_

Research Report
AP-R703-24

Methodology for Comparing Dangerous Goods

Risks in Road Tunnels to Those on Alternative
Routes: Best Practice Review
Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes:
Best Practice Review

Prepared by Publisher

Bernhard Kohl, Guido Zaunrith, Conrad Stacey, Arnold Dix Austroads Ltd.
Level 9, 570 George Street
Sydney NSW 2000 Australia
Project Manager Phone: +61 2 8265 3300
austroads@[Link]
Nigel Casey
[Link]

Abstract About Austroads

This report is one of the three reports produced as part of a project Austroads is the peak organisation of Australasian road
undertaken to establish a comparative risk assessment method for transport and traffic agencies.
assessing whether a tunnel route is safer than a surface route for the
road transit of dangerous goods (DG) materials. Austroads’ purpose is to support our member organisations to
deliver an improved Australasian road transport network. To
This report reviews best international practice in road tunnel DG risk succeed in this task, we undertake leading-edge road and
assessment and provides justification for the selection of a transport research which underpins our input to policy
Dangerous Goods Quantitative Risk Assessment Model (DG-QRAM) development and published guidance on the design,
as the tool most appropriate for comparative DG risk assessments. construction and management of the road network and its
associated infrastructure.
The second report AP-R704-24 Methodology for Comparing
Dangerous Goods Risks in Road Tunnels to Those on Alternative Austroads provides a collective approach that delivers value
Routes: Manual for Using DG-QRAM in the Australasian Context for money, encourages shared knowledge and drives
provides background on the development, capabilities and limitations consistency for road users.
of the DG-QRAM tool as well as guidance as to how the DG-QRAM
tool should be applied to road tunnels in Australia and New Zealand. Austroads is governed by a Board consisting of senior
executive representatives from each of its eleven member
The third report AP-R705-24 Methodology for Comparing Dangerous organisations:
Goods Risks in Road Tunnels to Those on Alternative Routes: Case • Transport for NSW
Study Using DG-QRAM in the Australasian Context illustrates the
stages of the comparative assessment, with incorporation of • Department of Transport and Planning Victoria
mitigation measures. The case study is a demonstration of the intent • Queensland Department of Transport and Main Roads
of the method that is detailed in the manual. • Main Roads Western Australia

Keywords • Department for Infrastructure and Transport South Australia

• Department of State Growth Tasmania
Road tunnel, dangerous goods, comparative risk, hazardous
material. • Department of Infrastructure, Planning and Logistics
Northern Territory
• Transport Canberra and City Services Directorate,
ISBN 978-1-922994-20-2 Australian Capital Territory
Austroads Project No. ART6122 • Department of Infrastructure, Transport, Regional
Development, Communications and the Arts
Austroads Publication No. AP-R703-24
• Australian Local Government Association
Publication date January 2024
• NZ Transport Agency Waka Kotahi.
Pages 52

© Austroads 2024 | This work is copyright. Apart from any use as permitted under the Copyright Act 1968, no part may be reproduced by
any process without the prior written permission of Austroads.

Acknowledgements
The technical review and guidance of Austroads’ Project Manager Nigel Casey is gratefully acknowledged.

This report has been prepared for Austroads as part of its work to promote improved Australian and New Zealand transport outcomes by
providing expert technical input on road and road transport issues.
Individual road agencies will determine their response to this report following consideration of their legislative or administrative
arrangements, available funding, as well as local circumstances and priorities.
Austroads believes this publication to be correct at the time of printing and does not accept responsibility for any consequences arising from
the use of information herein. Readers should rely on their own skill and judgement to apply information to particular issues.
 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Summary
This report is one of the three reports produced as part of a project undertaken to establish a comparative
risk assessment method for assessing whether a tunnel route is safer than a surface route for the road
transit of dangerous goods (DG) materials. This method will inform decisions on dangerous goods routes
with respect to tunnel assets.

This report should be read together with the other two reports produced as part the project. Any application
of this DG risk assessment method must ensure that the guidance given in all three reports is referenced and
applied. The three reports are:
1. AP-R703-24 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on
Alternative Routes: Best Practice Review (this document).
The report reviews best international practice in road tunnel DG risk assessment and provides justification
for the selection of a Dangerous Goods Quantitative Risk Assessment Model (DG-QRAM) as the tool
most appropriate for comparative DG risk assessments.
2. AP-R704-24 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on
Alternative Routes: Manual for Using DG-QRAM in the Australasian Context (Austroads 2024a).
The report provides background on the development, capabilities and limitations of the DG-QRAM tool as
well as guidance as to how the DG-QRAM tool should be applied to road tunnels in Australia and New
Zealand.
3. AP-R705-24 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on
Alternative Routes: Case Study Using DG-QRAM in the Australasian Context (Austroads 2024b).
The case study illustrates the stages of the comparative assessment, with incorporation of mitigation
measures. It is intended that transport authorities, competent authorities and risk specialists will find the
case study a useful demonstration of the intent of the method that is detailed in the manual.

A review of international practice on risk assessment of dangerous goods in tunnels has produced some
valuable insights to be considered when developing an Australasian approach to risk-based decisions. The
area of risk assessment related to road tunnels is most developed in Europe and so this report draws mostly
on experience from there.

The comparative risk assessment will give guidance as to the route of lower comparative risk regardless of
how small those risks are. An alternative concept that is applied in Austria, France and Germany is to accept
the tunnel route, if the risks from the route fall below an acceptable expected societal risk, which is
sometimes expressed in an F-N curve. The Australasian focus on reducing risks so far as is reasonably
practicable (SFAIRP), suggests that we eschew the ‘acceptability of low risk’ approach, to choose only the
outcomes which give the reasonably practical minimisation of risk. Subject to their local legal context, road
authorities seem able to choose either approach, noting also that the distinction only applies to how the
quantitative risk assessment (QRA) model output is used.

The wider decision required in a whole of government view also includes considerations that are not directly
comparable with risks to life or to the asset, and so neither SFAIRP nor risk acceptability can be taken all the
way to a final decision.

The Austrian approach to risk comparison requires the surface route risk to be “significantly” lower than the
tunnel risk before a particular dangerous cargo must be diverted away from the tunnel. A French (CETU)
paper quantifies “significant”, suggesting that at low risk levels, a factor of 10 between two alternatives is
significant, and below a factor of three, a disparity in risk is not significant. Besides the significance of
differences between two very low risks, there are also limits on the accuracy of QRA answers, not least
because of the difficulty in knowing the inputs, such that fine distinctions are insufficient upon which to base
a judgement.

Austroads 2024 | page i

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Dangerous goods risk assessments of this type are rarely undertaken in Australia, and when they are, the
documentation can be tightly held as it is deemed sensitive. The one available Australasian study which took
the quantification to the required level for decision seemed unsure whether the decision acceptance criteria
should be the establishment of the route of lower risk, or establishing whether all assessed routes achieved a
safety level better than a certain threshold. The answer is that no such conclusion is required, with the
comparative risk assessment outputs being presented as they are, to be considered alongside community,
environmental and economic issues.

On the detail of the comparative risk assessment, PIARC (World Road Association)’s DG-QRAM is selected
as the most appropriate quantitative tool to use within an Australasian approach, with the suggestion that a
screening assessment, as is common in Europe, be applied first. The purpose of a screening assessment
would be to shorten the assessment where the comparative risk result is so clear that a simple approach is
sufficient, and the time and expense of a more detailed approach is unnecessary. If such an assessment is
considered, it may use the same methodology, but with approximated or notional input data, appropriate to
the screening function and shorter time availability.

The application to a trial tunnel should demonstrate the suitability of DG-QRAM, and indicate any
modifications required for Australasia. The onerous requirements for sufficient data on traffic makeup, DG
movements and population density to make a detailed assessment may lead to increased importance of a
screening stage and may also require that appropriate Australasian default values be considered for some
inputs.

As noted above, the wider context of transport route decisions may determine the question of how to judge
or act on a quantitative risk assessment. The quantitative methods that are a primary focus of this document
are not an end in themselves, but a decision aid within a decision that is often more complex and judgement
based. Besides the risk to life from the QRA, a complete decision on routing might need to consider
environmental risks (including to water supply catchments), community and social benefits, economic
benefits, and economic risks. There is no magic formula for converting all risks to one measure (e.g. dollars,
lives, or ecological cleanliness) and so judgement will still be necessary. Because the big picture cannot be
entirely quantitative, the risks of both routes may be considered within the wider decision, and so perhaps it
matters less whether, within the risk model, we compare risks to a benchmark, or prefer only the lowest,
following a SFAIRP approach. That is: the outputs may be considered within the holistic decision without
necessarily being placed into either framework.

In the case study (Austroads 2024b), a hypothetical tunnel is used to demonstrate the intended application of
the proposed approach to the comparative assessment of risk to life from dangerous goods being routed
through the tunnel or via an alternative surface route. It follows the method outlined in the manual
(Austroads 2024a). The manual documents the established methodology, all in the context of road transport
and tunnel management practice in Australasia.

Austroads 2024 | page ii

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Contents
Summary ......................................................................................................................................................... i
1. Introduction ............................................................................................................................................ 1
2. Limitations of Quantitative Risk Assessment ..................................................................................... 4
3. Responsibilities of the Road Authority ............................................................................................... 6
3.1 Overview........................................................................................................................................... 6
3.2 Negligence ..................................................................................................................................... 10
3.3 Dangerous Goods – Placarded Loads ........................................................................................... 12
3.4 Conclusion ...................................................................................................................................... 13
4. Agreement Concerning the International Carriage of Dangerous Goods by Road ...................... 15
4.1 Classification of Dangerous Goods According to the ADR ............................................................ 15
4.2 Tunnel Categories (ADR Chap 1.9 p.85-87) .................................................................................. 16
4.2.1 Tunnel Category A............................................................................................................. 16
4.2.2 Tunnel Category B............................................................................................................. 16
4.2.3 Tunnel category C ............................................................................................................. 16
4.2.4 Tunnel Category D ............................................................................................................ 17
4.2.5 Tunnel Category E............................................................................................................. 17
4.3 European Tunnel Restriction Codes (ADR Chap 8.6 p. 593-594) ................................................. 18
4.4 Application of the ADR Tunnel Regulations in Europe .................................................................. 19
5. Examples of National Standardised Risk Assessment Methods .................................................... 20
5.1 France ............................................................................................................................................ 21
5.1.1 Step 1 ................................................................................................................................ 21
5.1.2 Step 2a .............................................................................................................................. 22
5.1.3 Step 2b .............................................................................................................................. 22
5.1.4 CETU Booklet .................................................................................................................... 23
5.1.5 Example Evaluation ........................................................................................................... 23
5.2 Germany ......................................................................................................................................... 24
5.2.1 Step 1a .............................................................................................................................. 25
5.2.2 Step 1b .............................................................................................................................. 25
5.2.3 Step 2a .............................................................................................................................. 25
5.2.4 Step 2b .............................................................................................................................. 26
5.3 Austria ............................................................................................................................................ 26
5.3.1 Step 1 ................................................................................................................................ 26
5.3.2 Step 2a .............................................................................................................................. 27
5.3.3 Step 2b .............................................................................................................................. 27
5.3.4 Step 3 ................................................................................................................................ 27
5.4 Other Countries .............................................................................................................................. 28
5.5 Discussion of the Various Approaches – with Focus on the Evaluation of Results ....................... 28
6. Quantitative Risk Assessment Applications Under the National Approaches ............................. 30
6.1 Dangerous Goods Quantitative Risk Assessment Model .............................................................. 31

Austroads 2024 | page iii

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

6.1.1 Scenario and Frequency Analysis ..................................................................................... 31

6.1.2 Consequence Analysis ...................................................................................................... 32
6.1.3 Output ................................................................................................................................ 34
6.2 The ‘German Model’ ....................................................................................................................... 36
6.2.1 Scenario and Frequency Analysis ..................................................................................... 36
6.2.2 Consequence Analysis ...................................................................................................... 38
6.2.3 Output ................................................................................................................................ 38
6.3 TuRisMo ......................................................................................................................................... 38
6.3.1 Scenario and Frequency Analysis ..................................................................................... 39
6.3.2 Consequence Analysis ...................................................................................................... 40
6.4 Comparison of DG-QRAM, the German Model, and TuRisMo-DG ............................................... 41
7. US Approach ........................................................................................................................................ 43
8. Prior Australasian Risk Assessments ............................................................................................... 45
9. Conclusion............................................................................................................................................ 46
References ................................................................................................................................................... 48
Appendix A Jurisdictional Differences .............................................................................................. 50
A.1 Tunnel Characteristics.................................................................................................................... 50
A.2 Dangerous Goods Fleet ................................................................................................................. 51
A.3 Operational Differences.................................................................................................................. 51

Tables

Table 4.1: DG classes according to the ADR.......................................................................................... 15

Table 5.1: Example multi-criteria assessment table for the final evaluation ........................................... 23
Table 5.2: German scenarios and limits for the DG-QRAM evaluation .................................................. 25
Table 6.1: DG-QRAM’s different possible approaches ........................................................................... 31
Table 6.2: Thirteen scenarios considered in DG-QRAM ......................................................................... 32
Table 6.3: DG-QRAM scenarios and their effects ................................................................................... 33
Table 6.4: Overview of the studied scenarios, also noting the relevant ADR tunnel category ............... 37

Figures

Figure 1.1: Flowchart of a generic risk assessment procedure.................................................................. 2

Figure 4.1: Restricted DG – Category B tunnel ........................................................................................ 16
Figure 4.2: Additional restricted DG – Category C tunnel ........................................................................ 17
Figure 4.3: Additional restricted DG – Category D tunnel ........................................................................ 17
Figure 4.4: Tunnel restriction codes ......................................................................................................... 18
Figure 5.1: Overview of the French method ............................................................................................. 21
Figure 5.2: German method ..................................................................................................................... 24
Figure 5.3: Overview of the Austrian method ........................................................................................... 27
Figure 6.1: Scope for the overarching route assessment ........................................................................ 30
Figure 6.2: Example of a FN diagram (absolute value approach) ........................................................... 35
Figure 6.3: Example of an F-N diagram – comparative approach ........................................................... 36
Figure 6.4: The transient 1D/3D smoke propagation model .................................................................... 41

Austroads 2024 | page iv

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

List of Abbreviations
1D one-dimensional
2D two-dimensional
AASHTO American Association of State Highway Transportation Officials
ADG Code Australian Code for the Transport of Dangerous Goods by Road & Rail
ADR European Agreement concerning the International Carriage of Dangerous Goods by Road
APTA American Public Transport Association
ASFINAG Autobahn and Schnellstraßen-Finanz-Aktiengesellschaft
BLEVE boiling liquid expanding vapour explosion
CFD computational fluid dynamics
d day
DG dangerous goods
DG-QRAM Dangerous Goods Quantitative Risk Assessment Model
DG HGV dangerous good heavy goods vehicle
EV expected (risk) value
F frequency
FFFS fixed fire-fighting systems
HGV heavy goods vehicle
HM hazardous materials (used in the US instead of “dangerous goods”)
IRAM Italian Risk Analysis Method
Kg kilogram
LPG liquefied petroleum gas
MW megawatt
N number of fatalities
No number
OECD Organisation for Economic Co-operation and Development
PIARC World Road Association
QRAM quantitative risk assessment model
S second
sc. scenario
sub. substance
SFAIRP so far as is reasonably practicable
t metric tonnes
TERN Trans-European road network
TuRisMo Tunnel Risk Model (risk model developed in Austria)
US United States (of America)
veh vehicle
VCE vapour cloud explosion
VSL value of statistical life

Austroads 2024 | page v

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

1. Introduction
The main motivation of applying risk-based assessment models to the routing of dangerous goods loads
through tunnels is to provide a rational, reliable and traceable basis for risk-based decision making. By
providing what is hoped is the most logical approach to making such decisions, road authorities may be
assisted to make the best decisions, and through that, also be seen clearly to have appropriately discharged
their responsibilities.

This Austroads project (ART6122) seeks to establish a standardised Australasian approach to comparing the
risk from dangerous goods (DG) in tunnels with that on the alternative surface route(s), to document the
method, and to apply it to an example tunnel. This report reviews best practice in tunnel DG risk assessment.

The idea of a comparative risk assessment necessarily starts with acknowledgement that there are also risks
on the alternative surface route. That in itself may be new to some. Often, discussion on what to permit
through tunnels focusses only on the tunnel, despite that some surface routes may not meet the
requirements in Section 13.1.4 of the Australian Code for the Transport of Dangerous Goods by Road & Rail
(ADG Code), which talk about DG route planning.

This document has an obvious European focus. That is entirely expected. Based on our experiences in
Europe, the United States, and to a lesser extent Asia, the Europeans prosecute risk analyses more widely
and more scientifically than others. This comes largely from the philosophical approach to safety, with first
principles thinking in which nothing is taken for granted, with everything thought about and justified. Risk
assessment is a clear way of thinking about the issues and provides both assurance and documentation that
risk has been managed appropriately and transparently.

The philosophy in the United States is significantly different. The provisions mandated in one or more codes
are often the start and finish of design conversations about safety, in that if a prescriptive approach is
followed, the level of residual risk is deemed to be acceptable. Consequently, risk analysis for tunnels with or
without dangerous goods vehicles (DGVs) is not so developed. US federal regulations do require risk
analyses for DG in tunnels, but no method is mandated, and the states individually decide how it is done,
apparently with varying levels of quantification. The one US example we have found applied a method
developed in Europe (DG-QRAM). The authors of this report were previously engaged on a comparative
assessment for DG routing involving a major US interstate tunnel. It is reflective of the DG risk assessment in
the United States that the acknowledged assessment method is to apply both DG-QRAM and the Austrian
Tunnel Risk Model (TuRisMo), with the latter applied to better capture detail arising from a complicated
ventilation system and from a fixed fire suppression system.

Asia seems to us to be somewhere in between in their development of risk assessments related to road
tunnels. There are many de facto or mandated standardised design approaches, but also some enlightened
developments such as the investigation by Singapore’s Land Transport Authority into effectiveness of water-
based suppression in tunnels. For that investigation, which sought to challenge the standard approach, they
looked to Europe, as we believe they do for leadership in risk analysis.

Risk-based approaches have become important tools in Europe, since a tunnel risk assessment became
mandatory for trans-European tunnels via EU Directive 2004/54/EC (European Union 2004). Many European
countries then adopted the requirement for all their tunnels. The main objective is to have a comparable and
uniform safety standard for all tunnels within the Trans-European road network (TERN). Particularly in that
context, risk-based approaches are important tools in the governance of design decision making, since they
allow a structured and transparent assessment of tunnel risks.

A risk assessment is a tool to identify the hazards and analyse the probability and magnitude of harm in
order to obtain a (quantifiable) risk indicator. Figure 1.1 illustrates the typical procedure for a risk
assessment. The blue cells correspond to the risk analysis steps, the brown cells to those of the risk
evaluation and the grey cell to the risk reduction.

Austroads 2024 | page 1

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Figure 1.1: Flowchart of a generic risk assessment procedure

The types of harm are the same for DG-carrying vehicles as for other vehicles, perhaps with differences in
the potential level of harm. In general, four different types of harm can be investigated:
• Harm to people (including injuries as well as fatalities)
This is the most important hazard type relating to the assessment of the transport of dangerous goods
through tunnels. The most commonly used indicator for the quantitative assessment is statistical fatalities.
The primary affected people are the tunnel users. However, people situated next to the tunnel portals and
possibly also those downwind and even above the tunnel (for shallow urban tunnels) could be affected.
Emergency responders may be at risk in responding to incidents. There is also a low probability that
some maintenance related inspection might be underway at the time of an incident. For surface routes,
the two affected groups are the road users, and the people close to the route or downwind or
downstream. DG-QRAM does not include risk to emergency responders, as that risk may be completely
different depending on the different approaches to emergency response that may be used. Responders
will undertake risk assessments in addressing each incident and adjust their response accordingly.
• Economic losses
Two types of economic loss may be distinguished. There are the (direct) capital losses due to the
damage caused by the event, and there are the (indirect) economic losses due to the tunnel closure
(longer or slower travel and transport routes, higher travel and transport costs, affected business losses,
possible loss of toll revenue, etc.).
• Impact on the environment
The impact on the environment may also be relevant to the transport of dangerous goods. Whilst the
environmental pollution expected from an event in a tunnel is normally limited by capture of in-tunnel
spillage and suppression water, the consequences of a liberation of noxious substances in the open can
be significant (pollution of soil, groundwater, habitat destruction, etc.). The routes may also vary
significantly in the required vehicle fuel use and resultant emissions.

Austroads 2024 | page 2

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

• Reputational damage
Spectacular events causing higher damage (to people and/or property) attract a high media attention,
causing public discussions and reactions at political level. A major road tunnel incident may result in
significant damage to a tunnel operator’s reputation. In comparison to conventional tunnel risks, like
collisions or vehicle fires not involving dangerous goods, incidents involving the release of dangerous
substances in a tunnel are characterised by very low probabilities but very high potential consequences.
Even unspectacular spills of the wrong material could close a tunnel for some time, generating similar
media interest. The harm under this heading could include loss of trust in the asset owner or authorities
(with unduly high expenditure on changes to recover trust), reduced tolls, reduced economic use of the
tunnel, or difficulty in gaining acceptance for new tunnel projects. Similar risks apply if the decision is
taken to exclude DGs from a tunnel and a high consequence event occurs on the surface route.

The risks linked to these types of harm can be analysed either quantitatively or qualitatively, or both. In
comparison to qualitative methods, which are mostly focussing on the interaction and the interdependence of
events, quantitative methods enable the calculation of characteristic risk values. Often, due to the lack of
sufficient data for complete quantitative analysis, qualitative and quantitative components might be combined
in practical applications.

For potential high-profile incidents, a closer look at reputational damage is required. Due to the specific
influence of risk perception described above, in the process of risk evaluation, special attention is to be paid
to incidents with potentially very high consequences (Kohl & Krieger 2012). This requires special techniques
for displaying the risk indicators (e.g. in a frequency-consequence diagram), for the definition of risk
acceptance parameters, and on the calculation of representative overall risk values (e.g. by overrating of
incidents with very high consequences by applying adequate weighting factors).

Further, we can differentiate between scenario-based and system-based risk assessment approaches. The
scenario-based approach is more qualitative, in considering broad classes of incidents, the relative
frequency and the likely consequences, leading to an understanding of the principal contributors to risk and
hence how the design might best be modified. The system-based approach categorises all events by the
primary parameters of the event. Then, with very specific parameters, probabilities and consequences may
be evaluated numerically with some credibility, permitting mathematical combination into a single overall risk
number (Kohl & Krieger 2012).

A complete comparative assessment of the tunnel route’s risks against those of the alternative route(s) – in
regard to the transport of DGs – consists at least of three major steps: an assessment of the tunnel route
(including the tunnel approaches), an assessment of the alternative route(s) and the comparison of the
obtained results.

Assessment of risk from terrorist events is generally kept separate from general infrastructure risks, partly for
security reasons related to document distribution. However, there is a question as to whether there is an
interaction which increases the risks from DG in tunnels in a way that is not relevant to surface routes. It is
noted though that terrorist risks in Australasia are generally extremely low.

Austroads 2024 | page 3

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

2. Limitations of Quantitative Risk Assessment

There are many approximations in risk assessment calculation methods. For DG risk assessment methods,
the simplifications and approximations in the methodology, as well as the uncertainties in the inputs will
always limit the accuracy of any risk estimate. First, the theoretically possible events and chains of events
are so numerous and situation dependent, that considering them all correctly is simply not feasible. Thus, in
any DG risk assessment, quantitative or qualitative, approximations and assumptions are inevitable.
Quantitative risk assessment results are a statistically-based approximation to what could most likely happen
in reality. It will always be only an approximation to reality and never a foretelling of exactly what will happen
in a specific event.

Second, it seems logical that the more encompassing an assessment is (e.g. by covering many scenarios),
the more realistic the results will be. However, the accuracy of any risk estimation is obviously limited by the
completeness and correctness of the initial information and input data processed. One must therefore be
cautious to not produce tremendously detailed analyses delivering results to many decimal places and
pretending a precision that is outrunning both the initial data quality and the modelling approximations.
Hence, the level of detail should be truncated where the assessor’s knowledge of the inputs determines that
no further accuracy or certainty can be achieved.

However, the benefit of including greater numerical precision than warranted by data certainty is that it
permits the calculations to be followed better by a reader, where otherwise rounding of numbers at every
stage could cause confusion. Further, to avoid compounding rounding errors where only a single significant
digit may be justified in the end result, additional digits must be retained through the calculations, and it is not
always clear whether there is further numerical use to be made of any apparently ‘end’ result. Thus, all
results may reasonably be displayed to greater precision than apparently justified by the inputs.

Nevertheless, a systematic analysis of potential hazards, its probabilities and consequences following the
guiding principles of good risk assessment practice should be more reliable than an intuitive expert
judgement which tends to be biased by personal experience and perspective. After more than a decade of
development and experience in practical application, the leading risk assessment tools for road tunnels are
reliable and have proven to deliver realistic, understandable and reproducible results. However, applying
these tools requires expert knowledge and experience in the complex matter of tunnel safety as well as an
excellent knowledge of the model structure and background, and the potential and limitations of the tool
(model) being applied. Most of the problems observed in this context are caused by inadequate application
or misinterpretation of results. For instance, some tunnel DG risk studies, are performed without any specific
knowledge of the amount or composition of DG transported through this tunnel, with default values being
applied. Or, to quote another example, for the evaluation of the effectiveness of safety measures, it is key to
know whether the physical effects of a measure under investigation can be modelled properly with the tool in
use.

The present work is about comparing risks between two competing routes for DG transport. That greatly
reduces the impact of inaccurate assumptions on those inputs that apply to both routes. For example, if the
probability of a major fire given a crash is overstated, it will be so for both the tunnel and alternative routes,
leaving the comparative risk unaffected. In that way, the result from the comparative quantitative risk
assessment (QRA) will always be more robust than absolute QRA.

Austroads 2024 | page 4

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

It is important also to acknowledge that QRA is one tool among many used in informing multifaceted
decisions. The scope of the present study does not include the assessments beyond the QRA. However, it is
important, in framing the QRA method, to place the QRA in context. The whole-of-government decisions
required on transport routes must also consider the social costs and benefits, the risks to the environment,
threats to municipal water supply, and the economics of road transport. There is no direct way of comparing
risk to life with risk to the environment, or the social benefit of town centre renewal. That is, the other
important considerations may not be quantifiable in the same way as risk to life and cannot be combined in
one quantitative assessment. Consequently, the life safety QRA will not be the only factor in determining the
decision.

Placing the QRA in that context also illuminates the discussion about risk benchmarks and the SFAIRP
approach. When risk to life is very small, differences in that risk between two options will be even smaller.
When faced with an extremely low risk difference in the context of a multifaceted decision, there is no direct
way of judging the significance of that extremely low difference, and so the quantifiable SFAIRP approach
may not be applicable when considering all types of potential harm.

The risk assessment is not only about the physical and the traffic throughput; operation of the tunnel is also
an influence on risk. A recent example is DGVs being marshalled before the tunnel and convoyed through a
tube with other traffic held back to minimise tunnel risk. In this case, the incident risk beyond the tunnel is
possibly higher as the resulting denser batch of cars overtake the dissipating convoy. The disruption to traffic
flow introduced in seeking to reduce tunnel risk could well increase overall risk. Of course, this depends also
on the road downstream of the tunnel and the operation of the convoy vehicles on that road.

The conclusion is perhaps that the risks from all route options (tunnel and surface) be stated clearly along
with those other considerations, with explicit statement that they must all be assessed together holistically
and that SFAIRP as a concept applies to that holistic judgement, not to the QRA as a subset of the decision
inputs.

Austroads 2024 | page 5

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

3. Responsibilities of the Road Authority

A risk assessment model, and the method within which it is used, must fit the needs of the road authorities
seeking to discharge their responsibilities. This section was prepared by scientist and barrister Arnold Dix, to
give a legal and governance context to the model and method development of this project. Note that the
referencing in this section only is done by footnote, with the references not listed in the references section at
the end of the document.

3.1 Overview
It is generally accepted in all Austroads legal jurisdictions that each member of the public is entitled, as of
right, to pass along a public road (whether on foot, in a vehicle or otherwise) and to drive stock or other
animals along the public road. 1 Any restriction on the passage of goods through road tunnels is a restriction
of that right.

Typically, such restrictions will be imposed by a road authority, dangerous goods transport regulator or a
combination of statutory authorities which may have overlapping jurisdiction. 2 The regulation of the transport
of goods prescribed as dangerous, or otherwise worthy of transportation restrictions, is an exercise of power
to restrict the general rights of passage of the public and must therefore be exercised in accordance with law.

Because each state in Australia is a different legal entity, and because each of the states has transferred
some of its powers to the Commonwealth of Australia, and because New Zealand is a separate country, the
exact identity of the authorities with power, or which claim to have power, over dangerous goods transport
varies from place to place and time to time.

Both Australia and New Zealand look to the United Nations’ model regulations as a basis for regulating
dangerous goods transport:
The UN Economic and Social Council (ECOSOC) Sub-Committee of Experts on the
transport of Dangerous Goods is responsible for developing and reviewing the UN
Recommendations on the Transport of Dangerous Goods – Model Regulations (UNMR).
The UN Recommendations are internationally accepted and form the basis for dangerous
goods transport codes across the world. Australia contributes to the development of the UN
Recommendations as a participating member state. 3

Notwithstanding the national intent of implementing the UN model regulations, 4 the reality is that each road
authority operates under its own statutory and common law regime and its activities occur in a complex
regulatory ecosystem with many statutory authorities as stakeholders. This means that in all states and
countries the implementation of the model regulations is subject to the local laws, and tunnels are a
specialised subset of the road network requiring special attention by road authorities.

1 Such rights arise from common law and from statute – see for example: Roads Transport Act No 18, NSW, 2013, Section 5.
2 See for example placarded goods restrictions Road Rules 2014 - REG 300-2, NSW rule: carriage of dangerous goods in prohibited
areas; and in addition, the area restrictions on transport of explosives: Explosive Regulation 2013 (NSW) Reg 89.
3 NTC 2020: Examining the legal framework for the land transport of dangerous goods, June 2020, National Transport Commission
Australia.
4 See for example for Australia: National Road Transport Commission Act 1991 of the Commonwealth of Australia.

Austroads 2024 | page 6

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Typically, a road authority is created by an act of parliament 5. Road authorities are uniquely empowered and
entrusted to safely build, operate, and maintain the road network. The exact mandate for a specific road
authority is determined through detailed legal analysis of the unique combination of parliamentary statutes
which define in detail their powers, duties, and obligations 6, 7, 8.

No two road authorities are the same. As new laws are enacted, so too can the obligation on a road authority
change. It is likely that in the future some parliaments will codify road authorities’ responsibilities for tunnels,
to remove uncertainty about their obligations, powers, and liability, in respect of modern road networks.

The regulation of the passage of goods which may be dangerous in road tunnels is primarily, but may not be
exclusively, the statutory responsibility of road authorities, 9 noting that in Victoria, prohibitions on the use of
tunnels are made and gazetted under the Dangerous Goods Act 1985. The general classification and
regulation of dangerous goods transport can be the subject of decisions made by other authorities. 10 The fact
that classes of radioactive materials are not dealt with under dangerous goods regulation, 11 and that the
propulsion systems of vehicles are excluded from dangerous goods classification, highlights that road
authorities must exercise their duties in relation to the passage of anything that could be dangerous goods in
tunnels, and not merely rely on classification by others.

Parallel and overlapping regulation of dangerous goods transport by road from other authorities does not
alter the road authorities’ responsibilities. Road authorities can utilise the classifications of other authorities
as a means of regulating the passage of placarded loads through tunnels 12 but they must do so by engaging
in a process to ensure that the outcome satisfies their unique statutory powers and duties.

The civil liability of road authorities is substantially modified by statute. While the exact modification may
differ between road authorities, both the common law and statutory modifications somewhat protect road
authorities from civil liability. These legal protections for road authorities arise from the recognition by courts
and parliament of the special risks to which road authorities are exposed, and the social utility of protecting
them from legal liability. 13 It is inevitable that the extent of this protection will change in time. 14

Tunnels are, by their nature, prone to creating hazards for users, vehicles and goods. The hazards are
uniquely the responsibility of road authorities and could not reasonably be expected to be within the
knowledge or control of other regulatory authorities. The unique physical properties of a tunnel when
combined with a road transportation function make the regulation of dangerous goods through tunnels a
responsibility of road authorities.

A decision to regulate the passage of dangerous goods through tunnels by a road authority must be both
reasonable and not negligent.

5 See for example Roads and Maritime Services, Roads Act 1993, NSW.
6 See for example Roads Act 1993, NSW, Section 78 Roads Authority has specific power to construct tunnels across navigable waters
(e.g. Sydney Harbour).
7 See for example ‘RMS is the road authority for all freeways in NSW’, Section 7, Roads Act 1993, NSW.
8 See for example Section 7(5) of the Roads Act 1993 NSW which broadly defines the functions of all road authorities (including RMS)
as [those] ‘conferred on it by or under this or any other Act or law’.
9 See for example Road Transport Act No 18 Section 4, NSW, 2013; Definition of ‘road infrastructure’ includes ‘a road, including its
surface or pavement, and … any bridge, tunnel, … or other work or structure forming part of a road system or supporting a road, and
… any … other work or structure located above, in or upon a road and maintained by the road authority … and … any practical control
device … electricity equipment, emergency telephone systems or any other facilities (whether of same or a different kind) in, on, over,
under or connected with [these things]’. Road Transport Act No18 NSW 2013’ Schedule 1 Section 7 provides ‘Examples of Statutory
Rule Making Powers’. It identifies one of the example powers as ’Mass, dimension, load restraint and access requirements with
respect to the use of roads by vehicles and combinations and the enforcement of such requirements’, It defines access requirement
as ‘a requirement that relates to the roads or class of roads on which a vehicle or combination may or may not be taken, or otherwise
limits the area in which a vehicle or combination may or may not operate‘.
10 See for example in NSW the Dangerous Goods (Road and Rail Transport) Regulation 2014 and the determination of routes – Part 1,
Division 4 Clause 25. By way of example in NSW The Competent Authorities are Safework NSW and Environment Protection
Authority NSW which have different statutory duties and responsibilities to NSW Road Authorities. In Victoria the competent authority
is the Victorian Workcover Authority, while in Western Australia it is the Department of mines.
11 Radiation Control Act 1990 - SECT 14.
12 See for example NSW ROAD RULES 2014 - REG 300-2
NSW rule: carriage of dangerous goods in prohibited areas which evokes placarded loads under the Dangerous Goods (Road and
Rail Transport) Regulation 2014 NSW, Explosives Regulation 2013 NSW and the Radiation Control Regulation 2013 NSW.
13 See for example Civil Liability Act, NSW, 2002 and Buckle v Bayswater Road Board (1936) 57 CLR 259; Gorringe v Transport
Commission (Tas) (1950) 80 CLR 357.
14 Brodie v Singleton Shire Council [2001] HCA 29; (2001) 206 CLR 512 at 577.

Austroads 2024 | page 7

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

The requirements for a ‘reasonable’ decision for the regulation of dangerous goods and the requirement that
a road authority not be ‘negligent’ are separate legal requirements. This means that a road authority must be
mindful of both its statutory obligations as an authority and its civil responsibilities not to do harm. The two
obligations are different, albeit that they may rely upon similar statutory provisions.

A road authority must be able to demonstrate that its decision to regulate the passage of dangerous goods
through tunnels is ‘reasonable’.

It is well established that reasonableness is a ground of judicial review of administrative action in Australia
and New Zealand. 15 One of its earliest and most often cited conceptualisations of statutory authorities is in
the judgment of Gibbs J in Parramatta City Council v Pestell where his Honour held that the Council’s
opinion would not be valid if it were so unreasonable that no reasonable council could have formed it. 16

High Court decisions which have expounded on the reasonableness ground of judicial review contain useful
commentary on how a court would likely assess the question of the reasonableness of a decision to regulate
dangerous goods through tunnels.

Legal unreasonableness can occur in two principal ways: first, where an identifiable error has occurred in the
decision-making process (i.e. where a finding of fact is not supported by evidence) 17 and secondly, where
the outcome of the decision-making process is said to be characterised as being legally unreasonable. This
second way is known as outcome focused unreasonableness and does not require the identification of a
specific error – instead, an assessment of the quality of the decision must be undertaken.

The High Court considered the scope of outcome focused unreasonableness in Minister for Immigration and
Citizenship v Li [2013] HCA 13. The Court found that the failure by the [court] to grant an adjournment to the
applicant was unreasonable and rendered the Tribunal’s refusal of her application invalid. French CJ held
that the concept of unreasonableness:
Reflects a limitation imputed to the legislature on the basis of which courts can say that
Parliament never intended to authorise that kind of decision. After all the requirements of
administrative justice have been met in the process and reasoning leading to the point of
decision in the exercise of a discretion, there is generally an area of decisional freedom.
Within that area reasonable minds may reach different conclusions about the correct or
preferable decision. However, the freedom left by the statute cannot be construed as
attracting a legislative sanction to be arbitrary or capricious or to abandon common sense.

French CJ’s judgment makes it clear that the courts understand that reasonable minds may reach differing
conclusions about the correct decision. In this way, road authorities have the ability to make a decision on
the transport of dangerous goods through tunnels even where there are alternative options which could
likewise be considered reasonable. In this way, the risk assessment methodology described within these
documents as best practice provides a foundation for making decisions even when other reasonable and
potential contradictory decisions could be made.

A decision which is made within the area of decisional freedom alluded to above is not affected by
jurisdictional error on the ground of legal unreasonableness. This means that the existence of a number of
options for regulating the passage of dangerous goods allows a road authority to make an informed choice –
a choice which is reasonable and therefore lawful.

In a joint judgment, Justices Hayne, Kiefel and Bell developed the notion that unreasonableness is linked to
rationality and logicality. They held that:
[un]reasonableness is a conclusion which may be applied to a decision which lacks an
evident and intelligible justification. 18

15 Parramatta City Council v Pestell [1972] HCA 59; Minister for Aboriginal Affairs v Peko-Wallsend (1986) 162 CLR 24
16 [1972] HCA 59.
17 For example: regulating a class of placarded dangerous goods without evidence of the risk they posed in the tunnel environment.
18 Minister for Immigration and Citizenship v Li (2013) 249 CLR 332; [2013] HCA 18.

Austroads 2024 | page 8

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Justice Gageler held that decision-making authority:

conferred by statute must be exercised according to law and to reason within limits set by
the subject-matter, scope and purpose of the statute. 19

In Minister for Immigration and Border Protection v SZFVW [2018] HCA 30, Nettle and Gordon JJ held that
the court’s task in determining a claim grounded upon legal unreasonableness:
requires the court to assess the quality of the administrative decision by reference to the
statutory source of power exercised in making the decision and, thus, assess whether the
decision was lawful, having regard to the scope, purpose and objects of the statutory
source of the power.

The determination of legal unreasonableness is heavily fact dependent and inherently sensitive to context. 20
It is therefore not capable of reduction to a formula and no single legal test of unreasonableness can be
stated. 21

In Li, Gageler J held that the test to be applied on judicial review grounded upon legal unreasonableness is
stringent. Accordingly, the cases in which it has been successfully established that an administrative decision
was affected by legal unreasonableness are rare.

A road authority that can clearly articulate its statutory powers and obligations to an identified and
transparent framework for making decisions about the regulation of dangerous goods through tunnels is in
an extremely strong position to demonstrate the legal reasonableness of its actions.

It is clear from an analysis of these and other judgments, that any finding of unreasonableness of a road
authority in any context, including in the context of a decision regarding transportation of dangerous goods
through tunnels, would be heavily dependent on the relevant statutory and regulatory context. It is this which
will provide the framework for determining what is to be considered reasonable/unreasonable. The fact that
there are other authorities that may regulate the transport of dangerous goods does not excuse a road
authority from exercising its power responsibly.

It is within this articulation of ‘unreasonableness’ that the methodologies for considering comparative and
absolute assessment for the acceptability of risk through risk analysis techniques must be analysed. While
the tools for risk analysis may be used in multiple overseas jurisdictions, it does not follow that their use by
Austroads members would withstand a reasonableness test if the outcomes appear to abandon common
sense. Similarly, if there are alternative tools that both meet Australasian reasonableness and common-
sense requirements, road authorities could choose either according to their preference.

It must be noted that the concept of reasonableness is not within European law (the legal framework within
which DG-QRAM and other models evolved). In Europe, the test for regulation of dangerous goods vehicles
in tunnels would activate legal principles of proportionality (the German word is ‘Verhältnismäßigkeit’ and in
France ‘the concept of balance’.)

This means that the application of European methodologies in an Austroads context needs to be challenged
and verified for local regulatory and factual conditions.

Obvious practical challenges in adjusting these foreign methodologies for managing risk include ensuring
that the value of a human life within the Austroads jurisdiction is factored into the models (if it is relevant), the
widespread use of active fire suppression systems in Australasia, the differing regulatory arrangements for
dangerous goods, to name but a few.

19 Ibid.
20 SZFW [2018] HCA 30.
21 Ibid and BFH16 v Minister for Home Affairs & Anor [2019] FFCA 730.

Austroads 2024 | page 9

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

The importance of ensuring that the methodology used by Austroads members is calibrated for local
conditions is entirely consistent with the European Union’s approach to risk management as reflected in the
EU Directive on Trans-European Road tunnels. 22 In that directive, member states are required to develop
their own methodology for their country – the EU does not impose a methodology 23. Although a harmonised
methodology was contemplated in 2004, 24 it has never been pursued.

The approach suggested in this report is considered best practice. So long as its use is validated and
calibrated (its outcomes tested to ensure they meet the “common sense” test), the outcomes would be
considered robust and reasonable.

3.2 Negligence
In order to ensure that decisions made on regulating dangerous goods movement through tunnels are robust
and survive accusations of negligence, road authorities must have regard to their special position as a road
authority and contrast it with other statutory authorities with apparently similar responsibilities for dangerous
goods.

Each road authority’s liability and vulnerability to such an attack is a function of its local legal context. The
exact circumstances of the allegation of negligence against a road authority are critical to determining the
road authority’s liability. 25 The most likely scenario for a negligence action against a road authority in a
dangerous goods context would be where fatalities occur either within a tunnel or on an alternative route to a
tunnel, where the cause of the harm is identified as the road authorities action (or inaction) with respect to
dangerous goods passage.

In Brodie v Singleton Shire Council [2001] HCA 29; (2001) 206 CLR 512 at 577 [150], Gaudron, McHugh and
Gummow JJ held that, under the common law of Australia, public authorities who had statutory powers of the
nature of those conferred by the Local Government Act 1919 upon the local councils involved in Brodie:
“… to design or construct roads, or carry out works or repairs upon them, are obliged to
take reasonable care that their exercise of or failure to exercise those powers does not
create a foreseeable risk of harm to a class of persons (road users) which includes the
plaintiff. Where the state of a roadway, whether from design, construction, works or non-
repair, poses a risk to that class of persons, then, to discharge its duty of care, an authority
with power to remedy the risk is obliged to take reasonable steps by the exercise of its
powers within a reasonable time to address the risk. If the risk be unknown to the authority
or latent and only discoverable by inspection, then to discharge its duty of care an authority
having power to inspect is obliged to take reasonable steps to ascertain the existence of
latent dangers which might reasonably be suspected to exist.)”

The importance of this case is that it demonstrates that a road authority may be found negligent for failing to
exercise its statutory powers. Gibbs CJ in Sutherland Shire Council v Heyman (1985) 157 CLR 424 at 443
confirmed that, in principle, failure to exercise statutory powers could also be the subject of an action in
negligence.

The position of road authorities is typically modified by statute. The exact modification varies from jurisdiction
to jurisdiction, again highlighting the importance of conducting a detailed legal assessment of the statutory
position of each road authority. 26

22 Tunnel Directive 2004/54/EC (minimum safety requirements for tunnels in the Trans-European Road Network).
23 Tunnel Directive 2004/54/EC, Article 13 Member States shall ensure that, at national level, a detailed and well-defined methodology,
corresponding to the best available practices, is used and shall inform the Commission of the methodology applied.
24 Tunnel Directive 2004/54/EC, Article 13 Where necessary, it shall make proposals for the adoption of a common harmonised risk
analysis methodology.
25 Road and Traffic Authority (NSW) v Refrigerated Roadways Pty Ltd (2009) 77 NSWLR 360.
26 See for example: Road Management Act 2004 (Vic) which has codified the extent of civil liability of Victorian road authorities to some
extent.

Austroads 2024 | page 10

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

For example, in New South Wales, the principles to be applied in determining whether the road authority’s
duty of care has been breached are specifically restricted by Section 42 of the Civil Liability Act of 2002,
which modifies the process of determining whether a road authority has a duty of care or has breached a
duty of care. The Act specifically provides that in judging the road authority the following apply:
“(a) the functions required to be exercised by the authority are limited by the financial and
other resources that are reasonably available to the authority for the purpose of exercising
those functions,

(b) the general allocation of those resources by the authority is not open to challenge,

(c) the functions required to be exercised by the authority are to be determined by

reference to the broad range of its activities (and not merely by reference to the matter to
which the proceedings relate),

(d) the authority may rely on evidence of its compliance with the general procedures and
applicable standards for the exercise of its functions as evidence of the proper exercise of
its functions in the matter to which the proceedings relate.” 27

It is within this legislative context that the role and potential power of an Austroads agreed assessment
methodology for the transport of dangerous goods through tunnels emerges.

An agreed methodology provides demonstrable evidence of a proper methodology for managing the risks
and the road authority exercising its powers. The weight to be attributed to the assessment methodology
adopted by a road authority will in part be driven by the rigour of the assessment process of the adopted
methodology for dangerous goods transport regulation through tunnels. At the heart of such an assessment
is the intellectual rigour of the process adopted.

The peculiar nature of tunnel risks and road authorities’ special skills in managing such risks place special
burdens on road authorities to discharge their statutory responsibilities and common law duties with diligence
and subject matter expertise.

At the heart of the exercise of this discretion is the common law question of what is ‘reasonably practicable’.
As noted above, the common law position of road authorities is modified by statute. The exact modification is
dependent upon which state or country the road authority is created, and the laws which apply to that road
authority at the relevant time.

The basic principle is that the common law duty of care is often interpreted as requiring any section of
roadway (surface road, bridge or tunnel) to be as safe as ‘reasonably practicable’ for tunnel users. The term
reasonably practicable is something narrower than ‘physically possible’ or feasible 28, is to be judged on the
basis of what is known at the relevant time 29, and to determine what is ‘reasonably practicable’ it is
necessary to balance the likelihood of the risk occurring against the cost, time and trouble necessary to avert
that risk. 30

27 Section 42 Civil Liability Act 2002 (NSW)

28 See for example, Edwards v National Coal Board [1949] 1 KB 704 at 712 per Asquith LJ; Marshall v Gotham Co Ltd [1954] AC 360
at 377 per Lord Keith of Avonholm; Auckland City Council v NZ Fire Service [1996] 1 NZLR 330 at 337-338 per Gallen J.
29 See for example, Edwards v National Coal Board [1949] 1 KB 704 at 712 per Asquith LJ; Marshall v Gotham Co Ltd [1954] AC 360
at 370 per Lord Oaksey, 377 per Lord Keith of Avonholm; Buchanans Foundry Ltd v Department of Labour [1996] 3 NZLR 112 at 118
per Hansen J.
30 See, for example, Coltness Iron Co v Sharp [1938] AC 90 at 94 per Lord Atkin; Edwards v National Coal Board [1949] 1 KB 704 at
710 per Tucker LJ, 712 per Asquith LJ, 715 per Singleton LJ; McCarthy v Coldair, Ltd [1951] 2 TLR 1226 at 1228 per Denning LJ,
1230 per Hodson LJ; Marshall v Gotham Co Ltd [1954] AC 360 at 370 per Lord Oaksey, 373 per Lord Reid; Austin Rover Ltd v
Inspector of Factories [1990] 1 AC 619 at 625 per Lord Goff of Chieveley, 635-636 per Lord Jauncey of Tullichettle; Auckland City
Council v NZ Fire Service [1996] 1 NZLR 330 at 338 per Gallen J.

Austroads 2024 | page 11

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Road authorities are not directed in their statute to follow any particular methodology for determining what is
reasonable or even reasonably practicable. There is no requirement that a specific methodology or
formulation for performing risk assessments is adopted, nor is there a specific requirement to follow a
specific standard or assessment methodology. 31 This means that road authorities are obliged to determine
which methodology is appropriate for discharging their statutory obligations.

There is a temptation to adopt strategies for risk management-based decision making that are imposed by
statute and regulation on other authorities and for other activities. 32 The proposed best practice management
of dangerous goods regulation in tunnels in this document is a bespoke tunnel safety focused mechanism to
assist road authorities in discharging their statutory duties. If found appropriate by a road authority within
their statutory duties, it offers a specialist methodology tailored to the highly technical and complex risks of
road tunnels and thereby provides a strong technical basis for justifiable, robust and thereby reasonable and
non-negligent decision making.

The requirement on each road authority (subject to local road authority legal confirmation) is to make
decisions on the transport of dangerous goods that are reasonably practical having due regard to the
statutory obligations of the road authority, and its unique allocation of resources and statute-modified duties.

Other statutory authorities can lawfully hold different opinions about the transport of dangerous goods
through tunnels. For example, fire authorities are not obliged to agree with a road authorities’ opinion, nor
are road authorities obliged to agree with fire authorities’ opinions about the transport of dangerous goods
through tunnels. Indeed, as is the case with road authorities – each separate fire authority has its own
special statutory and common law duties to discharge. It is to be expected that the lawful exercise of a fire
authority’s duties may result in a different decision on the appropriate regulation of dangerous goods in
tunnels than a road authority because they have differing statutory duties and obligations to uphold.

While the exact duties may vary between different fire authorities, invariably their legal duties will include
taking all practicable measures for preventing and extinguishing fires and protecting and saving life and
property. 33 The fire authorities’ duties are thereby different to the road authorities’ duties. Disagreement
between fire authorities and road authorities on the regulation of dangerous goods in tunnels from time to
time is to be expected and is evidence that both authorities are performing their functions in accordance with
parliament’s differing intent.

The assessment methodology proposed within this Austroads report is in substance the intellectual process
that can be relied upon by a road authority in exercising its powers to manage dangerous goods through
tunnels. The methodology ultimately provides a transparent mechanism for a road authority to exercise its
power and, subject to local statutory analysis, thereby defend itself from accusations of negligence.

3.3 Dangerous Goods – Placarded Loads

There are clear national agendas promoting the safe and efficient transport of dangerous goods in Australia
and New Zealand. Goods which must be placarded for road transport vary due to differences in the
regulatory regimes that deem placarding a requirement. 34,35 These are overarching economic and transport
policy objectives – not matters squarely within the authority and responsibility of road authorities.

31 See for example - Australian New Zealand Risk Management Standard (AS/NZS ISO 31000:2009) which is an organisational
generic approach created by a private company for profit and used in a broad range of contexts.
32 See for example – Slivak v Lurgi (Aust) Pty Ltd [2001] Q05CLR304 statutory requirement to reduce risk so far as is reasonably
practicable; Section 23 Occupational Health Safety and Welfare Act 1986 (SA); Bulga Underground Operations Pty Ltd v Nash [2016]
93NSWLR338; Occupational Health and Safety Act 1983 (NSW) Section 8, 28; Occupational Health and Safety Amendment Act 2011
(NSW) Section 7A; Rail Safety Act 2007 (SA) Section 28(1) safe systems of work and safe railway operations ‘so far as is reasonably
practicable’ [State implementation of Commonwealth National Law]; Rail Safety (Adoption of National Law) Act 2012 (NSW) Part 3,
Sections 47& 52 reducing risk ‘so far as is reasonably practicable’; Susan McCarrey, National Rail Safety Regulator v Sydney Trains
[2019] NSWDC 453.
33 See for example Fire Brigades Act 1989 (NSW) Section 6.
34 See for example: Radiation Control Act 1990 - SECT 14; Dangerous Goods (Road and Rail Transport) Regulation 2014 (NSW).
35 Explosive Regulation 2013 (NSW) Reg 89.

Austroads 2024 | page 12

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Road authorities are uniquely empowered as the principal responsible authority for road safety, yet they are
NOT a competent authority under the suite of dangerous goods legislation. 36 That legislation seeks to
regulate the passage of prescribed goods that are considered dangerous in a range of contexts – none of
which are specifically dealing with tunnels.

From the regulation of transport of dangerous goods by road and rail legislation 37 to the Radiation Control
Act 1990, there is no specific consideration of the unique risks posed by transporting goods through tunnels.
This risk is within the immediate and stated scope of the responsibilities of road authorities under statute.

Road authorities should not be dissuaded from their task of assessing the suitability of dangerous goods to
travel through tunnels because they are not a listed competent authority. Non road authority listed
Competent Authorities can rightly assert that the characterisation and placarding of goods for road transport
is generally within their power, but the unique attributes of tunnels demand special attention by road
authorities in the discharge of their statutory mandate. The fact that there may be an overlap of powers
between competent authorities and road authorities does not excuse either from performing their statutory
duties. It is incumbent upon the road authorities to come to their own reasonable view about the transport of
goods through tunnels on their road network – the unique and peculiar risks of such transport being squarely
within their expertise and authority.

It is to be expected, upon expert analysis by road authorities, that: some goods permitted to be transported
by road by non road authorities may be excluded from tunnels; some goods not permitted to be transported
by road by non road authorities could be transported by tunnels, and; that some goods not considered
dangerous at all by non road authorities in normal road transport contexts be found suited to regulation in a
tunnel context. An example is the bulk carriage of diesel fuel, which is not classed with dangerous goods and
yet could (with marginally lower probability) generate a fire comparable with classified dangerous goods,
which is perhaps equally significant in a tunnel.

Another example is the exclusion from dangerous goods control of vehicle propulsion systems at a time of
emerging high energy density alternative fuel systems and even conventional high volume diesel powered
vehicles. This highlights the need for road authorities to determine their own approach to regulating
Dangerous Goods in tunnels.

3.4 Conclusion
The proposed methodologies for the assessment of dangerous goods transport through tunnels in this report
form a robust framework for assisting road authorities to decide on the appropriate use of specific tunnels for
the transport of dangerous goods. The exact scope of each road authority’s powers and duties varies
between each of Austroads member road authority. The exact powers and duties may also vary in time.

The fact that a road authority is generally not also a competent authority in associated dangerous goods
legislation or radiation control legislation does NOT alter the responsibilities of road authorities to regulate
tunnel safety and dangerous goods within their jurisdiction.

No current Austroads road authority is required by statute to manage down the risk to road users so far as is
reasonably practicable. In fact, all parliaments have enacted specific legislation codifying the duty of care of
road authorities and directing the courts as to limitations on the exposure of road authorities to liability when
people are harmed – even in tunnels.

So long as the methodologies explored in this report are validated, calibrated and confirmed applicable to
each road authority’s operational context, they provide a robust platform for demonstrating both the
reasonableness of a road authority’s regulation of dangerous goods through tunnels and can be used to
robustly demonstrate that the standard of care to road users has been met and discharged.

36 See for example in New South Wales: Competent Authority Definition (NSW), Dangerous Goods (Road and Rail Transport)
Regulation 2014 (NSW), Part 1 Division 2 Clause 9.
37 Each state and territory implement its own legislation and regulations – see for example, Dangerous Goods (Road and Rail
Transport) Act 2008 NSW and Dangerous Goods Transport (Road and Rail) Regulations 2014 NSW.

Austroads 2024 | page 13

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

The requirement for reasonableness and that road authorities must discharge their statutory duties does not
mandate which of many methodologies and approaches should be used in managing the risks of
transporting dangerous goods through tunnels. Road authorities are required by statute to discharge their
obligations – how they discharge those obligations is currently open for their expert consideration and
decision.

The methodologies proposed in this report provide a basis for road authorities making robust and legally
responsible decisions on road tunnel safety risks. Subject to confirmation of their translation into the local
road authority regulatory context, they empower a road authority to responsibly regulate dangerous goods
passage through tunnels.

Austroads 2024 | page 14

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

4. Agreement Concerning the International

Carriage of Dangerous Goods by Road
Before going into the details of the different assessment methods and national approaches, it is necessary to
have a closer look at the UN Agreement concerning the International Carriage of Dangerous Goods by Road
(ADR) (United Nations 2018).

ADR stands for the French title ‘Accord européen relatif au transport international des marchandises
dangereuses par route’ and is a 1957 UN treaty concerning the international carriage of dangerous goods by
road, valid in 52 countries. The European reference in the title came about because it was developed under
the UN Economic Commission for Europe. It was renamed in January 2021, to be clearer that it is available
to all countries, both in Europe and elsewhere. It is available in English, French and Russian, with the
English title being “Agreement concerning the International Carriage of Dangerous Goods by Road”. The aim
of this regulation was, and still is, to permit the free movement of dangerous goods at an acceptable level of
safety. Two systems were defined: a classification system to regroup the large variety of dangerous goods,
and a categorisation system for tunnels that is based on the DG classification.

The latest version dates from 1 January 2021.

4.1 Classification of Dangerous Goods According to the ADR

Table 4.1 below shows the thirteen different classes and subdivisions of dangerous goods as defined in the
ADR. The Australian Code for the Transport of Dangerous Goods by Road & Rail (ADG Code) uses the
same set of descriptions (classes and sub-divisions) (NTC 2018) & (NTC 2020), as does the New Zealand
Standard 5433 Transport of dangerous goods on land. The only difference is that ADG Classes 4, 5 & 6
have divisions which in the ADR are called classes, even though they have the same decimal number
designations.

Individual products are classified in accordance with the criteria in Part 2 of the UN Model Regulations
(PIARC 2019b) and then assigned to a specific four-figure UN Number.

Table 4.1: DG classes according to the ADR

Class Description
1 Explosive substances and articles
2 Gases
3 Flammable liquids
4.1 Flammable solids, self-reactive substances, polymerizing
4.2 Substances liable to spontaneous combustion
4.3 Substances which, in contact with water, emit flammable
5.1 Oxidizing substances
5.2 Organic peroxides
6.1 Toxic substances
6.2. Infectious substances
7 Radioactive material
8 Corrosive substances
9 Miscellaneous dangerous substances and articles

Source: United Nations (2018).

Austroads 2024 | page 15

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

4.2 Tunnel Categories (ADR Chap 1.9 p.85-87)

The categorisation of a tunnel is based on the assumption that there are three major hazards that could
cause any of the types of harm (to people or the environment) (United Nations 2018):
• Explosions
• Release of toxic gas or volatile toxic liquid
• Fires.

The five established tunnel categories are described in the following sub-sections.

4.2.1 Tunnel Category A

There are no restrictions for the carriage of dangerous goods. No road sign or additional panel is necessary
(United Nations 2018).

4.2.2 Tunnel Category B

There are restrictions for the carriage of those dangerous goods which may lead to a very large explosion.
The goods concerned are given in Figure 4.1.

Figure 4.1: Restricted DG – Category B tunnel

Source: United Nations (2018).

4.2.3 Tunnel category C

There are restrictions for the carriage of those dangerous goods which may lead to either a very large
explosion, a large explosion, or a large toxic release. The goods concerned are the ones restricted in tunnel
category B plus those given in Figure 4.2.

Austroads 2024 | page 16

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Figure 4.2: Additional restricted DG – Category C tunnel

Source: United Nations (2018).

4.2.4 Tunnel Category D

There are restrictions for the carriage of those dangerous goods which may lead to either a very large
explosion, a large explosion, a large toxic release, or a large fire. The goods concerned are the ones
restricted in tunnel category C plus those given in Figure 4.3.

Figure 4.3: Additional restricted DG – Category D tunnel

Source: United Nations (2018).

4.2.5 Tunnel Category E

There are restrictions for almost all dangerous goods. Exceptions can be looked up in Table A, Chapter 3.2
of the ADR (“Dangerous Goods List”). In Column 15 ‘Transport category’ they are marked by a “( ̶ )” instead
of the restriction code.

Austroads 2024 | page 17

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

In general, if a transport unit carries more than one type of dangerous good, the most restrictive restriction
code for those goods shall be assigned to the whole load (United Nations 2018).

This assumes that the load is not made up of materials which when combined create a hazard worse than
any of the materials would on their own. The ADG Code Part 9 requires various degrees of segregation for
some reactive cargoes. Those requirements may not always be followed, but as it will be a very small
fraction of DG, it is taken as being within the ‘noise’ of mathematical risk estimates (United Nations 2018).

4.3 European Tunnel Restriction Codes (ADR Chap 8.6 p. 593-594)

Restricted tunnels have to be recognisable. They are required to be indicated by means of road signs and
signals including an additional panel bearing the restriction code letter. For larger corporate carriers
particularly, it is possible that restrictions could also be supervised by GPS tracking etc.

The restriction codes are given in Figure 4.4.

Figure 4.4: Tunnel restriction codes

Source: United Nations (2018).

Austroads 2024 | page 18

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

4.4 Application of the ADR Tunnel Regulations in Europe

The introduction of the tunnel categories described above created a unified system for regulating the
transport of dangerous goods through tunnels in the European Union. In parallel, some European countries
established their own national risk-based assessment approach, in order to define the tunnel category for
each individual tunnel on main transport routes, applying national risk acceptance criteria as well as risk
evaluation procedures. In this context, the following two aspects are worth mentioning:
• The principle of the tunnel categorisation is not risk based, but consequence based. The practical
experience of many risk assessment studies shows that, in most cases, the main contributors to the
overall dangerous goods transport risk in a tunnel are products of Class 3 (flammable liquids). Hence, in
order to achieve a significant reduction in risk, a tunnel has to be assigned to category D (or E) to prevent
flammable liquid tankers (which also means forbidding a major part of all hazardous products). So,
depending on the configuration, equipment, and traffic of the individual tunnel, a categorisation of the
tunnel in B or C is often not very effective in terms of reducing the absolute risk of the tunnel route.
Consequently, relatively few tunnels have the ADR categorisation B or C.
• The various European countries have different strategies regarding the acceptability of the dangerous
goods transport through tunnels on their trunk road networks. Some countries are quite restrictive, like
France. Others, like Austria, want to keep this kind of transport on routes at the top of the road network
hierarchy, because their tunnels have high safety standards and the alternative route(s) typically run
through urban areas or on secondary roads across mountainous regions. In Spain, there is a designated
subset of the road network (red de itinerarios para mercancías peligrosas – RIMP) for the transport of
dangerous goods. Sweden has designated primary and secondary DG routes.

Austroads 2024 | page 19

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

5. Examples of National Standardised Risk

Assessment Methods
Typically, the risk assessment and evaluation procedure for dangerous goods transports consists of two
levels:
• At the policy/overview level, there is the procedure of risk assessment and risk evaluation, including the
risk acceptance approach and criteria (addressed in this Section 5).
• The detailed level is the choice of risk model that is applied for the quantitative investigation. In fact, there
are only a limited number of models which find use in practice (addressed in Section 6).

In order to evaluate the ADR category that should be assigned to a tunnel, a (quantitative) risk assessment is
necessary. Experience has shown that submitting a tunnel to a complete examination is an extensive task
and requires a large amount of data. It is time and cost expensive.

To reduce these expenditures, and in view of the large diversity of tunnels (regarding tunnel parameters,
equipment and traffic parameters), some national authorities decided to divide the process into several
steps. The quantitative models discussed in Section 6 form steps or parts of steps required in their
respective national risk assessment methods. The particular procedures, although similar, differ from country
to country. Nevertheless, the main principle is to end the tunnel assessment as soon as it is proven that
sufficient tunnel safety measures have been included such that no restriction on the transport of dangerous
goods is necessary. Alternative routes are evaluated only in a last step.

This is conceptually different to the framework envisaged for Australasia, in which parallel assessments of
the tunnel and surface route would be compared.

In general, acceptance criteria are not harmonised within Europe. In fact, the definition of the acceptability
limits or boundary curves for the F-N diagrams is often a complicated (national) long-term process involving
multiple stakeholders. Adopting a comparative approach with F-N curves for a risk-based comparison of
alternatives is useful but needs to be done carefully, as the interpretation – especially where curves intersect
– is a highly sophisticated task. The complexities of all this make it attractive to apply a ‘deemed to satisfy’
approach as commonly applied in the United States.

The use of F-N curves to judge acceptability against criteria, is also a point of difference with the SFAIRP
approach to risks common in Australasia. If we follow the SFAIRP approach to risk reduction, we need to see
if we could reasonably reduce risk further by using the alternative route for DG, even if it has already been
made ‘acceptable’ against some criterion. European approaches generally are satisfied if the tunnel route
meets the criteria set, whether or not it is theoretically possible to reduce risk further. Only if the tunnel route
does not meet the criteria is the comparison with alternative routes required, to decide which is preferred.
Whether criteria are met or not, the French approach (see Section 5.1.4) allows other parameters to be
considered in the decision (environment, economic, non-DG incidents), alongside DG risk outcomes.

The following sections discuss the approaches in France, Germany and Austria, with some less detailed
comments on the treatment of tunnel DG risk in other countries.

Austroads 2024 | page 20

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

5.1 France
The French method (CETU, 2018) foresees a two-step process for the risk assessment, in which the second
step is again subdivided, as shown in Figure 5.1.

There are only two circumstances in which an assessment can be skipped:

• In the case of a new tunnel shorter than 300 m along a route where there is already a restriction on the
transport of dangerous goods.
• In the case of an old tunnel longer than 300 m, where there is already a restriction of dangerous goods
and it is not planned to change that restriction (CETU 2018).

Figure 5.1: Overview of the French method

Source: CETU (2018).

5.1.1 Step 1

According to the French regulation, before proceeding to a comparative study between the different tunnel
categories, it is advised, to avoid unnecessary work on detailed risk assessment, to check the tunnel’s
intrinsic risk first. PIARC’s DG-QRAM is used for that assessment. The limit value is fixed at 10-3 fatalities
per year (CETU 2018).

Furthermore, it is explicitly stated that, if the risk value is close to the threshold, the results of the assessment
must be accompanied by a sensitivity study (with particular attention given to the DG traffic).

If the risk is too high, the availability of alternative routes has to be checked. In the event that there are no
other options, the HGV traffic cannot be diverted but must continue to use the tunnel route, and so additional
risk mitigation measures need to be investigated. If there are alternatives available to a high-risk tunnel
route, the method proceeds to the next step of selecting and evaluating alternatives (CETU 2018).

Austroads 2024 | page 21

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

5.1.2 Step 2a

Within this step, the project owner/contracting authority, together with the consulting contractor, has to define
the possible alternative routes. It is generally advised to focus on not more than two routes. Then, all
retained routes, including the tunnel route, are examined in-depth. (The model used for the risk assessment
linked to the DG transport is again DG-QRAM.) The main characteristics to consider are:
• The route’s technical characteristics (geometry, equipment, etc.)
• Traffic regulation (speed limits, effective speed, etc.)
• Traffic composition (HGVs, buses, light vehicles, etc.)
• HGV-traffic with regard to its new distribution along the different routes (as a function of ADR’s tunnel
categorisation system)
• The routes’ accident rates
• Operation, supervision and management of the road
• The surrounding areas and their organisational structure
• The societal context (population density, sensitive zones, etc.)
• Environmental context (conservation areas, natural hazards, etc.)
• Economic context (CETU 2018).

5.1.3 Step 2b

This step consists of the comparison of the results obtained from the previous step, provided that there are
four criteria that have to be taken into account as part of a multi-criterion assessment shown at the bottom of
Figure 5.1:
• Mortality risk of “M”-type accidents
“M”-type accidents are HGV accidents including a dangerous goods event. The assessment is done using
DG-QRAM. As it is a comparative approach, the computation results of the simulation of a certain tunnel
category are compared to the results of the computation of the other categories.
• Mortality risk of “C”-type accidents
In analogy to the “M”-type accidents, the risk due to “C”-type accidents is calculated quantitatively. “C”-
type accidents are thereby purely ‘mechanical’ accidents involving a DG HGV but not the dangerous good
itself. In other words, it is the assessment of the effect on the ‘regular’ risk of (lethal) road accidents
resulting from the increase of the HGV traffic on the alternative route and the reduction of that traffic
through the tunnel.
• Vulnerability of the route in regard to the transport of dangerous goods
The assessment of the route’s vulnerability is a qualitative process taking into account a multitude of
parameters such as the presence of sensitive buildings along the route, the occurrence of conservation
areas, the distance of emergency services, noise emissions, etc.
• Economic impact
Lastly, the economic impact of the deviation of the dangerous goods traffic is evaluated. This includes, for
example, the additional costs for the transport company, the loss of income for the tunnel operator,
additional costs for (eventually) necessary environmental protection measures, etc. (CETU 2018).

The most important of these criteria remains the mortality risk for “M”-type accidents (CETU 2018).

It is also noteworthy that the method formally includes partially subjective criteria (route vulnerability)
alongside the ‘economic rationalist’ quantitative criteria.

Austroads 2024 | page 22

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

5.1.4 CETU Booklet

We include here reference to a booklet produced by CETU (Centre d'Etudes des Tunnels, in France)
(CETU 2005) and reviewed in Austroads Research Report AP-R589-19 “Dangerous Goods in Tunnels:
Literature Review”. It is interesting to us because it gives guidance on the significance of difference in risk
between two routes. If the expected fatality rate differs by a factor of 10 or more, that is a ‘significant’
difference, and there is a clear basis for a decision. If the factor is below 3, the difference is ‘insignificant’ and
the relevant authority can make a decision either way, considering other matters. Between those cases,
sensitivity analyses are looked at and, if the factor varies in a range that is less than 10 but not below 3, then
the risk difference is ‘inconclusive’, but may be still significant enough for a decision to be made.

The French method then allows for consideration of other criteria. Where the fatality risk is inconclusive, the
other criteria may come to determine the conclusion. This methodology is described in the next section.

Without knowledge of the CETU work, the authors of this report also previously proposed a factor of 10 as a
point where risk difference becomes significant or conclusive.

Key residual considerations are both the accuracy of quantitative risk assessments, and the significance of
risk differences when the risks being compared are already very small.

5.1.5 Example Evaluation

An example final multi-criterion evaluation is given in Table 5.1. It summarises the results from the individual
quantitative and qualitative assessments that have been undertaken on a French sample tunnel project. The
table format is used as a means of facilitating the risk comparison across the different criteria. The effect on
overall risk of each possible tunnel categorisation is assessed, with the aim of identifying the category most
appropriate to the risks. We don’t discuss the detail of the example project, or the rationale behind the table
entries, but simply provide it to substantiate that the French approach uses such a multicriteria assessment.
It is intended to be illustrative rather than referenced in detail. Further, this is a recent addition to the French
approach.

The example multi-criteria categorisation assessment, shown in Table 5.1, identified that a categorisation as
“A” has significant advantages over a categorisation as “D/E” according to three of the four criteria. However,
for the remaining criterion (‘M’ type accidents, a categorisation as “D/E” has a significant advantage over the
categorisation as “A”.

Table 5.1: Example multi-criteria assessment table for the final evaluation

Tunnel
“M”-type accidents “C”-type accidents Route vulnerability Economic impact
category
Significant difference Significant difference Significant difference
A
EVA < EVDE A < D/E A < D/E
Significant difference Significant difference
B
EVB < EVD/E B < D/E
C
Inconclusive difference
EVD/E ≤ EVB
D/E
Significant difference
EVD/E < EVA

Source: CETU (2018).

Note that ‘<’ should be read as ‘lower risk than’.

Austroads 2024 | page 23

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

The evaluation identified that a “B” categorisation is better than a categorisation as “D/E” according to two of
the four categories, and also a ‘B’ categorisation is not significantly worse than other categorisations across
the other criteria. Indeed for ‘M’ type accidents category ‘B’ is better than category ‘A’.

Evaluation against a “C” categorisation did not identify advantages for any criteria when compared to other
categories.

Blank cells in the table indicate insignificant differences between categories (CETU 2018).

In summary, the assessments indicate that it would be best to either classify the tunnel as either category ‘A’
or category ‘B’, (CETU 2018) but category ‘B’ is preferred given that it performs equal to or better than all
categories across all criteria.

It is noted that the “M”- type and “C”- type assessments are quantitative risk assessments, delivering
Expected Value (EV) as results. EV is a simple way of representing societal risk. In contrast, the evaluation
of the economic impact is based on a monetary assessment of the different DG transportation costs for the
freight forwarder and tunnel operator/owner (e.g. loss of toll). And the assessment of the route vulnerability is
finally a purely qualitative evaluation based on expert judgment; considering ecological aspects (e.g. national
parks, drinking water protection, etc.), cultural aspects (e.g. historic preservation of buildings), life quality of
the residents (e.g. noise pollution, air pollution, etc.), impact on local industries (e.g. would the provisioning of
a local industrial site be affected), and distance to emergency services.

5.2 Germany
The German method stipulates two major steps, themselves subdivided in two minor steps (see Figure 5.2).

Figure 5.2: German method

Source: Baltzer, et al. (2009).

Austroads 2024 | page 24

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

5.2.1 Step 1a

The first step consists of a rough review by means of a checklist. The following key factors are examined:
• Tunnel length < 400 m
• Longitudinal inclination ≤ 2%
• Curvature ≤ 225 degrees turn/km (255 m radius)
• No traffic weaving areas (no feeder and exit roads) in the tunnel or immediately next to the tunnel portals
• A traffic quality better than a certain level (for instance no excessive traffic jam frequency, etc.)
• A truck proportion ≤ 10%
• No indication of an unusually high proportion of dangerous goods transports (for example a chemical
industry in the area would already be a basis for exclusion)
• No building right above or right next to the tunnel (high people exposure)
• The technical and structural safety equipment complies with the applicable guideline (Baltzer, et al. 2009).

As soon as one of these requirements is not satisfied, a quantitative risk assessment is necessary.

5.2.2 Step 1b

The second step consists of an assessment with DG-QRAM (see Section 6.1). The German regulator
regroups DG-QRAM’s scenarios before comparing the expected risk values to the German thresholds as
follows:

Table 5.2: German scenarios and limits for the DG-QRAM evaluation

Effect Allocation of DG-QRAM’s Admissible expected value

(standardised to 1 km) scenarios [deaths/km/y]
All effects 3 to 13 6.2*10-3
Fire impact 4, 5 5.0*10-3
Pressure-/fire impact 7, 8, 9 1.2*10-3
Toxicity 6, 10, 11, 12 4.0*10-4
Pressure 3, 13 1.0*10-6

Source: Baltzer, et al. (2009).

Scenarios 1 (20 MW fire) and 2 (100 MW fire) are not considered, as they are non-DG scenarios. Scenarios
involving the detonation of solid explosives need eventually to be considered in case of a (shallow) tunnel with
relevant constructions (schools, tramway stations, supermarkets, etc.) on the surface (Baltzer, et al. 2009).

If at least one of these absolute target values is exceeded, a deeper analysis is required.

5.2.3 Step 2a

This step corresponds to a quantitative risk analysis according to the German model described in Section
6.2. As already mentioned, the idea behind that is to avoid an overly conservative assessment by increasing
the simulation accuracy and by adapting the key parameters to the German traffic specificities (e.g. ratio of
empty runs, average loaded quantities, etc.).

If the evaluation shows that a restriction is necessary, the tunnel has to be assigned to the corresponding
category according to the ADR and an alternative route has to be found in the next step (2b) (Baltzer, et al. 2009).

Austroads 2024 | page 25

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

5.2.4 Step 2b

Within this last step, an alternative route is chosen first. The selection is done with regard to the following
characteristics:
• Road type (highway, urban street, etc.)
• Number of lanes
• Intersections and junctions
• Accident rate or rather rate of accidents with DG release
• Density of population in the concerned area
• Special buildings such as hospitals, schools, retirement homes, etc.
• The surrounding area’s characteristics such as topography, noise barriers, etc.
• Sensitive facilities (fuel stations, etc.)
• Further relevant aspects in regard to the probability and severity of DG events (Baltzer, et al. 2009).

Once a route has been chosen, it is in respect to these same characteristics that the route is subdivided into
sections. Except for some adaptations in the event trees, the assessment procedure is basically the same as
for the tunnel assessment (Step 2a). So, the four main substances to consider are still gasoline, propane,
chlorine and TNT. The major difference to the tunnel assessment is probably the deepened consideration of
the meteorological influences. Also, attention shall be given to the local conditions (Baltzer, et al. 2009).

The evaluation of the risks is done by means of F-N diagrams in analogy to the tunnel assessment. The
section with the highest risk is the decisive one. In case a threshold value is exceeded, another route must
be tested. If the risk remains too high on all available alternative routes, another mode of transport shall be
chosen. Only if neither an alternative route nor a carrier change is possible, is it allowed to bring the transport
back on the initial road through the tunnel (Baltzer, et al. 2009).

5.3 Austria
The Austrian method (FSV 2016) is a multistage risk assessment process mainly based on DG-QRAM. It is
noteworthy that while the more detailed TuRisMo was developed in Austria and is in use for regular tunnel
risk assessments (FSV 2015), it is not part of the national method for DG transport assessments. The
procedure stipulates three major steps including a subdivision of the second step (see Figure 5.3).

5.3.1 Step 1

Step 1 corresponds to a simplified (pre-) assessment of the tunnel. A special assessment matrix has been
designed for this. The key factors reviewed are:
• Tunnel length [m]
• Single or bidirectional tunnel
• Traffic volume [veh/d]
• HGV ratio [%]
• Type of ventilation (natural ventilation, mechanical ventilation, etc.) (FSV 2016).

If the matrix shows that the risk is at an acceptably low level (expected value < 10-3 fatalities per year) the
assessment ends at this point. Otherwise, Step 2 is necessary.

Austroads 2024 | page 26

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

5.3.2 Step 2a

Step 2 is subdivided in two parts. During the first part (Step 2a) DG-QRAM is used to assess the transport of
dangerous goods through the tunnel as described in Section 6.1. If the computed results (expected values
and/or FN curves) are below the reference values/lines, the assessment is stopped at this point and the
tunnel is classified Category A. Otherwise Step 2b is necessary (FSV 2016).

(DG-QRAM’s scenarios 1 & 2 (20 MW-fire & 100 MW-fire) are not considered as DG scenarios and therefore
not taken into account.)

5.3.3 Step 2b

In Step 2b, mitigation measures are defined, and the tunnel parameters are adapted according to them. Then,
the tunnel, including the new features, is reassessed (via DG-QRAM). If the results allow an assignment to
Category A, the assessment is over and the evaluated mitigation measures must be implemented. Otherwise
further mitigation measures (primarily organisational and operational) need to be tested until the evaluation
results in a Category A tunnel. If all possible and economically reasonable mitigation measures have been
deployed and the risk is still too high, this implies that at least some DG vehicles should not use the tunnel, and
an alternative route, or routes, has to be checked (FSV 2015) & (FSV 2016).

Figure 5.3: Overview of the Austrian method

Source: FSV (2016).

5.3.4 Step 3

Within this last step, at least one alternative route has to be evaluated in regard to the transport of those
dangerous goods whose transport through the tunnel would be prohibited. Of course, the alternative route
has to be generally suited for the HGV traffic (regarding the number of lanes, slope, road width, curve radius,
weight limits, driving restrictions, etc.) (FSV 2016).

Austroads 2024 | page 27

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Once the optimal route has been selected, it has to be proven that the transport of the respective dangerous
goods, in regard to the risks for the road users and the local population, entails a lower overall risk than the
unrestricted transport through the tunnel.

The assessment of the alternative route is done analogously to the tunnel assessment. The route shall be
subdivided into sections with homogenous characteristics such as:
• Permitted speed
• Traffic volume and accident situation
• Special route characteristics (e.g. intersections zones)
• Number of lanes
• Population density
• Population characteristics (age distribution, students, commuters, etc.)
• Buildings with a high occupancy
• Meteorological conditions (FSV 2015).

For the computation, DG-QRAM is applied. The approach is to deviate parts of the dangerous goods traffic –
according to ADR’s tunnel categorisation system – from the tunnel route to the alternative route until the
tunnel requirements are met. Then the overall risk from the tunnel route is compared to the overall risk of the
alternative route.

To be accepted as the better option, according to the Austrian regulation, the risk of a transport on the
alternative route has to be “significantly” lower than the risk of the transport on the tunnel route (FSV 2016).

5.4 Other Countries

While the three sections above describe the national approaches most relevant to the current work, it is
worth mentioning why some others have not been presented in such detail.

In Italy, the European country with the highest number of road tunnels, there is no one specific approach for
dangerous goods risks. Instead, dangerous goods risks are included in the national risk model IRAM, which
addresses dangerous goods risks together with other tunnel risks like collisions and conventional fires. As
these other risks dominate the assessment, dangerous goods risks have little influence on decision making.

In Switzerland, DG-QRAM is applied in the context of the overall national hazardous incidents regulations
(applicable to all industries, not only to the transport sector), which include absolute reference criteria.
However, as part of this process, some elements (e.g. evacuation model) of DG-QRAM were modified to be
less conservative. This serves to emphasise the important difference between comparative risk assessment
and absolute risk assessment, where conservatism cannot be tolerated in the same way.

In other countries, for instance the United States, Great Britain and Ireland, the methodology to be applied is
decided on a case-by-case basis and there are no standard national approaches.

5.5 Discussion of the Various Approaches – with Focus on the

Evaluation of Results
With respect to the evaluation of the results of risk studies there are two fundamentally different approaches,
which are also part of the methods described in the previous chapters:
• An absolute approach, by defining an absolute, quantitative risk level as minimum acceptable risk limit
• A relative approach, by defining a ‘safe’ reference case as a risk limit.

Austroads 2024 | page 28

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

The definition of absolute risk acceptance criteria is a complicated process and needs to be based on a
comprehensive study including sensitivity analysis and expert judgment.

In general, a relative approach as sometimes used in the risk assessment methods demands the definition of
reference tunnel characteristics and hazards to be considered. In essence, a reference tunnel approach is
close to a deemed to satisfy code approach, in that there is implied acceptability of the reference tunnel in
the same way that there is implied or legislated acceptability of codes. Rather than comparison with hard
coded provisions, a tunnel is deemed safe enough if it is sufficiently close in risk characteristics to the
reference project.

An advantage of this approach is, as the same model is applied for the reference tunnel and the tunnel being
studied, the potential influence of inaccuracies and uncertainties in the assessment of frequencies and
consequences is of minor importance. Without any further steps, the raw comparison of two tunnels could
still deliver ambiguous results, in particular when there are slight differences between the reference tunnel
and the investigated tunnel parameters. Thus, the guidelines in some jurisdictions for the transport of
dangerous goods through road tunnels foresee the application of a combination of absolute and relative risk
criteria.

For the present task, on the comparison of DG risks between tunnel and surface routes, we are not so
obviously interested in the absolute criteria and appreciate that the arrangement of the methods to give
relative risk makes them better suited to our comparative task, even though the ‘reference tunnel’ is in fact a
surface route. However, the focus on evaluating absolute risk as accurately as possible is also useful for our
purpose, as, rather than slight differences between our tunnel and the ‘reference’ surface road, there will of
course be dramatic differences. Because of this, accuracy in absolute risk is desirable to control error in the
assessed relative risk.

Unlike the formally recorded French approach, the Austrian and German approaches have in common that
some of the tunnel’s key factors are checked before starting a quantitative risk assessment. We understand
that the French may often do such a check less formally. The complete quantitative assessment of a tunnel
(and eventually the alternative route(s)) is a complex and expensive process. Thus, implementing a process
step for a rough check of a tunnel’s key characteristics could be useful. Especially as these quick analyses
do not require a high level of expert knowledge from the user. It is likely that in Australasia the more detailed
method developed as part of this work would be applied to all proposed tunnels, at least for another decade
or more until sufficient data for a ‘reference tunnel’ is available. Should this be undertaken, even those future
tunnels that would fall within a ’pre-check’ risk threshold would then have a documented ‘reference tunnel’ in
terms of DG risk.

The Austrian and French approaches are mainly based on the DG-QRAM model as offered by PIARC. The
German approach includes the application of DG-QRAM also, but only in a first step. In fact, if the safety
requirements are not met (according to DG-QRAM results), the German regulator stipulates a second
assessment model within the next step. One of the ideas behind this is to apply a more precise and hence
less conservative model in this phase, which is better able to model the effects of relevant tunnel safety
features like tunnel ventilation.

Concerning the last step of all three approaches, namely the assessment of the alternative routes, it is the
French guideline that offers the most detailed description on how to proceed and how to integrate aspects
such as environmental hazards, economic losses, etc. into the assessment. A sample description of this
evaluation process is included in Section 5.1.5

Austroads 2024 | page 29

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

6. Quantitative Risk Assessment Applications

Under the National Approaches
This section describes the detail of the core quantitative risk assessment (QRA) steps in the national
methods mentioned above.

Usually, these are multi-step procedures, trying to achieve an adequate balance between the relevance of
the problems and the level of detail, and associated expenditure, of the risk study.

Risk analysis and evaluation is just one of several different tasks in tunnel safety management, and it will
typically be undertaken concurrently with all the other management or project activities, including design if it
is a new tunnel. So, the risk evaluation hinges not only on the chosen method of risk analysis, but also on
the scope and given circumstances of the risk assessment. As with all modelling, we note that risk models,
even if designed with highest accuracy and implementing realistic data, are not able to predict real events
exactly. Consequently, there will always remain some uncertainties in the results. This aspect must be taken
into account in the risk evaluation process, for instance by first selecting the most suitable risk evaluation
approach and possibly by complementing the assessment process by a sensitivity analysis for the most
relevant influence parameters.

Before describing the different models, we note that, independent from the chosen model, when comparing
an alternative route with the tunnel route, the tunnel route’s risk is not simply the risk linked to the transport of
dangerous goods through the tunnel but the (overall) risk linked to the transport through the tunnel and to the
road ahead of and behind the tunnel. That is; both analysed routes (tunnel and surface alternative) have to
have the same start and endpoint. Referring to Figure 6.1, the relevant section to take into account, if
assessing the tunnel route (blue) in comparison to the alternative 1 (green), lies between the points P1 & P2.
To assess the tunnel route (blue) in comparison to alternative 2 (purple), the relevant section to evaluate lies
between the points P3 & P4. It is also clear that as more surface road is included in the analysed route, the
tunnel’s contribution to the overall risk becomes relatively smaller.

Figure 6.1: Scope for the overarching route assessment

Austroads 2024 | page 30

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

6.1 Dangerous Goods Quantitative Risk Assessment Model

The Dangerous Goods Quantitative Risk Assessment Model (DG-QRAM) for the transport of dangerous
goods was first developed in a joint project of the Organisation for Economic Co-operation and Development
(OECD) and PIARC almost 20 years ago. It is a commercial software, which is widely used and serves often
as a basis for a country-specific method for the risk-based assessment of the transport of dangerous goods
through tunnels or on tunnel routes. The latest version dates from May 2019 and the model is now
compatible with Microsoft Windows versions in use at the time. Currently another upgrading process is under
way, embedded as a task force in PIARC’s Technical Committee 4.4 Tunnels, aiming at some methodical
improvements to the model. Importantly for the present work, DG-QRAM was developed as a tool for
deciding whether to forbid or authorise, partly or totally, the dangerous goods traffic through tunnels. It is
intended to be used for this purpose only. The risk model is not suitable for a general risk assessment study
for road tunnels. There are several reasons for this limitation such as:
• The model does not deliver a complete picture of the overall tunnel risk, as it does not include incidents
(collisions/fires etc) from a range of non-DG transports (i.e. passenger cars).
• The model includes a lot of different DG-scenarios, therefore to reduce complexity the consequence sub-
models in DG-QRAM for each individual scenario are rather coarse. This is particularly relevant to fire
and smoke propagation modelling inside a tunnel. Other modern tunnel risk models (like TuRisMo) offer a
much better and realistic representation of these effects, because they use complex combined and
transient 1D/3D CFD modelling.
• There are limitations in assessing the positive effects of many quite effective tunnel safety measures –
related to ventilation, evacuation, FFFS or organisational measures influencing the timeline of reactive
actions in fire incidents.

The tool’s general structure allows an assessment of the tunnel alone, as well as the assessment of an
open-air route in comparison with the tunnel route. This is a valuable feature, as this allows either a
comparative risk assessment or an absolute risk assessment or both.

Furthermore, the tool allows the user to select between the option of calculating fatalities plus injuries or
fatalities only, and to select between the consideration of road users plus local populations, or road users
only, or local population only. The resulting six review options are given in Table 6.1 (PIARC 2019b).

Table 6.1: DG-QRAM’s different possible approaches

Fatalities only for road users and the local Fatalities plus injuries for road users and the local
population population
Fatalities only for road users Fatalities plus injuries for road users
Fatalities only for local population Fatalities plus injuries for local population
Source: PIARC (2019b).

6.1.1 Scenario and Frequency Analysis

One part of the model consists of a quantitative frequency analysis. DG-QRAM encompasses an analysis of
the sequence of events or occurrences that stem from an initial DG event, to a set of consequence scenarios
(see Table 6.2) including the respective probabilities of occurrence.

Thirteen scenarios representing the hazards most threatening to human life have been defined. Scenarios of
other dangerous goods primarily affecting the environment are not considered, as the main focus is on
human well-being. Within modern Australasian tunnels, the environmental impact of the release of polluting
substances seems negligible, as all are equipped with sumps that capture spillage and either store it for
disposal or transfer it to surface storage.

For every scenario there is a probability implemented in the model, distinguishing between the different
boundary conditions (tunnel or open air, urban area or rural area, etc.) (PIARC 2019b).

Austroads 2024 | page 31

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

The thirteen retained scenarios are shown in Table 6.2. The consequences of scenarios involving those DGs
not appearing in the DG-QRAM scenario list is taken as being similar, or less significant than the
consequence of one already in the list. The chosen substances are representative of all other dangerous
goods. The exceptions are large quantities of explosives, and radioactive materials, which are not addressed
by DG-QRAM and are also regulated differently. Depending on the detonation assumption (rapid or delayed)
more moderate explosive loads could be considered through BLEVEs or vapour cloud explosions of the
appropriate size. It may not be appropriate to use those scenarios to model military quantities of explosives.

The general procedure to obtain a scenario’s probability of occurrence is to combine the route’s HGV
incident rate (or a national default value) with the traffic and the DG HGV ratio among that traffic along the
(tunnel) road, as well as the conditional probability of occurrence of the scenario. Those specific probabilities
are preset in the software and are a function of the dangerous goods traffic’s composition. The probabilities
used are based on a comprehensive (400-page) study by the University of Waterloo (Canada), which was an
annex to the original work documentation. The probability can be adjusted by aggravating or diminishing
factors related to the tunnel’s or/and the route’s characteristics, such as being an urban or rural road, the
presence of exit or on-ramps, the existence of crossroads, etc. For the tunnel, it is of course important to
know whether it is single tube or twin tube.

A more detailed description of the model is given in the DG-QRAM reference manual (PIARC 2019a).

Table 6.2: Thirteen scenarios considered in DG-QRAM

Scenario number Description

1 20 MW heavy goods vehicle fire
2 100 MW heavy goods vehicle fire
3 Boiling liquid expanding vapour explosion (BLEVE) of a liquefied petroleum gas (LPG) cylinder
4 Motor spirit pool fire
5 Vapour cloud explosion (VCE) of motor spirit
6 Chlorine release
7 BLEVE of LPG in bulk
8 VCE of LPG in bulk
9 Torch fire of LPG in bulk
10 Ammonia release
11 Acrolein release
12 Acrolein in cylinder release
13 BLEVE of liquefied refrigerated carbon dioxide

Source: PIARC (2019b).

6.1.2 Consequence Analysis

The other part of the DG-QRAM model consists of a quantitative consequence analysis. As already
mentioned, the model can calculate the consequences (fatalities and injuries) of the different representative
(dangerous goods) scenarios for road users and the population located close to the route. The extent of
those consequences is a function of various parameters:
• Quantity of the substance(s) involved as well as its (their) physical and chemical properties
• Type of release
• Physiological effects on the human body as a function of the exposure time
• Meteorological (wind, portal pressure) influences
• The number of people affected
• The route’s/tunnel’s characteristics as well as the implemented safety measures.

Austroads 2024 | page 32

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

As seen in Table 6.2, the model includes eleven scenarios which are intended to represent the most
important hazards from dangerous goods (namely: (large) explosions with or without fire, liberation of toxic
substances and fires), whereas Scenarios 1 and 2 in Table 6.2 represent non-DG HGV fires (without
involvement of dangerous goods) (PIARC 2019a). The non-DG scenarios were implemented in the model for
comparison reasons only and are not typically used for the assessment of DG-transport risks.

Concerning the type of release and the scenarios’ effects, a rough description is given in Table 6.3. For
tunnels, the more relevant hazard of the scenarios 1, 2 and 4 is the smoke inhalation rather than the burn
injuries. And for the scenarios 3, 5, 7 and 8 it is the temperature rather than the pressure effect.

For more detailed information about those effects and the models used to calculate the damage caused
(FNAP, PHAST 4.2, SHIELD, etc.), please refer to the ‘Applicability Report’ or DG-QRAM’s Reference
Manual (PIARC 2019a).

Table 6.3: DG-QRAM scenarios and their effects

Scenario Quantity Size of Liberation

Description Effects
No. involved breach [mm] rate [kg/s]
HGV fire, 20 MW (unladen or laden
1 - - - Thermal & smoke
with little combustible material)
HGV fire, 100 MW (loaded with
2 - - - Thermal & smoke
combustible material)
3 BLEVE of LPG in cylinder 50 kg - - Thermal & pressure
4 Motor spirit pool fire 28 t 100 20.6 Thermal & smoke
5 VCE of motor spirit 28 t 100 20.6 Thermal & pressure
6 Chlorine release 20 t 50 45 Toxic gas
7 BLEVE of LPG in bulk 18 t - - Thermal & pressure
8 VCE of LPG in bulk 18 t 50 36 Thermal & pressure
9 Torch fire of LPG in bulk 18 t 50 36 Thermal & pressure
10 Ammonia release 20 t 50 36 Toxic gas
11 Acrolein in bulk release 25 t 100 24.8 Toxic liquid
12 Acrolein in cylinder release 100 litres 4 0.02 Toxic liquid
13 BLEVE of liquefied refrigerated CO2 20 t - - Pressure

Source: PIARC (2019a) and PIARC (2019b).

Of course, there will be different consequences for an incident in the open air compared to the ‘same’
incident within a tunnel. While in a tunnel the affected people are mostly the tunnel users and possibly
people near the downstream portal, in an open-air incident the people near the incident site (‘local
population’) might be affected too. The consequences could vary greatly with the occupancy of land beside
the surface route. Also, the sequence of events might not be the same inside and outside the tunnel. For
example, the geometry and propagation of a ‘fireball’ is not the same in the open air as in a confined tunnel
environment. The dilution and propagation of smoke is completely different too, as are the site’s accessibility,
and the wind’s influence, etc.

DG-QRAM allows such differentiation between open air and tunnel sections. Depending on the
requirements, the user can choose between a 2D or a simplified 1D tool in the open. An extra 1D tool is
available to determine the physical and physiological consequences of incident scenarios within the
modelled tunnel (PIARC 2019a). In the case of a tunnel, additional information is necessary such as: the
tunnel length, its inclination and cross section as well as safety measures like emergency exits, ventilation
systems, activation times, etc., as they have a significant influence on the expected number of fatalities.

Austroads 2024 | page 33

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

One more decisive parameter influencing the extent of the harm is the traffic. The traffic and its composition
(e.g. permitted speed and proportion of buses) not only have an influence on the event’s frequency, as
discussed in the previous chapter, but also have an influence on the severity of an incident. The heavier the
traffic flow is, the more vehicles will be in proximity to the incident location and the more people will be
affected. Hence, another feature of DG-QRAM is that the model allows consideration of up to three different
periods of time (during a day). As a result, it is possible to look separately at time periods with comparatively
low traffic and those with high traffic. Similarly, the time-dependent composition and geographical distribution
of the local population and/or eventual time-dependent driving bans can also be taken into account.

Finally, we repeat that when comparing an alternative route to the tunnel, the tunnel route includes all of the
road alignment (tunnel and surface) from the routes’ nodal point before the tunnel to the route’s nodal point
beyond the tunnel (see Figure 6.1). The eventual consequences to consider on the tunnel route are therefore
not only the possible incidents within the tunnel, but also those in the open air between the tunnel and the
nodal points.

For the sake of completeness, we mention that DG-QRAM is additionally able to calculate the reinstatement
costs for the tunnel equipment as a percentage of the estimated capital costs of building a new tunnel of
similar specification (PIARC 2019a).

However, this feature is rarely used for evaluations, as the monetary aspect is often considered less
important than the safety aspects and potential adverse publicity. Costs also vary significantly with country
and escalation over time, rendering the features application questionable.

6.1.3 Output

The model’s structure enables its application in two steps. In the first step, the route that includes the tunnel
is assessed. The outcome corresponds to the statistically expected yearly number of victims (fatalities and
optionally injuries). In the second step, the risk of the dangerous goods transport over one or more
alternative route(s) is assessed. In Europe, the various options in assigning an ADR tunnel categorisation
need to be evaluated. The results can then be compared to each other (PIARC 2019b).

By combining the scenarios’ frequencies and consequences within the desired scope of risk (see Table 6.1),
the model is able to deliver the expected values (EV) and F-N (frequency-number) curves (see Figure 6.2)
for the societal risk for every individual scenario as well as for all scenarios together (PIARC 2019b).

An F-N diagram is a double-logarithmic plot of the statistically expected fatalities (plus injuries) (on the
abscissa) and the cumulative incident frequency over one year (on the ordinate). In other words, it shows the
probability of incidents with more than a certain number of statistically expected fatalities (plus injuries) within
a year on (and along) the examined route.

Austroads 2024 | page 34

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Figure 6.2: Example of a FN diagram (absolute value approach)

Source: FSV (2016).

If the risk assessment method foresees, or a regulation requires, an absolute criteria approach, a boundary
line can be added (orange line in Figure 6.2), so that it becomes easily visible if the overall risk or the risks of
one or more single scenarios are too high to be acceptable (crossing above the boundary line). In contrast to
the absolute risk approach, in a comparative approach the aim is to compare the (overall) risks of two or
more routes or to compare one and the same route but with different parameters (for example added safety
measures). In that case, the evaluation is done by contrasting the respective lines (see Figure 6.3).

Generally, an F-N curve shows the cumulative annual frequency for scenarios with different numbers of
fatalities in ascending order, starting from the frequency at which at least one fatality is statistically expected,
and ending at the frequency considered as sufficiently low to be negligible (in Figure 6.3 this value is 10-9).
Particularly if the F-N curves overlap, as is shown in Figure 6.3, the risk expectation value (EV) can be
calculated to differentiate between the alternatives.

The risk expectation value (EV) represents the sum of all single risk values statistically expected for the
respective (individual) scenarios:
𝐸𝐸𝐸𝐸 = ∑𝑖𝑖 𝑟𝑟𝑖𝑖 ,

with the individual risk value 𝑟𝑟𝑖𝑖 being the product of the annual scenario frequency 𝑓𝑓𝑖𝑖 and the average
scenario fatality number 𝑠𝑠𝑖𝑖 .

Austroads 2024 | page 35

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Figure 6.3: Example of an F-N diagram – comparative approach

Source: PIARC (2019a).

As mentioned above, independent of the output form (F-N curve or risk expectation value) the risk can be
assessed either by an absolute or by a relative approach.
• In the absolute assessment approach, the risk expectation value or the FN curve is compared to
predefined risk thresholds (tolerable risk expectation value or an F-N acceptance curve).
• In a relative risk assessment approach, the risk model is applied to a national standard tunnel, which is
fully in line with all regulations. The resulting risk profile of this reference tunnel serves as a reference risk
profile which must not be exceeded by the risk value / FN curve of the tunnel under investigation.
• For the present purposes, there will be no standard reference, with the comparison to be made between
the tunnel route and the alternative surface route.

6.2 The ‘German Model’

The German method (Baltzer, et al. 2009) is a complex methodology for the categorisation of road tunnels
referring to the ADR tunnel regulation, elaborated and documented in a research project in 2007. It includes a
quantitative risk assessment method as well as a multistage application procedure (documented in Section 5.2).
The main motivation for its development was to define a framework for deciding to forbid or authorise – partially or
totally – the dangerous goods transport through tunnels (primarily) and the alternative route(s).

The ‘German model’ (Stage 2 of the method described in Section 5.2) was designed to fit the dangerous
goods transport conditions in Germany and defines the parameters and thresholds to use. So, it is not an
existing (commercial) software but a description of the required set-up for any model development.

6.2.1 Scenario and Frequency Analysis

The scenario analysis in the ‘German model’ is based on an event tree approach including the following
branch points: released quantity, manner of release, immediate or delayed ignition, event location, day or
nighttime, traffic conditions, detection, road closure, activation of (emergency) ventilation, availability of
additional safety systems, increased damage extent, and external rescue.

Austroads 2024 | page 36

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

In contrast to DG-QRAM, the German model retains only four representative main substances. This is most
likely the result of a judgement that the four substances are appropriate proxies for the substances that
generate most of the risk. It appears to be a means of simplifying the method while minimising compromise
on accuracy. The four substances are:
• Motor spirit (DG-class 3; liquid and easily inflammable)
• Propane (DG-class 2; gaseous and easily inflammable)
• Chlorine (DG-class 2; gaseous and toxic)
• Trinitrotoluene (DG-class 1; solid and explosive) (Baltzer, et al. 2009).

Aligned to ADR’s main threats in regard to the transport of dangerous goods (very large explosion, large
explosion, large toxic release and large fire), this results in a total of eight representative scenarios; two per
main substance, as per Table 6.4.

Table 6.4: Overview of the studied scenarios, also noting the relevant ADR tunnel category

Considered Represented Tunnel

Main Scenario Released
Type of release propagation and main threat cat.
substance No. quantities
effects (ADR) (ADR)
Motor spirit • Immediate ignition
Tank transport; and pool fire
1 spontaneous or continuous 15 t • Pool formation and Large fire D
release of 20 m³ delayed ignition
• Smoke spread
• Immediate ignition
Transport in cylinders; and pool fire
2 Spontaneous or continuous 8t • Pool formation and - E
release of 10 m³ delayed ignition
• Smoke spread
Propane • Immediate ignition
Tank transport; and fireball/BLEVE Very large
3 Spontaneous release (400 kg/s) 12 t B
• Delayed ignition and explosion
or continuous release (30 kg/s)
VCE and/or jet fire
• Immediate ignition
Transport in cylinders; and fireball/BLEVE
4 spontaneous release (400 kg/s) 1t Large fire D
or continuous release (30 kg/s) • Delayed ignition and
VCE and/or jet fire
Chlorine • Heavy gas
Tank transport; Large toxic
5 4t propagation and C
spontaneous release of 4 t release
toxic effects
Transport in cylinders (total • Heavy gas
6 quantity of <50 kg); 50 kg propagation and - E
spontaneous release toxic effects
TNT Transport of 1000 kg TNT-
equivalent. After an accident the Very large
7 1t • Delayed detonation B
transport vehicle catches fire, explosion
triggering the explosion.
Transport of 100 kg TNT-
equivalent. After an accident the Large
8 100 kg • Delayed detonation C
transport vehicle catches a fire explosion
triggering the explosion

Source: Baltzer, et al. (2009).

Austroads 2024 | page 37

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

The German method offers national default values for accident rates (including the release of a dangerous
good), with a differentiation being made by road type (urban road, rural road, highway, and freeway) (Baltzer,
et al. 2009).

For the assessment of surface road sections, some additional factors have to be taken into account, such as
people along the route, intersections, special infrastructure (hospitals, schools, fuel stations, etc.) and
environmental characteristics (topography, meteorology, etc.) (Baltzer, et al. 2009).

6.2.2 Consequence Analysis

The German guideline provides only a general description of the requirements for the consequence analysis
and does not call for a specific model.

For an event within a tunnel, it foresees the application of high-resolution computational models, such as:
• flow and dispersion models
• impact models
• escape/evacuation models
• and eventually traffic flow models (Baltzer, et al. 2009).

For surface roads, the consequence analysis is done analogously to the tunnel. Additionally, some parameters,
as for example the local population (and its distribution) along the route or the protective effect of buildings,
have to be considered also. It is recommended to subdivide the route into sections. The section with the
highest risk determines the overall assessment of the alternative route, as the German methodology assumes
that the overall route can be adequately represented by the highest-risk section (Baltzer, et al. 2009).

The risk analysis and evaluation focus primarily on harm to people. Experience has shown that, in regard to
the transport of dangerous goods, material and environmental damage are of minor relevance. Thus, the
applied indicator to quantify the harm is fatalities (Baltzer, et al. 2009).

6.2.3 Output

The methodology for Step 2 of the German method foresees a risk model based on a quantitative
assessment of the tunnel user’s societal risk. The output encompasses F-N curves that are to be compared
to a given threshold. If the calculated risk exceeds the limits, the transport of dangerous goods through the
evaluated tunnel has to be restricted using ADR’s tunnel categorisation system. The selection of the right
category is facilitated as the main substances used in the assessment are aligned with ADR’s main threats
(see Table 6.4) (Baltzer, et al. 2009).

6.3 TuRisMo
The Austrian Tunnel Risk Model TuRisMo was one of the first (probability-based) risk models specifically
developed for general risk assessment studies of road tunnels on the basis of the EC-Directive
2004/54/EC (European Union 2004). The first version, developed by a working group of the Austrian Society
for Research on Road, Rail and Transport, was published in 2008 in the Austrian Tunnelling Guideline (RVS
09.03.11). An enhanced version (TuRisMo 2), which is able to address new and more complex problems,
particularly regarding the fire risk, was published in 2015 (FSV 2015). Additionally, an adapted version
(‘TuRisMo-DG’) fulfilling the requirements of the ‘German model’ description in the method’s Step 2a has
been generated.

A package of integrated software tools is available, which allows an efficient application (in tunnels only).

Austroads 2024 | page 38

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

TuRisMo(-DG) is a system-based quantitative risk model in accordance with PIARC’s definition, comprising a
range of methodical elements to analyse the whole tunnel system in an integrated manner. The model
addresses the personal risk of the individual tunnel users and hence calculates the expected risk values for the
groups of persons within the tunnel (societal risk). The risk value refers to the tunnel as a whole, reflecting
issues such as the collective risk of all tunnel users and the expected average number of fatalities per year.
The respective shares of risk due to mechanical effects, fires and hazardous goods are shown separately.

Typically, TuRisMo is applied in a comparative approach, in which the risk analysis is performed twice. First,
it is undertaken for the (planned) tunnel under investigation. Then, parameters describing the effects of
potential mitigation measure(s) are adapted and the risk calculation is repeated. The effectiveness of the
proposed mitigation measures is evaluated by comparing the two outcomes.

6.3.1 Scenario and Frequency Analysis

TuRisMo’s scenarios for non-DG incidents are the following:

• Breakdown or malfunction of a vehicle, causing a fire
• Breakdown or malfunction of a vehicle, causing a collision (with or without fire as a follow-up event)
• Single-vehicle collision (with or without fire as follow-up event)
• Collision between vehicles driving in the same direction (with or without fire as follow-up event)
• Head-on collision (with or without a fire as follow-up event) (not applicable to major tunnels).

The frequency of pre-defined damage scenarios is calculated by means of an event tree analysis. Starting
from an initial event (for which the frequency is known) alternative chains of events leading to different
consequence scenarios are developed step by step (branches of the event tree). These consequence
scenarios differ with respect to the scenario type, involvement of vehicle types, damage effects, etc. By
quantifying the event tree branch probabilities (initial event as well as bifurcations of the individual branches),
it is possible to estimate the frequency of each follow-up scenario.

In the adapted version ‘TuRisMo-DG’, the initial incident types have been substituted by specific DG-accident
scenarios. Those specific DG-accident scenarios are:
• Pool gasoline fires
• LPG-BLEVE scenarios
• VCE scenarios
• Explosions
• Release of toxic gases.

The incident frequencies can be adapted with reference to national accident rates.

Austroads 2024 | page 39

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

6.3.2 Consequence Analysis

Mechanical consequences

The method of determining the sequence of each follow-up scenario depends on the nature of the incident.
For collisions with exclusively mechanical consequences, the calculation is based on values derived from a
database on collisions with casualties in road tunnels. The collision consequence data was originally based
on a study prepared by the Austrian “Kuratorium für Verkehrssicherheit” (which collects all traffic safety
statistics in Austria) using incident data from Austrian motorway tunnels between 1999 and 2004. With the
implementation of the EC directive 2004/54/EC, a new much more accurate data collection regime was
established by the Austrian freeway operator (incl. tunnels) ASFINAG. During the update of TuRisMo in
2015, all the collision incident data were reviewed using the new data (covering the years around 2006 –
2012 or 2013). The consequence data were quite stable and so were not modified. However, the incident
rates were significantly lower and were therefore adapted.

Consequences due to tunnel fires (DG and non-DG)

A consequence model combining a smoke propagation model with an evacuation model is applied for
collisions, vehicle breakdowns or DG incidents resulting in a fire.

The actual fire risk model (TuRisMo 2) comprises the following sub-models:
• Transient, one-dimensional air flow simulation
This sub model gives the development of the longitudinal airflow in the tunnel for a set of influencing
parameters. As this can be performed in a very efficient way, a large number of cases can be handled,
resulting in a range of possible flow cases.
• Three dimensional CFD simulation
Based on the flow fields calculated in the one-dimensional simulation process, a (smaller) number of
representative scenarios is derived (and projection coefficients are set). These are simulated in a three-
dimensional CFD environment to deduce concentrations of smoke and toxic gases in the vicinity of the
fire location.
• Accumulation based evacuation simulation
The calculated gas concentrations and visibilities are then used as input for an evacuation simulation. In
this sub model, an accumulation-based intoxication model is used to determine the maximum distance a
person can cover after starting to evacuate from a given location at a given time. Comparisons of this
distance with the distances to the various emergency exits enable risk values to be determined for each
possible fire zone.
This distance, combined with the representative configuration of emergency exits calculated from their
given location, results in identification of ‘perilous’ zones, from which egress may not be successful.
• Exposure simulation and data projection
Projecting these life-threatening zones on the zones occupied by congested traffic (exposure) gives the
resulting risk for the assessed tunnel system. This projection is performed repetitively for the full number
of scenarios.

Austroads 2024 | page 40

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Figure 6.4: The transient 1D/3D smoke propagation model

Consequences due to DG-incidents

As for ordinary tunnel fires, an FDS CFD model is used for VCE and toxic gas incident scenarios. This is
done to determine the combustible mass within the explosive concentration range and toxin concentration
distributions, respectively.

In case of BLEVE incidents, the total fireball size, as well as the resulting overpressure, is calculated based
on the total combustible mass.

6.4 Comparison of DG-QRAM, the German Model, and TuRisMo-DG

The first difference is in the commercial status of the models. DG-QRAM is a commercial software based on
OECD/PIARC’s methodology. In contrast, the ‘German model’ encompasses only a description of the
methodology to follow and draws the framework (requirements and parameters) for the model development.
For the diverse sub-models (smoke propagation, evacuation simulation, etc.), different commercial software
tools can be used or have to be developed. (The many sub-models, plus the coordination of the different
tools generally requires a very large number of man-hours of skilled personnel.)

The main differences between DG-QRAM and the German model:

• In DG-QRAM, explosion effects are represented via a CO2 scenario, whereas the ‘German model’
stipulates TNT.
• The ‘German model’ foresees eight scenarios (in fact 4 basic scenarios with 2 different magnitudes for
each scenario) instead of thirteen, enabling a more precise consequence modelling of a more limited
representation of different DG hazards.
• The assumed quantities of released DG substances are different.
• Within the tunnel, the German methodology requests the application of a 3D dispersion model, enabling a
better indication of the effects of mitigation measures for the tunnel.
• The ‘German model’ assumes that the evaluation of the entire alternative route is defined by the risk of
the most endangered section of that route (because it is not realistic to do this detailed level of modelling
for a whole transport route).

Austroads 2024 | page 41

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

According to the German guideline (Baltzer, et al. 2009), the main reason for the German regulator to
stipulate a model deviating from DG-QRAM is that DG-QRAM seems to be too conservative (e.g. released
quantities). Regarding the escape model for example, parameters such as the distance between the
emergency exits, the cross-section, the transverse slope and the number of lanes are deemed to be
insufficiently considered/weighted (Baltzer, et al. 2009). The inference is that the German regulator believes
that a more sophisticated consideration of these items would lower the overall risk attributable to a tunnel.

It follows that when choosing to apply an absolute risk assessment, the coarserer the model is, the more
conservative the parameters have to be, with the consequence that risk thresholds are exceeded more
readily.

Some shortcomings of DG-QRAM will be addressed in the next functionality upgrade. A project to upgrade
DG-QRAM is underway via a PIARC working group (as explained in Section 5.1). Enhancements may be
made to the user interface, driving speed influence, and the egress model, as examples, with the precise
scope to be settled along with the budget.

TuRisMo is also a model put into practice and available as an integrated software tool (latest version
TuRisMo 2). With the corresponding adaptations TuRisMo-DG meets the ‘German model’ requirements and
can be used for the tunnel assessment (Step 2a) of the German method (see Section 5.2.3).

Furthermore, it is a well-developed tool, enabling the comparison of potential mitigation measures. Due to
the level of detail (regarding the 3D-simulation, evacuation models, etc.), TuRisMo(-DG) allows a much more
precise evaluation of fire scenarios in comparison to DG-QRAM. This is of importance as the biggest
contributors to risk from the release of dangerous goods are generally the (pool) fire scenarios.

The second advantage of TuRisMo(-DG) is the relative approach. Indeed, as already described, by
comparing the tunnel to a reference tunnel, both designed according to the same methodology, the
(eventual) modelling fuzziness becomes insignificant to some extent. That is; errors that affect both the
assessed tunnel and the reference project in the same way have no net effect on the risk comparison.

The reasons that TuRisMo(-DG) is not recommended as the principal tool for our present purposes is the
complexity, but more significantly, the inability to address the alternative surface routes appropriately.
However, it can be useful to assess the influence of mitigation measures that are not representable by other
simpler tools such as DG-QRAM.

Austroads 2024 | page 42

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

7. US Approach
We acknowledge contributions to the following information by Dr Matthew Bilson, an underground ventilation
and fire colleague from WSP formerly based in New York, and Gary English, formerly Deputy Chief Fire
Marshal of the Seattle Fire Department.

US federal regulations require a risk assessment for DG in tunnels, but specifically do not even suggest a
recommended risk model. This allows and forces the states to select their own model. The US lack of
centralised DG regulations and therefore adoption of a preferred national risk model, is a direct result of the
individual states requirements superseding federal requirements. The presumption is that all US tunnels
were required to use a version of the “Highway Routing of Hazardous Materials – Guidelines for Applying
Criteria”, Publication No. FHWA-HI-97-003 (HRHM) or its precursor. Note that Dangerous Goods are called
Hazardous Materials (HM) in the US.

Local jurisdictions (counties, cities) can also make rules around DG more restrictive, but they cannot be less
restrictive than required by the state.

In general, full quantitative risk assessment (QRA) for fires and dangerous goods in tunnels or surface routes
is hardly used in the US. It is conventional to comply with codes and standards to fulfil the mandatory
requirements, as they correspond to an assumed level of risk. These are commonly used as the benchmark
for setting the initial bar. Deviations from such requirements must be justified and usually measured against
that benchmark (Coles 2019).

While there is no industry approach, standard, or risk assessment method available, there are some
recommendations, papers or reports related to risk assessments applied on specific projects and design
objectives: (Coles 2019), (Flint, Woodburn, Kotsovinos, & Panev 2019), (Bilson, et al. 2018) & (Bilson &
Purchase 2014).

If risk is used, it is generally applied to compare two alternatives, or it only looks at likelihood and does not
cover consequences. In the past, only probability assessment was used to make a very qualitative decision
whether a scenario might be included or excluded in a risk assessment, based on a probability of
occurrence. A brief overview about risk related documents and recommendations in US is as follows.

The US Department of Transport (DOT) periodically publishes a Value of a Statistical Life (VSL), which is
defined as the additional cost that society considers reasonable expenditure for improvements in safety (that
is, reductions in risk) that, in the aggregate, reduce the expected number of fatalities by one (DOT 2016).
This VSL represents the valuation of reduction in risk rather than the valuation of life itself. That may be a
semantic point created because ‘value of life’ is a difficult thing to express. However, the VSL is not
commonly used in practice.

The AASHTO (a non-profit, nonpartisan association representing highway and transportation departments in
the US) recommended guidelines for emergency ventilation smoke control in roadway tunnels. (Maevski
2016) gives some information about risk assessment and evaluates the fire safety risk for different type of
tunnels depending on tunnel length and usage as well as providing general mitigation means and strategies.
However, AASHTO do not recommend that the document is used as the basis for decision making. Also, the
distinction between risk and hazard are not always clearly defined and it does not cover a risk assessment
for dangerous goods.

The American Public Transport Association (APTA) published a report (APTA 2001) that loosely advocated
risk assessment, but the document is mainly related to passenger railways and provides guidance for making
determinations as to which categories and levels of fire safety risk may be acceptable and unacceptable for
purposes of regulatory compliance.

Austroads 2024 | page 43

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

(Bilson & Purchase 2014) presents a risk-based methodology and acceptance framework to both help
stakeholders deal with residual uncertainty of a safety system (e.g. fixed fire-fighting systems, FFFSs), and
to capitalise on the benefits offered by installation of safety systems (e.g. FFFS). That paper demonstrates a
risk-based approach which is only to be applied on FFFSs in road tunnels. It has not yet made any real
impact in terms of application to a project and decision making in the US.

Other papers and articles consider risk assessment but they are not related to a quantitative approach in the
classical sense (Coles 2019), (Flint, Woodburn, Kotsovinos, & Panev 2019) & (Bilson, et al. 2018).

To sum up, QRA does not get very much use in the US, since codes and standards correspond to a risk
level that is inherently deemed acceptable, and there is a general mindset that codes are both necessary
and sufficient. There are some reports, articles and recommendations for risk and probability assessments,
but they are not commonly used in practice. A general standard for a QRA method or model does not exist in
the US for either dangerous goods or for fire risk in tunnels or surface routes, with PIARC’s DG-QRAM being
used in the one detailed QRA that we know of (Parson Brinckerhoff Quade & Douglas 2016).

Austroads 2024 | page 44

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

8. Prior Australasian Risk Assessments

Prior to the current project, Australasia has not had a standardised DG risk assessment method.

It is a feature of the risk area generally that by choice of method and inputs, a range of results can be
generated. For example, economic rationalism and the value of life can generally be used to avoid taking risk
reduction measures. Multi-criteria assessment can be used more generally to provide documentary backing
to any course of action. It is not generally the case that risk assessments are manipulated to achieve
particular outcomes. A standardised method will ensure that assessments are not subject to the variances
that might occur if individual projects develop their own assessment methods.

Most assessments, whether for DG or fire generally, are done with good intention. All serious assessments
strike the same issue; that it is a difficult thing to do well and meaningfully. The intention is that a
standardised method will reduce the difficulty.

We have been provided with and have reviewed risk assessments for several specific Australasian projects.
We are aware of others that cannot be specifically referenced as the information is tightly held. Of the reports
that are available to the authors, we refer to one that had the depth of analysis sufficient for a comparative
QRA. That report is discussed briefly here.

The transport of DG was looked at for the M5 East tunnel in 1995, as part of the planning and design. The
report (Parsons Brinckerhoff Quade & Douglas, 1995) noted a lack of detail in prior work looking at DG risk in
tunnels, and the paucity of data with which to make assessments. The study compared the risk in tunnels
against that on alternative surface routes, and emphasised the sensitivity of the result to values and
assumptions which might tend to favour the open air or the tunnel cases. The method applied was a full
quantitative method, considering frequency and consequences for representative DG loads, reaching distinct
conclusions on the preferred route for each DG class. The spectrum of DG loads was simplified in a manner
similar to that done in DG-QRAM.

The likelihood and consequence analysis appears to have been done in some detail, although the specifics
of the consequence analysis are not given. The methodology of assessing the alternative route
consequences is not readily deduced.

While suggesting an outcome of the comparative assessment for DG classes, the report also calls for the
development of risk acceptability criteria. That is; there is confusion as to whether the decision should be
SFAIRP-based or assessed on risk acceptability against some benchmark. From the present work, future
decisions may be informed by but not solely based on either, with the risk numbers considered holistically
with other risk inputs. This is a key point to be resolved in the overall approach.

Austroads 2024 | page 45

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

9. Conclusion
Currently there is no standardised risk assessment approach in Australia or New Zealand. Experience shows
that this lack of a uniform methodology and an associated validated assessment model has led to a certain
grade of arbitrariness in the evaluation of the permissibility of the transport of dangerous goods through
tunnels.

A glance, on how other developed countries face this issue, reveals that they still face (e.g. United States) or
faced (mostly European countries) the same or a similar problem. In fact, in order to bring more structure into
their evaluations, countries like Austria, Germany, France, Italy, the Netherlands, Spain and Switzerland
developed national methodologies describing an official evaluation procedure according to their needs.

While those methodologies may at first appear quite different, when reviewed in detail they have much in
common:
• they all have as their basis a quantitative risk assessment
• they are all based on the ADR
• most of the methods include a multi-stage approach
• most rely on DG-QRAM.

This DG-QRAM (dangerous goods quantitative risk assessment model) has in fact been developed
specifically for the needs of a QRA for the transport of dangerous goods. It has been developed in an
international context (PIARC/EU/OECD) and encompasses the statistical background of various countries.
Further, it enables “direct” comparison of the tunnel route and an alternative route. As the categorisation
system of dangerous goods in the ADR is the same as in Australia, a closer examination of DG-QRAM in the
Australasian context seems reasonable.

Independent of the assessment model, be it DG-QRAM, another tool or a combination of both, it is

necessary to have a (national) method encompassing the model. A clearly defined application framework will
help prevent an (accidental) misuse of the model, and only a well-thought-out evaluation process is able to
incorporate model results in the wider field of decision-relevant factors.

There is no national standardised risk assessment method in the US, and the detailed studies available from
the US and Australia either used DG-QRAM or followed a similar approach to DG-QRAM. The conclusion is
focused on the national standardised methods being applied in France, Austria and Germany, which use or
permit the use of DG-QRAM.

It is a core purpose of this work to conclude which of the available methods is the most appropriate for direct
application, or adaptation for application, in Australia and New Zealand for the purpose of comparing DG risk
in a tunnel with that on the alternative surface route. We recommend that DG-QRAM is the preferred base
method. It has been developed specifically for dangerous goods and is suited to the comparison with surface
routes. It is well-proven, internationally applied software and is the only one that can be applied, without
modification, to both the tunnel and alternative routes. Importantly, it is not overly complex, and so less likely
to be much more complex than can be justified by available input data.

In comparison to DG-QRAM, the ‘German model’ is very detailed, requiring extensive analysis, making it
perhaps more accurate, but much more complicated and time-consuming to use if indeed stages beyond the
initial screening are applied.

Austroads 2024 | page 46

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

TuRisMo is also much more detailed and hence more accurate but is very complicated to use and its
application is limited to the tunnel and so does not provide the required comparative assessment. Neither
TuRisMo nor the German method offer benefits for the base case comparative assessment which justify the
extra complexity in use, or the extra work in creating a solid comparison with a surface route. Turismo may
be useful in modifying the tunnel route risk following design modifications not addressed well in DG-QRAM
(e.g. FFFS).

In addition to building a method for Australasia based on DG-QRAM, it could be worthwhile to think about a
first ‘screening step’ as is included in both the Austrian and German methods, as a shortcut for cases where
decisions might be possible without further study. It is not yet clear how that would work, but it could facilitate
project development, and so we will keep it in mind through the case study tunnel assessment and see if
such a screening step could be generated. As noted earlier, it may follow the same method but using
nominal inputs to accelerate the screening.

Finally, it must be remembered that governments must consider many aspects, such as social and
community benefit, risks to environment, including water supplies etc. Within that broader context, the fine
details of a very detailed model may lose impact. Depending on the legal context applying to each road
authority, the distinction between acceptable risk and the SFAIRP approach for model outputs may lose
relevance, as the overall method outcome cannot be assessed one dimensionally.

Austroads 2024 | page 47

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

References
APTA (2001) Recommended Practice for Fire Safety Analysis of Existing Passenger Rail Equipment, Edited
3-22-04. Washington DC: American Public Transportation Association - Commottee 5 APTA-PR-PS-RP-
005-00. <[Link]
[Link]>.

Austroads (2024a) Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on
Alternative Routes: Manual for Using DG-QRAM in the Australasian Context, AP-R704-24, Austroads,
Sydney, NSW.

Austroads (2024b) Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on
Alternative Routes: Case Study Using DG-QRAM in the Australasian Context, AP-R705-24, Austroads,
Sydney, NSW.

Baltzer W, Riepe W, Zimmermann U, Zulauf C, Imhof D and Mayer G (2009) Verfahren zur Kategorisierung
von Straßentunneln gemäß ADR 2007, Bundesanstalt für Straßenwesen, Bergisch Gladbach, Germany.

Bilson M and Purchase A (2014) 'Determining Benefits of fixed Fire Fighting Systems in Road Tunnels - A
Risk-based Approach' [conference paper], Sixth International Symposium on Tunnel Safety and Security,
Marseille, France, pp 475-484.

Bilson M, Purchase A, Liu L, Nishimura N, Li S and Seo BR (2018) 'Rail System Ventilation Rehabilitation'
[conference presentation] Eighth International Symposium on Tunnel Safety and Security, Borås, Sweden
pp 333-348.

CETU (2005) Guide to Road Tunnel Safety Documentation - Booklet 3 - Risk Analyses Relating to
Dangerous Goods Transportation, CETU, Bron, France.

CETU (2018) Guide des dossiers de sécurité des tunnels routiers, Centre d'Études des Tunnels.

Coles A (2019) 'Performance-Based Fire Engineering of Road tunnels', Fire Protection Engineering (SFPE)
Q2(82):30-34.

DOT (U.S. Department of Transportation) (2016) Treatment of the Value of preventing Fatalities and Injuries
in Preparing Economic Analyses - Revised Department Guidance, Washington DC.
<[Link]
valuation-of-a-statistical-life-in-economic-analysis: U.S. Department of Transportation>.

European Union (2004) 'Directive 2004/54/EC of the European Parliament and of the Council of 29 April
2004 on Minimum Safety Requirements for Tunnels in the Trans-European Road Network', Official
Journal of the European Union 167(39).

Flint G, Woodburn P, Kotsovinos P and Panev Y (2019) 'Quantitative Fire Risk Assessment for Fire Safety
Design of Bridges', Fire Protection Engineering (SFPE) Q2(82):24-28.

FSV (2015) Tunnel Risikoanalysemodell - RVS 09.03.11, Österreichische Forschungsgesellschaft Straße -

Schiene - Verkehr, Vienna, Austria.

FSV (2016) Risikobewertung von Gefahrguttransporten in Straßentunneln - RVS 09.03.12, Österreichische

Forschungsgesellschaft Straße - Schiene - Verkehr, Vienna Austria.

Kohl B and Krieger J (2012) Current Practice for Risk Evaluation for Road Tunnels, PIARC World Road
Association, Paris.

Austroads 2024 | page 48

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Maevski I (2016) Recommended AASHTO Guidelines for Emmergency Ventilation Smoke Control in
Roadway Tunnels. New York.

NTC (National Transport Commission) (2018). Australian Code for the Transport of Dangerous Goods by
Road & Rail - Edition 7.6, National Transport Commission, Melbourne, VIC.

NTC (2020). Examining the Legal Framework for the Land Transport of Dangerous Goods, National
Transport Commission, Melbourne Vic.

Parson Brinckerhoff Quade & Douglas (2016) Risk Analysis Study of Hazardous Materials Trucks through
Eisenhower/Johnson Memorial Tunnel.

Parsons Brinckerhoff Quade & Douglas (1995) M5 Motorway East: Transport of Dangerous Goods -
Supplement to M5 East Motorway Environmental Impact Statement in Working Papers Vol. 2 of 3.

PIARC (2019a) Transport of Dangerous Goods through Road Tunnels: Quantitative Risk Assessment Model
(v. 4.04) - Reference Manual, PIARC - World Road Association, Paris, France.

PIARC (2019b) Transport of Dangerous Goods through Road Tunnels: Quantitative Risk Assessment Model
(v. 4.04) - User's Guide, PIARC - World Road Association, Paris, France.

Roads and Maritime Service (1993). Roads Act, RMS, Sydney, Australia.

United Nations (2017) Recommendations on the Transport of Dangerous Goods - Model Regulations,
twentieth revised edition, ST/SG/AC.10/1/Rev.20 (Vol.I & [Link]), United Nations, New York & Geneva.

United Nations (2018) ADR - European Agreement Concerning the International Carriage of Dangerous
Goods by Road, ECE/TRANS/275 (Vol.I & [Link]), applicable from 1 January 2019, United Nations, New
York and Geneva.

Austroads 2024 | page 49

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Appendix A Jurisdictional Differences

We note here differences between jurisdictions that need to be considered in the methodology and its
application. Primarily they are differences from European practice as DG-QRAM was initially set up for
Europe.

To begin with, we note the sources for Australasian regulations:

• The DG transport regulations in Australasia (Australian Code for the Transport of Dangerous Goods by
Road & Rail, Edition 7.8 ([Link]
code).
• 5433:2012 New Zealand Standard for the transport of dangerous goods on land (noting that this was
updated in 2020, but that the new version is not yet cited by regulations). It is incorporated by reference
into the Land Transport Rule – Dangerous Goods 2005
([Link]

The differences with European practice are noted here in three categories: those which are tunnel
characteristics resulting from the design and construction; those that relate to the DG fleet and the risks
related to the fleet; and those related to the operation of the tunnel and road system.

A.1 Tunnel Characteristics

Fixed fire-fighting systems (FFFS)

Deluge systems are standard in Australasia and either deluge or misting can be expected in all new or
significantly refurbished Australasian tunnels that may become dangerous goods routes. In contrast, fixed
suppression systems are newer in Europe, with many tunnels having no systems. DG-QRAM does not allow
for the fire suppression and cooling effects of water-based systems. Corrections will need to be made to
make it applicable in Australia. In particular, DG-QRAM assumes that all jet fires from LPG tanks will result in
a BLEVE. That is not the case, and with water cooling of the tank, it becomes unlikely.

It is suggested that reduction factors be applied to the risk calculated by DG-QRAM for each of the
DG-QRAM scenarios. Initially such factors could be estimated from other projects, such as comparative
studies with and without FFFS using the TuRisMo model. Longer term, the method may be adjusted with
comparative assessments of the risk improvement offered by FFFS for each type of DG in typical
Australasian tunnels.

Egress

The de facto standard in Australian road tunnels is to have a door to a place of safety (normally a cross
passage to the other tube) at 120 m or less. In New Zealand, this may be 150 m to be a multiple of the 50 m
emergency cabinet and hydrant spacing. Many European tunnels have significantly fewer egress
opportunities – egress routes are spaced significantly further apart. It is understood that DG-QRAM under-
estimates the positive benefits of egress routes. That will of course have less effect on the outcome if there
are not many egress routes, but the outcome may become quite wrong with the frequent cross passages that
are the norm in Australasia. That part of the DG-QRAM model will need to be revisited for Australasia. It is
not yet clear what form the adjustment should take.

Austroads 2024 | page 50

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

A.2 Dangerous Goods Fleet

Bulk diesel

In Australia, flammable liquids do not become dangerous goods unless the flashpoint is below 60°C. The
Australian specifications for diesel fuel give a flashpoint of 61.5°C (Fuel Quality Standards (Automotive
Diesel) Determination 2019). Hence, diesel is allowed through Australian tunnels. New Zealand does not
permit diesel through tunnels, treating it as DG. In Europe also, diesel is a dangerous good as it is often cut
back with lighter compounds for winter use, dropping the flashpoint below the threshold temperature. In this
way, DG-QRAM applied in Europe would recognise the count of diesel loads within the overall DG count, as
it would in New Zealand, whereas in Australia it would not. This is part of a broader consideration of whether
any of the division lines in the ADG between dangerous and non-dangerous loads are inappropriate for
tunnels. We should probably be considering assessing the comparative risk for diesel in the same way that
we do for (other) DG and making a decision on routing through tunnels in the same way, regardless of the
distinction created by 1.5°C difference in flashpoint.

B-double trucks

B-Double trucks are not permitted to carry DG in Europe, whereas petroleum particularly is regularly
delivered by B-Double in Australia, and even road trains (3 or 4 trailers) are known to carry DG sometimes.
There is a question as to whether that is a practical consideration for tunnels, which aren’t built in the
outback where road trains are commonly used.

While a B-double obviously carries more product, it is not an order of magnitude more, and the initial phases
of an incident may not be any different for a B-double. Consequently, there is perhaps no adjustment to be
made to the risk assessment for this, other than including in the DG vehicle count the counting of
combination vehicles by their number of trailers.

Driver training

We understand that European DG drivers get about twice the training of Australian DG drivers. There is a
need to assess how that could be accommodated in the model, and the extent that it needs to be.

Seasonal chemical transport

There is a question as to how to treat short term, seasonal risks. The example quoted to us is the start of
strawberry season in Western Australia when many tanks of agricultural chemical leave Fremantle Port.

A.3 Operational Differences

Lane use

In some (perhaps all) European countries, trucks are required to only use the slow lane on major routes with
two or more lanes each way. That is not the case in Australia, such that overtaking may occur on either side.
This could be addressed in that it changes accident rates, but only if such change is known. As local
accident rates are a key input to the model this should be readily captured.

Time of day restrictions

In some (perhaps all) European countries, heavy goods vehicles are not permitted to travel on Sundays. In
Australia, there are some time-of-day restrictions on the Sydney Harbour Bridge and the Western Distributor
for DG transport, but Australia generally does not have such restrictions. New Zealand has some such
restrictions, for example, on the Lyttleton and Homer Tunnels. Such restrictions can be assessed using
DG-QRAM.

Austroads 2024 | page 51

 Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Restrictions by categorisation

Australasia does not follow a categorisation procedure such as that set out in the ADR (sections 4.2, 4.3 and
4.4). Hitherto, dangerous goods transit in Australasian tunnels can be grouped as follows:
• Tunnels allowing all DGs
• Tunnels allowing some DGs
• Tunnels allowing no DGs.

In NSW, for example, the routes that exclude DGs are set out in the Road Rules (300-2). This list identifies
exclusions in the following three groups:
• All DGs are excluded all of the time
• All DGs are excluded most of the time
• Some DGs are excluded all of the time.

Where DGs are allowed through tunnels, such as at St Helena and Tugun tunnels, the exclusions are
identified by ADR Class and sub-division.

Supervision

All significant Australasian tunnels have 24/7 supervision from a control room with trained operators. That is
also true for major European tunnels. However, there are many older tunnels for which the level of
supervision and control may be lower. It is unclear whether any adjustment is appropriate for this.

Austroads 2024 | page 52

Methodology for Comparing Dangerous Goods Risks in Road Tunnels to Those on Alternative Routes: Best Practice Review

Austroads 2024 | page 53