Ask two people in a steel plant which costs are fixed and you will often get two different lists. Labour, maintenance and even some energy sit in a grey zone that changes with how hard the plant runs. That is why cost per tonne swings so much between a strong month and a weak one, and why budgets built on a single blended rate miss the mark. A CFO-grade model separates what moves with output from what does not, line by line. Teams building that model can see how iFactory AI classifies each cost line on live plant data before finalising next year's budget.
Fixed vs Variable Cost Modeling for Steel Plants
Learn how to classify labour, depreciation, maintenance and raw materials so cost per tonne reacts correctly when utilisation moves.
Why the Split Changes Every Decision
Steel is a capital-heavy business. A large share of cost stays put whether the furnace runs at 60 percent or 95 percent, so utilisation moves cost per tonne far more than most budgets assume.
This example assumes fixed cost is 20 percent of the full-load cost per tonne and the rest is variable. Only the fixed part spreads thinner as output rises, which is the whole effect.
The Cost Behaviour Spectrum
Few costs are purely fixed or purely variable. Place each line on the spectrum below before it goes into the model.
Does not change with tonnes in the planning period. Depreciation, insurance, salaried staff.
Holds steady, then jumps at a threshold. An added shift, a second crew, a new line.
Has a base plus a usage part. Maintenance, utilities, contract labour.
Rises and falls with every tonne. Ore, scrap, coal, alloys, consumables.
See Your Cost Lines Classified Live
Book a 30-minute session and iFactory AI will walk through how your own labour, maintenance, energy and raw-material lines would be split into fixed and variable parts.
Line-by-Line Classification Guide
Use this table as the starting point. Your plant's contracts and staffing rules may move a line one step along the spectrum, so validate the mapping in a guided model review.
| Cost Line | Typical Behaviour | Modeling Note |
|---|---|---|
| Depreciation | Fixed | Spread over planned tonnes, not actual, or cost per tonne looks better in strong months for the wrong reason |
| Raw materials | Variable | Track consumption per tonne, and price separately so the two effects stay visible |
| Energy | Semi-variable | Idle and holding energy is fixed; process energy scales with output |
| Direct labour | Step-fixed | Changes by shift pattern, not by tonne, unless overtime or contract labour flexes |
| Maintenance | Semi-variable | Planned upkeep is fixed; wear-driven repair follows running hours |
| Consumables | Variable | Refractories, electrodes and rolls wear with output, though not always in a straight line |
Two Lines That Break Most Models
Maintenance and labour cause the most classification errors, because each holds a fixed base and a flexible part at the same time.
Building the Model in Five Steps
A defensible model is built in order. Skipping the early steps is how classification errors reach the CFO's desk.
List every cost line by unit
Start from the ledger and map each line to a plant unit, so behaviour is judged where the cost occurs.
Place each line on the spectrum
Use contracts, staffing rules and history to decide the behaviour, not habit.
Split semi-variable lines
Separate the base from the usage part using regression on past periods or work order data.
Choose the planning volume
Fix the tonnage base for spreading fixed cost so the figure is comparable month to month.
Test with utilisation scenarios
Run high and low cases and check that cost per tonne behaves as the split predicts.
Classification Traps to Check Before Sign-Off
Run this list against the finished model. Each item has caused a wrong cost per tonne in real budgeting cycles.
Spreading depreciation over actual tonnes instead of planned tonnes.
Treating energy as fully variable when idle load is significant.
Mixing raw material price changes with usage changes in one variance.
Ignoring step costs when volume crosses a shift or crew threshold.
Plants that carry these traps often see the same symptom: a budget that looks right at planned volume and fails whenever output moves. Reviewing the model against live data with an iFactory AI specialist catches them early.
Where iFactory AI Fits
iFactory AI keeps the fixed and variable split current as real production, energy and maintenance data arrive, so the model does not go stale after the budget is approved.
Behaviour-based classification
Cost lines are tested against actual output history, so semi-variable lines are split with evidence rather than opinion.
Scenario view for the CFO
Change utilisation and see cost per tonne respond, with fixed and variable movement shown separately.
Price and usage kept apart
Raw material variances are split so buying decisions and furnace efficiency are not confused.
One model for finance and plant
Finance and operations review the same classification, which ends disputes about which number is right.
iFactory AI arrives pre-configured on an NVIDIA server that ships racked and ready with software pre-loaded. Scope covers cabling, network, ERP and MES integration, team training and 24x7 remote monitoring.
Frequently Asked Questions
Why does fixed vs variable classification matter for steel?
Steel plants carry heavy fixed costs, so cost per tonne depends strongly on utilisation. If the split is wrong, budgets and pricing decisions are wrong at every volume except the planned one. A correct split lets the CFO forecast margin at different output levels. You can arrange a session to test your current split on live data.
Is maintenance a fixed or variable cost?
It is usually both. Planned inspections and standing contracts behave as fixed, while wear-driven repairs follow running hours and tonnes. Splitting by work order type gives a more honest picture than treating the whole line one way. For a worked example on your data, request a maintenance cost walkthrough, or ask support how the split is configured.
How should depreciation be spread per tonne?
Depreciation is fixed, so it is best spread over planned or normal capacity tonnes rather than actual output. Using actual tonnes makes cost per tonne look better in busy months and worse in quiet ones without any real change in efficiency. Agreeing the planning base early keeps month-to-month comparisons fair. Look at how the planning base is set in a live model to see the effect.
Can labour be treated as variable?
Only the flexible part of it. Permanent crews and salaried supervision hold steady with output, while overtime, contract labour and incentives move with it. Modeling by shift pattern and adding the flex on top avoids overstating how quickly labour cost falls in a slow month. A short product tour with your staffing data shows where the base and the flex sit.
How often should the cost model be reviewed?
Review it at least once a year at budget time, and again whenever a major change hits the plant, such as a new line, a contract renewal or a shift change. Live tracking helps because drift in a semi-variable line shows up as it happens. To see a model that updates continuously, schedule a walkthrough with the iFactory AI team.
Build a Cost Model Your CFO Can Trust
iFactory AI classifies fixed and variable cost from real plant behaviour and keeps the split current. Book a walkthrough to see it built around your own cost lines.
