The future of evacuation drills: assessing and enhancing evacuee performance

K E Boyce, Steve Gwynne, Martyn Amos, Max Kinateder, Noureddine Benichou, Natalie Van der Wal, Enrico Ronchi

Research output: Contribution to journalArticlepeer-review

28 Citations (Scopus)
50 Downloads (Pure)


Evacuation drills are generally the main mechanism for improving or measuring occupant performance in emergency situations, but their effectiveness is often hard to measure, and there is limited evidence for sustained training benefits. However, innovations in technology (e.g., augmented/virtual reality, novel sensors and wearable tech) offer (when combined with new approaches to designing and delivering drills) significant opportunities for a “next generation” of evidence-based evacuation drills. In this paper, we present the findings of a recent trans-national research project; we establish the main limitations of existing drills, propose a framework for the assessment of both training and evaluation aspects of drills, make a number of recommendations, and suggest a programme of work for their implementation. The paper, therefore, provides a conceptual foundation for future work which will focus on (1) establishing an evidence-based methodology for assessing evacuation drills (and alternatives), (2) harnessing novel objective and automatable approaches to data capture/analytics in order to better characterize performance, (3), developing alternatives to the current drill model, based on emerging technologies, and (4) developing guidance for regulatory bodies on the costs and benefits of each approach for different scenarios.

Original languageEnglish
Article number104767
Number of pages10
JournalSafety Science
Early online date13 May 2020
Publication statusPublished (in print/issue) - Sept 2020


  • AR/VR
  • Assessment
  • Evacuation drill
  • Protocol
  • Simulation
  • Training


Dive into the research topics of 'The future of evacuation drills: assessing and enhancing evacuee performance'. Together they form a unique fingerprint.

Cite this