Demontration of of a Risk coverage Engine on a Territory – DEMOCRITE

The proposed work plan couples “upstream” scientific works (simplified model for a consequence of a blast in urban environments, multi-scale model of fire propagation in urban environments, correlation methods between probability and consequences to propose gravity criteria; correlation between intervention data and statistics to propose predictive analytics models for specific risks …) and operational applications (GIS links for the urban geometry and production od risk propensions maps and intrinsic vulnerability maps).
DEMOCRITE in structured in 3 complementary parts :
- An operational part [O] at the origin of the project specifications.
- A scientific part [S] for the modelling of risks and data analysis.
- An information technology part [I] for the development of the software prototype and the integration of models.
The project tasks below are linked to one or several of the O, S, I parts:
Task 01: Project management – Coordination [OSI]
Task 02: Data acquisition and data management [O]
Task 03: Prototype model for fire risks and consequences [S]
Task 04: Prototype model for explosion risks and consequences [S]
Task 05: Mapping of risk probabilities [I]
Task 06: Mapping of intrinsic vulnerabilities [S]
Task 07: Software development and integration [I]
Task 08: Dissemination [OSI]
Task 09: Social, legal and organizational stakes [O]
Task 10: Tests and generalization [OSI]

Main results at T0 + 18 months for eachh task :
T01: Consortium agreement available.
T02: Geocoding of 3 years of BSPP interventions. Preparation of a common GIS database for all partners.
T03: Statistical analysis of premises and on the two zones and definition of fire scenarios. Interface definition between the micro- and macro-scale models..
T04: Consequences model for a blast in urban environments available (first version).
T05: Improvements of the SYSTEL Risk Modelling Tool (RMT) in order to produce a posteriori risk probability maps. Definition of the modifications required to produce vulnerability maps.
T06: Maps of human vulnerability including various population types and time-dependency..
T07: First version of the technical specification and use cases. Assessment of work packages cost, coping with additional requirements.
T08: Website available. Poster and flyer available (1st version). Dissemination plan available (1st version).
T09: Analysis of the “risk basin” notion in relation to the BSPP territory. Analysis of the « equality principle » with respect to the public service of firefighting and rescue.

Main work planned until the end of year 2:
T02: Acquisition of additional GIS databases.
T03: Numerical simulations of fire propagation in a building. Coding of the cellular automata model for urban propagation (macro model).
T04: New version of the blast consequence model callable from the DEMOCRITE platform.
T05: New RMT version. Selection of specific risks to be studied by predictive analysis techniques.
T06: Coding of the vulnerability analysis maps into the RMT.
T07: First version of the DEMOCRITE platform.
T08: Update of the documents and website. Proposition of an article and an oral communication on the blast consequence model.

Scientific production
The overall project has been presented through two posters (WISG 2014 and WISG 2015) and vulgarization articles in safety magazines.
The simplified model of blast consequences in urban environments has been presented in an oral presentation during an international workshop (IPSF 2015).

DEMOCRITE is a software platform which integrates tools for the analysis and coverage of risks on a territory. They could be used in cold planning mode or in crisis management, and will be used to optimize the rescue response (nature, number, location) given a risk coverage level agreed by the Authority. Some tools will be tested on a limited territory (2,5 km²) but the extension at larger scale will be studied. These tools are meant to map risk probabilities and potential consequences as well as intrinsic vulnerabilities. Techniques for the optimization of resources will be studied.

Models for the development of complex risks:
These low probability risks imply a level 3 operational answer [BSPP 2011]. They are likely to cause large scale consequences and may require engaging numerous vehicles / crews. DEMOCRITE tackles for the moment two such risks: urban fire and explosion. Others (flood, epidemic...) will be studied in a future version. Fire propagation will be based on an urban representation given by a GIS. The propagation will be handled by a cellular automaton whose transition rules will be based on numerical simulations. A local model will be able to replicate the different phases of an indoor fire for different kinds of buildings. Explosion effects (accident, bombing ...) will be first computed will be computed by an hydrocode (reference results). The geometry will be imported from the GIS. Simplified approaches will be tested against the reference results in order to select the best one for DEMOCRITE. The explosion will be allowed to be either the cause or the consequence of a fire.

Risk propensity maps:
High probability risks (such as first aid to persons, representing more than 80% of the BSPP actions) may require a level 1/2 operational setup. The analysis of past events shows that risk propensities are far from being isotropic. Optimizing risk coverage thus requires a precise mapping of risks. The aggregation of unitary risks will be studied. Experience feedback will be coupled to statistical approaches in order to predict land use planning impact on territory risks. For instance, car-crash intervention statistics are not sufficient to predict risk evolution due to the creation of new roads: they must first be correlated to other data (traffic density, average velocity, meteorological conditions, etc.).

Intrinsic vulnerability maps:
Intrinsic vulnerabilities are characteristics of territory stakes having a direct link with consequences should an unwanted event happen. they may also vary with space and time. For instance, public access buildings with a high density of people (stadium during a sport meeting) will increase the local human vulnerability during a few hours. Another aspect, functional vulnerability, describes the functions (government, education ...) performed by a society and how they could be threatened. These functions rely on mappable items. Sometimes the localization of a vulnerable item (a transformer substation) may differ from the affected zone in case of failure (a whole district). Human and functional vulnerabilities will be mapped, and the vulnerability of networks will be tackled. Theses maps will aid in making operational decisions (priority evacuation zones, safety perimeters ...).