Why small disruptions cause major delays

Highly optimised production lines promise smooth output, yet this efficiency often hides a weakness: even the smallest disturbance can send an entire schedule off course. A brief machine breakdown, a missing worker or delayed materials may last minutes, but the knock-on effects can stretch across the shop floor. Despite growing interest in resilient manufacturing, most assessment methods overlook the place where disruptions actually unfold – the task level.

Researchers at TU Wien set out to answer a crucial question: how can manufacturers quantify the real impact of individual task failures and understand how well their systems recover?

A task-centred view of resilience

The team developed the Severity of Failure (SoF) method, designed specifically for the shop floor. Instead of broad frameworks, SoF focuses directly on planned tasks. Every disturbance ultimately prevents a task from running, so each task becomes a potential point of failure. By analysing both its probability of failure and the disruption it would cause, SoF provides a practical indicator of resilience.

Crucially, the method also evaluates the scheduler itself. When a simulated failure occurs, the scheduler must rebuild the plan around a repair period. This reveals not only how fragile the original schedule is but how quickly the system can respond.

What the results show

Applying SoF to schedules created by common dispatching rules, the researchers found that dataset size does not influence resilience and that no single scheduling rule consistently performs best. However, one trend is clear: adding buffer time significantly reduces the SoF value, improving resilience by giving the scheduler room to recover.

By identifying which tasks cause the greatest disruption, SoF supports more targeted planning and future work on intelligent scheduling. It offers manufacturers a practical, data-driven way to understand resilience where it matters most.

Based on “Severity of Failure: Resilience Assessment on the Shop Floor”, published in Procedia CIRP (Vol. 134, 2025).