Intro to Fire Sim

An introduction toFire Simulationwith FDS and Smokeview

Emanuele Gissi

Updated to FDS 5.5.1 and BlenderFDS

Development

Download a revised version of this document from:

http://www.emanuelegissi.eu

You can participate in the development of this document. Please, contact meat:

[email protected]

This manual was produced using LYX and OpenOffice.org on Ubuntu Linux. TheLATEX file intro_to_fire_sim.tex was compiled on July 16, 2010.

Copyright notice

This work is licensed under the Creative Commons Attribution-Share Alike 3.0License. To view a copy of this license, visit:

http://creativecommons.org/licenses/by-sa/3.0/

You are free to use, share, adapt this work. Remember to cite the sources andto use the same open license for derivative work.

About the Author

Emanuele Gissi ([email protected]) is a fire officer since 2002, servingat the Comando provinciale dei Vigili del Fuoco, Genova (Italy), a branch ofthe governmental fire safety national organization (http://www.vigilfuoco.it/).He is a mechanical engineer and obtained a doctorate in EngineeringPhysics (EP) in 2001. His work is focused on incident command, re-view of performance based fire safety design, and fire investigation. Heprovides basic and advanced training on fire simulation to fire officers andprofessionals.

iii

Preface

This manual was born as a small tutorial for students of fire safety engineeringcourses. Then it grew to the current state.The main goal of this manual is to introduce the student to the complex world offire simulation with Fire Dynamics Simulator and Smokeview, and complementsthe official documentation. The official documentation remains an invaluablesource of reference for advanced users and this manual is heavily based on it.Some large parts are even copy-pasted and adapted.In this manual, topics are organized in a strict logical order and the basics arethoroughly explained to improve the learning curve. Some advanced topics arecompletely omitted for the sake of simplicity.According to teaching experience, students understand the logic behind the FireDynamics Simulator and become autonomous learners after 16 hours of course:they learn to work independently and are able to develop reasonable engineeringlevel applications by themselves.

v

Acknowledgments

People

The National Institute of Standards and Technology (NIST), a federal agencywithin the Department of Commerce of the United States, is the major drivingforce behind Fire Dynamics Simulator development.The Fire Dynamics Simulator has been publicly released on 2000. Since its firstrelease, continued improvements have been made to the software based largelyon feedback from its users and on the hard work of some NIST employees.Fire Dynamics Simulator documentation is written and maintained by KevinMcGrattan, Bryan Klein, Simo Hostikka, and Jason Floyd fromvarious organizations. Their user support through the discussion group is alwayscomplete, fast, and precious. In a simple word: friendly.I owe to them large parts of this manual and most of my knowledge on firesimulation. Thus, I thank them once more.

Ideas

This manual is open content and free, because its published in a format thatexplicitly allows copying and modifying of its information by anyone.Fire Dynamics Simulator itself is free and in the public domain. Linux1 andOpenoffice.Org2 are free and open source.I thank the open source movement as a whole, as the best things in life are free.

1Linux is an open source operating system. See http://www.ubuntu.com/ for an userfriendly Linux distribution.

2Openoffice.org is an open source office and productivity program for Linux, MS Windows,and MacOS X. See http://www.openoffice.org/ for further details.

vii

Contents

1 Introduction 11.1 Complexity of fire phenomena . . . . . . . . . . . . . . . . . . . 11.2 Approaches to fire simulation . . . . . . . . . . . . . . . . . . . 21.3 Simplification . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Inside FDS 72.1 What are FDS ans Smokeview? . . . . . . . . . . . . . . . . . . 72.2 Engineering applications . . . . . . . . . . . . . . . . . . . . . . 82.3 Who develops FDS? . . . . . . . . . . . . . . . . . . . . . . . . 82.4 Who uses FDS? . . . . . . . . . . . . . . . . . . . . . . . . . . 92.5 How much does it cost? . . . . . . . . . . . . . . . . . . . . . . 92.6 How does FDS work? . . . . . . . . . . . . . . . . . . . . . . . 9

2.6.1 Hydrodynamic model . . . . . . . . . . . . . . . . . . . 92.6.2 Combustion model . . . . . . . . . . . . . . . . . . . . . 92.6.3 Radiation transport . . . . . . . . . . . . . . . . . . . . 102.6.4 Geometry and multiple meshes . . . . . . . . . . . . . . 102.6.5 Parallel processing . . . . . . . . . . . . . . . . . . . . . 112.6.6 Boundary conditions . . . . . . . . . . . . . . . . . . . . 11

2.7 Limitations of FDS . . . . . . . . . . . . . . . . . . . . . . . . . 112.7.1 Low speed flow assumption . . . . . . . . . . . . . . . . 112.7.2 Rectilinear geometry . . . . . . . . . . . . . . . . . . . . 112.7.3 Fire growth and spread . . . . . . . . . . . . . . . . . . 122.7.4 Combustion . . . . . . . . . . . . . . . . . . . . . . . . 122.7.5 Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . 13

ix

x CONTENTS

3 Running FDS 153.1 Online resources and user support . . . . . . . . . . . . . . . . . 153.2 Graphical user interface . . . . . . . . . . . . . . . . . . . . . . 163.3 Version numbers . . . . . . . . . . . . . . . . . . . . . . . . . . 163.4 Hardware requirements . . . . . . . . . . . . . . . . . . . . . . . 163.5 Serial and parallel calculations . . . . . . . . . . . . . . . . . . . 173.6 Installing on Windows XP . . . . . . . . . . . . . . . . . . . . . 173.7 Installing on Ubuntu Linux . . . . . . . . . . . . . . . . . . . . . 18

3.7.1 First install . . . . . . . . . . . . . . . . . . . . . . . . . 183.7.2 Installing a new version . . . . . . . . . . . . . . . . . . 20

3.8 Compiling an optimized binary . . . . . . . . . . . . . . . . . . . 203.9 Performing a calculation . . . . . . . . . . . . . . . . . . . . . . 20

3.9.1 Running serial FDS on Windows XP . . . . . . . . . . . . 203.9.2 Running serial FDS on Ubuntu Linux . . . . . . . . . . . 213.9.3 Running parallel FDS on Ubuntu Linux . . . . . . . . . . 21

3.10 Monitoring progress . . . . . . . . . . . . . . . . . . . . . . . . 233.11 Stop a calculation . . . . . . . . . . . . . . . . . . . . . . . . . 233.12 Visualizing results . . . . . . . . . . . . . . . . . . . . . . . . . 233.13 Output files . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

4 Input file basics 274.1 Syntax of the input file . . . . . . . . . . . . . . . . . . . . . . 274.2 Writing an input file . . . . . . . . . . . . . . . . . . . . . . . . 294.3 The logic behind most FDS input files . . . . . . . . . . . . . . 314.4 Keep it simple . . . . . . . . . . . . . . . . . . . . . . . . . . . 344.5 Each model, its input data . . . . . . . . . . . . . . . . . . . . . 354.6 Units of measurement . . . . . . . . . . . . . . . . . . . . . . . 364.7 Reference coordinate system . . . . . . . . . . . . . . . . . . . . 364.8 Prescribing geometric entities . . . . . . . . . . . . . . . . . . . 374.9 Prescribing orientations . . . . . . . . . . . . . . . . . . . . . . 384.10 Prescribing colors and aspect . . . . . . . . . . . . . . . . . . . 39

CONTENTS xi

5 General configuration 415.1 Naming the job, HEAD . . . . . . . . . . . . . . . . . . . . . . . 415.2 Simulation time, TIME . . . . . . . . . . . . . . . . . . . . . . . 425.3 Miscellaneous, MISC . . . . . . . . . . . . . . . . . . . . . . . . 42

6 Combustion and radiation 456.1 Combustion is not pyrolysis . . . . . . . . . . . . . . . . . . . . 456.2 Prescribing a fire . . . . . . . . . . . . . . . . . . . . . . . . . . 456.3 Modeling gas phase combustion, REAC . . . . . . . . . . . . . . 46

6.3.1 Ignition . . . . . . . . . . . . . . . . . . . . . . . . . . . 466.3.2 Burning . . . . . . . . . . . . . . . . . . . . . . . . . . 47

6.4 CO production in under-ventilated fires . . . . . . . . . . . . . . 516.5 Flame extinction . . . . . . . . . . . . . . . . . . . . . . . . . . 516.6 Radiation transport, RADI . . . . . . . . . . . . . . . . . . . . . 52

7 Computational domain 557.1 Defining a mesh, MESH . . . . . . . . . . . . . . . . . . . . . . . 557.2 Multiple meshes . . . . . . . . . . . . . . . . . . . . . . . . . . 577.3 Conformity to the mesh . . . . . . . . . . . . . . . . . . . . . . 587.4 Choosing the right mesh dimension:

a sensitivity study . . . . . . . . . . . . . . . . . . . . . . . . . 597.5 Initial conditions of the computational domain, INIT . . . . . . . 60

8 Materials 638.1 Defining a material, MATL . . . . . . . . . . . . . . . . . . . . . 638.2 Thermal properties . . . . . . . . . . . . . . . . . . . . . . . . . 648.3 Burning properties . . . . . . . . . . . . . . . . . . . . . . . . . 65

8.3.1 Solids . . . . . . . . . . . . . . . . . . . . . . . . . . . . 658.3.2 Liquids . . . . . . . . . . . . . . . . . . . . . . . . . . . 668.3.3 HEAT_OF_COMBUSTION in a MATL line? . . . . . . . . . . 67

8.4 Properties hell . . . . . . . . . . . . . . . . . . . . . . . . . . . 678.5 Resources for material property data . . . . . . . . . . . . . . . 68

xii CONTENTS

9 Extra gas species 69

9.1 Defining extra gas species, SPEC . . . . . . . . . . . . . . . . . . 69

9.2 CARBON DIOXIDE and carbon dioxide . . . . . . . . . . . . . 70

10 Lagrangian particles 73

10.1 Defining Lagrangian particles, PART . . . . . . . . . . . . . . . . 73

10.2 Massless particles . . . . . . . . . . . . . . . . . . . . . . . . . 74

10.3 Water droplets . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

11 Boundary conditions 77

11.1 Defining boundary conditions, SURF . . . . . . . . . . . . . . . . 77

11.2 Predefined boundary conditions . . . . . . . . . . . . . . . . . . 79

11.3 Coloring boundary conditions . . . . . . . . . . . . . . . . . . . 80

11.4 Examples of boundary conditions . . . . . . . . . . . . . . . . . 81

11.4.1 Adiabatic surface . . . . . . . . . . . . . . . . . . . . . . 81

11.4.2 Fixed temperature and heat flux . . . . . . . . . . . . . . 81

11.4.3 Fans . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

11.4.4 Fans injecting extra gas species . . . . . . . . . . . . . . 83

11.4.5 Dynamic pressure at an open boundary . . . . . . . . . . 84

11.4.6 Prescribing an heat release rate . . . . . . . . . . . . . . 84

11.5 Geometric conformity and rates . . . . . . . . . . . . . . . . . . 85

11.6 Boundary conditions for solids . . . . . . . . . . . . . . . . . . . 85

11.6.1 Backing . . . . . . . . . . . . . . . . . . . . . . . . . . 86

11.6.2 Setting an initial temperature . . . . . . . . . . . . . . . 88

11.7 Time dependent boundary conditions . . . . . . . . . . . . . . . 88

11.7.1 Simplified ramps . . . . . . . . . . . . . . . . . . . . . . 88

11.7.2 User defined ramps . . . . . . . . . . . . . . . . . . . . 89

11.8 Injecting Lagrangian particles . . . . . . . . . . . . . . . . . . . 91

CONTENTS xiii

12 Solid geometry 9312.1 Defining solid obstructions, OBST . . . . . . . . . . . . . . . . . 9312.2 Creating voids inside obstructions, HOLE . . . . . . . . . . . . . . 9612.3 Prescribing a different boundary condition, VENT . . . . . . . . . 9712.4 Default boundary condition . . . . . . . . . . . . . . . . . . . . 9912.5 How thick is a wall? . . . . . . . . . . . . . . . . . . . . . . . . 10012.6 Thin sheet obstructions . . . . . . . . . . . . . . . . . . . . . . 10112.7 Activating and deactivating objects . . . . . . . . . . . . . . . . 10212.8 Stair stepping complex geometries . . . . . . . . . . . . . . . . . 10212.9 Coloring individual objects . . . . . . . . . . . . . . . . . . . . . 10312.10Making burning solids disappear . . . . . . . . . . . . . . . . . . 104

13 Devices and control logic 10513.1 Devices, DEVC and PROP . . . . . . . . . . . . . . . . . . . . . . 10513.2 Examples of devices . . . . . . . . . . . . . . . . . . . . . . . . 107

13.2.1 Timer . . . . . . . . . . . . . . . . . . . . . . . . . . . 10713.2.2 Thermometer . . . . . . . . . . . . . . . . . . . . . . . 10713.2.3 Smoke detector . . . . . . . . . . . . . . . . . . . . . . 10713.2.4 Beam smoke detector . . . . . . . . . . . . . . . . . . . 10813.2.5 Sprinkler and heat detector . . . . . . . . . . . . . . . . 108

13.3 Basic control logic . . . . . . . . . . . . . . . . . . . . . . . . . 10913.4 Advanced control logic . . . . . . . . . . . . . . . . . . . . . . . 111

14 Output 11314.1 Prescribing output . . . . . . . . . . . . . . . . . . . . . . . . . 11314.2 Output control parameters, DUMP . . . . . . . . . . . . . . . . . 11514.3 Point measurement devices, DEVC . . . . . . . . . . . . . . . . . 115

14.3.1 Record a gas phase quantity . . . . . . . . . . . . . . . . 11614.3.2 Record a solid phase quantity . . . . . . . . . . . . . . . 11614.3.3 Record integrated quantities . . . . . . . . . . . . . . . . 117

xiv CONTENTS

14.4 Animated planar slices, SLCF . . . . . . . . . . . . . . . . . . . 11714.5 Animated boundary quantities, BNDF . . . . . . . . . . . . . . . 11814.6 Animated isosurfaces, ISOF . . . . . . . . . . . . . . . . . . . . 12014.7 Realistic smoke and fire . . . . . . . . . . . . . . . . . . . . . . 12114.8 Heat release rate . . . . . . . . . . . . . . . . . . . . . . . . . . 12214.9 Visibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12214.10Upper and lower layers . . . . . . . . . . . . . . . . . . . . . . . 12314.11Heat fluxes and thermal radiation . . . . . . . . . . . . . . . . . 12414.12Interfacing with structural models . . . . . . . . . . . . . . . . . 12414.13Visualizing Lagrangian particles . . . . . . . . . . . . . . . . . . 12514.14Frequently used output quantities . . . . . . . . . . . . . . . . . 125

15 Using BlenderFDS 12915.1 A faster way to writing input files . . . . . . . . . . . . . . . . . 12915.2 From the Community, to the Community . . . . . . . . . . . . . 13015.3 Online resources . . . . . . . . . . . . . . . . . . . . . . . . . . 131

16 Real world examples 13316.1 Building a ventilator . . . . . . . . . . . . . . . . . . . . . . . . 13316.2 Prescribing a simplified burning material . . . . . . . . . . . . . 13316.3 Simulation and revelation of smoke of a smoldering fire . . . . . 13416.4 A pan filled of ethanol . . . . . . . . . . . . . . . . . . . . . . . 13516.5 A simple car parking . . . . . . . . . . . . . . . . . . . . . . . . 135

16.5.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . 13516.5.2 Input file and results . . . . . . . . . . . . . . . . . . . . 137

16.6 An university building . . . . . . . . . . . . . . . . . . . . . . . 146

List of Figures

3.1 A cluster of Linux computers . . . . . . . . . . . . . . . . . . . 183.2 Starting a serial calculation on Windows XP and on Linux Ubuntu 213.3 Running a parallel calculation and monitoring the system on Ubuntu

Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223.4 Running Smokeview on Windows XP and on Ubuntu Linux . . . . 24

4.1 The structure of an FDS namelist group . . . . . . . . . . . . . 274.2 Modeling reality in FDS . . . . . . . . . . . . . . . . . . . . . . 354.3 The reference system, a volume, a face, a segment, a point, and

a plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

6.1 Combustion is not pyrolysis . . . . . . . . . . . . . . . . . . . . 466.2 Combustion and pyrolysis in a flaming match . . . . . . . . . . . 476.3 Flame extinction criteria . . . . . . . . . . . . . . . . . . . . . . 52

7.1 The computational domain composed by four meshes . . . . . . 577.2 Mesh connections: (a) ideal, (b) allowed, and (c) forbidden . . . 587.3 Geometric object: before and after automatic shifting . . . . . . 58

11.1 Extending the computational domain beyond the vent . . . . . . 8011.2 brick wall: multiple layers of different materials . . . . . . . . 8611.3 HRRPUA as function of time after SURF activation . . . . . . . . . 8911.4 VEL and TMP_FRONT as function of time after SURF activation . . 90

12.1 Boundary conditions prescribed with SURF_ID, SURF_IDS, andSURF_ID6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

xv

xvi LIST OF FIGURES

12.2 OBST, HOLE and VENT . . . . . . . . . . . . . . . . . . . . . . . 9712.3 Setting boundary conditions to exterior boundaries of the com-

putational domain . . . . . . . . . . . . . . . . . . . . . . . . . 9812.4 Flow velocity on two sides of an oblique wall. Lower side has

SAWTOOTH=.FALSE. set. . . . . . . . . . . . . . . . . . . . . . . 103

14.1 Output of animated planar slices SLCF as viewed in Smokeview . 11714.2 Output of animated boundary quantities, BNDF as viewed in Smoke-

view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11914.3 Output of animated isosurfaces, ISOF as viewed in Smokeview . . 12014.4 Output of soot MASS FRACTION and HRRPUV as viewed in Smoke-

view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

15.1 A BlenderFDS session . . . . . . . . . . . . . . . . . . . . . . . 13015.2 A BlenderFDS property panel . . . . . . . . . . . . . . . . . . . 13115.3 Exporting a BlenderFDS model to an FDS input file . . . . . . . 132

16.1 Car parking plan . . . . . . . . . . . . . . . . . . . . . . . . . . 13616.2 The entered geometry . . . . . . . . . . . . . . . . . . . . . . . 14116.3 Heat release rate curve . . . . . . . . . . . . . . . . . . . . . . . 14116.4 Thermocouples at the center of the car parking. . . . . . . . . . 14216.5 Gas temperatures in front of window panes vs time. . . . . . . . 14216.6 FED vs time. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14316.7 Layer height vs time. . . . . . . . . . . . . . . . . . . . . . . . . 14316.8 AST (Adiabatic Surface Temperature) boundary file at 1200 s. . . 14416.9 Visibility slice file at 120 s. . . . . . . . . . . . . . . . . . . . . . 14416.10Visibility (10 m) isosurface at 300 s. . . . . . . . . . . . . . . . . 14516.11Net heat flux on boundary surfaces at 1200 s. . . . . . . . . . . 14516.12The university building as modeled in BlenderFDS . . . . . . . . 14616.13The exported FDS input file, as shown by Smokeview . . . . . . 146

List of Tables

4.1 Systematic organisation of the input file . . . . . . . . . . . . . . 334.2 COLOR values . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

5.1 HEAD parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 425.2 TIME parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 425.3 MISC parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 43

6.1 Mixture fraction species . . . . . . . . . . . . . . . . . . . . . . 496.2 REAC parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 496.3 RADI parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 53

7.1 IJK values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 567.2 MESH parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 567.3 INIT parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 61

8.1 MATL parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 638.2 RAMP (temperature) parameters . . . . . . . . . . . . . . . . . . 65

9.1 Predefined extra species . . . . . . . . . . . . . . . . . . . . . . 699.2 SPEC parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 70

10.1 PART parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 74

11.1 SURF parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 7811.2 RAMP (time) parameters . . . . . . . . . . . . . . . . . . . . . . 90

xvii

xviii LIST OF TABLES

12.1 OBST parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 9512.2 HOLE parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 9612.3 VENT parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 99

13.1 DEVC parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 106

14.1 DUMP parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 11514.2 Output of devices in mycase_devc.csv file . . . . . . . . . . . . 11514.3 SLCF parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 11814.4 BNDF parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 11914.5 ISOF parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 12114.6 Output of HRR in mycase_hrr.csv file . . . . . . . . . . . . . 12214.7 Frequently used output quantities . . . . . . . . . . . . . . . . . 125

Chapter 1

Introduction

This chapter presents the problems embedded in fire simulation. Thetext has been adapted from [FDS5 users guide], Introduction byHoward Baum, NIST Fellow Emeritus.

1.1 Complexity of fire phenomena

The idea that the dynamics of a fire might be studied numerically dates backto the beginning of the computer age. Indeed, the fundamental conservationequations governing fluid dynamics, heat transfer, and combustion were firstwritten down over a century ago. Despite this, practical mathematical modelsof fire, as distinct from controlled combustion, are relatively recent due to theinherent complexity of the problem.Indeed, in his brief history of the early days of fire research, Hoyt Hottelnoted:

A case can be made for fire being, next to the life processes, themost complex of phenomena to understand [Hottel 1984].

The difficulties revolve about three issues:

First, there are an enormous number of possible fire scenarios to considerdue to their accidental nature.

Second, the physical insight and computing power required to perform allthe necessary calculations for most fire scenarios are limited. Any funda-mentally based study of fires must consider at least some aspects of bluff

1

2 CHAPTER 1. INTRODUCTION

body aerodynamics, multi-phase flow, turbulent mixing and combustion,radiative transport, and conjugate heat transfer; all of which are activeresearch areas in their own right.

Finally, the fuel in most fires was never intended as such.

Thus, the mathematical models and the data needed to characterize the degrada-tion of the condensed phase materials that supply the fuel may not be available.Indeed, the mathematical modeling of the physical and chemical transformationsof real materials as they burn is still in its infancy.In order to make progress, the questions that are asked have to be greatly sim-plified.To begin with, instead of seeking a methodology that can be applied to all fireproblems, we begin by looking at a few scenarios that seem to be most amenableto analysis. Hopefully, the methods developed to study these simple problemscan be generalized over time so that more complex scenarios can be analyzed.Second, we must learn to live with idealized descriptions of fires and approximatesolutions to our idealized equations.Finally, the methods should be capable of systematic improvement. As our phys-ical insight and computing power grow more powerful, the methods of analysiscan grow with them.

1.2 Approaches to fire simulation

To date, four distinct approaches to the simulation of fires have emerged. Eachof these treats the fire as an inherently three dimensional process evolving intime.The first to reach maturity, the zone models, describe compartment fires. EachZone modelscompartment is divided into two spatially homogeneous volumes, a hot upperlayer and a cooler lower layer. Mass and energy balances are enforced for eachlayer, with additional models describing other physical processes appended asdifferential or algebraic equations as appropriate. Examples of such phenomenainclude fire plumes, flows through doors, windows and other vents, radiativeand convective heat transfer, and solid fuel pyrolysis. Model development hasprogressed to the point where documented and supported software implementingthese models are widely available, such as CFAST1.

1The Consolidated Model of Fire and Smoke Transport (CFAST) is a two-zone fire modelused to calculate the evolving distribution of smoke, fire gases and temperature throughoutcompartments of a building during a fire. Visit http://cfast.nist.gov/ for further information.

1.2. APPROACHES TO FIRE SIMULATION 3

The relative physical and computational simplicity of the zone models has led totheir widespread use in the analysis of fire scenarios. So long as detailed spatialdistributions of physical properties are not required, and the two layer descriptionreasonably approximates reality, these models are quite reliable. However, by theirvery nature, there is no way to systematically improve them.The rapid growth of computing power and the corresponding maturing of com-putational fluid dynamics (CFD), has led to the development of CFD based fieldmodels applied to fire research problems. Virtually all this work is based on theconceptual framework provided by the Reynolds-averaged form of the Navier-Stokes equations (RANS). The use of CFD models has allowed the description RANSof fires in complex geometries, and the incorporation of a wide variety of physicalphenomena.However, these models have a fundamental limitation for fire applications theaveraging procedure at the root of the model equations.RANS models were developed as a time-averaged approximation to the conser-vation equations of fluid dynamics. While the precise nature of the averagingtime is not specified, it is clearly long enough to require the introduction of largeeddy transport coefficients to describe the unresolved fluxes of mass, momentumand energy. This is the root cause of the smoothed appearance of the resultsof even the most highly resolved fire simulations. The smallest resolvable lengthscales are determined by the product of the local velocity and the averaging timerather than the spatial resolution of the underlying computational grid.Unfortunately, the evolution of large eddy structures characteristic of most fireplumes is lost with such an approach, as is the prediction of local transient events.It is sometimes argued that the averaging process used to define the equations isan ensemble average over many replicates of the same experiment or postulatedscenario. However, this is a moot point in fire research since neither experimentsnor real scenarios are replicated in the sense required by that interpretation ofthe equations.The application of Large Eddy Simulation (LES) techniques to fire is aimed at LESextracting greater temporal and spatial fidelity from simulations of fire performedon the more finely meshed grids allowed by ever faster computers.The phrase LES refers to the description of turbulent mixing of the gaseousfuel and combustion products with the local atmosphere surrounding the fire.This process, which determines the burning rate in most fires and controls thespread of smoke and hot gases, is extremely difficult to predict accurately. Thisis true not only in fire research but in almost all phenomena involving turbulentfluid motion. The basic idea behind the LES technique is that the eddies thataccount for most of the mixing are large enough to be calculated with reasonable


accuracy from the equations of fluid dynamics. The hope (which must ultimatelybe justified by comparison to experiments) is that small-scale eddy motion caneither be crudely accounted for or ignored.The fourth approach is Direct Numerical Simulation (DNS). DNS is a simula-DNStion in computational fluid dynamics in which the Navier-Stokes equations arenumerically solved without any turbulence model. This means that the wholerange of spatial and temporal scales of the turbulence must be resolved in thecomputational mesh.The computational cost of DNS is very high, even at low Reynolds numbers.For the Reynolds numbers encountered in most industrial applications, the com-putational resources required by a DNS would exceed the capacity of the mostpowerful computers currently available.

1.3 Simplification

The equations describing the transport of mass, momentum, and energy by thefire-induced flows must be simplified so that they can be efficiently solved for thefire scenarios of interest. The general equations of fluid dynamics describe a richvariety of physical processes, many of which have nothing to do with fires.Retaining this generality would lead to an enormously complex computationaltask that would shed very little additional insight on fire dynamics. The simplifiedequations, developed by Rehm and Baum [Rehm 1978], have been widely adoptedby the larger combustion research community, where they are referred to as thelow Mach number combustion equations. They describe the low speed motionof a gas driven by chemical heat release and buoyancy forces.The low Mach number equations are solved numerically by dividing the physicalspace where the fire is to be simulated into a large number of rectangular cells.Within each cell the gas velocity, temperature, etc., are assumed to be uniform;changing only with time. The accuracy with which the fire dynamics can besimulated depends on the number of cells that can be incorporated into thesimulation. This number is ultimately limited by the computing power available.Present day, single processor desktop computers limit the number of such cellsto at most a few million. This means that the ratio of largest to smallest eddylength scales that can be resolved by the computation (the dynamic range of thesimulation) is on the order of 100. Parallel processing can be used to extend thisrange to some extent, but the range of length scales that need to be accountedfor if all relevant fire processes are to be simulated is roughly 104 to 105 because

1.3. SIMPLIFICATION 5

combustion processes take place at length scales of 1mm or less, while the lengthscales associated with building fires are of the order of tens of meters.

Chapter 2

Inside FDS

This chapter gives a concise insight on how Fire Dynamics Simula-tor works and its limitations. Possible engineering applications arepresented.

2.1 What are FDS ans Smokeview?

Fire Dynamics Simulator, version 5 (FDS) is a computational fluid dynamics FDS(CFD) model of fire-driven fluid flow. The model solves numerically a form ofthe Navier-Stokes equations appropriate for low-speed, thermally-driven flow withan emphasis on smoke and heat transport from fires. The partial derivatives ofthe conservation equations of mass, momentum and energy are approximatedas finite differences, and the solution is updated in time on a three-dimensional,rectilinear grid. Thermal radiation is computed using a finite volume techniqueon the same grid as the flow solver. Lagrangian particles are used to simulatesmoke movement, sprinkler discharge, and fuel sprays.Smokeview is a companion program to FDS that produces images and animations Smokeviewof the results. Smokeview is able to visualize fire and smoke in a fairly realisticway. Via its three-dimensional realistic renderings, Smokeview is an integralpart of the physical model, as it allows one to assess the visibility within a firecompartment in ways that ordinary scientific visualization software cannot.Although not part of the FDS/Smokeview suite maintained at NIST, there areseveral third-party and proprietary add-ons to FDS either available commerciallyor privately maintained by individual users.The first version of FDS was publicly released in February 2000.

7

8 CHAPTER 2. INSIDE FDS

2.2 Engineering applications

Throughout its development, FDS has been aimed at solving practical fire prob-lems in fire protection engineering, while at the same time providing a tool tostudy fundamental fire dynamics and combustion.It is generally recognized that FDS can be effectively used in engineering appli-Basic usecations to model the following phenomenas:

Low speed transport of heat and combustion products from fire;

Radiative and convective heat transfer between the gas and solid surfaces;

Sprinkler, heat detector, and smoke detector activation.

FDS can be used to model the following problems, too:Advanced use

Pyrolysis;

Flame spread and fire growth;

Sprinkler sprays and suppression by water.

Currently, the users interested in engineering applications should probably avoidusing FDS to model the latter problems, as they are still subject to intenseresearch study in academic environments.To date, about half of the applications of the model have been for design ofsmoke control systems and sprinkler/detector activation studies. The other halfconsist of residential and industrial fire reconstructions.

2.3 Who develops FDS?

Currently, FDS is maintained by the Building and Fire Research Laboratory(BFRL) of National Institute of Standards and Technology. The developers atNIST have formed a loose collaboration of interested stakeholders, including:

VTT Technical Research Centre of Finland, a research and testing labora-tory similar to NIST;

The Society of Fire Protection Engineers (SFPE);

Fire protection engineering firms that use the software;

Engineering departments at various universities with a particular emphasison fire.

2.4. WHO USES FDS? 9

2.4 Who uses FDS?

The use of fire models currently extends beyond the fire research laboratoriesinto the engineering, fire service and legal communities.FDS is intended for use only by those competent in the fields of fluid dynamics,thermodynamics, heat transfer, combustion, and fire science, and is intendedonly to supplement the informed judgment of the qualified user. The softwarepackage is a computer model that may or may not have predictive capability whenapplied to a specific set of factual circumstances. Lack of accurate predictionsby the model could lead to erroneous conclusions with regard to fire safety.Sufficient evaluation of any model is necessary to ensure that users can judge theadequacy of its technical basis, appropriateness of its use, and confidence levelof its predictions.

2.5 How much does it cost?

FDS is free and its source code is in the public domain.

2.6 How does FDS work?

A brief description of the major features of FDS follows.

2.6.1 Hydrodynamic model

FDS solves numerically a form of the Navier-Stokes equations appropriate for low-speed, thermally-driven flow with an emphasis on smoke and heat transport fromfires. The core algorithm is an explicit predictor-corrector scheme, second orderaccurate in space and time. Turbulence is treated by means of the Smagorinskyform of Large Eddy Simulation (LES). It is possible to perform a Direct NumericalSimulation (DNS) if the underlying numerical mesh is fine enough. LES is thedefault mode of operation.

2.6.2 Combustion model

For most applications, FDS uses a single step chemical reaction whose productsare tracked via a two-parameter mixture fraction model. The mixture fraction


is a conserved scalar quantity that represents the mass fraction of one or morecomponents of the gas at a given point in the flow field. By default, two com-ponents of the mixture fraction are explicitly computed. The first is the massfraction of unburned fuel and the second is the mass fraction of burned fuel, asfor example the mass of the combustion products that originated as fuel.A two-step chemical reaction with a three parameter mixture fraction decomposi-tion can also be used with the first step being oxidation of fuel to carbon monoxideand the second step the oxidation of carbon monoxide to carbon dioxide. Thethree mixture fraction components for the two step reaction are unburned fuel,mass of fuel that has completed the first reaction step, and the mass of fuel thathas completed the second reaction step. The mass fractions of all of the majorreactants and products can be derived from the mixture fraction parameters bymeans of state relations.Lastly, a multiple-step finite rate model is also available.

2.6.3 Radiation transport

Radiative heat transfer is included in the model via the solution of the radiationtransport equation for a gray gas, and in some limited cases using a wide bandmodel.The equation is solved using a technique similar to finite volume methods forconvective transport, thus the name given to it is the Finite Volume Method(FVM). Using approximately 100 discrete angles, the finite volume solver requiresabout 20% of the total CPU time of a calculation, a modest cost given thecomplexity of radiation heat transfer.The absorption coefficients of the gas-soot mixtures are computed using [Grosshandler 1993]narrow-band model. Liquid droplets can absorb and scatter thermal radiation.This is important in cases involving mist sprinklers, but also plays a role in allsprinkler cases.

2.6.4 Geometry and multiple meshes

FDS approximates the governing equations on a rectilinear mesh. Rectangularobstructions are forced to conform with the underlying mesh.It is possible to prescribe more than one rectangular mesh to handle cases wherethe computational domain is not easily embedded within a single mesh.

2.7. LIMITATIONS OF FDS 11

2.6.5 Parallel processing

It is possible to run an FDS calculation on more than one computer using theMessage Passing Interface (MPI).

2.6.6 Boundary conditions

All solid surfaces are assigned thermal boundary conditions, plus informationabout the burning behavior of the material. Heat and mass transfer to andfrom solid surfaces is usually handled with empirical correlations, although it ispossible to compute directly the heat and mass transfer when performing a DirectNumerical Simulation (DNS).

2.7 Limitations of FDS

Although FDS can address most fire scenarios, there are limitations in all of itsvarious algorithms. Some of the more prominent limitations of the model arelisted here.

2.7.1 Low speed flow assumption

The use of FDS is limited to low-speed1 flow with an emphasis on smoke and heattransport from fires. This assumption rules out using the model for any scenarioinvolving flow speeds approaching the speed of sound, such as explosions, chokeflow at nozzles, and detonations.

2.7.2 Rectilinear geometry

The efficiency of FDS is due to the simplicity of its rectilinear numerical grid andthe use of a fast, direct solver for the pressure field. This can be a limitation insome situations where certain geometric features do not conform to the rectan-gular grid, although most building components do. There are techniques in FDSto lessen the effect of sawtooth obstructions used to represent non rectangularobjects, but these cannot be expected to produce good results if, for example,the intent of the calculation is to study boundary layer effects. For most practical

1Mach numbers less than about 0.3


large-scale simulations, the increased grid resolution afforded by the fast pressuresolver offsets the approximation of a curved boundary by small rectangular gridcells.

2.7.3 Fire growth and spread

Because the model was originally designed to analyze industrial-scale fires, it canbe used reliably when the Heat Release Rate (HRR) of the fire is specified and thetransport of heat and exhaust products is the principal aim of the simulation. Inthese cases, the model predicts flow velocities and temperatures to an accuracywithin 10% to 20% of experimental measurements, depending on the resolutionof the numerical grid2.However, for fire scenarios where the heat release rate is predicted rather thanspecified, the uncertainty of the model is higher. There are several reasons forthis:

1. Properties of real materials and real fuels are often unknown or difficult toobtain;

2. The physical processes of combustion, radiation and solid phase heat trans-fer are more complicated than their mathematical representations in FDS;

3. The results of calculations are sensitive to both the numerical and physicalparameters. Current research is aimed at improving this situation, but itis safe to say that modeling fire growth and spread will always require ahigher level of user skill and judgment than that required for modeling thetransport of smoke and heat from specified fires.

2.7.4 Combustion

For most applications, FDS uses a mixture fraction-based combustion model.The mixture fraction is a conserved scalar quantity that is defined as the fractionof gas at a given point in the flow field that originated as fuel. In its simplestform, the model assumes that combustion is mixing-controlled, and that thereaction of fuel and oxygen is infinitely fast, regardless of the temperature.

2It is extremely rare to find measurements of local velocities and temperatures from fireexperiments that have reported error estimates that are less than 10%. Thus, the most accuratecalculations using FDS do not introduce significantly greater errors in these quantities thanthe vast majority of fire experiments.

2.7. LIMITATIONS OF FDS 13

For large-scale, well-ventilated fires, this is a good assumption. However, if a fireis in an under-ventilated compartment, or if a suppression agent like water mistor CO2 is introduced, fuel and oxygen are allowed to mix and not burn, accordingto a few empirically-based criteria.The physical mechanisms underlying these phenomena are complex, and are tiedclosely to the flame temperature and local strain rate, neither of which are readily-available in a large scale fire simulation.Subgrid-scale modeling of gas phase suppression and extinction is still an area ofactive research in the combustion community.Until reliable models can be developed for building-scale fire simulations, simpleempirical rules are used by FDS that prevent burning from taking place when theatmosphere immediately surrounding the fire cannot sustain the combustion.

2.7.5 Radiation

Radiative heat transfer is included in the model via the solution of the radiationtransport equation (RTE) for a gray gas, and in some limited cases using awide band model. The RTE is solved using a technique similar to finite volumemethods for convective transport, thus the name given to it is the Finite VolumeMethod (FVM).There are several limitations of the model:

First, the absorption coefficient for the smoke-laden gas is a complex func-tion of its composition and temperature. Because of the simplified com-bustion model, the chemical composition of the smokey gases, especiallythe soot content, can effect both the absorption and emission of thermalradiation.

Second, the radiation transport is discretized via approximately 100 solidangles, although the user may choose to use more angles. For targets faraway from a localized source of radiation, like a growing fire, the discretiza-tion can lead to a non-uniform distribution of the radiant energy. This erroris called ray effect and can be seen in the visualization of surface tem-peratures, where hot spots show the effect of the finite number of solidangles. The problem can be lessened by the inclusion of more solid angles,but at a price of longer computing times. In most cases, the radiative fluxto far-field targets is not as important as those in the near-field, wherecoverage by the default number of angles is much better.

Chapter 3

Running FDS

This chapter explains how to obtain, install and run FDS and Smoke-view. The explanation covers Ubuntu Linux and Windows XP.

3.1 Online resources and user support

The primary resource for detailed instructions on how to download executables,manuals, source code and related utilities, is the FDS-SMV website:

http://fire.nist.gov/fds

FDS has two separate manuals:

[FDS5 technical reference];

[FDS5 users guide].

The [FDS5 technical reference] guide is broken into three volumes:

Mathematical model;

Verification;

Experimental validation.

Smokeview has its own manual:

[Smokeview users guide].

15

16 CHAPTER 3. RUNNING FDS

The FDS and Smokeview users guides only describe the mechanics of usingthe computer programs. The technical reference guides provides the theory,algorithm details, and verification and validation work.Along with the FDS manuals, there are resources available on the Internet. Theseinclude an issue tracker, that allows you to report bugs, feature requests and askspecific clarifying questions, and group discussions, which support more generaltopics than just specific problems.Before using these on-line resources, it is important to first try to solve your ownproblems by performing simple test calculations, or debugging your input file.

3.2 Graphical user interface

FDS has an indepedently developed graphical user interface named BlenderFDS:

http://www.blenderfds.org/

See specific chapter 15 on page 129 for more information.

3.3 Version numbers

Each release of FDS and Smokeview is identified by a version number suchas FDS_5.0.1 or SMV_5.0.1, where the first number is the major release (5),the second number (0) is the minor release, and the third (1) indicates themaintenance release number.As a general pattern, major releases will occur every year or so. As the nameimplies, they dramatically change the functionality or capabilities of the model.Minor releases occur every few months, and may cause minor changes in func-tionality. Maintenance releases are more frequent and are just bug fixes or smallrefinements, and should not affect code functionality. The release notes can helpyou decide whether the changes should effect the type of applications that youtypically do.

3.4 Hardware requirements

FDS requires one or more fast CPUs and a substantial amount of random-accessmemory (RAM) to run efficiently. For minimum specifications, the system should

3.5. SERIAL AND PARALLEL CALCULATIONS 17

have at least a 1 GHz CPU, and 1 GB RAM. The CPU speed will determine howlong the computation will take to finish, while the amount of RAM will determinehow many mesh cells can be held in memory.1 GB RAM can hold around 106 cells. To understand the physical meaning, a20m x 10m x 5m computational domain contains 106 cells, when discretizedwith cubic cells of 10 cm side.A large hard drive is required to store the output of the calculations. It is notunusual for the output of a single calculation to consume many gigabytes ofstorage space.Smokeview needs an OpenGL graphic card. Look for graphics cards that specif-ically list OpenGL support for the operating system you intend to use.

3.5 Serial and parallel calculations

FDS can perform serial and parallel calculations:

Serial calculations are performed in a single process that uses one only core ofcurrent multi-core CPUs.

Parallel calculation splits the computational burden on many processes thatcan be assigned to many different cores or CPUs. These CPUs can resideon a single workstation or in a cluster of networked computers.

Setting up a cluster of computers is a complex task and it is out of the scope ofthis manual.

3.6 Installing on Windows XP

To install or reinstall FDS and Smokeview on a Windows XP system, downloadWindows installation package from official website.The Windows file is a self-extracting compressed archive that will install FDS,Smokeview and all associated files in the Program Files/FDS folder.Launch the installation by double-clicking on the downloaded file.At the end of the install process, your Windows XP system is ready to performserial calculations.


Figure 3.1: A cluster of Linux computers

3.7 Installing on Ubuntu Linux

These instructions require a basic knowledge of an Ubuntu Linux computer.Ubuntu Linux operating system comes in two basic flavors:

Ubuntu Linux 32 bit, that can be installed on any type of computer.

Ubuntu Linux 64 bit, that can be installed on all AMD 64 bit CPUs withAMD64 extension and all Intel CPUs with EM64T extension.

The 64 bit version has the ability to address more RAM memory than 32 bitversion (over 4 GB). The 32 bit version is limited to 4 GB of RAM memory.Be aware that Smokeview works much better on a good dedicated graphic card.Some cheap graphic cards can prevent you from using it on Linux.

3.7.1 First install

To install FDS and Smokeview on an Ubuntu Linux system, first, download thelatest version of the precompiled FDS and Smokeview for Linux from download

3.7. INSTALLING ON UBUNTU LINUX 19

page on the official web site. Depending on your Ubuntu flavor, download the32 bit or the 64 bit compressed archive of the FDS distribution.After downloading to your computer, extract the archive by right clicking on itsicon and selecting Extract here in the context menu.Then, move the extracted FDS folder to your preferred location, for exampleyour home directory. In my case the resulting path to FDS folder would be/home/egissi/FDS, as /home/egissi is my home folder.After that, look at hidden files of your home folder by selecting View . Showhidden files menu in the Nautilus file browser. Locate the .bashrc file inyour home directory, and open it for editing by double-clicking its icon.Append the following text to the .bashrc file in the editor window:

### FDS and Smokeview environment# Actual path to FDS folderFDS=/home/egissi set your actual path hereecho "FDS setup ($FDS)"# Setting limitsulimit -s unlimitedulimit -v unlimited# Setting executable and library pathsexport PATH=$PATH:$FDS/FDS5/FDS/binexport LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$FDS/FDS/FDS5/bin/lib32export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$FDS/FDS/FDS5/bin/lib64

Edit the emphasized line and set your actual path to the FDS folder. After that,save the .bashrc file and close the text editor.Open Synaptic package manager by selecting System . Administration . Synapticpackage manager menu and install:

On Ubuntu Linux 32 bit: lam-runtime software package;

On Ubuntu Linux 64 bit: lam-runtime and ia32-libs software packages.

Ubuntu Linux takes care for you of all the needed software dependencies.Close your session and log-in again to put into effect the environment modifi-cations. Your Ubuntu Linux system is now ready to perform serial and parallelsimulations on a multi-core or multi-CPU workstation.


3.7.2 Installing a new version

When installing a new version of FDS and Smokeview, just delete the old FDSfolder. After downloading the new version, extract the new FDS folder and dragit to the same position as before.Your Ubuntu Linux system is now ready to perform serial and parallel simulationswith the new version.

3.8 Compiling an optimized binary

As FDS is open source software, users can always download the latest version ofFDS source code and compile it by themselves using a Fortran 90 and C compiler.Direct compilation is often applied to obtain an FDS binary that is optimized forthe specific hardware and platform and to fully take advantage of its speed.Compilation is a complex task and is out of the scope of the present manual.

3.9 Performing a calculation

The typical procedure for using FDS and Smokeview is to:

1. Set up an FDS input file, as mycase.fds, and put it in a folder, as mycasefolder. See following chapters to learn how to do it.

2. Run FDS on the input files. FDS starts and creates many output files inthe mycase folder.

3. While FDS is running, monitor the development of the calculation.

4. Analyze the generated output files with Smokeview.

3.9.1 Running serial FDS on Windows XP

After having set up an input file, open up a command prompt window: selectStart . Run menu, then type cmd. Move into the mycase folder, where the inputfile for the case is located, with the cd command.Then run the code by typing the following command:

3.9. PERFORMING A CALCULATION 21

Figure 3.2: Starting a serial calculation on Windows XP and on Linux Ubuntu

fds5 mycase.fds

where fds5 is the name of FDS binary and mycase.fds is the input file name.A serial calculation starts and its progress is indicated by diagnostic output thatis written out onto the screen.

3.9.2 Running serial FDS on Ubuntu Linux

On an Ubuntu Linux computer select Applications . Accessories . Terminalmenu to open the command prompt. Move into the mycase folder, where theinput file for the case is located, with the cd command.Then execute one of the following commands:

On Ubuntu Linux 32 bit: fds5_intel_linux_32 mycase.fds

On Ubuntu Linux 64 bit: fds5_intel_linux_64 mycase.fds

If unsure, just look inside FDS/FDS5/bin folder to discover the right name forFDS binaries.Then the simulation starts and its progress is indicated by diagnostic output thatis written out onto the screen.

3.9.3 Running parallel FDS on Ubuntu Linux

FDS uses the Message Passing Library (MPI) for parallel computing.


Figure 3.3: Running a parallel calculation and monitoring the system on UbuntuLinux

MPI is a language-independent communications protocol used to program paral-lel computers. Both point-to-point and collective communication are supported.MPIs goals are high performance, scalability, and portability. MPI is the domi-nant model used in high-performance computing today.The input file for both single and parallel versions of FDS are the same. In fact,it is recommended that before embarking on parallel processing, you should runyour input file in serial mode to ensure that it is properly set up.To run FDS in parallel, you must break up the computational domain into multiplemeshes so that the workload can be divided among the available processors.For the parallel version to work well, there has to be a comparable number ofcells in each mesh, or otherwise most of the computers will sit idle waiting forthe one with the largest mesh to finish processing each time step.On an Ubuntu Linux computer as configured before, type the following commandsto perform a parallel calculation:

On Ubuntu Linux 32 bit:

lamboot -vmpirun -np 4 fds5_mpi_intel_linux_32 mycase.fdslamhalt -v

On Ubuntu Linux 64 bit:

lamboot -vmpirun -np 4 fds5_mpi_intel_linux_64 mycase.fdslamhalt -v

3.10. MONITORING PROGRESS 23

The lamboot command starts the MPI environment.The mpirun command starts an MPI application, in our case fds5_mpi_intel_linux_32or fds5_mpi_intel_linux_64, where -np 4 is the number of started processes.The number of processes must match the number of meshes that span the com-putational domain of the input case.At the end of the calculation, the MPI server is safely stopped with lamhaltcommand.

3.10 Monitoring progress

Diagnostics for a given calculation are written into a text file called mycase.out,contained in the case folder mycase. The CPU usage and simulation time arewritten here, so you can monitor it to see how far along the program has pro-gressed.The application System . Administration . Gnome System Monitor is a pro-cess viewer that provides a dynamic real-time view of a running system.

3.11 Stop a calculation

To stop a calculation before its scheduled time, create an empty file in themycase folder called mycase.stop. The mere existence of this file stops theprogram gracefully, causing it to dump out the latest flow variables for viewingin Smokeview.Since calculations can be hours or days long, FDS has a restart feature. See[FDS5 users guide] for broader details.

3.12 Visualizing results

Smokeview is used before, during and after model runs:

before, to check the input data;

during a calculation, to monitor a simulations progress;

in a post-processing step, to visualize FDS data after a calculation hasbeen completed.


Figure 3.4: Running Smokeview on Windows XP and on Ubuntu Linux

On Windows XP, Smokeview may be started by double-clicking on the file namedmycase.smv, contained in the case folder.On Ubuntu Linux, Smokeview is run from the command prompt by typing:

On Ubuntu Linux 32 bit: smv5_linux_32 mycase.smv

On Ubuntu Linux 64 bit: smv5_linux_64 mycase.smv

If unsure, just look inside FDS/FDS5/bin folder to discover the right name forSmokeview binary.Inside Smokeview, menus are accessed by clicking on the graphical window withthe right mouse button. The Load/Unload menu may be used to read in thedata files to be visualized. The Show/Hide menu may be used to change howthe visualizations are presented.For the most part, the menu choices are self explanatory. In case of need, helpon using Smokeview can be found on [Smokeview users guide].

3.13 Output files

FDS writes out many output files in the mycase folder:

Diagnostic output: the file mycase.out consists of a list of input parameters,and an accounting of various important quantities, including CPU usage.

Heat release rate and related quantities: the HRR of the fire, plus other globalenergy-related quantities, are automatically written into a text file calledmycase_hrr.csv

3.13. OUTPUT FILES 25

Device output data: data associated with particular devices (link tempera-tures, smoke obscuration, thermocouples, etc.) is output in comma de-limited format in a file called mycase_devc.csv

and many other types of files, used by Smokeview for visualisation.

The comma delimited format files can easily be imported into Openoffice.orgCalc or Microsoft Excel for further analysis.

Chapter 4

Input file basics

This chapter teaches the logic of FDS input file, its organization andits grammar. Then the standard reference system is explained andsome tips and tricks are exposed.

4.1 Syntax of the input file

All the necessary information to perform an FDS simulation has to be containedin a single text file. The input file is saved with a name such as mycase.fds.There should be no blank spaces in the job name.Data is specified within the input file by using namelist groups. Each namelist Namelist groupsgroup record occupies a line of text and begins with the & character. The & mustbe the first character of the line of text and should be followed by the nameof the namelist group. Then a comma-delimited list of the input parameters isinserted. Finally a forward slash / character closes the namelist group, as shownin Figure 4.1.

Figure 4.1: The structure of an FDS namelist group

Spaces and new lines can be freely inserted to visually format the namelist group.FDS uses the information contained between the & / delimiters, the rest is

27

28 CHAPTER 4. INPUT FILE BASICS

ignored. Thus, comments and notes can be written outside the & / delimiters. CommentsFor example:

&OBST XB=0.5,1.1,0.5,1.1,0.0,0.1 / A commentAnother comment

It is recommended that each namelist group be clearly commented to justify thechoice of its parameters and to link it to literature references or direct experi-mentation. For example, comments like these:

&REAC ID=polyurethane, SOOT_YIELD=0.1875, CO_YIELD=0.02775,C=1.0, H=1.75, O=0.25, N=0.065,HEAT_OF_COMBUSTION=25300., IDEAL=.TRUE. /Gas phase reaction: polyurethane flexible foam (means)from Tewarson SFPE Handbook 3rd ed,SFPE handbook table 3-4.14, p. 3-112.

help the reviewer keeping track of the sources of information employed by theuser.By deeply commenting the code, the input file becomes the complete and onlysource of information about the simulated case.The parameter values can be of the following types:Parameter values

Integers, as in T_END=5400

Real numbers, as in CO_YIELD=0.008

Groups of real numbers, as in XYZ=6.04,0.28,3.65

Groups of integers, as in IJK=90,36,38

Character strings, as in CHID=this_is_a_string

Groups of character strings, as in SURF_IDS=burner,steel

Logical parameters, as in POROUS_FLOOR=.FALSE. or POROUS_FLOOR=.TRUE.The periods must be included.

Sometimes the parameters are multidimensional arrays. For example:Parameter arrays

4.2. WRITING AN INPUT FILE 29

MATL_ID(2,3)=brick

indicates that the third material component of the second layer is brick.To speed up data input, you can use this notation:

MATL_ID(1:3,1)=plastic,insulation,steel

which means that the surface is composed by three different layers made re-spectively of plastic, insulation and steel. The notation 1:3 means arrayelement 1 through 3, inclusive.A simplified notation is accepted, too:

MATL_ID=plastic,steel

is equivalent to:

MATL_ID(1:2,1)=plastic,steel

These last surfaces are composed by two different layers made respectively ofplastic and steel.The code is case sensitive: my_burner is not the same as MY_BURNER. Case sensitivityTo ensure that FDS reads the entire input file, add &TAIL / or a comment as TAILthe last line at the end of the input file.

4.2 Writing an input file

When looking at a new scenario, first select a pre-written input file that resem-bles the case, make the necessary changes, then run the case at fairly low gridresolution to determine if the geometry is set up correctly.The following file is a slightly modified and simplified version of pplume5.fds,generally included in FDS software distribution:

!!! General configuration

&HEAD CHID=pplume5, TITLE=Plume case /name of the case and a brief explanation


&TIME T_END=10.0 /the simulation will end at 10 seconds

&MISC SURF_DEFAULT=wall, TMPA=25. /all bounding surfaces havea wall boundary conditionunless otherwise specified,the ambient temperature is set to 25C.

&REAC ID=polyurethane, SOOT_YIELD=0.10,N=1.0, C=6.3, H=7.1, O=2.1 /predominant fuel gas for the mixture fraction modelof gas phase combustion

!!! Computational domain

&MESH IJK=32,32,16, XB=0.0,1.6,0.0,1.6,0.0,0.8 /&MESH IJK=32,32,16, XB=0.0,1.6,0.0,1.6,0.8,1.6 /&MESH IJK=32,32,16, XB=0.0,1.6,0.0,1.6,1.6,2.4 /&MESH IJK=32,32,16, XB=0.0,1.6,0.0,1.6,2.4,3.2 /

four connected calculation meshesand their cell numbers

!!! Properties

&MATL ID=gypsum_plaster, CONDUCTIVITY=0.48,SPECIFIC_HEAT=0.84, DENSITY=1440. /thermophysical properties of gypsum plaster material

&PART ID=tracers, MASSLESS=.TRUE., SAMPLING_FACTOR=1 /a type of Lagrangian particles

&SURF ID=burner, HRRPUA=600.,PART_ID=tracers, COLOR=RASPBERRY /a type of boundary conditions named burner

&SURF ID=wall, RGB=200,200,200, MATL_ID=gypsum_plaster,THICKNESS=0.012 /a type of boundary conditions named wall

!!! Solid geometry

&VENT XB=0.5,1.1,0.5,1.1,0.1,0.1, SURF_ID=burner /the burner boundary conditionis imposed to a plane face

&OBST XB=0.5,1.1,0.5,1.1,0.0,0.1, SURF_ID=wall /

4.3. THE LOGIC BEHIND MOST FDS INPUT FILES 31

a solid is created, wall boundary conditionis imposed to all its faces

&VENT XB=0.0,0.0,0.0,1.6,0.0,3.2, SURF_ID=OPEN/&VENT XB=1.6,1.6,0.0,1.6,0.0,3.2, SURF_ID=OPEN/&VENT XB=0.0,1.6,0.0,0.0,0.0,3.2, SURF_ID=OPEN/&VENT XB=0.0,1.6,1.6,1.6,0.0,3.2, SURF_ID=OPEN/&VENT XB=0.0,1.6,0.0,1.6,3.2,3.2, SURF_ID=OPEN/

the OPEN boundary condition is imposed tothe exterior boundaries of the computational domain

!!! Output

&DEVC XYZ=1.2,1.2,2.9, QUANTITY=THERMOCOUPLE, ID=tc1 /send to output: the data collected by a thermocouple

&ISOF QUANTITY=TEMPERATURE, VALUE(1)=100.0 /3D contours of temperature at 100C

&SLCF PBX=0.8, QUANTITY=TEMPERATURE, VECTOR=.TRUE. /vector slices colored by temperature

&BNDF QUANTITY=WALL TEMPERATURE /surface WALL_TEMPERATURE at all solid obstructions

&TAIL / end of file

Not all kinds of FDS namelist groups are listed in this input files. In fact,another general rule of thumb when writing input files is to only add to the fileparameters that are to change from their default value. That way, you can moreeasily distinguish between what you impose and FDS defaults.In general, the namelist records can be entered in any order in the input file, butit is a good idea to organize them in some systematic way.Be aware that the order of identical namelist groups can be significant. When Order conventionproperties overlap the general rule is first-come, first-served.For the sake of clarity, users often group similar namelists in homogeneous sec-tions identified by heading comments, as shown in the former example inputfile.

4.3 The logic behind most FDS input files

This section presents the logic behind most FDS input files; this same logic isused for the organization of this manual:


First, general configuration is performed.

The case receives a name via the HEAD namelist group, the simula-tion time is set via the TIME namelist group. Other miscellaneousparameters are prescribed via the MISC namelist group.

Then gas phase combustion reaction is set up via the REAC namelistgroup, the radiation model is configured with RADI.

Second, the computational domain is defined via the MESH namelist group.All FDS calculations must be performed within a domain that is made upof rectilinear volumes called meshes. Each mesh is divided into rectangularcells, the number of which depends on the desired resolution of the flowdynamics.Some initial conditions are prescribed for the flow domain via the INITnamelist group.

Third, some properties are set up:

the properties of each material (MATL), the properties of extra gas species (SPEC), the properties of Lagrangian particles (PART), and the types of boundary conditions (SURF).

This is the most challenging part of setting up the simulation: first, for bothreal and simulated fires, the growth of the fire is very sensitive to the thermalproperties of the surrounding materials. Second, even if all the material proper-ties are known to some degree, the physical phenomena of interest may not besimulated properly due to limitations in the model algorithms or resolution of thenumerical mesh.It is your responsibility to supply the thermal properties of the materials, and thenassess the performance of the model to ensure that the phenomena of interestare being captured.

Fourth, the solid geometry is entered via OBST, VENT, HOLE namelistgroups.

A considerable amount of work in setting up a calculation lies inspecifying the geometry of the space to be modeled and applyingboundary conditions to these objects. The geometry is described interms of obstructions to the gas phase flow.

4.3. THE LOGIC BEHIND MOST FDS INPUT FILES 33

A boundary condition needs to be assigned to each bounding surfaceof the gas phase domain describing its thermal properties. Both solidobstruction faces and the exterior boundaries of the computationaldomain need a boundary condition assigned. A fire is just one typeof boundary condition.

Fifth, some control logic and automation is introduced via PROP, DEVC,CTRL namelist groups: devices can be used to control various actions, likecreating and removing obstructions, or activating and deactivating fansand vents.

Finally, the user prescribes the output quantities (DEVC, SLCF, BNDF, ISOF).All output quantities must be specified at the start of the calculation. Inmost cases, there is no way to retrieve information after the calculationends if it was not specified from the start. Much like in an actual experi-ment, the user must decide before the calculation begins what informationto save.

The table summarizes this logic and shows a proposed systematic organizationof an input file in sections:

Table 4.1: Systematic organisation of the input file

Section Content Namelist groups

General configuration General information required toperform a simulation, as its name,duration and other miscellaneousparameters.

HEAD, TIME, MISC

Main gas phase combustion reactionand radiation model.

REAC, RADI

Computational domain Computational domain:dimensions and grid.

MESH

Initial conditions of thecomputational domain.

INIT

Properties Materials, temperature dependentthermophysical properties.

MATL, RAMP(temperature)

Extra gas species properties. SPECcontinued on next page


from previous pageLagrangian particles properties. PART

Boundary conditions, time dependentboundary conditions.

SURF, RAMP (time)

Solid geometry Description of solid geometry,assignment of boundary conditionsto bounding surfaces.

OBST, HOLE, VENT

Control logic General properties of devices, devicesand control functions used to controlvarious actions, like creating andremoving solid obstructions oractivating and deactivating boundaryconditions.

PROP, DEVC, CTRL

Output List of calculated quantities tooutput.

DUMP, DEVC, SLCF,BNDF, ISOF

4.4 Keep it simple

Novice users tend to forget that FDS is not a Computer Aided Design (CAD)tool, but a CFD code.First, not all geometrical details, all physical and chemical properties of all in-volved objects need to be entered in the input file.Looking at the example proposed in Figure 4.2 on the facing page, chair and tableframe effect on the fluid flow can be considered negligible. On the contrary, theinfluences to fluid flow of the separating wall, the table top and seats can becomeimportant, depending on the objective of the analysis.So, the first step of the analysis process is to formulate the problem by seekinganswers to the following questions:

What is the objective of the analysis?

What is the easiest way to obtain that objective?

What input data needs to be included?

Approximations of the geometry and simplifications of the properties are alwaysrequired to allow an analysis with reasonable effort.

4.5. EACH MODEL, ITS INPUT DATA 35

Figure 4.2: Modeling reality in FDS

It is better to start off with a relatively simple file that captures the main featuresof the problem without getting tied down with too much detail that might maska fundamental flaw in the calculation.Initial calculations ought to be meshed coarsely so that the run times are lessthan an hour and corrections can easily be made without wasting too much time.As you learn how to write input files, you will continually run and re-run yourcase as you add in complexity.

4.5 Each model, its input data

When entering data into the input file, it is suggested to always consider howmodels inside FDS will use that data.For example:

The hydrodynamic model needs to know which cells of the computationaldomain are open to fluid flow and which are instead occupied by solidobstructions. The geometry is discretized and the maximum resolution isthe grid cell size.


The heat transfer model needs the characteristics and the thicknesses of thebounding surfaces of the flow domain to perform heat transfer calculation.

Imagine that the wall separating room 1 and room 2 of Figure 4.2 on the previouspage is 0.19m thick. Thus, during the calculation:

Taken the cell size equal to 0.30m, the hydrodynamic model considers thatwall as if it was 0.30m thick, because the geometry must conform to theunderlying grid. That information is used to obstacle the fluid flow.

The heat transfer model performs a one-dimensional heat transfer calcula-tion of the wall using the real 0.19m thickness and the material properties.

It may sound strange to novice users, but that wall is. . . both 0.30m and 0.19mthick for FDS.

4.6 Units of measurement

FDS employs the units of measurement from the International System (SI).

Lengths are expressed in m, time in s, mass in kg, temperature in C, pressurein Pa, heat in kJ, power in kW, conductivity in W/m/K, heat flux in kW/m2,molecular weight in g/mol. . .

This manual contains a comprehensive list of frequent namelist parameters andtheir units. For a complete list consult the [FDS5 users guide].

4.7 Reference coordinate system

FDS coordinate system conforms to the right hand rule. By default, the z axisis considered the vertical.

For computational reasons, it is always preferable for the longest horizontal di-mension of the model be aligned with the x axis. This often shortens the calcu-lation time.

4.8. PRESCRIBING GEOMETRIC ENTITIES 37

Figure 4.3: The reference system, a volume, a face, a segment, a point, and aplane

4.8 Prescribing geometric entities

Many namelist groups extend their action to volumes, faces, segments, points orplanes. As shown in Figure 4.3, FDS geometrical entities are always describedusing some conventional rules.A volume is always represented by a single right parallelepiped with edges parallel Volumesto the axis. Its position and dimensions are described by the coordinates of twoopposite vertices: if point A = (xA, yA, zA) and point B = (xB, yB, zB) arethe opposite vertices, its coordinates are entered as xA, xB, yA, yB, zA, zB. Forexample,

&OBST XB=0.5,1.5,2.0,3.5,-2.0,0., SURF_ID=wall /

uses the parameter XB to define a solid obstacle that spans the volume startingat the origin (0.5, 2.0,2.0) and extending 1m in the positive x direction, 1.5min the positive y direction, and 2m in the positive z direction.A face is represented by a right plane face with edges parallel to the axis. Its Facesposition and dimensions are described by the coordinates of two opposite vertices,that must lie on the same plane. For example:

&VENT XB=0.5,1.1,2.0,3.1,-2.0,-2.0, SURF_ID=fire /

uses the parameter XB to define a flat face perpendicular to the z axis imposinga particular boundary condition over a solid. Two of the six coordinates are thesame, denoting a flat face as opposed to a solid.A segment is bounded by two end points. If point A = (xA, yA, zA) and point SegmentsB = (xB, yB, zB) are the end points, its coordinates are entered following thesame convection valid for volumes. For example,


&DEVC XB=0.5,1.5,2.0,3.5,-2.0,0., QUANTITY=PATH OBSCURATION,ID=beam1, SETPOINT=0.33 /

is a beam smoke detector between (0.5, 2.0,2.0) and (1.5, 3.5, 0.) end points.A point is simply identified by its 3 coordinates. For example, the line:Points

&DEVC XYZ=2.,3.,4., QUANTITY=THERMOCOUPLE, ID=termo1 /

uses the parameter XYZ to insert a thermocouple at the point of coordinates(2., 3., 4.).A plane is represented by a right plane perpendicular to one of the reference axis.PlanesFor example, these lines:

&SLCF PBX=0.5, QUANTITY=TEMPERATURE /

is a plane perpendicular to the x axis and intersecting its point (.5, 0., 0.).

&SLCF PBY=1.5, QUANTITY=TEMPERATURE /

is a plane perpendicular to the y axis and intersecting its point (0., 1.5, 0.).

&SLCF PBZ=-.5, QUANTITY=TEMPERATURE /

is a plane perpendicular to the z axis and intersecting its point (0., 0.,.5).All use the parameters PBX, PBY, PBZ to specify the coordinate in the directionof the perpendicular axis.

4.9 Prescribing orientations

Some FDS entities need the prescription of a particular orientation. This is donewith one of the following parameters: IOR or ORIENTATION.The parameter IOR, index of orientation, is used to prescribe one of the sixIORpossible orientations parallel to axis: if the orientation is in the positive x directionset IOR=1, negative x direction IOR=-1, positive y IOR=2, negative y IOR=-2,positive z IOR=3, negative z IOR=-3.For example, the line:

4.10. PRESCRIBING COLORS AND ASPECT 39

&DEVC XYZ=0.7,0.9,2.1, QUANTITY=WALL TEMPERATURE,IOR=-2, ID=ST-1 /

designates the surface temperature of a wall facing the negative y direction.The parameter ORIENTATION is used for entities that require a free directional ORIENTATIONspecification, like a sprinkler. ORIENTATION is specified with a triplet of realnumber values that indicate the components of the direction vector. The defaultvalue of ORIENTATION is (0, 0,1).For example, the line:

&DEVC XYZ=23.91,21.28,0.50, PROP_ID=nozzle,ORIENTATION=1.,1.,0., ID=noz_1 /

designates a nozzle oriented towards the (1, 1, 0) vector.

4.10 Prescribing colors and aspect

Colors of objects can be prescribed with two parameters: RGB and COLOR.The RGB parameter is followed by a triplet of integer numbers in the range from RGB0 to 255, indicating the amount of red, green and blue that make up the color.The COLOR parameter calls the name of a predefined color that must be entered COLORexactly as it is listed in the color table:

Table 4.2: COLOR values

ACQUAMARINE, BANANA, BEIGE, BLACK, BLUE, BRICK, BROWN, CADMIUMORANGE, CARROT, COBALT, CORAL, CRIMSON, CYAN, FIREBRICK, FLESH,GOLD, GRAY, GREEN, INDIGO, MAGENTA, MAROON, MELON, MINT, NAVY,OLIVE, ORANGE, ORCHID, PINK, PURPLE, RASPBERRY, RED, SALMON,SEPIA, SIENNA, SILVER, TAN, TEAL, TOMATO, TURQUOISE, VIOLET,WHITE, YELLOW, ...

You can rapidly find the whole color table of more than 500 colors by googlingfor FDS COLOR TABLE on the Internet.For example, both the parameter RGB=0,0,255 and the parameter COLOR=BLUEcan be used to obtain a blue object.


Objects can be made semi-transparent by assigning a TRANSPARENCY parameter.TRANSPARENCYThe parameter value is a real ranging from 0 to 1, with 0 being fully transparent.The parameter should always be set along with RGB or COLOR.Using COLOR=INVISIBLE causes the object not to be drawn in Smokeview.The parameter OUTLINE=.TRUE. causes the object to be drawn as an outline.

Chapter 5

General configuration

First, general configuration is performed.The case receives a name via the HEAD namelist group, the simula-tion time is set via the TIME namelist group. Other miscellaneousparameters are prescribed via the MISC namelist group.

5.1 Naming the job, HEAD

The first thing to do when setting up an input file is to give the job a name. Thename of the job is important because often a project involves numerous simula-tions in which case the names of the individual simulations can help organize theeffort. The namelist group HEAD contains two parameters, as in this example:

&HEAD CHID=mycase, TITLE=This is a short description /

CHID is a string of 30 characters or less used to name the output files createdby FDS. No periods or spaces are allowed. TITLE is a string of 60 characters orless that describes the simulation. It is simply a descriptive text that is passedto various output files.It is always convenient to exactly use the same string for the name of the inputfile and the CHID. For example, if you name mycase.fds the input file, then setCHID=mycase in the HEAD namelist group.Only one HEAD line can be entered in the input file. The following table summa-rizes some HEAD parameters:

41

42 CHAPTER 5. GENERAL CONFIGURATION

Table 5.1: HEAD parameters

Parameter Type Description Unit Default

CHID String Job identifier output

TITLE String Short description of the job

5.2 Simulation time, TIME

TIME is the namelist group that define the time duration of the simulation.Usually, only the duration of the simulation is required on this line, via theparameter T_END. The default is 1 s.For example, the following line will instruct FDS to run the simulation for 5400 s:

&TIME T_END=5400. /

If T_END is set to zero, only the set-up work is performed, allowing you to quicklycheck the geometry in Smokeview.Only one TIME line can be entered in the input file. The following table summa-rizes some TIME parameters:

Table 5.2: TIME parameters


T_BEGIN Real Starting time for calculation s 0.

T_END Real Ending time for calculation s 1.

5.3 Miscellaneous, MISC

MISC is the namelist group of global miscellaneous input parameters. Manyparameters for MISC exist, some of them are explained later in this manual.For example:

&MISC SURF_DEFAULT=steel /

5.3. MISCELLANEOUS, MISC 43

establishes that all bounding surfaces are to be made of steel unless otherwisespecified.Only one MISC line can be entered in the input file. The following table summa-rizes some MISC parameters:

Table 5.3: MISC parameters


SURF_DEFAULT String Default boundary condition SURF forsurfaces

INERT

TMPA Real Ambient temperature C 20.

U0,V0,W0 Real Initial prevailing velocity field m/s 0.

GVEC(3) Real Gravity vector m/s2 0,0,-9.81

HUMIDITY Real Relative Humidity % 40.

CO_PRODUCTION Logical Start three-parameters mixturefraction model

.FALSE.

RESTART Logical Restart previous calculation .FALSE.

44 CHAPTER 5. GENERAL CONFIGURATION

Chapter 6

Combustion and radiation

. . . then gas phase combustion reaction is set up via the REAC namelistgroup, the radiation model is configured with RADI.

6.1 Combustion is not pyrolysis

A common source of confusion in FDS is the distinction between gas phasecombustion and solid phase pyrolysis.

Pyrolysis is the decomposition or transformation of a compound caused by heatthat produce the gaseous fuel. It is the first chemical reaction that occursin the burning of many solid fuels, like wood, cloth, paper, and plastic.

Gas phase combustion refers to the exothermic chemical reactions betweenthe gaseous fuel and oxygen accompanied by the production of heat andlight in the form of flames.

So solid phase pyrolysis refers to the generation of fuel vapor at a solid or liquidsurface, while the visible flames are not due to combustion of the solid fuel itself,but rather of the gases released by its pyrolysis.

6.2 Prescribing a fire

In FDS, a fire is a particular boundary condition applied to a surface boundingthe flow field. There are two ways of designating a fire:

45

46 CHAPTER 6. COMBUSTION AND RADIATION

Figure 6.1: Combustion is not pyrolysis

The first is to specify an heat release rate on a surface; this is the same asprescribing a well defined burner. How to do this is described in detail inSection 11.4.6 on page 84.

The other is to specify thermophysical properties of fuel materials andto let them pyrolyze. In this case the burning rate of the fuel dependson the net heat feedback to the surface. This approach is explained inSection 8.3.1 on page 65 for solid fuels and in Section 8.3.2 on page 66for liquids.

Both burners and pyrolyzing materials inject the calculated quantities of gaseousfuels in the flow field. In a realistic fire scenario, there may be various types ofgaseous fuels originating from the various burning objects in the building andinjected into the flow field.

6.3 Modeling gas phase combustion, REAC

6.3.1 Ignition

Once injected into the flow field, the gaseous fuels mix with air and burn. Thereis no need to prescribe an ignition source: the combustion model assumes that

6.3. MODELING GAS PHASE COMBUSTION, REAC 47

Figure 6.2: Combustion and pyrolysis in a flaming match

fuel gas and oxygen burn on contact. We can imagine that every grid cell hostsa virtual spark plug, that initiates combustion when temperature and local ratioof fuel gas and oxygen are appropriate (See Figure 6.3 on page 52).

6.3.2 Burning

The burning process releases heat and smoke.Whereas there can be many types of combustibles in an FDS fire simulation, oneonly gaseous fuel can be simulated by FDS. In general, you should set the chem-istry of the modeled burning gaseous fuel to coincide with the actual predominantburning gaseous fuel.This model simplification is due to computational cost: it is expensive to solvetransport equations for multiple gaseous fuels.FDS adjusts automatically the burning rates of solids and liquids to account forthe difference in the heats of combustion of the various combustibles. If thestoichiometry of the burning material differs from the global reaction, the heatof combustion of each burning material is used to ensure that an equivalentamount of fuel is injected into the flow domain from the burning object.FDS can describe the gas phase reaction in two ways.By default, a so-called mixture fraction model is used to account for the evolution Mixture fraction

modelof the fuel gas from its surface of origin through the combustion process.

48 CHAPTER 6. COMBUSTION AND RADIATION

The alternative is what is referred to as the finite-rate approach, where all of the Finite-rate approachindividual gas species involved in the combustion process are defined and trackedindividually. This is a costlier and more complicated approach than the mixturefraction model.This manual covers the mixture fraction model only, as it is simpler and commonlyemployed for engineering level problems.When the mixture fraction model is applied, a set of scalar variables, Zi, representthe state of the combustion process from pure fuel (Zi = 1) to pure air(Zi = 0).FDS provides two types of mixture fraction model:

Two-parameter mixture fraction model: the first parameter (Z1) is the massfraction of unburned fuel and the second (Z2) is the mass fraction of burnedfuel, as for example the mass of the combustion products that originatedas fuel. FDS uses the two-parameter model by default.

Three-parameter mixture fraction model: this combustion model simulatesa two-step chemical reaction with three parameters. The first step of thereaction is the oxidation of fuel to carbon monoxide and the second stepthe oxidation of carbon monoxide to carbon dioxide. The three mixturefraction components for the two step reaction are unburned fuel (Z1), massof fuel that has completed the first reaction step (Z2), and the mass offuel that has completed the second reaction step (Z3). See Section 6.4 onpage 51 to understand why and how to use the three-parameter model.

The mass fractions of all of the major reactants and products of combustion asfuel, O2, CO2, H2O, N2, CO and soot can be derived from the mixture fractionparameters by means of state relations: a set of pre-tabulated functions of themixture fraction parameters, Zi. In other words, the values of Zi in any givenmesh cell determines the mass fraction of all the gases listed.The stoichiometry of the predominant gas phase combustion reaction is pre-scribed in the input file by one only REAC namelist group: the specified parame-ters are used to generate the table associating the mass fractions with Zi. FDSdefaults to propane combustion if no REAC line is entered.In the mixture fraction model, each reaction is assumed to be of the form:

CxHyOzNvOtherw + O2O2 CO2CO2 + H2OH2O+ COCO+ sootSoot + N2N2 + H2H2 + otherOther

6.3. MODELING GAS PHASE COMBUSTION, REAC 49

You need only specify the chemical formula of the fuel along with the yields of CO,soot, and H2, and the amount of hydrogen in the soot, Hfrac. For completenessyou can specify the N2 content of the fuel and the presence of other species.FDS will use that information internally to determine the amount of combustionproducts that are formed.The species implicitly defined by FDS when doing a mixture fraction calculationfor gas phase combustion are as follows:

Table 6.1: Mixture fraction species

fuel, oxygen, nitrogen, water vapor, carbon dioxide, carbonmonoxide, hydrogen, soot, other

Note that these species are identified by a lowercase name, and are not to beconfused with the species identified by uppercase names defined by the SPECnamelist groups. See Section 9.2 on page 70 for further discussion.The Table 6.2 lists some of the parameters that may be prescribed on the REACline. Note that the various *YIELD are for well-ventilated, post-flame conditions.There are options

Intro to Fire Sim

Documents

Transcript of Intro to Fire Sim