(104as) Methodology for Uncertainty Reduction for FDS Simulations to Obtain Refined Results for Performance-Based Design

Methodology for uncertainty reduction for FDS simulations to obtain refined results for Performance-based design

J. Cadena¹, F. Munoz²

Chemical Engineering Department, Universidad de los Andes, Bogotá, Colombia

¹je.cadena21@uniandes.edu.co

²fmunoz@uniandes.edu.co

 Fire research safety has been researched and applied for around 30 years and has been an important element of Process Safety developments worldwide, which constitutes a clear response to the desire of minimizing the loss generated annually due to fire events in residential, commercial and process plants buildings and facilities. This research has led to enormous achievements such as the performance-based design, which consists basically of thinking ahead of how a construction will perform under a simulated fire and use those results to improve construction characteristics, fire protection systems and possibly impose operating restrictions (in case of a laboratory or a process plant facility). Nevertheless, fire is indeed the most or one of the most complex events that science has explored and there exist no model that always fits accurately the behavior of the diverse phenomena involved, and taking as case study one of the most advanced and researched model such as the FDS (Fire Dynamics Simulator – NIST development), one must take into account the different modules of phenomenological models that interact simultaneously: Mass and species transportation, Momentum transport and pressure, Combustion, Thermal Radiation, Solid phase. The great deal of research invested into this specific model has allowed obtaining a reliable tool to obtain Performance-based results useful for building designers and process safety engineers, but here an issue emerges. All the models involved and the previously mentioned interaction between them might not always be such that represent accurately specific complex fire scenarios, accumulating specific error on some key variables such as hot gases maximum temperature with 30% underestimates. This accumulated errors could be minimized by a modification of the models and their interactions, but such model has not been obtained yet, so a practical process safety tool to reduce uncertainty is developed, based on an FDS validation & error data base, obtained from sources such as the NRC (National Regulating Commission of Nuclear Energy), NIST (National Institute of Standards and Technology) and research papers focused on FDS validation. The tool allows performing an FDS simulation and then, based on the specific conditions of the scenario, a comparison and cross check with validation data is performed to identify key variables that are affected under those conditions and therefore establish statistic criteria for the reduction of uncertainty. This numeric criterion is then complemented by specific qualitative criteria established by the tool’s designers based on their process safety & fire research experience to obtain a final reduction of uncertainty and an enhanced data & information package that can then be used for a Performance-based design. The tool’s development need was identified in a previous work which consisted on a QRA of a fire scenario on a Chemical Engineering laboratory through the use of multiple simulators such as CFAST, FDS and a brand new zone model simulator developed for box-type compartments named FireLab. The results showed certain inconsistency between the field and zone models simulations, arising a doubt that finally was condensed on the development of the present research which proposes a new approach to analyze data obtained from fire simulations and improve fire systems design performance. The work related to this research is currently ongoing and trial computational tests are expected for early November 2012, and once this is accomplished the methodology will be refined if needed in order to begin using this tool as part of several tools already developed by the Safety Group of the Chemical Engineering Department of Universidad de los Andes, to improve the knowledge of technological risks and the role of process safety towards them.