Variability Risk vs Event Risk

Good QRA methodology separates two fundamentally different kinds of uncertainty. Variability risk (also called estimating uncertainty or inherent uncertainty) is the range of plausible outcomes for something that definitely will happen: the duration of excavation work will fall somewhere between X and Y, the cost of a cabling package will land between A and B. Variability is modelled as a distribution applied to the base estimate itself — typically a triangular, PERT or BetaPERT distribution with low, most-likely and high values.

Event risk is different in character: it is a discrete event that may or may not occur, and only applies to the estimate if it does. A ground conditions risk might have a 30% probability of materialising and a £1–5m cost impact if it does; 70% of the time the risk contributes nothing at all. Event risks are modelled as probability-weighted impact distributions attached to the activities or cost lines they would affect.

The distinction matters because mixing the two produces misleading output. If estimating uncertainty is bundled into the risk register as if it were discrete events, the model double-counts uncertainty and overstates contingency. If discrete events are absorbed into wider variability distributions, the model loses the ability to identify the specific risks driving exposure and the top of the tornado chart becomes uninformative. AACE Recommended Practice 57R-09 and 113R-20 both emphasise keeping variability and event risk separated — and it is one of the clearest markers of whether a QRA has been built to a professional standard.