You asked what the boiler plant replacement will cost. The first honest answer is a range, and the width of that range is information rather than evasion.
The most honest number in a first study is a range.
That is uncomfortable to hand an owner who wants a figure to take to a board. It is also the only defensible answer before anyone has drawn the mechanical room, confirmed your electrical capacity, or tested the market. What is available is a range, a label saying how wide it is, and a statement of what would narrow it.
The industry has a published system for this. AACE International maintains Recommended Practice 56R-08, the cost estimate classification system for the building and general construction industries, and it expressly covers renovation work rather than new construction alone. It sorts estimates into five classes, and what sets the class is how mature the project definition is. Not how much design has been drawn as a percentage. The maturity of the defining deliverables.
A number produced at the earliest stage lands in Class 5, whose stated end usage is functional area or concept screening. AACE puts the expected accuracy of a Class 5 building estimate as broad as minus 30 to plus 50 percent, or as narrow as minus 20 to plus 30 percent, at an 80 percent confidence interval. Vandecarb states its screening figures at the broad end, plus 50 and minus 30 percent, on the page where the capital figures sit rather than in a footnote.
The width of the band is not a measure of how careful the estimator was. It is a measure of how much has not been decided yet.
What it changes is the decision, not the arithmetic. A screening number sets direction: which building, which system, which sequence, and whether a project is worth developing at all. A ranked order survives a wide band, because the options move together when costs move. What does not survive a wide band is a go or no-go against a fixed budget, or a reserve fund contribution set to the dollar. Those need the later number, which in 56R-08 is Class 3, covering design development and budget authorization, with accuracy running minus 5 to minus 15 and plus 10 to plus 20 percent. It is earned with drawings, specifications and real pricing, and it should not be promised at screening prices.
An unlabelled point estimate hides which of the two you are holding. It presents an assumption in the visual language of a commitment, and it is the one that clears a board without a second question, which makes it the riskier document rather than the safer one.
Where this argument stops. A wide range is not automatically the rigorous one, and 56R-08 makes that case itself: an accuracy range should always be determined through risk analysis of the specific project and should never be pre-determined. That is a fair challenge to any firm quoting a standing band, including this one. A stated band is an opening disclosure about the stage of work, not a substitute for that analysis, and it should move when a building argues for it. The same document notes the high end can run two to three times the published figure on complex or poorly defined work, and that 56R-08 does not cover one-of-a-kind buildings such as hospitals, laboratories and stadiums. If that is your asset, the classes describe the stages but the published bands do not describe your risk.
What to do before your next capital decision. When a screening estimate arrives as a single number with no stated class and no band, the useful question is not whether it is wrong. It usually is not. The question is what was left out to make it look certain, and what that does to your plan when the missing work surfaces at tender.
Four questions get you most of the way, and they work on any advisor:
- What estimate class is this, and what accuracy band comes with it?
- Which defining deliverables exist today, and which are still assumed?
- What work would narrow the band, what does it cost, and when is it normally done?
- Which decisions in front of me can this number carry, and which need the later one?
If the first question draws a blank look, you have learned something for the price of asking.
info@vandecarb.ca