Dynamic Learning as the Follow-up from Accident Investigations
A +/-

Dynamic Learning from Accidents

Bridging the gap between accident investigations and learning

We need to investigate accidents. To get feedback from reality. To understand what went right and what went wrong. To determine how mishaps can be prevented in the future and successes can be disseminated.

We are willing to learn from mistakes and to avoid repeating them. We choose to start in reality, not in modeling or theories. We start with the unexpected, the unforeseen. We do not primarily create barriers for repetition of similar accidents, but try to remove barriers for learning in order to understand why systems function as they do and what can be done about the situation.

Why, then, do accidents repeat themselves? Why can accidents, incidents and undesirable events emerge as unforeseen surprises? Technology, organizations and people failed to execute what was originally planned, were unaware of limitations and assumptions and did what they thought was reasonable.


The ESReDA project group on Dynamic Learning has analyzed a number of case studies which illustrate learning deficiencies and the multilevel nature of necessary changes. In doing so, we have created a new model and metaphor, the “ESReDA Cube”. The ESReDA Cube is a tool to explore the learning potential of systems, to analyze events, incidents and accidents, assess accident investigation reports, to diagnose systems, layer-by-layer, actor by actor, dimension by dimension, lesson by lesson, intervention by intervention.

Case study analysis on dynamic learning from accidents — The ESReDA Cube, a method and metaphor for exploring a learning space for safety

This document introduces the ESReDA Cube, a method and metaphor for exploring a learning space for safety. The metaphor is illustrated on 5 accident case studies in high risk organizations, which are analyzed to identify learning barriers and opportunities at different system levels.


30 years of research into these questions highlight the existence of barriers to learning. Such barriers can be anywhere. Barriers can be in design assumptions, in risk assessment and operating limitations, in governance and control constraints. Barriers exist throughout the system and are open to exploration, analysis and understanding.

Barriers to learning from incidents and accidents

Provides an overview of knowledge concerning the barriers which exist to learning from incidents and accidents. While most research on learning focuses on individual cognition, the focus in this document is mainly on learning at an organizational level, while also taking into account a cross-organizational and even societal/cultural level. It concerns both organizational learning (the flow of lessons into new practices and modified procedures) and policy learning (impact of lessons on public policy, law, regulations and standards).

From an operational standpoint, the document aims to help practitioners to identify opportunities for improving their event learning process. It should be useful in the context of a process review of your organization’s learning system.


The ESReDA Project Group on Dynamic Learning brings together different people with experiences from different sectors of society, who all have a common interest in improving their learning from accidents, incidents and near misses.  The project group is an arena where the cross-fertilisation of experiences has resulted in guidelines for preparing a training toolkit in event investigation and dynamic learning.

Training toolkit: learning from incidents and accidents

We did not only focus on optimizing performance in existing systems. We also focused on adaptation to a more sustained, enhanced safe performance. Eventually, we had to incorporate innovation to eliminate barriers for safety enhancement. We studied not only the barriers in the system itself, but also the barriers in how to investigate and how to develop better solutions.


(All three documents may be printed in pdf-form.)

This journey to systemic learning has not been a single and simple trip. It took about 12 years, several working groups and seminars to compile our discussions, experiences and (dis-)agreements on how to proceed to enhance safety in complex and dynamic systems. The final seminar was held in Porto in 2013, and the results of the seminar have been compiled as a Summary of papers, visions and conclusions.

All the ‘products’ of the ESReDA project groups’ work 2003-2015 have been listed on the ESReDA project group webpage. These include not only the results of the current project group but also the results of the previous groups, e.g. the Guidelines for safety investigations of accidents.

Wishing you success,

The ESReDA Project Group on Dynamic Learning

Post scriptum

Safety as a science seems to be in crisis!

 To learn from the positive, safety has been established as a science in the 1970’s at several academia and within the R&D departments of major high-tech industries. To mention a few highlights: in the process industry, SHELL has taken the lead in safety management system development in occupational safety. In the nuclear industry, EDF has demonstrated its potential to safely produce nuclear energy for decades, based on probabilistic risk analysis and modeling. In aviation and the maritime, accident investigations have developed into a global network of independent safety investigation agencies. They have become non-plus ultra-safe systems.

However, with the emergence of new technologies, rapidly changing economic, social and market developments, globalization and privatization, new operating environment calls for a Next Gen approach. Such a call is heard in many industrial developments, stimulated by social awareness and acceptance of new risks and hazards, urging a need for a more sustainable and proactive society. Safety investigations provide a timely transparency in its factual functioning, disclosing both systemic and knowledge deficiencies and learning potential.

Unfortunately, major events have also triggered a sense of urgency to make a next step in safety:

  • Recent major accidents in the offshore, nuclear power and aviation sectors were unanticipated and remained unexplained;
  • Discussions in a Special issue of Safety Science (Oct 2014) on fundamentals revealed theoretical and methodological dilemmas;
  • Exploration of underlying fundamentals and notions seems inevitable to avoid disciplinary controversies and personal agonies;
  • Several safety communities each independently explore paradigm shifts: researchers, consultants, designers as well as investigators;
  • Academic and R&D safety experts and practitioners still have to rely on, to a large extent, ‘on the job’ training. A next step in scientific thinking and education seems to emerge.

To provoke your thoughts, an essay is written by John Stoop on the

Challenges to the investigation of occurrences

This essay proposes an integration of perspectives and a transition from a factor-based event approach to a vector-based systems approach. It introduces new working relations between safety investigations and forensic engineering, resilience engineering, design methodology, transition management, simulation and gaming. It advocates a shift in thinking from ‘human error’ to transforming classic notions of Good Airmanship into a general applicable and substantiated Good Operatorship.