You have a number. Your capital committee wants a sanction decision by Friday, and that number is the only thing standing between a green light and another year of nothing happening. Here's what nobody put in the slide: that number came from a Class 5 estimate, which by AACE International's own classification system is allowed to miss by roughly -20% on the low side and +50% on the high side, and up to +100% in volatile markets.
If your project lands at the wrong end of that range, nobody will remember that you explained the estimate class correctly. They'll remember that you sanctioned the overrun. That's not a forecasting failure, it's a classification failure: asking a number that was only ever built to screen an idea to instead defend a budget.
More than 80% of process-plant cost overruns don't start on the construction site. They start earlier, in the gap between an unvetted concept and a capital request. Structured AACE Class 5/4 estimation closes that gap: it turns a raw flowsheet into a defensible capital budget, protects the project against early scope creep, and builds an audit-ready bridge into FEED.
At Acufore, our Pre-FEED studies are built to produce a genuine Class 4 number the first time, not a Class 5 estimate wearing a Class 4 label.
Drawing from over 90,000 engineering hours across 1,400+ industrial projects, here's what has to exist before a number can honestly claim to be Class 4, and what happens when that gap gets skipped.
On this page ▾
01What a Class 5 Estimate Is Actually For
AACE's own recommended practice, 18R-97, is blunt about this: a Class 5 estimate is built from roughly 0–2% project definition, essentially a concept, a capacity number, and a comparison to something similar built before. It's meant for screening ideas, comparing options, and deciding whether a concept deserves another dollar of study. It was never meant to survive contact with a procurement department, and it wasn't designed to.
Used for what it's for, a Class 5 number is genuinely useful. Used as a budget, it's a liability wearing a budget's clothes.
| Class | Maturity | Methodology / Inputs | Accuracy Range | Decision Gate |
|---|---|---|---|---|
| Class 5 | 0–2% (Concept Screening) | Parametric modelling, capacity factoring, historical benchmarking | -20% / +50% (up to +100% volatile) | Fund initial study |
| Class 4 | 1–15% (Feasibility / Pre-FEED) | Equipment-factored math, initial BFDs, preliminary equipment lists | -15% to +30% | Sanction Pre-FEED; bound FEED scope |
| Class 3 | 10–40% (FEED Complete) | Semi-detailed line items, PFDs, preliminary P&IDs, plot plans | -10% to +20% | Final Investment Decision |
The Plain Business Consequence
Used as a budget instead of a screening tool, a Class 5 number can miss by -20% to +50%, or up to +100% in volatile markets. If the project lands at the wrong end of that range, nobody remembers the caveat about estimate class, only the overrun.
02Why Class 4 Is the Number Your Board Should Actually See
A Class 4 estimate asks for more: a preliminary equipment list, early vessel and pump sizing, a Heat & Mass Balance, and a conceptual plot plan, roughly the deliverable set a proper Pre-FEED & feasibility engineering study produces. In exchange, the accuracy range tightens to roughly -15% to +30%. That's still wide. It's also the first number precise enough that a capital committee is sanctioning a project, not a hunch.
A Class 4 estimate costs a few extra weeks up front. A Class 5 estimate presented as if it were Class 4 costs the project its second year.
This isn't a theoretical risk. McKinsey's 2022 review of more than 500 capital projects worldwide, each valued above $100 million, found average cost overruns running at roughly 79% above the initial budget, with schedule delays averaging around 52%. Not every one of those projects sanctioned off a Class 5 number dressed up as something firmer, but a meaningful share of them did, and the pattern is consistent enough to take seriously.
The Plain Business Consequence
In our experience, project teams rarely miss the math inside the accuracy band they actually earned. What goes wrong is the label: presenting Class 5 confidence in Class 3 language, because nobody wants to tell a capital committee "we're not sure yet" twice in one quarter. We'd rather hand over an uncomfortable number with an honest class stamped on it than a comfortable one that quietly isn't.
03Why Unvetted Concepts Actually Fail: Three Hidden Cost Drivers
When early feasibility estimates fail on the shop floor or in construction, it's rarely an arithmetic error. It's almost always a missing baseline assumption. Three show up more than any others.
- Unverified material traceability & tariff blind spots: estimators often apply generic regional benchmarks for structural steel, piping spools, or pressure vessels. For international procurement, failing to factor in BS EN 10204 Type 3.1 material certification or local duty regimes like the UK-India CETA corridor can trigger tariff penalties or customs delays that never show up in the original estimate.
- Ignoring site-specific constraints in brownfield plants: a Class 5 estimate assumes greenfield conditions unless proven otherwise. If your expansion ties into live pipe racks, tight plot boundaries, or active process corridors, unverified spatial clashes will consume your contingency budget the moment they surface in the field, the exact failure mode covered in our piece on 3D point cloud brownfield revamps.
- Factoring without baseline equipment quotes: capacity-scaling formulas are a reasonable starting point, but relying on them entirely, without securing budgetary vendor quotes for major long-lead items like pumps, reactors, and heat exchangers, compounds error margins in exactly the direction you don't want: down, then over.
Pre-FEED deliveryThe Pre-FEED Deliverables Register: What Acufore Delivers
Instead of a static spreadsheet, an Acufore Class 4 Pre-FEED package produces a defensible Basis of Design backed by an audit-ready Pre-FEED execution register.
Raw Concept / PFD
Process Validation
3D Spatial Framing
Cost Modelling
Handover
Three things happen at once inside that register:
Process & Fluid Validation
HMB calculations and preliminary PFDs pin down true utility and throughput demand before a number gets attached to it.
3D Spatial Coordination
Structural pipe racks, cable bridges, and equipment footprints modelled against scanned, real-world geometry, not an assumed greenfield layout. See our guide on 3D point cloud brownfield revamps.
Regulatory & Tariff Alignment
Structural frameworks calculated to local annexes (Eurocode 3 / UK National Annexes) and material callouts aligned with certified fabrication hubs (BS EN 1090 EXC2), covered in our UK-India CETA guide.
Equipment-Factored Cost Modelling
AACE Class 4 CAPEX (±20%) built from budgetary vendor quotes on long-lead items, not capacity-scaling formulas alone.
Bounded FEED Handover
An audit-ready Basis of Design and a bounded FEED Scope of Work, locking in equipment counts and battery limits before capital is committed.
Elevating a concept from an unverified Class 5 guess to a disciplined Class 4 Basis of Design produces three immediate commercial outcomes: TIC accuracy within ±20%, securing board sanctioning backed by defensible engineering math; bounded FEED scope, preventing EPC change orders by locking in equipment counts and battery limits early; and streamlined procurement, pre-qualifying material specifications and trade-corridor advantages before capital is committed.
Cost classification never happens in isolation from the rest of the plant. A Class 4 number is only as good as the spatial data and trade assumptions underneath it, which is why this piece sits alongside our companion guides on 3D point cloud brownfield revamps and the UK-India CETA trade corridor.