1. AACE estimate classes
The AACE International recommended practice 18R-97 classifies cost estimates by the maturity of the project definition. The class drives the expected accuracy range and the appropriate method:
| Class | Project definition | Typical accuracy (low / high) |
|---|---|---|
| 5 | 0–2% (concept screening) | −20 to −50% / +30 to +100% |
| 4 | 1–15% (study / feasibility) | −15 to −30% / +20 to +50% |
| 3 | 10–40% (budget authorization) | −10 to −20% / +10 to +30% |
| 2 | 30–70% (control / bid) | −5 to −15% / +5 to +20% |
| 1 | 50–100% (check estimate / definitive) | −3 to −10% / +3 to +15% |
The ranges are asymmetric — costs overrun more often than they underrun, so the positive tail is wider. A Class 1 definitive estimate (project ~50–100% defined) is the tightest. Quoting a single number without its class and range is the most common estimating error.
2. Estimating methods
Early classes (5/4) use stochastic/factored methods — capacity-factored analogies, equipment-factored (Lang/Hand factors that scale total installed cost from major-equipment cost), and parametric models. Later classes (2/1) use deterministic methods — quantity take-offs priced from vendor quotes and labour norms. The method must match the class; a take-off on a 2%-defined concept is false precision.
3. The capacity-factor ("six-tenths") rule
Scaling cost with size uses the power law:
The exponent n < 1 captures economy of scale — doubling capacity costs less than double (≈ 1.5×). n varies by equipment type (compressors ~0.7–0.8, tankage ~0.6, whole plants ~0.6–0.7); apply a cost index (e.g. CEPCI/IHS) to escalate the base cost to the estimate date.
4. Contingency & ranging
Contingency is the cost added to a base estimate to cover the statistically-expected (but individually unidentified) growth within the defined scope — it is not a fudge factor and does not cover scope changes. Rigorous practice ranges the estimate with a Monte-Carlo or range-analysis model and reports a probabilistic spread. The AACE 18R-97 accuracy range is an ~80% confidence interval ≈ P10/P90 (the low/high bounds), with contingency typically set at the P50 of the modelled distribution. The estimate is then a distribution, not a point.
5. Economic decision metrics
The capital estimate feeds the project economics. The core metrics:
- NPV = Σ CFt/(1+r)t − CAPEX — value created above the discount rate r (hurdle/WACC). NPV > 0 creates value.
- IRR — the discount rate where NPV = 0; compared against the hurdle rate.
- Payback — years to recover CAPEX (simple or discounted); a liquidity/risk screen, not a value measure.
For capacity-expansion decisions, these are evaluated against demand growth and the economy-of-scale of building ahead of need versus incremental additions — the trade-off between carrying idle capacity and paying repeated mobilization/scale penalties.
6. References
- AACE International Recommended Practice 18R-97 — Cost Estimate Classification System (process industries).
- AACE RP 17R-97 — Cost Estimate Classification System (general principles); RP 40R-08 / 42R-08 — contingency & risk-based ranging.
- Peters, Timmerhaus & West, Plant Design and Economics for Chemical Engineers (capacity factor, Lang factors).
- CEPCI / IHS process cost indices — escalation to estimate date.
Ready to use the calculator?
→ Launch Calculator