Acceptance, and what changes the day the structure exists

A construction policy has two lives, and the boundary between them is a date. During the works it covers a structure that does not yet exist: an assembly of materials, earthworks and equipment under erection, whose value grows every week and which nobody can price exactly on any given day. After handover it covers something else, or nothing at all, and other warranties take over. Every end-of-project dispute lies in the move from one regime to the other, and the question that triggers it is always the same: on what date was the works accepted?

Acceptance is a legal act, not a physical observation. Under French law it is the act by which the project owner declares that it accepts the works, with or without reservations, and it produces three effects that are anything but symbolic. It transfers custody of the works to the project owner, which shifts the burden of risk. It starts the statutory warranties running, including the ten-year one. And it ends, absent a clause to the contrary, the contractor's all risks cover over the accepted part. Three effects, one date, and that date is disputed all the more because it is rarely written down at the moment it occurs.

Acceptance has several faces, and that is what makes it hard. Express acceptance, recorded in a joint minute, is the simple case and the rarest on a large project. Acceptance with reservations is the norm: the works are accepted, the effects occur, and a list of outstanding items is appended. Sectional or phased acceptance carves an industrial complex into sections accepted on different dates, which multiplies the boundaries. And tacit acceptance is inferred from conduct: a project owner that takes possession of the site, pays the balance and operates the installation has, on settled case law, shown its intention to accept, even without any minute. That is the most dangerous form, because it is established after the fact, often before a judge, and always at a moment when somebody has an interest in placing it elsewhere.

Early taking of possession is the trap on industrial projects. The project owner wants to produce before the works end, it commissions one line while the contractor finishes the next, and both parties assume nothing has changed. Many contractor's all risks policies say the opposite: they stop covering the section put into operation, because the risks of a structure in service are not those of a structure under construction. A turbine spinning to sell electricity is no longer equipment under erection, it is a production asset, and the operator normally holds a property damage policy for that. Between the two, on the project's real calendar, there is almost always a zone where neither policy considers itself concerned.

The maintenance or defects liability period exists precisely to cover the aftermath. It comes in three shapes, whose names vary but whose logic does not. The narrowest guarantees only damage caused to the accepted works by the contractor itself when it returns to site to perform its making-good obligations: it breaks a pipe while redoing a weld, the policy pays. The intermediate shape adds damage that appears during the period and whose cause dates back to the execution of the works, which is very different, since the contractor may never have returned. The broadest covers both, plus damage of any origin occurring during the period, which brings it close to a temporary property damage policy.

That gradation decides whole files, and it reads badly because the three shapes often carry the same heading in the table of contents. The question to put to a wording is never 'is the maintenance period covered', which always answers yes. It is: must the triggering event be a contractor intervention during the period, or defective execution during the works, or neither? Depending on the answer, a weld defect that shows up eight months after acceptance, with nobody having returned to site, is covered or is not, and the amount at stake depends on no other parameter.

That leaves the costliest question, which is one of calendar rather than of cover. The maintenance period has a duration, twelve or twenty-four months depending on the project, and it runs from acceptance. On a project accepted in sections there are therefore as many periods as there are sections, each with its own start and expiry. An industrial complex whose first section was accepted eighteen months before the last can find its oldest section already out of warranty on the day the complete installation first runs at full load, that is, the very day execution defects are most likely to surface. Sectional carve-up, decided for scheduling reasons, thus produces a gap in cover that nobody wanted and nobody looks at until something happens.