The natural event that gets underwritten and the natural event that damages sites are not the same one. The one underwritten is the cyclone, the hundred-year flood, the tremor: it has a name, it is modelled, it takes up the placement discussion and it almost never happens on any given operation. The one that damages sites is an ordinary thunderstorm falling on works that are not finished. It has no name, it is not modelled, and it comes back several times a year on a long operation.
The reason for that vulnerability is simple and it is a matter of calendar. Every protection the finished works will have against water, the roof, the sealed envelope, the downpipes, the network draining to a sized outfall, is precisely what is not yet built. A completed building shrugs off the rain that filled its own excavation six months earlier. The switchover day module describes the moment that vulnerability ends; what comes before it is a period in which the works are, by construction, less resistant than they will ever be afterwards.
One must then look at what the policy actually excludes, because it is almost never the rain. It is normal climatic conditions, those to be expected in the season and the region, and that is a quite different object. The wording does not say water is not covered: it says what a careful site would have absorbed without damage is not fortuitous. The discussion, on the day, is therefore not about the event's intensity but about the boundary between what had to be borne and what went beyond what could reasonably be asked.
That boundary moves with the length of exposure, and it is the point files most often miss. A five-year return period event is rare on a given day and near certain over an excavation left open six weeks in the wet season. What must be looked at is not the statistics of the day but those of the window during which the works are vulnerable, and that window is a fact of the programme, not of meteorology. Two identical operations, one opening its excavations in June and the other in November, do not carry the same exposure, without a single line of the policy changing.
Then comes the family of clauses that decides most files, and which is not an exclusion. A pumping capacity to be maintained, a maximum length of open trench, backfilling within a given number of days: these are conditions whose performance is made a precondition. Their force lies in this, that an unmet condition precedent can defeat the claim without any need to show the breach played the least part in the damage. That is exactly what the drafting is for, and it is what separates it from an ordinary obligation whose breach is sanctioned only to the extent of its consequences.
The French natural catastrophe compensation regime brings almost nothing here, and it is better to know that than to hope for it. It requires an order recording the state of catastrophe for the commune and the period, and the storm that fills an excavation is not the subject of one. The event triggering that regime is of a different scale from the one this module is about. Counting on it means preparing an answer for the rare loss and having none for the one that happens.
That leaves the shape of the damage, which explains why nobody sees the whole. What costs is not destroyed material but making good: pumping out again, cleaning, digging out a disturbed formation, rebuilding a slope, checking and then replacing fouled reinforcement, and recovering the programme. Each event taken alone stays close to the deductible, often below it, and is therefore handled on site without ever becoming a file. The attritional loss module describes this mechanism: a sum nobody adds up because not one of its terms ever crossed the threshold that would have made it visible.
An urban excavation of 6,200 m² taken down to 14 m carries a dewatering condition precedent: four pumps of 120 m³/h, one of them standby. For nine days one pump has been under maintenance and three have run. On June 22, 71 mm of rain falls in three hours, a twelve-year return period at the nearby station. Run-off from the surrounding streets fills the excavation to 3.10 m. The raft reinforcement, laid over 2,400 m², is displaced and fouled. Material damage comes to €1,030,000: pumping and cleaning €240,000, reinforcement €610,000, rebuilding a slumped ramp €180,000. The water and earthworks deductible is €350,000 per event. The insurer relies on the condition precedent. What is the argument worth?
The missing pump's part must be quantified before the law is argued, and it is the calculation nobody performs because it looks lost in advance. The volume entering the excavation is 3.10 m over 6,200 m², that is 19,220 m³ in three hours, an inflow rate of 6,407 m³/h. Rain falling directly within the footprint counts for almost nothing in that: 71 mm over 6,200 m² makes 440 m³, that is 147 m³/h, which three pumps absorbed without effort. The water therefore came from the streets, and that is what would have had to be fought. Full contractual capacity, 480 m³/h, is 7.5% of the inflow; capacity actually in service, 360 m³/h, is 5.6%. The gap between them is 120 m³/h over three hours, that is 360 m³ out of 19,220, meaning 1.9% of the volume, or 5.8 cm of water depth. With four pumps the excavation was under 3.04 m instead of 3.10 m, and the reinforcement was exactly as fouled. The breach therefore played no material part, and that is precisely what makes the file instructive: if the clause is drafted as a condition precedent to cover, the absence of a causal role does not save it, because that drafting exists exactly to relieve the insurer of establishing a link. If it is drafted as an ordinary obligation, the 5.8 cm become the central argument and the claim stands. The discussion is about €680,000, that is €1,030,000 less the deductible, and its outcome was decided the day the clause was written, well before it rained.
- 01The natural event underwritten and the one that damages sites are not the same: the first has a name and almost never comes, the second is a thunderstorm and returns several times a year.
- 02Every protection the finished works have against water is precisely what is not yet built: the period before switchover is when the works are the least resistant they will ever be.
- 03The policy almost never excludes rain: it excludes the season's normal conditions, and the discussion is about the boundary between what had to be borne and what went beyond it.
- 04That boundary depends on the exposure window, not the statistics of the day: a five-year return period event is near certain over an excavation left open six weeks, and that window is a fact of the programme.
- 05An unmet condition precedent can defeat the claim with no showing of any causal role for the breach: that is what the drafting is for, and what separates it from an ordinary obligation.