Marianne C Ramey
Project Advisors International, Fairfax Station

When delay experts reach different conclusions, tribunals often treat this as a binary choice between analyses. This article argues that it is more useful to ask why the analyses differ, examining what records each uses, what assumptions it draws from them and where its critical choices are made. Comparing as-planned versus as-built analysis in windows with time-slice analysis, it shows that how each method uses contemporaneous programme updates explains divergence more often than the method itself, and that time-slice analysis, properly built, offers greater transparency and reproducibility, helping tribunals to assess competing analyses on their merits.

Competing delay analyses

Tribunals in international construction arbitrations are regularly presented with competing expert forensic delay analyses that use different methods and reach different conclusions. The natural response is to treat that as a simple choice: accept one, reject the other. In some cases, a tribunal will ultimately have to choose between the experts’ conclusions to decide the dispute, but even then, understanding why the analyses differ makes that choice better informed. That choice is often unnecessary.

A more productive approach is to examine what each analysis shows: what records it uses, what assumptions it draws from them and which choices drive its conclusions. Each analysis can then be tested against the same project record rather than evaluated on the name of the method. In many disputes, parts of each analysis will be useful even where the overall conclusions differ.

Projects change as they proceed: priorities shift, access opens or closes, design is released incrementally, procurement slips, resources are redeployed and recovery programmes are introduced. When prepared with reasonable care, programme updates set out the current plan for the works at each stage, and are the primary source for how those changes were reflected in the sequence and the duration of the remaining works; progress reports, correspondence and instructions support why change occurred. To set the programme updates aside, or treat them as background rather than primary material, is to set aside the record of how the project’s history unfolded and the events that shaped those changes.

On large international projects, including those under FIDIC contract forms, programmes are not produced once and then set aside. They are updated throughout the works and used by the contractor, engineer and employer as an active management tool.

Programme updates record the sequence being used, the assumptions then in play and the activities the network logic identified as controlling completion at that point in the works. Controlling completion means the sequence of activities whose combined duration and logic determine the earliest date by which the project can finish (the critical path); a delay to any activity on that path extends the completion date by the same amount.

How analyses use the project records

The principal difference between competing delay analyses is not usually the methodology. It lies in how each analysis uses the project records, specifically, which documents are relied upon, how the programmes and their revisions are treated, and what assumptions are drawn from the record as a whole. Two experts who use the same method can still reach substantially different conclusions if, for example, one validates the baseline programme logic before measuring departures from it, while the other applies the method to an unvalidated baseline.

A tribunal assessing competing analyses may find it useful to look not only at which method each expert uses but also at how each uses the project’s programmes and records, and what each expert draws from those records in identifying what was controlling completion in each period.

Why contemporaneous programmes matter

A project programme is a working tool for project execution, and, as it is updated through the course of a project, a contemporaneous record of how the work was planned and revised as the works progressed. It sets out all planned work sequences through to completion; identifies which sequence is critical; and is updated as the works progress to monitor performance, evaluate change and direct resources. Programme updates are valuable for analysis not because each update matters on its own, but because, together, they show when the sequence of the works changed, when planning for a new phase became more detailed and when the activities controlling completion changed. This is particularly true on projects managed through rolling programmes: programmes that are updated in successive phases as later work is planned in more detail. As the works advance, the programme for the next phase gains detail that earlier versions did not carry. That additional detail does not invalidate earlier programmes; it reflects the normal development of project planning as information becomes available and work sequences firm up. Not every update is analytically significant. Some record routine progress. Others capture a substantive shift – a major resequencing of work, a strategy to complete certain systems first, or a shift in which area of work was driving the end date – that changes which activities are controlling completion. Those substantive shifts are the ones a sound analysis will identify and explain.

Programme updates require scrutiny. Some are optimistic. Some contain logic errors. Some reflect claims pressure rather than actual project conditions. Those deficiencies call for expert review and correction, not outright rejection. If an update was being used to plan and execute the works, it records the planned sequence of work at the time, and that record remains useful despite its imperfections. An analyst who discards a contemporaneous update without accounting for how it was being used loses a primary source of project history.

A sound analysis also distinguishes between planning assumptions that genuinely reflect contemporaneous project management and later adjustments introduced without support in the project record. Planning assumptions recorded at the time remain analytically relevant – even where later events exposed their limits – because they explain the context in which decisions were made and resources were committed. A tribunal can apply this distinction when reading each expert’s report: is the assumption grounded in what the project record shows about how the works were being planned and executed at the time, or is it a later reconstruction?

The two methods: how each identifies criticality

The two methods most commonly applied in large-project arbitrations are: (1) as-planned versus as-built analysis in windows; and (2) time-slice analysis. Both examine delay across successive periods, identify what was controlling completion in each period and require the analyst to account for how the remaining works changed over time. The methods differ in how each identifies criticality.

An as-planned versus as-built analysis in windows uses the baseline programme’s late dates as the reference. The analyst measures actual progress against those late dates, which typically shows variance – the difference between planned and actual progress – on more than one activity in a given window. Identifying which of those variances controlled completion, rather than merely accumulated alongside it, is a judgment: the Society of Construction Law (SCL) Protocol describes this method as relying on ‘common-sense and practical analysis of the available facts’,¹ acknowledging that the conclusion depends on the analyst’s reasoning rather than a network calculation of float across all activities. This approach becomes unreliable where the remaining works are no longer following the baseline sequence because a large variance against the baseline does not establish that the activity was still controlling completion – it may simply have accumulated delay while a different sequence became the longest path to completion.

A time-slice analysis does not simply take contemporaneous contractor programme updates and run them as the analysis. Contractor updates vary widely in quality, logic integrity and analytical purpose – some are submitted to support claims; some are optimistic; and some contain open ends (activities with no logical successor in the network) or superseded logic (links between activities that no longer reflect how the work was executed). The analyst instead builds analysis programmes: schedules derived from the contemporaneous updates, but reviewed and corrected before being used as the basis for the analysis.

Holding the logic unchanged except where the project record supports a change is what distinguishes a time-slice analysis from a simple replay of contractor submissions. The analyst uses the contemporaneous updates as primary source material for understanding the sequence, logic and project team’s planning at each stage – but applies judgment to identify which changes reflect actual shifts in work approach, which reflect errors in the schedule or claims strategy and which reflect work that was executed out of sequence – that is, in a different order from the one planned. The critical path emerges from a network calculation applied simultaneously to all activity paths.

In both methods, the analyst’s choices – which records to use, which logic changes to accept and which sequence to treat as critical – determine the result. It is those choices, and the project record that supports or undermines them, that bear most directly on the reliability of each analysis.

Both methods require analytical judgment

It is common for one analysis to be characterised as factual and another as theoretical or model-based – particularly when an as-planned versus as-built approach is contrasted with a time-slice analysis. That characterisation is analytically unsound. Both methods examine the same project history through the same body of records, and in both, the analyst’s judgments about sequence, logic and what was controlling completion shape the result. Understanding what those judgments are and whether they are justified is more informative than treating one method as inherently more objective than the other.

An as-planned versus as-built analysis in windows compares planned and actual progress. Once it moves beyond describing that comparison and identifies what was driving delay, it is making an analytical judgment – no more purely factual than any other form of delay analysis. A time-slice analysis constructs analysis programmes from the project record and uses them to calculate criticality at successive points in time. Where those programmes are built from contemporaneous updates that were the tools through which the project was planned and managed, they are grounded in the project record. Neither method is exempt from judgment.

Transparency and reproducibility

In practice, the distinction between the two methods that bears most directly on how competing analyses can be evaluated is the combination of transparency and reproducibility – whether the basis for each critical path conclusion can be independently examined and tested. In an as-planned versus as-built analysis, that basis is the analyst’s judgment about which variance controlled completion in a given window. Analysts frequently describe this reasoning as common sense, but common sense is not a uniform standard. One analyst’s obvious critical path is another’s unsupported assertion. Whether the late delivery of structural steel controlled completion over a six-month window, rather than a concurrent shortage of formwork crews, is not obvious – it is a judgment, and the basis for it should be stated and capable of being checked.

Because the reasoning must be articulated step by step across each window, as-planned versus as-built analyses tend to produce reports that are very long, though that length does not always translate into transparency. Reports of 400 to 600 pages or more, supported by dozens of spreadsheets and progress curves, are common. The length reflects the effort required to justify each window’s conclusions in prose because the analysis itself does not produce a record of the analytical decisions that stands on its own. The spreadsheets and curves support the narrative, but the narrative carries the analysis. Despite this, the basis for the critical judgments within it is often only partially explained.

A time-slice analysis produces its results within the analysis programmes themselves. The network logic, the critical path calculation and the adjustments made at each window are embedded in the programme files, and can be opened in the scheduling software and tested by an analyst. A well-constructed time-slice analysis can typically be presented in a report of around 100 pages, with the analytical programmes and a targeted set of supporting schedules and curves carrying the detail. The tribunal can see not just the conclusion but the structure that produced it. The expert’s role shifts from narrating a series of judgment calls to explaining the analytical programmes – what changed between windows, why the project record supports those changes and what significance those changes carry for the critical path. That is precisely the form of explanation that allows a tribunal to engage with the analysis directly rather than simply accept the conclusions.

Where the Protocol fits

The SCL Delay and Disruption Protocol identifies recognised delay analysis methods and provides short descriptions of each. Those descriptions can create a false impression: that an as-planned versus as-built analysis in windows is a more practical, records-based exercise, while a time-slice analysis is software-driven and more removed from the project facts. The opposite is closer to the truth.

A time-slice analysis is built directly from the contemporaneous programme updates, which are reviewed, corrected and used as the primary analytical source; the network calculation is applied to that material, not substituted for it. An as-planned versus as-built analysis, by contrast, relies on the baseline programme mainly for its late dates and depends less on the contemporaneous updates that recorded how the sequence actually changed as the works progressed – it measures variance against a baseline that may no longer reflect the actual sequence of work, and fills the gap between that baseline and the as-built record with the analyst’s judgment rather than a network-derived critical path.

A named methodology is a framework: a structure for organising the analysis of project information. Its reliability depends entirely on how it is applied to the specific records that exist for a given project. A method does not become more reliable by having a recognised name, and it does not become less reliable by being less familiar. A debate about which method the Protocol prefers is less informative than an explanation of how each analysis uses the project record and what the project record actually shows.

Where contemporaneous programmes and progress records are comprehensive and well-maintained, they provide the primary basis for understanding when and why completion was delayed. Where records are sparse, inconsistent or demonstrably unreliable, the choice of method needs to reflect what the available records can actually support. A time-slice analysis depends on programme updates of sufficient quality to build reliable analysis programmes; where those updates are absent or cannot be relied upon, an as-planned versus as-built approach may be the more appropriate basis. The report should clearly state which records were unavailable or disregarded and why that affected the choice of method.

What was controlling completion?

Both methods address the same underlying question: what was controlling completion in the relevant period? But they reach that answer in different ways and understanding that difference is important when a tribunal reads each expert’s critical path conclusions.

In a time-slice analysis, the analyst holds the planned logic steady at the start of each window and advances actual progress through it. The logic is updated only at the start or end of a window where the project record documents a change in how the work was being approached: a mitigation measure, a revised sequence, a formal instruction, or a management decision recorded in the project’s correspondence or meeting records. The analyst adjusts for work already performed out of sequence because that is a fact but does not change the forward programme logic based on what the contractor later submitted, unless the reason for that change is independently recorded in the project record.

An as-planned versus as-built analysis identifies the activity or sequence whose departure from the baseline was most responsible for delaying completion in each window. It anchors criticality in variance from the original plan, period by period, rather than in a network calculation of the works as they were then being planned and executed. That distinction allows a tribunal to identify, in each window, whether the expert’s critical path finding rests on a network calculation applied to the analysis programme, with float calculated across all activity paths, or on the analyst’s judgment about which variances were most significant.

An as-planned versus as-built analysis will typically identify one primary as-built departure from the baseline as the delay driver for each window, even where the underlying critical path shifted during the period. A time-slice analysis identifies those shifts more directly by recalculating the network from the analysis programme at the start of each window. Which approach better reflects the project’s actual history depends on the quality and completeness of the records available and on how the project was planned and managed.

What matters in practice?

When experts present different methods and different conclusions, the questions that tend to be most productive are not about methodology. They are: what does each analysis show about what delayed completion, when it occurred and by how much; how well does that account fit the way the project was actually planned, managed and executed; and what assumptions does it rest on? Those questions can be applied to each analysis individually and used to identify where the analyses are aligned, where they diverge and what causes the differences between them.

In cases where the experts have been unable to agree on what their methodological differences mean for the identification and quantification of the project delay, the tribunal is not required to choose one methodology wholesale and discard the other. A tribunal can, in that situation, look at which analysis is better supported by the project record for a given window or issue, draw on the parts of each analysis where that support is strongest and weigh each expert’s critical path conclusions against the documents that were governing the project at the time. The more useful questions are whether the critical path conclusion follows from the programme data and records relied upon, and whether the reasoning that connects the data to the conclusion is explained clearly enough to be examined.

Methodology matters, but a recognised method does not become more reliable by having a familiar name. Its reliability depends on how carefully it is applied to the specific project record. Gaps in that record do not automatically invalidate an analysis – they require the analyst to state clearly what can and cannot be established from the available records. Expert analyses constructed on that basis – using the contemporaneous record as the primary source, explaining what controlled completion in each period and stating the basis for each analytical decision – put a tribunal in the best position to determine entitlement to an extension of time, whether or not the experts agree.

¹ Society of Construction Law, Delay and Disruption Protocol (2nd edn, SCL 2017) Guidance Part B, para 11.6(d).

Marianne C Ramey is the President of Project Advisors International in Fairfax Station, Virginia, United States, and can be contacted at ramey@paiworld.com.