QSRA for UK Highways: A66 Northern Trans-Pennine Schedule Risk Analysis

A £1.3 billion highway upgrade connecting the M6 to the A1(M) across the Pennines is already facing a £280 million cost increase, and construction has barely started. The A66 Northern Trans-Pennine project was selected for the government's Project Speed programme in 2020, with an ambitious goal: halve the construction time and open the road five years early, by 2029. But accelerated timelines and complex ground conditions across some of England's most challenging terrain have made the schedule a critical risk in its own right.

Quantitative Schedule Risk Analysis (QSRA) is a statistical method that stress-tests project timelines by modelling the impact of uncertainties and discrete threats using Monte Carlo simulation. It replaces single-point deterministic schedules with probabilistic forecasts, producing a range of possible completion dates at defined confidence levels such as P50, P80, and P90. This gives decision-makers a defensible basis for setting schedule contingency reserves instead of relying on optimistic planning assumptions or arbitrary buffers.

For a highways mega-project like the A66, where archaeological constraints, environmental legal challenges, and compressed construction windows are compounding the schedule pressure, QSRA provides the only rigorous method to quantify how much contingency the programme actually needs.

Here is how QSRA applies to the A66 Northern Trans-Pennine project, step by step.

Why the A66 Northern Trans-Pennine Needs QSRA

The A66 upgrade is not a single corridor improvement. It involves multiple separate sections across 80km of terrain between Penrith and Scotch Corner, each with distinct ground conditions, environmental sensitivities, and access constraints. National Highways secured Development Consent in March 2024 after environmental campaigners challenged the scheme, adding four months of delay before a single excavator moved.

The Office of Rail and Road reported that the cost acceleration strategy itself drove a £280 million overspend. Compressing the programme to meet Project Speed targets meant overlapping work packages, increasing resource demand, and removing the float that would normally absorb delay. Without a QSRA, the team could not quantify the trade-off between acceleration risk and the schedule benefit being promised to ministers.

IQRM recommends that every National Highways scheme above £500 million undergoes a formal QSRA before the Development Consent Order stage. This ensures the schedule baseline presented to planning authorities reflects genuine uncertainty, not the optimism inherent in deterministic planning tools.

Schedule Integrity: The Foundation of Any QSRA

Before running a Monte Carlo simulation, the schedule must be dynamically responsive to change. A schedule locked with hard constraints, excessive lags, or open-ended logic will not respond realistically when the simulation engine varies activity durations.

For the A66, the Primavera P6 schedule would be imported into Safran Risk and subjected to a health check. Key issues to resolve on a highways programme include:

First, remove all "Finish No Later Than" constraints tied to the 2029 opening target. These constraints mask the real schedule exposure by locking dates that should be free to float during simulation.

Second, replace lags greater than 5 days with explicit dummy activities. Lags fixate activities and prevent uncertainties from propagating through the network logic.

Third, resolve open-ended logic. Every activity must have a predecessor and successor to maintain a continuous critical path through the model.

Risk Identification: What Threatens the A66 Schedule

QSRA separates risks into two categories: estimated uncertainties (business-as-usual variation in activity durations) and discrete risk events (specific threats with a defined probability of occurring).

For the A66, the estimated uncertainties would apply three-point estimates (minimum, most likely, maximum duration) to every major activity. Earthworks across the Pennines, for example, face significant geological variability. A cutting operation planned for 12 weeks might realistically take anywhere from 10 to 18 weeks depending on ground conditions encountered.

Discrete risk events on the A66 include:

Archaeological Discoveries (probability 40-60%). Preparatory archaeological investigations are already underway. Significant finds would trigger statutory protection requirements, potentially halting earthworks for weeks on individual sections.

Environmental Legal Challenge Delays (probability 20-30%). Campaigners have already challenged the DCO. Further judicial review applications during construction could suspend works on specific sections.

Winter Weather Windows (probability 100%, variable severity). The Pennine corridor experiences severe winter weather from November to March. Earthworks and surfacing operations are weather-sensitive, and the calendar risk model must reflect reduced productivity during these months.

Utilities Diversion Delays (probability 50-70%). Major highways schemes routinely discover uncharted utilities during excavation. Each diversion requires statutory utility company response times that the contractor cannot compress.

Probability Distributions and the Risk Data Engine

Selecting the right probability distribution for each risk variable is critical. IQRM's Risk Data Engine (RDE) methodology ensures that distribution selection is grounded in historical data, not guesswork.

For a UK highways project, National Highways holds extensive benchmark data on earthworks productivity, surfacing rates, and utility diversion timescales from previous schemes such as the A14 Cambridge to Huntingdon improvement and the M4 Junctions 3-12 Smart Motorway programme. This data can be fitted to distributions using statistical methods (AIC/SIC) to identify the best-fit shape.

IQRM recommends the BetaPERT distribution for activity duration uncertainties where three-point estimates are available but large datasets are not. The BetaPERT gives more weight to the most likely value than a simple triangular distribution, producing more realistic outcomes. For archaeological delay events, a Bernoulli distribution models the binary "occurs or doesn't" nature, combined with a triangular distribution for the impact duration if it does occur.

Schedule Contingency = P80 Completion Date - Deterministic Baseline Date

Running the Monte Carlo Simulation

With risks mapped to activities and distributions assigned, the Monte Carlo simulation runs 10,000 iterations. Each iteration randomly samples from every distribution, calculates the schedule network, and records the resulting completion date. The aggregate of all 10,000 outcomes produces the S-curve (cumulative distribution function) that shows the probability of completing by any given date.

For the A66, the simulation settings in Safran Risk would include Latin Hypercube Sampling for faster convergence, a locked random seed for reproducible results, and convergence monitoring to confirm the P80 value has stabilised within a 3% tolerance band.

Correlation is essential. If earthworks on Section 1 overrun due to poor ground, earthworks on adjacent Sections 2 and 3 are likely to face similar conditions. IQRM recommends a Pearson correlation coefficient of 0.6 to 0.8 between geologically similar sections. Without correlation, the model underestimates total programme duration because it assumes independent outcomes across activities that share the same risk driver.

Reading the QSRA Outputs: What the A66 S-Curve Would Reveal

The deterministic schedule for a project like the A66 typically falls between P5 and P20 on the S-curve. This means the planned completion date has only a 5% to 20% chance of being achieved. The gap between the deterministic date and the P80 date represents the schedule contingency the programme needs to absorb realistic risk exposure.

The tornado chart would rank the top schedule drivers. Based on the risk profile, the top five drivers would likely be: winter weather productivity loss across all sections, archaeological discovery delays in the Stainmore Pass area, utilities diversion delays on the Kemplay Bank to Temple Sowerby section, ground condition variability in the Cross Fell foothills, and the environmental legal challenge risk.

Pre-Mitigation vs Post-Mitigation: Justifying Investment

QSRA delivers its greatest value when the model is run twice: once with the current risk exposure (pre-mitigation), and once with proposed response strategies priced in (post-mitigation). The shift between the two S-curves quantifies the return on investment for each mitigation action.

For the A66, a mitigation scenario might include: pre-construction archaeological trial trenching on all high-risk sections (cost: £3M, expected schedule saving: 8 weeks), advance utility surveys using ground-penetrating radar (cost: £1.5M, expected saving: 6 weeks), and extended summer working hours to offset winter productivity losses (cost: £5M, expected saving: 10 weeks).

Overlaying the pre- and post-mitigation S-curves in Safran Risk shows whether the £9.5M mitigation investment genuinely shifts the P80 completion date. If the P80 moves forward by 4 months, and each month of delay costs the programme £8M in prolongation, the ROI is compelling: £9.5M spent to avoid £32M of delay cost.

Best Practices for Highways QSRA

IQRM's experience with UK infrastructure risk analysis highlights several practices that distinguish a credible QSRA from a box-ticking exercise.

Model the sections independently then correlate them. A multi-section highway programme is not one project. Each section has its own critical path, and the programme completion is governed by the slowest section. Model each section as a sub-network, then apply correlation between geologically or logistically linked sections.

Use calendar risks for weather, not productivity factors. Applying a blanket 20% productivity reduction in winter is crude. Calendar risk modelling in Safran Risk generates simulated non-working days directly into activity calendars, reflecting the actual pattern of weather disruption without double-counting.

Run the QSRA before the DCO application, not after. The schedule presented in a Development Consent Order application becomes a baseline against which the Secretary of State measures delivery. If that baseline is deterministic and optimistic, every stakeholder is set up for disappointment.

Present outputs in terms executives and ministers can act on. Do not overwhelm stakeholders with histograms. Translate QSRA outputs into narrative: "We are 20% likely to open on time without additional investment. With £9.5M in targeted mitigation, our confidence rises to 80%."

Frequently Asked Questions

What is QSRA for highways projects?

QSRA for highways projects is a Monte Carlo simulation-based method that models schedule uncertainty across all construction activities and risk events. It produces probabilistic completion dates at confidence levels like P50, P80, and P90, replacing single-point deterministic schedules with defensible, data-driven forecasts.

How does QSRA quantify schedule risk on multi-section highways?

Each highway section is modelled as a sub-network within the overall programme schedule. Three-point duration estimates are applied to activities, discrete risk events are mapped with their probabilities and impacts, and correlation links related sections. The Monte Carlo engine then simulates thousands of scenarios to identify the programme completion distribution.

What confidence level should highways projects use for planning?

IQRM recommends P80 as the standard confidence level for schedule contingency on major highways schemes. P80 means there is an 80% probability of completing by that date. P50 is too aggressive for public infrastructure commitments, while P90 may over-allocate contingency.

Why is correlation important in highways schedule risk models?

Highway sections sharing the same geological corridor, contractor, or supply chain will experience correlated outcomes. If ground conditions are worse than expected on one section, adjacent sections are likely to suffer the same issue. Without correlation, the model underestimates total programme duration because it treats linked variables as independent.

What software is used for highways QSRA?

Safran Risk is the industry-standard tool for QSRA on major UK highways projects. It imports native Primavera P6 schedules, performs schedule health checks, supports all required distribution types, and produces the S-curves, tornado charts, and critical path analysis outputs that National Highways requires for investment decisions.

When should a QSRA be performed on a highways project?

IQRM recommends running the first QSRA before the Development Consent Order application, when the schedule baseline is being set. Subsequent updates should be run at each project gateway, after significant scope changes, and when new risk information emerges from site investigations or contractor procurement.

Written by Rami Salem, Quantitative Risk Management specialist, 15+ years in oil and gas, EPC/EPCM, and infrastructure projects.