SCA Guidance-June 21

Smoke Control Association
SCA guidance on CFD analysis for

Smoke Control design in Buildings
Edition 1.2: June 2021
Page 1 of 70
Acknowledgements
Contributions to this guide are gratefully acknowledged from the following people:
Marios Alexandrou Buro Happold

Cristina Paduano B-Fluid Building Fluid Dynamics Ltd.
Patrick Okolo B-Fluid Building Fluid Dynamics Ltd.
Conor Logan Colt International Ltd
Donal Sheridan Fire Design Solutions Ltd
Mike Duggan HEVAC
Stewart Miles International Fire Consultants Ltd
Ben Atkinson Jenson Hughes
Paul Kingston K8T Ltd
Paul Grimwood Kent Fire and Rescue Service
Gareth Steele London Fire Brigade
Klementina Gerova NHBC
Maulik Katkoria NHBC
Paul Hanson Royal Borough of Kensington & Chelsea
Matt Salisbury Salisbury Fire Ltd.
Nelson Godinho (Chairman of Working Group) SE Controls Ltd
James Bertwistle WSP Fire Engineering
Page 2 of 70
Contents
1 Introduction.................................................................................................................................... 4
2 Background ................................................................................................................................... 5
3 Terms and definitions .................................................................................................................. 7
4 Modeller Experience and Qualifications ................................................................................... 9
5 Qualitative Design Review .......................................................................................................... 9
6 Limitations ................................................................................................................................... 11
7 Modelling process ...................................................................................................................... 13
8 Preparing the CFD Model ......................................................................................................... 14
9 Presentation of Analysis and Results...................................................................................... 24
10 Modelling Lobbies, Corridors and Stairs ................................................................................. 27
11 Modelling Large Enclosure Buildings ...................................................................................... 28
12 Modelling Car Parks................................................................................................................... 32
13 Modelling guidance for designing smoke control system..................................................... 41
Appendix A: Physical sub models .............................................................................................. 49
Appendix B: Quality and Compliance ........................................................................................ 53
Appendix C: Fire and Smoke Source ......................................................................................... 57
Appendix D: Testing and Validation ........................................................................................... 66
Appendix E: CFD review process ............................................................................................... 67
Appendix F: Bibliography & Additional Reading ...................................................................... 69
Page 3 of 70
1 Introduction
Computational Fluid Dynamics, or CFD, is an established technique using numerical
methods to analyse problems involving fluid flows. The physical properties that define
the fluid flow, such as pressure, temperature and velocity are dependent variables in
a mathematical model describing the fluid flow. This mathematical model defines the
flow field at any point in space and is defined as a series of partial differential equations
(PDE’s).
The continuous increase in computational power has made CFD a popular tool among
practitioners, engineers and researchers in many fields, including fire and smoke
ventilation engineering. Performing CFD simulations and obtaining the results for a
specific test is not a difficult task thanks to the numerous available commercial CFD
packages.
The purpose of this guide is to give an outline of the basics requirements that should
be considered when developing a CFD simulation of a smoke control system. It is also
hoped that this guide will give a general understanding of the challenges involved in
preparing CFD simulations to help those tasked with approving such systems a better
understanding of how erroneous results can be identified and what supporting
information and documentation should be expected.
Above all it important that all parties are clear on the context of the CFD results – CFD
software alone is not a design tool, it only predicts the performance of a design given
a particular set of conditions. Other assumptions and design conditions may present
an entirely different set of results. The skill of the CFD engineer is to understand the
limitations of the software and present the results in such a way that all the
assumptions used in the model preparation are justified and relevant and above all,
traceable, without allowing the graphical nature of the outputs to conceal any
shortcomings in the design.
Disclaimer - Guidance from FETA

FETA does not guarantee, certify or assure the safety or performance of any products, components, or
systems tested, installed or operated in accordance with FETA's Standards, Guidelines or guidance or
that any tests conducted under its Standards, Guidelines or guidance will be non-hazardous or free
from risk. FETA disclaims all liability to any person for anything or for the consequences of anything
done or omitted to be done wholly or partly in reliance upon the whole or any part of guidance from
FETA.
Page 4 of 70
2 Background
The use of computer models for simulating fires in enclosures has increased
substantially in recent years. The increased emergence of performance-based
solutions within the construction industry together with more complex and unique
building geometries, has led designers, engineers and local authorities to rely on
computer modelling to analyse and evaluate various elements of design within
buildings, and fire safety is no exception. This is especially relevant in complex
structures where the validity of simple hand calculations is limited. However, using
computer modelling in fire safety engineering design is not always straightforward. The
user must have a fundamental understanding of the science associated with the
models in order to assess the validity and accuracy of the simulation results.
However, the ability to access CFD tools relatively easily and cheaply has led to an
increase in the number of inexperienced users with insufficient knowledge of the detail
that sits behind the colourful images and animations. This can be misleading, not only
to the user, who may not recognise the errors embedded in the information presented
but also to those tasked with approving designs which have used CFD as part of their
validation.
In fire and smoke modelling, it is strongly recommended that those tasked to perform
CFD modelling have a good knowledge of the underlying physics and thermodynamics
associated with fire dynamics and smoke behaviour. It is also equally important to
understand the implicit limitations of the CFD approach and sensitivity of a solution to
assumptions relating to both model and scenario definition.
The main advantage of using CFD for validation of a system design is that this can be
carried out prior to construction, providing confidence to all parties early in the project.
It is much easier and cheaper to agree expectations and correct any problems at this
stage than after completion of the installation. Due to concerns about damage it is rare
for testing to be carried out using a real fire or any significant heat source and a
correctly modelled CFD analysis can provide relatively accurate simulation of what
may happen within the analysed building.
The increased emergence of performance-based regulations and hence solutions

within the fire safety industry together with more complex and unique building
geometries, have led designers, architects and local authorities to rely on computer
modelling to analyse and evaluate the smoke and heat transfer within buildings. This
is especially relevant in complex structures where the validity of simple hand
calculations is limited. However, using computer modelling in fire safety engineering
design should not be considered simple or easy. It is important that the user who
creates and processes the CFD model has a fundamental understanding of the
physics and chemistry associated with the models in order to assess the validity and
accuracy of the simulation results.
Page 5 of 70
For the design and approval process to be successful it is strongly recommended that,
except perhaps in the simplest cases, the system objectives, the scenarios to be
modelled, the modelling criteria, the expected reporting and the success criteria are
all agreed and documented prior to commencement of the analysis. CFD modelling is
too expensive and time-consuming process to be carried out without this agreement.
Advice and guidance on these issues is provided in this document.
Page 6 of 70
3 Terms and definitions
The definitions include both terms that are used in this document and terms that could be
contained within a CFD report.
3.1 Blocks
Blocks are used to represent solid objects in the scenario being modelled. (e.g. walls, floors,
ceilings, down stands, cars, doors, etc)
3.2 Boundary
The boundary is the edges of the domain or space that the CFD calculation is being
performed within.
3.3 Boundary conditions

Boundary conditions are set up by the user and characterise what happens at the edges of
and in particular areas within the domain. (For example; Wall, vents etc)
3.4 CFD
Computational Fluid Dynamics (CFD) is a method used for analytical solution of
thermodynamic mathematical equations that simulate the flow of fluids, heat transfer and
other associated phenomena, using computing processing power and memory. (For the
purposes of this paper, CFD modelling can be used to predict fire, smoke movement, heat,
radiation, ventilation flow etc)
3.5 Comparative Analysis

This compares two or more scenarios. This is often used to show that the model is a good
as the minimum requirement of a Standard or the suggested guidance to the Building
Regulations.
3.6 Deterministic Analysis

This is a non-comparative study based on physical relationships derived from scientific
theories and empirical results, that given a set of initial conditions will always produce the
same outcome. It is often used to show that the conditions satisfy the functional requirement
of the Building Regulations.
3.7 Domain
The Domain is the area that is to be modelled. This may include; part of the building, all of
the building or all of the building and some surrounding areas.
3.8 Developing Fire

Fire Development is a function of many factors including: fuel properties, fuel quantity,
ventilation (natural or mechanical), compartment geometry (volume and ceiling height),
location of fire, and ambient conditions (temperature, wind, etc) changing with time.
Page 7 of 70
3.9 Impulse fan (Also known as Jet Fan or Induction Fan
Fan designed to transfer momentum into the air as part of an impulse ventilation system and used to
provide control of air direction and velocity.
3.10 Mesh (Grid)

The outcome of splitting up the computational domain (discretisation) into a number of
elements or cells defining the discrete points at which the numerical solution is computed.
The points are normally the cell centres or cell vertices.
3.11 Output Slice

This is a two-dimensional output of data across a plane in the domain and is used to show
visual conditions (i.e. temperature, pressure, velocity etc). Colours are used to represent
varying conditions.
3.12 Porosity
The condition of a boundary that allows a set amount of leakage that may not be
proportional to the size of the vent.
3.13 Sensitivity Analysis

Varying selected parameters in a model to investigate the extent of their effect (e.g.
changing the mesh size).
3.14 Steady State Fire

This is a fire with constant heat release rate.
3.15 Steady State Model

This is a model that has no associated time period. This shows what conditions would be like
if the scenario was run for infinity. (Within the scope of this document, it is usually used to
show that smoke or heat is being taken out by the ventilation system at the same rate that it
is being produced by the fire or demonstrates the constant flow profile of a fan or vent.)
3.16 Transient model (Also known as a time dependant model)

This is a model that is time dependent and shows how conditions vary with time.
3.17 Vector slice

This is a two-dimensional output of vector data across a plane in the domain. Arrows
represent the direction with arrow size indicating the vector quantity.
Page 8 of 70
4 Modeller Experience and Qualifications
When undertaking CFD modelling to design or validate a smoke control system,
there is no single definitive approach which can be used for all buildings since
interactions between the building, fire, and its occupants can be highly complex. As
such, a greater degree of care and responsibility by the designer is required,
therefore it is essential that application of CFD modelling be entrusted to a suitably
qualified and experienced personnel. This means that that the individual or group
entrusted to undertake the modelling has the relevant skills, qualifications, training,
experience and professional liability cover.
Normally the individual or group appointed to carry out the CFD modelling are
expected to demonstrate that they have experience of successfully working on
similar schemes, that they are appropriately qualified and have the appropriate
professional status or can prove they are adequately competent (e.g. showing
sufficient experience), whilst working within their scope of expertise and ethical
engineering practices. A designer unable to demonstrate any of these competence
attributes should be mentored or supervised and their work quality assured by
someone who does, following the principles as described in BS 7974.
There is a similar expectation on the individual or group tasked with reviewing the
study, that they are appropriately competent or employ a third party who is
competent in the relevant area. This is further discussed in Appendix E.
5 Qualitative Design Review

The early part of the design process, i.e. its conceptual stage is the ideal time to
optimise the design of a building and its fire safety features whilst minimising
disturbance elsewhere to the building. Although amendments to the fire safety
features or their optimisation does often become necessary within the build
programme, the frequency and scale of this can be minimised with the effective use
of the Qualitative Design Review (QDR) technique detailed in BS 7974.
In brief, the QDR is a qualitative process that allows the Fire Engineers / CFD
modeller supported by stakeholders to use their experience and knowledge to
critically analyse the design problem, develop fire safety objectives and quantitative
assessment criteria.
QDR team major stakeholders may include:
• Design Fire Engineer/ CFD Modeller

• Checking Fire Engineer (and/or Building Control Surveyor)
• Specialist Installing Engineer
Page 9 of 70
• Fire Service
• Architect
• Mechanical & Electrical Services Engineer
• Structural Engineer
• Operational Management
• Insurer/ Surveyors
• Client/ developer/ builder
The CFD modeller, supported by the team should work towards identifying
representative fire scenarios and design fires that can be regarded as worst-case
fires that may affect the fire safety objectives. A structured approach is required to
ensure that hazards are not missed or overlooked and that the final design and
strategy is able to meet the fire safety objectives.
Items for consideration:
• Review the architectural design. There may be architectural features or floor

plan layouts which cannot be justified in which case these would need to be
amended.
• Establish the fire safety objectives. This may be connected to the

requirements of the Building Regulations and guidance may be sought from
Approved Document B or another relevant standard.
• Agree tools and software that is fit for purpose and appropriate to model and
capture the physics in the resulting analysis.
• Identify fire hazards and possible risks, this should include possible ignition
sources, combustible fixtures and content, materials of construction,
• Establish trial fire safety designs
• Identify acceptance criteria and methods of analysis. Acceptance criteria

should be justified using established values where practicable rather than
individual standalone studies.
• Establish Fire Scenarios for analysis.
• Establish CFD review process
All findings from the QDR should be documented with clear reasoning so that all
stakeholders can understand it and either approve or comment upon it ahead of the
final design being submitted for building regulations approval.
Page 10 of 70
Note; apartment blocks are usually planned with ‘Computer Aided Design’ (CAD),
these drawings should be made available to the CFD modeller constructing the
model. The CFD model should be designed around the design intent and any
conflicts with other services, fixtures or the challenging geometry can be identified
(such as curved walls, columns, down-stand beams etc.) at an early stage. Similarly,
‘Building Information Modelling’ (BIM) is becoming widely used, where BIM exists it
should be shared to encourage spatial co-ordination between all stakeholders as
information can be updated from anywhere including site.
Further information on the QDR process is available in BS 7974.
6 Limitations
The application of CFD in any area of expertise requires a fair amount of knowledge
and experience with both CFD and the phenomena under investigation. Until
recently, only CFD researchers and design specialists held the amount of knowledge
regarding building properties, installations and CFD that was needed to successfully
apply these simulation techniques in building design.
However, in the past years some software and research establishments developed
tools aimed at less expert users. These tools include advanced techniques that
automate much of the data specification process for common situations. Some were
specifically developed for use in building practice and the built environment.
However, even user-friendly CFD applications still require a fair amount of input.
In addition, sufficient knowledge concerning airflow and heat transfer mechanisms is

needed in order to formulate a useful CFD model.
1) CFD solutions can only be as accurate as the physical models on which they
are based.
2) Solving equations on a computer invariably introduces numerical errors such

as:
a. Rounding off Error, due to finite word size available on the computer.
b. Truncation error, due to approximations in the numerical models.
These errors will tend to zero as the grid is refined but this is not
always feasible due to limited resources and the software being used.
3) Boundary conditions, as with physical models, can only be as accurate as the

initial and boundary parameters specified to the numerical model.
Page 11 of 70
4) Physical phenomenon, a definitive understanding of physical phenomenon is
needed so one can determine what assumptions can enable a convenient
case through computational domain (specifications of fluids, turbulence
model).
5) Experience in conducting CFD studies should make one aware of the

limitations to modelling turbulence. As turbulence, there are still lots of
phenomenon to discover to generate an analytic approach.
6) Numerical method is all about mathematical knowledge. Partial differential

equations based mathematical topics are important. Modellers have to conflict
optimal numerical method to solve those equations. They will need to
understand what the problem is and develop a method to solve it.
7) Computational grid (Meshing) considers the numerical approach based on

one of the several numerical approaches such as finite volume method for
example. Having the wrong generated mesh structure, could results through
the solution process that are non-physical.
8) CFD tool with a commercial code that should use any numerical solver to
generate reasonable solution grids. However, the use of free research and
shareware tools are not always subject to the same level of quality assurance.
To describe a physical phenomenon into the computational domain, it is recognised

that there are several limitations that will need to be optimize. An awareness of every
sub-sections from beginning to end is important. In case of a commercial program,
some of these sub-sections may be directly optimized into program. There is nothing
inexplicably needed to be sufficiently educated and capable of using CFD tools to
apply the most appropriate analysis however, an awareness of its limitations is
needed.
Page 12 of 70
7 Modelling process
The following describes the modelling process for a performance-based design. These
are further discussed in the sections that follow.
Stage 1 – Specification outline (QDR)
Item
Description
no.
1 Define objectives of the CFD modelling.
2 Determine type and number of simulation(s) to be prepared to demonstrate
design objectives.
3 Decide how the results will need to be presented to demonstrate objectives
i.e. using velocity, speed, temperature, visibility, etc.
4 Collate reference material and / or any previous test results for use later when
checking the credibility of results.
Stage 2 – CFD modelling
Item
Description
no.
1 Selection of computational fluid domain boundary.
2 Selection of geometric detail to be represented in the computational domain.
3 Creation of the geometric model(s).
4 Mesh / grid generation.
5 Define physics for the simulation(s).
6 Select appropriate 'Sub models' (if applicable) including definition of sources
(and / or species) within the model (i.e. fire, contamination etc.).
7 Define appropriate boundary conditions.
8 Define appropriate initial conditions.
9 Select solver time / number of iterations, results to be obtained from the
solver, monitor points etc.
10 Run the simulation(s)
11 Interpretation of results: Sanity check – Check the results provide a
reasonable representation of real-life events.
12 Technical Review – Confirm the performance objectives been achieved.
Decide what further actions should be taken if objectives not achieved.
Stage 3 – Report & presentation of results
Item
Description
no.
1 Description of the objectives.
2 Description of the geometric model(s) and simulation(s).
3 Description / justification of the input parameters
4 Presentation and interpretation of the results.
5 Conclusions – Confirm the objectives have been achieved.
Page 13 of 70
8 Preparing the CFD Model
8.1 Definition of Computational Domain
The starting point for the application of CFD to the simulation of air movement, fire
and smoke movement in any building is to establish the computational domain for
the simulation, i.e. the limit of the region to be modelled.
The primary considerations are itemised below.

- Three-dimensional Domain.
Even in the simplest geometries, the air and smoke flows are three-
dimensional.
- Boundaries of the Domain.
The boundaries of the domain are a function of the objective of the CFD
simulation. They should encompass the region of interest and be located
where the flow conditions are known. The influence of an external flow, e.g.
wind, may require that the boundaries of the domain extend beyond the area
of immediate interest.
The extent of the domain will be influenced by the nature of the building being
analysed. For example, the boundaries of the domain for a fully enclosed,
mechanically ventilated basement car park will be formed by the walls, doors
(usually assumed to be closed), floors and ceiling of the car park. In an ‘open’
naturally-ventilated multi-storey car park, the boundaries of the domain are
likely to extend beyond the car park (and may include a representation of
other buildings in the immediate vicinity) in order to represent the effect of
wind on internal air and/or smoke flows adequately.
- Additional Factors Affecting Domain Boundary Selection.
- Computational Limits.
A CFD simulation that accounts for all the possible influences on air
flows or a fire and the induced flows in an enclosure may be too large
for the computational resources (processing power, memory and time)
available. In these circumstances, it is necessary to focus on the key
features affecting the air and/or smoke flow whilst ensuring that the
influence of the omitted factors does not compromise the objective of,
or the conclusions to be drawn from, the CFD simulation.
- Attached Volumes.
The volume of immediate interest may be connected to, and influenced
by, other volumes, e.g. floors of a multi-storey car park which are not
immediately adjacent to the fire floor. In such cases, it may not be
Page 14 of 70
possible or necessary to model the entire car park. The boundaries of
the domain need to be defined at locations where the flows between
the attached volumes can be considered to be minimal or where they
may be approximated by boundary conditions derived from
measurement or calculation.
- Domain boundaries.
The boundaries of the domain should be located such that they do not
adversely affect simulated smoke movement. For example, open (or
‘free’) boundaries should not be located close to the source of the fire
as smoke may be lost, when in reality it might re-enter the fire affected
region. (If smoke leaves the computational domain, it should be at
locations sufficiently removed from any induced flow (as a result of the
fire, the wind or from mechanically assisted means) which might
subsequently allow the smoke to re-enter the domain).
8.2 Details Represented in the Computational Domain
Modeller should establish the boundary conditions to include in, and exclude from,
the geometric representation of the model.
Any object which may have a significant impact on air flows or fire induced flows and
smoke movement should be represented within the model.
Typically, this should include some or all of the following items.

- Structural walls, floors and ceilings.
- Structural openings, columns and beams.
- Services, e.g. the geometric representation of HVAC ductwork.
- Stationary vehicles (Car Park).
Note: For car parks, the size, number and distribution of vehicles represented
in the car park will be a function of the car park site and the design scenario.
In defining the number of vehicles to be included in the model, it is important
to consider:
i) The objective of the CFD simulation.
ii) That the potential for an accidental fire is likely to increase with an
increase in the number of cars within the car park.
iii) That the risk to life is likely to increase with an increase in the number
of cars in the car park.
iv) That the presence of vehicles will affect the fire, environmental and
mechanically induced flows.
Page 15 of 70
Further guidance on the inclusion of vehicles in car parks is provided under
Section 12.1.7.
Figure 1: An example of the distribution of cars within a car park model

(note that cars may often be represented satisfactorily by simple blocks)
- Environmental factors both internal and external, such as internal stack effect,
wind and other objects affecting such flows, e.g. external buildings.
- Mechanically induced flows, e.g. impulse fans and inlets / extracts vents.
- Significant sources of heat (other than the fire) which might create a natural
convection flow that could interact with the fire induced flow.
- The heat and smoke (fire) source.
Factors affecting the geometrical representation of the model include some or all of
the following.
- The location of the object / source with respect to the fire and environmental
and mechanical ventilation induced flows (and the effect that the object /
source may have on the induced flows).
An object remote from the fire may have a lesser impact on the (bulk)
movement of smoke than an equivalent sized object located close to the fire.
- The size of the object / source with respect to the space modelled and to the
anticipated flows.
Page 16 of 70
- The size of the object with respect to the computational mesh (or grid) size.
An object that is smaller than the size of the computational mesh in the vicinity
of the object cannot be represented within the model (unless a sub-mesh or
porosity sub-model can be utilised).
- The computational mesh (or grid).
The nature of the computational mesh adopted by the CFD package may
affect the geometrical representation: for example, a mesh based on a
rectilinear coordinate system will only approximate curved or sloping surfaces.
The modeller should generate and justify the geometric representation of the model
on a case-by-case basis. The flow should not be significantly affected by any
geometric simplifications made.
It is relevant to consider that a CFD simulation is an approximation of reality (a
description of what might happen in a fire event). Increasing the geometrical detail
within the model will not necessarily increase the understanding of the bulk flows in
the building enclosure being analysed.
8.3 Computation Mesh

The computational domain is sub-divided into a large number of smaller cells.
The computational mesh, i.e. the size and configuration of the mesh cells, should be
designed to ensure that the following requirements are satisfied.
- The geometric details (e.g. shape and size of objects) are represented
appropriately.
- Where a fire is being modelled, the flow phenomena driving smoke movement
are resolved adequately.
- Fire area and thermal plume-sufficient detail (fine mesh) is needed to capture
the rise of the hot gases.
- Where applicable, the mesh interfacing (mesh splitting) should be avoided
where large exchange of information occurs. For example, in FDS, a fire and
its immediate surrounding vicinity should ideally be within a single mesh.
Another example would be the internal space of an atrium with a fire at its
base. This should ideally not be split horizontally due to large exchange of
information at the vertical direction as buoyant hot gases and smoke rise to
the top. This also applies to smoke shafts. Where it is unavoidable, the CFD
user should ensure that there are no issues associated with mesh to mesh
connections and demonstrate there is no impact on the results.
- The region adjacent to the ceiling – including the ceiling hot gas layer (and,
particularly, ceiling layer flows) – should include:
o A significant number of cell layers normal to the ceiling when a
‘structured mesh’ is adopted.
Page 17 of 70
o Slowly inflated mesh cell sizes normal to the ceiling when an
‘unstructured mesh is adopted.
- Mechanically induced / assisted flows are represented appropriately. Ensure
that:
o Sufficient mesh cells are used to describe the dimensions of the fan /
inlet / extract in the plane normal to the direction of flow (typically,
several will be required – the guidance of the product developer should
be followed).
o Changes in the dimensions of the mesh cells in the direction of the flow
do not influence the flow characteristics.
- The guidance of the product developers is followed to ensure that mesh cell
size selection is consistent with the modelling approach adopted.
- Mesh cells should not be subject to significant distortion, i.e. they should have
low aspect ratios (preferably close to unity in the vicinity of the fire). Guidance
should be sought from the product developers to assess the maximum
permitted mesh cell distortion.
- Ideally, an investigation of the sensitivity of the results to the mesh cell size
adopted should be undertaken.
Further discussion regarding mesh quality is provided in Appendix B.
Figure 2: An example (using CFX) of a non-rectilinear mesh showing reduced

mesh size at the fire source and an inflated mesh size at the boundaries
Page 18 of 70
Figure 3: An example (using FDS) of a rectilinear mesh in a large car park
8.4 Physical Sub-models
The physical sub-models selected define the equations to be modelled within the
CFD simulation.
The important physical mechanisms governing the flow and which need to be
captured include the following.
- Combustion
- Buoyancy
- Turbulence
- Radiation
- Heat transfer at walls
In all cases, guidance should be sought from the developers of the CFD simulator as
to the appropriate model to adopt.
These models are discussed in more detail in Appendix A.
8.5 Fire Source Specification
Where the modelling includes a fire, the location, size and characteristics of the fire
need to be specified for the CFD simulation.
Modelling a fire is not usually necessary when considering environmental (daily)

ventilation or for smoke clearance.
Page 19 of 70
Fire Scenarios
Selection of the fire scenario(s) to be investigated is a complex process requiring:
- An assessment of the objective of the investigation.
- An understanding of the likely flow processes within the area being studied.
- An understanding of the fire hazards, i.e. the sources of fuel and ignition.
Typically, these lead directly to the definition of the fire location.
Experimental and / or published data can then be used to define the fire size and
characteristics. Recommendations for suitable design fires can be found in BS7974,
BS7346, SFPE Handbook and BRE publications.
For more unusual applications such as car parks containing car stacker systems
there may be little data available on fire loads and fire spread. In such cases it is
particularly important that any estimates for fire size and characteristics are agreed
with the approving authorities while developing the CFD model.
Fire Heat Release
The rate of heat release is a prescribed input to the CFD model for both the
volumetric heat source model and the combustion model.
The volume over which the heat is released (which is dependent upon the footprint
and area of the burning material) is an additional input when employing a volumetric
heat source model.
A combustion model predicts the heat distribution in the flaming region above the
seat of the fire; the area over which the heat is released must be specified.
Assessing heat release profiles (Heat Release Rate vs Time) in combustion models
is often necessary in determining whether the fire source is ventilation or fuel
controlled, when seeking specific objectives from the heat source
Fire Smoke Production
Production of smoke is dependent upon the properties and physical state of the
combustible materials, the quantity available and the availability of air supply to the
flame.
It is usual for the smoke production rate to be linked to the heat release rate by a
‘yield factor’ (representing the production of smoke) which has been determined
experimentally for a wide range of materials and conditions.
Smoke is then assumed to be generated uniformly over a volume (volumetric heat

source models) or an area (combustion models).
Page 20 of 70
This approach is suitable except for those cases where there is a significant change
in the rate of smoke production, e.g. from a well-ventilated fire to an under-ventilated
fire, as the CFD model will normally assume that it is constant.
Figure 4: Detail of smoke production and sprinkler droplet distribution
Fire Spread
In some scenarios, mainly involving combustion models and developing fire, flame
spread may be seen as an important factor in the total Heat Release Rate. In these
incidences it is important to identify the material properties of neighbouring materials
close to the seat of the fire to assess whether these materials will add to the flame
spread.
The source of fire and smoke is further discussed under Appendix C
8.6 Initial and Boundary Conditions

Any mechanism which is external to or within the computational domain, but which
significantly influences the behaviour of the flow within it must be represented.
Typical examples include the following.
- Initial flows present within the computational domain prior to the simulation.
- Flows in / out through doors, windows, openings, vents or mechanical
inlet/extract systems.
- Change of momentum and / or energy in simplified representations of
mechanical systems such as jet fans.
- Energy transfer (in the form of heat) at (to / from) walls.
- Sources of mass, momentum and / or energy, e.g. at the fire, or through the
release of a suppressant.
Page 21 of 70
The initial and boundary conditions must be defined by the user. Establishing
representative initial and boundary conditions can be a major challenge, particularly
where it is necessary to prescribe the level of turbulence associated with an initial /
boundary condition.
The level of detail available will vary with the CFD program used. Not all options
discussed are available in all CFD programs.
Initial Conditions
The initial flow conditions may need to be established by analysis and / or simulation
prior to undertaking the simulation.
Inlets and Outlets
At an open (or ‘free’) boundary where the flow will be mainly influenced by what
happens inside the computational domain, a constant pressure boundary (which
implicitly assumes that the flow is fully developed) is applicable. Such boundaries
have to be placed where the ‘fully developed’ assumption is either valid or has little
impact on flow inside the domain, i.e. away from any fire and at locations where the
flow is not expected to experience strong spatial variations.
Where flow is driven principally by mechanisms external to the computational

domain, such as forced ventilation, the boundaries of the computational domain
should be located where the flow conditions are known and can be specified.
Specification of the flow at the boundaries might require further analysis which can
be provided either by measurements or additional modelling (including CFD).
Detailed analysis is necessary when the flow across the boundary is expected to be
complex, e.g. in a space partially open to the atmosphere and for which the
surroundings influence the direction and velocity of the incoming wind.
Walls
In order to save computing time, universal wall laws are often applied as wall
boundary conditions. These functions preclude the need to resolve in detail the
large gradients of temperature and velocity near walls, which would necessitate a
large number of mesh cells.
Instead, momentum and convective heat fluxes between the near-wall nodes of the
computational mesh and the walls themselves are assumed to be described by
‘universal laws of the wall’. These laws include parameters that account for the
roughness of the walls and lead to lower velocities in the case of rough walls. For
these laws to be valid, the near-wall cell mesh size must be chosen such that the
first mesh nodes are specified to be at a distance from the wall which is related to
the local turbulent Reynolds number.
Page 22 of 70
Uncertainties associated with the use of wall laws are of two types: those due to the
difficulty in complying with restrictions on the location of the near-wall nodes across
the whole domain, and those due to the fact that the wall laws are strictly only valid
for idealised situations.
The CFD user has to specify how heat transfer is to be modelled at the walls. One
possibility is to assume nil heat transfer, i.e. an adiabatic wall. The other extreme is
to assume a constant wall temperature, leading to maximum rates of heat transfer.
The heating of the wall can also be modelled, by solving for thermal conduction
within the wall (requiring a much finer mesh resolution near the wall and specification
of the properties of the wall).
Typically, the adiabatic wall assumption leads to faster smoke propagation, with the
smoke being more concentrated in the hot gas layer near the ceiling and, therefore,
less smoke being predicted at lower levels, when compared with the fixed wall
temperature approximation. Appropriate assumptions should be made depending on
the scenario being modelled, the objectives and acceptance criteria. The
assumption made is likely to be more important in small enclosures with hot smoke
than in large volume spaces where smoke temperatures may be closer to ambient.
Radiation adds further complications, which are not dealt with here. Further
discussion on radiation is provided in Appendix A.
Fire-dependent Conditions
As the fire is developing, it may change the conditions inside the computational
domain. Window and / or structural component failure will lead to a change of
boundary conditions in the vicinity of the failure.
Guidance on the assessment of such behaviours is outside the scope of this

document.
The results of CFD simulations are influenced by the boundary conditions. It is

essential, therefore, that the user specifies boundary conditions appropriately and
understands the key role that they play. However, usually not all of the required
boundary conditions will be well-defined. For instance: turbulence parameters as
flow enters the computational domain are typically unknown; there may be
uncertainty in wall heat transfer coefficients; fire sources and fire growth rates or
heat loading may be ill-defined; events external to the selected computational
domain, such as pressure distributions arising from natural or forced ventilation, may
in reality affect flow inside the domain – these couplings should be encompassed if
they are likely to influence the outcome of the modelling. If doubt remains, the CFD
user should carry out a sensitivity analysis to evaluate the influence of a range of
plausible values for boundary conditions on the predictions.
Page 23 of 70
9 Presentation of Analysis and Results
9.1 General
The presentation of completed analysis should provide results which are clear and
unambiguous. Results need to demonstrate that the design, using either
comparative or absolute deterministic analysis, is acceptable. Therefore, the agreed
fire scenarios should be presented with evidence the performance explicitly is shown
in relation to the acceptance criteria set.
Visual and graphical results should be used in relation to the acceptance criteria with
reference to the fire scenario assessed. As CFD is deterministic the use of graphical
data plotted against acceptance criteria is the most useful method of demonstrating
compliance.
Where a time dependent CFD model is used to achieve a steady state result, a
timeline should be shown at regular intervals to demonstrate that the conditions have
become stable and steady state. Additional information at specific times may be
shown to highlight unusual flow behaviour, for example to demonstrate the action of
fans on start-up.
Care should be taken to ensure the results reflect the fire scenario being investigated
and where a sensitivity analysis is included, the comparative outcome should be
clearly shown.
Key items which should be considered in the presentation of results are:

a) Images or graphical representations of key output parameters such as
temperature and visibility (or smoke obscuration/distribution) should be
displayed with reference to the agreed tenable bounds.
b) Results which capture the relevant fire dynamics, in particular temperature,
smoke distribution and flow velocities at the relevant locations should be
checked.
c) Metrics to monitor and report should be chosen accordingly to the range of
data needed; velocities, temperatures etc. The chosen data ranges should be
universally adopted for every presented image to ensure all results are
directly comparable. However, it is recognised that in some cases of detailed
analysis this is not possible when highlighting specific phenomena. Where
the scale has change this should be clearly identified.
Page 24 of 70
d) The locations of the images taken must be relevant to the issues being
investigated, showing the flow regime at areas of interest, and at relevant
elevations or regions of activity.
e) It is useful to display images of velocity vectors to demonstrate their effects,
around fans or doorways for example.
f) Results should show mean values or where local hot spots exist;
g) Results (both graphical and deterministic) should be scaled to show relevant
temperature profile for example as being fairly compared. The scales and
units should be consistent across the cases analysed.
h) Use of repetitive imagery should be avoided.
i) Images must clearly explain what is being shown, and the location of the local
effects.
9.2 Presentation of Results
The results of the analysis should be documented and may be provided in the form
of a report, with any necessary animations attached in electronic form (memory stick
or FTP site).
The documentation should include at least the following information:

- A simple summary description of the case being assessed.
- The design criteria and objectives of the analysis agreed in the QDR.
- The rationale for the fire scenarios investigated
- Details of the CFD model used
- The results of the analysis with discussion points
- A conclusive statement as to whether the design criteria and objectives have
been met.
The reports should be fully referenced and should provide sufficient information for
the reviewer to complete the analysis independently and reach the same
conclusions.
Page 25 of 70
Figure 5: Image showing air residence time in a car park
For time dependent analyses, graphical results should be presented wherever

possible to qualitatively show conditions plotted against a timeline (although it is
recognised that this is not always possible).
Scales, ranges and units should be chosen accordingly to reflect the range of data
collected; velocities, temperatures etc. It is suggested that the chosen data ranges
should be universally adopted for every presented image to ensure all results are
directly comparable. However, it is recognised that in some cases of detailed
analysis this is not possible when highlighting specific phenomena.
When necessary, a sensitivity analysis should be carried out and presented such
that it allows important outputs between different scenarios to be easily compared.
The locations of the images taken must be relevant to the issues being investigated,
showing the flow regime at areas of interest, and at relevant elevations or regions of
activity.
It is useful to display images of velocity vectors at the mechanical fans to

demonstrate their effects. Images of temperature and visibility (or smoke
obscuration) should be displayed with reference to the agreed tenable bounds.
Page 26 of 70
Figure 6: Section showing velocities at jet fan locations
10 Modelling Lobbies, Corridors and Stairs

The modelling of residential, commercial lobbies and corridors is undertaken to
demonstrate that it satisfies the proposed design criteria outlined in the relevant
guidance for the application. As there are different approaches that can be applied to
demonstrate compliance when using CFD for residential corridors and commercial
lobbies, the modeller must ensure that the input parameters and approach adopted
forms a conservative assessment of the proposed design.
When undertaking CFD analysis, the primary objective is to protect the stairs from
ingress of smoke (or maintain smoke free) or may also involve returning or providing
tenable conditions to a space such as the common corridor or lobby depending on
the tenability criteria outlined in the guidance that the design has adopted. Typically,
acceptance criteria will be to demonstrate smoke free conditions in the stairs
throughout the simulation. In cases where corridors travel distances are in excess of
code requirements, tenability criteria will also be required in the corridors both
escape and fire-fighting phases.
Page 27 of 70
Modelling stairs, common corridors or lobbies can be referenced in the Smoke
Control Association’s – “Guidance on smoke control to common escape routes in
apartment buildings (Flats and Maisonettes)”. When an approach from the SCA
guide is not being applied, the modeller must provide suitable rationale and
justification for this deviation from the guidance.
Figure 7: Section showing smoke prevented from spreading into the stair from the corridor
11 Modelling Large Enclosure Buildings
Smoke movement in large volume buildings may involve smoke from a smaller
space flowing into the larger space via a spill plume at the void edge. However, in
some cases it may be appropriate to model a fire on the base of the large volume
with smoke flowing directly into the high-level space. Smoke control might be
provided for a number of reasons including to support the means of escape strategy
or to assist fire-fighting operations. The objectives of the modelling and fire
scenarios to be modelled should be identified by the fire engineer prior to beginning
the modelling. Examples of typical objectives for large volume buildings are
discussed below:
11.1 Storage, distribution or industrial buildings
Such buildings often have large plan areas and mezzanine levels which are open to
the levels below. The key fire strategy issue tends to be means of escape in cases
where travel distances exceed the recommendations of prescriptive guidance. In
Page 28 of 70
such cases a CFD analysis may form part of an ASET vs RSET analysis, e.g. with
the objective of showing occupants have time to escape before smoke builds down
to a level at which it affects the escape routes.
Large volumes provide an inherent margin of safety and smoke control might not be
necessary to support a means of escape strategy. However, smoke control might be
provided for asset protection or to support firefighter access into the building.
11.2 Atrium Buildings
11.2.1 Open Atria
In atria which are open to some or all of the floors the main fire strategy issue is the
impact of smoke spreading via the atrium to floors above the fire floor. Smoke rising
though the atrium might build down below the roof of the atrium to affect occupied
floors. More complex atrium geometries can result in more complex smoke
movement, for example with several spill plumes. A CFD analysis may form part of
an ASET vs RSET analysis, with the objective of showing occupants have time to
escape before smoke builds down to a level at which it affects the escape routes.
Smoke extract fans or natural smoke vents are typically required at the head of the
atrium to support such a strategy.
11.2.2 Enclosed Atria
Where atria are enclosed a fire is less likely to represent an immediate risk to the
levels above. However, in buildings with long evacuation times or residential
buildings with stay put strategies, smoke control may be provided to control the
temperature of smoke within the atrium with the objective of allowing the use of non-
fire rated glazing or glazing which is fire rated for integrity only. Examples of
buildings where this might be appropriate include offices with phased evacuation
strategies, hospitals or residential buildings.
11.3 Shopping Centres
Shopping centres are generally designed such that smoke is allowed to flow out of
the unit and into the mall where it is dealt with by a mall smoke control system. Mall
smoke control systems are normally designed to maintain a clear layer at least 3m
above the highest mall level for a steady state fire. Guidance recommends the size
of mall smoke reservoirs is limited because of the risk the smoke layer will lose
buoyancy and that smoke flowing under balconies is channelled by screens to limit
the width of the spill plume and therefore the volume entering the reservoir.
However, the prescriptive guidance was developed in the late 1980s and hasn’t kept
pace with trends in shopping centre designs which include larger volume mall
Page 29 of 70
spaces, malls of more than two storeys, a desire to minimise the number of smoke
screens and semi external malls.
CFD studies in shopping centres might be carried out to demonstrate that variations
from the prescriptive guidance will not compromise the safety of occupants or
firefighters.
Figure 8: Section showing velocities through shopping mall
11.4 Mesh Size
When modelling large volumes multiple meshes are generally required in order to
achieve practical processing times. For example, a relatively fine mesh should be
used in the near field. For example, this might include the fire location, the fire
plume and where smoke spills out of the fire room into the large volume space.
Outside that region a coarser mesh is generally adequate. However, it is essential
that the mesh has adequate resolution to model important features which might be
remote from the fire such as vents, mechanical extract points or replacement air
inlets.
11.5 Fire
11.5.1 Scenarios
Locations to be modelled should be identified by the fire engineer taking into

consideration the objectives of the modelling. If the objective of the modelling is to
demonstrate smoke does not build down to a level at which it affects escape routes,
then the most complex smoke path is likely to represent the most onerous scenario.
Page 30 of 70
However, if smoke temperature is the key criterion it may be appropriate to model a
fire with a less complex smoke path and less entrainment into the smoke plume.
This might apply, for example, to an atrium temperature control system where a fire
on the atrium base or on a higher floor might result in the highest smoke
temperatures.
11.5.2 Fire Size
The proposed fire size and fire growth rate should be identified by the fire engineer.
Sources of guidance on design fires include BS 7974, BS 7346 and the SFPE
Handbook.
If the CFD study is being carried out as part of an ASET vis RSET analysis a time
dependent development fire will generally be appropriate. Where a developing fire is
modelled it may be appropriate to cap the fire at a certain size to take account of any
suppression system provided.
However, there may be circumstances where it is adequate to assume a steady

state fire in order to simplify the modelling. For example, if carrying out an ASET vs
RSET analysis in a large storage or distribution building the margin of safety might
be such that modelling a steady state fire is adequate.
There are also applications where design has traditionally assumed a steady state
fire. These include smoke control systems in shopping centres and atria.
11.6 Activation of Smoke Control System
The timeline for the activation of the smoke control system should be identified by
the fire engineer prior to the modelling. Where systems operate automatically and
are linked to the fire alarm system this could be informed by a separate modelling
exercise such as modelling of smoke detector activation. This may also need to take
into account ramp up times for smoke extract fans as well as opening of dampers
and automatic opening ventilators (AOV).
However, the above factors are less critical in large volume spaces than in small
volumes and it may be appropriate to make some suitably conservative
assumptions. When considering steady state designs it would normally be adequate
to assume the smoke control system is active immediately.
11.7 Simulation Time
The simulation time needs to be chosen to give results that are appropriate to the
scenario being modelled. For a time dependent fire the simulation time should be
based on the time a particular action occurs. For example, this might relate to
evacuation time or fire brigade intervention.
Page 31 of 70
For a steady state scenario, the simulation time should be sufficient to demonstrate
that the model is in steady state and thermal equilibrium. This occurs when
conditions such as the smoke layer height and temperature are no longer changing.
In a large volume space this may require a simulation time of the order of 20
minutes.
11.8 Acceptance Criteria
Acceptance criteria should be defined by the Fire Strategy and ideally agreed with
the approving authorities prior to carrying out the modelling. For example,
acceptance criteria might be a smoke clear layer below which occupants can escape
unimpeded or conditions which allow escape or fire-fighting through cool dilute
smoke which remains tenable. Guidance on suitable tenability criteria is given in BS
7974.
The tenability criteria which are expected to be important should be identified prior to
the modelling and the model should be developed
12 Modelling Car Parks
Ventilation of covered car parks is usually recommended in order to limit

concentrations of carbon monoxide (CO) and other vehicle emissions in day to day
use of the car park and to remove smoke and heat in the event of a fire. The same
equipment is often used to satisfy both requirements.
Computational Fluid Dynamics (CFD) analysis is rarely used as the primary design
tool for car park ventilation systems. Many systems simply comply with the
prescriptive recommendations in Approved Documents B and F and do not require
performance analysis. When alternative systems are proposed, for example, the use
of impulse ventilation systems and in particular those designed to assist fire-fighting
access or protect means of escape, the design is usually initially developed using
other methods and may then be subjected to CFD analysis for fine tuning of the
design and to demonstrate to approving authorities that the system is likely to
perform satisfactorily.
The main advantage of using CFD for these procedures is that this can be carried
out prior to installation of the ventilation system, providing confidence to all parties
early in the project. It is much easier and cheaper to agree expectations and correct
any problems at this stage than after testing upon completion of the installation.
Page 32 of 70
Due to concerns about damage it is rare for testing to be carried out using a real fire
or any significant heat source and a correctly modelled CFD analysis can provide
relatively accurate simulation of what may happen within the car park.
Since car park ventilation systems are usually dual purpose, providing ventilation for
vehicle fume control in normal conditions and for smoke clearance or smoke control
in fire conditions, consideration should be given to which operational modes require
CFD analysis as the scenarios and operating conditions will be different depending
upon the choice made.
It is important to note that, while CFD modelling provides highly detailed outputs, the
results should be regarded as snapshots representing a likely outcome and an
indication of performance and not as definitive statements of conditions in use.
The scope on this guide is limited to car parks occupied by internal combustions
vehicles and is unable to offer guidance for vehicles powered by alternative fuel
sources (e.g. electrically battery powered, hydrogen fuel cell). Further consideration
is required in such instances and dealt with in the QDR process.
The SCA document “Design of Smoke Ventilation Systems for Loading Bays &
Coach Parks” offers further guidance for design of ventilation system for loading
bays, service yards and coach parks and lists the options available to the design
engineer.
12.1 Acceptance Criteria
Before detailed modelling is completed and results presented it is vital that outputs
are agreed (with respect to both the design objectives and the acceptance criteria)
and approval for the modelling methodology is attained. The acceptance criteria is
normally defined by the Fire Strategy and agreed with the approving authorities. Part
of this agreement should detail a method of assessing the model’s performance in
relation to prescribed values.
Where a comparative approach is used, results should be compared directly to the

agreed code compliant solution with variables and areas of comparison agreed.
Where a deterministic approach is used, limits for visibility, temperatures, radiation

etc should be agreed.
For car parks, there are seven specific main issues that require consideration.
These relate to:
- Vehicle emission ventilation
- Smoke clearance
- Safety of evacuating occupants
Page 33 of 70
- Safety of fire service personnel & their ability to attack the fire
- Fire spread and local effects
- Performance throughout the car park
- Error checking
Not all of these issues are necessarily relevant for all projects. The relevant issues
should be selected for each project. The following sections outline the key aspects
which should be considered in each area and be shown to be acceptable to the
approving authority.
It is not the intent of this document to set specific acceptance criteria. Rather these
should be agreed with the approving authorities based on recognised published
documents, e.g. BS7974, BS7346-7, CIBSE Guides, BS EN 12101-11, Approved
Documents B and F.
12.1.1 Vehicle emission ventilation
Objective: The objective should be to show that the whole car park is adequately
ventilated and that either the ventilation rate or the maximum CO level
meets the recommendations of Approved Document F to the Building
Regulations (or equivalent outside England and Wales).
Comparative analysis:
The Approved Document sets some basic prescriptive requirements for vehicle
emission ventilation. Where these are not followed CFD analysis can be used to
show equivalence.
The simplest way to show equivalence is to demonstrate that the overall ventilation
rate matches the basic prescriptive requirement and that the car park has no pockets
of stagnant air.
Deterministic analysis:
A deterministic approach would be to show that, under normal and peak traffic in the
car park, CO levels do not exceed the recommendations in the Approved Document.
This approach requires an understanding of likely traffic flows through the car park
and of vehicle CO emissions.
Page 34 of 70
12.1.2. Smoke clearance
Objective: The objective should be to show that the whole car park is uniformly
cleared of smoke by ventilation, without stagnate areas, and that the
ventilation rate meets the recommendations of prescribed guidance
(ADB, BS9991, BS9999).
The Approved Document sets some basic prescriptive requirements for smoke
clearance. Where these are not followed CFD analysis can be used to show
equivalence.
A way to show equivalence is to demonstrate that the overall ventilation rate

matches the basic prescriptive requirement and that the car park has no pockets of
stagnant air. An area with stagnant air is considered to be such areas where the age
of air or rate of purging is not appropriate, for example, where the air change rate is
less than 50% of the overall design value; velocity of air is less than 0.1m/s; or where
the mean age of air is greater than twice the average.
12.1.3 Safety of Evacuating Occupants
Objective: The objective of this stage should be to show that occupants can
reach a place of relative safety during a fire.
Primary escape routes should be included in any comparative analysis and

conditions shown to be equal to or better than the agreed code compliant solution
used for comparison. Comparisons should include, where appropriate, visibility,
temperature and radiation.
Under a deterministic approach the objective should be to show that the Available
Safe Egress Time (ASET) for the occupants will be greater than the Required Safe
Egress Time (RSET) plus a suitable safety margin in the particular scenario being
modelled.
In assessing the safety of occupants as they evacuate it is recommended that a

recognised approach be used. Potential methods include for example:
a) Clear layer assessment

b) Tenability criteria
Page 35 of 70
Clear layer assessment requires that as the occupants travel along the evacuation
routes they are not exposed to smoke and that any smoke above the occupants is
maintained at a temperature low enough that occupants are not affected by
untenable levels of heat radiation. This can be difficult to achieve in car parks, where
the headroom is usually restricted.
Figure 9: Maintenance of a clear layer below smoke
Tenability requires that it be shown that occupants escape in tenable conditions and
that their exposure to heat and smoke is limited. This approach typically requires
that visibility, temperature, radiation, CO and CO2 be assessed and that a Fractional
Equivalent Dose (FED) type analysis carried out.
Figure 10: Limiting spread of smoke using jet fans to maintain tenable conditions
Page 36 of 70
12.1.4. Safety of fire service personnel & their ability to attack the fire
Objective: The objective of this stage should be to show that fire service
personnel will be able to enter the car park and safely reach a position
where they can attack the fire.
Primary fire service access routes should be included in any comparative analysis
and conditions shown to be equal to or better than the agreed code compliant
solution used for comparison. Comparisons should include where appropriate
visibility, temperature and radiation.
Once occupants have been safely evacuated it should be shown that the proposed
system will allow fire service personnel to safely enter the fire zone and attack the
fire.
Unlike evacuating occupants, fire service personnel can be expected to have the
additional protection provided by breathing apparatus and protective clothing.
As such the tenability criteria and needs of fire service personal are very different
from evacuating occupants. These needs include the ability to:
a) Find the location of the fire.

b) Gain safe access to a position from which to deliver water to the base
of the fire.
c) Retreat to an area of relative safety if necessary.
It is recommended that the tenability limits for the fire service be considered at the
outset of the project and agreed with the local fire service prior to any modelling
being carried out as these may be dependent on the particular equipment local fire
service personnel have available.
General criteria may however be as suggested in Section 10 of BS 7346-7; designs

should be such that fire fighters can move through substantially clear smoke-free air
when approaching the fire up to a distance of 10m from that fire
12.1.5 Fire Spread and local effects
Objective: The objective is to show the effect of the fire on any nearby objects
including the structure, flammable materials and any required smoke
control systems such as fans.
Page 37 of 70
Key nearby objects should be included in any comparative analysis and conditions
shown to be equal to or better than the agreed code compliant solution used for
comparison. Comparisons should include, where appropriate, temperature, radiation
and the effects of flame spread.
Under a deterministic approach the intent should be to show that these nearby
objects will not be detrimentally affected or that, if they are, the safety of occupants
or the fire service will not be affected.
The assessment of the effect on nearby objects will depend on the type of objects
that are located nearby. The following outlines some of the potential areas that may
need consideration. It is by no means a comprehensive list and should be used as
indicative of the types of investigation that may take place.
Structural Stability: It should be shown that during an extended period of fire the
surrounding structure of the car park will maintain its structural
integrity to afford evacuating occupants time to escape and the fire
service sufficient time to fight the fire without collapse of the building
Fire Spread: The potential for fire spread to surrounding flammable materials
should be assessed. This may include the potential for the fire to
spread to adjacent vehicles, fixed insulation or storage areas. The
intent of this analysis is to confirm that the selected fire size is
appropriate and that the fire will be unlikely to grow further than that
modelled.
System Failure: It should be shown that any fire safety related system will either not
be detrimentally affected by the fire or, if affected, its failure will not
result in risk of life to evacuating occupants or the fire service.
Study of the above areas can be carried out using a quantitative and/or qualitative
approach depending on the particular configuration within the car park. In either
case, generally it will be necessary to consider conductive, convective and radiative
effects from the fire and smoke.
Page 38 of 70
12.1.6. Performance throughout the car park
Objective: The objective is to show that the proposed system will offer the
necessary levels of protection throughout the entire car parking area
for any credible fire location.
Comparative and Deterministic analysis:
Typically, this can be achieved by carrying out one or more CFD models with the
selected design fire located in the worst credible position(s). The position(s) should
be selected taking into account the car park geometry, selected smoke management
system and routes of escape. The objective should be to ensure tenable conditions
throughout the car park for evacuating occupants and fire service personal entry.
If the required conditions (identified previously in this section based on a

comparative or deterministic approach) can be shown to exist for both the escaping
occupants and the entering fire service personal for a fire in the selected position(s),
then it can be reasonably assumed that that the selected smoke management
system will provide adequate protection for a fire in other locations within the car
park. For this assumption to be made then it is critical that the fire locations selected
are indeed the worst credible positions.
The selection of the worst credible fire locations can be identified through the use of
quantitative and/or qualitative methods. However, it is recommended that the
locations be discussed and agreed with the relevant authorities at an early stage in
the project.
12.1.7. Error checking
Objective: The objective is to allow assessment of the model in terms of error

checking.
Comparative and deterministic analysis:
One of the most important aspects of any modelling presentation is that the
approving authority or checking party be provided with sufficient information to allow
a model to be checked for general errors. It is not intended that the modeller provide
sufficient information under this heading for the specific CFD package being used to
be assessed, (as the package being used should already have been assessed to
determine its suitability and its use been agreed with the relevant authorities) rather
that errors within the specific simulation have not occurred.
Flow patterns: It is important to confirm the flow pattern at the fans is correct
along with all the elements that could impact their flow field
Page 39 of 70
(silencers, deflection louvers tec.). The flow patterns should
show that there are no dead zones or stagnant areas. If there
are any ramps in the car park then the flow through between
floors should be carefully checked for anomalies & for overly
strong or weak flows. As general good working practice for the
modeller it is important to highlight any abnormal results early
on, before being committed to final runs.
Modelling cars
(obstacles): Modelling cars or other mobile obstacles is generally not
required. However, when unrealistic large-scale vortices are
observed, whereby large areas of the car park have air flow
speeds in excess of 2.5m/s, the same simulation should be
repeated with evenly distributed car obstacles occupying 50% of
the available car spaces. The obstacles ought to disturb the
vortex so to bring it in line with a more realistic scenario.
Convergence
criteria: The output of steady state models should be checked for
convergence criteria if applicable (it is noted that models vary in
the type of convergence criteria or convergence bounds used
and hence this should be related to the specific model).
Plausibility checks: Flame heights, smoke temperatures and plume mass flow rates
should be checked against historical data or hand calculations
where possible to ensure that the correct fire conditions are
being created.
Page 40 of 70
13 Modelling guidance for designing smoke control
system
The following tables are considered ‘best practice’ guidance in designing smoke
control systems through use of CFD analysis. The recommendations are intended to
build conservatism into the models using simplified assumptions to mitigate against
known or unknown variable characteristics in buildings which may have a significant
influence on the outcome of a study.
13.1 General modelling of smoke control systems:
Parameter Description Recommendation

Model Key elements, I-beams, - Building geometry should reflect the intended or as built
domain(s) pillars, ceiling bulk heads drawings.
to be included in the
model for accuracy - Ensure software is fit for purpose for the study, particularly in
high rise buildings.
- Creating extended boundary domain to external vents where

external conditions could influence the area of interest.
- Model domain should ensure all the different zones are

connected to each other.
Mesh sizing Suitable mesh sizing - Should be sufficient to capture the physics (velocity, pressure,
selection based on fire temperature) in the areas of interest.
and domain size.
Initial Ambient conditions - Material properties should be sufficient to capture the relevant
boundary (temperature, density, physics of the actual areas being studied. Default values may
conditions etc.) not produce realistic results.
Relevant external
air/wind
Seasonal climate
consideration
Fire source Suitable fire source in - Heat and Energy conservation balances during development or
and location terms of size, heat flux steady state should be checked to ensure numerical validity.
and heat release rate.
This also includes the
appropriate fire
location(s).
Oxygen Ventilation opening to - The fire size should be checked against the HHR profiles to
Source for sustain the design fire ensure steady state fire validity.
fire size (artificial or actual)
- Ensure fire development or in steady state is balanced with a
suitable boundary opening to support it. Incorporating window
breakage may affect the fire size, airflow dynamics and will also
influence the proportion of smoke and heat lost to outside rather
into the area being studied.
- Further discussion on artificial ‘low-level’ vents is provided in

Appendix C
Page 41 of 70
Heat Balanced heat source - Observe HRR profiles to validate the design fire performance
Release (fuel / ventilation
Rate controlled)
Smoke Suitable soot yield - Soot yield production should be equivalent to the combination of
Density (soot a 10% yield with heat of combustion between 19 000 to 20 000
yield) KJ/kg
Natural Air Natural source of - Domains should be extended beyond the area of immediate
inlet replacement air interest where external flows may influence inlet air through an
opening
- Avoid confusion between Geometric and Aerodynamic area.

Geometric area is represented as a hole in the wall whereas
aerodynamic is related to the vents performance efficiency which
is only obtainable through approved testing standards
(BS EN 12101-2).
- Incoming velocity from natural vents needs to be taken into

account as velocities in excess of 5m/s could impede escaping
occupants who need to travel past the vent(s).
Simulating Accounting for door - For Fire-Fighting access, fully open doors between the fire room
occupant opening times during and the FF access route (e.g. FF stair) should be modelled.
movement escape and; or fire-
fighting access - Airflow velocity along the escape route should be observed.
According to PD 7974-2, air velocity in excess of 5m/s could
impede escape.
Smoke Modelling airflow - Vents should be modelled with their actual characteristic width or
Extract / terminals as vents as an equivalent free area using the actual width dimension.
Exhaust
Terminals - The velocity of inlet air from a supply vent needs to be taken into
account as velocities in excess of 10 m/s could impede escaping
occupants who need to travel past the vent(s).
Plant start up Fan ramp up, damper & - Spontaneously starting fans and opening vents is not realistic,
time vent opening / closing therefore fan and damper(s) linked to mechanical extract system
times should include ramp up opening times of at least 10 seconds
following activation.
- For natural vents linked to a natural smoke exhaust system,

should include a minimum opening time of 30 seconds following
activation/detection.
Page 42 of 70
13.2 Lobbies, Corridors and Stairs:
The guidance given below is provided in addition to Section 13.1 and is relevant to
Lobbies, corridors and stairs.
Model Key elements, I-beams, - The entire stair enclosure height linking the relevant corridors
domain(s) pillars, ceiling bulk heads or/and lobbies should be included in the model. Where relevant,
to be included in the final exit routes should also be included.
model for accuracy
Mesh sizing Suitable mesh sizing - For smaller enclosures such as corridor and lobbies, a typical
selection based on fire mesh size of 0.1m near field and 0.2m far field to the fire source
and domain size. are considered suitable.
Initial Ambient conditions - Extreme temperatures differentials between different interacting

boundary (temperature, density, domains should be modelled. These should be particularly
conditions etc.) assessed in tall high-rise buildings where internal stack effect is
likely to occur.
Relevant external
air/wind
- Material properties should be sufficient to capture the relevant
Seasonal climate physics of the actual areas being studied. Default values may
consideration not produce realistic results.
Fire source Suitable fire source in - Fire parameters should be substantiated from available
and location terms of size, heat flux published design guides, experimental data and / or empirically
and heat release rate. calculated. SCA Residential Guide offers guidance appropriate
This also includes the sizing of residential building fires. Further information is provided
appropriate fire in Appendix C.
location(s).
- For sprinklered fires, it acceptable to assume that the fire will
develop until the suppression system activates and then remains
fixed thereafter. Depending on the context of what the modelling
objectives are, the cooling effect of water droplets may not need
to be accounted for in certain applications.
- Various locations should be considered. For corridors, a simple

approach would be to test fires closest and/or furthest from the
stair, particularly in extended single direction escape corridors. A
minimum two fire locations should be selected where a corridor
has a dead-end section (>5m) beyond and extract/inlet terminal.
In such case, a fire location furthest from the dead-end section
should be selected.
Oxygen Ventilation opening to - Using artificial opening requires careful consideration. The
Source for sustain the design fire opening should correlate to a steady state design fire.
- The artificial vent location should be conservatively set as low as
possible without having a significant impact of the airflow
dynamics (unrealistic short circuit of replacement / inlet air). High
level vents are not desirable for assessment. See Appendix C for
more information.
Page 43 of 70
Leakage Influential leakage from - Leakage from doors and windows should be included especially
doors, windows and where mechanical powered systems (smoke extract,
other pressurisation or depressurisation) are adopted.
- Leakage from closed doors or windows can be deemed

negligible where pressure differentials are lower than 5Pa
Natural Air Natural source of - Natural inlet paths, such as shafts or ducts, should be included
inlet replacement air for assessment.
- Natural inlet air profiles should be adjusted to account for

upstream resistance e.g. pressure drop through a natural inlet
shaft may affect depressurisation levels in the corridor to unsafe
levels
- Opening size(s) representing pressure relief vents should be

closely correlated to pressure differential near the vent.
Simulating Accounting for door - During initial escape, doors opening along the escape route
occupant opening times during should be modelled open for no less than 20 seconds to
movement escape and; or fire- simulate occupants escaping through the door.
fighting access
Smoke Modelling airflow - Consideration on the viability of the smoke control system.
Extract / terminals as vents Extract rates in corridors or lobbies that are in excess of 6m3/s,
Exhaust may be more hazards during escape and fire-fighting (e.g.
Terminals slamming doors, drawing excessive amount of fire and smoke
into protected spaces etc.) and may warrant further assessment.
Furthermore, installation of systems with excessive flow rates
may not be feasible to achieve.
- Assumptions used for inlet air in the model need to be feasible in

practice e.g. final exit door at the base of the stairs being open
by automation rather than assuming FF intervention or following
escape.
Simulation Model run time - CFD runtime should run for at least 10 minutes unless steady
run-time state conditions are observed sooner. Assessment criteria
should be assessed only after balanced conditions are achieved.
Page 44 of 70
13.3 Large buildings and enclosures:
large buildings and enclosures.

Model Key elements, I-beams, - In large open domains, all objects that may influence the flow
domain(s) pillars, ceiling bulk heads and the performance of the system should be explicitly modelled,
to be included in the e.g. down-stands and beams where these divert the bulk smoke
model for accuracy and air flow
Mesh sizing Suitable mesh sizing - For large enclosures, a higher mesh density should be used
selection based on fire within 2m of the fire source and a lower density throughout the
and domain size. remaining volume of the building
Initial Ambient conditions - Extreme temperatures differentials between different interacting

boundary (temperature, density, domains should be modelled. These should be particularly
conditions etc.) assessed in tall open voids where internal stack effect is likely to
occur.
Relevant external
air/wind
Seasonal Climate
Fire source Suitable fire source in - Fire parameters should be substantiated from published design
and location terms of size, heat flux guides, available experimental data and / or empirically
and heat release rate. calculated. See Appendix C for further guidance.
appropriate fire - For sprinklered fires, it acceptable to assume that the fire will
location(s). develop until the suppression system activates and then remains
fixed thereafter. Depending on the context of what the modelling
objectives are, the cooling effect of water droplets may not need
to be accounted for certain application.
- The selection of the worst credible fire locations can be identified

through the use of quantitative and/or qualitative methods. In
atriums for example, a simple approach would be to test fires at
the base of the atrium where smoke would spill into the atrium
void from a balcony edge; and fires on the highest level that
would fill the smoke reservoir to the designed clear layer height
in the shortest period.
- It is recommended that the locations be discussed and agreed

with the relevant authorities at an early stage in the project.
Oxygen Ventilation opening to - Using artificial opening requires careful consideration. The
Source for sustain the design fire opening should correlate to a steady state design fire.
- The artificial vent location should be conservatively set as low as
possible without having a significant impact of the airflow
dynamics (unrealistic short circuit of replacement / inlet air). High
level vents are not desirable for assessment.
Page 45 of 70
Density (soot a 10% yield with heat of combustion of 19 000 to 20 000 KJ/kg
yield)
Leakage Influential leakage from - Leakage from doors and windows should be included especially
doors, windows and where mechanical powered systems are adopted.
other
- Leakage from closed doors or windows can be deemed as
negligible where pressure differentials are lower than 5Pa
-
- Natural inlet air profiles should be adjusted to account for
upstream resistance e.g. pressure drop through a natural inlet
shaft may affect depressurisation levels in the corridor to unsafe
levels
occupant opening times during should be modelled open for less than 20 seconds to simulate
movement escape and; or fire- occupants escaping through the door. However, longer times
fighting access may need to be considered to simulate queue times for
simultaneous evacuation of large buildings
Smoke Modelling airflow - Consideration on the viability of the smoke control system.
Extract / terminals as vents
Exhaust
Terminals - Assumptions used for inlet air in the model need to be feasible in
practice e.g. Entrance lobby doors used for inlet air should be
open by automation rather than left open by escaping occupants
or FF access.
Page 46 of 70
13.4 Enclosed car parks:
enclosed car parks.
Model Key elements, I-beams, - All objects that may influence the flow and the performance of
domain(s) pillars, ceiling bulk heads the system should be explicitly modelled, e.g. down-stands and
to be included in the beams where these divert the bulk air flow.
model for accuracy
Modelling of Inclusion of cars as - Cars (obstacles) are not required unless significant vortices and
cars obstructions flow obstructions are generated. See Section 12.1.7 for further
(obstacles) guidance.
Mesh sizing Suitable mesh sizing - For large enclosures, a higher mesh density should be used
selection based on fire within 2m of the fire source and a lower density throughout the
and domain size. remaining volume of the building
Initial Ambient conditions - Material properties should be sufficient to capture the relevant
boundary (temperature, density, physics of the actual areas being studied. Default values may
conditions etc.) not produce realistic results. However, inclusion of material
properties is not necessary for smoke clearance or vehicle
Relevant external
emissions (pollution) studies.
air/wind
Fire source Suitable fire source in - Fire parameters should be substantiated from published design
and location terms of size, heat flux guides such as BS7346-7, available experimental data and / or
and heat release rate. empirically calculated.
appropriate fire - The selection of the worst credible fire locations can be identified
location(s). through the use of quantitative and/or qualitative methods.
However, it is recommended that the locations be discussed and
agreed with the relevant authorities at an early stage in the
project.
Oxygen Ventilation opening to - The model domain should match the planned geometry of the
Source for sustain the design fire car park. This includes all external openings. Artificial openings
fire size (artificial or actual) should not be adopted unless substantiated.
Density (soot a 10% yield with heat of combustion between 19 000 to
yield) 20 000 KJ/kg
Leakage Influential leakage from - Leakage from closed doors or windows can be deemed as
doors, windows and negligible for car park ventilation CFD studies.
other
- Incoming velocity from natural vents and entrance ramps need to

be taken into account as velocities in excess of 5m/s could
impede escaping occupants who need to travel past the vent(s)
or opening(s).
Page 47 of 70
occupant opening times during should be modelled open for less than 20 seconds to simulate
movement escape and; or fire- occupants escaping through the door. However, longer times
fighting access may need to be considered to simulate queue times for
simultaneous evacuation of large buildings
Smoke Modelling airflow - Extract and inlets should be meshed with a minimum of 5 cells
Extract / terminals as vents along each edge on the surface of the element.
Exhaust
Terminals - Consideration on the viability of the smoke control system.
- Assumptions used for inlet air in the model need to be feasible in

practice e.g. Entrance shutters at the car park entrance should
be open or close by automation rather assuming manual
operation during FF access.
Induction / Modelling jet fans and - Jet fans should be modelled with all the elements which may
Jet Fans their respective airflow impact their flow field e.g. silencers
profiles.
Page 48 of 70
Appendix A: Physical sub models
Turbulence Models
One of the most important physical models to consider when simulating relates to the
treatment of turbulence as this will define to a large extent how heat and soot are
transported around the computational domain.
Turbulence is generated across a wide range of length scales, but in ‘fire driven flows’
turbulence is generated across length scales typically of the order of a few meters and
representative of physical lengths associated with the generation of shear layers in the
flow. As the turbulent energy decays, the turbulent eddy sizes associated with this
energy become smaller until they are small enough for the energy to be dissipated by
viscous forces. It is at these finest length scales that fuel/air mixing takes place and at
which chemical reaction occurs. Thus, fire and smoke modelling involve a large range
of spatial and temporal scales, of which only a subset of these can be simulated and
the rest must be modelled. it is important to understand that different choices are
available for the treatment of turbulence in these problems.
A common way of modelling turbulence is termed large eddy simulation (LES), where
the time-dependent flow equations are solved and the larger eddy influences are
rigorously represented. This technique is aimed at extracting greater temporal and
spatial fidelity from the simulations of a fire performed on the more finely meshed grids.
The general philosophy behind LES simulations is that the eddies that account for
most of the mixing are large enough to be calculated with reasonable accuracy from
the equations of fluid dynamics. However, below the mesh size, certainly at length
scales associated with the chemical reaction, fluctuations are either ignored or
perhaps time averaged.
Reynolds-averaged Navier-Stokes (RANS) models were developed as a time-

averaged approximation to the conservation equations of fluid dynamics. In other
words, the RANS models solve only for time-averaged eddies. The smallest resolvable
length scales are determined by the product of the local velocity and the averaging
time rather than the spatial resolution of the computational grid. Unfortunately, the
evolution and behaviour of large eddy structures characteristic of most fire and smoke
plumes is lost with such approach. Over the years, the increase in CFD capabilities
has shown a general trend of moving from RANS models to LES techniques.
Another computational approach is the Direct Numerical Simulation (DNS).

DNS is a simulation in computational fluid dynamics in which the Navier-Stokes
equations are numerically solved without any turbulence model. This means that the
whole range of spatial and temporal scales of the turbulence must be resolved in the
computational mesh. In other words, all the turbulent motions within a certain flow
down to the smallest turbulent scales are resolved. The computational cost of DNS is
Page 49 of 70
very high also for non-turbulent flows, and as a consequence the computational
resources required by a DNS would exceed the capacity of the most powerful
computers currently available. Therefore, DNS techniques are not practical to fire
related flows.
The difference among DNS, LES and RANS models is graphically shown below,
representing the different approaches that the CFD methodology can employ for fire
simulation purposes for a given variable.
Figure 11 –Schematic representation of different turbulence resolutions among DNS, LES and RANS
models
Combustion Models
There are two modelling approaches that are commonly adopted to represent the
combustion processes.
Volumetric Heat Source Model:
This model does not predict the release of heat and smoke in the flame. The
quantities of heat and smoke released by the fire together with the volume of
flame where the releases occur are specified by the user. The distributions of
heat and smoke released are assumed to be uniform over the flame volume.
The model predicts the transport of heat and smoke away from the fire.
The need to prescribe the volume of the flaming region is a limitation of this
approach.
This approach is widely used to simulate the movement of smoke in large
spaces, i.e. car parks.
Combustion Model:
Typically, these models aim to predict, in a simplified manner, the chemical
reactions that happen in the flame. While the overall quantity of heat released
and the area of the fire have still to be specified by the user, the non-uniform
Page 50 of 70
distribution of heat in the flame region is predicted and aims to take into
account the influence of the local flow.
The model employed should represent the behaviour of the flaming region and the
thermal plume adequately as this will affect the transport of smoke away from the
fire.
Buoyancy Models
The Boussinesq approximation is commonly used to model buoyancy in flows in
which the temperature gradients are small (to a maximum of a few tens of Kelvin).
The model assumes that the density is constant in most of the momentum equations,
adopting a linear proportionality with temperature in the gravitational term only.
Fire and smoke movement studies require the simulation of flows involving heat
transfer and, therefore, the fluid properties, including density, are functions of
temperature. The temperature gradients are, typically, significantly greater than
those for which the Boussinesq approximation is valid.
It is recommended that an equation of state be used to represent the buoyancy
effect.
Radiation Models
Radiative heat transfer occurs between the emitters and receivers (i.e. between solid
surfaces, the soot / gas phase mixture of flames and smoke aerosol) and is an
important feature of the heat transfer processes in combusting flows when the
temperatures are above typically 600K.
The primary sources of radiation are CO, CO2, CH4 and H2O (which emit energy in
discrete bands) and soot (which emits radiation at all wavelengths).
There are four main modelling approaches.
a) Fractional Heat Loss Model
The heat output of the fire is represented by the convective heat fraction
only.
Heat loss from the fire as a result of radiation to the surroundings is,
typically, ignored. An alternative to ignoring the radiative fraction is to
define a priori a prescribed heat distribution to regions of the model to
represent the radiative heat transfer to the region from the fire. The
uncertainties arise from the specification of the prescribed radiative heat
transfer process.
Radiative heat transfer between the smoke and the walls can be
expressed as a function of the temperature of the wall and smoke and the
emissivity of the smoke. It is applied between the wall and the fluid cell
next to the wall.
Page 51 of 70
b) Six Flux Model
This method is applicable to structured meshes using quadrilateral cells to
form the mesh. Radiation is assumed to be transmitted along the local
axes of the cell such that the radiant flux across each of the six faces of
the mesh cells is uniform. This simplifies the set of equations to be solved
to calculate the radiative source term but the accuracy is highly
directionally dependent.
c) Discrete Transfer Model
This model aims to solve the representative discrete radiative rays only.
The directions of the rays are specified a priori. The solution for any
particular ray is restricted to the path between two boundary walls rather
than being partially reflected at walls and being tracked to extinction. The
accuracy of the discrete transfer model is dependent on the ray directions
chosen as well as the number of rays.
The discrete transfer model is not ideally suited to the body fitted meshes
likely to be seen in fire and smoke movement applications in complex
spaces. This is because it is computationally expensive, especially for
situations where a large number of rays are required to obtain an accurate
solution
d) Monte Carlo Simulation Model
A number of rays are ‘emitted’ in (pseudo-) random directions and are
then traced until they hit an obstacle / wall or exit the computational
domain. The quality of the heat transfer calculations is dependent on the
number of rays. This is a costly approach that potentially offers the most
accurate solution and flexibility for complex geometries.
All but the first approach requires the calculation of local emissive powers and
absorptivity’s, which depend on the composition of the soot / gas mixture. This is a
complex process for which guidance should be sought from the developer of the
CFD simulator.
When employing a volumetric heat source combustion model, the simplest method
of accounting for radiation loss, i.e. the fractional heat loss model, is adequate. This
approach, however, only accounts for the radiative loss of the flaming region and
ignores other radiative heat transfer, in particular the transfer from hot smoke to
walls and the transfer within the smoke. Including the radiative heat transfer at the
walls may improve the simulation prediction, but the simulation will still ignore the
radiative transfer within the gas. However, employing more sophisticated
approaches to account for radiation in combination with a volumetric heat source
model is unlikely to increase the accuracy of the simulation as a direct result of the
assumption of uniform heat distribution in the flaming region.
Any radiation model can be combined with any combustion model.
Page 52 of 70
Appendix B: Quality and Compliance
The equations of fluid dynamics are discretised through the mesh division and
transformed to a set of algebraic equations which are then solved through an iterative
process to arrive to a numerical solution. The quality of the mesh (i.e. its density and
distribution of the cells) plays an important role in the reliability and accuracy of the
numerical results.
Regardless of the type of mesh used for a specific simulation, checking the size,
refinement and numerical convergence of the solution obtained by using a certain
mesh is essential. In many cases, poor resolution in critical regions can dramatically
affect results of a CFD simulation.
This section presents an overview of the factors that can help in evaluating the quality
and correctness of the numerical solution obtained.
B.1 Mesh Quality
The indicators which identify mesh quality are Orthogonality, Aspect ratio and
Skewness:
- Orthogonality:
Is a quantity computed for cells and it is in general calculated by the software
package. The worst cells will have an orthogonal quality closer to 0, with the
best cells closer to 1. The minimum orthogonal quality for all types of cells
should be more than 0.01, with an average value that is significantly higher.
- Aspect ratio:
The aspect ratio is a measure of the stretching of a cell. Generally, it is best to
avoid sudden and large changes in cell aspect ratios in areas where the flow
field exhibit large changes or strong gradients. Truncation error is the difference
between the partial derivatives in the governing equations and their discrete
approximations. Rapid changes in cell volume between adjacent cells translate
into larger truncation errors.
- Skewness:
Defined as the difference between the shape of the cell and the shape of an
equilateral cell of equivalent volume. For example, optimal quadrilateral
meshes will have vertex angles close to 90 degrees, while triangular meshes
should preferably have angles of close to 60 degrees and have all angles less
than 90 degrees. Highly skewed cells can decrease accuracy and destabilize
the solution. A general rule is that the maximum skewness for a
triangular/tetrahedral mesh in most flows should be kept below 0.95, with an
Page 53 of 70
average value that is significantly lower. A maximum value above 0.95 may
lead to convergence difficulties.
B.2 Mesh Sensitivity Analysis Through Discretisation Error study
The most recent adopted and accepted method for assessing CFD results sensitivity
to the mesh size is based on calculation of the Grid Convergence Index (GCI). This
method provide that the Index is calculated using Richardson Extrapolation (RE)
method.
The main advantages of using this method is that it is a reliable method to all branches
of flow simulations, such as fluid motion, thermal analysis, fire and smoke driven flow
and furthermore, it can be applied to the results of different CFD software outputs.
Discretization errors for the CFD results can be analysed now by using the GCI (Grid
Convergence Index) method.
The process of the GCI assessment is as follows:
- A representative mesh size (cell c) is defined, which is provided by the equation

below:
𝑁𝑁
1
𝑐𝑐 = [ �(∆𝑉𝑉𝑖𝑖)]1/3
𝑁𝑁
𝑖𝑖=1
Where:
- c is computational mesh size (cell size);
- N is the total number of cell in the mesh;
- ∆Vi is the volume of the ith cell;
Three different sets of mesh (i.e. c1, c2 and c3) in decreasing order of fineness (c1 fine
mesh size and c3 coarser) are recommended for the sensitivity analysis. For these
different mesh sizes, simulations are run and any variables of the flow field f1, f2, f3 (i.e.
temperature, or pressure, or velocity, etc.) can be used for the discretization error
analysis.
For example, f1 is the selected variable obtained for the finest mesh, while f2 is
obtained for the medium mesh and the f3 is obtained for the coarse mesh. For these
flow variables the condition f2 /f1 > 1/3 and f3/f2 > 1/3 must always be satisfied.
Page 54 of 70
The Grid Convergence Index (GCI) is calculated using the following equation:
21
1.25 𝑒𝑒𝑎𝑎21
𝐺𝐺𝐺𝐺𝐺𝐺𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 = 𝑝𝑝
𝑟𝑟21 − 1
Where:
- 𝑟𝑟21 = 𝑐𝑐1 /𝑐𝑐2 is the refinement factor (from one mesh to the other)
𝑓𝑓1 −𝑓𝑓2
- 𝑒𝑒21
𝑎𝑎 = | 𝑓𝑓1
| is a
- p is the order of the method of the discretization scheme. It can be
found in the CFD software package details (usually range from 1-4).
The calculation of GCI in relation to a certain mesh size should result to be less than
10% for the mesh to be considered reliable and the results sufficiently accurate.
B.3 Convergence check
Discretisation of fluid dynamic equations yields to a large number of algebraic

equations (one set for each cell of the computational mesh). These equations are then
generally solved using an iterative method, starting with a first guess value for all
variables and completing a computational cycle.
When assessing the numerical correctness of a simulation it is fundamental to check

convergence of the solution. This evaluation can be done by checking the residual
values (i.e. the changes in the equation over an iteration) and/or by checking that
additional iterations produce negligible changes in the variable values. Information on
the criteria are given as follows:
B.3.1 Convergence check through residuals
The residuals of the equations are the change in the equations over an iteration. The
residual is one of the most fundamental measures of an iterative solution’s
convergence, as it directly quantifies the error in the solution of the system of
equations.
In a CFD analysis, the residual measures the “local imbalance” of a conserved variable
in each cell of the mesh. Therefore, every cell in a CFD model will have its own residual
value for each of the equations being solved.
In an iterative numerical solution, the residual will never be exactly zero. However, the
lower the residual value is, the more numerically accurate the solution. Each CFD code
will have its own procedure for normalizing the solution residuals.
Page 55 of 70
- Residuals of 10-4 are considered to be loosely converged,
- Residuals of 10-5 are considered to be well converged,
- Residuals of 10-6 are considered to be tightly converged.
It must be noticed that for complicated problems, however, it’s not always possible to
achieve well or tightly convergence.
B.3.2 Convergence identification through flow results
Every CFD simulation has the objective of determining some quantity such as
temperature, velocity, pressure, etc., of a certain flow field. In order to verify
convergence, it is possible to track the values of such variables with respect to iteration
and define iterative convergence when these quantities converge (i.e. remains
unchanged in the following iterations). Convergence can be defined when a monitored
flow value remains unchanged with respect to the number of iterations. The
convergence criteria is often defined by the acceptable error in these values.
Depending on the flow field simulated it can occur that certain quantities may reach
convergence at a different rate than other quantities.
B.4 Frequency for Mesh Sensitivity Testing
Mesh density is one of the significant metrics to control accuracy of a CFD model,
assuming that all the input parameters used are accurate. One of the ways to evaluate
quality of the mesh is by comparing its consequent results to actual test data, empirical
calculations or theoretical values. Otherwise refining mesh sizes and interpreting result
deviations would provide alternative methods. The problem with multiple remeshing
and rerunning the models, is that they can be time consuming, especially for complex
models.
Carrying out a mesh sensitivity testing is not always necessary, especially where well-
established industry norms exist. Such testing should be done in the context of the
results and other necessary sanity checks (mass flow rates, heat release rates etc).
Page 56 of 70
Appendix C: Fire and Smoke Source
C.1 Introduction
Specification of the smoke source is a key element in the application of CFD to
smoke control design and analysis. This section outlines various approaches
available and considers where each may be appropriate and the limitations that may
apply. The content is informative only; for detailed guidance the reader should
consult authoritative texts such as ‘Computational Fluid Dynamics in Fire
Engineering’ by Yeoh and Yuen (published by Butterworth-Heinemann 2009) or the
SFPE Handbook of Fire Protection Engineering (5th edition published by Springer
2016).
C.2 Pyrolysis and Combustion
In general, a fire involving flaming combustion includes the following processes:

a) heat transfer to the fuel surface, initially by the source of ignition and
subsequently from the flames;
b) heating of the fuel;
c) pyrolysis (release of combustible gases);
d) mixing of the pyrolysed gases and oxygen (in the air);
e) combustion (oxidation) of the pyrolysed gases, generating products of
combustion;
f) further mixing with air by entrainment and/or forced convection.
The fire may continue as a localised one with a defined smoke plume or may
progress to a fully developed one with an extended combustion zone possibly
involving the entire room or compartment.
For the fire to continue, the ‘fire triangle’ of fuel, heat and oxygen needs to be
maintained. The fire will decay or be extinguished if the fuel ‘runs out’ (items burn
out), the oxygen supply ceases (under-ventilation or suppression by water-mist etc)
or the heating process terminates (by suppression or limited combustibility of the
material).
C.3 Smoke and Visibility

Smoke comprises the products of combustion, any unburned fuel and the air
entrained into the fire plume. It is entrained air that, in general, constitutes the main
component of smoke as measured by volume; this means that away from the
combustion zone smoke can be considered as contaminated air from a fluid
dynamics perspective. For most fires, the primary products of combustion will include
Page 57 of 70
carbon dioxide and water vapour. Secondary products of combustion may include
soot particles and other gases such as carbon monoxide and hydrogen cyanide.
For smoke control design and analysis, soot is often the most relevant product of
combustion. The concentration of soot can provide an indication of the likely visibility
through the smoke. Reduction of visibility through smoke is a complex process,
caused by scattering of light by the aerosol of solid particles and liquid droplets in the
smoke, augmented by irritancy effects. Visibility is a vector quantity, meaning it
varies in each direction, i.e. line-of-sight. Calculation of light attenuation along a
given line-of-sight needs to account for variation in concentration of smoke aerosol
particles and droplets; however, this is beyond the scope of general fire engineering
analyses. A simplified, and widely adopted, approach is to gauge the reduction in
visibility from the concentration of soot. A ‘visibility scalar’ S [m] is defined using the
following empirical correlation [ref: SFPE Handbook of Fire Protection Engineering
5th Ed. (2016) – Chapters 33 and 61], where ρsoot is the local soot particulate density
(concentration) [kg m-3], Km [m2 kg-1], is the specific light extinction coefficient and C
is a constant (set to 3 for viewing light-reflecting elements of construction).
C
S= (1)
K m ρ soot
An appropriate value of Km for mixed burning of cellulose and plastic materials is

8700 m2kg-1 [ref: Mulholland and Croarkin (2000) Specific Extinction Coefficient of
Flame Generated Smoke, Fire and Materials, vol. 24, pp. 227–230]. Although S is
not a distance vector along the line-of-sight, it does provide an indication of the
visibility at each location if it is assumed the soot concentration remains constant
throughout space and visibility is isotropic. It might also be used as a qualitative
indicator of exposure to irritant products of combustion.
C.4 Modelling smoke in isolation

In some cases, it may be appropriate to include only air and smoke transport in the
CFD simulation without recourse to the fire source itself. Examples include smoke
clearance from a protected corridor or a basement car park, where the space is
exposed to smoke prior to the operation of the ventilation system. Two approaches
that might be considered are:
Purging an initial distribution of smoke
The space to be cleared (purged) is initially filled, or part-filled, with smoke or

contaminated air at a specified composition of species and temperature, e.g. warm
air containing a suspension of soot or other products of combustion. The CFD
model then calculates the purging of smoke or fume by air, allowing the time for the
Page 58 of 70
space to be returned various levels of ‘smoke-free’ or tenable conditions to be
determined.
The modeller will need to establish, by reasoning and calculation as appropriate, the
initial temperature and composition of smoke. A smoke-logged space might be
considered as composed of gases with the same composition as a representative
fire plume. Dilution of the initial smoke concentration to 1% may then be an
appropriate indicator that tenable conditions are achieved [ref: Tamura (1994)
Smoke Movement and Control in High-rise Buildings. NFPA, Dec 1994]. A lower
level of dilution may be appropriate in other scenarios.
In some applications, e.g. post-fire smoke clearance in an enclosed car park, it may
be appropriate to treat the initial smoke as ‘contaminated air’ at ambient or an
elevated temperature; the progressive purging of the initial atmosphere with outside
air could be examined to establish whether a required air change rate is achieved
throughout the space.
The concept of residence time and mean age of air may be useful. This is an output
available in some CFD models and provides a measure of how long a fluid element
remains within a space. This can be used to identify whether a purge ventilation
system works effectively, and there are no stagnation regions.
It may be necessary to consider whether a fire source (discussed below) is required

rather than assuming an initial smoke distribution, e.g. to account for the impact of
the thermal dynamics of a car fire on an escape lobby.
Smoke source specified as a boundary condition
It may be appropriate in some applications to specify the source of smoke and heat
on the boundary of the computational space, e.g. at an apartment doorway on a
common corridor. The modeller will need to define the flow rate, temperature and
composition of smoke at the boundary. This approach may lend itself, for example,
to the analysis of smoke clearance in an escape route following a limited time of
exposure, or to the modelling of the external flow of smoke from an open window or
shaft. Figure 12 illustrates an example of an open doorway boundary condition.
Page 59 of 70
Source of smoke and air at specified
temperature
Mass sink
Apartment side (outside CFD domain) Corridor side (inside CFD domain)
Figure 22 –Example of a doorway boundary condition
C.5 Modelling the fire source and smoke transport

Where the generation of heat and combustion products is required as part of the
simulation, it is necessary to incorporate one or both fuel pyrolysis (material burning)
and gas phase combustion. For example, the generation of heat and combustion
products will be an integral part of the analysis of a smoke and heat exhaust
ventilation system (SHEVS) for an atrium building or large space such as a
warehouse.
Figure 33 –Section showing temperature from a pallet fire in a warehouse
Page 60 of 70
There are various approaches available, with varying levels of complexity.
Volume (3 dimensional) heat source
In scenarios where the extent of the combustion (flaming) zone can reliably be
specified a-priori, then a volumetric source of heat may provide a sufficiently
accurate representation of the fire source. Here the fire is prescribed simply as a
source of heat; and if the calculation of smoke toxicity or visibility is required the
source terms for relevant products of combustion are required too. The modeller will
need to define the size and shape of the volume source, referring as necessary to
correlations for fire size and flame shape.
There may be limitations to this approach, and cases where it is inappropriate; for
example, where the shape of the combustion region is unknown (e.g. a leaning flame
due to asymmetric airflow), or where the fire is or under-ventilated.
Single-step combustion with prescribed pyrolysis (fuel release) rate
A relatively simple approach to combustion modelling in CFD is provided by the

Eddy Dissipation Concept [ref: Magnussen and Hjertager, On Mathematical
Modelling of Turbulent Combustion with Special Emphasis on Soot Formation and
Combustion, Proc. 17thSymposium (International) on Combustion (1977), pp. 719–
729.] whereby the rate of fuel consumption (combustion) is proportional to the
concentration of reactants and the local rate of mixing with oxygen. It provides the
foundation for the single step, mixing controlled combustion model available in
various commonly used CFD models.
Stoichiometric mixing of air and fuel and infinite rate kinetics are assumed. The
reaction takes the general form below, where the products comprises CO2, H2O and
inert N2.
Fuel + air → products (+ heat) (2)
The heat released by combustion drives the smoke transport and is a source of
thermal radiation. Heat release can be expressed in a simplified form as below,
where m is the fuel supply (pyrolysis) rate [kg s-1], q is the rate of heat release [kW]
and ΔHc is the effective heat of combustion [J kg-1].
q= m∆H c (3)
The modeller needs to define both m and ΔHc. Alternatively, they may opt to specify
the ‘required’ rate of heat release a priori, in which case m becomes, in effect, an
output parameter. In either case, the limiting assumption is that the fire size (burning
rate) is specified as an input to the model.
Page 61 of 70
If the transport of secondary products of combustion such as soot is to be modelled,
then an empirical fractional yield is required. For example, a 10% soot yield may be
specified where visibility through smoke is to be assessed. Note importantly that
where visibility is a key output, it is the combination of heat of combustion and soot
yield that determines the amount of soot (and hence reduction in visibility) for a given
heat release rate.
Advanced pyrolysis and combustion models
If a more detailed representation of the chemical reaction between fuel and air is
required, then a more advanced combustion modelling approach will be required.
This could be the case where effect of under-ventilated combustion is significant,
and the modelling of CO production, for example, is important.
The reader should refer to specialist publications for more information on advanced
pyrolysis and combustion modelling.
Note that advanced combustion modelling will not, in most cases, be necessary for
CFD modelling for smoke control design and analysis.
Suppression
In the simplest approach, suppression might be represented by restricting the heat

release rate such that the temperatures within the room or compartment are
commensurate with a sprinkler suppressed fire, e.g. the upper layer temperature is
circa 120°C (or the value considered reasonable for the scenario being modelled). In
effect, this approach involves calibrating the fire heat release to achieve the required
conditions. A more conservative approach might be to model a fuel bed fire with a
pre-defined heat release rate, e.g.1000kW for a residential fire or 2500 kW for a
retail fire. This will in general result in conditions more severe than would be
generally anticipated with an operating sprinkler system but can provide a higher
level of robustness as it allows for the sprinklers only part working.
A higher level of sophistication would be to include a sprinkler water particle model

for example, where the interaction of the water droplets with the hot gases is
incorporated and the cooling of the hot gases captured. This does not incorporate
effect on fire size/fuel release, however.
At an even more advanced level, the interaction of the suppression system with the
pyrolysis and combustion processes, could be modelled. This, however, remains a
research topic and is generally outside the scope of a CFD modelling for smoke
control design and analysis.
Page 62 of 70
Fire size and ventilation openings
For many smoke control applications, a localised fire source will be most
appropriate. The area of the fire source will be constrained, and flashover or fully
developed fire conditions will not occur.
In general, the fuel release (pyrolysis) for a pre-flashover fire will be defined either
using experimentally derived data, or by adopting a generic, ‘design fire’. In the case
of a steady fire, this is usually specified in terms of the surface area and the heat
release associated with the combustion of the fuel.
For time-dependent fires it is common practice for many applications to adopt a so-
called t-squared source (ref: ISO/TR 13387 Fire safety engineering – Part 2:1999
Design fire scenarios and design fires):
2
t 
q = 1000  
t
 g 
Here, q (kW) is the heat release rate associated with the fire, t (s) the time since the
start of the growing stage of the fire and tg the ‘characteristic time’ for q to reach
1000 kW. Fires are typically classified as slow, medium, fast or ultra-fast as indicated
below:
Growth rate Characteristic time tg (s)

Slow 600
Medium 300
Fast 150
Ultra-fast 75
The availability of air for combustion needs to be carefully considered when

modelling smoke control systems. If the fire is not to be allowed to become under-
ventilated, there will need to be sufficient opening(s) to provide the air required for
ventilated combustion. Examples of ventilation openings for a compartment include
open doors, windows, mechanical ventilation and louvres. Or, combustion
ventilation might be formed when glazing fails. Where there is a possibility that
compartment ventilation may not be readily available ( e.g. basements, specialist fire
glazing or specialist glazing protection sprinklers etc) then a sensitivity should be
undertaken with some form of generic ventilation allowing a physically realistic fire to
develop and a fire where the only ventilation is the open fire compartment door and
smoke extract system running. However, in some cases, such as tall buildings, this
may not be practical.
Page 63 of 70
A common approach is to incorporate an ‘artificial’ low-level vent to support the fire
with doors open and smoke control system running. This may be a good starting point
as the approach is often very conservative however, it can lead to impractical airflow
rate requirements. The low-level vent is not physically realistic but can give an upper
bound on the smoke control systems airflow rates needed. Careful attention to the
available ventilation in tall buildings is very important. A low-level ventilation strategy
supported by mechanical ventilation may be optimistic. Here, the extract system
located in the corridor/lobby, protects the staircase by drawing fresh air from the
staircase enclosure. The extract system in tall buildings will prefer to draw smoke from
the fire compartment as opposed to the staircase, which is made worse with large
openings, like doors and windows that will allow more hot smoke to be drawn from the
fire compartment. The low-level vent is considered a specialist ventilation strategy
and not a generic fire ventilation strategy.
A sensitivity study of the effects of the ventilation opening in the room of fire origin is
recommended under the context of the CFD objectives being studied to ensure an
element of realism in the results is balanced with a margin of safety.
Fully developed and post-flashover fires
In the case of flashover and post-flashover fires, a user-defined fuel release rate
(e.g. as in a design, t-squared fire) will not generally be appropriate. The pyrolysis
process is more complex than for a localised fire. The availability of air for
combustion is especially important, influencing whether the fire is fuel- or ventilation-
controlled.
There are publications and correlations available to guide the modeller on the size of
fully developed fire that can be expected in a room with given openings. For
example, e.g. see Karlsson & Quintiere, Enclosure Fire Dynamics, the maximum
heat release rate q (MW) within a room with a single opening with area A (m) and
height H (m) is given approximately as:
q = 1.5 A H
Boundary heat transfer to walls, ceilings etc. is also an important consideration post-
flashover. As the fire continues, the boundaries heat up and temperatures inside the
room or compartment increase, as is reflected in the time-temperature curves
employed in fire-resistance tests.
It is not possible to define, a-priori, a fire size or heat release rate for fires that reach
post-flashover. The fire size, and room conditions, depends on the distribution and
Page 64 of 70
nature of combustible items, the availability of openings (vents) and the properties of
the walls and ceiling.
Generally, modelling post-flashover or fully developed fires will be beyond the remit
of the smoke control analysis. When it is required, a greater level of understanding of
the fire physics and fluid dynamics is required compared to the localised fires
discussed above.
Page 65 of 70
Appendix D: Testing and Validation
Before using a CFD software product, particularly for fire modelling, it is incumbent
on the user to establish the suitability of that package. The issue is whether the
software has gone through a testing or validation process that shows that it is
suitable for the demands being placed upon it.
To give confidence in the software there ought to be information in the literature or

product documentation to support its use. Otherwise, the developers should be
contacted to establish a first view of applicability. This should be followed up by
tests, related to the proposed use, to compare simulation results with some expected
outcome.
If documented evidence exists that the software has been used successfully, such
evidence can be valuable in helping set user-controlled parameters to maximise the
effectiveness of the modelling.
It should be pointed out that validation is a continuing process, and that in the
present context “testing” is a more appropriate word.
The detailed capability-requirements for representing the physical processes thought

important in fire modelling are discussed elsewhere in this document. But in a real-
world application the question is whether the software is able to model what is
required to a reasonable degree of accuracy. In this context a reasonable degree of
accuracy might be +/- 20%. However, a first test is that of plausibility – do the results
look right? Results ought to confirm the expectation of experienced engineers. If not,
then more questions should be asked. If new or unexpected features are exposed,
then these features need to be understood in the light of established knowledge.
Although the emphasis is on application to fire modelling, there are often important
requirements for general day-to-day ventilation for dispersal of pollutants and also for
post-fire smoke clearance performance. The comments above also relate here.
A further requirement under the broad heading of validation relates to the user of the
software. Does the user have sufficient knowledge and experience? It is well-known
that to get the best from CFD requires both an understanding of the physical
processes of thermodynamics and fluid mechanics and the careful exercise of
engineering judgement. Whilst the former can be acquired in a university
environment the latter requires a broader exposure to the design process. The ability
to interpret practical engineering issues in the context of the sophisticated numerics
implicit in a CFD package is important. If the user does not have the comprehensive
experience necessary, then the advice of colleagues with a complementary
experience is vital.
Page 66 of 70
Appendix E: CFD review process
Introduction
The review process provides an essential independent check of a design, ensuring

both the analysis and documentation meet the agreed performance criteria. CFD
analysis and performance-based fire design require specialist engineering rigor,
particularly as the discipline and the adoption of alternative and innovative design
proposals increase.
The process should be remote from the immediate design team and provide a fresh
view – assessing the design intent, objectives, assumptions and acceptance criteria,
as well as verifying that the input data and results reflect the documented case.
It is essential that the whole team value the review process and appreciate the wider
objectives involved in it. A reviewer’s scope should be technical – to confirm input
and outputs for example; and non-technical such as providing a sense check on the
design. It is now becoming common practice that the reviewer assists in the
approval process, and in these cases the appointed reviewer is determining their
support for the design.
Scope
Where a third party and external reviewer is involved, they should be appointed
based on agreed scope. This may be a conceptual review of an initial design
through to numeric checking of input data and confirming documented results
represent the analysis provided.
It can include suitability of documentation to record the design process and

conclusions made. It is generally recommended in the reviewer’s scope to evaluate
the qualification and experience to the design team, rather focus on the technical
aspects presented. The scope of the review can include:
• Suitability of relevant codes and guidance;

• The design objectives;
• Assumptions regarding performance criteria, fire scenarios, material
properties etc.
• The technical approach documented;
• Appropriateness of models and methods used;
• Input data and the suitability of models and methods used. This also includes
error checking of relevant input data;
• The results presented in documentation and within analysis; and the suitability
of conclusions made.
Page 67 of 70
The review allows the relevant stakeholder (client, approver/ insurer etc) to make
informed decisions regarding a design, with a view that the design will ultimately be
agreeable by all relevant parties. It is essential the terms of reference are clear
between all parties, in particular the scope and responsibility of the Reviewer. There
is also an expectation on the wider team to understand and respond to the Review.
Competency
The competency of the review should reflect the scope required. Experience and
judgement are essential, as well as qualifications and expertise in the area the CFD
being modelled. CFD is an extensive discipline and it essential the reviewer has
competencies in the fire engineering field and the application of the designs being
assessed. This is discussed further in Section 4.
Process
The objective of the review is that the reviewer agrees with the design proposal, from
initiation through to completion. This often means the reviewer should be appointed
early in the design process. Disagreement of fundamental design parameters
through late consultation with a reviewer should be avoided.
The reviewers’ scope should be clearly defined, and they should be provided with all
relevant documentation and analysis to conduct their review. The delivery of the
review would generally be a clear itemised list of queries and clarifications
sought. The reviewer should substantiate their comments through reference to
relevant published information.
The designer should address the comments made. Where there is a technical
disagreement the designer should clearly and unambiguously respond to the
comment, with technical assessment to support the case. It is often not suitable to
provide narrative responses to technical disagreements – for example if
temperatures exceed an agreed limit suggesting they may be lower based on other
assumptions and fire scenarios would need to be analysed and presented.
The designer should expect to revaluate their design and provide reassessments as
necessary such that agreement on the design and its documentation can be
provided.
Page 68 of 70
Appendix F: Bibliography & Additional Reading
1 BS 7974:2019, Application of fire safety engineering principles to the design of

buildings – Code of practice.
2 BS 7346-4:2003, Components for smoke and heat control systems. Functional
recommendations and calculations methods for smoke and heat exhaust ventilation
systems, employing steady-state design fires – Code of practice.
3 BS 7346-5:2003, Components for smoke and heat control systems. Functional
recommendations and calculations methods for smoke and heat exhaust ventilation
systems, employing time-dependant design fires – Code of practice
4 BS 7346-7:2006, Components for smoke and heat control systems – Code of
practice on functional recommendations and calculation methods for smoke and heat
control systems for covered car parks.
5 BRE 368, Design methodologies for smoke and heat exhaust ventilation
6 BRE FB29, Design fires for use in fire safety engineering, Christopher Mayfield and
Danny Hopkin
7 SFPE Handbook of Fire Protection Engineering. 5th Edition, Society of Fire Protection
Engineers
8 CIBSE Guide E: Fire Engineering Handbook. 2019
9 BS EN 12101-11, Smoke and heat controls systems - Part 11: Horizontal flow
powered ventilation systems for enclosed car parks
10 Fire Safety – Approved Document B: Volume 1 and Volume 2, HM Government
11 BS9991 - Fire safety in the design, management and use of residential buildings –
Code of practice
12 BS9999 – Fire safety in the design, management and use of
11 Smoke Control Association, Guidance on smoke control to common escape routes in
apartment buildings (Flats & Maisonettes) – Rev 3.1 (July 2020)
12 Smoke Control Association, Design of Smoke Ventilation systems for loading bays &
coach parks – A guide for system designers
11 Gupta A. K, Kumar R, Yadav P. K, Naveen M. Fire Safety Through Mathematical
Modelling. Current Science, Vol 80, No 1, 10 January 2001
12 Grandison A. J, Barnett J. R, Patel M. K. Fire Modelling Standards/Benchmarks. Fire
Safety Engineering Group, University of Greenwich
13 Ferziger J. H, Peric M. Computational Methods for Fluid Dynamics. 3rd Edition. ISBN
3-540-42074-6
14 Versteeg H. K, Malalasekera W. An Introduction to Computational Fluid Dynamics.
The Finite Volume Method.
15 Guidance for HSE inspectors, Smoke movement in complex enclosed spaces –
Assessment of Computational Fluid Dynamics
16 Fire Dynamics Simulator User Guide, Kevin McGrattan et al; www.nist.gov/fds
Page 69 of 70
17 The Use of CFD Computer Models for Fire Safety Design in Buildings: Large
Warehouse Case Study; Brian Hume and Mick Eady
18 Development of Standards for Field Models; Fire Research Report number 85,
November 2003, Brian Hume, ODPM,
19 Assessment of Vehicle Fires in New Zealand Parking Buildings, Yuguang Li. Fire
Engineering Research Report 04/02, May 2004
20 Smoke Reservoirs – an evaluation of CFD modelling as a design tool; Jacob
Hagman, Fredrik Magnusson. Department of Fire Safety Engineering, Lund
University, Sweden, Brandteknik, Lunds tekniska högskola, Lunds universitet, Report
5130, Lund 2004
21 A Computer Model of Fire Spread from Engine to Passenger Compartments in Post-
Collision Vehicles; James A. Ierardi. May 1999
22 On the dynamics of vehicles and electrical equipment; Johan Mangs, VTT Building
and Transport. Academic Dissertation. University of Helsinki.
23 Natural Fire Modelling of Large Spaces; E S Korhonen, 21/08/2000. Academic
Dissertation, Helsinki University of Technology.
24 US Vehicle Fire Patterns and Trends; Marty Ahrens, Fire Analysis and Research
Division, National Fire Protection Association, August 2005. NFPA
25 Yeoh and Yuen, Computational Fluid Dynamics in Fire Engineering – Theory,
Modelling and Practice. Elsevier, 2009
26 Merci and Beji, Fluid Mechanics Aspects of Fire and Smoke Dynamics in Enclosures.
CRC Press, 2016
27 Andrew Basford and Michael Spearpoint (2021) Defining a velocity criterion for
stagnant air in buildings. The journal of the Institution of Fire Engineers (36), 32-37.
Page 70 of 70

SCA Guidance-June 21

Uploaded by

Copyright:

Available Formats

SCA Guidance-June 21

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SCA Guidance-June 21

Uploaded by

Copyright:

Available Formats

Smoke Control Association

SCA guidance on CFD analysis for

Edition 1.2: June 2021

Marios Alexandrou Buro Happold

3 Terms and definitions .................................................................................................................. 7

4 Modeller Experience and Qualifications ................................................................................... 9

5 Qualitative Design Review .......................................................................................................... 9

7 Modelling process ...................................................................................................................... 13

8 Preparing the CFD Model ......................................................................................................... 14

9 Presentation of Analysis and Results...................................................................................... 24

10 Modelling Lobbies, Corridors and Stairs ................................................................................. 27

11 Modelling Large Enclosure Buildings ...................................................................................... 28

12 Modelling Car Parks................................................................................................................... 32

13 Modelling guidance for designing smoke control system..................................................... 41

Appendix A: Physical sub models .............................................................................................. 49

Appendix B: Quality and Compliance ........................................................................................ 53

Appendix C: Fire and Smoke Source ......................................................................................... 57

Appendix D: Testing and Validation ........................................................................................... 66

Appendix E: CFD review process ............................................................................................... 67

Appendix F: Bibliography & Additional Reading ...................................................................... 69

Disclaimer - Guidance from FETA

The increased emergence of performance-based regulations and hence solutions

3.3 Boundary conditions

3.5 Comparative Analysis

3.6 Deterministic Analysis

3.8 Developing Fire

3.10 Mesh (Grid)

3.11 Output Slice

3.13 Sensitivity Analysis

3.14 Steady State Fire

3.15 Steady State Model

3.16 Transient model (Also known as a time dependant model)

3.17 Vector slice

5 Qualitative Design Review

QDR team major stakeholders may include:

• Design Fire Engineer/ CFD Modeller

Items for consideration:

• Review the architectural design. There may be architectural features or floor

• Establish the fire safety objectives. This may be connected to the

• Establish trial fire safety designs

• Identify acceptance criteria and methods of analysis. Acceptance criteria

• Establish Fire Scenarios for analysis.

• Establish CFD review process

Further information on the QDR process is available in BS 7974.

In addition, sufficient knowledge concerning airflow and heat transfer mechanisms is

2) Solving equations on a computer invariably introduces numerical errors such

3) Boundary conditions, as with physical models, can only be as accurate as the

5) Experience in conducting CFD studies should make one aware of the

6) Numerical method is all about mathematical knowledge. Partial differential

7) Computational grid (Meshing) considers the numerical approach based on

To describe a physical phenomenon into the computational domain, it is recognised

Stage 1 – Specification outline (QDR)

Stage 2 – CFD modelling

Stage 3 – Report & presentation of results

The primary considerations are itemised below.

8.2 Details Represented in the Computational Domain

Typically, this should include some or all of the following items.

Figure 1: An example of the distribution of cars within a car park model

8.3 Computation Mesh

Further discussion regarding mesh quality is provided in Appendix B.