Apr 9

Schedule Risk Analysis for Nuclear: Hinkley Point C QSRA Lessons

By Rami Salem | Published: April 9, 2026 Executive Overview: Hinkley Point C ballooned from £18 billion to £35 billion and slipped from 2025 to 2030+. A rigorous quantitative schedule risk analysis (QSRA) at project initiation could have flagged first-of-a-kind (FOAK) nuclear risk, supply chain constraints, and regulatory hold points—quantifying the most likely P50 and prudent P80/P90 completion dates that would have reflected reality.

Introduction: Why Hinkley Point C Missed Every Deadline

In 2016, EDF Energy and the UK government committed to delivering the Hinkley Point C (HPC) nuclear power station by 2025. By February 2026, that target had slipped by at least five years, with EDF taking a €2.5 billion impairment charge and announcing a 12-month commissioning delay on top of construction delays already baked in.

The plant's cost tripled: starting at £18 billion (2015 prices), it now approaches £49 billion in current money. Yet these failures were not inevitable. A mature, rigorous quantitative schedule risk analysis (QSRA)—commonly used in oil & gas mega-projects and aerospace—could have quantified the probability of every critical path delay, mapped supply chain bottlenecks, and flagged the 40–60% schedule buffer required for first-of-a-kind (FOAK) nuclear construction in a 20-year skills gap.

This article dissects HPC's schedule collapse through the lens of QSRA best practices, compares planned timelines to what risk-informed prediction would have shown, and extracts lessons that Sizewell C and future mega-projects can apply immediately.

The Hinkley Point C Context: A Perfect Storm of Schedule Risk

HPC began in 2017, targeting an 11-year build to operational status by 2025. The project inherited three compounding challenges that any QSRA would have surfaced as tier-one risks:

1. First-of-a-Kind (FOAK) Risk: HPC was the first new-build nuclear plant in the UK in 20 years. The supply chain—from specialist crane operators to nuclear-grade concrete suppliers—had to be rebuilt from scratch. FOAK premium in nuclear typically adds 30–40% to schedule and cost.

2. Supply Chain Restart: The 20-year gap meant no active nuclear construction workforce in the UK. Training, recruitment, and vendor qualification dragged timelines. Critical path items—steam generators, pressurizer modules, instrumentation—had single or dual-source suppliers globally, introducing supply chain correlation risk.

3. Regulatory Cadence: UK Office for Nuclear Regulation (ONR) hold points at design, fabrication, installation, and commissioning stages. Regulatory approval windows are not deterministic; delays of 6–18 months are common when rework is flagged.

In parallel, COVID-19 added 15 months of calendar disruption (2020–2021). By July 2025, the world's largest crane placed the 47-meter reactor dome on Unit 2, a visible milestone—yet commissioning timelines were already slipping by 12+ months.

What a Rigorous QSRA Would Have Predicted

A QSRA uses Monte Carlo simulation to model thousands of schedule scenarios, combining:

Activity duration distributions: Best-case, most-likely, worst-case for each task (civil, mechanical, electrical, commissioning).
Correlation matrices: Shared resource constraints (cranes, skilled labor, inspection teams) that make delays cascade across workstreams.
Probability trees: Regulatory approval cycles, rework scenarios, and supply chain failures modeled as risk events with frequency and severity.
Confidence levels: P50 (median), P80 (80th percentile), P90 (90th percentile) completion dates.

Applied to HPC at sanction (2016), a credible QSRA would have produced a comparison like this:


Phase/Metric	HPC Deterministic Plan (2016)	QSRA Predicted (P50)	QSRA Predicted (P80)	Actual (Feb 2026 forecast)
Civil Works Completion	Q4 2021	Q2 2022	Q4 2022	Q4 2023
Mechanical Installation	Q2 2023	Q4 2023	Q2 2024	Q2 2025
Electrical Commissioning	Q3 2024	Q2 2025	Q4 2025	Q4 2026
Full Operational (Unit 1)	Q4 2025	Q3 2027	Q2 2028	2030+
Buffer to P80 (days)	0 (deterministic only)	~550 (18 months)	~850 (28 months)	~1,800+ (actual overrun)

The table reveals the gap: HPC's 2025 operational target had zero schedule buffer. Any QSRA model would have predicted P50 completion no earlier than Q3 2027, and a prudent, defensible P80 close to Q2 2028. The actual 2030+ timeline, while later than even P80, validates the QSRA method's core insight: FOAK nuclear construction without 40–60% schedule buffer is a plan to fail.

Nuclear-Specific QSRA Risk Drivers

1. Civil-Mechanical-Electrical Sequencing Interdependencies

Unlike conventional power or industrial projects, nuclear plants have rigid, non-compressible sequencing mandates. Concrete cure times are fixed by physics. Pressure vessel installation cannot begin until the reactor cavity is finished and inspected by ONR. Any delay in civil work cascades forward with hard stops, not soft logic.

A QSRA must model correlation across these streams. If civil labor is constrained (which it was—UK construction market was tight in 2018–2019), then both civil and the downstream mechanical trades that depend on scaffolding and access suffer. Standard CPM (critical path method) miss this; Monte Carlo captures it.

2. Supply Chain Restart & Vendor Risk

HPC required dual-tracking of UK and overseas fabrication. Steam generators came from Japan/South Korea, reactor pressure vessels from France, instrumentation from specialist suppliers in Germany and Switzerland. Any single delay in overseas manufacture or port/transport constraint cascaded into the critical path.

A QSRA probability tree for supply chain risk might look like:

Supply Chain Event: Pressure vessel fabrication delay (frequency: 25% probability, duration impact: 6–12 months).

Correlation: If the pressure vessel is late, mechanical installation trades (piping, instrumentation) cannot start on schedule. These same trades are also labor-constrained. The QSRA correlates both risks in a single distribution.

3. Regulatory Hold Points & Rework Risk

The ONR approved design in stages, but commissioning review cycles added unexpected delays. In 2023–2024, ONR flagged instrumentation and control system issues that required 6–9 months of rework and re-inspection. Any QSRA should have modeled ONR approval hold points as discrete risk events, each with 10–15% probability of 3–6 month delays.

4. First-of-a-Kind (FOAK) Learning Curve & Rework Contingency

FOAK nuclear projects globally show 30–50% cost and schedule growth. This is not mismanagement; it reflects the reality that novel designs, first-time assembly of complex systems, and unforseeable interface conflicts demand rework. A credible QSRA applies a FOAK adder: every activity duration distribution is shifted right by 20–30%, and 3–5% of critical path activities are marked as "rework triggers" with secondary task networks.

HPC had no such adder in its baseline plan. It should have.

5. Commissioning & Testing Unpredictability

Commissioning is not a linear process. Testing often reveals minor defects (instrument calibration, valve seat leakage, control logic refinement) that require iteration. A single-unit commissioning in a new plant can take 18–24 months, not the 12 months HPC initially assumed. QSRA must model commissioning as a probabilistic phase with a median duration 20–30% longer than the deterministic plan and a long tail (P90 can be 50% longer).

The Nuclear Schedule Buffer Formula

Nuclear Schedule Buffer = P80 Completion Date − Deterministic Baseline Duration

For a typical first-of-a-kind nuclear project, this buffer is 40–60% of the base deterministic duration. For HPC's 11-year deterministic plan, a defensible P80 schedule buffer would have been 4.4–6.6 years—implying an operational date no earlier than 2028–2030. The actual 2030+ outcome aligns with the high end of this range, validating the formula.

Sizewell C: Learning from HPC's QSRA Lesson

Sizewell C (site in Suffolk, construction to follow) is explicitly being planned to avoid HPC mistakes. The project has adopted:

Modular Supply Chain Strategy: Pre-fabricating larger subsystems overseas to compress on-site duration and reduce labor bottlenecks. QSRA modeling shows 15–18% schedule acceleration vs. HPC's full site-build approach.

Parallel Design-and-Build Sequencing: Where ONR permits, detailed design continues while long-lead items (pressure vessels, steam generators) are being fabricated. This overlaps design and procurement risk windows.

Shared Workforce with HPC: HPC construction experience (2017–2030+) will directly feed Sizewell C labor planning. QSRA for Sizewell C can include a "learning curve" distribution that tightens duration estimates for activities that HPC has already executed.

These measures will be validated through a robust QSRA during Sizewell C's financial investment decision phase. The project will publish P50 and P80 timelines, not a single deterministic date.

Six Critical Practices for Nuclear QSRA

1. Distinguish FOAK from Nth-of-a-Kind: A second or third reactor at a multi-unit site (like HPC Unit 2) should have shorter durations and narrower distributions than Unit 1. QSRA must reflect this via learning curves.

2. Model Labor Correlation Explicitly: Nuclear projects are labor-intensive. If excavation, concrete, and structural steel all compete for cranes and skilled labor, the QSRA correlation matrix must capture shared resource contention. A delay in one activity increases the probability of delay in others.

3. Apply Monte Carlo with 10,000+ Iterations: Simple three-point estimation (best/most-likely/worst) is insufficient for mega-projects. A full Monte Carlo run with thousands of iterations reveals tail risk (P90, P95) more accurately than algebra.

4. Update the QSRA Quarterly: As-built progress, vendor delays, and regulatory decisions should feed back into the QSRA. A static baseline becomes irrelevant within 12 months in large nuclear projects.

5. Include Scenario Analysis: Run QSRA under different assumptions: optimistic (no major rework), base case (median FOAK assumptions), and pessimistic (supply chain disruption + regulatory rework). This provides decision-makers with a transparent range, not false precision.

6. Separate Calendar vs. Effort Duration: COVID taught the nuclear industry that calendar time and effort hours are decoupled. Remote inspections, supply chain gaps, and social distancing can add 15+ months of calendar time with no additional labor. QSRA must model calendar duration, not just work hours.

Frequently Asked Questions

Q: Why didn't HPC use QSRA from the start?

A: The project was sanctioned with a fixed cost-cap model designed to minimize public perception of risk. Presenting a P80 completion date of 2028 instead of 2025 would have triggered political pushback. Instead, the deterministic 2025 plan was adopted, embedding schedule risk into the baseline. This is a governance failure, not a technical one—but QSRA could have forced the conversation earlier.

Q: Is P80 the right confidence level for nuclear mega-projects?

A: P80 (80th percentile) is a pragmatic middle ground. It is aggressive enough to remain credible with finance and sponsors, yet conservative enough to buffer most foreseeable risks. For critical national infrastructure (like nuclear baseload), some argue P85–P90 is more appropriate. For HPC, a P85 commitment would have been ~2029.

Q: Can QSRA predict COVID-like black swans?

A: Traditional QSRA cannot model unknown unknowns (pandemic, geopolitical embargo). However, it can model resilience scenarios—"if supply from Country X is interrupted, what is the impact?" and "if on-site labor is reduced by 30%, how long does the project extend?" This scenario sensitivity is more useful than pretending to predict the unpredictable.

Q: Does QSRA replace project management?

A: Absolutely not. QSRA is an input to risk management and contingency planning. The project manager still executes the plan, track actuals, and makes decisions. But QSRA gives the PM a data-driven baseline for how much schedule buffer is realistic and where the highest-risk activities cluster—allowing prioritization of mitigation.

Q: How does HPC's P80 (2028) compare to the £35B cost?

A: The £35B figure includes schedule delay cost (extended staffing, overhead, financing charges). A more accurate accounting: base construction ~£20B, financing/financing charges on delay ~£8B, regulatory/quality assurance on extended timeline ~£4B, contingency/risk provisions ~£3B. QSRA informs the rightsize of schedule contingency (typically 8–12% of base duration, translating to 3–5% of total cost).

Q: Will Sizewell C benefit from QSRA?

A: Yes, explicitly. Sizewell C is being planned with a QSRA-informed schedule from day one. The project has committed to publishing P50 and P80 timelines as part of its financing documents, not a single deterministic date. This transparency allows stakeholders to understand and price schedule risk correctly.

Conclusion: From HPC's Failure to Industry-Wide Change

Hinkley Point C did not fail because the engineering was unsound or the team incompetent. It failed because the baseline plan was divorced from the reality of first-of-a-kind nuclear construction, 20-year supply chain gaps, and regulatory complexity. A mature QSRA would have surfaced these realities in 2015–2016, forcing a more honest conversation about P50 and P80 timelines.

The good news: QSRA is not new, esoteric, or expensive. Oil & gas mega-projects, aerospace, and industrial construction have used Monte Carlo schedule risk analysis for 20+ years. Nuclear mega-projects are now adopting it. Hinkley Point C's £35 billion lesson—and five-year slip—is already being learned by the next generation of projects.

The industry norm is shifting: from a single deterministic timeline to a transparent range (P50 to P90), from hidden schedule risk to quantified contingency, and from "hope for the best" to "plan for the realistic." Sizewell C will be the proof point.

Ready to understand schedule risk on your mega-project? IQRM's quantitative schedule risk analysis methodology is built on 20+ years of EPC and industrial construction experience.