Concurrent delay: the authority map

This page is not legal advice and does not state any jurisdiction's answer. It sets out what four published documents say about concurrent delay, where they disagree, and what this tool does with that disagreement — which is to report every reading and decline to pick one. Picking one is a determination that belongs to a person, and on US work no document at the head of the precedence makes it.

Most pages on this subject give you a confident summary. The summary is the part that is wrong, because the sources do not agree and the disagreement is the whole claim.

Concepts §5 publishes the two axes and the worked examples — the interval knob and the doctrine knob, each shown changing the day count on identical facts. This page extends that section rather than restating it: what each authority actually sets out, where the texts diverge, and what a reader is entitled to check.

Two axes, not one

"Were the delays concurrent?" is two questions wearing one coat.

They are independent. A dispute can turn on either, and the same facts can produce four different day counts depending on which pair you take. That is not an artefact of anyone's software; it is what the source documents leave open.

A third thing sits behind both and is not on either axis: apportionment — splitting the overlap between the parties rather than allocating it. One of the four documents below prefers it. It is neither concurrent-effect nor first-in-time, and treating it as a third value on axis 2 would misdescribe it.

Axis 1: literal and functional

Under the literal theory the delays must occur at the same time in the strict sense. It produces fewer concurrency findings.

Under the functional theory they need only fall within the same analysis period — or, in one document's framing, their effects need only be felt on the critical path at the same time, even where the events themselves arose weeks apart. It produces more.

The two converge as the analysis interval shrinks toward a single day, and diverge sharply at monthly granularity. Concepts §5 shows the arithmetic: the same two delay periods, three days of overlap under the literal theory at any interval, thirty days under the functional theory when the window is a month. One setting, an order-of-magnitude difference, and it is a property of the theory rather than an error in the count.

One caveat this project has not resolved, and states rather than smooths over. AACE RP 29R-03 frames the literal/functional distinction in terms of interval granularity. The SCL Protocol frames its equivalent distinction as events-versus-effects. This tool treats them as one axis because on the same facts they select the same two answers — but they are not textually the same distinction, and no source equating them has been found. A report that turns on the difference should cite the paragraph, not this tool.

Axis 2: concurrent effect and first in time

Take an employer risk event that begins after a contractor delay to completion is already running, and continues alongside it.

Concurrent effect. Older English appellate authority: both are effective causes over the overlap, because each would independently have caused delay in the absence of the other.

First in time. Adopted by the 2017 SCL Protocol at ¶10.10 on the strength of more recent lower-court decisions: only the contractor's delay is an effective cause, because the works were going to complete late regardless, so the employer event causes no incremental delay to completion — and on that reading there is no concurrency at all.

Concepts §5's second worked example is ten days of concurrent delay under one and zero under the other, on identical dates.

Concurrency as a fact, and concurrency as a consequence

These get run together constantly and they are different objects.

As a factual finding, concurrency is a statement about overlap: two delays, each independently affecting the critical path, over these dates. The arithmetic of the overlap is computable. Whether the overlap amounts to concurrency in the sense a tribunal recognises depends on which theory and which doctrine apply, and that is a choice, not a computation.

As an entitlement consequence, concurrency is a statement about what follows — extension of time, cost recovery, liquidated damages. That step is governed by the contract and the applicable law, and it is expressly outside what this tool produces.

Two axes of entitlement, not one, and collapsing them is a common and expensive error. The SCL Protocol is explicit that extension-of-time entitlement and compensation entitlement are independent, and its own core principles make the asymmetry operative: once concurrency is established the contractor's own concurrent delay does not reduce the extension otherwise due (CP10, ¶10.12, resting on the prevention principle at ¶10.16), while on the money side the contractor recovers only what it can separate from costs it would have incurred anyway from its own delay (CP14, ¶14.3). Same finding, two downstream calculations, opposite generosity. This tool's entitlement object refuses at construction to be compensable and non-excusable at once, which is a consistency guard on an input a person supplies, not a finding of its own.

What each authority sets out

The UFGS section is a US Government work and can be quoted directly. §3.8.1:

With multiple impacts consider any concurrency of delay.

That is a duty to consider concurrency with no definition and no resolution rule. The section then points at another document. §3.8:

ASCE 67-17 provides delay analysis guidelines. If there is a conflict between the contract and guidance provided in ASCE 67-17, the contract will govern.

And §3.8.4:

If a methodology is chosen from AACE 29R-03, the method must adhere to the principles identified in ASCE 67-17. If there is a conflict with the methodology chosen from AACE 29R-03 and ASCE 67-17, ASCE 67-17 will govern.

So the federal specification requires concurrency to be considered and commits the definition to a document that does not supply one — and the document that one defers to, in turn, endorses neither theory.

The two places the documents openly conflict

On what a concurrency finding does. One document prescribes an outcome (first-in-time), one sets out both readings without choosing, and a third prefers apportionment, which is neither. The rules here that touch concurrency carry that divergence into the finding rather than each asserting a house answer.

On pacing, and the notice question. A contractor who slows non-critical work in response to an employer delay is pacing, not delaying — the difference between a concurrent contractor delay and no contractor delay at all. AACE §4.2.G takes the position that contemporaneous notices of an intention to pace are exceedingly rare and must not be treated as a strict proof requirement; the SCL Protocol at ¶15.2 recommends giving one. They point opposite ways for a claimant with no notice in the file, so a rule that fails a pacing claim for want of a notice has silently chosen one. This one does not, and says so in the finding.

Of AACE §4.2.G's three tests, only the first — that a parent delay exists and precedes the pacing — is computable from a network. Contemporaneous ability to resume and contemporaneous evidence of intent are facts about the project record, so a pacing result here is never marked deterministic.

What the tool does

construct concurrency FILE --delays periods.json reports the answer under all four combinations of the two theories and the two doctrines, with the day count for each and the spread between them, and then declines to select.

Precisely:

• Under UK / SCL, the SCL Protocol sits at the head of the precedence and has an encoded position, so the governing reading is the functional / first-in-time one. Where another combination gives a different number, that divergence is recorded and both numbers are named.
• Under US federal, US state or private and neutral, the tool returns no governing answer, and says which document declined and why.

That None is the answer, not a missing feature, and the reason is a reading of the texts rather than an assumption about them. On the federal path the head of the precedence is UFGS, which requires concurrency to be considered and defines nothing; it defers to ASCE 67-17, which defines concurrency and selects no theory; below that sits AACE RP 29R-03, which sets out both and endorses neither. Three documents read, three abstentions. A tool that picked one to avoid returning nothing would be manufacturing exactly the certainty that gets an expert excluded rather than merely disagreed with.

What a US reader gets instead of a verdict: the specification's duty to consider concurrency and the two clauses by which it defers; the definition the standard it defers to supplies and the apportionment preference beside it; the statement that neither selects a theory; all four readings with their day counts and the spread between them; the open questions; and the route that would settle it.

Two further things the tool never computes, for the same reason:

• Float ownership. The authorities give different defaults for network float and for project float, and one of them makes the answer a contract question outright. The tool raises the question and does not answer it. See float ownership.
• Whose delay it was. The party is carried as an input you supply and is never inferred. If you do not know whose a period was, undetermined is a real answer: it keeps that period out of every pairing and lists it as an open question rather than filing it against somebody.

A unit warning worth repeating

Every day figure in the concurrency module counts calendar dates in an overlap window, inclusive of both ends — the right axis for "over which dates were both parties in critical delay" and the wrong axis for entitlement arithmetic, where a modelled delay counts working days on a named calendar. The two must never be added to or subtracted from one another.

One of the documents above also cautions that delay analysis is rarely precise to the day, which is a reason to treat a near-miss overlap as a question rather than as a bright line either way.

What would settle it for your project

In roughly the order they decide the question:

1. A contract clause defining concurrent delay, or defining what happens when employer and contractor delay coincide. It outranks every document here.
2. The governing law, and any authority in it on concurrency — the reason this page is an authority map rather than an answer.
3. An agreement between the parties on the theory, which AACE RP 29R-03 §4.2.A asks for where possible.
4. The analysis interval, stated explicitly: at monthly granularity the functional theory finds concurrency the literal theory does not.
5. The contemporaneous record for pacing. Neither of its two judgment tests is in a schedule file.

Related

• Concepts §5 — the two axes, the worked examples, and what the tool deliberately never computes
• Standards crosswalk — which documents say what, side by side
• Delay method comparison — the AACE RP 29R-03 methods and what each one needs
• Float ownership
• What it does — the commands, and which need a file you write
• Limitations

Source: web/pages/concurrent-delay.md. Source commit date: 2026-09-11.