Running QSRA and QCRA as separate models is the most common mistake on major UK and GCC capital programmes — and the most expensive.
Schedule slip drives cost. Cost overrun drives schedule. Modelling them in isolation systematically understates exposure. The fix: Joint Confidence Level (JCL) analysis — an integrated Monte Carlo model that simulates both dimensions simultaneously.
A project with siloed P80 cost and P80 schedule is not at P80 combined — it's typically at P55–P65. The integrated view is the only honest one.
The siloed-analysis trap
Most organisations operate in two silos:
- Planners run QSRA on the schedule.
- Cost engineers run QCRA on the estimate.
- The two models share no inputs, no correlations, no time-dependent cost drivers.
The board then sees “P80 schedule = +6 months” and “P80 cost = +£10M” — and assumes the project is P80 confident. It isn't. The probability of both being met simultaneously is much lower because slip and overrun are correlated.
What JCL actually is
Joint Confidence Level analysis runs a single Monte Carlo model in which:
- Schedule activities have duration ranges (the QSRA layer).
- Time-dependent costs (rent, supervision, prelims, financing) inherit those duration ranges automatically.
- Discrete risks impact both schedule and cost where appropriate.
- Correlations link commodity prices, labour productivity, and weather windows.
The output is a 2D scatter of (cost, schedule) outcomes from which you read joint P-values — e.g. “there is an 80% chance of completing by Q3 2027 AND staying under £54M”.
Figure 1 — Each dot is one Monte Carlo iteration. Hitting BOTH P80 cost and P80 schedule simultaneously is rarer than either alone.
The 5 building blocks of a JCL model
1. A risked schedule (QSRA layer)
Activity duration distributions, discrete schedule risks, and weather/permitting calendars. See our QSRA in Safran Risk walkthrough.
2. Time-dependent (TD) cost links
Costs that scale with duration: supervision, prelims, financing, equipment hire. In a JCL, these inherit the simulated duration directly — so a schedule slip auto-generates the right cost impact.
3. Quantity-dependent (QD) cost risks
Material rates, productivity, labour rates. These are independent of duration but vary across iterations.
4. Discrete risk events with cost AND schedule impact
A delayed permit might impact 3 weeks AND £0.4M. The same risk event drives both dimensions.
5. Correlations
Commodity correlations, labour pool correlations, sector-wide productivity. Without these, Monte Carlo correlation errors compound.
Tools that support JCL natively
- Safran Risk — first-class integrated cost/schedule simulation.
- Argo — integrated Monte Carlo (IQRM-aligned stack).
- Primavera Risk Analysis (legacy) — possible but with significant manual cost-side workarounds.
- @Risk + Excel — JCL achievable for small projects but heavy lifting on the schedule side.
This is exactly the gap the Risk Data Engine™ (RDE™) closes. RDE™ supplies empirically derived distributions for both duration uncertainty and cost uncertainty, drawn from your historical project data — so the JCL inputs reflect reality, not optimism.
Frequently asked questions
Is JCL only for NASA / aerospace?
No. JCL originated at NASA (post-Constellation) but is now standard practice for large infrastructure, energy, defence, and rail programmes in the UK and GCC.
What's a typical joint P-value to target?
Joint P70–P80 is the common target band. Some sponsors require joint P80 alongside individual P50 for cost and schedule.
Can I retrofit JCL onto an existing QSRA?
Yes, but it's usually cheaper to rebuild the model in a JCL-capable tool than to bolt on a cost layer late.
How long does a JCL take?
Allow 4–8 weeks for a first-time JCL on a £200M+ project — most of it integration and validation, not the simulation runtime.
Run JCL with confidence
The QRM Professional Programme covers integrated cost & schedule risk end-to-end — with live Safran Risk and Argo workshops.
Explore the Programme →Related: What is QCRA? · QSRA Explained · How to Calculate Cost Contingency
