Maintenance cost per megawatt-hour is the truest measure of how a power plant is running. It captures everything reactive repairs, overtime, expedited parts, lost generation, and asset health all collapsed into a single defensible number for the CFO. When a 320 MW combined-cycle plant in our case sample brought that figure down from $4.6 to $3.4 per MWh in 18 months, the change wasn't a single tool or training. It was a structured transition from spreadsheet-driven reactive work to an asset-linked storeroom, predictive PM schedules, and a mobile-first technician workflow built on OxMaint's CMMS platform, with measurable savings booked quarter by quarter.
$4.60
Maintenance cost per MWh at baseline (Month 0)
$3.40
Maintenance cost per MWh at Month 18
26%
Reduction in cost per MWh over 18 months
$3.1M
Annualized maintenance savings at run-rate
Plant profile
A typical mid-merit plant with a typical cost problem
The plant in this study is representative of dozens of facilities OxMaint has onboarded over the past three years. Gas-fired, combined cycle, mid-merit dispatch, aging balance-of-plant equipment, and a maintenance team that knew the numbers were drifting in the wrong direction but couldn't isolate why. Maintenance leadership had three competing pressures pulling against each other: corporate cost reduction targets, an aging asset base, and a maintenance technology stack stitched together from Excel, email, and a legacy work order tool.
Capacity
320 MW combined cycle
Configuration
2× GT + 1× HRSG + 1× ST
Capacity factor
58% mid-merit dispatch
Annual generation
1.62 million MWh
Maintenance headcount
34 FTE plus contractors
Asset population
4,200 tracked records
The baseline
Where the $4.60 per MWh was actually going
Before any platform change, the plant ran a 60-day baseline audit to break down its $7.45M annual maintenance budget into actionable cost categories. The result surprised the leadership team. Reactive work and emergency parts procurement together accounted for 42% of total spend. That single insight reframed the entire project, the goal stopped being "spend less" and became "shift the work from emergency to planned."
Reactive repair labor (1.5× to 2× overtime premiums)
$1.84M
Emergency parts procurement (2.4× planned pricing)
$1.27M
Lost generation revenue during forced outages
$1.18M
Scheduled preventive maintenance program
$1.02M
Contractor and OEM service calls
$0.84M
Inventory carrying and obsolescence
$0.66M
Compliance documentation and audit prep
$0.64M
Three pillars
The three changes that moved the cost-per-MWh needle
No single feature did this. The 26% reduction came from three coordinated workstreams running in sequence and reinforcing each other. Each pillar contributed measurable savings of its own, but the compound effect, where predictive insight feeds storeroom planning which feeds technician productivity, is what made the result durable rather than a one-quarter dip.
01
Asset-linked storeroom
Every spare part tied directly to the asset it serves, with consumption history feeding reorder points. Emergency procurement events fell from 38 per quarter to 9 within nine months. Parts spend dropped 22% in year one.
Cost impact
$0.42 per MWh saved
02
Predictive PM schedules
Calendar-based PMs replaced with condition-triggered work orders pulling from vibration, temperature, and runtime data. Six avoided failure interceptions in 18 months, including one HRSG tube event estimated at $680K in deferred outage cost.
Cost impact
$0.51 per MWh saved
03
Mobile technician rollout
Field technicians moved from paper work packs and clipboards to tablet-based execution with photos, meter readings, and parts confirmation logged in real time. Wrench time increased 11%, average work order close-out time dropped from 4.6 to 2.8 days.
Cost impact
$0.27 per MWh saved
See it on your own plant data
Map your current cost per MWh against this baseline
Bring your last 12 months of maintenance spend, overtime hours, and outage records. OxMaint's reliability team will walk you through the same breakdown that anchored this case study, your specific reactive ratio, your parts premium, your lost generation exposure, and the realistic 18-month trajectory for your plant.
Quarter by quarter
The 18-month trajectory: how the number actually moved
Cost-per-MWh did not fall in a straight line. The first quarter showed almost no change, which is typical and expected. Real reductions began in Q2 once the asset registry was cleaned, hit a sharp drop in Q3 when the storeroom rebuild went live, and compounded through Q4 and beyond as predictive PMs accumulated enough historical data to actually predict.
Quarter
Cost per MWh
Primary workstream
Plant condition
Q0 Baseline
$4.60
Audit and discovery
Reactive ratio 42% of total spend
Q1
$4.55
Data migration and asset hierarchy
4,200 records normalized in OxMaint
Q2
$4.28
Mobile rollout to two crews
Wrench time up 7%, paper logs retired
Q3
$3.92
Asset-linked storeroom live
Emergency parts orders down 58%
Q4
$3.71
Predictive PM on critical assets
First avoided failure interception logged
Q5
$3.54
Failure code library standardized
Diagnostic time down 47 minutes per event
Q6 Month 18
$3.40
Steady-state operation
Reactive ratio reduced to 19%
Before and after
The operational metrics that produced the financial result
A cost-per-MWh number on its own is not a defensible KPI. The CFO wanted to see the operational metrics underneath it, because those are the levers maintenance can actually pull. The table below shows what changed quarter over quarter at the asset and workflow level. Every operational improvement is traceable to a specific intervention rather than general organizational change.
The rollout sequence
What the implementation actually looked like, week by week
The rollout was deliberately staged. Trying to switch a 320 MW plant from spreadsheets to a full predictive platform on day one is how most CMMS projects fail. The team broke the 18 months into six phases, each with a defined go-live event, a measurable outcome, and a stop-and-stabilize period before the next phase began.
Weeks 1-4
Discovery and asset audit
Maintenance leadership and the OxMaint onboarding team walked the plant, validated the asset list against actual nameplate data, and identified 340 records that were duplicates, orphans, or missing critical fields.
Weeks 5-10
Migration and hierarchy
Excel registers, paper PM files, and legacy work order data were migrated. Asset hierarchy was rebuilt around system-level groupings (gas turbine train, HRSG, steam turbine, BOP) instead of the old equipment-flat list.
Weeks 11-18
Mobile rollout phase one
Two pilot crews moved to tablet-based work execution. Paper work packs were retired for routine PMs. Photos, meter readings, and parts confirmation logged in OxMaint within minutes of completion.
Weeks 19-30
Storeroom rebuild
Every active spare part was linked to its parent asset. Reorder points were recalculated based on actual 24-month consumption. 1,140 obsolete parts were written off, freeing $410K in tied-up working capital.
Weeks 31-48
Predictive PM activation
Top 60 critical assets moved from calendar-based to condition-triggered PM. Vibration thresholds, temperature limits, and runtime triggers were configured. First avoided failure interception logged at week 39.
Weeks 49-78
Steady-state optimization
Failure code library standardized, KPI dashboards rolled out to plant manager and corporate reliability team. Cost-per-MWh became a board-level reported metric. Reactive work ratio stabilized below 20%.
CFO view
The financial framing that got the project approved
Maintenance projects fail at the budget stage when they're presented as software purchases. This one was approved because the maintenance director presented it as a cost-per-MWh reduction program, the way a generation business actually thinks. Every dollar of platform spend was mapped to a specific savings lever with a conservative payback assumption. The CFO signed off in a single 45-minute review.
$3.10M
Annualized maintenance savings at run-rate (year 2 onwards)
7.2 months
Simple payback period on total platform and onboarding investment
$680K
Single avoided HRSG tube failure documented as interception event
$410K
Working capital released from obsolete inventory write-off
4.3×
Three-year ROI multiple against initial deployment cost
113 hrs
Forced outage hours avoided per year at $11,200 per hour exposure
What it took
The non-software factors that made this work
A platform is necessary but not sufficient. The plants that get stuck at 5% improvement and the plants that get to 26% reduction in cost per MWh have identifiable behavioral differences. The maintenance director on this project put four conditions in place before kickoff, and credited those conditions for the result more than any software feature.
A
Executive sponsorship at the plant manager level
The plant manager attended weekly status meetings for the first six months. When operations pushed back on outage windows for predictive PM activation, the plant manager broke the tie. Without that authority, maintenance projects get traded out for short-term generation targets.
B
A single owner for the cost-per-MWh metric
The reliability engineering manager was given accountability for the number and a weekly review cadence with the plant controller. Shared accountability would have meant no accountability. One owner, one number, every Friday.
C
Permission to write off bad data
Migration is where most CMMS projects bleed credibility. The plant chose to write off 340 asset records that were too corrupted to recover and rebuild them from nameplate data. This was unpopular but essential. Garbage in, garbage out applies more to maintenance than almost any other discipline.
D
Field technician input on mobile workflow
The mobile rollout was co-designed with two senior technicians who knew exactly where paper-to-tablet friction would emerge. The result was a workflow technicians used voluntarily by week three rather than one they resented for six months. To see the same mobile workflow on your own crew before committing,
book a walkthrough with the OxMaint team.
Common questions
Frequently asked questions from plant leadership teams
Is a $4.60 to $3.40 cost-per-MWh reduction realistic for a plant our size?
Yes for plants currently running reactive ratios above 35%. The 320 MW plant in this study started at 42% reactive, which is common in mid-merit fleets. Plants already at 20% reactive will see smaller absolute reductions but typically still 10 to 15% on cost per MWh.
How long before our CFO sees a measurable improvement on the maintenance budget?
First measurable cost reduction typically appears between months 4 and 6, driven by overtime reduction and emergency parts elimination. Full run-rate savings stabilize by months 14 to 18.
Start a free trial to begin baseline data capture immediately.
Do we need new sensors and IoT hardware before we start seeing results?
No. The case study plant achieved a 26% cost-per-MWh reduction without new sensor deployment. Existing SCADA tags and runtime meters provided enough condition data for the first 12 months. Sensor expansion happened only on the most critical 14 assets.
What happens to historical work order and PM data during the migration?
OxMaint supports direct CSV and Excel import for asset registers, PM schedules, work order history, and parts inventory. Migration typically completes in the first two weeks of deployment, with historical records preserved and searchable against each asset from day one.
Can we run the rollout in phases without disrupting current operations?
Yes, phased rollout is the recommended approach. The case study plant ran six phases over 18 months without a single operational disruption.
Book a 30-minute demo to discuss a phased deployment plan tailored to your plant.
Your plant has a cost per MWh number. We can help you move it.
The plants that move cost per MWh from $4.60 to $3.40 do not start with a software purchase. They start with a 30-minute conversation about where the money is actually going. Bring your last 12 months of maintenance spend, overtime hours, and outage records, and the OxMaint team will map your numbers against this case study, give you a realistic 18-month trajectory, and show you exactly which workstreams will move your specific cost line first.
