José L. Torero

University College London (UK)
Prof. José L. Torero is the Head of the Department of Civil, Environmental and Geomatic Engineering at University College London (UK). He was Director of the Center for Disaster Resilience at the University of Maryland (US) and Head of the School of Civil Engineering at the University of Queensland (Australia). He works in the fields of fire safety, combustion, environmental remediation and sanitation where he specializes in complex environments such as developing nations, complex urban environments, novel architectures, and critical infrastructure, among others topics. He has studied major environmental problems such as oil-spills or large underground coal fires where he has developed unique approaches towards impact mitigation. He is co-inventor of STaR, a technology for soil remediation that has been extensively commercialized and also extended to sanitation and the efficient management of wastewater.
Modelling Fire as a Natural Resource
Fire is generally perceived as a destructive force not a natural resource. This is mainly because the perception is that fire cannot be controlled, therefore it cannot be harnessed in a natural manner. A very different perception exists of combustion systems, from an internal combustion engine to an incinerator. While some fires, such as fire places, can be controlled in a natural and simple manner most fires seem to escape any form of management or control. At the core is the misunderstood complexity that disables any attempts to model fires. Combustion systems used as power sources can be isolated from their boundaries, therefore modelling of the combustion processes can be restricted to length and time scales consistent with the flame thickness, transport processes and the reaction kinetics. Once these can be resolved using detailed direct numerical simulation of chemistry and transport processes, different mathematical techniques can be used to extend the domain to realistic length scales (i.e. combustion chamber, burner, reactor, etc.). Thus, concepts such a s a “perfectly stirred reactor” (PSR) can be used when the resolution of the chemistry is most important and the turbulence details can be omitted or “flamelets” when the opposite is true and turbulence needs to be resolved while chemistry can be simplified. Numerical simulation of combustion systems has been demonstrated as viable and numerous techniques have been developed and reported. As such combustion is perceived as a natural resource. Fire is a combustion system where the reaction cannot be isolated from the boundary conditions, and as such the reaction details cannot be isolated from the boundary conditions. Furthermore, flow fields are mostly dominated by buoyancy, thus the flow field is strongly coupled with the combustion process. Fire is therefore a process that inevitably crosses length and time scales that cover several orders of magnitude. Furthermore, all time and length scales need to be resolved simultaneously. While simplifications are possible, these are very complex and conventional mathematical techniques used for combustion have proven to be unable to simulate fires. Fire modelling remains very computationally intensive and inaccurate. Therefore, fire has not been perceived as a harnessed resource but as a destructive force. This presentation explores the concept of data assimilation with a combination of simplified models and complex numerical simulations as a viable means to produce fire predictions at a reasonable computational cost and adequate precision. Given that fires are both initial and boundary value problems, conventional data assimilation techniques used for weather forecasting do not apply. Furthermore, data quality and quantity is poor when contrasted with the needs of such non-linear problems, thus most statistical techniques do not deliver sufficient accuracy and robustness. Therefore, novel mathematical approaches need to be implemented. Potential approaches that introduce physically based constraints rather than mathematical analysis of data will be discussed.