AACE International Recommended Practice 18R-97 is the most widely used framework for classifying cost estimates in the process industries. It links estimate accuracy to how well the project is defined.
- Estimate class depends on project definition, not the effort spent estimating.
- Accuracy ranges narrow as engineering progresses.
- Always state the class and basis of estimate with the number.
| Class | Project definition | Typical use | Indicative accuracy range |
|---|---|---|---|
| Class 5 | 0–2% | Screening | −50% to +100% |
| Class 4 | 1–15% | Feasibility / pre-FEED | −30% to +50% |
| Class 3 | 10–40% | Budget / FID | −20% to +30% |
| Class 2 | 30–75% | Control / bid | −15% to +20% |
| Class 1 | 65–100% | Check estimate | −10% to +15% |
The ranges above are indicative; the actual range for any estimate depends on technology maturity, location and the quality of the input data.
Contingency is not a fudge factor
A good estimate sets contingency using risk analysis — for example a Monte Carlo range analysis producing P50 and P80 values — rather than a flat percentage. This makes the number defensible when it goes to the board.
The basis of estimate
Every estimate should come with a basis of estimate document covering scope, battery limits, pricing sources, exclusions and assumptions. Without it, two estimates of the same plant cannot be compared.