THE SAFETY SYSTEMS RESEARCH GROUP performs fundamental research on reliability and resilience in complex systems. The aim is to investigate approaches outside current research norms, for both technological and human aspects of systems. We work closely with regulators, commercial organisations and the UK government to disseminate our methods for the benefit of public safety.
- Empirical methods for strong assurance of safe behaviour in computation. As progressively more critical tasks are handed over to automation, it remains unclear how much confidence we can reasonably hold in computation. Confidence needs to be highest where the possible consequences of system failure are extreme, and this implies that confidence should be measurable using scientific methods. We have been able to make major advances in statistical testing methods for this purpose.
- The cost benefit mathematics of risk mitigation. J-value analysis justifies new strategies in the management of major risks, from nuclear accidents to pandemics.
- The foundations of resilience to psycho-social and organisational causes of major disasters. Disasters continue to occur, and personal and human-organisational fallibility often plays a key causative role. A new approach is needed to understand and describe the causal mechanisms in order to find ways to mitigate them.
See more here.
CONTRIBUTIONS TO SCIENCE AND PUBLIC SAFETY
The world’s most cited academic journal paper in software testing
- https://dl.acm.org/doi/abs/10.1145/267580.267590
Our statistical test (ST) methods have been used for some of the UK’s most critical civil nuclear systems, and are now incorporated into UK nuclear regulation (ONR’s SAPs and TAGS)
- https://www.onr.org.uk/saps/2014 (2020 revision)
- https://www.onr.org.uk/operational/tech_asst_guides/ns-tast-gd-046.pdf
- cited in international standard IEC61508 (section C5)
ST methods are used in proposals for new nuclear build in the UK
- https://www.edfenergy.com/file/3863807/download – p. 36
- http://www.onr.org.uk/new-reactors/uk-abwr/reports/ro-abwr-0029-plan.pdf – see ‘Action 2 – statistical testing’
- http://www.onr.org.uk/new-reactors/ap1000/reports/assessment-reports/onr-nr-ar-16-033.pdf page 20, and search on ‘statistical testing’ for many more references
and our papers are cited in international policy synthesis projects
- NUREG/CR-7044 https://www.nrc.gov/docs/ML1329/ML13295A569.pdf
- NUREG/CR-7234 https://www.nrc.gov/docs/ML1715/ML17151A215.pdf
- https://doi.org/10.1016/j.net.2018.02.007 (2018)
- https://doi.org/10.1016/j.anucene.2020.107593 (2020)
Our research achieved the highest possible research ‘Internationally Leading’ impact rating in the Research Excellence Framework 2014
- the 2014 impact case was commercially sensitive (i.e. is not publicly available), but a second impact case submitted to REF 2021 will go public when released from the REF process
High Scientific Council of the European Nuclear Society recommendation of the J-value:
High academic impact papers using J-value analysis provide disruptive guidance on nuclear accident strategic planning, favouring less evacuation and relocation
- Foundational validation of the J-value safety assessment tool against pan-national data: https://ac.els-cdn.com/S0957582017302896/1-s2.0-S0957582017302896-main.pdf?_tid=e4b52f78-d2c5-11e7-b19e-00000aacb360&acdnat=1511713484_c1408cbb7d734eb48189c8bec89fcd68
- Deficiencies in the Value of a Prevented Fatality method currently used in UK cost benefit analysis: http://www.sciencedirect.com/science/article/pii/S0957582014000962 and https://doi.org/10.1016/j.measurement.2019.107044
- Process Safety and Environmental Protection Special Issue containing the closing papers of the multi-university NREFS project to find how best to cope with big nuclear accidents: https://www.sciencedirect.com/journal/process-safety-and-environmental-protection/vol/112/part/PA
- J-value assessment of how best to combat COVID-19: http://www.colbas.org/ntp/opnAxs/N02TH20A.pdf
- Why we need a new science of safety and a full set of J-value papers may be found here: http://jvalue.co.uk/
Guidance on organisational resilience to major accident events has been developed for the energy sector, disseminated by the Energy Institute
- https://publishing.energyinst.org/topics/human-and-organisational-factors/research-report-achieving-greater-resilience-to-major-events-organisational-learning-for-safety-risk-management-in-complex-environments
- http://www.bristol.ac.uk/policybristol/policy-briefings/minimising-industrial-accidents/
PROJECTS
See alsohttps://southwestnuclearhub.ac.uk/wp-content/uploads/2021/01/Systems-RRRS-final.pdf
The CINIF research programme. The Safety Systems Research group has had a 15 year and continuing collaboration with this programme. The statistical testing methods developed within it provide an probabilistic alternative to ‘guaranteeing the absence of faults’ (as in Dijkstra’s famous statement about software V&V). By providing solutions for stochastic system environment simulation and system design-for-test for certain classes of system it has been possible to translate the research into a major testing programme for UK nuclear systems, driving nuclear regulatory policy in this area.
Two recent BEIS-funded Nuclear Innovation Programme projects
- Improving C&I Design for Testability
- Reliability Limits of Programmable Protection Systems
were commissioned to horizon scan potential strategies for assurance of high integrity software in future nuclear applications. These completed in May 2021. Publication of the reports, delayed by Covid, to appear on a platform created by Fraser Nash Consultancy.
Studies funded by the Energy Institute
Two related major studies have been carried out to reduce the risk of major safety-related disasters across a range of major industries. In many cases, such disasters have led to loss of life, severe injury, environmental damage and/or major financial and reputational consequences.
Twelve major events occurring in a range of industrial settings have been studied in order to understand the organisational and cultural precursors which preceded the events and devise methods to avoid them. Diverse sectors were covered including petrochemical, nuclear, transport and civil engineering. These were classified under ten key ‘themes’: leadership, business pressures, safety culture, reporting and learning, safety management systems, competence, communication, risks assessment, contractor management and oversight and scrutiny. Substantial commonality of precursors across sectors was observed – suggesting wide applicability of the research.
A three-step process is proposed to facilitate better identification, understanding and, subsequently, control of the precursors. Firstly, sets of good practice requirements or ‘expectations’ have been developed based on the findings, to enable organisations to benchmark their current approach to organisational safety. Secondly, these are being developed into sets of ‘penetrating questions’ which should enable specific deficiencies to be identified in an organisation’s management and operations. Finally, causal loop diagrams are used to design more effective interventions and, for the first time, enabling behavioural phenomena (e.g. confirmation bias, normalisation of deviance, etc) to be addressed within a ‘systems’ approach.
RAEng Engineering X case study
The RAEng have initiated a major programme of case studies and collaboration in order to develop more effective approaches to designing safer complex systems. The Energy Institute work discussed above was selected as part of a competitive process for inclusion in this RAEng project.
J-value Japan fellowship
The Japan Society for the Promotion of Science (JSPS) appointed Philip Thomas an Invitational Fellow for 2020. He is travelling to Japan in early 2022 to develop proposals with Fukushima Medical University to apply the J-value to optimize remediation and return policies after the Fukushima Daiichi Nuclear accident.
PEOPLE
Dr John May (Associate Prof, group lead)
Dr Neil Carhart
Visiting Professor Philip Thomas
Visiting Professor Richard Taylor PhD, MBE
Visiting Fellow Dr Graeme Collinson
Visiting Fellow Richard Voke