Managing Large Project Schedules
Complex projects often involve extensive schedules with thousands of tasks. Typically, immediate tasks contain a high level of detail, while long-term activities are more broadly defined. As the project progresses, completed tasks are replaced with more granular breakdowns of upcoming work. Risk analysis for such projects relies on Monte Carlo simulation, a computationally intensive process that requires substantial memory and processing power. Historically, running simulations on large schedules posed challenges due to software limitations, forcing teams to maintain a simplified secondary schedule for risk analysis. However, as schedules grow in complexity, maintaining two separate versions becomes impractical and highly time-consuming.
Breaking a large task into smaller sub-tasks presents additional challenges in risk analysis. For example, consider a fencing project originally estimated to take Triangle(30,40,60) days. As the project nears execution, the task is divided into ten equal sections, each estimated at Triangle(3,4,6) days. This segmentation can create an unintended effect—suddenly, the total projected completion time appears significantly shorter in the risk model than in the original estimate.
Uncertainty contraction effect as a result of breaking a task down into several parts
The underlying issue is that many risk analysis tools treat each task as statistically independent. This means that if one section of the fence takes longer than expected (e.g., 6 days), the model does not account for whether subsequent sections will also be delayed. The Law of Large Numbers (LLN) explains that treating tasks as independent reduces overall uncertainty in the model, often leading to misleadingly optimistic results. Software solutions address this challenge in different ways: (1) some ignore the issue entirely, (2) others apply a simulation technique that enforces 100% correlation among related tasks, and (3) the most advanced tools identify the root cause of uncertainty, ensuring task durations depend on a shared influencing factor.
Another key consideration is the practicality of assigning uncertainty manually across thousands of tasks. A more efficient approach is bulk uncertainty assignment—selecting a group of tasks and applying a global percentage variation. However, caution is needed: if uncertainty is influenced by a common factor, such as contractor efficiency, failing to model this correctly will again lead to an underestimation of risk due to the LLN effect.
Conclusion: For large-scale project simulations, ensure the chosen risk analysis software can handle complex schedules efficiently, supports high-speed processing, properly accounts for statistical dependencies, and offers bulk uncertainty assignment to streamline the process.
Describing uncertainty
Probability distributions play a crucial role in quantifying uncertainty in project risk analysis, particularly for task durations and cost estimates. Selecting the appropriate distribution is essential for accurately modelling variability and ensuring realistic projections. Below are some of the most commonly used probability distributions in risk analysis:
Probability distributions typically used for describing task duration and cost uncertainty in a project
- Uniform – because it is so crude
- Split Triangle – because it doesn’t match what people want to say
- Normal– because the mean and standard deviation inputs are unintuitive, and life is not symmetric
- Lognormal – same mean, standard deviation issues, and the tail can be wildly uncontrolled
- Triangle – because people understand it, even if its crude, but it tends to overestimate uncertainty
- PERT – a curvy Triangle, easy to understand, but it tends to underestimate uncertainty
- Modified PERT –Last parameter alters how flat the curve is, controlling over/underestimation of uncertainty. I invented it, so a bit biased
- ThreePoint – a Modified PERT where you give a "practical max"(like a 95th percentile) instead of the absolute max which is very hard to elicit from an SME. It’s a distribution specific to VOSE software products, so perhaps a bit biased here
Conclusion: Ensure the software includes at least one of the well-suited probability distributions, while avoiding those that may lead to inaccurate risk estimations.
Single Risk Event with Multiple Consequences
Some risk events can trigger multiple consequences. For instance, if a key supplier goes out of business, it could lead to delays in procuring multiple materials, increased costs, and operational disruptions. These impacts either occur together or not at all. If a risk analysis model does not properly account for such dependencies, it may significantly underestimate the overall risk exposure and the importance of mitigation strategies.
Conclusion: Choose risk analysis software that allows linking a single risk event to multiple potential consequences, ensuring a more accurate assessment of project vulnerabilities.
Modelling Risk Event Frequency
Risk events such as strikes, equipment failures, and accidents can occur sporadically, introducing delays and additional costs. Some risks, like the complete destruction of a building, can only happen once, while others may repeat over time. The probability of a one-time event is typically expressed as a likelihood, whereas recurring risks are better represented using an expected frequency (e.g., 0.7 occurrences per year). If a risk analysis tool does not differentiate between these two types, it may lead to inaccurate uncertainty estimates and ineffective risk mitigation strategies.
Conclusion: Ensure that the software can properly distinguish between single-event probabilities and recurring risk frequencies to enhance the accuracy of your risk model.
Impact of Adverse Weather
Weather conditions can significantly affect project timelines, particularly in industries reliant on outdoor work, such as construction, offshore operations, and infrastructure development. Extreme weather events, seasonal variations, and climate-related disruptions must be factored into risk models to ensure realistic scheduling and budgeting.
Conclusion: If weather variability is a critical factor in your projects, choose software that incorporates weather calendars or similar mechanisms to account for these uncertainties.
Managing Complex Cost Structures
Project schedules typically incorporate various cost elements, such as:
- Fixed costs assigned at different project stages (start, mid, or completion)
- Resource-based costs (e.g., hourly rates for workers, daily charges for equipment), linked to task durations
- Overhead costs (e.g., office rentals), independent of specific tasks
Risk analysis software integrates these costs, applies uncertainty, and considers risk-driven cost variations to provide a realistic project cost estimate. However, many additional expenses, such as legal fees, materials, insurance, and financing costs, often fall outside scheduling tools. These are typically assessed separately by cost engineers and financial analysts, requiring integration for a complete financial overview.
Conclusion: If understanding total project cost uncertainty is critical, ensure the software supports data export for external analysis or includes a built-in financial simulation tool.
Evaluating Investment Risks
Projects are typically undertaken with financial returns in mind, whether it's launching a new product or constructing infrastructure for revenue generation. Investment risk analysis relies on discounted cash flow (DCF) modelling to assess financial viability, using metrics like net present value (NPV) and internal rate of return (IRR). Project delays can significantly impact financial performance—additional costs lower NPV, while delayed revenues further reduce profitability.
These calculations are often performed in Excel, with risk analysis tools like ModelRisk enhancing precision. Given that investment success depends on factors like discount rates (typically 10-15%) and IRR thresholds (e.g., 20-25% for strong projects, 5% for marginal ones), integrating project cost and schedule uncertainty into financial models is essential.
Conclusion: If investment evaluation is part of your project decision-making, choose software that can seamlessly export risk-adjusted cost profiles or, ideally, include financial modelling capabilities.
Ensuring Schedule Suitability for Risk Analysis
Effective project risk analysis requires a schedule structure that reflects realistic uncertainties. Fixed milestones and rigid task dependencies (e.g., forcing Task B to start exactly 30 days before Task A finishes) can conflict with probabilistic modelling. Similarly, assumptions that two tasks will finish simultaneously (Finish-Finish dependencies) may oversimplify actual project risks.
Conclusion: Planners should prioritize Finish-Start dependencies and avoid locked milestones to allow realistic risk modelling. Additionally, project risk analysis tools should perform logic checks to identify potential scheduling inconsistencies.
Considering Spreadsheet-Based Risk Modelling
For simpler projects, a full-fledged scheduling tool may not be necessary. In some cases, a well-structured spreadsheet model can effectively capture key risks and uncertainties. To explore practical examples, you can download and experiment with models such as this one and this alternative.