Navigation and service


For modelling flame spread, pyrolysis is a major process that needs to be modelled. It describes the gasification of a solid, where the emitted combustible gases feed the flame above. The energy needed for pyrolysis is provided by the flame itself, as the released heat of combustion is also transferred back to the solid.

Our contribution to this research field is twofold:

  • We conduct small scale experiments, i.e. thermogravimetric analysis (TGA), cone calorimeter and tube furnace, to analyse and characterise the pyrolysis process.
  • In order to model pyrolysis, material parameters, like heat capacity or reaction parameters, are needed. As they cannot be directly measured for the materials of interest, we develop an inverse modelling framework to implicitly derive them.


During pyrolysis volatiles from the sample are released into the gas phase. The released species react with oxygen which starts the burning process. In order to understand this process, it is crucial to understand which specimen are taking part in this process. Therefore, it is of importance to determine what is released from the sample into the gas phase during pyrolysis.

By heating the solid under an inert atmosphere, it is prevented that the pyrolysis products immediately oxidise. This allows the determination of the pyrolysis products. In these experiments the energy needed for pyrolysis does not originate from a flame but from an external heat source.

We conduct pyrolysis experiments in the thermogravimetric analyser (TGA) and in the tube furnace. The TGA allows to heat small scales samples (order of mg) under a chosen atmosphere, while measuring the mass loss of this samples. The combination of a TGA with a mass spectrometer allows to determine which species are released during pyrolysis. Large scale experiments are conducted in a tube furnace. This experimental set-up allows to heat samples of the order of several 100’s of grams.

Fire Dynamics Research Topic PyrolysisTube furnance used for experimental investigation of generic and cable samples.

Inverse modelling framework PROPTI

When conducting fire simulations, a decision on what material parameter values to use, either for combustible materials or structural elements of buildings, is needed. As often a complete set of information is not available, textbook values are used. However, these values are mostly generalised and are provided as a value range. Thus, the question arises, which value, or set thereof, describes the behaviour of the particular specimen of interest. Furthermore, an optimal parameter set does not guarantee a perfect match of the target data as the respective simulation model may not be able to reproduce the expected behaviour, due to limitations and simplifications of the model. In order to derive material parameters based on small scale experiments, we created a tool called PROPTI. It follows an inverse modelling approach to find effective parameters to represent the observed material behaviour.

Bild2General idea of direct vs. inverse modelling. While in direct modelling the parameters are explicitly state, the inverse modelling appraoch estimates the model parameter based on target, here experimental, data.

PROPTI is an open source tool implemented in Python. With its generalised formulation, it is tailored to enable communication between arbitrary simulation models, different optimisation algorithms, as well as various experimental data series as optimisation targets. The framework aims to facilitate high performance computing resources, via multi-threading and the Message PassingInterface (MPI), in order to speed up the overall process.

Bild3Workflow of the PROPTI framework. The optimiser used in PROPTI is SCE-UA (shuffled complex evolution) provided by the SPOTPY library. A common scenario is to use cone calorimeter data as target data.

Movie3Example of fitness and parameter evolution (left) and the adoption of the model respose to the target data (right).

Cable Fires

Various technical buildings, like power plants, research facilities, or transportation systems, need large amount of cables. In case of fire, these cables can lead to a fast flame spread, e.g. across technical galeries. These fires pose a modelling challenge due to their complex geometrical and chemical structure. Each component, i.e. jacket, filling material and insulator, has its own properties that need to be determined. In the conducted studies, bench scale experiments of the CHRISTIFIRE project are used to derive material parameters using PROPTI. The resulting parameter sets are then used for simulation of real scale experiments of cable trays. The setups focus on the experiments done in CHRISTIFIRE.

Movie4Flame spread across multiple cable trays. The pyrolysis model is based on material parameters only.

Related Publications

Numerical Fire Spread Simulation Based on Material Pyrolysis—An Application to the CHRISTIFIRE Phase 1 Horizontal Cable Tray Tests
Hehnen, T.; Arnold, L.; Mendola, S.
Fire 3, 2020, [10.3390/fire3030033]

Application cases of inverse modelling with the PROPTI framework
Arnold, L. ; Hehnen, T. ; Lauer, P. ; Trettin, C. ; Vinayak, A.
Fire safety journal 108, 102835 - (2019) [10.1016/j.firesaf.2019.102835]

PROPTI – A Generalised Inverse Modelling Framework
Arnold, L. ; Hehnen, T. ; Lauer, P. ; Trettin, C. ; Vinayak, A.
Third European Symposium on Fire Safety Sciences, ESFSS2018, Nancy, France, 12 Sep 2018 - 14 Sep 2018 6 pp. (2018)

Simulation of Fire Propagation in Cable Tray Installations for Particle Accelerator Facility Tunnels
Hehnen, T. ; Arnold, L. ; van Hees, P. ; La Mendola, S.
Proceedings from the 8th International Symposium on Tunnel Safety and Security, Eighth International Symposium on Tunnel Safety and Security, ISTSS 2018, Borås, Sweden, 14 Mar 2018 - 16 Mar 2018 Stockholm : RISE Research Institutes of Sweden AB 503 - 514 (2018)