Alternative Fuel Vehicles in Tunnels: Sandia Report

SANDIA REPORT
SAND2020-5466
Printed May 2020
Alternative Fuel Vehicles in

Tunnels
Chris B. LaFleur, Austin M. Glover, Austin R. Baird, Cyrus J. Jordan, Brian D. Ehrhart
Prepared by
Sandia National Laboratories
Albuquerque, New Mexico
87185 and Livermore,
California 94550
Issued by Sandia National Laboratories, operated for the United States Department of Energy by National
Technology & Engineering Solutions of Sandia, LLC.
NOTICE: This report was prepared as an account of work sponsored by an agency of the United States
Government. Neither the United States Government, nor any agency thereof, nor any of their employees, nor any of
their contractors, subcontractors, or their employees, make any warranty, express or implied, or assume any legal
liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or
process disclosed, or represent that its use would not infringe privately owned rights. Reference herein to any specific
commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily
constitute or imply its endorsement, recommendation, or favoring by the United States Government, any agency
thereof, or any of their contractors or subcontractors. The views and opinions expressed herein do not necessarily
state or reflect those of the United States Government, any agency thereof, or any of their contractors.
Printed in the United States of America. This report has been reproduced directly from the best available copy.
Available to DOE and DOE contractors from

U.S. Department of Energy
Office of Scientific and Technical Information
P.O. Box 62
Oak Ridge, TN 37831
Telephone: (865) 576-8401

Facsimile: (865) 576-5728
E-Mail: [email protected]
Online ordering: http://www.osti.gov/scitech
Available to the public from

U.S. Department of Commerce
National Technical Information Service
5301 Shawnee Rd
Alexandria, VA 22312
Telephone: (800) 553-6847

Facsimile: (703) 605-6900
E-Mail: [email protected]
Online order: https://classic.ntis.gov/help/order-methods/
2
ABSTRACT
Many types of vehicles using fuels that differ from typical hydrocarbons such as gasoline and
diesel are in use throughout the world. These include vehicles running on the combustion of
natural gas and propane as well as electrical drive vehicles utilizing batteries or hydrogen as
energy storage. These alternative fuels pose hazards that are different from traditional fuels
and the safety of these vehicles are being questioned in areas such as tunnels and other
enclosed spaces. Much scientific research and analysis has been conducted on tunnel and
garage hazard scenarios; however, the data and conclusions might not seem to be immediately
applicable to highway tunnel owners and authorities having jurisdiction over tunnels. This
report provides a comprehensive, concise summary of the literature available characterizing
the various hazards presented by all alternative fuel vehicles, including light-duty, medium- and
heavy-duty, as well as buses. Research characterizing both worst-case and more plausible
scenarios and risk-based analysis is also summarized. Gaps in the research are identified in
order to guide future research efforts to provide a complete analysis of the hazards and
recommendations for the use of alternative fuel vehicles in tunnels.
3
ACKNOWLEDGEMENTS
This work was initiated as a result of feedback from multiple stakeholders, including government,
industry, authorities having jurisdiction, and the Hydrogen and Fuel Cell Technical Advisory
Committee to the U.S. Department of Energy (DOE). Sandia National Laboratories led this effort,
in coordination with the U.S. DOE Office of Energy Efficiency and Renewable Energy (EERE)
Hydrogen and Fuel Cell Technologies Office (HFTO) and the U.S. Department of Transportation
(DOT) Federal Highway Administration (FHWA). This report was reviewed and input was provided
by Mark Smith of the U.S. DOE EERE Vehicle Technologies Office (VTO), William Bergeson of
the U.S. DOT FHWA, and Laura Hill and Mark Richards of HFTO. In addition to the reviewers
listed previously, the authors also acknowledge those who reviewed the Hydrogen Fuel Cell Vehicles
in Tunnels white paper which forms the basis for the hydrogen section in this work: members of the
workshop on hydrogen vehicles in tunnels and Alice Muna (Sandia). Financial support was provided
by the U.S. DOE HFTO.
4
CONTENTS
1. Introduction ............................................................................................................................................... 17
1.1. Definitions for Hazard Metrics..................................................................................................... 17
1.2. Tenability Criteria ........................................................................................................................... 21
1.3. Vehicle Classifications .................................................................................................................... 24
1.4. Report Organization ....................................................................................................................... 26
2. Tunnel Research Highlights for Traditional Fuels ............................................................................... 27
2.1. Overview of Traditional Fuels ...................................................................................................... 27
2.2. Properties of Traditional Fuels ..................................................................................................... 27
2.3. Associated Hazards of Traditional Fuels..................................................................................... 28
2.4. Pertinent Regulations and Safety Standards................................................................................ 30
2.4.1. National Fire Protection Association Standard 502 ..................................................... 30
2.4.2. ASHRAE HVAC Applications Ch. 16: Enclosed Vehicular Facilities (2019) ......... 30
2.4.3. NCHRP Guidelines for Emergency Ventilation Smoke Control in Roadway
Tunnels (2017) ................................................................................................................... 30
2.4.4. NCHRP Synthesis 415: Design Fires in Road Tunnels (2011) .................................. 30
2.5. Research Summary of Traditional Fuels in Tunnels .................................................................. 31
2.5.1. Experiments ....................................................................................................................... 31
2.5.1.1. The Rijkswaterstaat (RWS) and Other Time-Temperature Curves............31
2.5.1.2. The Runehamar Full Scale Tests .....................................................................32
2.5.1.3. Large Scale Fire Tests in the Second Benelux Tunnel .................................35
2.5.2. Modeling ............................................................................................................................. 36
2.5.2.1. Simulation of Tunnel Fires Using a Zone Model .........................................36
2.5.3. Analysis ............................................................................................................................... 39
2.5.3.1. Determination of critical parameters in the analysis of road tunnel fires ..39
2.5.3.2. An analysis of tunnel fire characteristics under the effects of blockage ....40
2.6. Traditional Fuels Research Gaps .................................................................................................. 42
3. Battery Electric Vehicles .......................................................................................................................... 45
3.1. Overview of Technology ............................................................................................................... 45
3.2. Properties of Lithium-ion Batteries.............................................................................................. 45
3.3. Associated Hazards......................................................................................................................... 47
3.3.1. Vent Gas Hazards ............................................................................................................. 50
3.3.2. Fire Hazards ....................................................................................................................... 51
Tunnels (2017) ................................................................................................................... 55
3.4.4. UL 2580 Standard for Safety Batteries for Use in Electric Vehicles ......................... 55
3.4.5. SAE J2464 Electric and Hybrid Electric Vehicle Rechargeable Energy Storage
System Safety and Abuse Testing ................................................................................... 55
3.4.6. SAE J2929 Electric and Hybrid Vehicle Propulsion Battery System Safety
Standard - Lithium-based Rechargeable Cells ............................................................... 56
3.4.7. FreedomCAR Battery Test Manual for Power-Assist Hybrid Electric Vehicles ..... 56
3.5. BEV Research Summary in Tunnels ............................................................................................ 57
3.5.1. Experiments ....................................................................................................................... 57
5
3.5.1.1. Reported BEV Failures and Incidents ............................................................57
3.5.1.2. Fire Analysis of BEVs in a Road Tunnel .......................................................57
3.5.1.3. Electric Vehicle Crash and Fire Damage .......................................................60
3.5.1.4. Comparison of the Fire Consequences of a BEV and ICE ........................61
3.5.2. Modeling ............................................................................................................................. 62
3.5.3. Analysis ............................................................................................................................... 62
3.5.3.1. Safety Test Methods for EV Batteries ............................................................62
3.6. BEV Research Gaps ....................................................................................................................... 63
4. Natural Gas Vehicles ................................................................................................................................ 65
4.2. Properties of Methane and Natural Gas...................................................................................... 66
4.3. Associated Hazards......................................................................................................................... 67
Tunnels (2017) ................................................................................................................... 69
4.4.4. National Fire Protection Association 52 ....................................................................... 69
4.4.7. National Fire Protection Association 59A .................................................................... 70
4.4.8. SAE J1616: Recommended Practice for Compressed Natural Gas Vehicle Fuel ... 70
4.4.9. SAE J2406: Recommended Practices for CNG Powered Medium and Heavy-
Duty Trucks ....................................................................................................................... 70
4.5. NGV Research Summary in Tunnels .......................................................................................... 71
4.5.1. Experiments ....................................................................................................................... 71
4.5.1.1. Vapor Cloud Explosions in a Long-Congested Region ...............................71
4.5.1.2. Heat Transfer to Ceiling and Impinging Diffusion Flame ..........................74
4.5.1.3. Vapor Cloud Explosions from Ignition of Gaseous Mixtures in a
Congested Region .............................................................................................................76
4.5.2. Modeling ............................................................................................................................. 80
4.5.2.1. Dispersion of CNG Fuel Releases in Naturally Ventilated Tunnels ..........80
4.5.2.2. Gaseous release, dispersion, and combustion for automotive scenarios ...82
4.5.2.3. Natural Gas Vehicle Explosion Risk in Tunnels...........................................87
4.5.2.4. Harm effects of Cryo-compressed hydrogen versus natural gas ................89
4.5.3. Analysis ............................................................................................................................... 92
4.5.3.1. Compressed Natural Gas Bus Safety: A Quantitative Risk Assessment ...92
4.5.3.2. LANL Risk Analysis ..........................................................................................94
4.6. NGV Research Gaps ...................................................................................................................... 95
5. Propane Vehicles ....................................................................................................................................... 99
5.2. Properties of Propane Storage ...................................................................................................... 99
5.3. Associated Hazards....................................................................................................................... 100
5.4. Pertinent Regulations and Safety Standards.............................................................................. 101
5.4.1. National Fire Protection Association Standard 502 ................................................... 101
5.4.2. ASHRAE HVAC Applications Ch. 16: Enclosed Vehicular Facilities (2019) ....... 101
6
Tunnels (2017) ................................................................................................................. 101
5.5. LPG Research Summary in Tunnels .......................................................................................... 102
5.5.1. Experiments ..................................................................................................................... 102
5.5.1.1. Experimental investigation and CFD modelling of the internal car park
environment in case of accidental LPG release ......................................................... 102
5.5.1.2. Smoke Control in Sloping Tunnels .............................................................. 105
5.5.2. Modeling ........................................................................................................................... 106
5.5.2.1. LPG Dispersion Analysis .............................................................................. 107
5.5.2.2. Explosion Risks and Consequences for Tunnels ....................................... 108
5.5.3. Analysis ............................................................................................................................. 108
5.5.3.1. LANL Risk Analysis ....................................................................................... 109
5.6. LPG Research Gaps ..................................................................................................................... 110
6. Hydrogen Fuel Cell Electric Vehicles .................................................................................................. 113
6.1. Overview of Technology ............................................................................................................. 113
6.2. Properties of Hydrogen ............................................................................................................... 113
6.3. Associated Hazards....................................................................................................................... 114
6.4. Pertinent Regulations and Safety Standards.............................................................................. 116
6.4.1. Global Technical Regulation No. 13 ............................................................................ 116
6.4.2. National Fire Protection Association Standard 502 ................................................... 116
6.4.3. ASHRAE HVAC Applications Ch. 16: Enclosed Vehicular Facilities (2019) ....... 116
Tunnels (2017) ................................................................................................................. 117
6.4.5. National Fire Protection Association 55 ..................................................................... 117
6.4.6. National Fire Protection Association 2........................................................................ 117
6.5. FCEV Research Summary in Tunnels ....................................................................................... 118
6.5.1. Experiments ..................................................................................................................... 118
6.5.1.1. Spontaneous Ignition of Pressurized Releases of Hydrogen into Air..... 118
6.5.1.2. Large-scale Hydrogen Deflagrations and Detonations ............................. 119
6.5.1.3. Releases from Hydrogen Fuel-cell Vehicles in Tunnels ............................ 121
6.5.1.4. HyTunnel Project to Investigate Hydrogen Vehicles in Road Tunnels.. 123
6.5.1.5. Deflagration and Detonation of Hydrogen Under a Tunnel Ceiling ...... 125
6.5.1.6. Fire experiments of carrier loaded FCEV in full-scale model tunnel ..... 126
6.5.1.7. Vapor Cloud Explosions from Ignition of Gaseous Mixtures in a
Congested Region .......................................................................................................... 127
6.5.2. Modeling ........................................................................................................................... 131
6.5.2.1. Hydrogen FCEV Tunnel Safety Study ........................................................ 132
6.5.2.2. Hydrogen Vehicle Explosion Risk in Tunnels ........................................... 133
6.5.2.3. CFD Modeling of Hydrogen Deflagration in a Tunnel ............................ 136
6.5.2.4. Releases from Hydrogen Fuel-cell Vehicles in Tunnels ............................ 139
6.5.2.5. Hydrogen Release and Combustion in Subsea Tunnels............................ 141
6.5.2.6. Hydrogen Jet Flame Hazard in Tunnels ...................................................... 145
6.5.2.7. Diffusion of Leaked Hydrogen in Tunnels ................................................. 145
6.5.2.8. Gaseous release, dispersion, and combustion for automotive scenarios 147
6.5.3. Analysis ............................................................................................................................. 151
6.5.4. Hydrogen FCEV Tunnel Risk Analysis ....................................................................... 152
7
6.5.4.1. Hydrogen Vehicle Explosion Risk in Tunnels ........................................... 153
6.5.4.1. Fire and Explosion Hazards in Tunnels of Alternative Fuel Vehicles.... 153
6.6. FCEV Research Gaps .................................................................................................................. 157
7. Conclusions .............................................................................................................................................. 161
7.1. Battery Electric Vehicles .............................................................................................................. 161
7.2. Natural Gas Vehicles .................................................................................................................... 161
7.3. Propane Vehicles .......................................................................................................................... 162
7.4. Hydrogen Fuel Cell Vehicles ....................................................................................................... 162
7.5. Closing Remarks and Future Work ............................................................................................ 162
8. References ................................................................................................................................................ 163
LIST OF FIGURES
Figure 1: Vehicle Types by Weight Classification (from [16])...................................................................25
Figure 2: Standard time-temperature fire design curves at tunnel structure interfaces .........................32
Figure 3: Heat release rates from the four large-scale fire tests (from [49]) ............................................34
Figure 4: Test 1 temperature compared with standard fire curves (from [49]).......................................34
Figure 5: Test Overview for Second Benelux Tunnel (from [51])............................................................35
Figure 6: Fire development for small truck fires (from [51]) ....................................................................36
Figure 7: Compartment Configurations (from [53]) ...................................................................................37
Figure 8: Smoke Characteristics Prediction of a Truck Fire (from [53]) .................................................38
Figure 9: Smoke Characteristics Prediction of a School Bus Fire (from [53]) ........................................39
Figure 10: Tunnel Dimension Effect on Average Temperature (from [54]) ..........................................40
Figure 11: Mean Effect on HRR Flux (from [55]) ......................................................................................41
Figure 12: Battery Technology Energy Densities (from [57]) ...................................................................46
Figure 13: Lithium-ion Battery Chemistry (from [59]) ...............................................................................46
Figure 14: 18650 Cell Defect (from [60]) .....................................................................................................47
Figure 15: BEV Failure Event........................................................................................................................49
Figure 16: Battery Vent Gas Species Compositions SOC (from [69]) .....................................................50
Figure 17: Cell Component Breakdown by % Mass (from [73]) ..............................................................51
Figure 18: HRR Curves for various Electrolyte Compositions and Mixtures (from [77]) ....................53
Figure 19: Test tunnel area layout (from [86]) .............................................................................................59
Figure 20: Cost of common fuel types over an 18-year span (from [98]) ...............................................65
Figure 21. Density contours for methane (created using data from [101]) .............................................66
Figure 22: Minimum ignition energy for common fuels types (from [111]) ...........................................67
Figure 23: Flammability limits for common fuels types (from [111]) ......................................................68
Figure 24: Schematic of Experimental Configuration (from [120]) .........................................................72
Figure 25: Spatial variation of flame speed and overpressure within the congestion region [120] ......73
Figure 26: Experimental Configuration (from [125]) .................................................................................74
Figure 27: Configuration of thermal sensors on ceiling plate (from [125]) .............................................75
Figure 28: Total heat flux at stagnation point vs. flame height over ceiling height (from [125]) .........75
Figure 29: Congestion region or grid where gas was filled then ignited (from [123]) ...........................76
Figure 30: Pressure Sensors distributed in and around grid structure (from [123]) ...............................77
Figure 31: Image of pure methane combustion right after ignition (from [123])...................................78
Figure 32: Overpressure vs. of distance parallel to the wall in the near field region (from [123]).......78
Figure 33: Overpressure vs. distance perpendicular to the wall in the far field region (from [123]) ...79
Figure 34: Volume of flammable region versus ventilation speed from simulation (from [112]) .......81
8
Figure 35: Gasoline Vapor and CNG Release Rates vs Ventilation Velocity (from [112]) ..................82
Figure 36: Fuel tanks configuration for both CNG and CH2 gas (from [126]) .....................................83
Figure 37: Tunnel Cross Section (from [126]) .............................................................................................84
Figure 38: Flammable mass and available energy of released gas in Case 1 (from [126]) .....................85
Figure 39: Flammable mass and available energy of released gases in Case 3 (from [126]) ..................85
Figure 40: Overpressure values up and down tunnel of the bus for release Case 3 (from [126]) ........86
Figure 41: Tunnel cross-sectional dimensions (from [127]) ......................................................................87
Figure 42: Mass flow of release for CNG and various H2 simulations (from [127]) ............................88
Figure 43: Exceedance curves for overpressure values per fuel type (from [127]) ................................89
Figure 44: Harmful/fatal distances per incident for instantaneous release (from [128]) ......................91
Figure 45: Harmful/fatal distances per incident per fuel type for continuous releases (from [128]) ..92
Figure 46. Fuel supply system of a CNG bus (from [135]) .......................................................................93
Figure 47: Example of Two-Phase Propane Storage (from [143]) ...........................................................99
Figure 48: Event Tree for Propane Event Scenario (from [149]) .......................................................... 102
Figure 49: Experimental Setup (from [149]) ............................................................................................. 103
Figure 50: CFD Results based on Ventilation (from [149]).................................................................... 104
Figure 51: Concentration LPG Comparing Simulation vs. Experimental Results (from [149]) ....... 104
Figure 52: Tunnel Cross Section (from [151]) .......................................................................................... 105
Figure 53: Critical Velocity vs. Tunnel Slope (from [150]) ..................................................................... 106
Figure 54: Critical Velocity Correlations (from [150]) ............................................................................. 106
Figure 55: Relevant Phenomena in LPG Spill & Vaporization Mathematical Model (from [137]).. 107
Figure 56: Minimum Ignition Energy for Different Fuels vs. Concentration (from [165])............... 115
Figure 57: Schematic of Experimental Configuration (from [169]) ...................................................... 119
Figure 58: SRI Corral Hollow Experiment Site Tunnel Facility (from [170])...................................... 120
Figure 59: Model of Vehicles in Tunnel (from [170]) ............................................................................. 120
Figure 60: 1/5th Scaled Tunnel Impulse and Overpressure (from [170]) ............................................ 121
Figure 61: Comparison of Experimental and Modeling Results (from [173]) ..................................... 122
Figure 62: Comparison of Time-dependent Hydrogen Concentration Values (from [173]) ............. 122
Figure 63: Configuration in Ignition Experiments (from [174]) ............................................................ 123
Figure 64: Experimental Setup for Deflagration Experiments (from [175]) ........................................ 125
Figure 65: Concentration and Layer Height Effect on Combustion (from [175]) .............................. 126
Figure 66: Experimental Tunnel Configuration and Carrier (from [176]) ............................................ 126
Figure 67: Comparison of Experimental & Simulated HRR for High Pressure Case (from [176]) . 127
Figure 68: Congestion region or grid where gas was filled then ignited (from [123]) ........................ 128
Figure 69: Pressure Sensors distributed in and around grid structure (from [123]) ............................ 129
Figure 70: Image of pure methane combustion right after ignition (from [123])................................ 130
Figure 71: Overpressure vs. of distance parallel to the wall in the near field region (from [123]).... 130
Figure 72: Overpressure vs. distance perpendicular to the wall in the far field region (from [123]) 131
Figure 73: Tunnel cross-sectional dimensions (from [127]) ................................................................... 134
Figure 74: Mass flow of release for CNG and various H2 simulations (from [127]) ......................... 134
Figure 75: Exceedance curves for overpressure values per fuel type (from [127]) ............................. 135
Figure 76: Overpressure Results without Vehicles (from [177]) ............................................................ 137
Figure 77: Overpressure Results with Vehicles (from [177]) ................................................................. 138
Figure 78: Tunnel Model with Transverse Ventilation (from [173]) ..................................................... 139
Figure 79: Simulation Results Showing Evolution of Flammable Hydrogen Volume (from [173]) 140
Figure 80: Simulation of Peak Ignition Overpressures vs. Ignition Delay (from [173]) .................... 140
Figure 81: Subsea Tunnel Model with FCEV and Monitoring Points (from [178]) ........................... 141
Figure 82: Subsea Tunnel Model Cross Section (from [178]) ................................................................ 141
9
Figure 83: Longitudinal H2 Distribution Various Ventilation Conditions (from [178]) .................... 142
Figure 84: Traverse H2 Distribution Various Ventilation Conditions (from [178]) ........................... 143
Figure 85: Overpressure History at Ignition Time of 3.1 Seconds (from [178]) ................................. 144
Figure 86: Overpressure History at Ignition Time of 6.1 Seconds (from [178]) ................................. 144
Figure 87: Tunnel geometry and Boundary Conditions of CFD Model (from [179]) ........................ 145
Figure 88: Case A Simulation Tunnel Geometries (from [180]) ............................................................ 145
Figure 89: Case B Simulation Tunnel Geometry (from [180]) ............................................................... 146
Figure 90: Case C Simulation Tunnel Geometry (from [180]) ............................................................... 146
Figure 91: Fuel tanks configuration for both CNG and CH2 gas (from [126]) .................................. 147
Figure 92: Tunnel Cross Section (from [126]) .......................................................................................... 148
Figure 93: Flammable mass and available energy of released gas in Case 1 (from [126]) .................. 149
Figure 94: Flammable mass and available energy of released gases in Case 3 (from [126])............... 149
Figure 95: Overpressure values up and down tunnel of the bus for release Case 1 (from [126])..... 150
Figure 96: Overpressure values up and down tunnel of the bus for release Case 3 (from [126]) ..... 150
Figure 97: Event Sequence Diagram for a Hydrogen FCEV in a Tunnel (from [113]) ..................... 152
Figure 98: Liquefied Fuel Vehicle Event Tree for Incidents in Tunnels (from [134]) ....................... 154
Figure 99: Compressed Gas Vehicle Event Tree for Incidents in Tunnels (from [134]) ................... 155
Figure 100: Overpressure vs. Distance for Liquid H2 tank at 350 bar (from [127]) ........................... 156
Figure 101: Overpressure vs. Distance for Gaseous H2 tank at 350 bar (from [127]) ....................... 156
LIST OF TABLES
Table 1: Human Injury Criteria from Overpressure (from [13]) ..............................................................21
Table 2: Effects from Heat Flux (from [13]) ...............................................................................................22
Table 3: Human Injury Criteria from Temperature (from [12]) ...............................................................23
Table 4: Vehicle Weight Classifications (from [15]) ...................................................................................24
Table 5: Properties of Traditional Fuels .......................................................................................................27
Table 6: Major ICE Road Tunnel Fire Incidents (from [34])....................................................................29
Table 7: List of Significant Full-Scale Tunnel Fire Tests (from [42]) .......................................................31
Table 8: Commodities used as fuel in the HGV tests (from [49]) ............................................................33
Table 9: Comparison of Lithium and Lithium-Ion Cells ...........................................................................45
Table 10: Overview of Test Standards for Lithium-ion Battery Abuse (from [61]) ..............................48
Table 11: Cell Deconstruction Breakdown by % Mass ..............................................................................52
Table 12: Summary of AEGL Values (from [79]).......................................................................................54
Table 13: Lithium-ion Battery Characteristics (from [85]) .........................................................................58
Table 14: Test Scenarios (from [85]) .............................................................................................................58
Table 15: Release Quantities (from [85]) ......................................................................................................60
Table 16: BEV vs. ICE Fire Characteristics (from [88]) ............................................................................61
Table 17. Physical and Chemical Properties of Natural Gas & Methane (from [100]) .........................66
Table 18: Flammability Properties of Hydrogen and Other Fuels ...........................................................67
Table 19: Test Results Summary for CH4 Vapor Cloud Combustions (from [120]) ............................72
Table 20: Initial Conditions of Experiment (from [123]) ..........................................................................77
Table 21: Tunnel dimensions and averaged velocities (from [112]) .........................................................81
Table 22: Storage Configurations (from [126]) ............................................................................................83
Table 23: Storage Configurations Details (from [126]) ..............................................................................84
Table 24: Combustion results within tunnel (from [126]) .........................................................................86
Table 25: Summary of gas cloud & overpressure for various vehicles in both tunnels (from [127]) ..88
Table 26: Assumed modeling parameters for simulation (from [128]) ....................................................90
10
Table 27. Quantitative Risk Assessment (from [135]) ................................................................................94
Table 28: Probabilistic Outcomes of a Tunnel Accident (from [136]) ....................................................94
Table 29. Natural gas fuel amounts by vehicle class. ..................................................................................97
Table 30: Physical and Chemical Properties of Propane (from [147]) .................................................. 100
Table 31: Flammability Properties of Propane ......................................................................................... 100
Table 32: Gas Flow Test Parameters (from [149])................................................................................... 103
Table 33: Transient Dispersion for Transverse Ventilation (from [137])............................................. 108
Table 34: Probabilistic Outcomes of a Tunnel Accident (from [136]) ................................................. 109
Table 35: Physical and Chemical Properties of Hydrogen (from [163]) ............................................... 114
Table 36: Flammability Properties of Hydrogen and Other Fuels ........................................................ 114
Table 37: Results of Steady State Ignition Experiments (from [174])................................................... 124
Table 38: Results of Quiescent Ignition Experiments (from [174]) ...................................................... 125
Table 39: Initial Conditions of Experiment (from [123]) ....................................................................... 129
Table 40: Results Summary of Hydrogen FCEV in Tunnels Risk Modeling (from [113]) ................ 133
Table 41: Summary of gas cloud & overpressure for various vehicles in both tunnels (from [127]) 135
Table 42: Storage Configurations Details (from [126]) ........................................................................... 148
Table 43: Storage Configurations Details (from [126]) ........................................................................... 148
Table 44: Combustion results within tunnel (from [126]) ...................................................................... 151
11
This page left blank
12
EXECUTIVE SUMMARY
Vehicles of all sizes use internal combustion engines powered by hydrocarbons such as gasoline,
diesel, ethanol, and various blends. As alternative fuel vehicles and the infrastructure to support
them become more widely available, safety concerns become a more important topic because these
fuels are different from the typical safety hazards that have been accepted since the widespread
adoption of the automobile. Road tunnels and other enclosures have additional concerns due to
limited access and egress, ventilation system capacity, and emergency response limitations.
Therefore, tunnel owners, authorities, and other stakeholders have raised concerns about alternative
fuel vehicles traveling through tunnels, particularly in urban and high commuter areas. Natural gas
and propane are used to replace traditional fuels as cleaner or more efficient alternative fuel in
combustion engines. Alternative fuel vehicles also include those powered by electricity using lithium-
ion batteries or hydrogen fuel cells. The intent of this document is to help illustrate the level of risk
for all types of fuel, provide a full understanding of what codes & standards are applicable to these
alternative fuel vehicles, and to review relevant research that has been conducted to date. Risks to
life safety and infrastructure damage always exist, regardless if the fuel type is traditional or
alternative. The goal for acceptance of alternative fuel vehicles may be to maintain the same level of
risk that is generally accepted for traditional fuel vehicles.
For each fuel type, including traditional fuels such as gasoline and diesel, an overview with a
summary of fuel properties and hazards, applicable codes and standards, and tunnel-specific
research in the literature is provided. Additionally, information to compare the various classes of
vehicles such as passenger, light-duty, heavy-duty, and cargo are reviewed where available. This
information helps characterize the severity of the hazard for each classification of vehicle. The
different codes and standards applicable come from organizations such as National Fire Protection
Association, Society of Automotive Engineers, Underwriters Laboratories, and more. While these
organizations provide various codes and standards, it is up to state and local jurisdiction to help
adopt, regulate, and enforce them. The literature review provides information organized by
experiments, modeling, and theoretical calculations/analyses performed regarding the specific fuel in
tunnels. Some of the fuel types are not as well studied as others, specifically in tunnels. In these
cases, research that is applicable to tunnels is reviewed. For example, battery electric vehicles real-
world incidents from the last few years are presented because few full-scale experiments have been
conducted. Research gaps identified are provided for each fuel type.
Battery
There are two distinct hazards from battery electric vehicles (BEVs): deflagration from the
flammable vent gas and a unique fire hazard due to thermal runaway propagation between cells.
Unlike flammable gases with well-known properties, characteristics of the vent gases from a battery
cell failure are not as well defined. Variations in cell chemistry, capacity, thermal runaway
propagation between cells, state of charge, form factor, and other variations affect hazards such as
vent gas species, volume production, and production rates. These are further discussed in the
associated hazards for BEVs. Due to these factors, the hazards associated with BEVs are not as well
characterized as some of the other alternative fuel vehicles. Studies to understand the heat release
rate compared with traditional fuel vehicles has been reviewed. Also, studies characterizing fire
spread within a BEV and tactics to slow or stop thermal propagation based using water and other
methods has been studied. Testing and analysis to understand the failure scenarios and modes
pertaining to BEVs is limited but also included. This includes understanding thermal, mechanical,
13
and electrical failures. A variety of research gaps are discussed for BEVs in tunnels and lithium-ion
batteries in general.
Natural Gas
Most of the studies surveyed involve compressed natural gas (CNG) in gaseous form, showing the
need to further understand liquefied natural gas (LNG) hazards and the differences for this lower
pressure, liquid fuel. Harmful and lethal distances due to fire or overpressure from a CNG release
can exceed comparable distances for LNG due to the higher storage pressures of CNG. A variety of
tunnel studies have been summarized that characterized CNG releases of vapor clouds that are
considered equivalent to amounts of CNG in city bus and passenger car configurations. One study
looks at different initiating events of a natural gas leak from a vehicle in a tunnel and how the
flammable mass and overpressures change based on those events. Another study shows the
difference in characteristics such as harmful distances from a vapor cloud explosion for CNG and
LNG. This helps understand the fundamental differences for this fuel based on its state. As more
research is conducted, the focus specifically for tunnels should include experimental studies of
natural gas dispersion and overpressure in actual or scaled down tunnels as well as large scale natural
gas flames heat transfer analysis.
Propane
There were relatively few studies evaluating the failure modes and consequences associated with
liquefied propane gas (LPG) vehicles in tunnels. One study used a failure tree to inform the
experimental setup. The experimental data was then used and compared with a CFD model for
validation. This model can be further used to understand the gaseous dispersion characteristics of
propane vehicles failing in tunnels and other confined spaces. Additionally, an experiment was also
conducted to understand smoke dispersion in a tunnel using a propane fire to understand the effect
of ventilation and tunnel slope. Modeling was done to help understand the dispersion and
evaporation of an LPG spill. Future work should investigate thermal consequences of failures, which
have not been reported for many of the studies included in this literature survey. This includes the
heat release rate, temperature, and structural damage resulting from different failure modes.
Hydrogen
There are a number of studies evaluating the failure modes and consequences associated with
hydrogen fuel cell electric vehicles (FCEVs) in tunnels. Multiple studies have identified possible
release events that could occur. Other studies have begun to quantify probabilities and likelihoods
for these various events. Modeling on the consequences of releases has also been performed;
multiple studies have investigated hydrogen accumulation followed by ignition, resulting in an
overpressure. One study looked at the thermal effects on tunnel components of a jet fire rather than
an overpressure. Multiple experimental studies have also investigated overpressures resulting from
delayed hydrogen ignition, though some have investigated thermal effects of jet fires as well. Some
research gaps identified for FCEVs in tunnels are the need for an increased focused on thermal
effects of jet fires on tunnel structures. Similarly, while some studies have investigated the extent of
overpressures in tunnels, more information is needed on what type of critical tunnel infrastructure
might be affected by such hazards. More information is needed on how likely hydrogen is to ignite
in different release configurations which will greatly inform risk analyses. Most current studies have
focused on light-duty vehicles, and so more modeling and experiments on larger fuel-capacity
vehicles would be important to consider. Some of these gaps as well as others will be addressed in
the current HyTunnel-CS project.
14
ACRONYMS AND DEFINITIONS
Abbreviation Definition
AEGL acute exposure guideline levels
AHJ authority having jurisdiction
AIT auto-ignition temperature
ASHRAE American Society of Heating, Refrigerating and Air-Conditioning Engineers
BEV battery electric vehicle
BLEVE boiling liquid expanding vapor explosion
CANA Central Artery North Area
CFAST Consolidated Model of Fire and Smoke Transport
CFD computational fluid dynamics
CNG compressed natural gas
DDT deflagration-to-detonation transition
DEC diethyl carbonate
DMC dimethyl carbonate
DOE Department of Energy
DOT Department of Transportation
EC ethylene carbonate
EMC ethyl methyl carbonate
FCEV fuel cell electric vehicle
FDS Fire Dynamics Simulator
FHWA Federal Highway Administration
FL flammability limit
FLACS FLame ACceleration Simulator
FMEA failure mode and effect analysis
FMVSS Federal Motor Vehicle Safety Standards
GGE gasoline-gallon-equivalent
GVWR gross vehicle weight rating
HF hydrogen fluoride
HGV heavy goods vehicle
HRR heat release rate
ICE internal combustion engine
IDLF immediate danger to life or health
ISO International Organization for Standardization
15
Abbreviation Definition
LCO lithium cobalt oxide
LEL lower explosive limit
LFL lower flammability limit
LFP lithium iron phosphate
LMO lithium manganese oxide
LNG liquefied natural gas
LPG liquefied petroleum gas
NCA nickel cobalt aluminum
NCHRP National Cooperative Highway Research Program Guidelines
NFPA National Fire Protection Association
NGV natural gas vehicle
NHTSA National Highway Traffic Safety Administration
NIST National Institute of Standards and Technology
NMC nickel manganese cobalt
MIE minimum ignition energy
MPC methyl propyl carbonate
PC propylene carbonate
PHEV plug-in hybrid electric vehicles
PIARC Permanent International Association of Road Congresses
PRA probabilistic risk assessment
PRD pressure relief device
PTC positive temperature coefficient
QRA quantitative risk analysis
RVP Reid vapor pressure
RWS Rijkswaterstaat, a time-temperature curve used in safety standards
SAE Society of Automotive Engineers
SUV sport utility vehicle
THC total hydrocarbons
THR total heat released
TPRD thermally activated pressure relief device
UEL upper explosive limit
UFL upper flammability limit
16
1. INTRODUCTION
The purpose of this report is to present tunnel owners and other stakeholders with a summary of
the current body of scientific information about the use of alternative fuel vehicles in tunnels. This
will allow tunnel regulators to determine requirements for alternative fuel vehicle transit through
tunnels on a national level. This objective is accomplished through performing a comprehensive
literature review of publicly available tunnel research experiments and modeling focused on the
hazards associated with alternative fuel vehicles. Gaps in the existing research are identified and
suggestions to address these gaps are presented. This includes a review of the scenarios and failure
modes and the range of consequences associated with the failures. This work expands upon a
previously published hydrogen-specific white paper [1] by including other alternative fuels.
The scope of this report is to summarize hazard research in tunnels for road vehicles powered by
traditional fuels, battery electric systems, compressed natural gas, liquefied natural gas, propane, and
hydrogen. The volume of fuel in each vehicle is an important consideration, so the class of vehicle is
noted where applicable to help further understand the potential consequences of light-duty,
medium-duty, or heavy-duty vehicle incidents.
Note that some of the research presented in this document assess theoretical scenarios that are
implausible in the real world, such as large stoichiometric mixtures. These situations are included in
this paper to recognize the scientific principle, but also point out the improbability of encountering
these conditions outside a laboratory. However, consequence is only part of the overall risk. The
likelihood of these scenarios is extremely low or virtually impossible, which significantly lowers the
risk of the theoretical hazards.
1.1. Definitions for Hazard Metrics

The following section defines the various hazard metrics discussed in the literature:
 Flammability Limits
Flammability limits are important for fire and explosion analysis because it defines the volume
fraction and conditions range of fuel required to create a flammable environment. Codes, standards,
and practices have specific requirements regarding the flammable gas concentration permitted in any
given environment. The lower flammability limit (LFL) is the lowest fuel concentration that will
allow flame or flash propagation from an ignition source within the mixture. The upper flammability
limit (UFL) is the highest fuel concentration that will allow flame or flash propagation from an
ignition source within the mixture. Outside of these limits, no flame can occur [2]. Standard units of
flammability limits are in volume percent (%) or volume fraction.
 Explosion
Explosion is a general term for the elevated release of energy that generates high temperatures. This
causes expansion of a gas volume and leads to an overpressure [3]. A broad class of pre-mixed
flames can cause this overpressure. To further define an explosion, the speed of the flame front
determines whether it is a deflagration (subsonic speeds) or detonation (supersonic speeds).
17
 Explosion Limits
Explosion limits refer to the range of pressure and temperature for which an explosive reaction can
occur for a fixed composition mixture. The explosion limit is given as a minimum autoignition
temperature (AIT) which is a strong function of the fuel type, pressure, and overall fuel
concentration [4]. The explosion limits are within the same concentration range as the flammability
limits. Standard units of explosion limits are in volume percent (%) or volume fraction.
 Auto-Ignition Temperature
The auto-ignition temperature (AIT) is the lowest recorded temperature in which ignition occurs
spontaneously or in the absence of piloted ignition source in a material. This applies to solids, liquids
and gases. If the rate at which heat evolves in a gas or vapor is greater than the rate of heat loss to
the surrounding area, ignition can result [2] [5]. Standard units for the auto-ignition temperature is
degrees Celsius (°C) or degrees Fahrenheit (°F).
 Detonation/Deflagration
Deflagration is defined as a flame front which propagates through a gas at subsonic speeds.
Typically, obstructions such as piping and conduit as well as confinement cause a flame front to
accelerate to speeds greater than the speed of sound (343 m/s in ambient air). This causes the
deflagration-to-detonation transition (DDT). A detonation is defined as a flame front which
propagates through a gas at supersonic speeds. A deflagration will have flame speeds that vary from
less than 1 m/s up to 3 m/s whereas detonation flame speeds are anywhere from 1.5 to 2.8 km/s
[6].
 Detonation Limits
Detonation limits is the range at which a detonation can self-sustain. These limits typically have a
narrower range within the flammability/explosion limits. This is due to the stronger dependence on
confinement, mixture composition, and initial temperature and pressure compared with the
flammability/explosion limits [4]. Standard units for detonation limits are in volume percent (%) or
volume fraction.
 Laminar Flame Speed

The critical parameter controlling the rate of pressurization is the burning speed. The burning speed
is correlated to a fundamental flame propagation rate into the unburned premixed gas. This flame
speed generally increases with increasing temperature and decreases with increasing pressure [7]. The
laminar flame speed is dependent on the chemical kinetics along with the thermal and mass diffusion
[8]. Standard units for the laminar flame speed are meter-per-second (m/s) or feet-per-minute
(ft/min).
 Overpressure
Overpressure defines the pressure wave that a flammable mixture generates during combustion.
This pressure wave is caused by the energy released from initial deflagration/detonation. The
maximum theoretical overpressure (Pmax) is the pressure that is generated when the gas is combusted
in a perfectly adiabatic process in a closed chamber. This is a value generated theoretically or at
optimal conditions in a laboratory. Pmax depends on the composition of gas produced as well as the
concentration and other factors such as confinement [9]. Standard units for overpressure can be
kilopascals (kPa), pounds-per-square-inch (psi), or bar.
18
 Equivalence Ratio
The ratio of actual molar fuel/air ratio to the stoichiometric molar fuel/air ratio is the equivalence
ratio. If this ratio is below 1, this is a lean mixture with excess air after combustion. If the ratio is
above 1, the mixture is fuel rich leading to incomplete consumption of the fuel. Combustion with an
equivalence ratio equal to 1 leads to complete consumption of oxygen and fuel during a reaction.
 Stoichiometric Ratio
The stoichiometric ratio is the mixture of fuel and air in which there is exactly enough air to
completely burn all of the fuel. Combustion in stoichiometric conditions (when the mixture is at the
stoichiometric ratio) leads to complete consumption of both oxygen and fuel.
 Adiabatic Flame Temperature

Adiabatic flame temperature is the maximum temperature that can result from combustion of
reactants. Heat transfer away from the reaction, incomplete combustion, and other dissociations will
result in a lower temperature. The maximum adiabatic flame temperature occurs when a mixture is
stoichiometric [10]. Standard units for flame temperature are Kelvin (K), degrees Celsius (°C), or
degrees Fahrenheit (°F).
 Flash Point
The flash point is the lowest temperature at which a liquid solvent can form a mixture above the
surface or within a container that is flammable or ignitable. Lower flash point temperatures indicate
that it is easier for the mixture to ignite. There are two types of tests to measure the flash point:
closed cup and open cup. These methods test flash point in an open pool type configuration and a
closed container configuration [2]. Standard units for the flash point temperature degrees are degrees
Celsius (°C) or degrees Fahrenheit (°F).
 Heat of Combustion
Heat of combustion is the amount of heat released when a substance is burned. The heat of
combustion can be further defined as the higher and lower heating values. The lower heating value
of a fuel is defined by combustion of a fuel at 25°C and returning the resulting mixture of
combustion products down to 150°C. This assumes the latent heat of vaporization of water in the
reaction products is not recovered. The higher heating value is similar, but the products have
returned to a temperature of 25°C, which considers the latent heat of vaporization of water in the
combustion products [11]. Standard units for the heat of combustion are Megajoule per Kilogram
(MJ/kg), British Thermal Units per-pound-mass (Btu/lbm), or kilojoule-per-mole (kJ/mol).
 Heat Release Rate

Heat release rate (HRR) is the rate of energy released from a fire. This rate is typically defined by a
plotted curve, with time on the horizontal axis and energy released on the vertical axis [12]. The heat
release rate curve is used to characterize fires by understanding the peak heat release rate as well as
the duration of the fire. Standard units for the heat release rate are kilowatt (kW) or British Thermal
Unit per hour (Btu/hr).
 Heat Flux
In addition to the HRR, the heat release rate flux or heat flux is the total energy flow over time per
unit of surface area. Standard units for the heat flux are kilowatt per square meter (kW/m2) or British
Thermal Unit per square foot (Btu/hr-ft2).
19
 Total Heat Release
The total area under the heat release rate curve defines the total heat released (THR) [12]. Total heat
released is used to characterize the size of a fire. Standard units for the total heat released are
megajoule (MJ) or British Thermal Unit (Btu).
20
1.2. Tenability Criteria
To better understand the consequence and take-away values from the literature review, the following
tables illustrate the effects of overpressure, heat flux, and temperature hazards. The severity of
injuries and damage with the increase in these metrics is shown in Table 1 through Table 3. Table 1
lists the human injury criteria due to overpressure from National Fire Protection Association
(NFPA) 921: Guide for Fire and Explosion Investigations [13].
Table 1: Human Injury Criteria from Overpressure (from [13])

Overpressure
Effects or Injuries
psi kPa bar
0.60 4.14 0.04 Threshold for injury from flying glass
1.00- 6.90-
0.070-0.140
2.00 13.80 Threshold for skin laceration from flying glass
2.40- 16.50- Threshold for eardrum rupture/10% probability of eardrum
0.170-0.19
2.80 19.30 rupture
2.00- 13.80-
0.140-0.21
3.00 20.70 Threshold for serious wounds from flying glass
3.00 20.70 0.21 Overpressure will hurl a person to the ground
3.40 23.4 0.23 1% eardrum rupture
4.00- 27.60-
0.28-0.35
5.00 34.500 Serious wounds from flying glass near 50% probability
5.80 40.00 0.40 Threshold for body-wall penetration from flying glass (bare skin)
6.30 43.40 0.43 50% probability of eardrum rupture
7.00- 48.30-
0.48-0.55
8.00 55.20 Serious wounds from flying glass near 100% probability
10.00 68.95 0.69 Threshold for lung hemorrhage
14.50 99.97 1.00 Fatality threshold for direct blast effects
16.00 110.30 1.10 50% eardrum rupture
17.50 120.70 1.21 10% probability of fatality from direct blast effects
27.00 186.20 1.86 1% mortality: A high incidence of severe lung injuries
29 199.9 2.00 99% probability of fatality from direct blast effects
21
Table 2 illustrates the levels and exposure durations at which blistering (second-degree burn) injuries
occur due to heat flux exposure from NFPA 921: Guide for Fire and Explosion Investigations [13].
Both the heat flux and exposure time help in understanding consequences.
Table 2: Effects from Heat Flux (from [13])

Radiant Heat Flux
Effects or Injuries
kW/m2 Btu/hr-ft2
Common thermal radiation exposure while firefighting. This energy level may
cause burn injuries with prolonged exposure.
2.5 793
Human skin experiences pain with a 33-second exposure and blisters in 79
seconds with second-degree burn injury.
Human skin experiences pain with a 13-second exposure and blisters in 29
5 1,586
seconds with second-degree burn injury.
Human skin experiences pain with a 5-second exposure and blisters in
10 3,172
10 seconds with second-degree burn injury.
15 4,758
20 6,344
80 25,377 Heat flux for protective clothing Thermal Protective Performance (TPP) Test.
100 31,720 Steel structure collapse (>30 min exposure) (from [14])
22
Table 3 illustrates various effects and injuries from temperature exposure from National Institute of
Standards and Technology (NIST) [12]. Both the heat flux and exposure time help in understanding
consequences.
Table 3: Human Injury Criteria from Temperature (from [12])

Temperature
Effects or Injuries
Celsius (°C) Fahrenheit (°F)
37.0 98.6 Average normal human oral/body temperature
38 101 Typical body core temperature for a working fire fighter
43 109 Human body core temperature that may cause death
44 111 Human skin temperature when pain is felt
48 118 Human skin temperature causing a first-degree burn injury
54 130 Hot water causes a scald burn injury with 30 second exposure
55 121 Human skin temperature with blistering and second degree burn injury
62 140 Temperature when burned human tissue becomes numb
72 162 Human skin temperature at which tissue is instantly destroyed
100 212 Temperature when water boils and produces steam
250 482 Temperature when charring of natural cotton begins
>300 >572 Modern synthetic protective clothing fabrics begin to char
≥400 ≥752 Temperature of gases at the beginning of room flashover
≈1000 ≈1832 Temperature inside a room undergoing flashover
23
1.3. Vehicle Classifications
Vehicle classifications are defined by the vehicle’s gross vehicle weight rating (GVWR) which is the
maximum operating weight of the vehicle. The GVWR includes all vehicle fluids, passengers, and
the cargo capability but does not include trailers. Definitions from various administrations such as
the U.S. Department of Transportation Federal Highway Administration (FHWA) and U.S.
Environmental Protection Agency (EPA) are included along with examples and applications of each
vehicle class.
Table 4: Vehicle Weight Classifications (from [15])

Vehicle Examples/
DOT FHWA EPA
Class Applications
Heavy-Duty Engine Light Light-Duty Truck: <6,000 lbs.
Class 1: <6,000
Light-Duty Vehicle: <8,500 lbs.
Sedans, lbs.
Light- Light-Duty Trucks: <8,500 lbs.
SUVs,
Duty
Pickups, Light-Duty Truck 3 and 4 and
Vehicle
Utility Van Class 2: 6,001 – Heavy Engines Heavy Light-Duty Truck: 6,001 – 8,500 lbs.
10,000 lbs.
Medium-Duty Vehicle: 8,501 – 10,000 lbs.
Heavy-Duty Vehicle Heavy-Duty Engine: >8,500 lbs.

Class 3: 10,001 –
14,000 lbs.
Heavy-Duty Vehicle 3: 10,001 – 14,000 lbs.
Delivery
Medium- Truck, Class 4: 14,001 –
Duty Bucket Heavy-Duty Vehicle 4: 14,001 – 16,000 lbs.
16,000 lbs.
Vehicle Truck,
School Bus Class 5: 16,001 –
19,500 lbs.
Class 6: 19,501 –
26,000 lbs.
City Bus, Class 7: 26,001 – Medium Heavy-Duty Engine: 19,501 – 33,000 lbs.
Refuse, 33,000 lbs.
Moving Heavy-Duty Vehicle 7: 26,001 – 33,000 lbs.
Heavy-
Truck, Heavy Heavy-Duty Engine Urban Bus: >33,000 lbs.
Duty
Truck, Fuel
Vehicle Class 8: >33,000 Heavy-Duty Vehicle 8a: 33,001 – 60,000 lbs.
Vehicle,
lbs.
Heavy Semi
Tractor Heavy-Duty Vehicle 8b: >60,000 lbs.
From Table 4, there are three distinct vehicle classes: light-duty, medium-duty, and heavy-duty
vehicle. These classes are broken down into sub-classes for each specific administration.
24
Figure 1: Vehicle Types by Weight Classification (from [16])
Figure 1 above gives further examples of various vehicles and what FHWA weight classification they
fall under.
25
1.4. Report Organization
This report is organized into separate chapters based on the specific fuel type. Each of the chapters
has a section that provides an overview of the fuel type, along with the general properties such as
density, flammability limits, etc. Additionally, applicable regulations, codes, and standards are listed
for reference. The literature review is broken down into three main section: experiments, modeling,
and analysis. The experiment section reviews what work has been done that uses testing and
measurement techniques to simulate fuel properties and characteristics such as dispersion,
flammability limits, and overpressure in intermediate to full scale tunnels or confined areas. The
modeling section shows computations done using various CFD software and other programs to
simulate the characteristics and effects of fuels dispersion, fires, overpressure, and more in a tunnel.
The analysis section describes work that studies hazard and risk of these fuels in tunnels. This work
is done using engineering calculations and physics equations with experimental data and fuel
properties as inputs to compare and further understand the effect of these fuels in tunnels. Each
section has a research gap sections that goes into detail on what work has been completed and what
work would be useful to help understand the overall hazard and consequence of these fuels in
tunnels.
26
2. TUNNEL RESEARCH HIGHLIGHTS FOR TRADITIONAL FUELS
2.1. Overview of Traditional Fuels

Internal combustion engines (ICE) are the most commonly used powertrains in passenger, light-
duty, heavy-duty and cargo vehicles. The main fuels are gasoline, diesel, and ethanol. Different
ethanol-gasoline blends such as E15 (containing 15% ethanol by volume) or E85 Flex Fuel
(containing 51% to 85% ethanol by volume) [17] are commonly found at gas stations. Usage of
ethanol has increased in the U.S. from 1.7 billion gallons in 2001 to about 14.4 billion in 2016 [18].
In 2018, the U.S. consumed approximately 143 billion gallons of gasoline [19]. Diesel fuel is more
common in larger commercial grade vehicles such as medium-duty, heavy-duty, and cargo, but is
also becoming more common at the passenger and light-duty markets. In 2014, 78% of medium-
duty and heavy-duty trucks sold were diesel powered, while only 1.5% of passenger and light-duty
sold were diesel powered [20].
2.2. Properties of Traditional Fuels

Various properties of these traditional fuels help understand and compare with the properties of
alternative fuels. Characteristics such as the liquid density and higher/lower heating values can be
used to define the fuel loading. The gas density can be used to understand the flammability limits.
The adiabatic flame temperature can be used to estimate the theoretical pressure rise from a
confined explosion. The laminar flame speed or burning velocity determines how quickly the
unburned mixture is consumed in a flame front. This value is affected by obstructions, temperature,
fuel concentration and, other factors which can cause turbulent flame speeds leading to the
deflagration-to-detonation transition.
Table 5: Properties of Traditional Fuels
Property Gasoline Diesel Ethanol
Chemical Formula C4-C12 [21] C8-C25 [21] C2H5OH [22]
Molecular Weight 95-120 g/mol [23] 204 g/mol [23] 46.07 g/mol [24]
Gas Density (25 °C, 1 atm) 1.227 kg/m3 [22] 1.46 kg/m3 [23] 1.214 kg/m3 [22]
Liquid Density 0.742 g/m3 [22] 0.87-0.95 g/m3 [25] 0.79 g/m3 [22]
Boiling Point 25-215 °C [22] 282-338 °C [25] 78.2 °C [24]
Flash Point -45 °C [21] 55 °C [21] 13 °C [24]
Auto-Ignition Temperature 258 °C [23] 316 °C [25] 420 °C [22]
Flame Speed- Stoichiometric
0.33 m/s [22] 0.40 m/s [26] 0.41 m/s [22]
(φ=1)
Adiabatic Flame Temperature 2289 K [23] 2327 K [23] 2234 K [22]
Flammability Limits (vol % in air) 1.2-7.1% [27] 0.6-6.5% [25] 3.5-15% [27]
27
Each fuel has variation and uncertainty in the properties, as different sources report slightly different
values for each metric. This is because the chemical formulas and content in the fuel changes based
on regional regulations, seasonal additives, refinery additives and detergents, water content, and
other factors [28] [29]. One familiar example is winter gasoline blends verse summer gasoline blends.
The winter blend must have a higher Reid vapor pressure (RVP) than summer blends to account for
lower temperatures during the engine startup and to run smoother in colder conditions. The RVP
varies from 9.0 psi down to 7.8 psi for summer months. Since ethanol has a lower vapor pressure,
there is a 1.0 psi allowance for gasoline containing 9% to 10% ethanol [30]. Additionally, blends
such as E85 have a range of 51% to 85% ethanol by volume [17] contributing further to the
uncertainty or variations in the combustion properties. Since gasoline and ethanol have distinct
properties, these factors should be considered when performing risk evaluations.
2.3. Associated Hazards of Traditional Fuels

Among traditional fuels, gasoline has the lowest flash point. It also has a higher vapor pressure,
which typically causes the vapor mixture inside of a vehicles fuel tank to stay above the UFL [31].
Since the vapor pressure of ethanol is lower, in mixtures such as E85, this vapor pressure must be
considered when both designing and adding this fuel to tanks. Based on the ASTM standard
classification, the vapor pressure of E85 mixtures vary between 5.5 to 15.0 psi [17] which might
cause the vapor mixture to fall within the flammable range. Since traditional fuel tanks have a high
fuel concentration and use break-away and structurally weak plastic components, it is necessary to
prevent high pressures from building up due to vapor pressure. While internal fire or deflagration
hazard in a fuel tank might be possible, a fire hazard due to a leak is more of a concern. This is due
to the volatility of gasoline and the hot surfaces of an ICE vehicle, such as the exhaust system. The
associated HRR for gasoline peaks within seconds. Studies that have characterized pool fires for
gasoline and ethanol have shown that as the amount of ethanol in a gasoline-ethanol blend increases,
the heat flux and temperature of the pool fire decreases. A mixture with a higher ethanol content
burns slower and has a lower flame height [32]. Diesel is much less volatile than gasoline and ethanol
with a corresponding higher flash point. At normal ambient temperatures, diesel cannot be ignited
without a very strong ignition source. Therefore, a diesel spill may not present as great a risk as a
gasoline spill on the ground due to the lower chance of ignition from a higher vapor pressure. Yet
leakage into the engine compartment after an impact or failure could still lead to a fire. This due to
the hot surfaces such as the exhaust system or turbocharger assembly that has a large surface area.
Also, diesel and gasoline have comparable auto-ignition temperatures [31].
With the fire and explosion hazards known, characterizing these hazards in tunnels can be addressed
from reviewing and analyzing various incidents. There have been numerous major reported fire
events in tunnels, in which many are caused by vehicle accidents and fires [33].
28
Table 6: Major ICE Road Tunnel Fire Incidents (from [34])
Year Tunnel Information Country Comments
Los Angeles Road Multiple big-rig trucks and passenger vehicles
2007 Tunnel Interstate 5 USA involved. Fire caused concrete spalling. 3
167 m long Built in 1975 fatalities and 10 injuries [33].
Vehicle collision due to shutdown lane and
Melbourne Burnley Road Tunnel
merging. The impact led to a reported fireball
2007 3,400 m Built between 1996 and Australia
and one car bursting into flames. 3 fatalities
2000
[35].
Gleinalm Road Tunnel Head on crash that led to a fire in the middle of
2001 Austria
8.3 km Completed in 1978 the tunnel. 5 fatalities and 4 injuries [36].
Gotthard Motorway Tunnel Head on crash that led to a fire in the tunnel.
2001 Switzerland
Opened in 1980 11 fatalities and 19 injuries [33].
Tauern Road Tunnel
6 km First bore completed in Vehicle crash that led to a fire in the tunnel. 12
1999 Austria
1975. A second, parallel tube was fatalities and 49 injuries [37].
officially opened in 2011
Table 6 shows more recent fire incidents in tunnels due to ICE vehicle collisions. The main cause of
vehicular fires in road tunnels are engine fires, short circuits, ignition of flammable/combustible
materials, collisions, and other defects. Collisions are mostly due to driver error. These major tunnel
fires can result in a HRR above 20 MW. Fire temperatures can exceed 1000 °C and lead to quicker
developing and spreading fires [34].
29
2.4. Pertinent Regulations and Safety Standards
Traditional fuel vehicles have robust safety standards and regulations with regard the vehicle itself
and the roadway structures on which they operate for all ICE vehicles types.
2.4.1. National Fire Protection Association Standard 502

NFPA 502, Standard for Road Tunnels, Bridges, and Other Limited Access Highways, provides fire
protection and life safety requirements as well as design criteria for authorities having jurisdiction
(AHJs) to use in ensuring tunnel safety. Section 7.3.2 states that a tunnel shall be capable of
withstanding the temperature exposure represented by the Rijkwaterstaat (RWS) time-temperature
curve or other recognized standard time-temperature curve that is acceptable to the AHJ, as shown
by an engineering analysis. The assumption is that every part of the tunnel should withstand these
temperature exposures, irrespective of the fire location, ventilation rate, or ventilation type [38].
2.4.2. ASHRAE HVAC Applications Ch. 16: Enclosed Vehicular Facilities (2019)
American Society of Heating, Refrigerant, and Air-conditioning Engineers (ASHRAE) 2019 HVAC
Applications Chapter 16: Enclosed Vehicular Facilities provides guidance on vehicular facilities that
store vehicles and through which vehicles travel. These vehicles can be driven by an internal
combustion engine or electric motors. Ventilation requirements including mechanical systems and
natural ventilation, climate and temperature control, contaminant level control, and emergency
smoke control. Additionally, ventilation concepts including normal operations and emergency
operations are covered in this chapter [39].
2.4.3. NCHRP Guidelines for Emergency Ventilation Smoke Control in

Roadway Tunnels (2017)
National Cooperative Highway Research Program (NCHRP) Guidelines for Emergency Ventilation
Smoke Control in Roadway Tunnels Chapter 2: Road Tunnel Fires provides guidance on fire design
parameters for tunnels. This includes consideration of the geometric parameters of the tunnel, fire
protection features, and response times that leads to decision making using NFPA 502. Chapter 2
provides a framework on how to understand and determine fire and hazardous materials
management in tunnels [40].
2.4.4. NCHRP Synthesis 415: Design Fires in Road Tunnels (2011)

National Cooperative Highway Research Program Synthesis 415: Design Fires in Road Tunnels
provides review on current practices and knowledge for road tunnel fire designs. Additionally, a
survey was completed by numerous transportation agencies and tunnel owners to understand their
experiences and what practices they use for ventilation, fire protection, and detection [41].
30
2.5. Research Summary of Traditional Fuels in Tunnels
This section documents the results of the evaluations regarding ICE vehicle failure in a tunnel.
2.5.1. Experiments
A large variety of tests in tunnels have been conducted in response to the various incidents listed in
Table 6. The tests in Table 7 are used to understand both the characteristics of a tunnel fire (such as
temperature and HRR) and how different ventilation techniques affect these characteristics.
Whether the fire is oxygen rich or oxygen lean will affect these characteristics, and these will drive
the fire time and total damage done.
Table 7: List of Significant Full-Scale Tunnel Fire Tests (from [42])

Name Type Year Country Fire Source Area HRR Ventilation
natural,
Ofenneg Rail 1965 Switzerland Gasoline Pool 24 m2 15-25 MW longitudinal,
semi-transversal
natural,
Zwenberg Rail 1976 Austria Gasoline Pool 24 m2 15-25 MW longitudinal,
semi-transversal
Train wagons,
Mining 1990- natural,
Rapperfijord Norway cars, HGV, 30-40 m2 15-100 MW
Gallery 1992 transversal
calibrated fires
1993- natural,
Memorial Road USA Diesel Oil Pool 60 m2 10-100 MW
1995 transversal
Petrol/diesel
Colli Berici Road 1999 Italy oil pools, car 60 m2 2-5 MW natural
mockup
Calibrated fires, natural,

Rosa Road 2002 Italy 60 m2 2-20 MW
cars, van longitudinal
Pellets, plastic,
Runehamar Road 2003 Norway tires, HGV 32.5 m2 70-200 MW longitudinal
mockup
2.5.1.1. The Rijkswaterstaat (RWS) and Other Time-Temperature Curves
While tunnel design requirements are not specified for metrics such as overpressure, standard fire
curves are used to design for road tunnel safety. This process typically involves selecting an expected
type and size of fire and determining the distribution of temperature exposure to the construction
materials. For example, the International Organization for Standardization (ISO) 834 curve
represents a fully developed fire in a compartment, based on materials found in standard buildings.
The ISO curve is what the World Road Association (PIARC) and the International Tunneling
Association recommend for defining tunnel design criteria for personal vehicles and vans [43].
Because the ISO 834 curve does not represent all materials, especially chemicals which escalate fire
growth, a hydrocarbon curve was developed in the 1970s for use in the petrochemical and off-shore
31
industries and began to be applied to tunnels [44]. The hydrocarbon curve (HC curve) exhibits a
faster fire development and consequently is associated with faster temperature increase than the
standard ISO 834 curve. The modified hydrocarbon curve (Mod. HC curve) exhibits an even faster
fire development and is more conservative than the ISO 834 or the standard HC curve. The Mod.
HC Curve is used for stricter regulations and has a much more severe temperature gradient over the
first few minutes [45].
The RWS curve was developed during extensive testing conducted by the Dutch Ministry of
Transport in cooperation with the Netherlands Organization for Applied Scientific Research (TNO)
in the late 1970s. The RWS curve simulates an accident involving a gasoline tanker loaded with
45,000 liters (45 m3) of gasoline with a fire load of approximately 300 MW released over two hours.
The ISO 834, hydrocarbon, and RWS fire curves are illustrated in Figure 2. Also included in Figure
2 are the ASTM E119, Standard Test Method for Fire Test of Building Construction and Materials, and the
UL 1709, Standard for Rapid Rise Fire Tests of Protection Materials for Structural Steel, time-temperature fire
curves for comparison [46] [47].
1400
1200
1000
Temperature (°C)
800
600
400
200
0
0 30 60 90 120 150 180
Time (minutes)
ASTM E119 ISO 834 HC Curve UL 1709 Mod. HC RWS
Figure 2: Standard time-temperature fire design curves at tunnel structure interfaces
2.5.1.2. The Runehamar Full Scale Tests
In September 2003, a European research program on tunnel safety conducted comprehensive large-
scale fire tests in the abandoned Runehamar road tunnel in Norway [44] [48] [49]. These fire tests
were intended to analyze fires from the cargo of heavy goods vehicle (HGV) trailers, which might
32
contain a large fuel source for fire. The fuel load of HGV and traditional vehicles contain
hydrocarbons and hydrocarbon-based materials (e.g., tires, plastics, etc.) which form very sooty fires
where radiation is the over-riding method of heat transfer to the surrounding materials.
The Runehamar tunnel is approximately 1,600 m long, 6 m high, and 9 m wide. The center of the
fire was located 172 m from one entrance. Two mobile fan units were added to simulate ventilation,
providing a velocity of about 3 m/s (centerline) in the tunnel. Because of the exposure to high
temperatures, the tunnel was protected using PROMATECT–T fire protection boards over 75 m
that were supported with a light steel structure. Fire sprinklers were not installed in the tunnel. Two
small ignition sources, consisting of fiberboard cubes soaked with heptane, were placed within the
lowest wood pallets. A total of four tests were performed with a fire in a semi-trailer set-up. In the
trailer, four different commodities were tested, shown in Table 8. All tests produced time-
temperature developments in line with the RWS curve, as stated in NFPA 502.
Table 8: Commodities used as fuel in the HGV tests (from [49])

Test # Description of the fire load Target Peak HRR (MW)
0 200 L diesel pool fire with a 2.27 m diameter - 6
360 wood pallets (1,200 x 800 x 150 mm)
20 wood pallets (1,200 x 1000 x 150 mm) 32 wood pallets and
1 200
74 polyethylene (PE) plastic pallets (12,200 x 6 PE pallets
800 x 150 mm)
216 wood pallets
20 wood pallets and
2 240 polyurethane (PUR) mattresses (1,200 x 160
20 PUR mattresses
800 x 150 mm)
Furniture and fixtures (tightly packed plastic
Upholstered sofa and
3 and wood cabinet doors, household items) 135
arm rest
10 large rubber tires (800 kg)
600 corrugated paper cartons with interiors
4 wood pallets and
(600 mm x 400 mm x 500 mm)
4 40 cartons with PS 65
15% of total mass of unexpanded polystyrene
cups (1,800 cups)
(PS) cups (18,000 cups) and 40 wood pallets
Test 0 used a 200 L diesel pool fire with a 2.27 m diameter. Four other tests included various
commodities such as wood pallets, mattresses, furniture, and more. Comparing the diesel pool fire
with the other commodities shows that diesel might not have a large peak HRR, but it does quickly
reach a peak and plateaus. While the intent of the diesel pool fire was for a baseline for calibration
and checking instruments, it does show how large commodities are more of a hazard in terms of the
peak HRR compared to the diesel pool fire. This also shows that since the HRR of diesel peaks
quicker than the commodities, a diesel fire might lead to ignition of other fuel loads in a tunnel
during a leak or spillage.
Test 1 with wood pallets and plastic pallets had the highest HRR, with a peak of 200 MW. The HRR
is the most important variable in characterizing the flammability of products and their consequent
fire hazard because it captures the driving force for the fire (i.e., power). Most other variables
(temperature, smoke, toxic gases) are correlated to HRR which can also be linked to the severity of
the fire [50]. Figure 3 illustrates the HRRs for the four large-scale tests. Figure 4 illustrates gas
temperatures in the first test, which had the highest temperatures out of the four tests, compared
33
with four different standard fire curves. The maximum gas temperatures beneath the ceiling were
approximately 1350°C.
Figure 3: Heat release rates from the four large-scale fire tests (from [49])
Figure 4: Test 1 temperature compared with standard fire curves (from [49])
Figure 4 shows the gas temperature from the wood and plastic pallets test shown in Figure 3 (labeled
Tgas,T1) along with the HC (THydrocarbon), RWS (TRWS), RABT ZTV (TRABT/ZTV) and ISO 834 (TStandard)
fire curves. The results from the Runehamar tunnel tests show that non-hazardous, solid
commodities can give a fast increase in temperatures to significantly higher temperatures than had
34
been measured in connection with solid material in tunnel fire tests previously [44]. The
temperatures measured in the post-flame gases downstream of the fire were high and the
measurements indicate that the flaming zone could expand up to a length of 70-100 m. The high
surface temperatures affected the entire tunnel ceiling downstream of the fire causing considerable
spalling of the unprotected tunnel ceiling after the first test, which resulted in considerable rock
debris completely covering the road. The long flames and high temperatures could also cause the
fire to spread to other vehicles.
2.5.1.3. Large Scale Fire Tests in the Second Benelux Tunnel

Testing was conducted in 2000-2001 in the Second Benelux Tunnel in the Netherlands [51]. A total
of fourteen full-scale tests were conducted. The intent of these tests was to determine conditions for
escaping motorists along with how well ventilation, detection, and suppression systems operate. A
multitude of measurements were collected such as temperature, radiant heat, smoke velocity, smoke
density, and HRRs. The first four tests were pan fires using a mixture of 60% n-heptane and 40%
toluene by mass. The next six tests consisted of cars and covered truck loads, with the loads
consisting of 800 kg of wooden pallets and four tires. Ventilation through the tunnel varied stepwise
from 0 m/s to 6 m/s. The final set of tests determined the effect of a deluge sprinkler system and
examined the effect of delayed activation, effectiveness of the sprinkler system, and if the sprinkler
system would prevent fire spread to other vehicles.
Figure 5: Test Overview for Second Benelux Tunnel (from [51])
Figure 5 shows the test configurations in the Second Benelux Tunnel. The report does not specify if
any fuel was in the vehicles during the testing. Since the intent was to compare ventilation and
obstructions of ventilation, the total commodity fuel load stayed nominally the same between tests.
35
Figure 6: Fire development for small truck fires (from [51])
Figure 6 shows the effect of ventilation on the HRR of a small truck fire. As the ventilation
increases, the peak HRR increases, but the total time from ignition to extinction decreases. This type
of testing helps provide data for determining emergency ventilation controls.
2.5.2. Modeling
Different models for tunnel fires have been created using both computational fluid dynamic (CFD)
programs and using other tools such as the zoning modeling tool Consolidated Model of Fire and
Smoke Transport (CFAST), an open-source software package National Institute of Standards and
Technology (NIST). CFAST is a zoning model that is used to evaluate the evolving distribution of
smoke, fire, gases, and gas temperatures [52].
2.5.2.1. Simulation of Tunnel Fires Using a Zone Model

Modeling by Chow et al. [53] used CFAST to understand how zoning the tunnel into multiple
sections compares with other similar CFD studies and experimental data. Various modeling has
been conducted to understand the effect of smoke movement and temperature effects [53].
Different zoning methods were used with CFAST 2.0. Five different fire simulations were used for
each zoned model: wood fire, passenger train fire, subway coach fire, truck fire, and a school bus
fire. Smoke layer and temperature predictions were created with the different zoned sections: single
compartment along with multiple compartment configurations designated as 2-room and 3-room.
Figure 7 below gives a visual for each of these configurations as well as the spatial relationship
between them.
36
Figure 7: Compartment Configurations (from [53])
Figure 8 and Figure 9 below show the smoke temperature and layer height for the various zoned
configurations for the truck fire. The colored and numbered circles are to help identify the smoke
temperature and layer heights based on which configuration was used (single compartment, two-
room, or three-room). The averages for the two and three-room configurations are very close to the
single compartment configuration for both the truck and bus fire characteristics.
37
1
2
3
3 2
2
1
1
1
1
1
2
1
2
3
Figure 8: Smoke Characteristics Prediction of a Truck Fire (from [53])
38
1
3
2
1
1
1
2
1
2
3
Figure 9: Smoke Characteristics Prediction of a School Bus Fire (from [53])
The CFAST models were validated with a tunnel fire experiment in Norway and compared with
other models. It was determined the CFAST model gives similar results to other known fire models
such as Consolidated Compartment Fire Model (CCFM.VENTS) on predicting the smoke
temperature and layer interface heights. The CFAST zone model is a good baseline prior to using a
more computationally heavy CFD program. By using a zoning technique as described by Chow et al.
[53], the fire environment can be predicted. This can be used to design tunnel ventilation systems
and suppression systems and where to add heat and smoke detecting devices based on the smoke
temperature and spreading characteristics.
2.5.3. Analysis
Different studies have analyzed the various parameters of tunnels and how they affect the fire and
smoke characteristics. A study by Haghighat et al. [54] goes over the effect of ventilation and tunnel
cross-sectional area on fluid properties down-stream of a fire using CFD data. A study by Shafee et
al. [55] determines how tunnel inclination, blockage, and ventilation effects the HRR of a fire.
2.5.3.1. Determination of critical parameters in the analysis of road tunnel fires

The intent of this study was to use CFD model data to determine fluid characteristics downstream
from a fire in a tunnel. The selected characteristics to study in this analysis were the average
temperature, the average density, the average viscosity, and the average velocity. How different
39
parameters interact with these response variables is determined [54]. The HRR was selected based
on NFPA 502 recommended peak HRR for a bus or van. The HRR varied between 10 MW and up
to 30 MW. It was determined the overall physical fire size, regardless of the HRR and fire intensity,
did not influence the response variables downstream of the fire. The tunnel dimensions do affect the
average temperature and density up to 220 m downstream and are insignificant at 400 m
downstream. The inlet velocity does influence the average temperature and the average density
drastically at 20 m and 400 m downstream of the fire source. The average velocity downstream of a
fire is dependent on both the inlet velocity and the HRR.
Figure 10: Tunnel Dimension Effect on Average Temperature (from [54])
This study by Haghighat et al. [54] proves to be useful for understanding what parameters effect the
overall characteristics downstream of a tunnel fire for different burning vehicles. This helps analyze
tunnel design, ventilation sequences, and firefighting techniques.
2.5.3.2. An analysis of tunnel fire characteristics under the effects of blockage

Ethanol pools were used in a study by Shafee et al. to understand how blockage, slopes, and
ventilation rates effect the overall HRR, burning rate, and smoke back-layering [55]. The downhill
and uphill slopes effect the fire-induced buoyancy effects. This study helps understand the downhill
inclination effect on critical ventilation velocity and compares the results of this inclination with a
horizontal study. Using a small-scale or reduced-scale test, these different parameters could be
changed for each test.
40
Figure 11: Mean Effect on HRR Flux (from [55])
The plot above is from Shafee et al. [55] and shows the overall effects of the inclination, blockage
ratio, and ventilation velocity on the HRR. This study shows blockage and inclination do influence
the overall HRR, but the ventilation velocity is a more important factor (45% overall mean effect) in
controlling the overall HRR. The higher the mean effect, the more sensitive the HRR is to a smaller
increase in that specific variable. The tunnel blockage effect accounted for 25% of the mean effect
and the inclination accounted for 19%. The other factors that contribute to 11% mean effect include
the blockage distance from the fire. The HRR flux decreases as that blockage distance increases
from 5 cm to 30 cm. But there is less than a 0.2 MW/m2 change with a 25 cm increase in the
blockage distance. An increase in the ventilation velocity from 0.0 to 1.5 m/s multiples overall HRR
flux by four from 0.4 MW/m2 to 1.6 MW/m2.
41
2.6. Traditional Fuels Research Gaps
Traditional fueled vehicles have been thoroughly studied for many years, with studies on the various
time-temperature curves and how they are applied. Additionally, different studies have supported
model development as well as understanding ventilation requirements. This section went through a
variety of studies, but there are many more available for traditional fuels. As different fuel blends
become available, such as ethanol and biodiesel, more work should be completed to further
understand the consequence of these fuels. Additionally, as engine technology advances, combustion
processes may require future work to help determine how higher temperatures and leaner mixtures
might be a scenario that leads to a fire. This is discussed further in the research gaps below.
The following criteria were evaluated to determine where research gaps may exist regarding
Traditional Fuel ICE powered vehicles in tunnels.
1. Scenario Identification
2. Failure Modes
3. Consequences
4. Validation
Scenarios that lead to failure modes have been determined as engine bay fires initiated by collisions
or various vehicle defects (such as short circuits) as well as flammable or combustible materials
igniting on the vehicle. A variety of incidents involving ICE vehicles in tunnels has led to a greater
understanding of what the failure modes are. Collisions are mostly due to driver error while other
initiating events are either due to collision or defects. Experiments such as those listed in Table 7
show the failure modes of various fuel spillages along with commodities that have been
characterized in tunnels.
The consequence of these failures has led to major tunnel fires that can result in a HRR above 20
MW. Fire temperatures can exceed 1000 °C and lead to quicker developing and spreading fires [34].
Using real scenarios such as vehicle crashes or malfunctions has driven studies and experiments on
tunnel fire characteristics of ICE vehicles. Some tunnel studies just involve fuels such as diesel and
gasoline [49] [55] while other studies include other commodities or whole vehicles [51]. Typically,
experiments are compared with the standard modeled time-temperature curves in order to validate
the models. This comparison also helps understand how the model characteristics are different than
that of the experiment. Preliminary findings from traditional fuel fires are shown in the study by
Shafee et al. [55] in which a variety of parameters such as ventilation were increased to understand
how the heat release rate changed (shown in Figure 11). Additionally, Lemaire and Kenyon [51]
show that as the ventilation increases, the HRR also increases but the total time of the fire decreases.
Various sizes or classifications of vehicles have been involved in both real scenarios as well as testing
to further understand how a passenger vehicle compares to a cargo-type vehicle regarding tunnel
fires. The properties of gasoline, diesel, ethanol, and blends of these fuels are known and well
regulated. For example, vapor pressure is closely regulated to ensure emissions and engine startup
are controlled. This is important when designing fuel tanks systems to ensure a flammable mixture is
not developed above the liquid fuel. Typically, the concentration above the liquid fuel exceeds UFL.
Most of the failures and consequences from ICE vehicle fires in tunnels are well known through
actual scenarios and further understood through testing and modeling. Yet the following research
gaps have been identified:
42
 As ICE emission technology evolves, such as using technologies such as particulate filter
regeneration, recirculating exhaust gases, and running vehicles with a leaner fuel mixture,
further study of the effect on exhaust system component temperatures may be needed. Fuel
spillage might accumulate directly under the vehicle and it is important to understand how
higher exhaust temperatures might lead to potential ignition sources.
 Future technologies, such as compression-ignition gasoline engines or advanced forced air
induction, might also lead to increased exhaust component temperature.
 As engine components advance, the ability to run hotter from leaner mixtures and higher
pressures could lead to an increase in engine bay temperatures, which might increase
potential ignition sources.
43
44
3. BATTERY ELECTRIC VEHICLES
3.1. Overview of Technology

Battery electric vehicles (BEVs) offer consumers an alternative transportation option to
conventional internal combustion engines (ICE). In these vehicles, the internal combustion engine
and powertrain system is replaced by an electrical powertrain. Plug-in hybrid electric vehicles
(PHEVs) offer an electric motor powered by batteries with an onboard combustion engine that can
charge the battery system when it gets depleted. The battery is charged through a cable connection
when the vehicle is parked and not operating. Generally, BEVs offer many benefits such as high
efficiency, no harmful tailpipe emissions, good performance, and low cost “refueling” (electricity
through charging) [56].
BEVs utilize an electrochemical storage system in which energy is converted between electrical and
chemical energy through a reversible process. This is done with a battery system consisting of
lithium-ion or lithium secondary cells. The battery cells generally consist of a case, an anode and a
cathode which are electrodes, a separator, and an electrolyte. Each of the different battery types
consist of different materials for the casing, electrodes, and electrolyte, which result in varying
performance and cost.
There are several different battery technologies in BEVs that offer a range of energy density, power
density, cycle life, and calendar life. The lithium-ion or lithium secondary battery technology is
regarded as the most promising because of its high energy density, high efficiency, and long lifespan.
Unlike lithium or lithium primary cells that are disposable, lithium-ion or lithium secondary cells are
rechargeable, making them suited for BEVs.
Table 9: Comparison of Lithium and Lithium-Ion Cells
Lithium (Primary) Lithium-Ion (Secondary)
One-time use Rechargeable
Smaller light weight applications Custom, larger scale application
Metallic lithium as the Anode Porous carbon as Anode
In addition to lithium-ion batteries, there are also lead acid, nickel metal hydride (Ni-MH), nickel
cadmium (Ni-Cd), and sodium nickel chloride (Na-NiCl2) battery technologies that have been
used/considered for use in BEVs. However, there are limitations associated with each of these
battery types that make them less desirable for future designs than the lithium-ion battery due to
lower specific energy or specific power capabilities [56]. There are several different types of BEVs
currently in operation, including buses, trucks, vans, and cars.
3.2. Properties of Lithium-ion Batteries

As discussed previously, lithium-ion batteries are considered the most promising battery technology
due to the high energy density and light weight compared with other battery technologies mentioned
above. This is further illustrated in Figure 12 below:
45
Figure 12: Battery Technology Energy Densities (from [57])
There are several different types of electrodes and electrolytes used in lithium-ion batteries.
However, Figure 13 shows the general chemistry of the battery. As shown, during recharging
operations, the positively charged lithium travels from the cathode to the anode through the
electrolyte then combines with the charging electrons, which forms a lithium atom that gets
deposited between carbon layers. This process is reversed during discharging activities [58].
Figure 13: Lithium-ion Battery Chemistry (from [59])
Technology improvements for lithium-ion battery types are the most promising. A battery system is
broken down into various sub-systems and sub-components as follows:
46
 Lithium-Ion (Lithium Secondary) Cell
o Varies capacities, chemistries, form factors, manufacturers, etc.
 Module
o Houses cells in various configurations
 System Casing/Housing
o Houses complete battery system comprising of modules
 Battery Management System
o Thermal management
o Determines charging/discharging routines
o Monitors system health
o Controls State of Charge (SOC)
Different cathode chemistry varieties of the lithium-ion battery offer different battery characteristics:
lithium cobalt oxide (LCO), nickel cobalt and aluminum (NCA), nickel manganese cobalt (NMC),
lithium iron phosphate (LFP), lithium manganese oxide (LMO), lithium polymer and lithium ion
phosphate offer different advantages and disadvantages in terms of power, energy density (Wh/kg),
specific volume (m3/kg), safety, and calendar and cycle life. Additionally, the various electrolytes
such as ethylene carbonate (EC), dimethyl carbonate (DMC), ethyl methyl carbonate (EMC), diethyl
carbonate (DEC), propylene carbonate (PC), and methyl propyl carbonate (MPC) offer different
performance characteristics and have varying safety metrics. These are discussed in Section 3.3.2.
3.3. Associated Hazards

A catastrophic failure of the battery system can occur due to manufacturing defects, thermal abuse,
electrical abuse, or mechanical damage.
Figure 14: 18650 Cell Defect (from [60])
Figure 14 shows defects in an 18650 cell in both the cathode and the anode. An 18650 cell is a
designation for a cylindrical cell that compares to a AA battery in terms of the form factor.
Manufacturing defects could cause cells to have shorter overall life cycles, and when used in a
47
battery module, this could cause an imbalance of charge amongst cells and premature cell failure.
Thermal abuse can be caused by failing or malfunctioning cooling or thermal management systems.
Electrical abuse might occur if the battery management system incorrectly charges/discharges/cycles
the cells. Mechanical damage could occur during an impact. Other mechanical abuse such as long-
term vibrations might cause cells to have excess wear and create an internal or external short circuit.
Table 10 below gives an overview of the types of abuse testing that each standard provides guidance
on by Ruiz et al. [61]. These are general practices and examples of standards that should be used to
better understand the safety implications of these battery systems at various levels. The letter
designations are the following: C-Cell level testing, M-Module level testing, P-Pack level testing and
V-Vehicle level testing.
Table 10: Overview of Test Standards for Lithium-ion Battery Abuse (from [61])
SAE SAE UL 2580 FreedomCAR
Test J2464 [62] J2929 [63] [64] [65]
Mechanical Shock C M P C M P V C M P - M P
Drop - - P - - P - C - P - - P
Penetration C M P - - - - - - - C M P
Mechanical Immersion - M P - - P - - M P - M P
Crush/crash C M P - - P V C M P C M P
Rollover - M P - - P - - - P - M P
Vibration - - - C M P - C M P C M P
External Short
Circuit C M P - - P - C M P C M P
Electrical
Over
Charge/Discharge C M P - - P - C M P - M P
Thermal Stability C - - - - - - C - - C M P
Thermal
Environmental Shock/Cycling C M P C M P - C M P C M P
Overheat - M P - - P - - - - - - -
Fire - M P - - P - C M P C M P
Emissions C M P - - P - C M P C M P
Chemical
Flammability C M P - - P - C M P C M P
Certain failure modes within lithium-ion cells can lead to an exothermic reaction within the sealed
cell. Examples of these failure modes include mechanical damage, manufacturing defect,
overcharging/discharging, and/or over cycling. These reactions can lead the cell into thermal
runaway. This is due to a series of exothermic reactions that increase the cell temperature, resulting
in internal generation of gases. This builds pressure in the cell and can ultimately lead to rupture and
a release of vent gas. Propagation of failure from the cell-to-cell chain reactions may occur due to
48
the thermal energy release from the failed cell. This can cause the entire module, pack, and system to
go into thermal runaway. Studies such as the one conducted by Lopez et al. [66] in 2015 studied
18650 cells and prismatic cells in order to determine separation distances that would prevent cell-to-
cell thermal runaway propagation. Based on various cell and tab configurations, it was recommended
to maintain at least 2 mm of space in between cells to minimize the chance of thermal runaway
propagation.
There are examples of safety devices to help prevent failures that lead to thermal runaway. One such
example is the internal positive temperature coefficient (PTC) current limiting device used in the
18650 cells in a Tesla Roadster [67]. The only downside is that once the PTC activates, the state of
charge can no longer be measured when disposing of this cell. A study by Balakrishnan et al. [68] in
2006 goes over other safety mechanisms associated with lithium-ion batteries. Safety vents relieve
internal pressure build up in a cell that has failed. Built-in internal thermal fuses will melt when
excess current flows through the cell, leaving the cell permanently disabled. Cell charge balancing
can also be controlled by measuring and terminating charging/discharging to a specific cell in a
module. This helps protect cells in a module from being overcharged/overdischarged. These are
some examples of ways to help mitigate or slow thermal runaway between cells.
Figure 15: BEV Failure Event
Figure 15 shows how system design can lead to the generation of a flammable mixture and an
eventual fire or deflagration from a BEV failure. In the event a failure occurs in a BEV and based on
the confinement, the vent gas can accumulate. Based on factors such as gas species, gas
concentration, release rates, and total vent gas volume, a flammable mixture can occur. With
immediate ignition, a fire may occur that consumes surrounding oxygen and can lead to under-
ventilated fire extinction. More vent gas can be produced from other cells that have failed. Based on
confinement, this will determine if a fire scenario or deflagration/explosion scenario will exist. If the
vehicle is in a well-ventilated area, with immediate ignition, a fully developed fire may occur. If vent
gas can continue accumulating in a confined space, secondary ignition may cause a deflagration.
There might be both a fire and deflagration that occurs or just one of them.
49
3.3.1. Vent Gas Hazards
It is already well known that battery systems such as lead-acid batteries can produce off-gas or vent
gas such as hydrogen. What makes lithium-ion batteries unique is the large variation in not only the
species of gases produced when venting, but also the variation in volume production and rate.
Figure 16: Battery Vent Gas Species Compositions SOC (from [69])
Figure 16 by Baird et al. [69] shows the vent gas composition for various cell chemistries and states
of charge (SOC). In some cases, the state of charge is greater than 100%, indicating that the battery
has been overcharged (where current is applied and there is an increase in voltage over the nominal
50
capacity). This could simulate an instance where the battery management system failed and allowed
the charging system to continue even after charging is complete. The vent gas released from these
failed lithium-ion cells contain a flammable gas mixture that contains various species such as
hydrogen (blue), carbon monoxide (orange) which is also toxic, various hydrocarbons such as
methane and propane (green), and carbon dioxide (red) which is an asphyxiant. The species
composition and vent gas volume production vary for other tests based on SOC, cathode
chemistries, cell form factor, capacity, etc. In addition, Somandepalli et al. [70] reported an estimated
vent gas production rate of 0.32 L/Wh from a 7.7 Wh cell. This is at the cell level, so further
experiments would need to be studied to determine if this estimated vent gas production holds on a
larger scale. Current BEVs have a capacity of 30 kWh to 100 kWh [71]. The rate at which the battery
cells fail in a module is also important to understand. This will ultimately lead to the gas production
rate which will vary based on the system parameters and the failure mode.
A deflagration hazard can exist due to lithium-ion cell failure. One example of this occurring was in
a substation explosion in Surprise, AZ, where a lithium-ion battery energy storage system failed,
releasing this flammable vent gas. The flammable vent gas ignited after a delay, and the resulting
shock wave threw multiple firefighters back [72]. Explosion hazards are based on five parameters
that must be present: oxidizer, ignition source, fuel, confinement, and dispersion of the gas. The
confinement of the gas and the amount of gas produced by a BEV battery failure will determine the
severity of the deflagration/explosion. These parameters will affect key deflagration metrics such as
flame speed and maximum overpressure.
3.3.2. Fire Hazards

In addition to the deflagration/detonation hazard, BEVs present a unique fire hazard as well due to
the potential for cascading failure. When a lithium-ion cell fails, flammable gases are usually ejected
due to the liquid electrolyte reacting during the combustion process. These gases can remain
unburned or might be ignited and burned as a jet flame. This jet flame from one cell in the battery
system can heat other cells, converting the potential chemical energy rapidly into thermal energy. If
there is enough oxygen present, fuel from the flammable gases along with heat provided by the
failed cells (as well as the potentially flammable packaging) could lead to a fire hazard.
Figure 17: Cell Component Breakdown by % Mass (from [73])
51
Figure 17 gives the material breakdown of each part of the cell in a study by Golubkov et al. [73].
Below, in Table 11, further breakdowns are provided for comparison.
Table 11: Cell Deconstruction Breakdown by % Mass
Nominal
Form Cathode Cathode Anode Electrolyte Separator Packaging
Reference Capacity
Factor Chemistry (%) (%) (%) (%) (%)
(Ah)
Ribiere
(2012) [74]
Pouch 2.9 LMO 44.0 35.0 11.0 2.0 6.0
Somandepalli
(2014) [70]
Pouch 2.1 LCO 42.4 34.9 9.5 6.4 6.8
Golubkov
(2014) [73]
18650 2.6 LCO/NMC 45.1 24.8 10.4 2.7 16.9
Golubkov
(2014) [73]
18650 1.5 NMC 33.4 31.8 10.2 3.2 21.3
Golubkov
(2014) [73]
18650 1.1 LFP 30.3 23.3 16.4 3.1 26.9
Not only is there the hazard of the flammable gases, but the various materials used to construct
lithium-ion cells (such as nylon and polypropylene) can add to the fuel load. Table 11 shows the cell
deconstruction by Ribiere et al. [74] in 2012, Somandepalli et al. [75] in 2014, and Golubkov et al.
[73] in 2014. There are five major components: cathode, anode, electrolyte, separator, and the
packaging.
The specific materials can be used to understand the heat of combustion and what the total heat
release will be at the cell level. For example, Somandepalli et al. [75] report nylon and polypropylene
as some of the various assumed compounds for the packaging. The actual heat of combustion
various for those two compounds are from 27.1 kJ/g for nylon to 38.6 kJ/g for polypropylene from
Quintiere et al. [6]. In addition to the various well-known materials used in the packaging of the
lithium-ion cells, the electrolyte composition is also required to understand the complete fuel load.
This component is not well known due to the mixtures not being published consistently from
manufacturers. The study by Somandepalli et al. [75] provides a value of 19.31 kJ/g for the heat of
combustion of the electrolyte. A study by Zhang et al. [76] has various electrolyte heat of
combustion values, with 13.2 kJ/g for EC, 20.9 kJ/g for DEC, and 14.5 kJ/g for the DMC
electrolyte. This shows the range that the heat of combustion for the electrolyte might fall under,
which can be used to estimate the total fuel load.
52
Figure 18: HRR Curves for various Electrolyte Compositions and Mixtures (from [77])
The HRRs of each electrolyte and various mixtures are all unique. A study by Eshetu et al. [77]
shows how the HRR peaks and characteristics vary for each individual and mixture of electrolyte.
The total heat released for the electrolyte mixture is based on the combustion conditions, mixture
ratios, and electrolytes in the mixture and can vary from 12 to 24 kJ/g, which is much lower than
some of the packaging materials.
A study by Larsson et al. [78] looked at toxic emissions from battery fires. Various tests of a variety
of cells at a SOC from 0% up to 100% were conducted to measure the HRR and the hydrogen
fluoride (HF) and phosphoryl fluoride production. The hydrogen fluoride production ranged
anywhere from 15 to 198 mg/Wh. Larsson noted the immediately dangerous to life or health
(IDLH) level for HF is 25 mg/m3 (30 parts per million) and the lethal 10 minute HF toxicity value
(Acute Exposure Guideline Levels 3) is 139 mg/m3 (170 ppm). Table 12 below gives ranges based
on the exposure time for the different AEGL levels from [79].
53
Table 12: Summary of AEGL Values (from [79])
If the production rate of 198 mg/Wh is scaled for a 100 kWh BEV fire, the results could be as much
as 20 kg of HF produced. Based on confinement and ventilation the IDLH or lethal threshold could
be reached.
54
BEVs have robust safety standards and practices regarding the battery system, the vehicle itself, and
the roadway structures on which they operate in.

protection and life safety requirements as well as design criteria for AHJs to use in ensuring tunnel
safety. Section 7.3.2 states that a tunnel shall be capable of withstanding the temperature exposure
represented by the RWS time-temperature curve or other recognized standard time-temperature
curve that is acceptable to the AHJ, as shown by an engineering analysis. The assumption is that
every part of the tunnel should withstand these temperature exposures, irrespective of the
ventilation rate, type of ventilation, or location of fire [38].
ASHRAE 2019 HVAC Applications Chapter 16: Enclosed Vehicular Facilities provides guidance on
vehicular facilities that store and/or through which vehicles travel. These vehicles can be driven by
an internal combustion engine or electric motors. Ventilation requirements including mechanical
systems and natural ventilation, climate and temperature control, contaminant level control, and
emergency smoke control. Additionally, ventilation concepts including normal operations and
emergency operations are covered [39].

NCHRP Guidelines for Emergency Ventilation Smoke Control in Roadway Tunnels Chapter 2:
Road Tunnel Fires provides guidance on fire design parameters for tunnels. This includes
consideration of the geometric parameters of the tunnel, fire protection features, and response times
that leads to decision making using NFPA 502. This is chapter provides a framework on how to
understand and determine fire and hazardous materials management in tunnels [40].
3.4.4. UL 2580 Standard for Safety Batteries for Use in Electric Vehicles
UL Standard 2580 Batteries for Use in Electric Vehicles covers requirements for electrical energy storage
assemblies including battery packs, sub-assemblies, and modules that make up the main assembly for
electric-powered vehicles. The electrical energy storage assemblies are tested to determine the ability
to withstand abuse testing and conditions. The manufacturer’s specified charging and discharging
parameters are used to test the assemblies and modules at the specified temperatures [64].
3.4.5. SAE J2464 Electric and Hybrid Electric Vehicle Rechargeable Energy
Storage System Safety and Abuse Testing
Society of Automotive Engineers (SAE) J2464 Electric and Hybrid Electric Vehicle Rechargeable Energy
Storage System Safety and Abuse Testing gives guidance on abuse testing performed to characterize
rechargeable energy storage systems and response to off-normal conditions and environmental
impacts. The response information collected can be used to determine various hazards due to abuse
conditions. This data can be used to create hazard mitigation efforts for specific energy storage
designs [62].
55
3.4.6. SAE J2929 Electric and Hybrid Vehicle Propulsion Battery System Safety
Standard - Lithium-based Rechargeable Cells
SAE J2929 Electric and Hybrid Electric Vehicle Rechargeable Energy Storage System Safety and Abuse Testing
gives guidance on minimum safety criteria for the complete battery system, including cells, modules,
packs and complete battery system. This includes understanding how physical support, enclosure,
thermal management and electronic controls operate. This standard focuses on evaluating the
battery system alone [63].
3.4.7. FreedomCAR Battery Test Manual for Power-Assist Hybrid Electric

Vehicles
FreedomCAR Battery Test Manual for Power-Assist Hybrid Electric Vehicles gives guidance testing for
cycle life behavior and performance for batteries in hybrid electric vehicle applications. These tests
are applicable to the full battery system as well as using scaling to apply to cells and modules [65].
56
3.5. BEV Research Summary in Tunnels
This section documents the results of evaluations regarding BEV failure in a tunnel.
3.5.1. Experiments
A large variety of experiments for BEVs in tunnels have been conducted. Additionally, real world
failures have also occurred. These different reported incidents and experiments conducted provide
an understanding of both what might cause a failure and the characteristics of that failure.
3.5.1.1. Reported BEV Failures and Incidents

Various instances of failed BEVs have occurred due to crash damage, factory defects, and battery
management issues while charging. One example of crash damage was during the 2011 Chevrolet
Volt crash testing. After the test was completed, the vehicle caught fire over the weekend when no
lab personnel were present. The Chevrolet Volt had been involved in a New Car Assessment
Program (NCAP) pole test three week prior to the fire. This is an impact test with a solid pole for
crash testing ratings. During the investigation, it was determined the battery was damaged but only
over time did it catch fire [80]. Additionally, a Tesla Model 3 collided with a parked tow truck in
Moscow, Russia. The BEV ignited after a short time, though the time is not specified. During the
fully developed vehicle fire there were two distinct ‘bursts’ that occurred [81].
One instance of a BEV failure due to either a defect or battery management error occurred in Los
Angeles in 2018. A Tesla Model S was driving slowly through traffic and began to vent gases and the
vehicle ignited. The fire department responded and consulted Tesla via telephone to advice on how
to safely contain the vehicle. Approximately 300 gallons of foam and water was used to extinguish
the flames and the vehicle was transported to Tesla for inspection [82]. A more recent example is a
Tesla Model S bursting into flames while parked in a Shanghai parking garage [83]. An image of this
is shown in Figure 15. Another example comes from a BMW i8 hybrid that began venting inside of a
Dutch showroom. The car was quickly driven outside where it was then immersed in a giant tank of
water where it remained for 24 hours [84].
3.5.1.2. Fire Analysis of BEVs in a Road Tunnel

An experimental analysis was performed to evaluate the worst-case effects of a damaged electric
vehicle battery in a road tunnel [85]. The batteries were evaluated without the chemical and electrical
safety modules or the protective battery housing to eliminate influencing factors on the test. Because
lithium-ion batteries are the most promising battery technology, these batteries were the focus of
these experiments. The consequences of a large battery fire are large energy release, leakage, and gas
venting. Table 13 shows the characteristics of the lithium-ion batteries used in this experimental
analysis [85].
57
Table 13: Lithium-ion Battery Characteristics (from [85])
Characteristic Description
Number of Cells 96 cells (8 modules)
Anode Active Material Graphite
Cathode Active Material NMC
Electrolyte Lithium hexafluorophosphate (LiPF6)

Energy (gross) 33.182 kWh
Energy (net) 27.2 kWh
Specific Energy (gross) 0.14 kWh/kg
Specific Energy (net) 0.12 kWh/kg
Thermal Runaway From 210 °C typical. High charge promotes thermal runaway.
Four test scenarios were evaluated to induce a fire. Three scenarios focused on mechanical damage
to the battery: wedge-shaped penetration with an explosively accelerated steel plate, blunt impact
with an explosively accelerated steel plate, and central puncturing with an explosively formed
projectile. The fourth scenario focused on thermal stress in which the battery module was exposed
to a propane gas fire until the module caught fire. Illustrations and additional descriptions can be
seen in Table 14. These experiments showed that the fire hazards of BEVs are similar to those of
conventional vehicles. However, a different hazard is introduced in the release of potentially more
severe chemical aerosols such as cobalt, lithium, and manganese which are toxic [85].
Table 14: Test Scenarios (from [85])
58
Figure 19 shows the layout of the test site where the experiment was conducted, indicating the
battery location (labeled Test site), ventilated section (colored blue), closed bypass section (pink),
and measuring site. The tunnel had a cross-sectional area of 56 m2 at the test site and 43 m2 at the
measuring site. Due to the difference in a fire involving a BEV when compared to a conventional
ICE vehicle, pollutants and aerosols were measured in addition to thermal parameters. The bypass
to the main ventilation duct was closed during all tests to control air flow and prevent dilution [86].
Figure 19: Test tunnel area layout (from [86])
Table 15 gives the release quantities for the four different tests listed above. Also listed is the total
test duration. It was discussed that toxic aerosols such as cobalt, lithium, and manganese are
released, which is unlike traditional ICE vehicle fires. One thing the authors noted was while
hydrogen fluoride gas was expected to be detected, no large quantities were measured. An additional
note was that the three mechanical damage tests caused each cell in the battery to go into thermal
runaway almost simultaneously, while the thermal test caused a chain reaction of thermal runaway
between cells.
59
Table 15: Release Quantities (from [85])
Parameter Test 1 (wedge) Test 2 (plate) Test 3 (puncture) Test 4 (fire)
PH3 [g] < 0.4 --- < 0.4 ---
F- as HF [g] 1.1 3.1 <1 < 0.5
PO4-P as H3PO4 [g] < 1.5 < 1.5 11.3 <1
Co [g] 457 567 190 364
Li [g] 107 124 42 92
Mn [g] 445 536 184 349
F- Aerosol [g] 152 160 68 126
NO [g] <1 1.1 <1 1.5
NO2 [g] <1 <1 <1 <1
CO [g] 76 181 97 141
CO2 [g] 8500 6000 2000 7800
TVOC [g] 20 196 93 32
∑ Aromate [g] 1.6 8.6 3.2 3.1
Benzene [g] 1.1 3 1.6 1.7
Toluene [g] 0.2 1.1 0.5 0.4
Xylene [g] 0.1 0.6 0.3 0.2
Styrene [g] 0.1 3.0 0.5 0.6
Duration 16 min 21 min 16 min 26 min
3.5.1.3. Electric Vehicle Crash and Fire Damage

An experiment was performed to show the effect of both crash and fire damage using a Tata Indica
GLX Electric Vehicle that uses a 26 kWh, 12 module NMC lithium-ion battery [87]. Overall, the
goal was to better understand if a BEV battery system will ignite with heavy crash damage and how
much water is required to extinguish a BEV fire after thermal runaway. The first test was a drop test
from 20 meters that simulated a 70 km/h rear impact. After seven minutes, the vehicle had visible
flames appear. After 2.5 hours of free burning, the temperatures around the battery containment
ranged between 310°C to 540°C. This test showed that with severe mechanical damage, the battery
can ignite. It is concluded that many factors such as the angle and energy of impact will affect the
outcome, and this is good knowledge to have for emergency responders and even towing companies
hauling wrecked or damaged EVs through tunnels.
60
The next test used a propane flame under the vehicle to attempt to ignite the battery pack externally.
The windows were rolled down for this test. After ten minutes, ignition occurred, and two attempts
had to be made to extinguish the fire. The first attempt used 100 L of water to extinguish the flames.
After a short time period, re-ignition occurred that led to a fully developed car fire. It took an
additional 550 L of water to fully extinguish the second fire, which was declared extinguished when
temperatures were no longer increasing. The major observation with this second test was the actual
battery pack did not go into thermal runaway. After investigating the battery pack, it was determined
that while the external fire did not cause the battery pack to ignite, it did ignite other combustibles of
the vehicle. The fire had characteristics comparable with a traditional ICE vehicle. This is unlike the
mechanical damage that caused the cells to go into thermal runaway and ignite [87].
3.5.1.4. Comparison of the Fire Consequences of a BEV and ICE

Testing to compare ICE and BEV fires was performed in a 50 m long, 3.5 m high fire gallery by the
National Institute of Industrial Environment and Risks [88]. There was a total of five different fire
tests performed on the following:
1. Modular assembly of battery cells to represent a portion of an EV battery
2. The same battery assembly with firefighting operation
3. A full battery pack with late firefighting attempts
4. EV with a fully charged battery
5. An analogous diesel vehicle with a full fuel tank
A 6-kW burner was used for an ignition source for each test, and an exhaust system with a
volumetric flow of 25,000 m3/hr would collect the products for analyzing. The HRR, total heat
released (THR), and heat of combustion (Δhc) were captured for two different EV vehicles and ICE
vehicles presented in Table 16 below:
Table 16: BEV vs. ICE Fire Characteristics (from [88])
BEV BEV ICE ICE
Manufacturer 1 Manufacturer 2 Manufacturer 1 Manufacturer 2
Nominal Voltage (V) 330 355 N/A N/A
Capacity (Ah) 50 66.6 N/A N/A
Energy (kWh) 16.5 23.5 N/A N/A
Mass (kg) 1122 1501 1128 1404
Mass Loss (kg) 212 278.5 192 275
Max HRR (MW) 4.2 4.7 4.8 6.1
THR (MJ) 6314 8540 6890 10000
Δhc (MJ/kg) 29.8 30.7 35.9 36.4
HF Production (g) 1540 1470 621 813
61
The total hydrogen fluoride production for the BEV was about double that of the ICE vehicle. This
was due to the combustion of the lithium-ion batteries and is noted in other cell-level studies [78].
Additionally, the heat of combustion for the EV was less than the ICE vehicle. This could be due to
the lithium-ion battery electrolyte having a lower Δhc, which lowers the overall heat of combustion
of the vehicle. The intent of this test was to use the data as an input for modeling toxic gas
dispersion and the thermal effects in a confined space such as a tunnel or underground parking
facility [88].
3.5.2. Modeling
Modeling for BEVs in tunnels has not been explicitly researched and developed. Currently, the
majority of modeling for batteries takes place to understand cell failures and modeling battery
management systems [89] [90]. Once a better understanding of how BEVs fail is developed, models
to characterize these failures in tunnels can be developed.
3.5.3. Analysis
While no specific studies to analyze BEVs in tunnels have been conducted, one analysis goes over
the various failures that might occur.
3.5.3.1. Safety Test Methods for EV Batteries

A study analyzing three different failure modes in BEVs was performed by Davidsson et al. [91] to
help understand the safety that should be maintained through the whole life of the battery: assembly,
usage, servicing, accidents, and recycling.
The first test was a short circuit test, which consisted of shorting out various batteries and measuring
the current discharge. This short circuit could come from for example crash damage or chassis
flex/deformation. One cell exploded, one ruptured and vented, and the last one was not physically
affected. The fire test consisted of a battery pack with a 25-kW propane burner located underneath
within a furniture calorimeter. The fire effluents were collected and analyzed to understand the
toxicology. Hydrogen fluoride and carbon monoxide were both measured in parts per million by
volume (PPMV) [91]. This helps understand required ventilation rates, and as more tests are
performed, the total production of these chemicals can be determined as a function of the battery
system capacity.
62
3.6. BEV Research Gaps
At a cell level, lithium-ion batteries are still being studied to understand how cells might fail and
what the impact is. With battery technology constantly changing to improve performance metrics,
the safety and consequences are also changing. This drives the need for further work which are
included in the research gaps below. There are two distinct hazards: the flammable vent gas and
unique hazards due to thermal runaway and propagation between cells. Studies conducted to
understand the safety of BEVs in tunnels is limited and ranged from real world vehicle crash tests
such as the Chevrolet Volt NCAP pole test to actual reported incidents on public roadways.
Attention should be given to the size or class of the vehicle. As vehicular class increases so does the
amount of stored energy. Currently, most BEVs are passenger vehicles. As the technology and
energy density of lithium-ion cells improve, medium- and heavy-duty BEVs will be developed.
Hybrid-ICE vehicles are becoming more common using larger battery packs in each scale of vehicle
and should also be evaluated. In addition to the scale of a BEV increasing, the transportation of
damaged BEVs should be considered. Reported incidents of damaged and failed BEVs reigniting
shows that studying and understanding how to mitigate this is important [92].
The following criteria were evaluated to determine where research gaps may exist regarding BEVs in
tunnels.
2. Failure Modes
3. Consequences
4. Validation
The scenarios that lead to a failure mode have been identified by real world examples. Section
3.5.1.1 goes over the various incidents that have been reported. These involve but may not be
limited to impacts that cause mechanical damage and thermal issues such as internal cooling systems
failure that leads to cell thermal runaway as well as internal shorts and/or incorrect
charging/discharging rates and cycles that lead to a failure mode.
The failure modes can lead to an exothermic reaction in the cell(s) which leads to self-heating. This
reaction can propagate failure throughout a module and release of flammable vent gas from multiple
cells. For each of the failure modes, there are several variables that effect the magnitude of the
consequence such as the following: battery vent gas volume production, type of cell, vent gas
species, rate of failure propagation through cells and modules, whether there is a delayed ignition,
tunnel geometry, etc. The failures described in Section 3.5.1.1 and as shown in Figure 15 show
trends of delayed fire or ignition during a BEV incident. With the complexity of these vehicles, most
research is at the bench-scale to study lithium-ion cell safety. More complete evaluations on failure
modes needs to be completed as more data becomes available.
In contrast to the some of the other alternative fuels, there has not been significant research on
BEV vehicles in tunnels or BEV fires/deflagrations in general. The measurements of the
consequences include vent gas deflagration metrics, HRR, battery vent gas dispersion, and resulting
structural damage. One experiment on has been performed to understand the consequence of
mechanical impact and external fires on BEVs [91]. Yet most experiments conducted to understand
mechanical, thermal, overcharging and cycling abuse tests have been performed at the cell level.
These types of tests need to be scaled from bench-scale to intermediate- and large-scale tests. One
large scale test [88] shows that BEVs are very similar in terms of the fire hazard compared with
63
ICEs when compared the HRR and THR. One of the differences found in this study is the
hydrogen fluoride production is over double during a BEV fire compared to an ICE vehicle fire.
Validation amongst various experiments to understand the vent gas composition has also been done
(see Figure 16).
Additional testing needs to be performed to help understand how to characterize these failures and
the consequences in tunnels. Examples would be understanding the HRR and THR of a BEV fire
and how ventilation and confinement in a tunnel effects these metrics. The following research gaps
were identified:
 Fully evaluate the deflagration metrics, such as the vent gas volume production, lower
flammability limit, laminar flame speed, and adiabatic overpressure.
 Deflagration metrics are dependent on the fuel concentration as well as ambient conditions for
well-known combustible gases. Battery vent gas is based on these parameters as well as effects of
cell chemistry, state of charge, cell capacity, cell form factor, electrolyte chemistry and
composition, cell to cell propagation, and failure mode.
 Fully evaluate the fire characteristics of both cells, cell arrays, and modules to understand the
HRR characteristics, temperatures, time to ignition, and effect of different cell configurations in
modules.
 Understand the effects on fire characteristics of cell chemistry, state of charge, cell capacity, cell
form factor, electrolyte chemistry and composition, cell to cell propagation, and failure mode.
 Fully evaluate the failure initiating events that are risk significant in terms of BEV vehicles in
tunnels such as mechanical damage or cell defects.
 Evaluate different BEV classes (light, medium, and heavy duty), including how the deflagration
metrics and fire characteristics scale up, as well as how does propagation of failure between cells
change.
 Assess how flame speed and overpressure from various vent gas compositions change in tunnels
with different geometric parameters (such as length/diameter ratio) and the effect on structural
components of tunnels.
 Validation of modeling efforts through direct comparison to experimental results, especially
since the battery vent gas contains multiple species (hydrocarbons, carbon monoxide, hydrogen,
carbon dioxide).
 Understand the amount of toxic chemicals release rate and total volume released during a battery
fire to understand emergency ventilation.
 Assess required ventilation rates for battery off-gas without ignition (deflagration hazard) and
fire effluent from combustion of battery system (toxicity hazard).
64
4. NATURAL GAS VEHICLES

The natural gas vehicle (NGV) is an alternative fuel vehicle that uses compressed natural gas (CNG)
or liquefied natural gas (LNG). Natural gas is comprised primarily of methane (with concentrations
ranging from about 85% to 96% by volume) and combustion of this fuel produces less emissions
compared to other hydrocarbon fuels (e.g. gasoline or diesel). In the transportation sector, NGVs
span the range of light- to heavy-duty vehicles. Natural gas (NG) can be used to run internal
combustion engines (ICE) as a dedicated fuel or as a bi-fuel mixed with gasoline, diesel, etc. CNG
can also be utilized to spin small gas turbines to power electric generators and therefore can support
hybrid vehicles as well. In 2019 there were a reported 27.7 million NGVs on the road [93]. Also, as
seen in Figure 20, over the years CNG has become a less expensive alternative compared with more
popular fuels such as gasoline and diesel in dollars per gasoline-gallon-equivalent (GGE). With fast-
fill stations for retail use and time-fill for commercial use, there are a variety of refueling options for
the range of consumers [94]. Fast-fill stations use a series of storage tanks that store high pressure
CNG that is then used to fast-fill a vehicle 20-gallon equivalent tank in less than five minutes, for
example. A time-fill station fills directly from the compressor rather than using high pressure tanks
and requires less fueling equipment at the expense of longer fill times.
According to the DOE Alternative Fuels Data Center [95] as of 2019 there are 945 CNG and 71
LNG fueling stations in the continental U.S. Stronger global demand for cleaner burning fuel drives
rapid growth in CNG and LNG [96]. Due to the existing fleets of NGVs and the forecasted
continued growth [97] we are reviewing the hazards associated with NGVs specifically in the context
of accidents that might occur in tunnels.
Figure 20: Cost of common fuel types over an 18-year span (from [98])
65
4.2. Properties of Methane and Natural Gas
To obtain the energy density required for onboard storage, CNG is stored as a pressurized gas while
LNG is stored cryogenically (see Figure 21). NG under standard conditions is roughly half as dense
as air and has a range of chemical properties shown in Table 17. Additionally, methane properties
are listed since some experiments and models discussed below are for pure methane gas. The energy
released upon ignition (heating value) of methane is higher than that of natural gas, which would
result in a conservative risk estimate for fires from natural gas systems when using methane as a
surrogate. Liquid methane has an expansion ratio of 621.3 which is the ratio of volume occupied by
one unit of mass in liquid form to that in gas form. The energy density of NG (both CNG and
LNG) per unit mass is approximately 43 MJ/kg. This is similar to that of gasoline which is 45 MJ/kg
[99]. The energy density per unit volume for CNG (at 250 bar) is 9 MJ/L versus the 22.2 MJ/L for
LNG at -162 °C compared with 34.6 MJ/L for conventional gasoline [100]. Figure 21 below shows
the density contour for methane. Two points on the graph are highlighted – the density at CNG
tank conditions (250 bar, atmospheric temperature), and LNG tank conditions (atmospheric
pressure, -162 °C).
Table 17. Physical and Chemical Properties of Natural Gas & Methane (from [100])
Property Methane Natural Gas
Molecular weight 16.043 g/mol 19.5 g/mol
Gas Density 0.657 g/L, 25°C, 1 atm 0.7–0.9 g/L, 25°C, 1 atm
Relative Vapor Density 0.5536 0.5809–0.7468
Liquid Density 422.62 g/L, −162°C 470 g/L, −162°C
Melting Point −182.5°C −182.0°C
Boiling Point −161.50°C −162.0°C
Auto-ignition Temperature 580°C 723°C
Flammability Limits (vol % in air) 5–15% 4.3–15%
Figure 21. Density contours for methane (created using data from [101])
66
The primary safety hazards associated with natural gas is the same as most other fuels, namely
flammability and uncontrolled combustion. There are additional hazards are associated with LNG
due to the storage temperature and the potential for rapid expansion, which will be covered in more
detail later. Naturally occurring NG is odorless and for safety reasons the gas is odorized prior to
distribution. According to [102], the odorized gas is detectable at concentrations as low as 0.3% by
volume in air. As shown above in Table 18, the lower flammability limit (LFL) is 5.3% (vol% in air).
Thus, a leak is detectable by odor at concentrations roughly 15 times lower the concentration
required for combustion. Flammability properties of methane along with other common fuel types
are shown in Table 18. Note that these properties vary based on lab testing uncertainties along with
regional and various seasonal additives.
Table 18: Flammability Properties of Hydrogen and Other Fuels
Gasoline
Property Hydrogen Methane Propane
Vapor
Flammability in LFL 4.0% 5.0% 2.1% 1.2%

Air (vol%) [27] UFL 75.0% 15.0% 9.5% 7.1%
Most easily ignited mixture in air
(vol%) 29% [103] 8.5% [104] 5% [105] 2% [106]
Adiabatic Flame Temperature
[107] 2483 K 2236 K 2250 K 2289 K
Buoyancy (ratio to air) 0.07 0.54 1.52 4
MIE [108] [109] 0.011-0.017 mJ 0.28-0.30 mJ 0.25-0.26 mJ 0.8 mJ
Autoignition Temperature [110] 500 °C 580 °C 455 °C 246 – 280 °C
Figure 22 and Figure 23 graphically represent the minimum ignition energy and flammability limits
(FL) of methane compared to other fuel types. As shown methane has a similar minimum ignition
energy to other hydrocarbons when the concentration lies within the FL.
Figure 22: Minimum ignition energy for common fuels types (from [111])
67
Figure 23: Flammability limits for common fuels types (from [111])
An important consideration in safety analyses is the fact that methane is less dense than air at
standard conditions. This means that during an event that yields a leak or rupture of a NGV fuel
container the flammable mass will in most cases dissipate upwards and away from the ground due to
buoyancy. This improves safety for cases where potential ignition sources are located near the
ground. Buoyant diffusion is also preferred unless there exists a restrictive surface above the release
leading to an accumulation of gas at the surface. At this point, the flammable mass must be
dissipated through proper ventilation and air flow, otherwise the hazard can linger for quite some
time. This hazard is reviewed in detail in Section 4.5.
Note that due to the high storage pressures of CNG and the small flammable range, this lowers the
chance of ignition. This was confirmed with modeling in a study by Zalosh et al. [112] that
compared flammable gas dispersion of CNG and gasoline fuel leaks in a tunnel. Using the CFD
code Fluent, the models showed that leaked CNG from vehicle storage containers dilute to levels
outside of the FL shortly after release. Additionally, Zalosh concluded that CNG fueled vans are
much less likely to ignite compared to those of gasoline when there is ventilation of 0.10 m/s or
higher in a tunnel. This is based on the model results having a smaller flammable vapor cloud
compared with the gasoline vapor clouds shown in the dispersion model results.
LNG releases, similar to CNG, can also produce flammable vapor clouds contained by restrictive
structures. But in the case of an LNG vehicle release, the vapor cloud has the potential to be larger
because of the typically larger amount of LNG stored onboard a vehicle [31]. In the event of a
release at the bottom of an LNG container, the liquid will cool the surface below which could allow
an LNG pool to form which will prolong the dissipation period. In addition, the density of recently
vaporized LNG is higher, reducing buoyant dispersion. Initially the vapor is much heavier than air
but as the gas temperature warms up and the density decreases, it will become lighter than air. Both
factors also increase the chances of ignition by a source on the ground.
Many tests have been carried out looking into the conditions required to cause DDT with methane
vapor clouds (see 4.5.1). Specific to LNG, liquefied fuels come with the risk of a boiling liquid
expanding vapor explosion (BLEVE). BLEVE corresponds to a rapid heating of the super cooled
liquid causing a rapid change in phase. As LNG vaporizes, its volume increases over 600 times. If
the rate at which the gas is released is less than the rate of expansion, the vessel can over-pressurize
and rupture. According to [31] this scenario has been witnessed under the extreme conditions of a
68
tanker with compromised insulation in contact with an external fire. For BLEVE to occur the
superheat temperature limit must be reached by the liquid. In the case of LNG this is roughly -93 °C
[31].

NGVs have robust safety standards and regulations regarding the fuel storage system, the vehicle
itself, and the roadway structures on which they operate.

temperature exposures, irrespective of the fire location, ventilation rate or type [113]. With regards
to NGVs in tunnels, appendix G.2.1 states that CNG fuel systems have a superior safety record than
that of current conventional systems (i.e. gas and diesel).
emergency operations are covered.

consideration of the geometric parameters of the tunnel, fire protection features and response times
4.4.4. National Fire Protection Association 52

NFPA 52, Vehicular Natural Gas Fuel Systems Code, provides design, installation, operation, and
maintenance requirements for CNG and LNG fuel systems, storage containers, and dispensing
systems [114].

NFPA 55, Compressed Gases and Cryogenic Fluids Code, provides storage, use, and handling requirements
for both compressed and cryogenic liquid hydrogen in portable containers, cylinders, and tanks.
Sections 10 and 11 deal with bulk hydrogen compressed gas systems and bulk liquefied hydrogen
systems, respectively [115].
69
NFPA 57, Liquefied Natural Gas (LNG) Vehicular Fuel Systems Code, provides storage, use, and
handling requirements for liquid natural gas fuel in storage containers [116].
4.4.7. National Fire Protection Association 59A

NFPA 59A, Standard for the Production, Storage, and Handling of Liquefied Natural Gas (LNG), provides
guidance for construction and operation equipment for production, storage, and handling of LNG
[117].
4.4.8. SAE J1616: Recommended Practice for Compressed Natural Gas Vehicle
Fuel
SAE J1616, Recommended Practice for Compressed Natural Gas Vehicle Fuel, provides recommended
practices for fuel systems on vehicles that are powered by CNG [118].
4.4.9. SAE J2406: Recommended Practices for CNG Powered Medium and
Heavy-Duty Trucks
SAE J2406, Recommended Practices for CNG Powered Medium and Heavy-Duty Trucks, provides
recommended practices for construction, maintenance, and operation of CNG powered medium-
and heavy-duty trucks [119].
70
4.5. NGV Research Summary in Tunnels
A significant amount of work exists evaluating the risks of NG ignition in various scenarios. The
following section documents some of the most pertinent studies.
4.5.1. Experiments
Due to the high storage pressures of natural gas onboard NGVs, compromised fuel containers could
produce large flammable masses. This flammable mass is unlikely to ignite without an external
ignition source [112].
Multiple studies review the outcomes of ignited NG vapor clouds and high-pressure jets. An
experiment determined that congestion had a strong influence on the maximum overpressure
produced by the combustion of vapor clouds (see section 4.5.1.1) [120]. The occurrence of DDT,
even in confined environments with obstructions was shown to be highly unlikely [121] [120] [122].
Specifically, Harris et al. [122] showed that even over the length of a 45 m pipe with repeated
obstructions, DDT for NG did not occur. A strong correlation for heat transfer to a ceiling based
on the flame height to ceiling distance was shown for large scale flames impinging on a ceiling (see
Section 4.5.1.1) [120]. Royle et al. [123] studied the distribution of overpressure resulting from the
ignition of NG contained in a congested region. An array of sensors was used to obtain the pressure
as a function of distance from the ignition source for an uncontained but congested explosion. More
details are provided in the sections below.
4.5.1.1. Vapor Cloud Explosions in a Long-Congested Region

A series of large-scale natural gas vapor clouds were ignited in a 3 m x 3 m x 18 m long region with
variable congestion by Lowesmith et al. [120]. The aim of the study was to determine the risk of
DDT when flame speed was accelerated by congestion. Flame speed and overpressure values were
measured. Initial flame speed prior to entering the congestion was varied from 45 m/s to 156 m/s.
Note the speed of sound for NG is approximately 446 m/s at standard conditions.
The experiment took place inside of a 26 m long enclosure. The enclosure contained two regions, a
congested region of 18 m length and 3 m x 3 m cross section and a free region of 8.25 m length and
3 m x 2.8 m cross section (W x H). The entire enclosure was filled with a gas mixture of air and
methane at 1.16 and 1.09 equivalence ratios. These mixtures were ignited in the free region at
variable locations to obtain specific flame speeds prior to congestion. The congested region was
formed from 12 racks (designated R1–R12), spaced 1.5 m apart, supporting alternately six or seven
horizontal pipes, each 0.18 m diameter. This created a cross-sectional area blockage ratio of 0.36 to
0.42. Images of the enclosure are shown in Figure 24.
71
Figure 24: Schematic of Experimental Configuration (from [120])
The gas within the confinement flowed left to right and was recirculated until a desired uniform
mixture was obtained. Note that within this study NG as well as NG-hydrogen mixes were
evaluated. Lowesmith et al. carried out two tests pertaining to NG air mixtures. These correspond to
test numbers VCE01 and VCE04. The test details are highlighted in Table 19:
Table 19: Test Results Summary for CH4 Vapor Cloud Combustions (from [120])
72
For VCE01, Lowesmith et al. report a maximum witnessed flame speed of ~130 m/s and max
overpressure of 0.34 bar. For the high initial speed test VCE04, the flame speed accelerated to a
maximum of 300 m/s with an over-pressure value reaching 2 bar. These results are graphically
shown in Figure 25.
Figure 25: Spatial variation of flame speed and overpressure within the congestion region [120]
Lowesmith et al. explain that despite the high flame speeds and amount of obstructions, the
transition to detonation did not occur. This aligns with the work of Harris and Wickens [122] who in
a similar study only witnessed deflagration as the flame front accelerated along the length of a highly
congested 45 m tube. For cases where methane-air combustions yield DDT severe test conditions
are required. The experiments ran by Zipf et al. [124] demonstrated DDT under the conditions of a
uniform pre-mixed gas, 0.5 cross-sectional area blockage ratios, high energy ignition sources, and
significant geometric confinement. Under these conditions DDT occurred with maximum flame
velocities of 812 m/s (Mach 1.82) and greater. These conditions are unlikely to develop from a
NGV release in a tunnel due to non-uniform mixing, a lack of confinement/blockage, and no high-
energy ignition sources. Thus, the chances of DDT occurring are unlikely.
73
4.5.1.2. Heat Transfer to Ceiling and Impinging Diffusion Flame
This experiment was designed to study heat transfer from impinging buoyant natural gas jet flames
by Kokkala et al. [125]. Heat transfer was measured at the stagnation point of an impinged diffusion
flame for various ceiling to flame height ratios. Flame powers ranging from 2.9 to 10.5 kW were
studied. Figure 26 displays the experimental setup.
Figure 26: Experimental Configuration (from [125])
As shown, the artificial ceiling consists of a circular plate held in place by four support rods. The
support rods also support the floor panel where the burner rests. To protect the experiment from
the fluctuating conditions of the lab, the apparatus was surrounded by double screens on all sides.
The fuel for the diffusion jet was reported to contain 96% volume of methane gas, which is a typical
concentration for CNG.
The thermal readings taken off the ceiling plate consisted of thermocouples for surface temperature
and both Gardon type and Schmidt-Boelter type heat flux gauges which provided material
independent heat flux readings. The configuration of these sensors on the ceiling plate are shown in
Figure 27.
74
Figure 27: Configuration of thermal sensors on ceiling plate (from [125])
The heat flux gauge is located at the stagnation point of the impinged diffusion flame. This should
represent the location of maximum heat flux. Results for the measured heat transfer rates are shown
in Figure 28 as a function of the ratio of flame to ceiling height.
Figure 28: Total heat flux at stagnation point vs. flame height over ceiling height (from [125])
As shown, the rate of heat transfer is varying with the ratio of the flame height to ceiling height. The
gradient of heat transfer begins to steepen at the point where 𝐻𝑓 /𝐻 reaches unity and plateaus
around 𝐻𝑓 /𝐻 > 1.5. Regardless of the flame power, the maximum heat transfer rate was measured
to be approximately 60 kW/m2. In other words, for the worst-case scenario, the maximum expected
heat flux is around 60 kW/m2 for the type of leak sizes and studied. Thus, tunnels capable of
withstanding a 60 kW/m2 heat flux over the duration of a NGV fuel release should have a low risk
of failure from an impinged jet flame.
75
4.5.1.3. Vapor Cloud Explosions from Ignition of Gaseous Mixtures in a Congested
Region
A series of studies were carried out by Royle et al. [123] to measure the overpressure produced from
methane and methane/hydrogen mixtures premixed with air when ignited within congested spaces.
The experimental space was a 3 x 3 x 2 m region containing multiple layers of pipes. An image of
the congested region is shown in Figure 29. A concrete wall sits adjacent to the one side of the
congested region. The wall is positioned there to protect the control room and has been shown to
not interfere with free field overpressure [123]. Additionally, the wall has embedded pressure sensors
at different heights. For this series of experiments, the blockage ratio was reported as 4.40% the
total volume. The outside of the grid was covered in a 23 µm thin plastic film which contained the
gas prior to ignition. The plastic film was used only to contain the premixed gas mixtures prior to
combustion and did not significantly restrict the outflow of gas or the pressure wave.
Figure 29: Congestion region or grid where gas was filled then ignited (from [123])
Methane gas was mixed with air to form a stoichiometric ratio of 1.1 which reportedly produces the
highest overpressures. Other gases evaluated in this study were mixtures of methane, air, and
hydrogen which are all included in some of the figures and tables below. For this section, only
results pertaining to methane alone are discussed. Section 6.5.1.7 discusses the hydrogen portion of
this experiment.
The ignition source was located at a height of 0.5 m off the ground and positioned at the center of
the grid. For ignition a 2.25 J capacitor was discharged through a spark gap of 6 mm. It was noted
that the spark exhibited lower energy than what was discharged from the capacitor. For the
instrumentation, overpressure values were measured by an array of low- and high-pressure pressure
sensors. The location of the pressure sensors can be seen in Figure 30. All pressure sensors were
positioned 500 mm above the ground except for the far field pressure sensors, which were mounted
at higher locations due to the topology of the testing pad.
76
Figure 30: Pressure Sensors distributed in and around grid structure (from [123])
Pressures were measured across a wide span of locations including up the adjacent wall. Table 20
lists the initial conditions prior to ignition. The pure methane is labeled as NatHy_02. For the results
of hydrogen/methane mixture experiments, we refer the reader to the paper [123].
Table 20: Initial Conditions of Experiment (from [123])
Measurement Test Conditions: NatHy_02
Methane (vol. %) 100
Number of Layers 9
Free Volume 17.207
Gas mixture temperature (°C) 4.8
Relative Humidity (%) 85.1
Atmospheric Pressure (kPa) 97.71
Mean Oxygen Concentration (%) 18.71
Partial Oxygen Pressure (kPa) 0.1871
Partial Nitrogen Pressure (kPa) 0.7059
Partial Water Vapor Pressure (kPa) 0.0076
Partial Fuel Gas Pressure (kPa) 0.0994
Mass of Methane (kg) 1.160
77
It was noted that during experiment the humidity was uncontrolled but was assumed to have a
minor effect on the resultant explosion overpressure values. Figure 31 displays an image of the
explosion immediately after ignition.
Figure 31: Image of pure methane combustion right after ignition (from [123])
Figure 32 and Figure 33 show the measured overpressure values at various locations for all mixtures.
Recall that NatHy_02 corresponds to the hydrogen-free methane gas. Pressures were reported in the
near-field (within and just outside of the grid) and far-field regions.
Figure 32: Overpressure vs. of distance parallel to the wall in the near field region (from [123])
78
Figure 33: Overpressure vs. distance perpendicular to the wall in the far field region (from [123])
In the near field, overpressure values were 11 to 15 kPa. Inside of the rig, overpressure reached 11.8
kPa and 1.4 kPa just outside the rig. At 32 m away, the overpressure was 1.2 kPa. Referring to Table
1, 11.8 kPa is the threshold for skin laceration from flying glass. Anything below 4 kPa is below the
threshold for injury from flying glass. Note that this experiment represented the ignition of a pre-
mixed, near-stoichiometric 18 m3 region with a high level of congestion—both the stoichiometric
mixture size and level of congestion are probably unlikely to occur in a tunnel, especially
simultaneously.
79
4.5.2. Modeling
A variety of literature exists on simulations evaluating the risk and consequences of CNG and LNG
releases. In one paper specific to CNG vehicles in tunnels, the goal was to capture the likelihood and
worst-case scenarios associated with failures of a NGV. The dispersion of CNG was simulated with
Fluent to compute the effect ventilation had on mitigating flammable masses (see Section 4.5.2.1).
Other simulations have been used to evaluate the release of CNG from a bus in a tunnel, the
hazardous overpressures from the combustion of these releases, as well as the flame lengths for
several failure scenarios (see Section 4.5.2.2). In Section 4.5.2.3, overpressures were calculated due to
the combustion of NG vapor clouds in tunnels, along with a risk analysis study. We also summarize
modeling that compared the risk associated with equivalent CNG and LNG vehicle failures in terms
of harm and lethality distance (see Section 4.5.2.4). In this work, both the flammable and non-
flammable effects on occupants within the tunnel were considered. Overall it has been shown that
the associated risks of harm to people and the structure of the tunnel itself is low in most cases of
failure. In absolute worst-case scenarios where full instantaneous fuel releases of commercial
vehicles occur with no tunnel ventilation and perfect mixing, hazardous overpressures can develop.
However, the likelihoods of the worse-case scenarios are implausible, and the hazard can be reduced
with proper ventilation.
4.5.2.1. Dispersion of CNG Fuel Releases in Naturally Ventilated Tunnels

Three naturally ventilated tunnels were studied with respect to the transient dispersion of vapor or
gas clouds produced during an accidental fuel release by Zalosh et al. [104]. The purpose of this
study was to determine the effect ventilation speed had on flammable volume formation and
dissipation. Gasoline and CNG releases were compared, sourced from CNG vans with equivalent
fuel capacity/driving range to conventional gasoline. For CNG, 24 kg of gas stored at 20,684 kPa
(3,000 psig) was released through a 6.35 mm (¼”) pipe under choked flow conditions. For gasoline,
a tank of 35 gallons was released in a liquid state through a 12.7 mm (½”) pipe. The gasoline pooled
and subsequently evaporated forming a vapor cloud. The specifics of the vaporization and flow
models can be found in [112].
Zalosh measured the average velocity profile during “low traffic” conditions for several tunnels in
Boston, including the Rutherford Avenue, Storrow Drive, and Prudential tunnels. The velocity
profiles for each tunnel was measured using a hotwire probe on a telescopic pole at various heights
(~1 - 5.5 m). During measurements, one lane was closed so readings could be taken from the center
of the tunnel. The probe was also moved along the length of the tunnel to obtain the variation along
the length. Table 21 provides the tunnel dimensions as well as the cross-sectionally averaged
velocities for each tunnel.
80
Table 21: Tunnel dimensions and averaged velocities (from [112])
Tunnel Tunnel Dimensions (L x W x H) [m] Avg Velocity [m/s]

Rutherford
121.9 x 15.2 x 5.4 1.24
Avenue
Storrow Drive 304.8 x 9.1 x 15 1.00
Prudential
224.0 x 18.3 x 4.6-7.1 1.10
Tunnel
Average 1.11
The CNG release rate was calculated as choked isentropic flow, as described in detail in [112].
Gasoline was assumed to be 53 % pentane, 22% hexane, and 25% benzene which results in a vapor
pressure of 5.3 psia at 20 °C and a viscosity of 5.0×10-7 kg/m-s.
Fluent was utilized to simulate the dispersion of the two fuels when the van was located at the center
of the tunnel. The dispersion models simulated the release occurring inside a computation domain
of 100.0 × 8.25 × 4.27 m (L × W × H). The size and shape of the flammable cloud was defined by
the lower flammability limit (LFL) for each fuel. The LFL is 5.0 % by volume in air for CNG while
the LFL is 1.6 % by volume in air for gasoline vapor. The volume of the flammable region was
measured from the release point (van location) to the end of the computational domain set for the
model. Figure 34 displays the volume of flammable region obtained for each fuel as a function of
ventilation speed.
3000
2500
Volume of Flammable Region (m3)
Gasoline CNG
2000
1500
1000
500
0
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70
Ventilation Speed, (m/s)
Figure 34: Volume of flammable region versus ventilation speed from simulation (from [112])
81
An equivalent CNG release produces a much smaller flammable volume when there is minimal
ventilation (0.1 m/s). As the ventilation speed increases, so does the gasoline pool evaporation. This
relation can be seen in Figure 35. Note that the CNG release rate is constant because the flow is
choked regardless of the ventilation speed.
1.2
1
Average Fuela Release Rate (kg/s)
0.8
0.6
Gasoline CNG
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
Ventilation Speed (m/s)
Figure 35: Gasoline Vapor and CNG Release Rates vs Ventilation Velocity (from [112])
The gasoline release rate, on the other hand, increases as the ventilation velocity increases due to
increased evaporation. This increase in severity begins to reduce as ventilation speed increases
beyond 0.2 m/s. The CNG release rate is independent of the ventilation velocity. Note these values
are much lower than the measured average vapor release. Overall, with enough ventilation, CNG
releases produce smaller flammable masses than that of gasoline when released in a naturally
ventilated tunnel during low traffic ventilation velocity. Zalosh et al. [112] concludes that a CNG
vehicle poses a smaller overall flammable region.
4.5.2.2. Gaseous release, dispersion, and combustion for automotive scenarios

Venetsanos et al. [126] CFD was used to study the effects of a compressed gas release from a
commercial vehicle in urban areas. One urban area simulated was a tunnel with a single deck city bus
located centrally along the length. Variable releases from both hydrogen and CNG fuel tanks were
evaluated. The fuel storage systems modeled represented that of a typical European bus with fuel
containers located along the roof, forward from the midpoint. The system consisted of 2 sets of 4
82
tanks connected as displayed in Table 34. The specific CFD solvers utilized were ADREA-HF for
dispersion and REACFLOW for combustion.
Figure 36: Fuel tanks configuration for both CNG and CH2 gas (from [126])
The tanks contained a total of 168 kg at 20 MPa of CNG. This is representative of a standard CNG
bus. The main vent lines are controlled by thermally activated pressure relief devices (TPRD). As
shown, each TPRD is attached to a manifold connected to 4 tanks (2 TPRDs per set of 8 tanks).
Multiple release scenarios were evaluated by varying TPRD orifice size and tank evacuations. A
summation of storage parameters is shown in Table 22. Note this study and the figures below have
results for hydrogen systems as well.
Table 22: Storage Configurations (from [126])
The computational domain was modeled as a tunnel of 212 m length with a cross-sectional area
displayed in Figure 37. As mentioned, the bus was located along half the length of the tunnel in a
centralized location. Besides the bus, the tunnel was assumed to be empty and the walls were
modeled as smooth surfaces. Additionally, the air was assumed quiescent to represent a worst-case
scenario.
83
Figure 37: Tunnel Cross Section (from [126])
Two release cases were evaluated for CNG. Table 23 lists the descriptions of all cases, Case 1, and 3
were selected since Case 2 lies between them.
Table 23: Storage Configurations Details (from [126])
The results for Case 1, where only one cylinder was released through a single TPRD, are shown in
Figure 38. The left frame shows the flammable mass and the right frame the total available energy.
The available energy was computed by multiplying the released mass of fuel by the lower heat of
combustion. The flammable mass was calculated from the amount of fuel/air mixture released
which was within the FL. Figure 39 displays the results for Case 3, where all 8 cylinders released
through all 4 TPRDs simultaneously. This case may be implausible.
84
Figure 38: Flammable mass and available energy of released gas in Case 1 (from [126])
Figure 39: Flammable mass and available energy of released gases in Case 3 (from [126])
Note the change in scales between Figure 38 and Figure 39. For Case 1, both the flammable mass
and available energy maintained lower values and dissipated rapidly in time. For Case 3, the
flammable mass and available energy reached dangerous levels which persisted over the length of the
simulation.
It was assumed the Case 1 overpressures would be negligible because the total flammable mass was
less than 0.5 kg. For Case 3, the overpressure values are displayed in Figure 40. The overpressure
was calculated assuming that the cloud was ignited after 40 seconds, corresponding to maximum
flammable mass. The ignition point was assumed to be at the center of tunnel at the top of the bus.
85
Figure 40: Overpressure values up and down tunnel of the bus for release Case 3 (from [126])
When reviewing these overpressure values, it is important to keep in mind what pressure values
correlate to what level of property damage or human hazard. The tables listed in Section 1.2 help
understand the damage and tenability thresholds. Case 3 scenario yields overpressure values capable
of rupturing eardrums and creating harmful glass splinters up to a distance 80-100 m from the
ignition point. Note that the range for the eardrum rupture threshold reported in Table 1 is 16.5-
19.3 kPa. Venetsanos et al. [126] states that the blast wave maintains its strength for long distances
inside of tunnels due to the high levels of confinement compared with urban environments where
blast waves decay quicker. In addition to overpressure, Venetsanos et al. [126] reported the fireball
length along the tunnel, the results for each scenario are displayed in Table 24.
Table 24: Combustion results within tunnel (from [126])
For Case 3, the NG combustion produced a flame length which traveled nearly the length of the
tunnel, 198 of the total 212 m. The flame length for Case 1 NG combustion was reported to be
negligible. It is worth noting that in this study, Case 3 represents an implausible scenario of rapid
and complete fuel release, ignition when the peak flammable mass is present, and static air within the
86
tunnel. This study further identifies the importance ventilation plays on mitigating risk during an
accidental release of fuel in a tunnel.
4.5.2.3. Natural Gas Vehicle Explosion Risk in Tunnels

Middha et al. [127] used CFD modeling to support the evaluation of explosion risks for NGV (both
cars and buses) in tunnels. The objective of the modeling was to predict a quantitative explosion risk
for CNG in tunnels. The solver utilized for simulations is commercial code FLACS.
Both NGV and H2 vehicles were studied in the simulations but this section will only make
comment on results pertaining to CNG. As mentioned, both a CNG car and bus fuel release was
modeled. The CNG car and bus parameters are described by Middha et al. as follows:
1. City bus with a storage pressure of 200 bar and a mass of 26 kg in each cylinder. It is
assumed that the release occurs from a set of four cylinders with a total inventory of 104 kg.
2. Car with a storage pressure of 200 bar and a total gas mass of 26 kg.
The values described above represent configurations of standard CNG vehicles as described in
Section 4.1.
As for the tunnels, two different cross sections were evaluated, rectangle and horseshoe shape, see
Figure 41 for cross-sectional dimensions. Both tunnels were modeled with a length of 500 m.
Figure 41: Tunnel cross-sectional dimensions (from [127])
In addition to the cross-sectional dimensions and length, the geometry of the modeled tunnel
included vehicles. The tunnels were dual lane with traffic running a single direction. The tunnel was
assumed to be full of cars and buses spaced out evenly with 1.5 m between each. The vehicle
distribution was a repeated pattern of 6 cars follow by 1 bus. The vehicles were placed such that one
bus and one car were at the exact center of the tunnel length in separate lanes. The same geometry
was used for both the car and bus release. The releases were assumed to be choked flow. The mass
flow rate of the CNG for choked flow at 200 bar is displayed in Figure 42. Note that this was
computed assuming a discharge coefficient of 0.8 and a 6 mm opening.
87
Figure 42: Mass flow of release for CNG and various H2 simulations (from [127])
Ignition points were varied from the center of the vapor cloud to the outer edges (length wise of the
tunnel). Ventilation velocities were also varied between the models.
A dispersion model simulating the release of the fuel systems was carried out. Table 25 lists the
maximum flammable gas cloud size for each configuration as well as the equivalent stoichiometric
cloud or the Q9 quiescent cloud. This is a scaled smaller stoichiometric gas cloud that represents the
same explosion load as the non-homogenous larger cloud. It is scaled based off the weighted
volume expansion, flammable volume, and laminar burning velocity. The flammable cloud and its
stoichiometric and Q9 equivalents along with the maximum pressures for the combustion of the
flammable gas clouds are listed in Table 25 below:
Table 25: Summary of gas cloud & overpressure for various vehicles in both tunnels (from [127])
Maximum equivalent Max. pressure for max. equiv.
Maximum flammable gas
stoichiometric flammable cloud Q9 Quiescent/Pre-
cloud size in m3 (kg)
Vehicle/Release Inventory gas cloud size in m3 (kg) ignition turb.
Characteristics (kg) Maximum Q9 Maximum
Horseshoe Rectangular Horseshoe Rectangular
Equivalent overpressure
Tunnel Tunnel Tunnel Tunnel
Volume (m3) (barg)
Bus CNG 200 bar 26 3.4 (0.15) 4.6 (0.19) 1.15 (0.08) 1.18 (0.08) 1.2 0.01/0.01
Bus CNG 200 bar 104 45 (2.01) 647 (26.0) 13.47 (0.90) 113.48 (7.60) 113.0 0.03/0.30
Car CNG 200 bar
26 2.1 (0.10) 3.4 (0.15) 0.85 (0.06) 1.03 (0.07) 1.0 0.01/0.01
(vent up)
Car CNG 200 bar
26 17 (0.78) 15 (0.65) 6.31 (0.42) 5.25 (0.35) 6.3 0.01/0.07
(vent down)
From the coupled dispersion combustion simulations, it is predicted that overpressure values can
produce minor damage to people and property within the tunnel. Referring to Table 1, these
overpressures fall between the threshold for injury from flying glass and in the pre-ignition
88
turbulence case up to potentially serious wounds from flying glass near 50% probability. The data
presented in Table 25 were combined to create a frequency of exceedance curve for overpressures
during combustion of CNG vapor clouds, shown in Figure 43. Overpressure values outside the
hazardous range (less than 0.1 barg) are much more likely than higher overpressures that can be
hazardous to tunnel occupants.
Figure 43: Exceedance curves for overpressure values per fuel type (from [127])
4.5.2.4. Harm effects of Cryo-compressed hydrogen versus natural gas

Zhiyong et al. [128] evaluated CNG, LNG, and cryo-compressed hydrogen in terms of harm,
lethality, and separation distances. Both flammable and non-flammable effects were considered. The
potential hazards to people included cold hazard effects from cryo-release, thermal effects from
combustion with both immediate and delayed ignition, and overpressure or projectile effect due to
explosion. The simulated comparison was completed by coupling methodologies from several prior
works. Specifically, dispersion was modeled by Witlox et al. [129], jet fires by Cook et al. [130], and
overpressure from explosion by Tang et al. [131].
Two quantifications metrics were applied: a harm criterion which corresponds to a 1% probability of
fatality and a fatality criterion which corresponds to a 100% probability of fatality. The criteria for
cold effects was taken in reference to a European Industrial Gases Association document [14] which
states which states that a temperature of -40°C corresponds to harmful effects. The thermal effects
on people within the flammable mass region during ignition or in direct contact with an ignited jet
flame is considered fatal. In terms of heat radiant heat transfer, harm is quantified by thermal dose
units, which is a combination of heat flux intensity and time of exposure. From the literature [132]
values for a dangerous thermal dosage range from 420 (kW/m2)4/3 to 1655 (kW/m2)4/3. The lower
value was utilized to remain on the conservative side of harm analysis. Additionally, for flash fires,
no standard criterion is present but Marangon et al. [133] suggests that a flash fire occurring even at
½ LFL could have lethal effects.
The modeling assumption for each fuel type as well as other release conditions are summarized in
Table 26. Conservative model choices were selected for release situations, such as the release
direction was assumed to be horizontal and in the downwind direction with respect to a 5 m/s wind.
Additionally, the release orifice was set to 10 mm diameter which corresponded to an expected leak
89
size in the connecting parts of the fuel systems where leaks are more likely to occur. The release
height was also assumed to be 1 m from the ground which was assumed to be a good average effect
height for a person.
Table 26: Assumed modeling parameters for simulation (from [128])

Item Cryo-Comp. H2 CNG LNG
Release pressure (bar) 350 250 1
Release temperature (°C) -210 15 -152
Release inventory (kg) 5.6 28.8 28.8
Release direction Horizontal and downwind for continuous release
Release hole size (mm) 10
Atmospheric Pressure (atm) 1
Wind velocity (m/s) 5
Result output height (m) 1
Also note the fuel capacities were selected based on equivalent mileage instead of energy density.
The results are shown in Figure 44 where the two graphs display the harmful and fatal distances for
each incident per fuel type based on an assumed instantaneous release of the inventory, which is
implausible. The non-flammable effects are without ignition of the flammable cloud, whereas the
flammable effects correspond to immediate or delayed ignition depending on which produced the
larger hazard.
90
Figure 44: Harmful/fatal distances per incident for instantaneous release (from [128])
For non-flammable effects, LNG posed long distance hazards due to the potential for cryogenic
injuries. For flammable effects, overpressure damage from flash fires or vapor cloud explosions
were larger with LNG compared to CNG when ignition was delayed. The hazards associated with
LNG are intuitive due to the low storage temperature (111 K) and the larger volume of stored fuel
forming larger flammable masses. For CNG the risk of physical explosion and thermal hazards are
higher due to the high storage pressure and high heat flux associated with a jet fire (see Figure 45
below). As mentioned in Section 4.3, it should be noted that BLEVE is a risk associated more
closely with cryogenic fuels due to the higher chance of rapid expansion due to phase change during
a release event.
Below Figure 45 shows the hazard distances for each fuel when the release occurs at a continuous
rate rather than instantaneous:
91
Figure 45: Harmful/fatal distances per incident per fuel type for continuous releases (from [128])
Due to the high storage pressure, CNG will release at a more rapid rate and thus produce higher
hazard distances for continuous releases. Li [134] shows that LNG vehicles pose their own unique
hazards which exceed CNG in risk in specific situations. It should be understood there is a low
probability of a BLEVE, and instantaneous releases are implausible.
4.5.3. Analysis
Scenario identification as well as probabilistic risk assessments have been carried for CNG vehicles
in the context of vehicular failure on the roadways. To our knowledge, there are no specific
evaluations of NGV risks within tunnels including event sequence diagrams. However, a study
comparing a CNG powered buses related hazards to that of current conventionally diesel-powered
buses showed that CNG posed higher probabilistic risks.
4.5.3.1. Compressed Natural Gas Bus Safety: A Quantitative Risk Assessment

The analysis carried out by Chamberlain and Modarres [135] compared the fire safety risk associated
with a CNG bus against that of a more conventional diesel-fueled bus. Since the data involving
critical failures of heavy-duty CNG vehicles is limited, a probabilistic risk assessment (PRA) was
performed with a two-step approach. The first step was to perform a qualitative risk analysis (QRA)
92
followed by quantitative PRA. The QRA was used to determine the possible fire scenarios that
should be further studied in the PRA.
Chamberlain and Modarres identified the associated fire-safety hazards used to assess the risk of
using CNG fuel:
 Fire potential from fuel leakage.
 Explosion potential from uncontrolled dispersion and mixing of CNG in the presence of an
ignition source.
 Impacts and missile-generated hazards due to fuel being stored at high pressure.
 Chemical hazards (gas toxicity, asphyxiation potential, and higher hydrocarbons in CNG
may be considered neurotoxins even though CNG is relatively nontoxic).
 Electrostatic discharge.
A failure mode and effect analysis (FMEA) was used to identify the most significant failure modes.
These failure modes were determined based on the frequency of occurrence as well as the overall
consequence. From there, a PRA with event trees and fault tree modeling to describe the events
leading to various fire and explosion hazards was performed. Parameter uncertainty was also
discussed and how it affected the PRA results. The entire bus fuel system as well as the supporting
infrastructure (such as fueling stations) was included in the evaluation. Figure 46 represents a typical
CNG bus fuel system.
Figure 46. Fuel supply system of a CNG bus (from [135])
Through the FMEA performed, the severity of an accident scenario was categorized, and a ranking
matrix was created for CNG systems. Based on a variety of inputs regarding the expected
occurrence frequency, scenarios, ignition potential, etc. detailed in the literature, the results in terms
of frequency per year, per distance traveled, and total risk are shown in Table 27.
93
Table 27. Quantitative Risk Assessment (from [135])
From this QRA Chamberlain and Modarres approximated the total mean fire risk to be 2.5 times
larger for CNG buses compared to diesel-powered buses. The overall results identified CNG buses
as higher risk in terms of fire and explosion hazards due to failure rates of relief valves, CNG
cylinders, and piping. This conclusion differs compared to the study by Zalosh et al. [112] which
compared CNG and gasoline vans. The CNG vans were shown to pose lower risks regarding
flammable gas dispersion compared to the flammable vapor dispersion from a gasoline leak from an
equivalent van. The difference is comparing CNG with gasoline and diesel. Zalosh concludes CNG
is a lower risk than gasoline while Chamberlain and Modarres concludes it is a higher risk than
diesel.
4.5.3.2. LANL Risk Analysis

A comparative risk analysis was performed by Los Alamos National Laboratories to put into
perspective the relative hazards of alternative fuels, including LNG and CNG, compared to gasoline
and diesel fuels [136]. As part of this analysis, data was collected about the physicochemical
properties of natural gas and general petroleum and transportation information. Subsequently, the
technical properties of the alternative fuels were ranked, safety data and vehicle accident statistics
were reviewed, and specific accident scenarios were evaluated. One of the accident scenarios
evaluated was a vehicle collision with fuel loss in a tunnel. An expert panel estimated the
probabilities of different outcomes for the different alternative fuels compared to gasoline and
diesel. Table 28 shows the results of the risk analysis. These probabilities are just expert judgement
rather than data. As shown, the natural gas (both liquid and compressed) was estimated to be less
likely to result in a fire with or without injury, or more likely to not ignite, and less likely to result in
an explosion when compared to gasoline [136].
Table 28: Probabilistic Outcomes of a Tunnel Accident (from [136])
Probability
Consequence CNG LNG LPG Gasoline Diesel
Fire without Injury 0.1 0.1 0.15 0.12 0.02
Fire with Injury 0.05 0.05 0.08 0.1 0.01
Explosion 0.05 0.05 0.12 0.2 0.01
No Ignition 0.8 0.8 0.65 0.58 0.93
94
4.6. NGV Research Gaps
Through this literature study, it was found that there are several existing studies that evaluate NGV
vehicles in tunnels. There were several experiments conducted that evaluated the consequences of
ignited NG vapor clouds and high-pressure jets and the relevant variables. Specifically, the
likelihood of DDT, overpressure, and heat transfer of NGV accidents were evaluated. Also, there
were a variety of literature studies that evaluate the risk and consequences of CNG and LNG
releases. Specifically, modeling studies evaluated dispersion, failure modes, overpressure, and flame
lengths. Analytical studies evaluated scenario identification through probabilistic risk assessments
for CNG vehicles in the context of vehicular failure on roadways.
There are several conclusions about important variables and consequences from an NGV accident
in a tunnel. One study showed that hazards associated with a CNG-powered bus posed higher
probabilistic risks that that of a conventional diesel-powered bus. Also, it has been shown that the
associated risks of harm to people and the structure of the tunnel itself is low in most cases of NGV
vehicle failure. In implausible scenarios where full instantaneous fuel releases of commercial vehicles
occur with no tunnel ventilation and perfect mixing, hazardous overpressures can develop. The
occurrence of DDT was shown to be highly unlikely in multiple experiments, even in confined
environments with obstructions. Also, a strong correlation for heat transfer to a ceiling based on the
flame height to ceiling distance was shown for large scale flames impinging on a ceiling [120].
The following criteria were evaluated to determine where research gaps may exist regarding NGV’s
in tunnels.
2. Failure Modes
3. Consequences
4. Validation
The scenarios that lead to a failure mode have been identified as impacts to the vehicle or failure of
the TPRD, hardware degradation or failure, and operator error which may lead to a release of fuel.
Based on the vehicle and fuel composition state, a BLEVE scenario can be a risk. Failure modes
include immediately ignited releases or delayed ignition of releases that might accumulate in tunnels.
Component-level failure modes are beyond the scope of this document. Regulations on valves,
pressure containers, and other overall fuel storage/delivery systems govern these risks. The failure
modes are broken down for a CNG vehicle by Chamberlain and Modarres [135] as the following:
 Fire potential from fuel leakage.
 Explosion potential from uncontrolled dispersion and mixing of CNG in the presence of an
ignition source.
 Electrostatic discharge.
Multiple experiments and simulations were reviewed pertaining to fuel releases from NGVs, the
consequences, and the effects on tunnels. It was shown that the congestion at the roof of a tunnel
can accelerate the flame speed during a vapor cloud combustion, but the chances of this yielding
significant damage or DDT occurring are very low [121] [120] [122]. Furthermore, [137] showed that
with proper ventilation, the risk from overpressure during a combustion are significantly mitigated.
In terms of heat transfer, 60 kW/m2 was shown to be around the maximum expected rate for a jet
fire impinging on the ceiling. It was shown this applies to flames of a wide range of power [125].
Additional consequences of these failure modes include impacts/projectile-generated hazards due to
95
fuel being stored at high pressure, chemical hazards (gas toxicity, asphyxiation potential, and higher
hydrocarbons in CNG may be considered neurotoxins even though CNG is relatively nontoxic),
peak overpressure due to ignition ranging from 30 to 45 kPa based on the pressure of the release
and release amount as shown in [126] and [127], and fireballs that can reach nearly 200 m down a
tunnel based on the tunnel geometry and the release pressure [126].
Initial observations show that flammable cloud formation for CNG becomes negligible as
ventilation increases [112]. The harmful and lethal effect distance of CNG for a continuous release
event is significantly lower than an LNG release. A risk analysis for a CNG bus system shows the
highest risk comes from degradation of the system and components [135].
Despite the large quantities of studies carried out pertaining to the risks and hazards associated with
NG, research gaps were identified. As discussed thoroughly in Sections 4.5.1 and 4.5.3, the
consequences of scenarios have been evaluated and validated in multiple experiments and
simulations. Research gaps regarding consequences are the same as for scenario identification:
limited evaluations of NGVs specifically in tunnels have been carried out. Several specific research
needs include:
 The risk of spalling of tunnel surface from flame impingement or heat from a NG jet flame.
 Experimental studies of NG dispersion and overpressure in actual or scaled down tunnels
(similar to HyTunnel evaluations).
 Partially pre-mixed (realistic extents of pre-mixing) ignition in tunnels to determine
maximum overpressure.
 Large scale NG flames heat transfer analysis. So far only lab-scale or simulated data found in
the literature.
 Validation of the experiments and modeling studies.
Additional attention should be given to the size or class of the vehicle. As vehicular class increases
so does the amount of stored fuel. This can lead to an increased flame duration or formation of a
larger flammable vapor cloud. Table 29 (below) shows the approximate equivalent ambient gas
volumes of stored NG per vehicle class.
The vehicle class groups used were those defined by the Federal Highway Administration [138]. The
storage volumes used for this approximation are representative of a typical tank size based on
vehicle class. The actual storage volume varies from vehicle to vehicle within the same class based
on design choices of the manufacturer. Since the vast majority of LNG vehicles are large scale
commercial vehicles, the approximate LNG tank sizes are based off reported values from this
vehicular class.
96
Table 29. Natural gas fuel amounts by vehicle class.
FHWA Class - CNG Characteristic Mass Characteristic Ambient Unmixed Volume
2 22.0 kg 34 m3
3 132.0 kg 201 m3
4 225.0 kg 343 m3
5 125.0 kg 191 m3
7 550.0 kg 838 m3
FHWA Class - LNG Characteristic Mass Characteristic Ambient Unmixed Volume
5 or greater 200.0 kg 305 m3
5 or greater 250.0 kg 381 m3
Per the Federal Highway Administration, class 2 represents SUVs, pickups and utility vans, 3
delivery trucks and vans, 4 passenger buses, and 5-7 represent a wide range of multi axle commercial
trucks. CNG classes 4 and 7 vehicles typically have multi-tank configurations. The listed LNG
configurations consist of variously sized single tanks. There are many combinations of tank number
and size variations and the values listed above are only representations of some typical
configurations listed on select manufacturers websites [139] [140] [141].
97
98
5. PROPANE VEHICLES

What is commonly called propane is liquefied petroleum gas (LPG) that has constituents aside from
the chemical propane. The terms propane and LPG will be used mostly interchangeably in the
remainder of this report (depending in part on the source of information presented). Propane
vehicles offer consumers an alternative transportation option to conventionally fueled vehicles.
Propane vehicles are available as both dedicated and bi-fuel types. In dedicated propane vehicles,
propane is the sole energy source. In bi-fuel propane vehicles, there are two separate fueling systems
that allow the vehicle to use either propane or gasoline. Generally, propane vehicles are comparable
to conventionally fueled vehicles in terms of power, acceleration, cruising speed, and driving range.
There are several different types of propane vehicles currently in operation, including light, medium,
and heavy-duty vehicle classes. Lower maintenance costs, good performance in cold-start
conditions, and low carbon and low oil contamination characteristics make propane vehicles popular
for trucks, taxis, street sweepers, and school buses [142].
5.2. Properties of Propane Storage

Generally, propane is stored under pressure in a two-phase state. When a storage tank is full, the
propane is largely liquid with some gaseous vapor. The gaseous vapor in the storage tank is what is
used as fuel.
Figure 47: Example of Two-Phase Propane Storage (from [143])
The key properties that make propane an attractive auto fuel is its high-octane value and
compatibility with spark-ignited internal combustion engines. Propane naturally occurs as a three-
carbon alkane gas with the chemical formula C3H8. Propane can be liquefied through moderate
pressurization at 1,220 kPa (177 psia) [144], which increases its energy density by a factor of 270
over the gaseous form [145]. In terms of vehicle usage, the propane fuel must consist of at least 90%
propane, no more than 5% propylene, and no more than 2.5% other gases such as butane and
butylene [146]. Table 30 shows some of the physical and chemical properties of propane [147].
99
Table 30: Physical and Chemical Properties of Propane (from [147])
Property Value
Chemical Structure C3H8
Energy Density 23,648 MJ/m3
Stoichiometric Air/Fuel Ratio 15.5
Gas Density at 15 °C 1.85 kg/m3
Liquid Density at 15 °C 505 kg/m3

Auto-ignition Temperature 450 °C (724 K)
Specific Gravity at 15 °C 0.85

There are two main safety considerations related to propane vehicles. The first is tank over-
pressurization due to either tank overfill or environmental changes. To address this issue, propane
vehicles are equipped with overfill prevention devices (e.g., bleed valves) and pressure release
devices to vent the tank if pressure rises beyond safe levels [146]. The second safety consideration is
ignition of propane fuel that has either been released through the overfill prevention device/
pressure release device or released from a vehicle crash. Propane gas is heavier than air at standard
temperature and pressure (STP) unlike natural gas or hydrogen. Therefore, vapors can collect in low
areas such as service pits. Unlike hydrogen and natural gas, propane vapors will dissipate primarily
based on air movement rather than buoyancy and will dissipate faster in windy conditions than in
still conditions [146]. Table 31 shows relevant flammability properties of propane.
Table 31: Flammability Properties of Propane

Property Propane
Flammability LFL 2.1%
Concentration in
Air (vol%) [27] UFL 9.5%
Most Easily Ignited Mixture in Air (vol%) 4% [105]
Adiabatic Flame Temperature [107] 2250 K
Buoyancy (ratio to air) 1.52
MIE [108] [109] 0.25-0.26 mJ
Autoignition Temperature [110] 455 °C
100
Propane vehicles have robust safety standards and regulations with regard to the energy storage
system, the vehicle itself, and the roadway structures on which they operate.

temperature exposures, irrespective of the fire location, ventilation rate or type [113].

consideration of the geometric parameters of the tunnel, fire protection features and response times

NFPA 58, Liquefied Petroleum Gas Code, provides storage, use, and handling requirements for liquid
propane fuel in storage containers [148].
101
5.5. LPG Research Summary in Tunnels
This section documents the results of the evaluations regarding propane vehicle failure in a tunnel.
5.5.1. Experiments
This literature survey includes limited experiments that evaluate the hazard of propane vehicles in a
tunnel. Because of this, other experiments are reviewed that have relevance to propane vehicles in
tunnels and can help characterize the hazard.
5.5.1.1. Experimental investigation and CFD modelling of the internal car park
environment in case of accidental LPG release
Brzezinska and Markowski [149] performed a series of full-scale experiments to help understand
emissions and flammable cloud formation in order to help validate models using the CFD code Fire
Dynamics Simulator (FDS) by NIST. While the focus of these experiments are releases into parking
areas such as garages, it is still a confined space and characteristics and take-aways can be applied to
help understand the hazard in a road tunnel. An event tree was used to help understand the initiating
events and possible outcomes from an LPG vehicle. These outcomes are based on the leak type and
location as well as the ignition sources and safety systems such as detection and ventilation. This can
be seen below in Figure 48:
Figure 48: Event Tree for Propane Event Scenario (from [149])
Based on the event tree, release scenarios were determined which included three different release
rates (from 1 mm, 3 mm, and 6 mm diameter holes) with and without the ventilation for a total of
six different tests. The experimental setup featured concentration measurement points at both 10
and 30 cm off the ground and is shown in Figure 49 below:
102
Figure 49: Experimental Setup (from [149])
The experiment was conducted in a test setup with dimensions of 23.7 m × 4.2 m and a height of 6
m. Semiconductor-type gas sensors are used to measure the concentration of LPG vapor. A total of
0.17 L of LPG was fitted in the car for each test. The ventilation system draws 0.61 m3/s of air.
Table 32 below gives more details for each individual test:
Table 32: Gas Flow Test Parameters (from [149])
Diameter of Release Release Gas Outflow Ventilation Switch
Test Set #
Hole (mm) Duration (s) Rate (L/s) on Time (s)
1 1 20.75 0.008 Not Active
2 3 5.30 0.032 Not Active
3 6 3.95 0.043 Not Active
4 1 20.75 0.008 150
5 3 5.30 0.032 25
6 6 3.95 0.043 27
Once the experiments were conducted, the data was used to compare with the various CFD
simulations. These simulations are shown below in Figure 50 with emissions from the 6 mm hole.
The color scale at 4 x 10-3 kg/m3 (red) corresponds to 10% of LFL.
103
Figure 50: CFD Results based on Ventilation (from [149])
The results from the experiment show that the concentration varies considerably over time and
space. The 6 mm experiment yielded the highest fuel concentration. The concentration was in the
flammable range for approximately 10 seconds before dissipating. 10 cm above and 3 m away from
the source, the concentration was 180% LFL, whereas 30 cm above it was only 33% of LFL. At a
distance 9 m away, the concentration ranged between 10% to 20% LFL based on the height. These
results can be compared with the equivalent CFD simulation in Figure 51 below (at 3 m from the
source on the left, or 3L) with emissions from the 6 mm hole.
Figure 51: Concentration LPG Comparing Simulation vs. Experimental Results (from [149])
104
The results show very similar magnitudes and characteristics when comparing the model with the
experimental data. Both the 10 cm and the 30 cm measurements off the ground have very similar
characteristics to the model. This shows the value of using the CFD simulations to model the
phenomenon of a LPG vehicle failure that causes gas dispersion.
5.5.1.2. Smoke Control in Sloping Tunnels

Atkinson and Wu [150] studied the effects of ventilation and smoke development due to a propane
fire in a tunnel. Tunnel slopes from 0° to 10° were used to help develop a slope factor for smoke
development and movement. A model tunnel with a height of 244 mm with an arch shape cross
section which comprises a semicircular head on walls splayed out at 7 ° per Oka and Atkinson [151]
shown in Figure 52.
A 100 mm diameter porous bed propane burner was used with the top set flush with the tunnel
floor. The flow rate ranged from 2 to 10 L/min. This corresponded to a 15 to 75 MW fire. The
propane output velocity was about 0.4 to 2 cm/s. Understanding the gradient of a tunnel with a fire
downhill is important in designing emergency smoke control to keep evacuation routes clear of
smoke. The critical velocity is explored to help understand the minimum required ventilation
velocity that does not allow the smoke to back flow past the fire. This can be shown in Figure 53
below:
105
Figure 53: Critical Velocity vs. Tunnel Slope (from [150])
The critical velocity is a function of the slope of the tunnel θ, the volumetric flow of propane, and
the dimensionless variables Q* and V*. These transitional values are determined via Froud scaling as
shown by Oka and Atkinson [151]. In Figure 53 the black square indicate a flow of 10 L/min of
propane & Q*=0.44, the circles indicate 5 L/min & Q*=0.22, the + indicates 2 L/min & Q*=0.088,
and the line ‘a’ indicates SES (Subway Environment Simulations) computer program predications
from a former methane experiment whereas the line ‘b’ indicate the predications based on this work.
From there, the expressions for the critical velocity are given as the following:
Figure 54: Critical Velocity Correlations (from [150])
Figure 54 shows the critical velocity relationship in a tunnel as a function of the slope of the tunnel
θ, the volumetric flow of propane V*, and the transitional value Q* as well as the tunnel height H
and gravity constant g.
5.5.2. Modeling
An analysis of the release rates and dispersion from a propane vehicle showed that the maximum
volume of the resulting flammable cloud is dependent on the ventilation type (see Section 5.5.2.1).
106
Also, the propane vehicle explosion in a tunnel was modeled and the explosion load and
consequence were evaluated (see Section 5.5.2.2). Additionally, refer to Section 5.5.1.1 for CFD
modeling information.
5.5.2.1. LPG Dispersion Analysis

A hazard analysis of LPG, LNG, and gasoline fueled vehicles, was performed for the Massachusetts
Highway department by Zalosh et al. [137]. The study evaluated the release rates and subsequent
dispersion from a LPG fueled vehicle in a tunnel with various types of ventilation systems. The
primary accident scenario evaluated a fuel line break at the junction to the fuel tank of an LPG van.
The liquefied gas was modeled to flow directly onto the tunnel road surface from the tank, which
was assumed to be located under the van. The cross-section of the tunnel is rectangular with
dimensions of 8.25 m wide by 4.27 m high. Both transverse and longitudinal ventilation was
evaluated for release from tanks with and without excess flow valves, which limits the fuel release
rate to 100 g/s [137].
The fuel release rates and fuel vaporization histories were calculated using a mathematical model
similar to the work by Webber and Jones in 1987 [152] that evaluated the relevant phenomena (see
Figure 55). The vaporization rate calculated from the mathematical model was input into a CFD
simulation (Fluent) that calculated the vapor dispersion as a function of ventilation type [137].
Figure 55: Relevant Phenomena in LPG Spill & Vaporization Mathematical Model (from [137])
For the cases without an excess flow valve, the maximum volume of the resulting flammable cloud
is comparable to that of a gasoline spill for transverse ventilation. In the case of longitudinally
ventilated tunnels, the vapor cloud is smaller than that of gasoline when the ventilation velocity is 1
m/s. For the case with an excess flow valve, in which the release rate is limited to 100 g/s, the LPG
cloud would be no larger than that of gasoline. The growth and decay of the vapor cloud with 20
cfm per lane foot normal transverse ventilation showed that LPGs growth rate is like gasoline and
slower than LNG. Each result is further summarized in below [137].
107
Table 33: Transient Dispersion for Transverse Ventilation (from [137])
Max Half Max Volume
Depth of Length of of
Vaporization Cloud Flammable Flammable Flammable
Fuel Rate (kg/s) Duration Cloud Cloud Cloud
Max: 0.477 0.77 m under the 840 m3 at 100
LNG 260 sec 43 m
Avg: 0.344 ceiling at 100 sec sec
Max: 0.542 0.4 m under the 722 m3 at 492
LPG 900 sec 60 m
Avg: 0.435 van sec
Gasoline Max: 0.621 940 m3 at 420
>500 sec 0.5 m 69 m at 353 sec
A Avg: 0.398 sec
Gasoline Max: 0.339 800 m3 at 570
980 sec 0.45 m 80 m at 589 sec
B Avg: 0.139 sec
Max: 0.78
CNG 240 sec 0.55 to 0.82 m 50 m at 80 sec 530 m3
Avg: 0.35
5.5.2.2. Explosion Risks and Consequences for Tunnels

Weerheijm and van den Berg [153] developed engineering models to quantify the explosion load and
consequence of an LPG explosion in a tunnel. The release rate of the gas, the duration of the flow,
geometric parameters, and ventilation are inputs to a CFD code developed by TNO. The result is a
model of gas dispersion and gas cloud size concentration as a function of time. The explosive loads
for the gas explosion are then estimated by relating the cloud length and concentration to the
overpressures of an equivalent stoichiometric cloud. The development of a flammable cloud is
dependent on the ratio between the ventilation air speed and the leak rate of the LPG. The
explosion strength is largely dependent on the length of the flammable cloud. Moreover, detonation
could occur if the flammable cloud is sufficiently long [153].
The first case modeled a nearly full vessel in which a boiling liquid expanding vapor explosion
(BLEVE) may occur. The tunnel cross-section rectangular with dimension of 5 m by 14.4 m (72 m2).
In the area around the bursting vessel (within 8 m), the explosion results in a high-pressure impact
with the tunnel lining (600-800 kPa). The flow is then redirected in the axial direction of the tunnel
and the magnitude of the pressure is reduced. The second case evaluated a blast load resulting from
the rupture of a nearly empty LPG vessel. This case resulted in similar phenomena as the BLEVE;
however, the magnitude of the pressure impact with the tunnel lining in the area around the bursting
vessel (within 6 m) was much lower (200-500 kPa) [153].
5.5.3. Analysis
A risk analysis was performed which ranked the probabilistic outcomes of a tunnel accident for
different fuel types (see Section 5.5.3.1). Additionally, refer to Section 5.5.1.1 which includes a risk
informed failure tree.
108
5.5.3.1. LANL Risk Analysis
A comparative risk analysis was performed by Los Alamos National Laboratories to put into
perspective the relative hazards of alternative fuels, including LPG, compared to gasoline and diesel
fuels [136]. As part of this analysis, data was collected about the physicochemical properties of LPG
and general petroleum and transportation information. Subsequently, the technical properties of the
alternative fuels were ranked, safety data and vehicle accident statistics were reviewed, and specific
accident scenarios were evaluated. One of the accident scenarios that was evaluated was a vehicle
collision with fuel loss in a tunnel. An expert panel estimated the probabilities of different outcomes
for the different alternative fuels compared to gasoline and diesel. Table 34 shows the results of the
risk analysis. These probabilities are just expert judgement rather than data. As shown, the LPG was
estimated to be more likely to result in a fire without injury or not ignite, and less likely to result in
an explosion when compared to gasoline [136].
Table 34: Probabilistic Outcomes of a Tunnel Accident (from [136])
Probability
Consequence CNG LNG LPG Gasoline Diesel
Fire without Injury 0.1 0.1 0.15 0.12 0.02
Fire with Injury 0.05 0.05 0.08 0.1 0.01
Explosion 0.05 0.05 0.12 0.2 0.01
No Ignition 0.8 0.8 0.65 0.58 0.93
109
5.6. LPG Research Gaps
Through this literature study, it was shown that studies of LPG vehicles in tunnels is limited.
Experiments that explicitly evaluate the hazards of propane vehicles in tunnels were scarce.
However, there were other experimental studies included that may have relevance to propane
vehicles in tunnels and can help characterize the hazard. One such experiment evaluated emissions
and flammable cloud formation in confined spaces such as parking areas and garages from an LPG
vehicle [149]. Also, the effects of ventilation and smoke development due to a propane fire in a
tunnel were evaluated [149]. Modeling studies have been conducted on the release rates and
dispersion from a propane vehicle and the consequence of an explosion [137] [153]. Additionally, a
risk analysis was performed which ranked the probabilistic outcomes of a tunnel accident for
different fuel types, including propane [136].
The conclusions about important variables that can be derived from these studies is limited. As far
as dispersion and cloud formation, the flammable propane cloud is dependent on the ventilation
type and tunnel geometry. Initial findings show how much ventilation makes a difference on the
overall concentration of a LPG spill. Figure 50 shows how after nearly two minutes the LPG
concentration is negligible in compared to the case without ventilation. Analysis shows that the
probabilistic outcome of a various consequences from LPG is very similar to that of gasoline [136].
Dispersion properties are very similar to that of gasoline as well [137].
The following criteria were evaluated to determine where research gaps may exist regarding LPG
vehicles in tunnels.
2. Failure Modes
3. Consequences
4. Validation
In terms of scenario identification and failure modes, explicit studies on these topics were not found
for propane vehicles. However, the failure modes can be determined from the event tree in Figure
48 by Brzezinska and Markowski [142]. These include release scenarios that lead to flash fires or
explosions based on the amount of fuel released, ventilation and cloud dispersion, as well as when
there is ignition. Also, a limited scenario identification study was conducted that documented the
likelihood of different consequences of a tunnel accident. However, a more complete evaluation of
failure modes would need to be completed.
For each of the initiating events, there are several variables that effect the magnitude of the
consequence: LPG quantity released, ventilation, obstructions, ignition time, tunnel geometry, etc.
The measurements of the consequences of the failure mode include overpressure, HRR, dispersion,
and resulting structural damage. The study by Brzezinska and Markowski [142] completed validation
by understanding the failure tree to inform the experimental setup. From there the experimental data
was used and compared with a CFD model for validation and to use the model for other release
events.
In contrast to the some of the other alternative fuels, there has not been significant research on LPG
vehicles in tunnels. The following research gaps were identified:
 Fully evaluate the initiating events that are risk significant in terms of LPG vehicles in
tunnels.
110
 Evaluate different classes of LPG vehicles, including buses, and multiple LPG vehicles.
 Evaluate the HRR, temperature, and structural damage resulting from different failure
modes.
 Evaluate the effect that overpressure, deflagration, and DDT of released propane has on
structural components of tunnels.
 Evaluate the effect that variables such as ventilation, obstructions, and tunnel geometry have
on the consequence of an LPG vehicle failure.
 Investigate thermal consequences of failures, which have not been reported for many of the
studies included in this literature survey.
 Perform additional validation of modeling efforts through direct comparison to experimental
results.
 Additional attention should be given to the size or class of the vehicle. As vehicular class
increases so does the amount of stored fuel.
111
112
6. HYDROGEN FUEL CELL ELECTRIC VEHICLES

Hydrogen fuel cell electric vehicles (FCEVs) are part of a comprehensive portfolio of technologies
and can offer consumers an alternative transportation option to conventional options such as
internal combustion engines (ICEs). Fuel cells are more efficient than combustion technologies and
FCEVs qualify as zero emission vehicles, emitting no pollution  only water vapor and air through
the tailpipe. Additionally, these vehicles offer fast fueling times, and comply with both
manufacturer’s requirements and consumer expectations for driving range [154]. There are several
types of hydrogen FCEVs available to support the diversification of U.S. energy sources in the
transportation sector. While there is growing interest in medium and heavy duty FCEVs, production
of light-duty hydrogen FCEVs has been ongoing since the Hyundai ix35 fuel cell vehicle rolled off
of the assembly line in February of 2013 [155]. An infrastructure of refueling stations has been
developed both regionally in the U.S. and in several locations internationally [156]. As of late-2019,
there were over 7,800 commercial (sold/leased) fuel cell passenger FCEVs on US roads, mostly in
California, with that number projected by industry to exceed 23,000 in 2021 and 47,000 in 2024 [98].
Although these dates and number of deployments may be updated, global industry manufacturers
have made a number of plans for commercial expansion, particularly for larger vehicles and trucks,
to complement other vehicle platforms such as battery electric vehicles, plug in hybrids, and
biofueled ICEs. In addition to FCEVs there are a number of other hydrogen fuel cell applications.
For example, there are over ~30,000 fuel cell-powered forklifts operating in commercial warehouses
and distribution centers by companies such as Amazon, Coca-Cola, FedEx, Kroger, Walmart, and
more as of late-2019 and over 20 million hydrogen fuelings to date [157] [158] [159]. Buses and
medium-/heavy-duty vehicles have utilized hydrogen fuel cell technology for public transportation
and commodity distribution [160]. The implementation of hydrogen for these larger scale vehicles is
expected to increase due to the difficulty in fully decarbonizing these modes of transport. Because of
the growing market and diverse applications, a robust safety analysis of hydrogen FCEVs is
necessary to ensure public safety.
6.2. Properties of Hydrogen

As an energy carrier, hydrogen fuel can either be a compressed gas or a low-pressure cryogenic
liquid. Hydrogen is the lightest gas (~1/14 as dense as air) and at standard temperature and pressure
exists in the form of a hydrogen molecule with two atoms: H2. Liquid hydrogen has a boiling point
of -252.88 °C and is much more dense than gaseous hydrogen [161]. Gaseous hydrogen can be
stored in high pressure tanks to provide large amounts of energy; however, even more energy can be
stored in low pressure cryogenic liquefied form. Hydrogen has an expansion ratio of 1:848, which
means that gaseous hydrogen at atmospheric conditions occupies 848 times more volume than
liquid hydrogen [162]. Table 35 shows physical and chemical properties of hydrogen [163].
113
Table 35: Physical and Chemical Properties of Hydrogen (from [163])
Property Value
Molecular weight 2.0159
Gas Density 0.08988 g/L @ 0°C, 1 atm
Relative Vapor Density 0.07
Liquid Density 70.8 g/L @ -253°C
Melting Point -259.35°C
Boiling Point -252.88°C
Auto-ignition Temperature 500°C
Flammability Limits 4-75% (vol % in air)

The primary safety hazard associated with hydrogen is that it is flammable. Hydrogen properties
require that the fuel delivery system be designed to mitigate all relevant safety hazards. Table 36
shows relevant flammability properties of hydrogen as compared to other common fuel sources.
Table 36: Flammability Properties of Hydrogen and Other Fuels
Gasoline
Property Hydrogen Methane Propane Vapor
Flammability LFL 4.0% 5.0% 2.1% 1.2%
Concentration in
Air (vol%) [27] UFL 75.0% 15.0% 9.5% 7.1%
Easily Ignited Mixture in Air (vol%) 29% [103] 8.5% [104] 4% [105] 2% [106]
Adiabatic Flame Temperature [107] 2483 K 2236 K 2250 K 2289 K
Buoyancy (ratio to air) 0.07 0.54 1.52 4
MIE [108] [109] 0.011-0.017 mJ 0.28-0.30 mJ 0.25-0.26 mJ 0.8 mJ
Autoignition Temperature [110] 500 °C 580 °C 455 °C 246 – 280 °C
Although hydrogen’s lower flammability limit is comparable to the other fuels, it has a much higher
upper flammability limit. Also, its minimum ignition energy is an order of magnitude lower than the
other fuel types. This introduces the possibility of ignition even from weak electrostatic discharged.
Sources such as NFPA 77 [164] give discharge ranges showing that even a corona type discharge at
the end of a wire or other point could lead to enough energy to exceed the MIE for hydrogen.
Figure 56 illustrates the MIE of different fuels as a function of concentration in air by volume. As
shown in the figure, between approximately 10% and 60% volumetric concentration, hydrogen has a
lower ignition energy than methane and gasoline over a much wider range of concentration.
However, for hazard evaluation the MIE of lean mixtures is more relevant, and hydrogen does not
differ from other fuels. At the LFL concentrations for each of the fuels, the ignition energies are
much more similar between fuels.
114
However, it should be noted that these characteristics have led to robust system safety requirements
to reduce the likelihood of hydrogen release after an accident.
Figure 56: Minimum Ignition Energy for Different Fuels vs. Concentration (from [165])
To mitigate the ignition hazards of hydrogen, sensors are placed in indoor and enclosed locations
where hydrogen has the potential to be trapped and accumulate flammable concentrations. These
sensors can be programmed to alert when the hydrogen reaches some fraction of the LFL. Because
hydrogen is lighter than air, sensors should be placed above potential release points but below
ceiling height to avoid elevated temperatures. Consideration should be given to understand the
effect of ventilation systems and how air flow might be altered [166]. Most of the hydrogen fuel
system will be at a pressure that will result in momentum driven jets of hydrogen. In outdoor
locations, hydrogen releases rise away from ignition sources because it is more buoyant than air. This
means that hydrogen leaks can dissipate readily, potentially avoiding a concentrated, explosive
atmosphere.
115
Hydrogen FCEVs have robust safety standards and regulations regarding the fuel storage system,
the vehicle itself, and the roadway structures on which they operate.
6.4.1. Global Technical Regulation No. 13

The Global Technical Regulation No. 13 (GTR #13) establishes vehicle requirements for hydrogen
FCEVs that can attain equivalent levels of safety as those for conventional gasoline powered
vehicles. GTR#13 is intended to be applied globally. However, it is up to specific regulatory bodies
in each country to adopt the GTR. Because of the large number of countries implementing
hydrogen vehicles and developing their jurisdiction-specific requirements, especially for tunnels, this
report does not attempt to catalogue these requirements and regulations. However, during the IPHE
RCSS Working group meeting in September 2018, many of those present shared the regulations
regarding tunnels and enclosed spaces.
Hydrogen vehicle fuel is contained in a composite overwrapped pressure vessel and stored in the
gaseous state. The pressure vessel includes a thermal pressure relief device which, in the event of a
fire, releases the hydrogen to prevent the vessel from over-pressurizing. Current storage systems
have pressures of up to 10 ksi (70MPa). GTR #13 provides requirements for the integrity of
compressed and liquid hydrogen motor vehicle fuel systems, including pressure cycling tests, a burst
test, a permeation test, and a bonfire test. The pressure cycling test evaluates a container’s durability
to withstand, without burst, 22,000 cycles of pressurization and depressurization. The burst test
evaluates a container’s initial strength and resistance to degradation over time. The bonfire test
evaluates the ability of the container’s thermal pressure relief device to open in a fire scenario
(localized and engulfing) [167].
For Crash testing, GTR #13 specifies that participating countries will use existing national crash
tests but develop and agree on maximum allowable levels of hydrogen leakage. In the U.S., these
national crash tests are found in the Federal Motor Vehicle Safety Standards (FMVSS) which
includes specified tests for barrier impacts, rear collisions, and side impact crashes. In a later phase
of the requirement, the international crash test requirements are planned to be unified for FCEVs
[113].

temperature exposures, irrespective of the fire location, ventilation rate or type [113]. With regards
to hydrogen vehicles in tunnels, appendix G recommends on-board detection and incident shutoff
systems be provided in fuel-cell vehicles.
116

National Cooperative Highway Research Program Guidelines for Emergency Ventilation Smoke
Control in Roadway Tunnels Chapter 2: Road Tunnel Fires provides guidance on fire design
parameters for tunnels. This includes consideration of the geometric parameters of the tunnel, fire
protection features and response times that leads to decision making using NFPA 502. This is
chapter provides a framework on how to understand and determine fire and hazardous materials
management in tunnels [40].

NFPA 55, Compressed Gases and Cryogenic Fluids Code, provides storage, use, and handling requirements
for both compressed and cryogenic liquid hydrogen in portable containers, cylinders, and tanks.
Sections 10 and 11, specifically, deal with bulk hydrogen compressed gas systems and bulk liquefied
hydrogen systems, respectively [115].

NFPA 2, Hydrogen Technologies Code, establishes the necessary requirements for hydrogen
technologies. This includes requirements associated with general fire safety, explosion protection,
fueling facilities, fuel cell power systems, hydrogen generation systems, combustion applications,
laboratory operations, and enclosed spaces [168]. All hydrogen requirements from other NFPA
documents, including NFPA 55, are included by reference in NFPA 2 to provide a single source of
hydrogen requirements in in NFPA.
117
6.5. FCEV Research Summary in Tunnels
There has been substantial work in evaluating the effects of a failure of a hydrogen tank on an
FCEV in a tunnel. This section documents the results of these evaluations.
6.5.1. Experiments
Several experiments have evaluated the consequences of a hydrogen FCEV failure in a tunnel. A
series of experiments were performed to determine what would happen if hydrogen is released from
the onboard pressure vessel. It was determined that spontaneous ignition is the most likely
consequence (see Section 6.5.1.1). Qualitatively, this is the least severe and most likely consequence
to a hydrogen release in a tunnel. However, there were also several experiments performed to
evaluate more severe consequences. Multiple experiments were conducted to evaluate deflagration
of hydrogen within a tunnel. These experiments investigated the consequences to delayed ignition of
the released hydrogen, considered a worst-case scenario because if ignition does not occur
immediately, a large volume of flammable gas could build up and impart more energy into the
confined space of a tunnel once it does ignite. An immediate ignition scenario involves less
accumulation of hydrogen involved in the ignition event. The concentration of hydrogen and
presence of ventilation had a significant effect on the measured pressure pulses (see Section 6.5.1.2).
A variety of quiescent and steady-state hydrogen ignition experiments were performed to evaluate
the effect of congestion, hydrogen release rates, along with ventilation rates on overpressure. In
general, congestion increased overpressure; however, low hydrogen leakage rates and increased
ventilation air velocity resulted in lower overpressure (see Section 6.5.1.4). Also, deflagration was
examined in stratified hydrogen layers to evaluate the potential of self-sustained detonation in flat
layer hydrogen-air mixtures. The results indicated that a DDT was possible, however, a minimum
layer thickness and sufficient congestion was required (see Section 6.5.1.5). A series of fire
experiments and simulations of a car carrier in a tunnel loaded with hydrogen FCEVs simulated the
HRR and showed similar results when compared to the experimental results (see Section 6.5.1.6).
Also, experiments have been performed to validate the results of CFD models (see Section 6.5.1.3).
6.5.1.1. Spontaneous Ignition of Pressurized Releases of Hydrogen into Air

A series of experiments were performed to show that the spontaneous ignition of released hydrogen
is caused by transient shock formation and mixing associated with rupture of a burst disk between
compressed hydrogen and air [169]. Several different variables were evaluated through these
experiments, including rupture pressure and internal geometry downstream of the burst disk. The
rupture pressure of the burst disk was evaluated with both commercial and in-house manufactured
disks with different rupture pressures. The majority of experimentation was performed outdoors,
with ambient conditions (between 280K and 305K, with between 60-90% relative humidity). Figure
57 shows a schematic of the experimental configuration [169]
118
Figure 57: Schematic of Experimental Configuration (from [169])
Over 200 experiments were conducted with hydrogen failure pressures between 11.2 atm and 113.25
atm, with various upstream and downstream geometries. These experiments demonstrated that
spontaneous ignition of compressed flammable hydrogen repeatedly occurs at the range of pressures
seen in FCEV applications, given that sufficient mixing occurs as well. The short mixing time scales
are provided by the pressure boundary failure geometry, multi-dimensional shock-boundary, the
shock-shock interactions, and the molecular diffusion. Continued combustion can occur because the
turbulent free jet hydrogen flames can be stabilized at sufficiently high jet velocities. The reflected
shock and shock-shock interactions determine the minimum compressed hydrogen pressure at
which spontaneous ignition occurs. Due to the repeatability of the ignition and the characteristic
time scale, it was determined that alternative ignition sources, such as electrostatic discharge, did not
contribute to these experimental results [169].
6.5.1.2. Large-scale Hydrogen Deflagrations and Detonations

A scaled down tunnel was used to perform spark-initiated deflagration tests using homogeneous
hydrogen mixtures by Groethe et al. [170]. A 1/5-scale tunnel was used to perform multiple
experiments with varying released quantities of hydrogen with and without ventilation. This was
done to simulate the release from a fuel cell vehicle or storage cylinder on a hydrogen transport.
Additionally, selected tunnel tests contained obstacles representing traffic to investigate turbulent
enhancement. The cross-area blockage ratio was 0.03. Figure 58 and Figure 59 show the tunnel
facility and model vehicles in the tunnel, respectively. Hydrogen was contained in a 37 m3 volume at
the center of the tunnel by HDPE plastic film barriers in homogeneous mixtures ranging from 9.5%
(0.32 kg) to 30% (1 kg) hydrogen mixed with air in that volume. Prior to the spark ignition, the
plastic barriers were cut. Additional experiments evaluated different release rates of hydrogen both
with and without forced ventilation [170].
119
Figure 58: SRI Corral Hollow Experiment Site Tunnel Facility (from [170])
Figure 59: Model of Vehicles in Tunnel (from [170])
The results of the experiments showed that the 9.5% homogeneous hydrogen mixture produced
pressure pulses that were too small for sensors to detect. When the hydrogen content in the mixture
was increased to 20% and 30%, the pressure pulses measured 35 kPa and 150 kPa, respectively. It
was shown that the presence of the vehicles had an insignificant effect on the deflagration as shown
by the pressure pulse, but that ventilation during a release reduces the hazard dramatically. Also,
release of hydrogen through a source like the vehicle fuel tank safety release valve produced very
lean hydrogen concentrations which created very small pressure pulses [170]. Figure 60 below shows
the pressure impulse and overpressure associated with the 30% hydrogen experiment:
120
Figure 60: 1/5th Scaled Tunnel Impulse and Overpressure (from [170])
As the Groethe et al. points out, larger vehicles should be studied in this scaled experiment to
understand the results. The scaling on how the overpressure changes with the size of tunnel and
vehicle would need to be further investigated. As noted by Groethe, the tunnel has a larger aspect
ratio than a normal tunnel, which might affect how these results scale up. This directly correlates to
the L/D ratio or the length of the tunnel compared to the effective hydraulic diameter. Additionally,
to compare with the other literature, the location of the overpressure measurements would need to
be well known and using known scaling laws such as Hopkinson Blast Scaling and Sachs Blast
Scaling could be used to help understand the total explosive energy. Sachs scaling law states that
pressure, time, impulse, and other parameters can be expressed as functions of this scaled distance
but assumes that air behaves as a perfect gas and assumes gravity and viscosity are negligible [171].
Additionally, how effects of confinement that could lead to turbulent flame speeds might not carry
over at full scale. Further understanding these scaling laws would allow for better comparisons
between experiments. Studies such as the one by Tamanini [172] provide additional insight on
various scaling methods for sizing deflagration vents which helps understand important scaling
factors.
6.5.1.3. Releases from Hydrogen Fuel-cell Vehicles in Tunnels

In order to validate the dispersion/deflagration modeling described later in Section 6.5.2.4 a set of
experiments were performed at the SRI Corral Hollow Experiment Site (see Figure 58) by Houf et
al. [173]. A set of scaled tunnel tests were performed to approximate the full-scale dimensions of the
tunnel from the modeling effort. The hydrogen mass, release rate, initial tank pressure, and TPRD
release diameter were scaled to approximate the modeling parameters. Figure 61 shows a
comparison of the peak overpressures from the experiments with the results from the model
121
simulations. The peak overpressure from the experiments is in good agreement with the modeling
results [173].
Figure 61: Comparison of Experimental and Modeling Results (from [173])
Figure 62 shows a comparison of the hydrogen concentration at discrete locations within the tunnel
as a function of time. As shown, the predicted and measured values are generally in good agreement
[173]. While the simulation does approximate the overpressure there are some points in the data that
might be considered outliers.
Figure 62: Comparison of Time-dependent Hydrogen Concentration Values (from [173])
122
6.5.1.4. HyTunnel Project to Investigate Hydrogen Vehicles in Road Tunnels
A set of experiments were performed by Kumar et al. [174] at the Health and Safety Laboratory to
evaluate the influence of congestion and ventilation flow rates on the over-pressure produced from
ignition of hydrogen stoichiometric clouds. Quiescent experiments were performed in a sealed
enclosure with a stoichiometric hydrogen/air mixture and different congestion
volumes/configurations. The congestion configurations consisted of different arrangement of pipes
with variable spacing and orientation. Arrangement A is the tight configuration, consisting of four
rows of pipes with a spacing of three diameters between pipes, with adjacent rows oriented at right
angles and the pipes staggered between every other row. Arrangement B is the loose configuration,
consisting of 3 rows of pipes with a spacing of five pipe diameters between pipes, and the same
orientation of pipes as Arrangement A. Figure 63 shows the configuration of the ignition
experiments. The enclosure (left) shows two modules; however, for the ignition experiments, a total
of six modules were combined to give the enclosure a total length of 14.9 m and a volume of 93.1
m3. The arrangement of the tight congestion setup is also shown (right) [174]. A single obstacle
setup is used in each experiment which is shown in Figure 63.
Figure 63: Configuration in Ignition Experiments (from [174])
Table 37 and Figure 38 show the results of the quiescent ignition experiments. A non-uniform
pressure field resulted from these hydrogen ignition experiments. An increase in the volume of
hydrogen/air mixture increased the maximum explosion overpressure. However, as shown in the
table below, an initial increase in the congestion level increased the maximum explosion
overpressures (from none to congestion configuration B). Further increase in congestion (from
configuration B to configuration A) resulted in a reduction in overpressure [174].
Also, a set of steady-state experiments were performed with various hydrogen leak rates and
ventilation flow rates (while also evaluating congestion arrangements A and B). Ventilation in the
enclosure was produced through a variable speed fan producing suction through an end plate with
324 circular holes to create a homogeneous flow. Table 37 shows the results of the steady state
ignition experiments. As shown, the maximum explosion overpressures increased with increasing
hydrogen release rate and decreasing ventilation air velocity. At the lowest leakage rates, the highest
explosion overpressures were seen for the more congested configurations. However, at the highest
hydrogen leakage rates, the highest explosion overpressures were seen for the less congested
configuration (except at the lowest ventilation rate) [174].
123
Table 37: Results of Steady State Ignition Experiments (from [174])
Overpressure from Transducer (mbar)

Hydrogen
Air Congestion
Release
Velocity Configuration Congested
Rate Enclosure Left- Enclosure Right-
Volume Cage
Hand Wall Hand Wall
Wall Center
A 28.2 124.2 63.5
1 m/s
B 16.2 63.4 19.6
A 13.6 66.6 12.6
1.5 g/s 2 m/s
B 8.8 20.6 7.5
A 12.1 39.5 10.5
4 m/s
B 6 13.1 5
A 32.4 123.3 55.4
1 m/s
B 27.5 106 46.6
A 23.2 117.7 39.6
2.0 g/s 2 m/s
B 25.7 66.3 46.6
A 14.1 53.6 14.7
4 m/s
B 39.4 25.4 28.9
A 48.9 255.8 71.2
1 m/s
B 48.5 136.9 91.7
A 37.3 222.5 66
4.0 g/s 2 m/s
B 48.1 196.4 85.8
A 26 160.4 39.2
4 m/s
B 28.9 126.2 51.2
124
Table 38: Results of Quiescent Ignition Experiments (from [174])
Overpressure from Transducer (mbar)

Congested Congestion
Volume Configuration
Enclosure Left- Enclosure Right-
Hand Wall Hand Wall
None 28.2 24.7
0.098% of total enclosure volume B 37.2 42
A 27.4 24.2
None Over-range 85
0.55% of total enclosure volume
B Over-range 114.6
6.5.1.5. Deflagration and Detonation of Hydrogen Under a Tunnel Ceiling

A set of experiments were performed at Research Centre Karlsruhe in Germany by Friedrich et al.
[175] that examined deflagration in stratified hydrogen layers to evaluate the potential of self-
sustained detonation in flat mixture layers. Figure 64 shows the main experimental set up used in
these evaluations. The chamber had dimensions of 5.7 m x 1.6 m x 0.6 m with layering heights of
0.15 m, 0.3 m, and 0.6 m. The hydrogen concentrations used in these experiments ranged between
15% and 25% (by volume in air). Also, variation in obstacles and hydrogen layer thickness were
evaluated [175].
Figure 64: Experimental Setup for Deflagration Experiments (from [175])
In the set of experiments with no obstacles, slow flame propagation regimes were observed. The
experiments with obstacles showed three distinct combustion regimes. The obstructions in the
ceiling may have added turbulence to the flame propagation, which would make the explosions more
severe. These results indicate that ceiling design and mitigation measures in tunnels are important
which can be understood in the volume and layer height matrix shown in Figure 65 [175].
125
55
Figure 65: Concentration and Layer Height Effect on Combustion (from [175])
6.5.1.6. Fire experiments of carrier loaded FCEV in full-scale model tunnel

A series of fire experiments and numerical simulations of a carrier loaded with hydrogen FCEVs in a
full-scale tunnel were conducted to calculate heat release and smoke generation rates by Seike et al.
[176]. As shown in Figure 66, the experimental tunnel is 80 m long, 12.4 m wide, and 7.36 m wide
with a horseshoe cross-section. The total HRR of the carrier loaded with hydrogen FCEVs was
determined from the experimentally obtained temperature variation near the fire [176].
Figure 66: Experimental Tunnel Configuration and Carrier (from [176])
The total HRR was also estimated through numerical simulation. The individual HRR of each part
of the car was calculated and summed to determine the total HRR. The different parts of concern
were the carried vehicles without fuel, the hydrogen fuel which was approximately 17.6 m3 of low-
pressure hydrogen and another case of 43.6 m3 of high-pressure hydrogen, the rear wheels, the
driver’s seat in the carrier vehicle, and a 1 m2 gasoline pool fire. The methodology of estimating the
HRR of each part and superimposing them to obtain the total HRR was then compared to the
experimental results. As shown in Figure 67, for a vehicle containing 43.6 m3 of compressed
hydrogen, the numerical method and experimental results are in fairly good agreement [176].
126
Figure 67: Comparison of Experimental & Simulated HRR for High Pressure Case (from [176])
This methodology was extended to predict the HRR of a carrier loaded with eight hydrogen FCEVs.
It was determined that, when compared to a large bus fire, the HRR was larger after 10 minutes and
the maximum HRR was 1.5 times greater [176].
6.5.1.7. Vapor Cloud Explosions from Ignition of Gaseous Mixtures in a Congested

Region
A series of studies were carried out by Royle et al. [123] to measure the overpressure produced from
methane and methane/hydrogen mixtures premixed with air when ignited within congested spaces.
The experimental space was a 3 x 3 x 2 m region containing multiple layers of pipes. An image of
the congested region is shown in Figure 68. A concrete wall sits adjacent to the one side of the
congested region. The wall is positioned there to protect the control room and has been shown to
not interfere with free field overpressure [123]. Additionally, the wall has embedded pressure sensors
at different heights. For this series of experiments, the blockage ratio was reported as 4.40% the
total volume. The outside of the grid was covered in a 23 µm thin plastic film which contained the
gas prior to ignition. The plastic film was used only to contain the premixed gas mixtures prior to
combustion and did not significantly restrict the outflow of gas or the pressure wave.
127
Figure 68: Congestion region or grid where gas was filled then ignited (from [123])
Hydrogen gas was mixed with air to form a stoichiometric ratio of 1.2 which reportedly produces
the highest overpressures. Other gases evaluated in this study were mixtures of methane, air, and
hydrogen which are all included in some of the figures and tables below. For this section, only
results pertaining to hydrogen alone are discussed. Section 4.5.1.3 discusses the methane portion of
this experiment.
The ignition source was located at a height of 0.5 m off the ground and positioned at the center of
the grid. For ignition a 2.25 J capacitor was discharged through a spark gap of 6 mm. It was noted
that the spark exhibited lower energy than what was discharged from the capacitor. For the
instrumentation, overpressure values were measured by an array of low- and high-pressure pressure
sensors. The location of the pressure sensors can be seen in Figure 69. All pressure sensors were
positioned 500 mm above the ground except for the far field pressure sensors, which were mounted
at higher locations due to the topology of the testing pad.
128
Figure 69: Pressure Sensors distributed in and around grid structure (from [123])
Pressures were measured across a wide span of locations including up the adjacent wall. Table 39
lists the initial conditions prior to ignition. The pure hydrogen is labeled as NatHy_01. For the
results of hydrogen/methane mixture experiments, we refer the reader to the paper [123].
Table 39: Initial Conditions of Experiment (from [123])
Measurement Test Conditions: NatHy_01
Hydrogen (vol. %) 100
Number of Layers 9
Free Volume 17.207
Gas mixture temperature (°C) 11.0
Relative Humidity (%) 30.7
Atmospheric Pressure (kPa) 97.72
Mean Oxygen Concentration (%) 13.59
Partial Oxygen Pressure (kPa) 0.1359
Partial Nitrogen Pressure (kPa) 0.5127
Partial Water Vapor Pressure (kPa) 0.0041
Partial Fuel Gas Pressure (kPa) 0.3474
Mass of Hydrogen (kg) 0.498
129
It was noted that during experiment the humidity was uncontrolled but was assumed to have a
minor effect on the resultant explosion overpressure values. Figure 70 displays an image of the
explosion immediately after ignition.
Figure 70: Image of pure methane combustion right after ignition (from [123])
Figure 71 and Figure 72 show the measured overpressure values at various locations for all mixtures.
Recall that NatHy_01 corresponds to the hydrogen gas. Pressures were reported in the near-field
(within and just outside of the grid) and far-field (further out from the grid) regions.
Figure 71: Overpressure vs. of distance parallel to the wall in the near field region (from [123])
130
Figure 72: Overpressure vs. distance perpendicular to the wall in the far field region (from [123])
In the near field, values were well over 100 kPa and up to 450 kPa based on the distance parallel to
the wall. At 32 m away, the overpressure drops to less than 20 kPa in the perpendicular direction
and just above 50 kPa in the parallel direction. Referring to Table 1, 100 kPa is the fatality threshold
for direct blast effects. Anything above 200 kPa has a 99% probability of fatality from direct blast effects.
Note that this experiment represented the ignition of a pre-mixed 18 m3 region with a high level of
congestion— both the stoichiometric mixture size and level of congestion are probably unlikely to
occur in a tunnel, especially simultaneously.
6.5.2. Modeling
A series of modeling efforts have been undertaken to understand hydrogen dispersion, deflagration,
and hydrogen jet flame hazard in tunnels. Modeling was used to support a risk analysis of a
hydrogen FCEV accident in a tunnel. The objective of the modeling was to predict the thermal
expansion of the structural members and the temperature of the epoxy when a hydrogen jet flame
impinges on the suspended tunnel ceiling (see Section 6.5.2.1). Furthermore, a CFD evaluation
showed that the flame resulting from hydrogen release had the potential to damage tunnel
equipment and structure (see Section 6.5.2.6). In another study, CFD modeling was performed in
support of the evaluation of explosion risk of hydrogen vehicles (both cars and buses). A dispersion
analysis determined realistic cloud sizes and hydrogen concentrations expected after a tunnel
accident of various hydrogen vehicles (assuming delayed ignition). It was determined that the
resulting overpressure is insignificant in terms of risk to human life (see Section 6.5.2.2). Another
effort involved a turbulence modeling study evaluating hydrogen release and combustion, variable
tunnel ventilation, and variable delayed ignition time. These results showed that larger ventilation
rates decreased the growth rate of overpressure after ignition and the attenuation rate after reaching
the peak while increased ignition time delay had the opposite effect (see Section 6.5.2.5). Also, a
series of CFD simulations were performed to evaluate diffusion of leaked hydrogen in tunnels.
These simulations showed that in tunnels without ventilation, the geometry effects the hydrogen
diffusion (see Section 6.5.2.7). Finally, CFD models of hydrogen deflagration in a tunnel were
compared with the results from experiments to validate the results of the CFD code (see Section
6.5.2.3 and 6.5.2.4).
131
6.5.2.1. Hydrogen FCEV Tunnel Safety Study
CFD, heat transfer, and solid mechanics modeling was performed by Sandia National Laboratories
[113] in support of the risk analysis of a hydrogen FCEV accident in a tunnel. The scenario modeled
in support of the risk analysis was a hydrogen vehicle in an accident exposed to a resulting fire (see
Section 6.5.4 for additional details).
The objective of the modeling was to predict the thermal expansion of the structural members and
the temperature of the epoxy when a hydrogen jet flame impinges on the tunnel ceiling. The analysis
was divided into three parts: 1) a CFD simulation of the flame, 2) a heat transfer simulation of the
structural members, and 3) a solid mechanics analysis of the structural members. A Sandia-
developed code called Sierra was used to perform the simulations. Sierra is divided into different
modules that can interact with each other. The Fuego module was used for the CFD simulation, the
Aria module was used for the heat transfer model, and the Adagio module was used to calculate the
deflection of the structural members. The CFD simulation provided the boundary conditions for the
heat transfer simulation, specifically the radiative and convective heat flux on the tunnel ceiling.
Note that due to computational limitations, the smallest reasonable tank orifice diameter that can be
modeled is 5.25 cm. This is conservative because the velocity was kept constant for the larger
diameter, so the mass flow and total heat release are larger than what is expected for the realistic
2.25 mm tank orifice diameter. While the velocity could be decreased in order to compensate for the
larger release diameter, the flame impingement would be underestimated [113].
These boundary conditions served as input to Aria to calculate the temperature profiles across the
structural members. Specifically, the reference temperature, heat transfer coefficient, and the
irradiation from the CFD model were used as boundary conditions on the surface in direct contact
with the heated gases. The temperature profiles on the structural members were input into Adagio
to calculate the deflection due to thermal strain on each structural member. A simplified analysis was
also performed to determine if the stainless-steel hangers can hold the concrete panels when the
hydrogen jet is impinging the stainless-steel bar surface. Note that the different tunnel structures
(Central Artery North Area or CANA tunnel and Ted Williams Tunnel) were each evaluated with
and without ventilation [113].
Table 40 shows a summary of the maximum temperature and deflection for the CANA and Ted
Williams (TW) structures. The worst-case scenarios were seen when the ventilation is not operating.
Both the CANA and Ted Williams Tunnel results show that the thermal conditions may result in
localized concrete spalling in the area where the hydrogen jet flame impinges the ceiling. If the
ventilation is operating, the maximum temperature is significantly lower, and spalling is not expected
to occur. The total stress on the steel structure was significantly lower than the yield stress of
stainless steel and ASTM A36 at the maximum steel temperature even when the ventilation was not
on. Therefore, the steel structure is not expected to be compromised. Also, the epoxy remains at
ambient temperature and so should not degrade or fail due to this exposure. The maximum
deflection of the steel hanger is 7 mm, which will not impact the structural integrity of the beam.
Note that several conservative assumptions were made in this modeling, so the temperature
observed should be lower than what which results in spalling [113]. Table 40 shows the results of the
modeling. Each jet flame fire curve is created with ventilation (V) and without ventilation (NV). The
hydrocarbon and ISO 834 curves are discussed in Section 2.5.1 in more detail.
132
Table 40: Results Summary of Hydrogen FCEV in Tunnels Risk Modeling (from [113])
Maximum Maximum Yield

Temperature Deflection Stress
Fire Curve (°C) (mm) (MPa)
Hydrocarbon ~750 ~5 -
ISO 834 ~750 ~10 -
H2 Jet Flame CANA (NV) 592 19.4 -
H2 Jet Flame CANA (V) 336 7.6 -
Concrete 1,088 <200 -
Stainless Steel 706 ~7 147.79
H2 Jet Flame
TW (NV) ASTM A36 - - 399.9
Concrete 805 43.5 -
Stainless Steel 436 1.3 214.76
H2 Jet Flame
TW (V) ASTM A36 - - 172.37
6.5.2.2. Hydrogen Vehicle Explosion Risk in Tunnels

CFD modeling was performed by Middha and Hansen [127] in support of the evaluation of
explosion risk of hydrogen vehicles (both cars and buses) in a tunnel (see Section 6.5.4.1 for
additional details) [127]. The objective of the modeling was to predict a quantitative explosion risk
for hydrogen vehicles in tunnels. All the scenarios described in Section 6.5.4.1 were evaluated using
the CFD code FLACS. Both NGV and H2 vehicles were studied in the simulations but this section
will only comment on results pertaining to H2. The H2 car and bus parameters are described by
Middha and Hansen as follows:
1. Compressed hydrogen gas city bus with 40 kg H2 stored in 8 cylinders (two sets of 4 each) –
5 kg per cylinder at a storage pressure of 350 bar. The vehicle was represented as a
rectangular block (12.0 m x 2.55 m x 2.9 m) with the distance to the top of the tanks being
3.1 m.
2. Compressed hydrogen gas car with 5 kg H2 stored in 1 cylinder at a storage pressure of 700
bar. The car was represented as a simple rectangular block (5.0 m x 1.9 m x 1.5 m) located
0.3 m above the ground.
As for the tunnels, two different cross sections were evaluated, rectangle and horseshoe shape, see
Figure 73 for cross-sectional dimensions. Both tunnels were modeled with a length of 500 m.
133
Figure 73: Tunnel cross-sectional dimensions (from [127])
In addition to the cross-sectional dimensions and length, the geometry of the modeled tunnel
included vehicles. The tunnels were dual lane with traffic running a single direction. The tunnel was
assumed to be full of cars and buses spaced out evenly by 1.5 m. The vehicle distribution was a
repeated pattern of six cars follow by one bus. The vehicles were placed such that one bus and one
car were at the exact center length wise of the tunnel in separate lanes. The same geometry was used
for both the car and bus release. The releases were assumed to be choked flow. The mass flow rate
of the H2 for choked flow at 350 and 700 bar is displayed in Figure 74. Note that this was computed
assuming a discharge coefficient of 0.8 and a 4 mm opening.
Figure 74: Mass flow of release for CNG and various H2 simulations (from [127])
Ignition points were varied from the center of the vapor cloud to the outer edges (length wise of the
tunnel). Ventilation velocities were also varied between the models.
134
A dispersion model simulating the release of the fuel systems was carried out. Table 41 lists the
maximum flammable gas cloud size for each configuration as well as the equivalent stoichiometric
cloud or the Q9 quiescent cloud. This is a scaled smaller stoichiometric gas cloud that represents the
same explosion load as the non-homogenous larger cloud. It is scaled based off the weighted
volume expansion, flammable volume, and laminar burning velocity. The flammable cloud and its
stoichiometric and Q9 equivalents along with the maximum pressures for the combustion of the
flammable gas clouds are listed in Table 41 below:
Table 41: Summary of gas cloud & overpressure for various vehicles in both tunnels (from [127])
Maximum equivalent Max. pressure for max. equiv.
Maximum flammable gas
stoichiometric flammable cloud Q9 Quiescent/Pre-
cloud size in m3 (kg)
Vehicle/Release Inventory gas cloud size in m3 (kg) ignition turb.
Characteristics (kg) Maximum Q9 Maximum
Horseshoe Rectangular Horseshoe Rectangular
Equivalent overpressure
Tunnel Tunnel Tunnel Tunnel
Volume (m3) (barg)
Car LH2 10 1.4 (0.007) 1.8 (0.009) 0.02 (0.003) 0.02 (0.004) 0.0 <0.05/0.1
Car H2 Gas 700
bar (vent up) 5 281 (1.14) 273 (1.21) 4.42 (0.07) 4.31 (0.09) 4.4 0.05/0.10
Car H2 Gas 700
bar (vent down) 5 268 (1.33) 308 (1.39) 17.75 (0.29) 8.77 (0.18) 17.8 0.11/0.34
Bus H2 Gas 350
bar 5 213 (0.89) 190 (0.81) 2.16 (0.04) 1.94 (0.04) 2.2 0.05/0.10
Bus H2 Gas 350
bar 20 1795 (7.46) 3037 (13.97) 27.46 (0.45) 24.67 (0.49) 27.5 0.11/0.34
From the coupled dispersion combustion simulations, it is predicted that overpressure values can
produce minor damage to people and property within the tunnel. The data presented in Table 41
were combined to create a frequency of exceedance curve for overpressures during combustion of
gaseous hydrogen clouds, shown in Figure 75. Overpressure values outside the hazardous range (less
than 0.1 barg) are much more likely than higher overpressures that can be hazardous to tunnel
occupants. This is important information to perform a risk analysis, but the method used to create
the exceedance curves is not discussed in detail.
Figure 75: Exceedance curves for overpressure values per fuel type (from [127])
135
6.5.2.3. CFD Modeling of Hydrogen Deflagration in a Tunnel
Deflagration in homogenous, near stoichiometric hydrogen/air mixtures in a model of a tunnel were
simulated through CFD modeling techniques by Tolias et al. [177]. The ADREA-HF CFD code was
used for this modeling. The purpose of this modeling was to baseline the results from the ADREA-
HF CFD code with that of the experiment discussed in Section 6.5.1.2. Specifically, the time-
dependent overpressure data generated from the CFD modeling was compared directly with the
experimental data. The two cases that were examined were the empty tunnel and tunnel with
simulated traffic with a homogeneous hydrogen/air mixture with a 30% hydrogen concentration by
volume. While this experiment used this concentration for each scenario, more plausible
concentration would need to be used in future work. Figure 77 shows the experimental and
computational overpressure results at different locations along the tunnel. As shown, the
computational results are in general agreement with the results of the experiment. Therefore, the
CFD code was able to simulate the combustion process and estimate the resulting overpressures.
136
Figure 76: Overpressure Results without Vehicles (from [177])
137
Figure 77: Overpressure Results with Vehicles (from [177])
138
6.5.2.4. Releases from Hydrogen Fuel-cell Vehicles in Tunnels
Houf et al. [173] modeled the consequence of hydrogen TPRDs being activated, the flammable gas
venting to the environment and the time-delay to ignition within a tunnel. Multiple simulation tools
were used to perform the evaluation. To model the TPRD releases inside ventilated tunnels, Sandia’s
computational fluid mechanics code, Fuego, was used. An FCEV was modeled with three separate
tanks, each containing 1.67 kg of hydrogen at 70 MPa. For these simulations, high-pressure
hydrogen gas was vented simultaneously from three separate onboard tanks through three separate
TPRD vents located on the bottom of the FCEV. Figure 78 shows a diagram of the tunnel model
layout with transverse ventilation. The evolution of the hydrogen/air mixture was modeled after
blowdown from the TPRDs on the FCEV. A Sandia developed code, NETFLOW, was used to
model the transient nature of the tank blowdown [173].
Figure 78: Tunnel Model with Transverse Ventilation (from [173])
Figure 79 shows the simulation results of the hydrogen release and mixing in the tunnel model. Note
that the solid lines are the total flammable volume in both the tunnel and ventilation plenum, and
the dashed lines represent the flammable volume in the ventilation plenum only. As shown, a range
of ventilation rates were evaluated, and the flammable volume decreases with increasing ventilation
rate. Also, the flammable volume disperses quicker with a higher ventilation rate [173].
139
Figure 79: Simulation Results Showing Evolution of Flammable Hydrogen Volume (from [173])
A FLACS model was developed to perform ignition overpressure simulations for the simulations
evaluated in Fuego. Figure 80 shows the results from the simulation modeling at different ignition
times and locations of the ignition source after the beginning of the TPRD release. As shown,
overpressure peaks at an ignition delay of around 5 seconds [173]. Referring to Table 1, ignition
delays of about 4 to 8 seconds result in overpressures approaching or above the fatality threshold
level. These results show the importance and sensitivity to various ignition locations and delays.
Figure 80: Simulation of Peak Ignition Overpressures vs. Ignition Delay (from [173])
140
6.5.2.5. Hydrogen Release and Combustion in Subsea Tunnels
Turbulence modeling was used to evaluate hydrogen release events from vehicles in subsea tunnels
by Bie and Hao [178]. As part of this study, variable tunnel ventilation conditions and the resulting
hydrogen cloud sizes, as well as delay in ignition time, were assessed to fully characterize this risk.
The partially averaged Navier-Stokes turbulence model was used to research the hydrogen release
and combustion phenomena as it related to the risk inside highway tunnels. The physical tunnel used
as the basis of this modeling effort was the Bay subsea tunnel, a three-lane highway. The model of
the tunnel was 13.5 meters wide, 5 meters high, and 500 meters long. A typical mid-sized hydrogen
FCEV was modeled containing 4.955 kg H2 at 70 MPa. Varying ventilation conditions of 0 m/s (no
ventilation), 1 m/s, 3 m/s, and 6 m/s were evaluated with five monitoring points spaced at 5 meter
horizontal intervals (see Figure 81) [178].
Figure 81: Subsea Tunnel Model with FCEV and Monitoring Points (from [178])
Figure 82: Subsea Tunnel Model Cross Section (from [178])
Figure 83 and Figure 84 show the longitudinal and traverse hydrogen concentration contours at
three seconds after event initiation for different ventilation conditions. As shown, the ventilation
141
rate has a significant influence on the hydrogen distribution after the TPRD release event. The
upstream monitoring points showed less hazardous concentrations of hydrogen than the
downstream monitoring points [178].
Figure 83: Longitudinal H2 Distribution Various Ventilation Conditions (from [178])
142
Figure 84: Traverse H2 Distribution Various Ventilation Conditions (from [178])
143
Figure 85 and Figure 86 show the overpressure history for different ventilation conditions at ignition
times of 3.1 s and 6.1 s, respectively. There are four monitoring points (P1, P2, P3, and P4) that
measure the overpressure. P1 was arranged 5 m away from the leakage location along the direction
of traffic, P2 was arranged 10 m away from the leakage location along the direction of traffic, P3 was
arranged 5 m away from the leakage location in the inverse direction of vehicle, and P4 was arranged
10 m away from the leakage location in the inverse direction of vehicle per Bie and Hao [178]. The
literature specifies that P1 and P2 are downstream, while P3 and P4 are upstream [178].
Figure 85: Overpressure History at Ignition Time of 3.1 Seconds (from [178])
Figure 86: Overpressure History at Ignition Time of 6.1 Seconds (from [178])
The overpressure shown in Figure 85 and Figure 86 shows the peak overpressures for the upstream
locations (P3 & P4) are appreciably reduced with ventilation from 10-12 kPa at 3.1 seconds down to
7.5-8 kPa and 12 kPa down to 9 kPa at 6.1 seconds. Only P2 downstream shows a small reduction in
the overpressure measured comparing with and without ventilation.
144
6.5.2.6. Hydrogen Jet Flame Hazard in Tunnels
An evaluation of the possible fire scenarios of hydrogen cars in tunnels was conducted to assess the
implications on a tunnel ventilation system. To accomplish this, CFD simulations were evaluated on
a tunnel with a length of 102 m and a cross-section of 5 m by 5 m. Fluent was used to simulate the
smoke flow in the tunnels after a fire. Figure 87 shows the tunnel geometry and boundary conditions
of the CFD case. As shown, the hydrogen FCEV is located 40 m from the air inlet [179].
Figure 87: Tunnel geometry and Boundary Conditions of CFD Model (from [179])
Two scenarios were evaluated: a) 6 MW hydrogen fire with 2.5 m/s ventilation, and b) 30 MW
hydrogen fire with 2.5 m/s ventilation. This study selected these two specific scenarios based on
realistic hydrogen release conditions from a hydrogen car. Hydrogen was released at a rate of 0.05
kg/s and at a velocity of 10 m/s which resulted in a 6 MW hydrogen fire lasting about 1 minute in
the first scenario. In the second scenario, hydrogen was released at a rate of 0.25 kg/s and a velocity
of 50 m/s, which resulted in a 30 MW fire for a shorter duration. The results of the CFD evaluation
show that the ventilation in the 6 MW fire can fully eliminate the backlayering of smoke. However,
this is not true of the 30 MW fire. Moreover, the 30 MW fire resulted in the flame reaching the
tunnel ceiling and spreading under the ceiling for large distances. This could result in serious damage
to the tunnel equipment and structures along the ceiling [179].
6.5.2.7. Diffusion of Leaked Hydrogen in Tunnels

A series of CFD simulations were performed to evaluate diffusion of leaked hydrogen in tunnels
[180]. Multiple tunnels with variations in slope, leak location, cross-section geometry, ventilation
rate, and ventilation type were evaluated for a 60 m3 (unmixed, approximately 5 kg) hydrogen leak.
In the vehicle tunnel simulations (Case A), long model tunnel and an underwater model tunnel were
evaluated. Figure 88 illustrates the differences between these tunnels. Each tunnel was evaluated
with varying ventilation flow rates [180].
Figure 88: Case A Simulation Tunnel Geometries (from [180])
145
The general flow modeling software code STAR-CD was used in the calculation model. It was
found that in tunnels without ventilation, the geometry effects hydrogen diffusion. The slope of the
long tunnel model resulted in hydrogen collecting in the tunnel for several dozen minutes. In tunnels
with the underwater model tunnel slope, hydrogen is rapidly cleared from the tunnel. For each Case
A tunnel geometry with ventilation, the hydrogen is removed by the ventilation flow within several
dozen seconds. In the Case B simulations, there is a brief time in which hydrogen with a
concentration at about LFL flows into the power collector. For the Case C simulations, there is no
concern about inflow of hydrogen at concentrations greater than LFL since that time is very short
[180].
Figure 89 illustrates the tunnel evaluated in the Case B simulations. In these simulations, an
electrostatic dust collector is in a branch off the main tunnel. The location of the leaked hydrogen is
varied [180].
Figure 89: Case B Simulation Tunnel Geometry (from [180])
Figure 90 illustrates the tunnel evaluated in the Case C simulations. In these simulations, an
underground ventilation facility is in a branch off the main tunnel and air is released through a
vertical shaft opening to the outside. The location of the leaked hydrogen is varied [180].
Figure 90: Case C Simulation Tunnel Geometry (from [180])
146
6.5.2.8. Gaseous release, dispersion, and combustion for automotive scenarios
Venetsanos et al. [126] CFD was used to study the effects of a compressed gas release from a
commercial vehicle in urban areas. One urban area simulated was a tunnel with a single deck city bus
located centrally along the length. Variable releases from both hydrogen and CNG fuel tanks were
evaluated. The fuel storage systems modeled represented that of a typical European bus with fuel
containers located along the roof, forward from the midpoint. The system consisted of 2 sets of 4
tanks connected as displayed in Figure 91. The specific CFD solvers utilized were ADREA-HF for
dispersion and REACFLOW for combustion.
Figure 91: Fuel tanks configuration for both CNG and CH2 gas (from [126])
The tanks contained a total of 40 kg at 20, 35, and 70 MPa of H2. This is representative of a standard
CGH2 bus and the 70 MPa case exceeds normal bus configurations. The main vent lines are
controlled by thermally activated pressure relief devices (TPRD). As shown, each TPRD is attached
to manifold connected to 4 tanks (2 TPRDs per set of 8 tanks). Multiple release scenarios were
evaluated by varying TPRD orifice size and tank evacuations. A summation of storage parameters is
shown in Table 42. Note this study and the figures below have results for hydrogen and methane.
147
The computational domain was modeled as a tunnel of 212 m length with a cross-sectional area
displayed in Figure 92. As mentioned, the bus was located along half the length of the tunnel in a
centralized location. Besides the bus, the tunnel was assumed to be empty and the walls were
modeled as smooth surfaces. Additionally, the air was assumed quiescent to represent a worst case
scenario.
Two release cases were evaluated for CNG. Table 43 lists the descriptions of all cases, Case 1, and 3
were selected since Case 2 lies between them.
The results for Case 1, where only one cylinder was released through a single TPRD, are shown in
Figure 93. The left frame shows the flammable mass and the right frame the total available energy.
148
The available energy was computed by multiplying the released mass of fuel by the lower heat of
combustion. The flammable mass was calculated from the amount of fuel/air mixture released
which was within the FL. Figure 94 displays the results for case 3, where all 8 cylinders released
through all 4 TPRDs simultaneously.
Figure 93: Flammable mass and available energy of released gas in Case 1 (from [126])
Figure 94: Flammable mass and available energy of released gases in Case 3 (from [126])
Note the change in scales between Figure 93 and Figure 94 For Case 1 both the flammable mass and
available energy maintained lower values and dissipated rapidly in time. For Case 3 the flammable
mass and available energy reached dangerous levels which persisted over the length of the
simulation. It was assumed the Case 1 overpressures would be negligible because the total flammable
mass was less than 0.5 kg. The overpressure values are displayed in Figure 96. The overpressure was
149
calculated assuming that the cloud was ignited after 40 seconds, corresponding to maximum
flammable mass. The ignition point was assumed to be at the center of tunnel at the top of the bus.
150
When reviewing these overpressure values, it is important to keep in mind what pressure values
correlate to what level of property damage or human hazard. The tables listed in Section 1.2 help
understand the damage and tenability thresholds. Case 3 scenario yields overpressure values capable
of rupturing eardrums and creating harmful glass splinters up to a distance 80-100 m from the
ignition point. Note that the range for the eardrum rupture threshold reported in Table 1 is 16.5-
19.3 kPa. For the 35 MPa scenario, the overpressure goes into the threshold for fatality per Table 1.
Venetsanos et al. [126] states that the blast wave maintains its strength for long distances inside of
tunnels due to the high levels of confinement compared with urban environments where blast waves
decay quicker. In addition to overpressure, Venetsanos et al. [126] reported the fireball length along
the tunnel, the results for each scenario are displayed in Table 44.
Table 44: Combustion results within tunnel (from [126])
For the Case 3 model, the hydrogen combustion produced a flame length which traveled farther
than the length of the tunnel, 220 m for the 20 MPa and 285 m for the 35 MPa case of the total 212
m. The flame length for Case 1 hydrogen combustion was reported as 58 m for the 20 MPa and 47
m for the 35 MPa. It is worth noting that in this study, Case 3 represents an implausible scenario of
rapid and complete fuel release, ignition when the peak flammable mass is present, and static air
within the tunnel. This study further identifies the importance ventilation plays on mitigating risk
during an accidental release of fuel in a tunnel.
6.5.3. Analysis
Multiple analyses have been performed to evaluate the risk associated with hydrogen vehicles in
tunnels. One analysis was performed to characterize the most likely consequence of an accident by
developing an event sequence diagram and further characterizing severe consequence scenarios. It
was shown that the most likely consequence is no additional hazard from the hydrogen (see Section
6.5.4). Another assessment was performed to evaluate the consequence of a delayed ignition of
hydrogen released from several different types of vehicles (both cars and buses with compressed
hydrogen, as well as a car with LH2 fuel). It was shown that the maximum pressure loads resulting
from ignition from a hydrogen cloud would be insignificant (see Section 6.5.4.1). Another analysis
evaluated the possible incidents and consequences of hazardous events in a tunnel for several
different alternative fuel vehicles. This analysis showed that although the HRR is higher for
hydrogen when compared to other fuels, the overpressure is relatively low (see Section 6.5.4.1).
151
6.5.4. Hydrogen FCEV Tunnel Risk Analysis
A risk analysis was performed to estimate what scenarios were most likely to occur in the event of a
hydrogen FCEV accident in a tunnel by Ehrhart et al. [113]. An event sequence diagram for a
hydrogen vehicle accident was developed for a hydrogen FCEV accident in a tunnel, including all
outcomes along with associated values and probabilities. Figure 97 shows the event sequence
diagram developed in the risk analysis.
Figure 97: Event Sequence Diagram for a Hydrogen FCEV in a Tunnel (from [113])
Each event was evaluated to determine whether the respective scenario warranted further
characterization with heat transfer and CFD models. Based on an evaluation of the risk of each
scenario (both the likelihood and consequence), the scenario evaluated further with modeling was a
hydrogen vehicle in an accident exposed to a resulting fire. A typical hydrogen FCEV was
considered in this analysis, with a 125 L, 70 MPa tank of hydrogen with a typical TPRD orifice of
2.25 mm. Note that due to computational limitations, the smallest reasonable tank orifice diameter
that can be modeled is 5.25 cm. This is conservative because the velocity was kept the same for the
larger diameter, and so the mass flow and total heat release are larger than what is expected for the
realistic 2.25 mm tank orifice diameter. Taking this into account, the worst-case scenario is based on
a 5.25 cm release diameter with a constant velocity of 700 m/s. Conservative assumptions were
made in terms of the hydrogen fuel released from the TPRD, including having the vehicle flipping
over in the crash to orient the jet flame toward the ceiling of the tunnel. Three Boston tunnels with
different structural configurations were investigated: the CANA Tunnel, the Ted Williams Tunnel,
and the Sumner Tunnel [113]. See Section 6.5.2.1 for information on the CFD, heat transfer, and
solid mechanics modeling used for this risk analysis.
152
The results show that the most likely consequence is no additional hazard from the hydrogen,
although some factors need additional data and study to validate. This includes minor crashes and
scenarios with no release or ignition. When the hydrogen does ignite, it is most likely a jet flame
from the pressure relief device release due to a primary hydrocarbon fire. This scenario was
considered in detailed modeling of specific tunnel configurations. Localized concrete spalling may
result where the jet flame impinges the ceiling, but this is not expected to occur with ventilation.
Structural epoxy remains well below the degradation temperature. The total stress on the steel
structure will not be compromised. It is important to note that this study took a conservative
approach in several factors, so observed temperatures should be lower than predicted by the models
[113].
6.5.4.1. Hydrogen Vehicle Explosion Risk in Tunnels

An assessment was performed to evaluate the risk of explosion for hydrogen vehicles in tunnels. For
all accident scenarios, the hydrogen release is attributed to the activation of the pressure relief
device. Two different hydrogen vehicles were evaluated: 1) a city bus with 40 kg H2 at a storage
pressure of 350 bar, and 2) a car with 5 kg H2 at a storage pressure of 700 bar. Additionally, two
different tunnel layouts (horseshoe and rectangular) and several longitudinal ventilation conditions
were considered. The following hydrogen release scenarios were evaluated [127]:
1. Hydrogen Passenger Vehicle (vent up) releasing 5 kg of H2 for 84s
2. Hydrogen Passenger Vehicle (vent down) releasing 5 kg of H2 for 84s
3. Hydrogen Bus releasing 5 kg of H2 for 147s
4. Hydrogen Bus releasing 20 kg of H2 for 147s
5. Hydrogen Passenger Vehicle releasing 10 kg of LH2 for 900s
The ignition probabilities and intensities were developed from information relevant to the oil and
gas industry. Initially, stoichiometric gas clouds of different sizes are considered to explode to
calculate the maximum overpressure near the tunnel ceiling [127]. As a refinement, dispersion
modeling was performed to determine the gas cloud size and hydrogen concentration that can be
realistically expected. CFD modeling was used to evaluate both the dispersion and explosion
simulations for each of the scenarios described previously (see Section 6.5.2.2).
The worst-case deterministic evaluation of each of the scenarios involved the tunnel filling with
stochiometric hydrogen gas clouds of varying size. This showed unacceptable results in terms of
very high overpressures. However, a dispersion study was performed to determine a more realistic
gas cloud from hydrogen release and their subsequent ignition. This more realistic evaluation
showed that the worst-case overpressures were reduced by two orders of magnitude. Moreover, a
probabilistic study was performed that reduced the expected risk of an explosion due to a hydrogen
vehicle even more. The maximum pressure loads (between 0.1 barg [threshold for skin laceration
from flying glass] and 0.3 barg [serious wounds from flying glass near 50% probability]) predicted by
the simulations could be significant [127].
6.5.4.1. Fire and Explosion Hazards in Tunnels of Alternative Fuel Vehicles

An analysis of the possible incidents and consequences of hazardous events in a tunnel was
evaluated for several different alternative fuel vehicles, including gaseous and liquid hydrogen
153
vehicles by Li [134]. The likelihood of the events was not evaluated, but event trees were defined for
both liquefied fuel vehicles and compressed gas vehicles (see Figure 98 and Figure 99).
Figure 98: Liquefied Fuel Vehicle Event Tree for Incidents in Tunnels (from [134])
154
Figure 99: Compressed Gas Vehicle Event Tree for Incidents in Tunnels (from [134])
Each event was evaluated through simple modeling to determine the potential consequence. An
analysis of spilled fuel fires for liquid hydrogen vehicles showed that the heat release rate per unit
fuel area of liquid hydrogen is around 60 times higher than ethanol and methanol. For jet fires, the
analysis showed that the heat release rates for hydrogen vehicles were significantly higher than those
of compressed natural gas tanks, while the flame length was only slightly greater. Figure 101 shows
the peak overpressure as a function of distance resulting from rupture of the pressure vessel. As
shown, the overpressure decreases rapidly within the first 50 m [134].
155
Figure 100: Overpressure vs. Distance for Liquid H2 tank at 350 bar (from [127])
Figure 101: Overpressure vs. Distance for Gaseous H2 tank at 350 bar (from [127])
Finally, the peak overpressure resulting from a gas cloud explosion was evaluated for both gaseous
and liquefied hydrogen. The overpressures for each case were relatively low when compared to the
other alternative fuels such as natural gas due to the small fuel mass for the hydrogen case which
leads to lower explosion energy [134].
156
6.6. FCEV Research Gaps
Through this literature study, it was found that there are several existing studies that evaluate the
failure modes and consequences associated with hydrogen fuel cell electric vehicles (FCEVs) in
tunnels. There have been multiple experimental studies that have investigated overpressures
resulting from delayed hydrogen ignition, HRR, hydrogen dispersion, and thermal effects of jet fires.
Modeling studies have been conducted on the consequences of release, including hydrogen
accumulation followed by ignition and the resulting overpressure. Also, risk analysis has been
conducted on the thermal effects on tunnel components from a jet fire rather and identification of
release events that could occur. Additionally, analysis has been performed to quantify probabilities
and likelihoods for these various events.
Conclusions about important variables can be derived from comparison of the different literature.
As ventilation increases, the overpressure decreases in a congested area [173] [174]. However, CFD
results of a tunnel including vehicles as blockage show that the overpressure stays about the same.
This could mean that ventilation is a stronger factor in varying the overpressure. The exceedance
curve shown in Figure 75 shows the various hydrogen leak scenarios and the frequency of each one.
In the event of a leak, the layer height and concentration were used to create a matrix to understand
how different leak scenarios might lead to various deflagrations or even detonations [175].
Although significant work has been accomplished, there are still areas that should be evaluated
further. The following criteria were evaluated to determine where research gaps may exist regarding
hydrogen FCEVs in tunnels.
2. Failure Modes
3. Consequences
4. Validation
In terms of scenario identification included in this literature survey, the scenarios that lead to a
failure mode have been identified as impacts to the vehicle or failure of the TPRD, hardware
degradation or failure, and operator error which may lead to a release of fuel. Fault trees for both
liquid and gaseous fuel release can be seen in Figure 98 and Figure 99. Failure modes were addressed
through several studies evaluating the mechanism and consequences associated with hydrogen
FCEVs in tunnels. The failure modes with potentially hazardous consequences identified in the
scenario identification effort included a release with either immediate or delayed ignition.
The measurements of the consequences of the failure mode include overpressure, HRR, hydrogen
dispersion, and resulting structural damage determine the extent of the hazard. There are several
variables that effect the magnitude of the consequence: hydrogen quantity released, ventilation,
obstructions, ignition time, tunnel geometry, etc. Results of the consequences include overpressures
that range from 34 kPa [127] to over 100 kPa [126], HRR that can peak near 16 MW [176], and
fireballs that can exceed 250 m long [126]. Also, validation of the results has been achieved through
comparison studies between the modeling and experiments with regard to various consequences like
HRR [176].
The research of hydrogen FCEVs in tunnels has evaluated, in some manner, a significant
combination of failure modes, consequences, and influencing variables. Despite this, the following
research gaps were identified:
157
 Temperature and thermal effects to structures. A diagnostic to the consequence of a failure
mode is temperature or thermal effects. Although this has been addressed in a single
modeling/analysis report, additional research should be conducted on this topic.
 Ventilation effects. The study of spontaneous ignition was conducted at ambient conditions
outdoors. The effect that ventilation in a tunnel has on the results could be evaluated.
 Hydrogen-specific fires. The power associated with the fires in Section 6.5.2.6 should be
related to specific hydrogen vehicle types (e.g., cars, buses, etc.).
 The effect of deflagration/detonation on structural components of a tunnel for each of the
different hydrogen vehicle classes.
 The effect of overpressure effects on life safety to people within the tunnel.
 The extent to which hydrogen can accumulate due to partial confinement and restriction,
rather than complete confinement.
 Additional attention should be given to the size or class of the vehicle. As vehicular class
increases so does the amount of stored fuel. Several different classes of vehicles were
evaluated in the studies, including hydrogen cars and buses, liquid hydrogen cars, and
multiple hydrogen cars on a cargo truck.
The International Conference on Hydrogen Safety (ICHS) in 2019 showcased a variety of topics for
hydrogen safety [181]. As the papers are published from this event, some of the identified research
gaps may be addressed. Some of these gaps as well as others will be addressed in the output of the
current HyTunnel-CS project [182]. The intent of this project is to perform research regarding
hydrogen powered vehicle safety in tunnels and confined spaces. The goal is for hydrogen vehicles
entering underground environments to maintain comparable risk as fossil fuels. Experiments in
tunnels, modeling using tools such as CFD, and analysis using risk assessment methodologies will be
covered in this project.
The HyTunnel-CS project has identified the following work packages (WP) to address research gaps:
WP1 – The state-of-the-art in safety provisions for underground transportation systems and
accident scenarios prioritization
WP1 will review the state-of-the-art in safety provisions for underground transportation systems and
the accident scenarios prioritization will aid at identifying the knowledge gaps in both safety science
and regulations, codes, and standards to be addressed
WP2 – Effect of mitigation systems on hydrogen release and dispersion in confined spaces
An intensive experimental program empowered by theoretical and numerical studies will be performed
under this work package. The work addresses the knowledge gaps highlighted in WP1 and the
development of novel engineering solutions for the prevention and mitigation of accident involving
hydrogen releases.
WP3 – Thermal and pressure effects of hydrogen jet fires and structure integrity
Under this work package jet fires will be investigated through a comprehensive set of experimental,
theoretical, and numerical studies to improve the principal understanding of hydrogen jet fire on life
safety provisions in underground transportation systems and their structural integrity.
158
WP4 – Explosion prevention and mitigation
WP4 investigates explosion prevention and mitigation through numerous experimental tests realized
in tunnels and other confined spaces and theoretical and numerical studies on accident scenarios
involving hydrogen tanks. The aim of the WP is to provide engineering tools to evaluate the associated
hazards, as well as innovative preventive and mitigation solutions and to improve the principal
understanding of hydrogen explosion hazards in tunnels and similar confined spaces using
complementarities of theoretical, numerical and experimental studies.
WP5 – First responders’ intervention strategies and tactics for hydrogen accidents in
underground transportation systems and risk assessment
Under WP5, the research findings from WPs 2-4 will be translated into suitable information,
guidelines, and recommendations for first responders intervening in an accident involving hydrogen-
powered vehicles in tunnels or other confined spaces. This includes examining and supplementing
available knowledge in such a way that it can be taught to all first responders and can also be practically
applied by them.
WP6 through WP8 –Outreach/Dissemination, Management, and Ethics are not summarized since
these are not technical research gaps.
159
160
7. CONCLUSIONS
In this report, as a result of requests from multiple stakeholders across government and industry,
including code officials, Sandia National Laboratories has completed a preliminary assessment of
safety research and incidents for the use of alt fueled vehicles in tunnels. While a lot of research for
traditional fuels has been completed through studies, modeling, and experiments, alternative fuels
have limited tunnel research that has been completed. Various experiments have been completed to
understand release of hydrogen, natural gas, and propane in both the gaseous and liquid form into
confined space which can be applicable to alternative fuels in tunnels. These studies and experiments
can be used to understand the hazard characteristics of alternative fuels in tunnels.
7.1. Battery Electric Vehicles

Overall, BEVs have a variety of research gaps listed due to the complexity of scaling battery systems
to power the various classes of vehicles. Additionally, the wide variety of cell chemistries, battery
pack designs, and battery management systems makes it difficult to assess specific safety metrics for
BEVs as a whole. One large-scale test [86] showed that BEVs have comparable HRR and THR to
ICE vehicles, but with a higher hydrogen fluoride production during a fire. Specific modeling for
BEVs in tunnels has not been conducted, and modeling for lithium-ion batteries is still at the cell
and module level. A tunnel experiment conducted looked at various failure scenarios and the effect
inside a tunnel downstream of the failed battery cells [86]. The release quantities were measured
which included toxic aerosols such as cobalt, lithium, and manganese. There is some analysis
conducted to further understand the consequences. Data from mechanically and thermally failed
modules was analyzed to understand the fire effluents and required ventilation in the space [91]. A
systems safety approach from the cell level, to modules, battery system, and complete BEV should
be studied to further understand how the components and sub-components interact. Using a system
safety V-diagram to define testing and safety requirements for each system level (cell, module, and
battery system) would help further understand how to safely design these systems. This will directly
tie into understanding consequence metrics for further evaluations in tunnels. Future work
characterizing fire spread within a BEV and tactics to slow or stop thermal propagation would be
beneficial.
7.2. Natural Gas Vehicles

Multiple studies for both CNG and LNG powered vehicles have been conducted, including
experiments to understand flame speed and overpressure and the effects of congestion on these
hazard metrics. Vapor cloud explosions and heat flux from flames have also been studied. Modeling
to understand CNG dispersion has been conducted for various tunnels. An FMEA of CNG-
powered buses has also been completed to understand the risk of these vehicles. However, the
majority of these studies involve CNG only, showing the need to further understand LNG hazards
and the differences for this liquefied fuel. Continuous release of LNG compared to CNG showed
that the harmful and lethal distances for CNG exceed those for LNG due to the higher storage
pressures. Further studies to understand and compare the hazard difference between LNG and
CNG should be considered. A variety of tunnel studies have used CNG release quantities that are
considered equivalent to the amount of CNG used in a city bus and passenger car. Other classes of
vehicles should be further studied for release characteristics for both CNG and LNG.
161
7.3. Propane Vehicles
There are relatively few studies evaluating the failure modes and consequences associated with LPG
vehicles in tunnels. A scenario identification study was conducted that estimated the likelihood of
different consequences of a tunnel accident through expert elicitation, but a more rigorous
evaluation of failure modes should be completed. One study focuses on creating and using a failure
tree to inform release scenarios for an LPG vehicle. This then helped inform the experimental setup
where six different tests using different ventilation and releases were used. The experimental data
was then compared with a CFD model for validation. This model can be further used to understand
the gaseous dispersion characteristics of propane from vehicles failing in tunnels and other confined
spaces. Additionally, experiments have been conducted to determine the effect of ventilation and
tunnel slope on smoke dispersion in a tunnel using a propane fire. Modeling helps understand the
dispersion and evaporation phenomena of an LPG spill.
7.4. Hydrogen Fuel Cell Vehicles

A variety of studies and experiments have been completed for hydrogen fuel cell electric vehicles,
specifically in tunnel applications. Just as for CNG, batteries, and other fuels, industry plans to use
hydrogen fuel cells for larger class vehicles (e.g. class 8 trucks). Therefore, studies to understand how
the increase of vehicle class affects the hazard should be considered. Consequence models for tunnel
safety studies have been conducted using CFD models and should be further used to evaluate larger
classes of vehicles. Future work and studies to improve characterization of the risks include the
HyTunnel-CS project [182]. This project will use experiments in tunnels, modeling using tools such
as CFD, and analysis using risk assessment methodologies. Through this work, over-conservatism
will be reduced which will help increase effectiveness of safety systems along with cost savings of
tunnel and confined space safety systems.
7.5. Closing Remarks and Future Work

In this report, Sandia National Laboratories has compiled the first comprehensive overview of key
studies and experiments to date on the safety of alternative fuel vehicles specifically within tunnels.
While there have been various studies at different levels of rigor and complexity, and several real-
world incidents with alternative fuel vehicles in tunnels, it is clear that hazards can never be
completely prevented regardless of the type of vehicle or fuel being used. Different classes of
vehicles and the different hazards represented by each fuel need to be considered during
development of regulations. The phase of the fuel (solid, liquid, or gas) plays a role in the hazards
associated with each fuel type and should be considered when determining tunnel design
specifications required.
These studies help to develop relevant information on how best to construct, site, and maintain
tunnels and other enclosed spaces and determine key specifications such as ventilation requirements,
etc. to enable the safe use of emerging technologies. Analysis of these studies will also enable
research gaps to be identified and research prioritized to close the gaps. This overview will allow
stakeholders, AHJs, and tunnel owners to actively participate in the discussions on further studies
and tunnel regulations. As more studies and experiments are conducted, they will help to provide a
complete analysis of the hazards and recommendations for the use of alternative fuel vehicles in
tunnels and other confined spaces such as parking garages and locomotive tunnels.
162
8. REFERENCES
[1] A. M. Glover, A. R. Baird and C. B. LaFleur, "Hydrogen Fuel Cell Vehicles in Tunnels,"
Sandia National Laboratories (SAND2020-4507 R), Albuquerque, 2020.
[2] A. K. Coker, "9 - PROCESS SAFETY AND PRESSURE-RELIEVING DEVICES," in
Ludwig's Applied Process Design for Chemical and Petrochemical Plants (Fourth Edition), Burlington,
Gulf Professional Publishing, 2007, pp. 575-770.
[3] Los alamos National Laboratory, "What’s the difference between an explosion and a
detonation?," [Online]. Available:
https://www.lanl.gov/museum/news/newsletter/2018/08/detonation.php. [Accessed 19 05
2020].
[4] California Institute of Technology, "Flammability and Explosion Limits," [Online]. Available:
https://shepherd.caltech.edu/EDL/PublicResources/flammability.html. [Accessed 05 12
2019].
[5] W. D. Manha, "Chapter 20 - Propellant Systems Safety," in Safety Design for Space Systems,
Oxford, Butterworth-Heinemann , 2009, pp. 661-694.
[6] J. G. Quintiere, in Fundamentals of Fire Phenomena, John Wiley & Sons, 2006, pp. 77-115.
[7] M. Huth and A. Heilos, "14 - Fuel flexibility in gas turbine systems: impact on burner design
and performance," in Modern Gas Turbine Systems, Cambridge, Woodhead Publishing Limited,
2013, pp. 635-684.
[8] G. Jomaas, "Fundamentals of Premixed Flames," in SFPE Handbook 5th Edition, New York
City, Springer, 2016, pp. 373-395.
[9] NOAA Office of Response and Restoration, "Overpressure Levels of Concern," NOAA
Office of Response and Restoration, [Online]. Available:
https://response.restoration.noaa.gov/oil-and-chemical-spills/chemical-
spills/resources/overpressure-levels-concern.html. [Accessed 20 09 2019].
[10] MIT Thermodynamics and Propulsion, "Thermodynamics and Propulsion," MIT
Thermodynamics and Propulsion, [Online]. Available:
https://web.mit.edu/16.unified/www/FALL/thermodynamics/notes/node111.html.
[Accessed 20 09 2019].
[11] DOE Office of Energy Efficiency and Renewable Energy's Fuel Cell Technologies Office,
"H2 Tools," DOE Office of Energy Efficiency and Renewable Energy's Fuel Cell
Technologies Office, [Online]. Available: https://h2tools.org/hyarc/calculator-tools/lower-
and-higher-heating-values-fuels. [Accessed 20 09 2019].
[12] National Institute of Standards and Technology, "Fire Dyanmics," NIST, 17 07 2018.
[Online]. Available: https://www.nist.gov/el/fire-research-division-73300/firegov-fire-
service/fire-dynamics. [Accessed 01 10 2019].
[13] National Fire Protection Association, NFPA 921: Guide for Fire and Explosion
Investigations, Quincy, 2017.
[14] EUROPEAN INDUSTRIAL GASES ASSOCIATION AISBL , "DETERMINATION OF
SAFETY DISTANCES IGC Doc 75/07/E," Brussels, 2007.
[15] DOE Energy Efficiency & Renewable Energy, "Maps and Data - Vehicle Weight Classes &
Categories," DOE Energy Efficiency & Renewable Energy, [Online]. Available:
https://afdc.energy.gov/data/10380. [Accessed 19 09 2019].
163
[16] DOE Energy Efficiency & Renewable Energy, "Alternative Fuels Data Center," [Online].
Available: https://afdc.energy.gov/data/. [Accessed 25 09 2019].
[17] DOE Energy Efficiency & Renewable Energy, "Alternative Fuels Data Center," DOE
Energy Efficiency & Renewable Energy, [Online]. Available:
https://afdc.energy.gov/fuels/ethanol_e85_specs.html. [Accessed 06 09 2019].
[18] Office of Energy Efficiency & Renewable Energy, "Ethanol," Office of Energy Efficiency &
Renewable Energy, [Online]. Available: https://www.fueleconomy.gov/feg/ethanol.shtml.
[Accessed 06 09 2019].
[19] US Energy Information Administration, "How much gasoline does the United States
consume?," US Energy Information Administration, 04 09 2019. [Online]. Available:
https://www.eia.gov/tools/faqs/faq.php?id=23&t=10. [Accessed 06 09 2019].
[20] Bureau of Transportation Statisitics, "Diesel-powered Passenger Cars and Light Trucks,"
Bureau of Transportation Statisitics, 10 2015. [Online]. Available:
https://www.bts.gov/archive/publications/bts_fact_sheets/oct_2015/entire. [Accessed 06
09 2019].
[21] V. Honig, M. Pexa and Z. Linhart, "Biobutanol Standardizing Biodiesel from Waste Animal
Fat," Jounrla of Enviromental Studies, vol. 24, no. 6, pp. 2433-2439, 2015.
[22] D. Gi and C. Lee, "Combustion and emissions behaviour for ethanol–gasoline-blended fuels
in a multipoint electronic fuel injection engine," International Journal of Sustainable Energy, 2014.
[23] M. I. Khan, T. Yasmin and A. Shakoor, "Technical overview of compressed natural gas
(CNG)," Renewable and Sustainable Energy Reviews, vol. 51, pp. 785-797, 2015.
[24] P. Martinez, E. Rus and J. Compana, "FLASH POINT DETERMINATION OF BINARY
MIXTURES OF ALCOHOLS, KETONES AND WATER," Departamento de Ingeniería
Química. Facultad de Ciencias. Universidad de Málaga, Málaga.
[25] IPCS INCHEM, "Diesel Fuel No. 2," IPCS INCHEM, 10 2004. [Online]. Available:
http://www.inchem.org/documents/icsc/icsc/eics1561.htm. [Accessed 06 09 2019].
[26] D. Alviso, F. Krauch, R. Roman, M. Hernando, R. Goncalves dos Santos, J. C. Rolon and N.
Darabiha, "Development of a diesel-biodiesel-ethanol combined chemical scheme and
analysis of reactions pathways," Fuel, vol. 191, pp. 411-426, 2017.
[27] Matheson Tri Gas, "Lower and Upper Explosive Limits for Flammable Gases and Vapors
(LEL/UEL)," Matheson Tri Gas, [Online]. Available:
https://www.mathesongas.com/pdfs/products/Lower-(LEL)-&-Upper-(UEL)-Explosive-
Limits-.pdf. [Accessed 04 12 2019].
[28] DOE Office of Energy Efficiency and Renwable Energy, "Many Factors Affect Fuel
Economy," DOE Office of Energy Efficiency and Renwable Energy, [Online]. Available:
https://www.fueleconomy.gov/feg/factors.shtml. [Accessed 20 09 2019].
[29] J. S. Bartlett, "Study Shows Top Tier Gasoline Worth the Extra Price," Consumer Reports, 03
04 2019. [Online]. Available: https://www.consumerreports.org/car-maintenance/study-
shows-top-tier-gasoline-worth-extra-price/. [Accessed 20 09 2019].
[30] US Enviromental Protection Agency, "Gasoline Reid Vapor Pressure," US Enviromental
Protection Agency, [Online]. Available: https://www.epa.gov/gasoline-standards/gasoline-
reid-vapor-pressure. [Accessed 09 09 2019].
[31] J. Gehandler, P. Karlsson and L. Vylund, "Risks associated with alternative fuels in road
tunnels and underground garages," 2017.
164
[32] J. Sjostrom, G. Appel, F. Amon and H. Persson, "Experimental results of large ethanol fuel
pool fires," Science Partner, 2015.
[33] D. Schütz, "Fire protection in tunnels: Focus on road & train tunnels," SCOR Global P&C,
2014.
[34] National Academies of Sciences, Engineering, and Medicine, "Signiticant Fire Incidents in
Road Tunnels- Literature Review," in Design Fires in Road Tunnels, Washington, The National
Academies Press, 2011, pp. 21-26.
[35] P. Johnson and D. Barber, "The Brunley Tunnel Fire- Implications for Current Design
Practice," in Third International Symposium on tunnel Safety and Security, Stockholm, 2008.
[36] BBC News, "Five killed in Austria tunnel fire," 07 08 2001. [Online]. Available:
http://news.bbc.co.uk/2/hi/europe/1476385.stm. [Accessed 10 09 2019].
[37] A. Leitner, "The fire catastrophe in the Tauern Tunnel: experience and conclusions for the
Austrian guidelines," Tunnelling and Underground Space Technology, vol. 16, pp. 217-223, 2001.
[38] National Fire Protection Association, NFPA 502: Standard for Road Tunnels, Bridges, and
Other Limited Access Highways, Quincy, 2020.
[39] ASHRAE, Handbook- HVAC Applications, Atlanta, 2019.
[40] I. Maevski, "Road Tunnel Fires," in Guidelines for Emergency Ventilation Smoke Control in Roadway
Tunnels, Washington, NATIONAL COOPERATIVE HIGHWAY RESEARCH
PROGRAM, 2017, pp. 16-22.
[41] National Cooperative Highway Research Program, "Design Fires in Road Tunnels," 2011.
[42] E. Cafaro and V. Bertola, "Fires in Tunnels: Experiments and Modelling," The Open
Thermodynamics Journal, vol. 4, pp. 156-166, 2010.
[43] Working Group No.6 Maintenance and Repair, "Structural Fire Protection For Road
Tunnels," International Tunneling Association, 2017.
[44] A. Lönnermark and H. Ingason, "Gas temperatures in heavy goods vehicle fires in tunnels,"
Fire Safety Journal, vol. 40, no. 6, pp. 506-527, 2005.
[45] Promat Tunnels, "Fire curves," Promat , [Online]. Available: https://www.promat-
tunnel.com/en/advices/fire-protection/fire%20curves. [Accessed 17 12 2019].
[46] ASTM International, West, "E119-18ce1, Standard Test Methods for Fire Tests of Building
Construction and Materials," Conshohocken, 2018.
[47] Underwriters Laboratory, "1709: Standard for Rapid Rise Fire Tests of Protection Materials
for Structural Steel," 2017.
[48] J. Brekelmans, R. ven den Bosch and K. Both, "Summary of large scale fire tests in the
Runehamar tunnel in Norway," UPTUN, TNO, PROMAT, 2003.
[49] H. Ingason, A. Lonnermark and Y. Z. Li, "Runehamar tunnel fire tests," fire Safety Journal, vol.
71, pp. 134-149, 2015.
[50] V. Babrauskas and R. D. Peacock, "Heat release rate: the single most important variable in
fire hazard," Fire safety journal, vol. 18, no. 3, pp. 255-272, 1992.
[51] T. Lemaire and Y. Kenyon, "Large Scale Fire Tests in the Second Benelux Tunnel," fire
Technology, vol. 42, pp. 329-350, 2006.
[52] NIST, "CFAST, Fire Growth and Smoke Transport Modeling," [Online]. Available:
https://www.nist.gov/el/fire-research-division-73300/product-services/consolidated-fire-
and-smoke-transport-model-cfast. [Accessed 17 12 2019].
165
[53] W. Chow, "Simulation of Tunnel Fires Using a Zone Model," Tunneling and Underground Space
Technology , vol. 11, no. 2, pp. 221-236, 1996.
[54] A. Haghighat and L. Kray, "Determination of critical parameters in the analysis of road tunnel
fires," International Journal of Mining Science and Technology, vol. 29, pp. 187-198, 2019.
[55] S. Shafee and A. Yozgatligil, "An analysis of tunnel fire characteristics under the effects of
vehicular blockage and tunnel inclination," Tunnelling and Underground Space Technology, vol. 79,
pp. 274-285, 2018.
[56] A. Andwari, A. Pesiridis, S. Rajoo, R. Martinez-Botas and V. Esfahanian, "A Review of
Battery Electric Vehicle Technology and Readiness Levels," Renewable and Sustainable Energy
Reviews, vol. 78, pp. 414-430, 2017.
[57] University of Washington Institue of Clean Energy, "What is a lithium-ion battery and how
does it work?," [Online]. Available: https://www.cei.washington.edu/education/science-of-
solar/battery-technology/. [Accessed 03 02 2020].
[58] M. Hannan, M. Hoque, A. Mohamed and A. Ayob, "Review of Energy Storage Systems for
Electric Vehicle Applications: Issues and Challenges," Renewable and Sustainable Energy Reviews,
vol. 69, pp. 771-789, 2017.
[59] X. Zhang, L. Ji, O. Toprakçı, Y. Liang and M. Alcoutlabi, "Electrospun Nanofiber-Based
Anodes, Cathodes, and Separators for Advanced Lithium-Ion Batteries," Polymer Reviews, vol.
51, no. 3, pp. 139-264, 2011.
[60] Y. Wu, S. Saxena, Y. Xing, Y. Wang, C. Li, W. K. C. Yung and M. Pecht, "Analysis of
Manufacturing-Induced Defects and Structural Deformations in Lithium-Ion Batteries Using
Computed Tomography," Energies, vol. 11, no. 4, p. 925, 2018.
[61] V. Ruiz, A. Pfrang, A. Kriston, N. Omar, P. Van den Bossche and L. Boon-Brett, "A review
of international abuse testing standards and regulations for lithium ion batteries in electric and
hybrid electric vehicles," Renewable and Sustainable Energy Reviews, vol. 81, pp. 1427-1452, 2018.
[62] SAE International, J2464: Electric and Hybrid Electric Vehicle Rechargeable Energy Storage
System (RESS) Safety and Abuse Testing, 2009.
[63] SAE International, "Electric and Hybrid Vehicle Propulsion Battery System Safety Standard -
Lithium-based Rechargeable Cells," 18 02 2011. [Online]. Available:
https://saemobilus.sae.org/content/J2929_201102/#scope. [Accessed 20 09 2019].
[64] Underwriters Laboratory, 2580: Batteries for Use In Electric Vehicles, 2013.
[65] D. Doughty and C. Crafts, "FreedomCAR Electrical Energy Storage System Abuse Test
Manual for Electric and Hybrid Electric Vehicle Applications," Lithium Battery Research and
Development Department Sandia National Laboratories (SAND2005-3123), Albuquerque,
2005.
[66] C. F. Lopez, J. A. Jeevarajan and P. P. Mulherjee, "Experimental Analysis of Thermal
Runaway and Propagation in Lithium-Ion Battery Modules," Journal of the Electrochemical Society,
vol. 162, no. 9, pp. 1905-1915, 2015.
[67] G. Berdichevsky, K. Kelty, J. Straubel and E. Toomre, "The Tesla Roadster Battery System,"
Tesla Motors, 2006.
[68] P. Balakrishnan, R. Ramesh and T. P. Kumar, "Safety mechanisms in lithium-ion batteries,"
Journal of Power Sources, vol. 155, pp. 401-414, 2006.
[69] A. R. Baird, E. J. Archibald, K. C. Marr and O. A. Ezekoye, "Explosion hazards from lithium-
ion battery vent gas," Journal of Power Sources, vol. 446, pp. 227-257, 2020.
166
[70] V. Somandepalli, K. Marr and Q. Horn, "Quantification of Combustion Hazards of Thermal
Runaway Failures in Lithium-Ion Batteries," SAE International Journal of Alternative Powertrains ,
vol. 3, no. 1, pp. 98-104, 2014.
[71] D. Yoney, "Come count kilowatt-hours with us," INSIDEEVs, 02 06 2018. [Online].
Available: https://insideevs.com/features/336680/7-electric-cars-with-the-biggest-batteries/.
[Accessed 19 09 2019].
[72] P. Vandell, "First report of Suprise APS battery explosion that hospitalized firefighters offers
few answers," azcentral: Part of the USA Today Network, 09 08 2019. [Online]. Available:
https://www.azcentral.com/story/news/local/surprise/2019/08/09/report-surprise-aps-
battery-explosion-hosptialized-hazmat-offers-few-answers/1951399001/. [Accessed 16 09
2019].
[73] A. W. Golubkov, D. Fuchs, J. Wagner, H. Wiltsche, C. Stangl, G. Fauler, G. Voitic, A. Thaler
and V. Hacker, "Thermal-runaway experiments on consumer Li-ion batteries with metal-
oxide and olivin-type cathodes," Royal Society of Chemistry, vol. 4, pp. 3633-3642, 2014.
[74] P. Ribière, S. Grugeon, M. Morcrette, S. Boyanov, S. Laruelle and G. Marlair, "Investigation
on the fire-induced hazards of Li-ion battery cells by fire calorimetry," Energy & Enviromental
Science, vol. 5, pp. 5271-5280, 2012.
[75] V. Somandepalli and H. Biteau, "Cone Calorimetry as a Tool for Thermal Hazard Assessment
of Li-Ion Cells," in SAE 2014 World Congress & Exhibition, 2014.
[76] W. Zhang, C. X, Q. Chen, C. Ding, J. Liu, M. Chen and J. Wang, "Combustion calorimetry of
carbonate electrolytes used in lithium ion batteries," Journal of Fire Sciences, vol. 33, no. 1, pp.
22-36, 2015.
[77] E. G. Gebresilassie, S. Grugeon, S. Laruelle, S. Boyanov, A. Lecocq, J.-P. Bertranda and G.
Marlair, "In-depth safety-focused analysis of solvents used in electrolytes for large scale
lithium ion batteries," Royal Society of Chemistry, vol. 15, pp. 9145-9155, 2013.
[78] F. Larsson, P. Andersson, P. Blomqvist and B.-E. Mellander, "Toxic fluoride gas emissions
from lithium-ion battery fires," Scientific Reports, vol. 7, 2017.
[79] Board of the National Research Council, Acute Exposure Guideline Levels for Selected
Airborne Chemicals Volume 4, Washington: National Academy of Sciences, 2004.
[80] National Highway Traffic Safety Administration, "Chevrolet Volt Battery Incident Overview
Report DOT HS 811 573," National Highway Traffic Safety Administration, 2012.
[81] D. Orf, "Tesla Model 3 Blows Up ... Twice ... on Busy Highway," Popular Mechanics, 13
August 2019. [Online]. Available: https://www.popularmechanics.com/cars/hybrid-
electric/a28687473/tesla-model-3-explosion/. [Accessed 29 August 2019].
[82] National Highway Safety Board, "Battery Fire in Electric-powered Passenger Car
HWY18FH014 Preliminary," National Highway Safety Board, 2018.
[83] V. Tangermann, Futurism, 23rd April 2019. [Online]. Available: https://futurism.com/the-
byte/tesla-fire-shanghai-parking-garage.
[84] E. Shilling, "Jalopnik," Jalopnik, 26 03 2019. [Online]. Available:
https://jalopnik.com/firefighters-drop-smoking-bmw-i8-into-a-giant-tub-of-wa-1833582453.
[Accessed 05 09 2019].
[85] L. Mellert, U. Welte, M. Hermann, M. Kompatscher and P. X., "Electric Mobility and Road
Tunnel Safety Hazards of Electric Vehicle Fires," in 9th International Conference 'Tunnel Safety and
Ventilation', Graz, 2018.
167
[86] L. Mellert, U. Welte and M. Hermann, "Electric Mobility and Road Tunnel Safety Hazards of
Electric Vehicle Fires," in 9th International Conference 'Tunnel Safety and Ventilation', Graz, 2018.
[87] A. S. Boe, "Full-scale fire test of electric car," SP Fire Research AS, 2017.
[88] A. Lecocq, M. Bertana, B. Truchot and G. Marlair, "Comparison of the fire consequences of
an electric," International Conference on Fires In Vehicles, pp. 183-194, 2012.
[89] R. Xiong, H. He, H. Guo and Y. Ding, "Modeling for Lithium-Ion Battery used in Electric
Vehicles," Procedia Engineering, vol. 15, pp. 1869-2874, 2011.
[90] D. P. Finegag, J. Darst, W. Walker, Q. Li, C. Yang, R. Jervis, T. M. Heenan, J. Hack, J. C.
Thomas, A. Rack, D. J. Brett, P. R. Shearing, M. Keyser and E. Darcy, "Modelling and
experiments to identify high-risk failure scenarios for testingthe safety of lithium-ion cells,"
Journal of Power Sources, vol. 417, no. 31, pp. 29-41, 2019.
[91] K. Davidsson, I. Karlsson, P. Leisner, M. Bobert and P. Blomqvist, "Safety test methods for
EV batteries," Wolrd electric Vehicle Journal, vol. 4, pp. 414-420, 2010.
[92] "Train car carrying Lithium batteries explodes near downtown Houston," KHOU-11, 24 04
2017. [Online]. Available: https://www.khou.com/article/news/local/train-car-carrying-
lithium-batteries-explodes-near-downtown-houston/433576556. [Accessed 23 04 2020].
[93] NGV Global, "Current Natural Gas Vehicle Statistics," NGV Global, 31 07 2019. [Online].
Available: http://www.iangv.org/current-ngv-stats/. [Accessed 04 09 2019].
Available: https://afdc.energy.gov/fuels/natural_gas_cng_stations.html. [Accessed 23 12
2019].
Available: https://afdc.energy.gov/. [Accessed 09 07 2019].
[96] NGV America, "Enviroment," [Online]. Available:
https://www.ngvamerica.org/environment/. [Accessed 23 12 2019].
[97] NGV Global News, "U.S. EIA’s 2017 Outlook Forecasts Strong Growth for Natural Gas,"
[Online]. Available: https://www.ngvglobal.com/blog/u-s-eias-2017-outlook-forecasts-
strong-growth-natural-gas-0917. [Accessed 23 12 2019].
[98] California Air Resources Board, 2018 Annual Evaluation of Fuel Cell Electric Vehicle
Deployment & Hydrogen Fuel Station Network Development, July 2018.
[99] World Nuclear Association, "Heat Values of Various Fuels," World Nuclear Association, 08
2018. [Online]. Available: https://www.world-nuclear.org/information-library/facts-and-
figures/heat-values-of-various-fuels.aspx. [Accessed 05 09 2019].
[100] The Engineering ToolBox, "Engineering ToolBox," [Online]. Available:
https://www.engineeringtoolbox.com/. [Accessed 23 12 2019].
[101] I. H. Bell, Wronski, Jorrit, Quoilin, Sylvain and V. Lemort, "Pure and Pseudo-pure Fluid
Thermophysical Property Evaluation and the Open-Source Thermophysical Property Library
CoolProp," Industrial & Engineering Chemistry Research , vol. 53, no. 6, pp. 2498-2508, 2014.
[102] NGV Global, "Natural Gas Vehicle Database," [Online]. Available: https://www.iangv.org/.
[Accessed 08 07 2019].
[103] R. Ono, N. Masaharu, S. Fujiwara, S. Horiguchi and T. Oda, "Minimum ignition energy of
hydrogen–air mixture: Effects of humidity and spark duration," Journal of Electrostatics, vol. 65,
no. 2, pp. 87-93, 2007.
168
[104] C. J. Coronado, J. A. Carvalho Jr., J. C. Andrade, E. V. Cortez, F. S. Carvalho, J. C. Santos
and A. Z. Mendiburu, "Flammability limits: A review with emphasis on ethanol for
aeronautical applications and description of the experimental procedure," Journal of Hazardous
Materials, vol. 32, no. 54, pp. 241-242, 2012.
[105] R. Eckhoff, M. Ngo and W. Olsen, "On the minimum ignition energy (MIE) for
propane/air," Journal of Hazardous Materials, vol. 175, no. 1-3, pp. 293-297, 2010.
[106] Air Products and Chemicals Inc., Ignition Energy of H2, CH4, and gasoline with Air, Air Products,
2001.
[107] The Engineering ToolBox, "Adiabatic Flame Temperatures," [Online]. Available:
https://www.engineeringtoolbox.com/adiabatic-flame-temperature-d_996.html. [Accessed 04
12 2019].
[108] H. Haase, Electrostatic Hazards: their evaluation and control, Verlag Chemie, 1977.
[109] V. Babrauskas, Ignition Handbook, 2003.
[110] The Engineering ToolBox, "Fuels and Chemicals - Autoignition Temperatures," [Online].
Available: https://www.engineeringtoolbox.com/fuels-ignition-temperatures-d_171.html.
[Accessed 04 12 2019].
[111] J. M. Kuchta, "Investigation of Fire and Explosion Accidents in the Chemical, Mining, and
Fuel-Related Industries: A Manual," US Bureau of Mines, vol. Bulletin 680, 1985.
[112] R. G. Zalosh, J. Amy, C. E. Hofmeister and W. Wang, "Dispersion of CNG fuel releases in
naturally ventilated tunnels," Center for Firesafety Studies, Worcester Polytechnic Institue , vol. 1, 1994.
[113] B. D. Ehrhart, D. M. Brooks, A. B. Muna and C. B. LaFleur, "Risk Assessment of Hydrogen
Fuel Cell Electric Vehicles in Tunnels," Fire Technology, pp. 1-22, 2019.
[114] National Fire Protection Association, NFPA 52: Vehicular Natural Gas Fuel Systems Code,
Quincy, 2019.
[115] National Fire Protection Association, NFPA 55: Compressed Gases and Cryogenic Fluids
Code, Quincy, 2013 Edition.
[116] National Fire Protection Association, NFPA 57: Liquefied Natural Gas (LNG) Vehicular Fuel
Systems Code, Quincy, 2002.
[117] National Fire Protection Association, NFPA 59A: Standard for the Production, Storage, and
Handling of Liquefied Natural Gas (LNG), Quincy, 2019.
[118] SAE International, J1616: Recommended Practice for Compressed Natural Gas Vehicle Fuel,
1994.
[119] SAE International, J2406: Recommended Practices for CNG Powered Medium and Heavy-
Duty Trucks, 2002.
[120] B. Lowesmith, G. Hankinson and D. Johnson, "Vapour cloud explosions in a long congested
region involving methane/hydrogen mixtures," Process Safety and Environmental Protection, vol.
89, no. 4, pp. 234-247, 2011.
[121] D. Bjerketvedt, J. R. Bakke and K. v. Wingerden, "Gas explosion handbook," Journal of
Hazardous Materials, vol. 52, no. 1, pp. 1-150, 1997.
[122] R. Harris and M. Wickens, "Understanding vapour cloud explosions-an experimental study,"
The Institution of Gas Engineers, vol. 1408, 1989.
169
[123] M. Royle, L. Shirvill and T. Roberts, "Vapour cloud explosions from the ignition of
methane/hydrogen/air mixtures in a congested region," in International Conference on Hydrogen
Safety, 2007.
[124] R. K. Zipf, V. N. Gamezo, K. M. Mohamed, E. S. Oran and D. A. Kessler, "Deflagration-to-
detonation transition in natural gas–air mixtures," Combustion and Flame, vol. 161, no. 8, pp.
2165-2176, 2014.
[125] M. Kokkala, "Experimental study of heat transfer to ceiling from an impinging diffusion
flame," Fire Safety Science, vol. 3, pp. 261-270, 1991.
[126] A. Venetsanos, D. Baraldi, P. Adams, P. Heggem and H. Wilkening, "CFD modelling of
hydrogen release, dispersion and combustion for automotive scenarios," Journal of Loss
Preventation in the Process Industries, vol. 21, no. 2, pp. 162-184, 2008.
[127] P. Middha and O. Hansen, "CFD Simulation Study to Investigate the Risk from Hydrogen
Vehicles in Tunnels," Int J Hydrogen Energy, vol. 34, pp. 5875-5876, 2009.
[128] L. Zhiyong, X. Pan, K. Sun and J. Ma, "Comparison of the harm effects of accidental releases:
Cryo-compressed hydrogen versus natural gas," Internation Journal of Hydrogen Energy, vol. 38,
no. 25, pp. 11174-11180, 2013.
[129] H. Witlox and A. Holt, "Unified Dispersion Model," DNV Software Product and
Development, London, 2000.
[130] J. Cook, Z. Bahrami and R. Whitehouse, "A comprehensive program for calculation of flame
radiation levels," Journal of Loss Prevention in the Process Industries, vol. 3, no. 1, pp. 150-155, 1990.
[131] M. Tang and Q. Baker, "A new set of blast curves from vapor cloud explosion," American
Institute of Chemical Engineers, vol. 18, no. 4, pp. 235-240, 2004.
[132] J. LaChance, A. Tchouvelev and A. Engebo, "Development of uniform harm criteria for use
in quantitative risk analysis of the hydrogen infrastructure," International Journal of Hydrogen
Energy, vol. 36, no. 3, pp. 2381-2388, 2011.
[133] A. Marangon, M. Carcassi, A. Engebo and S. Nilsen, "Safety distances: Definition and
values," International Journal of Hydrogen Energy, vol. 32, no. 13, pp. 2192-2197, 2007.
[134] Y. Z. Li, "Fire and Explosion Hazards of Alternative Fuel Vehicles in Tunnels," Brandforsk,
Stockholm, 2018.
[135] S. Chamberlain and M. Modarres, "Compressed natural gas bus safety: a quantitative risk
assessment.," Risk Analysis, vol. 25, no. 2, 2005.
[136] M. Krupka, A. Peaslee and H. Laquer, "Gaseous Fuel Safety Assessment for Light-Duty
Automotive Vehicles," Los Alamos National Laboratory Report LA-9829-MS, November
1983.
[137] R. Zalosh, W. Wang and J. Amy, "Hazard Analysis of Alternative Fueled Vehicles in Tunnels
Part 2: LPG and LNG Fueled Vehicles," Center for Firesafety Studies, Worcester Polytechnic
Institute, November 1995.
[138] US Department of Transpotation Federal Highway Administration, "Verification,
Refinement, and Applicability of Long-Term Pavement Performance Vehicle Classification
Rules FHWA-HRT-13-091," Federal Highway Administration, 2014.
[139] Power Solutions International, "GM G2500/G3500 Express & Savana," Power Solutions
International, [Online]. Available: https://psiengines.com/wp-
content/uploads/2018/02/PSI_SVM_G2500-G3500_Van_CNG_Spec.pdf. [Accessed 05 09
2019].
170
[140] O. Topal and I. Nakir, "Total Cost of Ownership Based Economic Analysis of Diesel, CNG
and Electric Bus Concepts for the Public Transport in Istanbul City," Energies, vol. 11, no. 9,
p. 2369, 2018.
[141] Daimler Truck Cascadia Freightliner, "Natural Gas Fuel Your Future," Daimler, 13 06 2019.
[Online]. Available: https://freightlinerads.azureedge.net/3642-
new_cascadia_natural_gas_sell_-2019-06-13-1.pdf. [Accessed 05 09 2019].
[142] "Alternative Fuels Data Center," US Department of Energy: Energy Efficiency & Renewable
Energy, [Online]. Available: https://afdc.energy.gov/vehicles/propane.html. [Accessed July
2019].
[143] E. Hahn, "Is Propane a Liquid or Gas? Liquid Propane vs Gas Propane," ELGAS - LPG Gas
for Home & Business, 20 10 2019. [Online]. Available: https://www.elgas.com.au/blog/726-
what-is-the-difference-between-liquid-and-vapour-lpg. [Accessed 12 3 2020].
[144] The Engineering ToolBox, "Propane - Vapor Pressure," [Online]. Available:
https://www.engineeringtoolbox.com/propane-vapor-pressure-d_1020.html. [Accessed 24
12 2019].
[145] Argonne National Laboratory, "Propane Vehicles: Status, Challenges, and Opportunites
ANL/ESD/10-2," May 2010.
[146] A. Brecher, A. Epstein and A. Breck, "Review and Analysis of Potential Safety Impacts of and
Regulatory Barriers to Fuel Efficiency Technologies and Alternative Fuels in Medium- and
Heavy-Duty Vehicles," U.S. Department of Transportation, June 2015.
[147] A. Pundkar, S. Lawankar and S. Deshmukh, "Performance and Emissions of LPG Fueled
Internal Combustion Engine: A Review," International Journal of Scientific and Engineering Research,
vol. 3, no. 3, March 2012.
[148] National Fire Protection Association, NFPA 58: Liquefied Petroleum Gas Code, Quincy,
2017.
[149] D. Brzezinska and A. S. Markowski, "Experimental investigation and CFD modelling of the
internal car park environment in case of accidental LPG release,"
ProcessSafetyandEnvironmentalProtection, vol. 110, pp. 5-14, 2017.
[150] G. Atkison and Y. Wu, "Smoke Control in Sloping Tunnels," Fire Safety Journal, pp. 335-341,
1996.
[151] Y. Oka and G. T. Atkinson, "Control of Smoke Flow in Tunnel Fires," Fire Safety Journal, vol.
25, pp. 305-322, 1995.
[152] D. Webber and S. Jones, "A Model of Spreading Vaporising Pools," in International Conference
on Vapor Cloud Modeling, 1987.
[153] J. Weerheijm and B. van den Berg, "Explosion Risks and Consequences for Tunnels," in Sixth
International Symposium on Tunnel Safety and Security, Marseille, France, March 12-14, 2014.
[154] National Renewable Energy Laboratory, On-Road Fuel Cell Electric Vehicles Evaluation:
Overview, March 2019.
[155] Hyundai Motor America, "Hyundai ix35 Fuel Cell," Hyundai Media Center, [Online].
Available: https://www.hyundainews.com/en-us/releases/1624. [Accessed 04 04 2020].
[156] J. Kurtz, S. Sprik, G. Saur and S. Onorato, "On-Road Fuel Cell Electric Vehicles Evaluation:
Overview NREL/TP-5400-73009," National Renewable Energy Laboratory, March 2019.
[157] Plug Power, "WHO’S USING GENDRIVE?," [Online]. Available:
https://www.plugpower.com/customer/whos-using-gendrive/. [Accessed 03 02 2020].
171
[158] Fuel Cell & Hydrogen Energy Association, "[Company] Celebrates National Hydrogen and
Fuel Cell Da," [Online]. Available: http://www.fchea.org/sample-press-release-2019.
[Accessed 03 02 2020].
[159] International Partnership for Hydrogen and Fuel Cells in the Economy, [Online]. Available:
https://www.iphe.net/united-states. [Accessed 20 April 2020].
[160] N. LePan, "The Evolution of Hydrogen: From the Big Bang to Fuel Cells," Visual Capitalist,
23 4 2019. [Online]. Available: https://www.visualcapitalist.com/evolution-of-hydrogen-fuel-
cells/. [Accessed 3 1 2020].
[161] K. Kunze and O. Kircher, "CRYO-COMPRESSED HYDROGEN STORAGE," BMW
Group, Oxford, 2012.
[162] College of the Desert, Hydrogen Fuel Cell Engines and Related Technologies, Module 1:
Hydrgoen Properties, December 2001.
[163] I. Abe, Energy Carriers and Conversion Systems - Vol. I - Physical and Chemical Properties
of Hydrogen.
[164] National Fire Protection Association, NFPA 77: Recommended Practice on Static Electricity,
Quincy, 2019.
[165] C. LaFleur, "Large Scale Hydrogen System Safety Issues," Chemical Engineering Progress, vol.
115, no. 8, pp. 42-46, 2019.
[166] National Fire Protection Association, NFPA 2: Hydrogen Technologies Code, Quincy, 2020
Edition.
[167] Global Registry, "Addendum 13: Global technical regulation No. 13 Global technical
regulation on hydrogen and fuel cell vehicles," UNITED NATIONS, 2013.
[168] National Fire Protection Association, NFPA 2: Hydrogen Technologies Code, Quincy, 2016
Edition.
[169] F. Dryer, M. Chaos, Z. Zhao, J. Stein, J. Alpert and C. Homer, "Spontaneous Ignition of
Pressurized Releases of Hydrogen and Natural Gas into Air," Combust. Sci. and Tech. 2007,
2007.
[170] M. Groethe, E. Merilo, J. Colton, S. Chiba, Y. Sato and H. Iwabuchi, "Large-scale Hydrogen
Deflagrations and Detonations," Int J Hydrogen Energy , vol. 32, no. 13, pp. 2153-2133, 2007.
[171] J. M. Chock, "Review of Methods for Calculating Pressure Profiles of Explosive Air Blast and
its Sample Application," Virginia Polytechnic Institute and State University, Blacksburg, 1999.
[172] F. Tamanini, "Scaling parameters for vented gas and dust explosions," Journal of Loss Prevention
in the Process Industries, vol. 14, pp. 455-461, 2001.
[173] G. Houf, G. Evans, E. Merilo, M. Groethe and S. James, "Releases from Hydrogen Fuel-cell
Vehicles in Tunnels," Int J Hydrogen Energy, vol. 37, pp. 715-719, 2012.
[174] S. Kumar, S. Miles, P. Adams, A. Kotchourko, D. Heldey, P. Middha, V. Molkov, A.
Teodorczyk and M. Zenner, "HyTunnel Project to Investigate the use of Hydrogen Vehicles
in Road Tunnels," in 3rd International Conference on Hydrogen Safety (ICHS3), 2009.
[175] A. Friedrich, J. Grune, T. Jordan, A. Kotchourko, N. K. M. Kotchourko, K. Sempert and G.
Stern, "EXPERIMENTAL STUDY OF HYDROGEN-AIR DEFLAGRATIONS IN FLAT
LAYER," in Proc. 2nd ICHS International conference on hydrogen safety, San Sebastian, 2007.
172
[176] M. Seike, Y. Ejiri, N. Kawabata, M. Hasegawa and H. Tanaka, "Fire experiments of carrier
loaded FCV in full-scale model tunnel - Estimation of heat release rate and smoke generation
rate," in Third International Converence on Fire in Vehicles, Berlin, Germany, 2014.
[177] I. Tolias, A. Venetsanos, N. Markatos and C. Kiranoudis, "CFD Modeling of Hydrogen
Deflagration in a Tunnel," Int J Hydrogen Energy, vol. 39, pp. 20538-20546, March 2014.
[178] H. Bie and Z. Hao, "Simulation Analysis on the Risk of Hydrogen Releases and Combustion
in Subsea Tunnels," Int J Hydrogen Energy, 2016.
[179] Y. Wu, "Assessment of the impact of ject flame hazard from hydrogen cars in road tunnels,"
Transportation Research Part C, vol. 16, pp. 246-254, 2008.
[180] S. Mukai, J. M. H. Suzuki, K. Oyakawa and S. Watanabe, "CFD simulation on diffusion of
leaked hydrogen caused by vehicle accident in tunnels," in Proceedings of the International
Conference on Hydrogen Safety (ICHS’05), 2005.
[181] ICHS, [Online]. Available: https://hysafe.info/ichs2019/. [Accessed 24 03 2020].
[182] HyTunnel-CS, "About HyTunnel-CS," HyTunnel-CS, 2019. [Online]. Available:
https://hytunnel.net/?page_id=31. [Accessed 26 09 2019].
[183] I. Abe, Energy Carriers and Conversion Systems - Vol. I - Physical and Chemical Properties
of Hydrogen.
[184] Global Registry, "Addendum 13: Global technical regulation No. 13 Global technical
regulation on hydrogen and fuel cell vehicles," UNITED NATIONS, 2013.
[185] Sandia National Laboratories, Hydrogen Fuel Cell Electric Vehicle Tunnel Safety Study,
October 2017.
[186] F. Dryer, M. Chaos, Z. Zhao, J. Stein, J. Alpert and C. Homer, "Spontaneous Ignition of
Pressurized Releases of Hydrogen and Natural Gas into Air," Combust. Sci. and Tech. 2007,
2007.
173
DISTRIBUTION
Email—External
Name Company Email Address Company Name
Laura Hill [email protected] DOE Office of Energy Efficiency & Renewable Energy
Mark Richards [email protected] DOE Office of Energy Efficiency & Renewable Energy
Neha Rustagi [email protected] DOE Office of Energy Efficiency & Renewable Energy
Stephen Bartha [email protected] DOT Federal Highway Administration
Cyrus J. Jordan [email protected] North Carolina State University (Former Sandia Intern)
Email—Internal
Name Org. Sandia Email Address
Alice B. Muna 05814 [email protected]
Ethan S. Hecht 08367 [email protected]
Jonathan A. Zimmerman 08367 [email protected]
Myra L. Blaylock 08751 [email protected]
Austin R. Baird 08854 [email protected]
Brian D. Ehrhart 08854 [email protected]
Austin M. Glover 08854 [email protected]
Chris B. LaFleur 08854 [email protected]
Technical Library 01977 [email protected]
174
175
Sandia National Laboratories
is a multimission laboratory
managed and operated by
National Technology &
Engineering Solutions of
Sandia LLC, a wholly owned
subsidiary of Honeywell
International Inc. for the U.S.
Department of Energy’s
National Nuclear Security
Administration under contract
DE-NA0003525.

Alternative Fuel Vehicles in Tunnels: Sandia Report

Uploaded by

Copyright:

Available Formats

Alternative Fuel Vehicles in Tunnels: Sandia Report

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Alternative Fuel Vehicles in Tunnels: Sandia Report

Uploaded by

Copyright:

Available Formats

SANDIA REPORT

Alternative Fuel Vehicles in

Available to DOE and DOE contractors from

Telephone: (865) 576-8401

Available to the public from

Telephone: (800) 553-6847

1.1. Definitions for Hazard Metrics

 Laminar Flame Speed

 Adiabatic Flame Temperature

 Heat Release Rate

Table 1: Human Injury Criteria from Overpressure (from [13])

Table 2: Effects from Heat Flux (from [13])

Table 3: Human Injury Criteria from Temperature (from [12])

Table 4: Vehicle Weight Classifications (from [15])

Heavy-Duty Vehicle Heavy-Duty Engine: >8,500 lbs.

2.1. Overview of Traditional Fuels

2.2. Properties of Traditional Fuels

2.3. Associated Hazards of Traditional Fuels

2.4.1. National Fire Protection Association Standard 502

2.4.3. NCHRP Guidelines for Emergency Ventilation Smoke Control in

2.4.4. NCHRP Synthesis 415: Design Fires in Road Tunnels (2011)

Table 7: List of Significant Full-Scale Tunnel Fire Tests (from [42])

Calibrated fires, natural,

2.5.1.1. The Rijkswaterstaat (RWS) and Other Time-Temperature Curves

Figure 2: Standard time-temperature fire design curves at tunnel structure interfaces

2.5.1.2. The Runehamar Full Scale Tests

Table 8: Commodities used as fuel in the HGV tests (from [49])

2.5.1.3. Large Scale Fire Tests in the Second Benelux Tunnel

Figure 5: Test Overview for Second Benelux Tunnel (from [51])

2.5.2.1. Simulation of Tunnel Fires Using a Zone Model

Figure 8: Smoke Characteristics Prediction of a Truck Fire (from [53])

Figure 9: Smoke Characteristics Prediction of a School Bus Fire (from [53])

2.5.3.1. Determination of critical parameters in the analysis of road tunnel fires

Figure 10: Tunnel Dimension Effect on Average Temperature (from [54])

2.5.3.2. An analysis of tunnel fire characteristics under the effects of blockage

3.1. Overview of Technology

One-time use Rechargeable

Smaller light weight applications Custom, larger scale application

Metallic lithium as the Anode Porous carbon as Anode

3.2. Properties of Lithium-ion Batteries

Figure 13: Lithium-ion Battery Chemistry (from [59])

3.3. Associated Hazards

Figure 14: 18650 Cell Defect (from [60])

Figure 15: BEV Failure Event

3.3.2. Fire Hazards

Figure 17: Cell Component Breakdown by % Mass (from [73])

Table 11: Cell Deconstruction Breakdown by % Mass

3.4.1. National Fire Protection Association Standard 502

3.4.3. NCHRP Guidelines for Emergency Ventilation Smoke Control in

3.4.7. FreedomCAR Battery Test Manual for Power-Assist Hybrid Electric

3.5.1.1. Reported BEV Failures and Incidents

3.5.1.2. Fire Analysis of BEVs in a Road Tunnel

Cathode Active Material NMC

Electrolyte Lithium hexafluorophosphate (LiPF6)

Table 14: Test Scenarios (from [85])

Figure 19: Test tunnel area layout (from [86])

PH3 [g] < 0.4 --- < 0.4 ---

F- as HF [g] 1.1 3.1 <1 < 0.5