Nobody was negligent. Nobody had done the math.

NFPA 70B 2026 now sets intervals by assessed condition and recognizes continuous monitoring as a valid input. It isn’t tolerating condition-based maintenance. It’s steering you into it.

The money: a calendar program pays to open healthy gear and still misses the gear that’s degrading. Condition-based programs usually spend less — they spend it where the degradation is — and they turn outages into shutdowns you planned. Circuit Yield in REALTIMEais puts the NPV and payback on it with your numbers.

Part 6 of the REALTIMEais 70B READINESS SCORE series: Condition-Based & Predictive Maintenance — 10% of your score.

One question for today: has a condition finding ever changed one of your intervals? Ever?

Free assessment: REALTIMEais.com/readiness — The Electrical Intelligence System™

#NFPA70B #PredictiveMaintenance #ConditionMonitoring #Reliability #REALTIMEais