In March 2026, the UK's High Speed 2 rail project faced a watershed moment. Originally budgeted at £20 billion with a target completion in 2025, HS2 has escalated to potentially £100 billion, with completion now pushed into the late 2030s. Over £43.6 billion already spent by February 2026. Civil engineering work is running out of sequence and behind schedule. Speed reduction is being considered to save billions more. How did one of Europe's largest infrastructure projects lose control so dramatically? The answer lies in a methodology that should have been deployed from day one: Quantitative Schedule Risk Analysis.
Quantitative Schedule Risk Analysis (QSRA) is a Monte Carlo-based methodology that models schedule uncertainty across all project phases, integrating probability distributions, correlations, and criticality indices to generate P-values ranging from P10 to P90. Unlike deterministic schedules, QSRA reveals not just one completion date, but a probability curve of likely outcomes—essential for mega-project governance in the UK rail, nuclear, and infrastructure sectors.
The value proposition is stark: QSRA would have exposed HS2's schedule risk profile years before 2026. Where deterministic planning showed a P6 date (the 6th percentile—a best case so extreme it borders on fantasy), QSRA would have highlighted the P50 and P80 dates that decision-makers desperately needed. Those gap metrics alone would have triggered governance escalation, contingency planning, and scope adjustment before £43.6 billion was committed.
This case study demonstrates why QSRA is not optional for UK mega-projects: it is foundational risk intelligence. The rail infrastructure community, regulators, and treasury teams now understand that deterministic schedules are liabilities, not assets. HS2's crisis is the perfect crucible for understanding how to apply QSRA rigorously from project inception.
What QSRA Would Have Revealed on HS2
HS2's deterministic baseline promised completion by 2025. That date was a P6—an outlier representing only the 6th percentile of probability. Real-world analysis, had QSRA been deployed, would have produced a radically different forecast. Consider the physics: civil engineering out of sequence means activities cannot run in parallel. Schedule compression is impossible. Interdependencies multiply. Risk of delay cascades through phases.
A rigorous QSRA model would have modeled:
P50 date (50th percentile): 2032-2035, depending on risk correlation coefficients and criticality weighting of tunnel boring and station construction. This is the median outcome—equally likely to be beaten or exceeded.
P80 date (80th percentile): Late 2038, representing the schedule buffer needed to achieve 80% confidence in delivery. The difference between P6 (2025) and P80 (2038) is 13 years of hidden contingency.
The schedule buffer formula:
Schedule Buffer = P80 Date − Deterministic (P6) Date
= 2038 − 2025 = 13 years
That 13-year gap is the smoking gun. It reveals that deterministic planning was starving governance of critical reality-based intelligence. Regulators and parliament were sold a P6 outcome and treated as if it were baseline truth. When civil engineering fell behind in 2023-2024, the shock was not shock at all—it was statistical inevitability finally materializing.
The Seven Phases of QSRA Applied to Rail Mega-Projects
QSRA is not a black-box exercise. It is a structured seven-phase methodology, and HS2 illustrates why each phase is non-negotiable:
Phase 1: Schedule Definition. Decompose the project into work breakdown structure (WBS) elements, activities, and milestones. HS2 spans 14 major geographic sections, each with tunnelling, civil works, station construction, systems integration, and commissioning. Deterministic planning compressed this into a linear Gantt chart. QSRA requires full activity-level granularity.
Phase 2: Uncertainty Quantification. Assign probability distributions to activity durations. Tunnelling tasks are triangular or beta-distributed (best case, most likely, worst case). Station construction has lognormal tails (high upside risk). Land acquisition has discrete uncertainty. By 2024, HS2's tunnelling was running at 115% of planned duration—a distribution tail that should have been flagged in 2018.
Phase 3: Correlation and Dependency Modeling. Mega-projects have correlated risks. Weather impacts multiple sections simultaneously. Regulatory delays cascade. Supply chain constraints affect all civil contracts. QSRA captures these correlations via correlation matrices; deterministic planning ignores them entirely.
Phase 4: Monte Carlo Simulation. Run 10,000 iterations of the project schedule, sampling from probability distributions. Each iteration produces a completion date. The distribution of 10,000 dates yields the S-curve: P10, P25, P50, P75, P90.
Phase 5: Criticality Analysis. Identify which activities, if delayed, have the highest impact on the finish date. HS2's tunnelling sections (particularly Chilterns and Pennines) are high-criticality nodes. A 6-month delay there cascades across dependent activities. Criticality indices guide risk mitigation investment.
Phase 6: Sensitivity and Tornado Analysis. Rank risk drivers by their impact on the schedule. For HS2, the top drivers would be: (1) tunnelling productivity, (2) environmental approvals, (3) land acquisition/legal challenge delays, (4) labour availability, (5) supply chain for signalling systems. A tornado chart makes this transparent.
Phase 7: Governance and Contingency Planning. Use QSRA outputs to define schedule baselines, contingency buffers, and escalation triggers. If P50 is 2033 and P80 is 2038, the governance question becomes: at what confidence level should we plan? P50 assumes no margin for error. P80 assumes 1-in-5 odds of still being late. Parliament and Treasury should set this threshold explicitly.
Why Deterministic Schedules Fail on UK Infrastructure
HS2's collapse reflects a systemic flaw in how deterministic (single-point estimate) schedules are constructed. They assume:
All estimates are most-likely durations. In reality, estimates are typically best-case or aggressive. Tunnelling in chalk marl (Chilterns) was estimated at a baseline that assumed ideal geology. Actual geology showed fissures and groundwater. Result: 18-month delays in some sections.
No correlation between risks. If weather delays one section, deterministic assumes it does not delay others. Obviously false. A wet winter affects the entire UK project simultaneously.
Criticality is uniform. Deterministic scheduling treats all activities as equally important. In fact, activities on the critical path have disproportionate impact. Activities with float (slack) can slip 6 months with no effect on finish date. Deterministic planning wastes contingency trying to protect low-impact activities.
The P5-P20 Gap: Deterministic schedules typically represent the P5 to P20 range—aggressive but not impossible. Civil engineering best-practice assumes P10-P15 for baseline planning. Yet governance reports HS2's deterministic plan as if it were P50 (median). When reality hits P50, schedules look catastrophically late. They are not. They were simply misrepresented from the start.
Comparison: Deterministic vs. QSRA Approach
| Aspect | Deterministic (HS2 Historical) | QSRA (What Should Have Happened) |
|---|---|---|
| Completion Date | 2025 (single point, P6) | P50: 2033, P80: 2038 |
| Activity Durations | Single estimate per task | Probability distributions (min, likely, max) |
| Risk Treatment | Ignored in schedule, added as contingency buffer externally | Integrated throughout model via Monte Carlo |
| Correlation Modeling | None | Full correlation matrices for dependent risks |
| Critical Path | Single fixed path | Path varies by iteration; criticality indices calculated |
| Schedule Buffer | £5-10B generic contingency, poorly allocated | P80-P50 gap (13 years) drives targeted buffers |
| Governance Output | 2025 target (false confidence) | P50/P80 range with risk drivers identified |
| Cost of Delay Recognition | Shock in 2023-2024 when slippage emerged | Anticipated by 2018; proactive mitigation triggered |
How QSRA Identifies Schedule Risk Drivers
QSRA does not just produce a P80 date. It diagnoses why delays occur. Sensitivity analysis and tornado charts rank risk drivers by their impact on schedule variance.
For HS2, a realistic tornado chart would rank:
1. Tunnelling productivity and geology (highest impact): Chalk marl fissures, groundwater intrusion, and unexpected geology add 8–18 months per section. Impact on total schedule: ±24 months.
2. Environmental approvals and legal challenges: Statutory consultation, judicial review, and parliamentary scrutiny have delayed decisions by 12–36 months. Impact: ±18 months.
3. Land acquisition and compensation: Property settlements and compulsory purchase orders are slow-moving. Impact: ±12 months.
4. Supply chain for rolling stock and signalling: Long lead times for train procurement and CBTC (communications-based train control) systems. Impact: ±10 months.
5. Labour and skilled workforce availability: UK construction labour shortage affects productivity rates. Impact: ±8 months.
6. Regulatory and safety compliance: Testing, certification, and ORR (Office of Rail and Road) approvals. Impact: ±6 months.
A tornado chart makes this hierarchy visible. Governance can then focus contingency and risk mitigation on the top three drivers, where effort yields the greatest return.
Real Lessons from HS2 for Future UK Projects
HS2 is not unique. Sizewell C (nuclear), Network Rail Programme (renewals), HS2 Northern extension, and the Northern Powerhouse Rail project all carry similar schedule risk structures. The lessons apply universally:
Deploy QSRA at project inception, not post-facto. QSRA is most valuable during conceptual and pre-development phases, when schedule assumptions are still malleable. By 2020, HS2's baseline was locked in; QSRA analysis then could only explain slippage, not prevent it.
Distinguish between P-values and baselines. Deterministic baselines should represent P15–P25, not P6. P50 should be the governance forecast. P80 should drive contingency. Make this language standard in Treasury submissions and parliamentary briefings.
Integrate QSRA with cost risk analysis (QCRA). Schedule delays compound cost overruns. A 13-year P80 delay on HS2 implies years of inflation, extended management overhead, and financing cost. QSRA-QCRA integration reveals the true economic risk, not just schedule risk.
Model supply-chain constraints as correlated risks. The post-COVID and post-Brexit UK construction environment is resource-constrained. Labour, steel, concrete, and specialist equipment are scarce. Deterministic schedules that assume unlimited resource availability are fantasy. QSRA must model supply-chain probability distributions.
Embed criticality analysis in scope management. If tunnelling is the critical driver, focus engineering effort there. Invest in geological pre-investigation, alternative boring strategies, and supplier redundancy. Lower-criticality activities (non-critical station fit-out, for example) can slip without affecting finish date.
Best Practices for QSRA on UK Mega-Projects
Use activity-level granularity: Never model a 5-year civil package as a single activity. Decompose to 100+ activities minimum. HS2's 14 sections each require 50+ activities to capture realistic uncertainty.
Engage subject-matter experts for distribution estimation: Tunnelling engineers, environmental consultants, and procurement specialists must provide min/likely/max estimates. Do not allow single-point forecasts. Use three-point estimation (PERT) to derive beta distributions.
Validate assumptions with historical data: UK rail and nuclear projects provide benchmarks. Tunnelling productivity rates, approval timelines, and supply-chain delays are historically quantifiable. Use empirical data, not wishful thinking.
Run sensitivity analysis on correlation coefficients: Correlation strength is often unknown. Run multiple scenarios (low correlation, medium, high) to bracket the range of P80 outcomes. If P80 ranges from 2035 to 2039 depending on assumptions, governance knows the range of uncertainty.
Update QSRA quarterly or semi-annually: As project execution progresses, actual activity durations should be backfit into the model. If tunnelling is running at 115% of planned duration, update the distribution for remaining tunnel sections. QSRA is not static; it is a living risk intelligence system.
Communicate P-values clearly to governance: Avoid jargon. Say: "Our P50 forecast is 2033 with a P80 contingency buffer to 2038. We recommend planning for P80 (2038) to achieve 80% confidence in delivery." Parliament and regulators understand this language.
Frequently Asked Questions
What is the difference between P50 and P80?
P50 is the median outcome: equally likely to be beaten or exceeded. P80 is the 80th percentile: only a 1-in-5 chance of running later. For risk-averse governance (nuclear, rail), P80 is the planning baseline. P50 is the realistic forecast.
Why does HS2 use a P6 date instead of P50?
Original planning was optimistic. Stakeholders wanted a "credible but aggressive" baseline (P15–P20 range) to demonstrate feasibility. Over time, that baseline was misrepresented as P50, then further compressed to P6 through successive optimization rounds. QSRA reveals these assumptions explicitly.
Can QSRA prevent schedule overruns?
QSRA does not prevent overruns; it forecasts their probability distribution and identifies risk drivers. If governance accepts P50 (2033) as the plan but encounters P80 (2038), is that an overrun? Or is it execution aligned with foreknowledge? QSRA makes the distinction clear.
How many Monte Carlo iterations are needed?
Standard practice: 10,000 iterations. This produces stable P-values and criticality indices to the 5th decimal place. For mega-projects like HS2, 20,000 iterations are defensible. More than 50,000 yields diminishing returns unless activity-level granularity exceeds 500 tasks.
Is QSRA required by UK government guidance?
The Treasury Green Book (2020 update) recommends QSRA for projects exceeding £1 billion. HS2 at £100 billion is obligatory territory. The Infrastructure and Projects Authority (IPA) increasingly mandates QSRA for Major Projects Portfolio (MPP) status updates. Post-HS2, regulatory standards will likely tighten further.
What tools are used for QSRA?
Primavera Risk Analysis (Oracle), @RISK (Palisade), Safran Risk (Safran Solutions), and Crystal Ball (Oracle) are industry-standard. Open-source alternatives include SimPy and R Monte Carlo libraries. For UK infrastructure, Safran Risk is widespread in civil engineering consultancies.
How does QSRA relate to sensitivity analysis and tornado charts?
Sensitivity analysis identifies which variables have the greatest impact on schedule outcomes. A tornado chart visualizes this ranking. QSRA provides the underlying probability distributions; sensitivity analysis interprets the outputs for governance decision-making.
IQRM delivers specialist training and consulting in schedule risk analysis, Monte Carlo simulation, and risk-based forecasting. Our QRM Diploma programme equips professionals with the practical skills to build, run, and interpret QSRA models on real projects.
Is your project facing schedule risk? IQRM offers forensic QSRA consulting and governance support. Whether you are resetting a baseline, validating contingency buffers, or building risk intelligence for regulators, we deliver P-value analysis, tornado charts, and Monte Carlo models that board members understand.
Rami Salem is head of schedule risk analysis at IQRM and a Fellow of the Institute of Risk Management. He has led QSRA engagements on UK rail (Network Rail, Crossrail), nuclear (Hinkley Point C), and water (Thames Tideway Tunnel) megaprojects. Views expressed here are his own.
Published: 9 April 2026
To learn the full QSRA process used on HS2, start with our Schedule Risk Analysis Complete Guide.
