Hazard and consequence reasoning
Urban and industrial fire dynamics, fire spread, tenability, consequence estimation, zoning, emergency planning and standards alignment.
Research biography
Dr Abderrahim Zermane
Fire and process safety researcher focused on physics-based urban fire modelling, occupational accident prevention, explainable machine learning, and research software that turns complex evidence into defensible decisions.
PhD in Process Safety and ReliabilityFormer Research Fellow in Physics-based Fire Modelling, University College London
A researcher working between hazard physics, incident evidence and practical decisions
My work is built around a simple question: how can complex safety information be converted into outputs that engineers, regulators, organisations and communities can actually use? I combine physical modelling, structured risk analysis, statistical methods and explainable machine learning to identify the pathways that produce harm and the controls most likely to interrupt them.
At University College London, I worked as a Research Fellow in Physics-based Fire Modelling and contributed to urban fire-spread modelling within a wider multi-hazard and fragility decision framework. The work required more than producing a simulation. It involved defining assumptions, reasoning about tenability and thresholds, maintaining validation and versioning discipline, and communicating outputs in a form suitable for scenario planning.
My doctoral research at Universiti Putra Malaysia focused on fatal falls in the Malaysian construction sector. Across related studies, I used statistical pattern analysis, fault tree analysis and machine-learning classification to examine thousands of accident scenarios and translate administrative incident records into prevention priorities.
Research approach
I favour methods that are reproducible, traceable and explicit about uncertainty. A model should show where its evidence came from, what assumptions were necessary, which outputs are strongly supported, and where professional judgement is still required. This approach also shapes the research software I develop, including tools for environmental review, design-safety assurance and regulatory evidence mapping.
Current research direction
My programme of work connects four areas: explainable risk analytics for prevention and response; urban fire-spread modelling for the built environment; hydrogen safety readiness and standards evidence; and education-ready resources that help students and practitioners reason from first principles.
Expertise at a glance
A combination of engineering risk methods, fire modelling, data analysis and applied system design.
Structured causal analysis
Fault Tree Analysis, Event Tree Analysis, Bow-Tie, accident causation, critical controls and incident learning.
Explainable analytics
Python, pandas, scikit-learn, classification models, class imbalance, calibration, statistical patterning and interpretable outputs.
City-scale scenarios
QGIS, building exposure data, urban fire-spread workflows, fragility integration and probabilistic uncertainty maps.
Evidence-led software
Review frameworks, auditable evidence chains, controlled outputs, reviewer correction and human-in-the-loop analysis.
First-principles learning
Fire engineering, process safety, ergonomics, risk methods, safety analytics and defensible calculation habits.
Education and professional background
Universiti Putra Malaysia. Thesis on predictive modelling of fatal falls in the Malaysian construction sector.
Université Batna 2, with work in environmental systems, hazards, toxicology and reliability.
Université Batna 1, Algeria.
Physics-based urban fire modelling and teaching support from 2023 to 2025.
Fires and explosions, ergonomics, practical teaching and postgraduate research support.
Safety roles with DMA Construction, Bonatti SPA and Sonatrach.
Research profiles and contact
Review the current publication record on Google Scholar and ORCID, or contact me regarding research, teaching and technical collaboration.
Contact Dr Abderrahim Zermane