Prolongation and disruption: delay costs money twice

Model version

1.0

The two measurements, in words

• Prolongation is a period, not a multiplication. The question is which resources were held longer and when, so it is taken from the cost records of the period the delay took effect; multiplying every monthly cost line by the length of the extension is the error the whole page is about.
• Disruption is a rate, not a total. Compare what was produced per hour on affected work against comparable unaffected work, convert the difference to hours, and price the hours.
• The comparison must be with the same crew, not with the tender. A tender is an estimate, and the gap between an estimate and what a crew ever achieved is not something an event caused.
• The two never merge. The same hour cannot sit in both, and where a rate for changed work already carries supervision or overhead, that content is already recovered once.

A worked example

Prolongation, and the month it is taken from. Daily rates come from the month, so September runs at $\frac{102{,}000}{30}=\$3{,}400$ a day and November at $\frac{54{,}000}{30}=\$1{,}800$. Six weeks is 42 days. Where the delay took effect, from 1 September: 30 days of September at $3,400 is $102,000, plus 12 days of October at $\frac{66{,}000}{31}=\$2{,}129.03$ is $25,548.39, so $127,548.39. In the overrun window, from 1 November: 30 days of November at $1,800 is $54,000, plus 12 days of December at $\frac{46{,}000}{31}=\$1{,}483.87$ is $17,806.45, so $71,806.45. The same six weeks, $55,741.94 apart, a factor of 1.78, and nothing about the delay changed between the two lines.

March is the row that stops this being an argument for one method: on a slow start the overrun window overstates. The point is not that one window is bigger, it is that the two differ and the difference is decided by when the programme was spending.

Disruption, and the comparison that is not a method. 2,400 units of unchanged work. The crew's own unimpacted rate, 1,800 units in 900 hours, is 2.00 an hour, so it should have taken 1,200 hours. It took 1,600, a 25% fall in output rate and a 33.3% rise in hours, the two ways of saying it that do not sound like the same number. Against the crew's own rate: $1{,}600 - 1{,}200 = 400$ hours, at $58, $23,200. Against the tender's 2.50 an hour, which allowed $2{,}400 \div 2.50 = 960$ hours: $1{,}600 - 960 = 640$ hours, $37,120. The difference is 240 hours, $13,920, and it is not disruption. It is the distance between what was tendered and what this crew ever achieved on a good day, and it belongs where it fell.

The measured comparison goes to zero, correctly, because nothing was lost. The tender comparison still reads $13,920. The gap remains 240 hours: a method that reports a loss when there was no event is not measuring the event.

Where the method descriptions come from

The method descriptions are paraphrased from the Society of Construction Law Delay and Disruption Protocol, AACE International practitioner guidance, and secondary practitioner commentary summarised for this lesson. They are not quotations. Several source positions are one author's argument rather than settled practice, and the page does not present the measured mile and direct time-lost records as a complete list of methods.

What this does not model

Entitlement of any kind, notice, contractual routes, concurrency between named events, float ownership, acceleration, escalation, head-office overhead and its formulae, the comparability tests a real measured mile has to pass, the exclusion of competing causes, and every question about whose delay it was, which is the one this page most deliberately never asks.