Improved modelling of fire

Researcher: Mahmoud Assad

Supervisor(s): Dr. R Prosser and Dr. A. Revell
Sponsor: The University of Manchester, and EDF
Start Date: September 2009 End Date: September 2013
Keywords: Fire Modelling, non-premixed combustion, Code_Saturne, %$C_{as}$% model, URANS

Overall Research Aim

This project basically aims to develop a simplified %$C_{as}$% model in Code_Saturne to predict fire behaviour. The objective of this research is also to capture some of the anisotropy and unsteadiness effects in fire which are not captured in the classical linear EVM models.

Research Progress


Fires are unwanted and uncontrolled combustion events that occur everyday around the world and cause immense material damages and loss of life. For example, in
the United Kingdom alone, the direct fire losses exceeded 1600 million and more than 515 people die in 438 thousand fire events have been reported for year 2006. Fire is one of the main risks in property and environment, especially in nuclear power plants wherein fire is considered as the main threat to safety. Hence, the study of fire’s behaviour and its physics play significant roles in fire protection engineering and building design engineering. Hence, the study of fire physics plays a significant role in reducing fire risks. The prediction of fire and smoke behaviour has been basically investigated using four methods; full-scale fire experience, analytical studies (zone model studies), fine-scale numerical modelling (CFD), and physical modelling at reduced scale ( Meroney, 2007). In the past 30 years, field modelling (CFD) has been extensively used in fire prediction. More than one hundred and sixty computer models for fire and smoke studies have been identified by Olenick and Carpenter (2003) in their survey study. Since 2003, more computer models have been developed due to the large growth in computing facilities and CFD models.

Fire codes generally use LES model or RANS model to simulate the fire behaviour. Most LES and RANS models are expensive methods and do not quantify the heat release from the fire chemistry ( Torero and Steinhaus, 2004). Thus, an alternative method will be used in our project “Improved modelling of fire”. The improved modelling of fire is a collaborative project between EDF and University of Manchester. This project aims to develop a simplified %$C_{as}$% model in Code_Saturne to predict fire behaviour. The suggested model basically codifies the misalignment between stress and strain fields was developed by Revell, (2006) for non-reacting cold flows (constant density)

Project Progress

The project started in October 2009 with supervisor Dr. R. Prosser and co-supervisor Dr. A. Revell. The work in this project has the main steps as follows:

  • A basic literature survey will be performed including theoretical background for turbulence modelling and the existing turbulent non-premixed models. This survey will be extended to include a critical literature survey for fire modelling.
  • The %$C_{as}$% model for cold flow will be used as a basis to develop a %$C_{as}$% model for reacting flows such as fire. The resulting model will include the heat release from chemical.
  • The %$C_{as}$% model will be coupled with simple non-premixed combustion models such as the flamelet model.
  • The coupled models will be implemented in Code_Saturne.
  • The model will be applied to some standard test cases to compare its performance with existing numerical and experimental models.

Test Case

Turbulent Diffusion Flame

Sandia flame is a simple turbulent diffusion flame which is used as a initial test case for combustion simulation in Code_Saturne. In this test case, Propane (fuel) and air (oxidizer) enter a cylindrical combustion chamber separately with different flow rate, as shown in Fig.1 . The simulation done using version V1.4.0 of Code_Saturne. the 2D temperature profile for different turbulence models are shown in Fig 2. The results are compared with experimental data and computational data resulted from different version of Code_Saturne such as in Fig 3.



Fig.1: Schematic diagram of plane furnace( 3D Co-flow ) Fig.2: Visualization of flame stream line( temperature profile via different turbulence models)

Fig.3: Distribution of temperature at r=0 line (via different turbulence models)

Steckler Room

The Steckler room test case is a large and real fire experiment that has been investigated by the National Institute of Standards and Technology (NIST) ( Steckler et. al, 1982). Code_Saturne was used to simulate the fire inside the Steckler room, with fire strength 105.3 KW established in a 30-cm diameter diffusion burner as shown in Fig.4. The fire was simulated for 2000 second, and mean profiles for temperature and velocity were obtained as in Fig.5, Fig.6:

Fig.4: (a) Schematic drawing of the Steckler compartment fire (b) top view of Steckler compartment

tempy140mm-allv140.png veloy140mm-allv140.png
Fig.5: Measured and predicted gas temperatures on doorway centreline for different turbulence models Fig.6: Measured and predicted gas velocity on doorway centreline for different turbulence model

Fig.7: The predicted velocity vector and the Mixture fraction for the k-eps model.

Last Modification: r23 - 2013-02-27 - 18:13:26 - MahmoudAssad

Current Tags:
Code_Saturne_res_topic1Add my vote for this tag create new tag
, view all tags
Topic attachments
I Attachment Action Size Date Who Comment
pngpng Schematic_diagram_of_plane_furnace.png manage 41.1 K 2010-12-02 - 17:39 MahmoudAssad  
pngpng Temp.0009.png manage 21.1 K 2010-04-14 - 09:55 MahmoudAssad  
pngPNG co-flow.PNG manage 28.9 K 2010-04-12 - 15:46 MahmoudAssad  
pngpng computional_domin-1.png manage 152.2 K 2010-12-02 - 17:23 MahmoudAssad  
gifgif mixture_fraction.gif manage 28.3 K 2011-02-03 - 15:08 MahmoudAssad  
pngpng tempr0-V140stnd.png manage 22.8 K 2010-12-02 - 13:57 MahmoudAssad  
pngpng tempy140mm-allv140.png manage 21.2 K 2010-12-02 - 17:26 MahmoudAssad  
pngpng veloy140mm-allv140.png manage 22.2 K 2010-12-02 - 17:27 MahmoudAssad  
pngpng visulization_of_flame_stream_line.png manage 61.9 K 2010-12-02 - 13:56 MahmoudAssad  
Topic revision: r23 - 2013-02-27 - 18:13:26 - MahmoudAssad

Computational Fluid Dynamics and Turbulence Mechanics
@ the University of Manchester
Copyright & by the contributing authors. Unless noted otherwise, all material on this web site is the property of the contributing authors.