Energy infrastructure megaprojects rarely stay within their initial budgets. More often than not, the gap is created long before construction begins, due to an initial estimate that is set too low. This article focuses on that specific moment: costing and provisioning before launching a major offshore or nuclear project.
It is intended for CFOs, project directors, cost controllers, and estimation teams. It explains why estimates go off track with such regularity and which levers can be used to improve accuracy.
Cost overrun: a statistical phenomenon in energy megaprojects
Before looking for someone to blame, we must measure the regularity of the phenomenon. Budget overruns on megaprojects are too frequent and too systematic to be attributed to bad luck. They follow a pattern.
What the data says: chronic underestimation in nuclear and offshore
Based on a database of over 16,000 projects, Bent Flyvbjerg observes that only 8.5% meet both their budget and schedule. The energy sector is among the worst performers. The study by Sovacool, Gilbert, and Nugent (journal Energy, 2014), covering 401 power projects, confirms this: costs are underestimated in about 75% of cases, with an average overrun of 117% for nuclear. Such consistency points to a structural cause.
Initial estimate vs. Estimate at Completion (EAC)
A common confusion often clouds this subject. Estimating a project's cost before launch and managing that cost during execution are two different exercises. Ongoing project monitoring, or the Estimate at Completion (EAC), provides real-time corrections. However, when the initial estimate is flawed, all subsequent management starts from a biased foundation.
Optimism bias in offshore wind projects
The primary cause is cognitive. When estimating a project, we spontaneously calculate the scenario where everything goes right and underweight risks, simply because we lack the cost distributions of past projects. This is the optimism bias, documented across thousands of projects. The environment exacerbates it: a young, non-standardized industry where data from previous projects is rarely captured.
Inside view vs. outside view: two ways to estimate
The distinction, popularized by Daniel Kahneman and Bent Flyvbjerg, sheds light on the mechanism. The "inside view" estimates the project from within, based on its own assumptions and plans. The "outside view" positions it relative to a family of comparable projects already completed. The former underestimates because it ignores what history could have taught us. The latter brings the estimate back toward observed reality.
The offshore case: what is under-provisioned
At sea, under-provisioned items are identifiable. The standby costs for installation vessels, which range from 150,000 to 500,000 euros per day—and up to a million for a latest-generation jack-up—are rarely provisioned accurately. Weather windows and maritime force majeure, often poorly covered in contracts, turn into unforeseen costs. An estimate that ignores these specific risks is doomed to be exceeded. You can find more information in this article: Offshore wind vessel standby fees: why CFOs find out too late.
The same logic applies to end-of-life obligations, for which provisions are built up from the moment of commissioning. You can find more information in this article: Decommissioning and the lifecycle of an offshore wind farm: how much to provision and how to avoid underestimation.
Reference class forecasting: estimating based on comparable projects
The most direct remedy for optimism bias is to estimate based on truly comparable projects, rather than relying solely on internal plans. This is reference class forecasting, the practical application of the "outside view."
How reference class forecasting works
The method consists of three steps. First, identify a reference class—a set of comparable projects with known actual costs. Next, derive the distribution of observed cost overruns from this set. Finally, adjust the initial estimate based on this distribution. The method was first used for the Edinburgh tramway, drawing on about forty reference projects.
Its limitations
Reference class forecasting reduces bias, but it does not eliminate it, and it is best to be clear about that. Its reliability depends on the quality of the reference class: if it is too narrow, it lacks data points; if it is too broad, it loses relevance. It also assumes the availability of credible data on past projects, which is rare when those projects are not standardized.
Sizing contingency: from flat percentages to P50/P90
An accurate estimate is more than just a central value. It includes a contingency provision, or contingency, commensurate with the actual level of uncertainty.
Why a flat percentage doesn't hold up
Adding a "10% margin" to a major energy infrastructure project is reassuring but misleading. Estimation standards, such as the AACE classification, acknowledge this: a very preliminary estimate can have a range of -50% to +100%. A fixed flat rate does not reflect this reality.
Monte Carlo, P50, and P90: choosing a confidence level
The probabilistic approach simulates thousands of cost scenarios (Monte Carlo simulation) to produce a distribution rather than a single figure. We then identify the cost that has a 50% probability of not being exceeded (P50) and the one that has a 90% probability (P90). The gap between the two provides the contingency justified by the risk. The target budget is set based on an accepted confidence level, often around P80, rather than a round number.
Moving from reactive costing to proactive provisioning
Reliable estimation is a methodical process that must be carried out before launch.
A pre-launch estimation checklist
Four key habits define this approach. Use an "outside view" by comparing with real-world projects. Provision for project-specific risks, whether in offshore wind or nuclear. Size your contingency based on probability—P50 versus P90—rather than a flat fee. And keep revising these assumptions until the scope is finalized.
A fifth habit is worth adding: cost the work breakdown structure itself, as every interface created between two contracts carries its own cost risk. You can find more information in this article: Managing interfaces in offshore wind: mastering work packages before they become disputes.
The skills that make the difference
These methods are only as good as the people behind them: cost controllers who think in terms of final outcomes, and estimators who understand the real-world risks of offshore and nuclear projects and know how to build a cost distribution model.
Renergy strengthens execution control for major projects through two complementary levers: the deployment of experts in estimation, project control, and project finance and the training of internal teams.
If these topics align with your needs, you can book a meeting with a Renergy consultant to discuss them. The conversation is confidential and aims to identify solutions tailored to your specific exposure.
To learn more, discover the four engineering-phase decisions that save the construction of a major energy project.
