A plant can be exactly the right size while something else is holding the asset back. Expansion capital gets proposed against the constraint that is easiest to measure, not the one that binds.
Nameplate capacity is a number on a sign. It is written down, everybody knows it, and it is the first thing anyone reaches for when asked why production is not higher.
The constraint that is actually binding is frequently something with no sign on it. Water disposal capacity. Compression on a gathering line. Truck availability. A permit condition on hours. Power supply. Any of these can be the reason the asset produces what it produces, and none of them are on the plant nameplate.
The economics of this are unforgiving. A constraint that is not binding has a shadow price of zero, meaning an extra unit of it is worth exactly nothing. Expansion capital aimed at a non-binding constraint does not earn a poor return. It earns no return, because the asset was never limited by the thing that got bigger.
Worse, relieving one constraint usually just exposes the next. Sizing without knowing the order of the queue produces an expansion that hits a new wall a few months after commissioning.
Perturb each constraint one at a time. Relax it by a small increment in the full model, re-run, and read the change in value. That change is the shadow price. Constraints that return zero are not binding and no amount of capital aimed at them will help.
Check utilisation against the sign. A plant running well below nameplate while production is capped means the cap is somewhere else entirely.
Ask the operators before the engineers. The people who run the asset day to day usually know exactly what stops them. It is rarely the thing in the capital proposal.
Shadow price every constraint, not just the candidate. Relax and tighten each one against the full model. The output separates what binds from what is merely large.
Couple capital to the dial being turned. Sizing a debottleneck without its capital cost attached produces a recommendation that ignores whether the increment is worth buying.
Find the next constraint before committing. Relieve the binding one in the model, re-run, and see what binds after that. If it binds immediately, the expansion is smaller than it looks.
An Asset Health Check finds this one on a producing asset in two weeks, fixed fee.
The Asset Health Check is the cheapest possible way to find out whether we are worth the larger engagement. Two weeks, fixed fee, and a written answer either way.