What belongs in a pre-FEED if you want to reach FID
Most projects that stall before the investment decision were underdefined two gates earlier. What a pre-FEED has to deliver, and which estimate class it can honestly support.
October 22, 2026 · 5 min read · Marcel Deurhof
A project rarely dies at the investment decision. It dies earlier, in a pre-FEED that answered the easy questions thoroughly and left the expensive ones open. By the time a number reaches the decision makers, the room to fix that has closed.
The pattern is consistent across industrial and energy development: the study was delivered on time, the report looks complete, and nobody can say what the project will cost within a band anyone will sign for.
What a pre-FEED has to settle
A pre-FEED is not a small FEED. Its job is to reduce a set of options to one selected configuration, put a defensible cost and schedule range around it, and produce an honest list of what is still open.
That last part is the one that gets dropped. Stage gate systems built for public capital are explicit about it. In the US Department of Energy critical decision process, the gate that follows concept selection approves an alternative and a cost range rather than a single number, because the range allows for the uncertainty that still sits in the estimate and the scope. A pre-FEED that hands over one number has quietly removed the thing the decision was supposed to weigh.
The owner test is also different from the contractor test. A contractor needs enough definition to price a scope. An owner needs enough definition to know whether the project is worth doing at all, and what it would take to make it fundable. Those are not the same deliverable, and that distinction sits behind our energy transition and pre-FEED work. The project types this applies to are listed at a glance.
The estimate class a pre-FEED can support
Estimate accuracy is a function of how much of the project is actually defined, not of how much effort went into the estimate. AACE International Recommended Practice 18R-97 sets the bands for the process industries. Class 5 sits at 0% to 2% project definition, with an accuracy range of 20% to 50% on the low side and 30% to 100% on the high side. Class 4, the usual pre-FEED product, sits at 1% to 15% definition, with 15% to 30% low and 20% to 50% high. Class 3, the first class intended for budget authorisation or control, needs 10% to 40% definition.
Read that as a sequencing problem, not an estimating problem. If the investment decision needs a Class 3 number and the pre-FEED leaves definition near the bottom of the Class 4 band, no estimator closes that gap. It closes with engineering, vendor data and site information that somebody has to go and get. Naming that work, costing it and putting it on a schedule is the real output of a pre-FEED.
Where pre-FEED scope usually runs thin
The same gaps come back:
- Brownfield tie-ins and the shutdown window they need, tested against the operating unit's actual turnaround plan instead of a generic allowance
- Utility and infrastructure balance across the whole site, not the new unit in isolation
- The permitting basis, including which permit drives the critical path and what data that permit requires
- Feedstock and offtake terms firm enough to survive a sensitivity run
- Operations and maintenance input on layout, access and sparing, collected before the plot plan freezes
One question per discipline usually finds the hole: what would have to be true for this cost to be wrong by 30%? If the answer is a study nobody has commissioned, that study belongs in the pre-FEED scope, not in FEED. And contingency does not rescue it. Contingency covers quantified uncertainty inside a defined scope. It was never meant to cover scope that nobody has written down.
Definition maturity, not document count
The Project Definition Rating Index takes the same view and makes it measurable. In the version documented by NASA for building projects, 64 scope definition elements are scored out of 1000, with 200 as the threshold. Projects scoring at or below 200 ran 1% over budget and 2% behind schedule. Projects above it ran 6% over budget and 11% behind schedule, with change orders at 11% of budget instead of 7%. That index was built for buildings rather than process plants, so the absolute numbers do not transfer, but the mechanism does: scored definition predicts outcomes better than the volume of paper a study produces.
In energy transition development the timing pressure makes this sharper. The IEA's Global Hydrogen Review 2026 records 4.3 Mt of committed low emissions hydrogen production by 2030 against an announced pipeline of 27 Mt, with new investment decisions falling below 0.8 Mtpa in 2025 from around 1 Mtpa in each of the two years before. It also warns that more than 100 GW of announced electrolysis capacity loses any chance of operating by 2030 if decisions are not taken before the end of 2027. Projects on that clock do not get a second development cycle.
Take the current pre-FEED deliverable list and write next to each item the decision it feeds and the estimate class it supports. Anything that feeds no decision comes off the list. Anything a decision needs and nobody owns goes on the schedule this week, with a name against it. The exercise takes an afternoon, and it changes what the gate meeting is about: not whether the report is finished, but whether the project is defined.
Is your project ready for FID?
We take early-stage projects from concept to executable.
More from M2DP
Stakeholder alignment is a structure, not a workshop
Chaos on a capital project is usually blamed on people. The research points elsewhere: unranked objectives, no written rule for conflicts, and nobody who owns the ranking. Here is how to test yours.
EPC Project Management Strategies That Actually Reduce Chaos
Most EPC project problems are visible long before they become expensive. This article sets out the strategies we apply on energy and industrial projects: governance that gets used, contracting strategies that fit the scope, and recovery.
Open book estimate: when it works, and when it does not
Opening the contractor's cost build-up gives you the right to see how a number was made. It does not tell you whether the number is reasonable, and those are two different capabilities.
If this article matches a project you are running, one email is enough to start.
Get in touch