Every power plant knows what maintenance cost. Every power plant knows what it generated. Very few can put those two numbers on the same page per unit, per week, and tell you which direction the ratio is moving. Maintenance cost per MWh is the one KPI that reveals whether your reliability programme is a cost centre or an efficiency lever — and the reason it's hard to see is that the data sits in three different systems that were never designed to talk. This guide walks the construction using OXMAINT AI, the AI-powered CMMS for power generation operations.
Power Generation · Maintenance KPIs · Cost per MWh
The Fleet Made 12 GWh Last Month. Maintenance Cost You What Per MWh?
OXMAINT AI, the AI-powered CMMS/maintenance management software, connects the full generation workflow on one platform — work orders and labour in, parts and contractor spend linked to the asset, generation output pulled from the historian, and cost per MWh calculated per unit, per week, per driver.
Labour + Parts + Contractor Roll-Up
Historian-Linked Generation
Per-Unit, Per-Week View
5 Inputs
the true cost per MWh needs — labour, parts, contractor, corrective, outage
3 Systems
the data lives in — CMMS, ERP, historian — before OXMAINT AI unifies it
4 Drivers
that move the ratio — reactive mix, deferred PM, contractor rate, outage overrun
$/MWh
the one number every plant manager can defend to the board
Why Cost Per MWh Is Hard to See
The reason most plants can't put this number on a screen isn't accounting — it's plumbing. Maintenance labour lives in the CMMS, parts spend lives in the ERP, contractor invoices land in AP, and generation output sits in the historian. Reconciling them by hand takes a controller a week. OXMAINT AI closes the loop so the number is live, not lagged. Sign up free and connect your first data source in OXMAINT AI.
CMMS
Labour Hours
Craft-hour bookings against WOs — usually the only complete data set
►
ERP
Parts Spend
Material issue against WO — reconciliation lag is where the ratio blurs
►
AP
Contractor $
Invoices posted after service — rarely allocated to the WO that caused them
►
HIS
Generation MWh
Historian data by unit and interval — the denominator no one links to the numerator
The 5 Cost Inputs — What the Numerator Must Include
A cost-per-MWh number that only counts labour is a comforting number, not a true one. Five inputs make the ratio defensible. OXMAINT AI ties each one back to the WO that generated it. Book a demo to see the 5-input cost roll-up live.
01
Direct Labour
Craft-hour bookings × loaded labour rate — CMMS-native, easiest to capture
02
Parts & Materials
Store issues plus direct-charge purchases linked to the WO
03
Contractor & Vendor
T&M and lump-sum invoices allocated to the WO — usually the biggest miss
04
Corrective Rework
Re-visits and warranty callbacks on prior WOs — the hidden multiplier
05
Outage-Attributable
Planned + forced outage work — usually the largest single monthly line
The MWh Denominator — Choose It Deliberately
"Cost per MWh" depends on which MWh you divide by. Net vs gross, produced vs sent-out, all-hours vs available-hours — each answers a different management question. OXMAINT AI lets you pick per view; most plants run all three side-by-side. Sign up free and configure your denominator basis in OXMAINT AI.
GROSS
Gross Generation
Total MWh at generator terminals — highest denominator, lowest ratio. Best for asset-efficiency framing.
NET
Net Sent-Out
Gross minus auxiliary load — what the plant actually delivered to the grid. Best for commercial framing.
AVAIL
Available-Hour MWh
Only counts hours the unit was declared available — reveals whether reliability spend is paying for actual availability.
The 4 Drivers That Move the Ratio
A rising cost-per-MWh trend is a diagnostic prompt, not a verdict. Four drivers explain almost every movement — knowing which one is pushing the number is what turns the KPI into an action. OXMAINT AI breaks the ratio down by driver on the same dashboard. Book a demo to see the driver breakdown live.
R
Reactive Mix Creep
Reactive % rising against planned — each unplanned WO costs 3–5× a planned equivalent
Watch: reactive/planned trend per month
D
Deferred PM Backlog
Missed PMs accumulate → forced outages later → cost spike months after the deferral
Watch: PM compliance % + deferred WO count
C
Contractor Rate Shift
Vendor rates re-priced or scope creep on T&M — invisible until the monthly reconciliation
Watch: contractor $ per WO trend
O
Outage Overrun
Planned outage extends → double hit: extra cost + lost MWh in the denominator
Watch: outage plan vs actual per event
One Number. Four Drivers. Zero Reconciliation Lag.
OXMAINT AI ties every dollar to the WO that generated it and every MWh to the unit that produced it — so cost per MWh becomes a live operating metric, not a monthly finance report.
A Worked Example — One Unit, One Month
Numbers land differently than words. Here's a worked-example scorecard for a single unit — how the five cost inputs and three denominator choices combine into three ratios that each answer a different question. OXMAINT AI generates this per unit, per week, automatically. Sign up free and generate your first scorecard.
Unit 3 — October Scorecard (Illustrative)
Cost Roll-Up (Numerator)
Direct labour$182,000
Parts & materials$94,000
Contractor & vendor$156,000
Corrective rework$28,000
Outage-attributable$240,000
Total$700,000
Generation (Denominator)
Gross generation145,000 MWh
Net sent-out137,500 MWh
Available-hour MWh120,000 MWh
Ratios
vs Gross$4.83/MWh
vs Net$5.09/MWh
vs Available-hour$5.83/MWh
Illustrative values only — actual numbers vary by fuel, technology and market.
The Trend View — What Actually Gets Actioned
A single month's ratio is context. A rolling trend against a baseline is the actionable view. OXMAINT AI plots the ratio per week per unit with driver overlays so a spike carries its own explanation. Book a demo to see the trend view live.
Cost per MWh — Rolling 8-Week Trend (Illustrative)
$8$6$4$2
Baseline · $5.00
W1W2W3W4W5W6W7W8
Below/on baseline
Above baseline · driver investigation
Below baseline · confirm cause before celebrating
W4–W5 spike attributable to contractor rate + reactive mix drivers per breakdown view.
The Reduction Play — 5 Moves That Actually Shift the Number
Once the number is visible and the drivers are named, five moves consistently shift it downward across generation fleets. OXMAINT AI holds each as a workflow, not a slide. Start free and configure the reduction playbook in OXMAINT AI.
01
Shift the Reactive/Planned Mix
Every 10-point shift from reactive to planned typically pulls cost/MWh down measurably — the leading lever
02
Retire the Deferred-PM Backlog
Structured burn-down over 2 quarters — cost spikes briefly, then flattens permanently
03
Contractor Scope Discipline
Vendor SLA ledger, T&M cap per WO, quarterly rate review — invoice surprises stop
04
Outage Plan Discipline
Critical-path scheduling, parts-kitted 30 days out, day-by-day burn-down against plan
05
Kill Corrective Rework
Warranty-callback flag on every closed WO — repeat-visit rate becomes a tracked KPI itself
What OXMAINT AI Gives a Power Generation Team
Purpose-built for the reliability-plus-commercial reality — one platform behind the WOs, the parts, the contractors, the historian and the board report. Start free and load your generation fleet into OXMAINT AI.
5-Input Cost Roll-Up
Labour, parts, contractor, corrective and outage-attributable spend all tied to the WO.
Historian Integration
Generation MWh pulled per unit per interval — no manual reconciliation lag.
Three Ratio Views
Gross, net, available-hour — pick per question, run all three side-by-side.
Driver Breakdown
Reactive mix, deferred PM, contractor rate, outage overrun — each driver overlaid on the trend.
Unit-Level Scorecard
One page per unit per period — the exact view the plant manager takes to the board.
Baseline & Alert Rules
Rolling baseline per unit — configurable alerts when the ratio breaks the band.
"
We used to publish cost per MWh as a lagging quarterly slide — by the time we saw it, the drivers were three months in the rearview. Once the CMMS pulled MWh from the historian and cost from every source in real time, it became a weekly operating conversation. The number itself didn't matter as much as the fact that the whole team was now watching the same one.
Plant Manager · 850 MW Combined-Cycle Gas Facility
Frequently Asked Questions
Can OXMAINT AI pull generation data from our plant historian?
How does the platform handle contractor invoices that arrive weeks late?
What labour rate should we use — loaded or raw?
How is corrective rework attributed?
Every WO carries a warranty-callback flag. If a related WO reopens within a defined window, both are grouped and the second is tagged as rework — cost rolls up under the corrective bucket.
Book a demo to see rework attribution.
How long before we see a defensible ratio?
Cost. Generation. One Number. Weekly.
Move your power plant cost visibility onto OXMAINT AI — 5-input roll-up, historian-linked denominator, driver breakdown and unit-level scorecard the whole team watches together.
