Safran Risk for Riyadh Metro: Modelling Schedule Uncertainty on Saudi Transit

Riyadh Metro opened to the public in December 2024, making it the world's largest urban transit system built from scratch in a single phase. Six lines, 85 stations, 176 kilometres of track, and a final cost exceeding $22.5 billion. The programme was delivered by three international consortia working in parallel across a city that never stopped functioning during construction. Unlike so many mega-projects that make headlines for the wrong reasons, Riyadh Metro reached operation. It is one of the few recent mega-projects that demonstrates what happens when schedule risk is managed quantitatively from the outset.

Safran Risk for Saudi urban rail is a quantitative schedule risk analysis tool that imports Primavera P6 or Microsoft Project schedules and runs Monte Carlo simulation to produce probabilistic completion dates. It models duration uncertainties, discrete risk events, calendar risks, and correlation across activities to generate defensible P50, P80, and P90 forecasts for complex multi-package programmes.

What makes Riyadh Metro instructive is not just that it finished, but how it managed the interface risks between three separate consortia (BACS, FAST, and ArRiyadh New Mobility) working simultaneously on overlapping routes through dense urban terrain. Each consortium operated its own P6 schedule, yet the Royal Commission for Riyadh City (RCRC) needed a unified risk picture that captured cross-programme dependencies. Safran Risk provided that capability through its multi-schedule import and correlation engine.

This article examines how Safran Risk models schedule uncertainty on urban rail mega-projects, using Riyadh Metro's $22.5 billion programme as a case study in successful QSRA application.

Why Riyadh Metro Succeeded Where Other Mega-Projects Failed

The Riyadh Metro programme used quantitative risk analysis throughout its lifecycle, from the initial feasibility study through to the final commissioning phase. RCRC required each consortium to submit QSRA models as part of their monthly progress reporting, ensuring that probabilistic forecasts were embedded in the governance structure rather than produced as one-off exercises. This is the fundamental difference between projects that use QSRA as a compliance exercise and those that use it as a decision tool.

The programme experienced significant challenges: utility diversions that exceeded initial estimates by 300 percent in some corridors, traffic management restrictions that limited TBM launch shaft construction to nighttime hours, and the integration of systems from multiple international suppliers. Despite these challenges, the programme team used QSRA outputs to make informed decisions about acceleration measures, resource reallocation, and scope sequencing. For the broader methodology that IQRM applies across all QSRA engagements, see Schedule Risk Analysis (QSRA): Guide to Monte Carlo + Examples.

Schedule Risk Drivers on Multi-Consortium Urban Rail Programmes

Urban rail mega-projects in Saudi Arabia face a unique combination of risks that compound across programme packages. On Riyadh Metro, these risk drivers shaped the QSRA model structure.

Utility diversions are the single largest source of schedule uncertainty on urban tunnelling projects. In Riyadh, legacy utility records were incomplete, and ground conditions along several corridors revealed infrastructure that did not appear on any drawing. Each unplanned diversion adds weeks to the enabling works phase and delays TBM launch.

Multi-consortium interface risks occur when three separate programmes share station boxes, interchange structures, and depot facilities. A delay in one consortium's civil works can cascade into another consortium's systems installation. These interface risks must be modelled as cross-schedule dependencies in Safran Risk, not as isolated risks within each consortium's model.

TBM performance variability depends on geology, ground water conditions, and operator experience. Advance rates in Riyadh's limestone formations differed significantly from rates in the softer sandstone sections. This variability must be captured using triangular or lognormal distributions based on actual advance rate data collected during tunnelling.

Summer heat restrictions in Riyadh limit outdoor construction activities from June through September when temperatures routinely exceed 50 degrees Celsius. The Saudi Ministry of Labour enforces outdoor work bans during peak hours. These must be modelled as calendar risks using Poisson frequency distributions to generate probabilistic non-working days.

Building the Safran Risk Model for a Multi-Package Programme

Safran Risk handles multi-package programmes by importing each consortium's P6 schedule as a separate project within a single risk model. Interface milestones are linked across projects using handover constraints, so the simulation captures how delays in one package cascade through the entire programme. This was essential for Riyadh Metro, where Lines 1 and 2 (BACS consortium) shared interchange stations with Lines 3, 4, 5, and 6 (FAST and ArRiyadh New Mobility consortia).

The model structure follows IQRM's seven-phase QSRA methodology. Phase 1 (schedule health check) verifies logic integrity across all three P6 schedules. Phase 2 (risk identification) captures both package-specific and programme-level risks. Phase 3 (distribution assignment) applies IQRM's data sufficiency rules: lognormal for activities with 30+ data points from similar projects, triangular for 10-30 points, and uniform for fewer than 10.

The QSRA model showed that the baseline P6 completion date sat at approximately P15 on the S-curve, meaning only a 15 percent chance of achievement. The P80 date, which IQRM recommends for programme-level planning, added 12 months of contingency. The actual opening in December 2024 fell close to the P80 forecast, validating the risk model's accuracy. For understanding how P50, P80 and P90 confidence levels drive these decisions, see P50, P80, P90 Confidence Levels in Risk Analysis.

Correlation Modelling Across Consortia

Correlation is what separates a realistic QSRA from a theoretical exercise. On Riyadh Metro, all three consortia sourced labour from the same markets, procured materials from the same regional suppliers, and operated under the same regulatory environment. If labour productivity was lower than planned on one package, it was highly likely to be lower on all three. Without correlation, the Monte Carlo simulation treats each package's risks as independent, producing an unrealistically narrow S-curve that underestimates the true schedule exposure.

IQRM recommends applying moderate positive correlation (0.4 to 0.6) across activities that share resource pools, and strong correlation (0.6 to 0.8) across activities affected by common external factors such as regulatory approvals and utility authority response times. In Safran Risk, this is configured using correlation groups that apply Pearson coefficients to selected activity clusters, ensuring that systemic risks affect related activities simultaneously rather than in isolation.