top of page

Improving Fire Safety by Modelling Buildings’ Fire Behaviour with SAFIR 2019 – with Dr Gernay

Fire Safe Europe

8 May 2020

Dr Thomas Gernay is an Assistant Professor of Engineering at John Hopkins University. He has developed expertise on Structural Fire Engineering, Fire Safety Engineering, Resilient Infrastructure and Computational Mechanics. His research aims to better the multi-hazard resilience of cities and civil engineering infrastructure. He is the co-author of SAFIR, a software, developed for modelling the behaviour of building structures in fire.

  • Could you please explain what SAFIR is?


SAFIR is a computer software for modelling the behaviour of building structures subjected to fire. It is based on the finite element method. The structure can be made of a 3D skeleton of linear elements such as beams and columns, in conjunction with planar elements such as slabs and walls. Volumetric elements can be used for the analysis of details in the structure, such as connections. Different materials such as steel, concrete, timber, aluminium, gypsum or thermally insulating products can be used separately or in combination in the model.


The thermal attack from the fire is entered as input data, with multiple possibilities ranging from time-temperature curves to interfacing with the computational fluid dynamics software FDS. From this fire input, SAFIR will first calculate the evolution of the temperatures in the structure. Then, the mechanical behaviour of the structure is calculated based on its geometry, its support conditions, the loads that it must withstand and the strength of the materials, taking into account the progressive increase of temperature. The elevation of temperature in the materials produces thermal elongations together with a reduction of strength and stiffness. As a consequence, the displacements of the structure and the distribution of forces in it vary continuously during the course of the fire.


The simulation may result in the prediction of collapse after a certain time, including the way it will collapse, or possibly the demonstration that the structure is able to survive until full burnout (for a fire with a cooling phase).

One of the main strengths of SAFIR is that, because it was specifically developed for modelling structures in fire, the transition from the fire input to the thermal analysis to the mechanical analysis is accurate and efficient for the user.


  • How did the idea of a software like SAFIR emerge?


My PhD advisor at Liege University, Professor Jean-Marc Franssen, started developing code in the 1980s to analyze the fire behaviour of some specific members. At the time, there was no software covering this aspect for structural fire engineers, but it soon became clear that there was a need to understand the response of entire structures in fire, notably following the large-scale BRE Cardington tests in the UK, and later the WTC building collapses in the US. He thus assembled and extended pieces of code he had developed over the years to launch the first version of SAFIR in response to this need for a user-friendly specialized software for structures in fire. SAFIR quickly found a user community among scientists, researchers as well as practitioners, and since then we have been continuously developing it to add new functionalities with the goal to propose the most up-to-date software for performance-based structural fire engineering.


  • Who uses SAFIR?


SAFIR is used by more than 250 institutions and companies worldwide. These include academic institutions, public and private research centres, and design offices on five continents. In addition, close to 1000 free demonstration versions have been distributed to students and engineers who want to try the software or use it for small studies. 


  • How does your software help them in their work?


Engineers and researchers use our software to model accurately and predict a structure’s response to fire. SAFIR supports researchers who derive new design methods or develop new structural systems for improving the fire safety of the built environment. The scientific impact of SAFIR is attested by the fact that the two scientific papers describing the software total about 600 citations on Google Scholar. SAFIR also supports practising engineers who design exceptional structures that require an explicit assessment of the fire response. For example, SAFIR was recently used for the fire design of the JTI Building in Geneva (by INGENI), as well as the Four Pancras Square in London (by Trenton Fire). In the US, SAFIR was used by three of the four engineering design offices in the recent project funded by the ASCE and the CPF to advance performance-based structural fire engineering design through exemplar procedural guidance. 


  • So SAFIR enables its users to evaluate the fire resilience of a structure and model the behaviour of a building in fire. On what definition of fire resilience do you base this evaluation?


Our software enables the detailed assessment of the response of entire structures under physically-based fire scenarios. SAFIR will yield the time history of temperatures, deflections and changes in load paths in the structure under the full course of the fire, including the cooling phase. SAFIR will not decide whether the predicted response amounts to “failure”; the user has to interpret the results in light of the performance objectives, and his/her appreciation of what a fire resilient design means for a particular structure. For example, researchers and designers can evaluate whether a building can withstand a fire event for sufficient time to allow safe evacuation, whether it will collapse toward the inside or toward the outside – thus impacting nearby other buildings or firefighters intervening from the outside – or whether it can survive the full burnout with limited damage and remain functional after the fire.


  • What are the improvements of the 2019 version of the software?


Recently, new improvements in SAFIR have focused on material models, finite elements, and interfacing with sophisticated fire models. For the 2019 version, the main improvement was the modification of the beam finite element to allow the introduction of semi-rigid or pinned connections at their ends. With this method, users can easily define temperature-dependent semi-rigid connections between the beams and columns of their frame structures, for example, without the need to introduce any specific “connection-type” element. Other improvements included the introduction of probabilistic material models for steel and sprayed fire-resistive material, as well as the incorporation of the HCM and RWS fire curves for tunnel fires.


  • What kind of challenges do you face when developing the software?


The main challenge lies in the fact that the two developers of SAFIR, Jean-Marc Franssen and myself, are Faculty members. We are not professional software engineers, and we have to combine the software development with our other duties in terms of conducting research, teaching and mentoring, and providing service to the university. As a consequence, we cannot provide the same type of 24/7 service as a professional software company. While we work hard to provide user-friendly tools and support to the community, we cannot compete on these aspects with general-purpose commercial software developed by large corporations. Yet we believe that being active scientists in the field is a strength because it means, first, that the software is state of the art, it incorporates the latest research in the field. Secondly and equally important, the support that we provide to the users extends well beyond simply the introduction of input data to have the software running. We very often accompany our advice by recommendations about the best way to model the structure or even about general concepts linked to structural fire safety engineering.


  • What kind of upgrade is envisioned for the next version?


The next version will have some new material models. Notably, the revised Eurocodes will come out shortly, and we will update the models in SAFIR to align with the modifications of the code with no delay (Jean-Marc Franssen was the chairman of the draft team of Horizontal Fire Group for the Eurocode revision). We will also implement developments stemming from our recent research, including on probabilistic structural fire engineering and on CFD-FEM interfacing. Finally, we have a long wish list of developments, with many great suggestions from users, and we try to cross an item out of the list whenever we have the opportunity!

bottom of page