The module on accumulation describes a project as a point: all the value in one place, one storm to reach it. A whole family of works contradicts that image. A road, a railway, a pipeline, a power line, a subsea cable extend over tens or hundreds of kilometers, and value there is spread rather than concentrated. It is too quickly concluded that the risk is smaller because it is dispersed. That is true for some perils and false for others, and that asymmetry is the subject.
Dispersion protects against what is local. A fire on a linear project burns a works compound and not the three hundred kilometers; a flood reaches a section in a valley floor and leaves the rest dry. The probable maximum loss of a linear structure is therefore structurally smaller, relative to its total value, than a plant's. That is the argument that gets these projects viewed favorably, and it is sound. It stops being sound as soon as the peril considered is itself extended: a storm crosses a route over its full width, an earthquake concerns a region, and a two-hundred-kilometer route is far more likely than a point to cross a dangerous zone.
That last remark reverses the reasoning, and it must be held firmly. On an extended structure, the probability of meeting a given hazard rises with length, even as the consequence of each meeting falls. Frequency goes up and unit severity goes down. A linear project is therefore not underwritten like a point project with a smaller figure: it is underwritten on a different profile, where attritional loss weighs more than the extreme event, which shifts the whole reasoning onto the deductible and the limit rather than onto accumulation alone.
A second feature stems from the nature of linear works: they cross things, and what they cross rarely belongs to them. A route crosses roads, rivers, buried services, railways, private property and protected areas. Each crossing is a point of contact with a third party, therefore a liability exposure, and there are hundreds of them. The module on adjoining property described a site with two neighbors; here the same mechanics repeat along the whole length, and the liability head weighs proportionally far more than on a point project of the same value.
Buried services deserve a place of their own because they are the family's leading cause of loss and because their treatment is poorly understood. A machine severs a gas main, an optical fiber, a supply cable. The physical damage is often modest, a few tens of thousands of euros of repair, and the consequences are not: interruption of a service, third-party business interruption, sometimes thousands of subscribers. Those consequences come under liability and not under cover for the works, they count as non-physical loss, and many policies sub-limit them precisely where they make the amount.
The project's progression introduces a last difference, which favors the project and which people forget to argue. A linear project is almost never exposed over its whole length at once: crews advance, one section is finished and backfilled while another is open. The value at risk at any given moment is therefore a moving window, far smaller than the total value, and that is a technical argument documented with a schedule rather than a commercial one. The policy still has to be built on it, since a sum insured set on total value does not reflect that reality and a declaration of values by section can.
That leaves the question of the unit of loss, which decides amounts on these works more than anywhere else. Does a storm damaging twelve points of a route over three days produce one loss or twelve? The module on program layers showed what a badly matched hours clause does between floors; here the question arises within the policy itself, and it is doubled by a question of place that point works do not know. Twelve deductibles applied to twelve modest repairs can absorb the whole of a loss the limit covered comfortably, and that is the mechanism by which a properly sized program pays nothing.
A 180-kilometer transmission pipeline is under construction, total value 240 million euros. The policy carries a 60 million limit, a 250,000 euro deductible per loss, and a clause defining a loss as all damage occurring in one place and arising from one cause. A storm crosses the route on November 8 and 9 and damages nine distinct work zones, 4 to 60 kilometers apart: collapsed trenches, displaced pipe, flooded works compound. Repairs total 3.1 million euros: three zones at 180,000 euros each, and six zones sharing 2.56 million. Separately, three weeks earlier a machine severed an optical fiber at a crossing: 40,000 euros of repair, and 1.9 million claimed by the operator for its service interruption. What is left?
On the storm, the reflex is to count 3.1 million under a 60 million limit and conclude the program absorbs it. The opposite happens, and the word that decides is *place*. Nine zones 4 to 60 kilometers apart are not one place, whatever the unity of the cause: the clause requires both, and the storm supplies only one. So there are nine losses and nine deductibles. The count is not a matter of subtracting nine times 250,000 from the total: the three zones at 180,000 euros stay below the deductible and yield nothing, without a full deductible being chargeable to them either. That leaves six zones carrying 2.56 million, of which 1.5 million is deductibles, so 1.06 million for the insurer on a 3.1 million loss. Two million stays with the insured. The 60 million limit played no part: a properly sized program pays almost nothing, and the mechanism lies entirely in the definition of a loss. On the fiber, the error would be to look at the 40,000 euros of pipe. The physical damage is modest and the consequences are not: the 1.9 million claimed by the operator is non-physical liability loss, not repair of the works, and that is where the sub-limit one should have read sits. The non-obvious point joins the two: on a linear structure, neither the limit nor the sum insured decides the outcome. It is the definition of a loss on one side and the non-physical sub-limit on the other, two lines nobody negotiates because they carry no large figures.
- 01Dispersion protects against the local peril and not the extended one: a two-hundred-kilometer route is far more likely than a point to cross a dangerous zone.
- 02The profile inverts: frequency rises with length while unit severity falls, so attritional loss weighs more than the extreme event.
- 03A linear structure crosses what it does not own: hundreds of crossings, therefore a liability head proportionally far heavier than at equal value on a point.
- 04On a severed service, the physical damage is modest and the non-physical loss makes the amount: that is where the sub-limit sits, and that is what must have been read.
- 05The definition of a loss decides more than the limit: a clause requiring one PLACE and one cause turns a storm into nine losses and nine deductibles.