AACE 18R-97 (Cost Estimate Classification System)

AACE 18R-97 is the AACE International Recommended Practice titled "Cost Estimate Classification System — As Applied in Engineering, Procurement, and Construction for the Process Industries" — and despite the process-industry title, it is the foundational reference for cost estimate classification across UK infrastructure, defence and energy programmes. Published in its original form in 1997 (and revised multiple times since), it defines a five-class system that ranks cost estimates by the level of project definition behind them, the typical methodology used to produce them, the end-usage they support, and the expected accuracy range each class can credibly claim.

The five classes, summarised: Class 5 — concept screening / rough order of magnitude (ROM), 0-2% project definition, accuracy range -50% to +100%, used for portfolio screening and early sanction decisions. Class 4 — study or feasibility, 1-15% project definition, accuracy -30% to +50%, used for strategic planning and concept selection. Class 3 — budget authorisation, control or bid/tender estimate, 10-40% project definition, accuracy -20% to +30%, used for funding submissions and stage-gate decisions. Class 2 — control or bid/tender estimate, 30-75% project definition, accuracy -15% to +20%, used for project control and competitive bidding. Class 1 — check estimate or bid/tender, 50-100% project definition, accuracy -10% to +15%, the maximum accuracy class.

The classification system serves three practical purposes. First, it gives the estimator a defensible vocabulary for explaining what an estimate is and isn't capable of supporting. An estimate that is genuinely Class 3 should not be presented as Class 2 to make a sponsor more comfortable — the accuracy range is part of the deliverable, not a separate caveat. Second, it gives the sponsor a calibration framework for understanding the level of certainty they are looking at. A Class 5 estimate carrying -50% to +100% accuracy is a portfolio-screening tool, not a funding figure. Third, it gives the QRA practitioner the baseline-uncertainty calibration for the Monte Carlo simulation — the three-point estimate ranges on individual cost line items should be consistent with the AACE class of the parent estimate.

The AACE classes map approximately onto the UK IPA Cost Estimating Requirements stage-gate tolerance bands, though the mapping is approximate rather than one-to-one. Strategic Outline Case (SOC) typically uses Class 5 or Class 4 estimating with the IPA's -20% to +50% tolerance band (the IPA band is tighter than AACE Class 5's -50% to +100%, reflecting the IPA's expectation that SOC estimates are already informed by some preliminary engineering). Outline Business Case (OBC) typically aligns with Class 3 estimating and the IPA's -15% to +30% band. Final Business Case (FBC) aligns with Class 2 estimating and the IPA's ±10% band. The two frameworks are complementary — the AACE class describes the estimating method and underlying definition; the IPA band describes the gateway tolerance — and both should be cited in a UK business case.

Common misuses to watch for: presenting a Class 3 estimate with a Class 2 accuracy claim (typically by trimming the upper bound of the range to make the figure look tighter than the methodology supports); confusing the AACE class with the IPA band (different frameworks, different meanings); and treating the class as a property of the project rather than the estimate (a Class 5 estimate of a mature project still has Class 5 accuracy if the input data is at that definition level). The class is about the estimate's methodology and inputs, not the project's actual maturity.