Concurrent delay: two delays at once, five different answers

An employer delay and a contractor delay land on the same three weeks. The facts do not change and neither do the dates, but every analysis method awards a different number of days.

Live result

The job finished 35 days late, and five methods split that overrun five different ways.

Days both sides were delaying25 days
Impacted as planned30 days
Time impact analysis30 days
As planned against as built17.5 days
Collapsed as built5 days
Windows5 days

What separates the five answers is the treatment of the days on which both sides were delaying.

The two delays

Try a scenario

Contractor delay

The contractor's delay starts on daydays

The contractor's delay runs fordays

Employer delay

The employer's delay starts on daydays

The employer's delay runs fordays

They overlap for: 25 days

The numbers below are what five methods do to one deliberately simple pair of delays. They do not decide entitlement and are not evidence for a real programme or contract.

What actually happened

Days only the contractor was delaying	5
Days only the employer was delaying	5
Days both were	25
Days late in total	35

What each method awards

Impacted as planned

Insert the employer delay into the baseline and measure the movement without considering the contractor delay.

Baseline finish: day 75
Insert the employer delay: 30 days
Projected movement: 30 days

Time impact analysis

Insert the employer event into the programme as it stood when that event began and measure the projected movement.

Programme when the employer event begins: day 25
Insert the employer event: 30 days
Projected completion movement: 30 days

As planned against as built

Give the employer-only days in full and share the concurrent days evenly.

Employer only days: 5 days
Concurrent days, shared evenly: 12.5 days
Award: 17.5 days

Collapsed as built

Remove the employer delay from the as-built record and measure the time saved.

As built finish: day 110
Remove the employer delay, contractor delay alone: day 105
The difference: 5 days

Windows

Split the overrun at every event boundary, give employer-only windows, and give concurrent windows to neither side.

Employer only windows: 5 days
Concurrent windows, awarded to neither: 25 days
Award: 5 days

Why the method is a choice

None of these five methods is wrong. Each answers a slightly different question: what the employer's delay would have done to the plan, what it did to the programme on the day, what the record shows the two of them did between them, and what would have happened without one of them.

The uncomfortable part is that the method is usually chosen after the outcome is known, by whichever side benefits from it, and then defended as though it had been the obvious one all along. A claim that does not name its method, say why that method suits these facts, and show the working is asking to be rejected, and it usually is.

The version above is deliberately simple: one chain, two events, no movement of the critical path. Real programmes have hundreds of activities and a critical path that changes hands several times, which makes the gap between these five numbers wider rather than narrower.

Here is the maths

The duration comparison holds overlap at 10 days, rework at 25%, extra crews at 2, each at 60% efficiency, and available crews at 2.

Model version: 1.0

The formula, in words

The baseline is a three activity chain. It starts on day 0 and finishes on day 75 under the shared critical path engine.
A delay window includes its start and excludes its end. The overlap is the part inside both windows. The total overrun is both window lengths with that overlap counted once.
$U=E+C-O$
where $U$ is the total overrun, $E$ the employer window, $C$ the contractor window, and $O$ their overlap.
Employer only days are the employer's window minus the overlap. Contractor only days are the contractor's window minus the overlap. The actual finish is day 75 plus the total overrun.
$E_0=E-O,\qquad C_0=C-O,\qquad F=75+U$ $\begin{gathered}E_0=E-O, \\ C_0=C-O, \\ F=75+U\end{gathered}$
where $E_0$ is employer only days, $C_0$ contractor only days, and $F$ the actual finish day.
Impacted as planned and time impact analysis each award the employer's full window here. On this single chain, neither looks at what the contractor was doing, so they agree.
As planned against as built adds the employer only days to half the overlap. Collapsed as built and windows each award the employer only days under the single rules used on this page.
$A=E_0+\frac{O}{2}$
where $A$ is the award from sharing the overlap equally.
When the events do not overlap, every method awards the employer's full delay. When they overlap, the difference between the longest and shortest award is exactly the length of that overlap.
These rules stand in for five families of practice. Variants inside each family also disagree. The method is part of the analysis, not a neutral wrapper around it.
A duration is elapsed time, and a finish is the moment the work ends. An activity of five days starting at day 0 finishes at day 5, not day 4. A finish to start successor starts at day 5.
This is the continuous convention rather than the inclusive day convention. It is chosen because every off by one error in a hand check of a critical path comes from the other one, and because both lesson pages ask readers to check the arithmetic themselves.
Overlap gains the days brought forward and gives back the share that must be redone. Adding crews divides the 40 day construction activity by one plus the effective extra crew contribution. Its crew time is the crews used over the shorter window minus the original 40 crew days.
$R=or,\qquad G=o-R$ $\begin{gathered}R=or, \\ G=o-R\end{gathered}$
where $R$ is rework days, $o$ days brought forward, $r$ the rework fraction, and $G$ the net gain.
$d=\frac{40}{1+n\eta},\qquad K=(1+n)d-40$ $\begin{gathered}d=\frac{40}{1+n\eta}, \\ K=(1+n)d-40\end{gathered}$
where $d$ is the shortened construction duration, $n$ extra crews, $\eta$ each extra crew’s efficiency, and $K$ extra crew days.
Durations are elapsed working days and every day counts. There is no calendar, no date and no imposed completion date.

A worked example

The plan was 20 days of enabling works, 40 of main construction and 15 of testing, so day 75.

20+40+15=75

The contractor's delay runs from day 20 to day 50. The employer's runs from day 25 to day 55. Between them, work stopped on every day from 20 to 55, which is 35 days, so the job finished on day 110. Of those 35 days, 5 were the contractor alone, 5 were the employer alone, and 25 were both at once.

30+30-25=35,\qquad 75+35=110

\begin{gathered}30+30-25=35, \\ 75+35=110\end{gathered}

Impacted as planned takes the employer's 30 day delay, drops it into the original programme and reports 30 days. It never looks at the contractor at all.

Collapsed as built starts from the actual finish on day 110, takes the employer's delay out of the record, and finds that the contractor's delay alone would still have finished on day 105. The employer's contribution is the difference, 5 days.

110-105=5

Sharing the concurrent period evenly gives the employer's 5 days plus half of the 25 concurrent days, which is 17.5 days.

5+\frac{25}{2}=17.5

Same programme. Same two delays. Same 35 days late. Thirty days, five days, or seventeen and a half, depending on which of these you open with.

Overlapping 10 days with 25 percent rework gives back 2.5 days, so the net gain is 7.5 days at 2.5 crew days.

10\times0.25=2.5,\qquad 10-2.5=7.5

\begin{gathered}10\times0.25=2.5, \\ 10-2.5=7.5\end{gathered}

Two extra crews at 60 percent each shorten the 40 day activity to 18.18 days, gaining 21.82 days at 14.55 crew days.

\frac{40}{1+2\times0.60}\approx18.18,\qquad 40-\frac{40}{2.2}\approx21.82,\qquad 3\times\frac{40}{2.2}-40\approx14.55

\begin{gathered}\frac{40}{1+2\times0.60}\approx18.18, \\ 40-\frac{40}{2.2}\approx21.82, \\ 3\times\frac{40}{2.2}-40\approx14.55\end{gathered}

What this does not model

A moving critical path, which is where real analyses diverge most; float and whether either delay consumed it; mitigation and acceleration beyond the two resource sketches shown here; notice, and whether it was given in time; the contract form, which decides more of this than the arithmetic does; money cost, prolongation, disruption and damages of any kind; whether either delay actually affected critical work, which is what most real disputes are about; and the variants inside each of these five method families, which disagree with each other as much as the families disagree.

Check it yourself

Enter day 55 as the start of the employer's delay. It no longer overlaps the contractor's, and all five methods award 30 days. Every difference on this page is the overlap and nothing else.

Two ways to buy the same days

same 75 day baseline

Overlapping 10 days of testing into construction gains 7.5 days once 2.5 days of rework is taken back, at 2.5 crew days of crew time. Adding 2 crews gains 21.82 days at 14.55 crew days of crew time, and creates no rework.

That profile asks for 3 crews for 18.2 days, against the 2 in the fixed example. The bar chart above does not know that, which is why this one is here.

It is not a delay analysis or an entitlement calculation. Each method here is a single stated rule standing in for a whole family of practices, real programmes have a critical path that moves and this one does not, and the question of which method suits a given set of facts is a matter for the contract and for a delay analyst. No number here belongs in a notice or a claim.