Public Comment No. 4-NFPA 30A-2016 [ Section No ] Statement of Problem and Substantiation for Public Comment

Transcription

1 National Fire Protection Association Report 1 of 42 10/26/2016 3:45 PM Public Comment No. 4-NFPA 30A-2016 [ Section No ] ASTM Publications. ASTM International, 100 Barr Harbor Drive, P.O. Box C700, West Conshohocken, PA ASTM D5, Standard Test Method for Penetration of Bituminous Materials, ASTM D56, Standard Test Method for Flash Point by Tag Closed Cup Tester,2005, reapproved ASTM D93, Standard Test Methods for Flash Point by Pensky-Martens Closed Cup Tester, 2015a. ASTM D323, Standard Method of Test for Vapor Pressure of Petroleum Products (Reid Method), 2015a. ASTM D3278, Standard Test Methods for Flash Point of Liquids by Small Scale Closed-Cup Apparatus, 1996, reapproved ASTM D3828, Standard Test Methods for Flash Point of Liquids by Small Scale Closed Cup Tester, 2012a. ASTM D4359, Standard Test for Determining Whether a Material in a Liquid or a Solid, 1990, reapproved ASTM E648, Standard Test Method for Critical Radiant Flux of Floor-Covering Systems Using a Radiant Heat Energy Source, 2015e1. Statement of Problem and Substantiation for Public Comment Add standard proposed in public input and comment Related Public Comments for This Document Related Comment Public Comment No. 3-NFPA 30A-2016 [Section No ] Related Item Public Input No. 56-NFPA 30A-2015 [Section No ] Relationship Submitter Information Verification Submitter Full Name: Marcelo Hirschler Organization: GBH International Street Address: City: State: Zip: Submittal Date: Thu Mar 24 20:04:59 EDT 2016 Page 1 of 129

2 National Fire Protection Association Report of /26/2016 3:45 PM Committee Statement Committee Action: Resolution: Rejected ASTM E648 was not added to the body of the code so it not added to the references. Page 2 of 129

3 National Fire Protection Association Report of /26/2016 3:45 PM Public Comment No. 14-NFPA 30A-2016 [ Section No [Excluding any Sub-Sections] ] Except as modified by the provisions of this subsection chapter, aboveground storage tanks shall meet all applicable requirements of Chapters 21 and 22 of NFPA 30. Statement of Problem and Substantiation for Public Comment clarify that compliance with the entire chapter is required, not just this subsection. Related Item Committee Input No. 15-NFPA 30A-2015 [New Section after ] Submitter Information Verification Submitter Full Name: Marcia Poxson Organization: Michigan Bureau of Fire Servic Street Address: City: State: Zip: Submittal Date: Tue May 10 16:05:48 EDT 2016 Committee Statement Committee Accepted Action: Resolution: SR-2-NFPA 30A-2016 Statement: This revision clarifies that compliance with the entire chapter is required, not just this subsection. Page 3 of 129

4 National Fire Protection Association Report of /26/2016 3:45 PM Public Comment No. 15-NFPA 30A-2016 [ New Section after ] * All storage tank appurtenances shall be installed, calibrated, operated, and maintained in accordance with either the manufacturer s instructions, recognized engineering practices, or equivalent methods approved by the AHJ. Routine maintenance and service checks for operability and running condition shall be subject to this requirement. A See A Statement of Problem and Substantiation for Public Comment The addition of this language will ensure that storage tank systems are installed and maintained in accordance with recognized practices. Related Item Committee Input No. 15-NFPA 30A-2015 [New Section after ] Submitter Information Verification Submitter Full Name: Marcia Poxson Organization: Michigan Bureau of Fire Servic Street Address: City: State: Zip: Submittal Date: Tue May 10 16:07:38 EDT 2016 Committee Statement Committee Rejected but see related SR Action: Resolution: SR-4-NFPA 30A-2016 Statement: The addition of this language ensures that storage tank appurtenances are installed in accordance with recognized practices. Page 4 of 129

5 National Fire Protection Association Report of /26/2016 3:45 PM Public Comment No. 8-NFPA 30A-2016 [ Section No [Excluding any Sub-Sections] ] Where liquid is supplied to the dispensing device under pressure, a listed, rigidly anchored, double-poppet type emergency shutoff valve incorporating a incorporating a pressure relief valve on the dispenser side with a rating of minimum 10 PSI and less than the dispenser's rated burst pressure and a fusible link or other thermally actuated device, designed to close automatically in the event of severe impact or fire exposure, shall be installed in the supply line at the base of each individual island-type dispenser or at the inlet of each overhead dispensing device. The emergency shutoff valve shall be installed in accordance with the manufacturer's instructions. The emergency shutoff valve shall not incorporate a slip-joint feature. Exception: As provided for in Statement of Problem and Substantiation for Public Comment Without additional pressure relief in dispenser side a hazardous condition maybe created by fire heating fuel captured in dispenser leading to an explosion. A controlled release will limit fuel leakage and control pressure in dispenser side limiting the risk of explosion. Related Item First Revision No. 32-NFPA 30A-2015 [Section No [Excluding any Sub-Sections]] Submitter Information Verification Submitter Full Name: Peter Manger Organization: OPW Retail Fueling Street Address: City: State: Zip: Submittal Date: Wed May 04 14:47:25 EDT 2016 Committee Statement Committee Action: Resolution: Rejected Insufficient information was provided to substantiate the need for a double poppet with pressure relief. Page 5 of 129

6 National Fire Protection Association Report of /26/2016 3:45 PM Public Comment No. 17-NFPA 30A-2016 [ Section No. 7.1 ] 7.1 Scope. This chapter shall apply to the construction of buildings and portions of buildings that are motor fuel dispensing facilities or repair garages. Repair facilities that also repair vehicles powered by hydrogen shall also meet the requirements of NFPA 2. Statement of Problem and Substantiation for Public Comment Recommendation of the Joint NFPA 2/30A Task Group in reference to Committee Inputs # 13 and 25. Related Item Committee Input No. 25-NFPA 30A-2015 [Section No. 2.4] Committee Input No. 13-NFPA 30A-2015 [Section No ] Submitter Information Verification Submitter Full Name: Ronald Laurence Organization: Stantec Consulting Services, I Affilliation: NFPA 2/30A Task Group Street Address: City: State: Zip: Submittal Date: Fri May 13 14:45:45 EDT 2016 Committee Statement Committee Rejected but see related SR Action: Resolution: SR-3-NFPA 30A-2016 Statement: This revision clarifies where the requirements of NFPA 2 and NFPA 30A are applicable. Page 6 of 129

7 National Fire Protection Association Report of /26/2016 3:45 PM Public Comment No. 3-NFPA 30A-2016 [ Section No ] Drainage. In areas of repair garages used for repair or servicing of vehicles, floor assemblies shall be constructed of noncombustible materials or, if combustible materials are used in the assembly, they shall be surfaced with approved, nonabsorbent, noncombustible material, except as indicated in 7. Exception: Slip Slip -resistant, nonabsorbent, interior floor finishes having a critical radiant flux not more than 0.45 W/cm 2 (9.87 Btu/in. 2 ), as determined by NFPA 253 or by ASTM E648, shall be permitted Floors shall be liquidtight to prevent the leakage or seepage of liquids and shall be sloped to facilitate the movement of water, fuel, or other liquids to floor drains In areas of repair garages where vehicles are serviced, any floor drains shall be properly trapped and shall discharge through an oil/water separator to the sewer or to an outside vented sump. Additional Proposed Changes File Name ASTM_vs_NFPA_vs_UL_vs_ISO_vs_IEC_tests_with_titles_Nov_2015.pdf Description Approved Comparison between ASTM and NFPA tests Statement of Problem and Substantiation for Public Comment ASTM E648 is identical to NFPA That has been recognized by the NFPA technical committee on fire tests and by ASTM committee E05 on fire standards. It has also been recognized by NFPA 101 and 5000 and by the International Building Code, all of which allow the standards interchangeably. In recent years ASTM and NFPA have been efficient in ensuring that both standards be maintained to be identical. Related Public Comments for This Document Related Comment Public Comment No. 4-NFPA 30A-2016 [Section No ] Related Item Public Input No. 56-NFPA 30A-2015 [Section No ] Relationship Submitter Information Verification Submitter Full Name: Marcelo Hirschler Page 7 of 129

8 Fire Test Methods (November 2015) ASTM NFPA UL ISO IEC D2859 (Standard Test Method for Ignition Characteristics of Finished Textile Floor Covering Materials) E84 (Standard Test Method for Surface Burning Characteristics of Building Materials) E108 (Standard Test Methods for Fire Tests of Roof Coverings) E119 (Standard Test Methods for Fire Tests of Building Construction and Materials) E136 (Standard Test Method for Behavior of Materials in a Vertical Tube Furnace at 750 C) E162 (Standard Test Method for Surface Flammability of Materials Using a Radiant Heat Energy Source) E648 (Standard Test Method for Critical Radiant Flux of Floor Covering Systems Using Radiant Heat Energy Sources) E662 (Standard Test Method for Specific Optical Density) ASTM E (Standard Test of Surface Burning Characteristics of Building Materials, WD) 256 (Standard Fire Tests of Roof Coverings, WD) 251 (Standard Tests of Fire Endurance of Building Construction and Materials, WD) 723 (Standard Test for Surface Burning Characteristics of Building Materials) 790 (Standard Tests for Fire Resistance of Roof Covering Materials) 263 (Standard Fire Tests of Building Construction and Materials) (External exposure of roofs to fire Part 1: test method) 834 (Fire resistance tests Elements of building construction) 1182 (Reaction to fire tests for products Noncombustibility test) * 253 (Standard Test for Critical Radiant Flux of Floor Covering Systems Using a Radiant Heat Energy Source) 258 (Recommended Practice for Determining Smoke Generation of Solid Materials WD) (Reaction to fire tests for floorings Part 1: Determination of the burning behaviour using a radiant heat source) (Fire hazard testing Part 6 30: Smoke obscuration Small scale static method Apparatus) Page 8 of 129

9 ASTM NFPA UL ISO IEC ASTM E05 (continued) E814 (Standard Test Method for Fire Tests of Through Penetration Fire Stops) 1479 (Standard Fire Test of Through Penetration Stops) E906 (Standard Test Method for Heat and Visible Smoke Release Rates for Materials and Products) E970 (Standard Test Method for Critical Radiant Flux of Exposed Attic Floor Insulation Using a Radiant Heat Energy Source) E1317 (Standard Test Method for Flammability of Marine Surface Finishes) E1321 (Standard Test Method for Determining Material Ignition and Flame Spread Properties) E1352 (Standard Test Method for Cigarette Ignition Resistance of Mock Up Upholstered Furniture Assemblies) E1353 (Standard Test Methods for Cigarette Ignition Resistance of Components of Upholstered Furniture) 263 (Standard Method of Test for Heat and Visible Smoke Release Rates for Materials and Products, WD) (Reaction to fire tests Spread of flame Part 2: Lateral spread on building and transport products in vertical configuration) 261 (Standard Method of Test for Determining Resistance of Mock Up Upholstered Furniture Material Assemblies to Ignition by Smoldering Cigarettes) 260 (Standard Methods of Tests and Classification System for Cigarette Ignition Resistance of Components of Upholstered Furniture) Page 9 of 129

10 ASTM NFPA UL ISO IEC ASTM E05 (Continued) 271 (Standard Method of Test for Heat and Visible Smoke Release Rates for Materials and Products Using an Oxygen Consumption Calorimeter, WD) E1354 (Standard Test Method for Heat and Visible Smoke Release Rates for Materials and Products Using an Oxygen Consumption Calorimeter) E1474 (Standard Test Method for Determining the Heat Release Rate of Upholstered Furniture and Mattress Components or Composites Using a Bench Scale Oxygen Consumption Calorimeter) E1529 (Standard Test Methods for Determining Effects of Large Hydrocarbon Pool Fires on Structural Members and Assemblies) E1537 (Standard Test Method for Fire Testing of Upholstered Furniture) E1590 (Standard Test Method for Fire Testing of Mattresses) 272 (Standard Method of Test for Heat and Visible Smoke Release Rates for Upholstered Furniture Components or Composites and Mattresses Using an Oxygen Consumption Calorimeter, WD) 1709 (Standard Rapid Rise Fire Tests of Protection Materials for Structural Steel) * 266 (Standard Method of Test for Fire Characteristics of Upholstered Furniture Exposed to Flaming Ignition Source, WD) 267 (Standard Method of Test for Fire Characteristics of Mattresses and Bedding Assemblies Exposed to Flaming Ignition Source, WD) (Reaction to fire tests Heat release, smoke production and mass loss rate Part 1: Heat release rate (cone calorimeter method) and smoke production rate (dynamic measurement)) 1056 (Standard Fire Test of Upholstered Furniture, WD) 1895 (Standard Fire Test of Mattresses, WD) Page 10 of 129

11 ASTM NFPA UL ISO IEC ASTM E05 (Continued) E1623 (Standard Test Method for Determination of Fire and Thermal Parameters of Materials, Products, and Systems Using an Intermediate Scale) (Reaction to fire tests Determination of fire and thermal parameters of materials, products and assemblies using an intermediatescale calorimeter (ICAL)) E1678 (Standard Test Method for Measuring Smoke Toxicity for Use in Fire Hazard Analysis) E1725 (Standard Test Methods for Fire Tests of Fire Resistive Barrier Systems for Electrical System Components) E1740 (Standard Test Method for Determining the Heat Release Rate and Other Fire Test Response Characteristics of Wall Covering or Ceiling Covering Composites Using a Cone Calorimeter) E1822 (Standard Test Method for Fire Testing of Stacked Chairs) E1966 (Standard Test Method for Fire Resistive Joint Systems) E1995 (Standard Test Method for Measurement of Smoke Obscuration Using a Conical Radiant Source) 269 (Standard Test Method for Developing Toxic Potency Data for Use in Fire Hazard Modeling) UL 1724 (Fire Test for Electrical Circuit Protective Systems) (WD) 2079 (Standard Tests for Fire Resistance of Building Joint Systems) 270 (Standard Method of Test for Measurement of Smoke Obscuration Using a Conical Radiant Source in a Single Closed Chamber) (Plastics Smoke generation Part 2: Determination of optical density by a singlechamber test) Page 11 of 129

12 ASTM NFPA UL ISO IEC ASTM E05 (Continued) 257 (Standard Fire Test of 9 (Standard Fire Tests of Window and Glass Block Window Assemblies) Assemblies) E2010 (Standard Test Method for Positive Pressure Fire Tests of Window Assemblies, WD) E2058 (Standard Test Methods for Measurement of Synthetic Polymer Material Flammability Using a Fire Propagation Apparatus (FPA)) E2074 (Standard Test Method for Fire Tests of Door Assemblies, Including Positive Pressure Testing of Side Hinged and Pivoted Swinging Door Assemblies, WD) E2102 (Standard Test Method for Measurement of Mass Loss and Ignitability for Screening Purposes Using a Conical Radiant Heater) E2187 (Standard Test Method for Measuring the Ignition Strength of Cigarettes) E2257 (Standard Test Method for Room Fire Test of Wall and Ceiling Materials and Assemblies) ** 287 (Standard Test Methods for Measurement of Flammability of Materials in Cleanrooms Using a Fire Propagation Apparatus (FPA)) 252 (Standard Methods of Fire Tests of Door Assemblies) 10B (Standard Fire Tests of Door Assemblies) and 10C (Standard for Positive Pressure Fire Test of Door Assemblies) (Plastics Simple heat release test using a conical radiant heater and a thermopile detector) (Standard test method for assessing the ignition propensity of cigarettes) 265 (Standard Methods of Fire Tests for Evaluating Room Fire Growth Contribution of Textile or Expanded Vinyl Wall Coverings on Full Height Panels and Walls), 286 (Standard Methods of Fire Tests for Evaluating Contribution of Wall and Ceiling Interior Finish to Room Fire Growth) ** 1715 (Standard Fire Test of Interior Finish Material) ** 9705 (Fire tests Fullscale room test for surface products) ** Page 12 of 129

13 ASTM NFPA UL ISO IEC ASTM E05 (Continued) E2307 (Standard Test Method for Determining the Fire Endurance of Perimeter Fire Barrier Systems Using the Intermediate Scale Multi Story Test Apparatus) E2336 (Standard Test Methods 2221 (Test for Fire for Fire Resistive Grease Duct Enclosure Systems) Resistance of Grease Duct Enclosure Assembly) E2632 (Standard Test Method for Evaluating the Under Deck Fire Test Response of Deck Materials) E2652 (Standard Test Method for Behavior of Materials in a Tube Furnace with a Cone shaped Airflow Stabilizer, at 750 C) E2707 (Standard Test Method for Determining Fire Penetration of Exterior Wall Assemblies Using a Direct Flame Impingement Exposure) E2768 (Standard Test Method for Extended Duration Surface Burning Characteristics of Building Materials (30 min Tunnel Test)) E2816 (Standard Test Methods for Fire Resistive Metallic HVAC Duct Systems) 1182 (Reaction to fire tests for products Noncombustibility test) Page 13 of 129

14 ASTM NFPA UL ISO IEC ASTM E05 (Continued) E2837 (Standard Test Method for Determining the Fire Resistance of Continuity Head of Wall Joint Systems Installed Between Rated Wall Assemblies and Nonrated Horizontal Assemblies) E2886 (Standard Test Method for Evaluating the Ability of Exterior Vents to Resist the Entry of Embers and Direct Flame Impingement) E2912 (Standard Test Method for Fire Test of Non Mechanical Fire Dampers Used in Vented Construction) E2957 (Standard Test Method for Resistance to Wildfire Penetration of Eaves, Soffits and Other Projections) E2965 (Standard Test Method for Determination of Low Levels Heat Release Rate for Materials and Products Using an Oxygen Consumption Calorimeter) DTS (draft) (Reaction to fire tests Heat release, smoke production and mass loss rate Part 4: Measurement of heat release for determination of low levels of heat release) Page 14 of 129

15 ASTM NFPA UL ISO IEC ASTM D09 D3874 (Standard Test Method for Ignition of Materials by Hot Wire Sources) (Fire hazard testing Part 2 20: Glowing/hot wire based test methods Hot wire ignition test Apparatus, confirmatory test arrangement and guidance) D5424 (Standard Test Method for Smoke Obscuration of Insulating Materials Contained in Electrical or Optical Fiber Cables When Burning in a Vertical Cable Tray Configuration) *** 1685 (Vertical Tray Fire Propagation and Smoke Release Test for Electrical and Optical Fiber Cables) *** (Tests on electric cables under fire conditions Part 3: Test for vertical flame spread of vertically mounted bunched wires or cables) *** D5485 (Standard Test Method for Determining the Corrosive Effect of Combustion Products Using the Cone Corrosimeter) D5537(Standard Test Method for Heat Release, Flame Spread, Smoke Obscuration, and Mass Loss Testing of Insulating Materials Contained in Electrical or Optical Fiber Cables When Burning in a Vertical Cable Tray Configuration) *** 1685 (Vertical Tray Fire Propagation and Smoke Release Test for Electrical and Optical Fiber Cables) *** (Tests on electric cables under fire conditions Part 3: Test for vertical flame spread of vertically mounted bunched wires or cables) *** Page 15 of 129

16 ASTM NFPA UL ISO IEC ASTM D09 (Continued) D6113 (Standard Test Method for Using a Cone Calorimeter to Determine Fire Test Response Characteristics of Insulating Materials Contained in Electrical or Optical Fiber Cables) D6194 (Standard Test Method for Glow Wire Ignition of Materials) (Fire hazard testing Part 2 13: Glowing/hot wire based test methods Glow wire ignition temperature (GWIT) test method for materials) D635 (Standard Test Method for Rate of Burning and/or Extent and Time of Burning of Plastics in a Horizontal Position) D1929 (Standard Test Method for Determining Ignition Temperature of Plastics) ASTM D20 94 (HB portion) (Standard for Safety of Flammability of Plastic Materials for Parts in Devices and Appliances) 1210 (Plastics Determination of the burning behaviour of horizontal and vertical specimens in contact with a small flame ignition source 871 (Plastics Determination of ignition temperature using a hotair furnace) (Fire hazard testing Part 11 10: Test flames 50 W horizontal and vertical flame test methods) Page 16 of 129

17 ASTM NFPA UL ISO IEC ASTM D20 (Continued) D2843 (Standard Test Method for Density of Smoke from the Burning or Decomposition of Plastics) D2863 (Standard Test Method (Plastics for Measuring the Minimum Oxygen Concentration to Support Candle Like Combustion of Plastics (Oxygen Index)) Determination of burning behaviour by oxygen index Part 2: Ambienttemperature test) D3014 (Standard Test Method for Flame Height, Time of Burning, and Loss of Mass of Rigid Thermoset Cellular Plastics in a Vertical Position) D3675 (Standard Test Method for Surface Flammability of Flexible Cellular Materials Using a Radiant Heat Energy Source) D3801 (Standard Test Method for Measuring the Comparative Burning Characteristics of Solid Plastics in a Vertical Position) 94 (V portion) (Standard for Safety of Flammability of Plastic Materials for Parts in Devices and Appliances) 1210 (Plastics Determination of the burning behaviour of horizontal and vertical specimens in contact with a small flame ignition source (Fire hazard testing Part 11 10: Test flames 50 W horizontal and vertical flame test methods) Page 17 of 129

18 ASTM NFPA UL ISO IEC ASTM D20 (Continued) D4804 (Standard Test Method for Determining the Flammability Characteristics of Nonrigid Solid Plastics) 94 (VTM portion) (Standard for Safety of Flammability of Plastic Materials for Parts in Devices and Appliances) 9773 (Plastics Determination of burning behaviour of thin flexible vertical specimens in contact with a small flame ignition source) D4986 (Standard Test Method for Horizontal Burning Characteristics of Cellular Polymeric Materials) D5048 (Standard Test Method for Measuring the Comparative Burning Characteristics and Resistance to Burn Through of Solid Plastics Using a 125 mm Flame) D7309 (Standard Test Method for Determining Flammability Characteristics of Plastics and Other Solid Materials Using Microscale Combustion Calorimetry) D1230 (Standard Test Method for Flammability of Apparel Textiles) D4151 (Standard Test Method for Flammability of Blankets) 94 (HF portion) (Standard for Safety of Flammability of Plastic Materials for Parts in Devices and Appliances) 94 (5V portion) (Standard for Safety of Flammability of Plastic Materials for Parts in Devices and Appliances) 9772 (Cellular plastics Determination of horizontal burning characteristics of small specimens subjected to a small flame) 9773 (Plastics Determination of burning behaviour of thin flexible vertical specimens in contact with a small flame ignition source) ASTM D13 Page 18 of 129

19 ASTM NFPA UL ISO IEC ASTM D13 (Continued) D5238 (Standard Test Method for Smoldering Combustion Potential of Cotton Based Batting) D6413 (Standard Test Method for Flame Resistance of Textiles (Vertical Test)) D6545 (Standard Test Method for Flammability of Textiles Used in Children's Sleepwear) D7016 (Standard Test Method to Evaluate Edge Binding Components Used in Mattresses After Exposure to An Open Flame) D7140 (Standard Test Method to Measure Heat Transfer Through Textile Thermal Barrier Materials) Other NFPA Fire Test Standards 259 (Standard Test Method for Potential Heat of Building Materials) 262 (Standard Method of Test for Flame Travel and Smoke of Wires and Cables for Use in Air Handling Spaces) 910 (Standard Test for Flame Propagation and Smoke Density Values for Electrical and Optical Fiber Cables Used in Spaces Transporting Environmental Air, WD Page 19 of 129

20 ASTM NFPA UL ISO IEC Other NFPA Fire Test Standards (Continued) 268 (Standard Test Method for Determining Ignitability of Exterior Wall Assemblies Using a Radiant Heat Energy Source) 274 (Standard Test Method to Evaluate Fire Performance Characteristics of Pipe Insulation) 275 (Standard Method of Fire Tests for the Evaluation of Thermal Barriers) 276 (Standard Method of Fire Test for Determining the Heat Release Rate of Roofing Assemblies with Combustible Above Deck Roofing Components) 288 (Standard Methods of Fire Tests of Horizontal Fire Door Assemblies Installed in Horizontal Fire Resistance Rated Assemblies) Page 20 of 129

21 ASTM NFPA UL ISO IEC Other NFPA Fire Test Standards (Continued) 289 (Standard Method of Fire Test for Individual Fuel Packages) 290 (Standard for Fire Testing of Passive Protection Materials for Use on LP Gas Containers) 701 (Standard Methods of UL 214 (Standard for Tests Fire Tests for Flame Propagation of Textiles and Films) for Flame Propagation of Fabrics and Films ) WD 705 (Recommended Practice for a Field Flame Test for Textiles and Films) Other ISO Tests (Plastics Determination of burning behaviour by oxygen index Part 3: Elevatedtemperature test) * Test methods address the same property(ies) but are not technically equivalent, either through apparatus or procedure ** Test method ASTM E2257 and ISO 9705 are technically equivalent the others are similar in concept only *** Test methods ASTM D5424 and D5537 contain two test methods, technically equivalent to those in UL 1685 but different from that, although measuring similar properties to, the one in IEC Page 21 of 129

22 National Fire Protection Association Report of /26/2016 3:45 PM Organization: GBH International Street Address: City: State: Zip: Submittal Date: Thu Mar 24 19:58:58 EDT 2016 Committee Statement Committee Rejected but see related SR Action: Resolution: SR-1-NFPA 30A-2016 Statement: The exception has been replaced with new to comply with the manual of style. The reference to ASTM E648 was not added due to the possibility of conflicts with NFPA 253. Page 22 of 129

23 National Fire Protection Association Report of /26/2016 3:45 PM Public Comment No. 21-NFPA 30A-2016 [ Section No ] Gas Detection System. Repair garages used for repair of vehicle engine fuel systems fueled by non-odorized gases, such as hydrogen and non-odorized LNG/CNG, shall be provided with an approved flammable gas detection system. Gas detection systems in repair garages used for the maintenance of hydrogen vehicles shall be in accordance with NFPA System Design. The flammable gas detection system shall be calibrated to the types of fuels or gases used by vehicles to be repaired. The gas detection system shall be designed to activate when the level of flammable gas exceeds 25 percent of the lower flammable limit (LFL). Gas detection shall also be provided in lubrication or chassis repair pits of repair garages used for repairing non-odorized LNG/CNG-fueled vehicles Operation. Activation of the gas detection system shall result in all of the following: (1) Initiation of distinct audible and visual alarm signals in the repair garage (2) Deactivation of all heating systems located in the repair garage (3) Activation of the mechanical ventilation system, when the system is interlocked with gas detection Failure of the Gas Detection System. Failure of the gas detection system shall result in the deactivation of the heating system and activation of the mechanical ventilation system and, where the ventilation system is interlocked with gas detection, shall cause a trouble signal to sound in an approved location The circuits of the detection system required by shall be monitored for integrity in accordance with NFPA 72. Statement of Problem and Substantiation for Public Comment Recommendation of the Joint NFPA 2/30A Task Group Related Item Committee Input No. 25-NFPA 30A-2015 [Section No. 2.4] Committee Input No. 13-NFPA 30A-2015 [Section No ] Submitter Information Verification Submitter Full Name: Ronald Laurence Organization: Stantec Consulting Services, I Affilliation: NFPA 2/30A Task Group Street Address: City: Page 23 of 129

24 National Fire Protection Association Report 0 of 42 10/26/2016 3:45 PM State: Zip: Submittal Date: Fri May 13 15:35:55 EDT 2016 Committee Statement Committee Rejected but see related SR Action: Resolution: SR-6-NFPA 30A-2016 Statement: This revision clarifies where the requirements of NFPA 2 and NFPA 30A are applicable. Page 24 of 129

25 National Fire Protection Association Report 1 of 42 10/26/2016 3:45 PM Public Comment No. 22-NFPA 30A-2016 [ Section No ] Exhaust duct openings shall be located so that they effectively remove vapor accumulations at floor level from all parts of the floor area. Where LTA gaseous fuel vehicles are repaired, exhaust duct openings shall be located so that they effectively remove vapor accumulations at the ceiling level from all parts of the ceiling area. Statement of Problem and Substantiation for Public Comment Recommendation of the Joint NFPA 2/30A Task Group Related Item Committee Input No. 25-NFPA 30A-2015 [Section No. 2.4] Committee Input No. 13-NFPA 30A-2015 [Section No ] Submitter Information Verification Submitter Full Name: Ronald Laurence Organization: Stantec Consulting Services, I Affilliation: NFPA 2/30A Task Group Street Address: City: State: Zip: Submittal Date: Fri May 13 15:38:18 EDT 2016 Committee Statement Committee Rejected but see related SR Action: Resolution: SR-7-NFPA 30A-2016 Statement: This revision specifies the appropriate location for exhaust duct openings where lighter-than-air fuels are present. Page 25 of 129

26 National Fire Protection Association Report 2 of 42 10/26/2016 3:45 PM Public Comment No. 23-NFPA 30A-2016 [ Section No ] * Where major repairs are conducted on CNG LTA -fueled vehicles or LNG-fueled vehicles, open flame heaters or heating equipment with exposed surfaces having a temperature in excess of 399 C (750 F) shall not be permitted within 18 in. of the ceiling or in areas subject to ignitible concentrations of gas. Statement of Problem and Substantiation for Public Comment Recommendation of the Joint NFPA 2/30A Task Group Related Item Committee Input No. 25-NFPA 30A-2015 [Section No. 2.4] Committee Input No. 13-NFPA 30A-2015 [Section No ] Submitter Information Verification Submitter Full Name: Ronald Laurence Organization: Stantec Consulting Services, I Affilliation: NFPA 2/30A Task Group Street Address: City: State: Zip: Submittal Date: Fri May 13 15:40:27 EDT 2016 Committee Statement Committee Action: Rejected but see related SR Resolution: SR-8-NFPA 30A-2016 Statement: This revision is made so the requirement extends to hydrogen. Page 26 of 129

27 National Fire Protection Association Report 3 of 42 10/26/2016 3:45 PM Public Comment No. 25-NFPA 30A-2016 [ Section No ] * In major repair garages where CNG vehicles are repaired or stored, the area within 455 mm (18 in.) of the ceiling shall be designated a Class I, Division 2 hazardous (classified) location. Exception: In major repair garages, where ventilation equal to not less than four air changes per hour is provided, this requirement shall not apply. Statement of Problem and Substantiation for Public Comment The NGVAmerica Technology & Development Committee supports Public Input No. 25-NFPA 30A-2015 (Sect. No ), and the deletion of section The NGV industry has not had any recorded failures (premature releases) from thermally activated pressure relief devices (PRDs), mounted on hundreds of thousands of cylinders, since The existing requirement in section is based on the assumption for the credible release of natural gas from a vehicle being 150% of the largest on board cylinder (Annex A8.2.1). This assumption was based on failures of thermally activated PRDs, which the industry experienced in the 1990s. The suspect thermally activated PRDs have gone through many generations of new designs, and there have been no documented failures (premature releases) since The fact that the industry has not had any premature failures since 2002 is a result of the phase out of older designs and excellent performance record of the new thermally activated PRD designs. Sandia National Laboratory has conducted HAZOP and CFD analysis on indoor release of natural gas from LNG and CNG vehicles (SAND ). The HAZOP analysis shows that more likely releases from CNG and LNG vehicles are a much lower volume than the 150% case and pose less of a hazard for maintenance facilities. The Sandia report also includes CFD gas dispersion modeling of the smaller releases showing the extent of the ignitable mixture. The modeling of the more credible releases indicates that the ignitable mixture is only within an area no more than ten feet from the point of release. This limited volume of release also does not allow for an increased concentration in the ceiling area due to the dispersion of the gas. The full release that was noted in the 1990 assumption was also modeled for background information. The report indicates that ventilation rates within the facility have little or no impact on the volume of ignitable mixture and the dispersion of the released gas. Related Item Public Input No. 25-NFPA 30A-2015 [Section No ] Submitter Information Verification Submitter Full Name: Dan Bowerson Organization: NGVAmerica Street Address: City: State: Zip: Submittal Date: Sun May 15 12:36:22 EDT 2016 Committee Statement Page 27 of 129

28 National Fire Protection Association Report 4 of 42 10/26/2016 3:45 PM Committee Action: Resolution: Rejected The modeling performed does not address all credible release scenarios and is therefore too limited to make a change to the current requirements in the code. Page 28 of 129

29 National Fire Protection Association Report 15 of 42 10/26/2016 3:45 PM Public Comment No. 27-NFPA 30A-2016 [ Section No ] 8.2.1* In major repair garages where CNG vehicles are repaired or stored, the area within 455 mm (18 in.) of the ceiling shall be designated a Class I, Division 2 hazardous (classified) location. Exception: In major repair garages, where ventilation equal to not less than four air changes per hour is provided, this requirement shall not apply. Additional Proposed Changes File Name Description Approved SAND _CNG_PRD_Failure.pdf CVEFSANDIAFinalReportPhase1.pdf SAND : Background report to support the Rational of the proposed change. SAND : Background report to support the Rational of the proposed change. Statement of Problem and Substantiation for Public Comment Sandia National Labs supports Public Input No. 25-NFPA 30A-2015 (Sect. No ), and the deletion of section Rationale: The electrical classification area depth of 18 seems to have no rigorous justification. The most likely scenarios have been modeled and have shown flammable concentrations near the release point at the vehicle, well below 18 from the ceiling. Models of the much less likely but more catastrophic release of the entire contents of a CNG cylinder show that the flammable region could extend from the release point to the ceiling and produce a flammable layer thicker than 18. Modeling results comparing ventilation versus no ventilation show little reduction in the flammable volume. The ventilation exception therefore, is also not supported. To develop a comprehensive analysis into existing regulatory issues regarding NGV maintenance facility operations, the Clean Vehicle Education Foundation (CVEF) partnered with Sandia National Laboratories (SNL) to take advantage of Sandia s extensive experience performing similar analyses in support of hydrogen refueling infrastructure. A report on the findings from Phase I of these investigations can be found in Analyses in Support of Risk-Informed Natural Gas Vehicle Maintenance Facility Codes and Standards: Phase I by Ekoto, et al. (SAND ). A synopsis of relevant information is presented below. For the hazard analysis work, detailed Computational Fluid Dynamics (CFD) simulations were performed at Sandia to examine the three release scenarios identified from a hazard and operability study (HAZOP) analysis: (1) a dormant LNG blow-off, (2) indoor CNG fuel system purge downstream of the storage isolation valves, and (3) a full-scale CNG tank blow-down due to a failure of the pressure relief device (PRD). The analysis also included investigation of the impact of ventilation and the presence of beam pockets formed by solid roof support beams. Methane was used as a proxy for natural gas in the simulations. The reference NGV facility had dimensions of 100 feet long, 50 ft wide and 20 ft tall, with pitched roof. Geometries with and without solid, evenly spaced roof rafters were examined. The impact of active ventilation at the commonly prescribed rate of 5 air changes per hour versus a facility with passive ventilation was considered. For conditions with mechanical ventilation, air was forced into the enclosure 750 seconds before the start of the release to ensure internal steady flows. The vehicle was modeled as a cuboid ( ft3) and placed in the center of the NGV Page 29 of 129

30 SANDIA REPORT SAND Unlimited Release Printed Month and Year Analysis of a Full Scale Blowdown Due to a Mechanical Failure of a Pressure Relief Device in a Natural Gas Vehicle Maintenance Facility Myra Blaylock Radoslav Bozinoski Isaac Ekoto Prepared by Sandia National Laboratories Albuquerque, New Mexico and Livermore, California Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL Approved for public release; further dissemination unlimited. Page 30 of 129

31 Issued by Sandia National Laboratories, operated for the United States Department of Energy by Sandia Corporation. 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 Telephone: (865) Facsimile: (865) Online ordering: Available to the public from U.S. Department of Commerce National Technical Information Service 5301 Shawnee Rd Alexandria, VA Telephone: (800) Facsimile: (703) Online order: Page 31 of 129

32 SAND Unlimited Release Printed Month Year Analysis of a Full Scale Blowdown Due to a Mechanical Failure of a Pressure Relief Device in a Natural Gas Vehicle Maintenance Facility Myra Blaylock Radoslav Bozinoski Thermal/Fluids Science and Modeling Isaac Ekoto Sandia National Laboratories P.O. Box 696 Livermore, CA Abstract A computational fluid dynamics (CFD) analysis of a natural gas vehicle experiencing a mechanical failure of a pressure relief device on a full CNG cylinder was completed to determine the resulting amount and location of flammable gas. The resulting overpressure if it were to ignite was also calculated. This study completes what is discussed in Ekoto et al. [1] which covers other related leak scenarios. We are not determining whether or not this is a credible release, rather just showing the result of a possible worst case scenario. The Sandia National Laboratories computational tool Netflow was used to calculate the leak velocity and temperature. The in-house CFD code Fuego was used to determine the flow of the leak into the maintenance garage. A maximum flammable mass of 35 kg collected along the roof of the garage. This would result in an overpressure that could do considerable damage if it were to ignite at the time of this maximum volume. It is up to the code committees to decide whether this would be a credible leak, but if it were, there should be preventions to keep the flammable mass from igniting. Keywords: Natural Gas Vehicle Maintenance Facility, Pressure Relief Device Failure, CFD Page 32 of 129

33 ACKNOWLEDGMENTS The authors would like to thank Bill Houf (SNL-Retired) for his assistance with setting up and post-processing of the numerical simulations. We would also like to thank the Clean Vehicle Education Foundation and DOE s Vehicle Technologies Office Clean Cities program for funding this research. Page 33 of 129

34 CONTENTS 1. Introduction Method Mechanical Failure of a Thermally Activated PRD PRD Leak Calculation Garage Description Flow Solver Results Corrected Netflow Results CFD Results Overpressure calculations Conclusions and Future Work Recommendations References Appendix A: Time Laps of Leak Distribution FIGURES Figure 1. Schematic of the NGV maintenance facility used for the simulations. The roof had a 1:6 pitch and had layouts with and without 9 evenly spaced, horizontal supports. Two circulation vents were located on the smaller building side-walls (shown in yellow), with one placed low and the other high to maximize room currents Figure 2. Leak exit velocity for simulation with incorrect thermal conductivity, 3.5e-5 W/m-K, and correct thermal conductivity, 3.5e-2 W/m-k Figure 3. CNG tank temperature history comparison between incorrect and corrected thermal conductivity Figure 4. The CNG tank mass inventory remains fairly consistent with the correction of the thermal conductivity Figure 5. The ventilation was run for 720 seconds before the leak was started, so this image shows what happens within the first second of the leak. The boundary of flammable mass is shown in white. A plume has already reached between the leak location at the top of the vehicle and the ceiling of the garage Figure 6. The maximum flammable volume of natural gas in the garage is reached ~220 seconds after the start of the leak. At this point, most of the mass has left the tank. The gas dispersing out of the ventilation and throughout the garage starts to bring the concentration back below the flammable limit of 15%. The flammable mass is still in the plume above the leak and has spread to cover most of the ceiling. In places it is more than 80 inches thick Figure 7. As the tank empties, the flammable mass clears from the garage and has completely dispersed by 680 seconds Page 34 of 129

35 TABLES Table 1. Consequences of overpressures in an enclosed space [10] Page 35 of 129

36 NOMENCLATURE CFD CNG DOE EOS HAZOP LNG PRD SNL Computational Fluid Dynamics Compressed Natural Gas Department of Energy Equation of State Hazardous and Operability Study Liquid Natural Gas Pressure Relief Device Sandia National Laboratories Page 36 of 129

37 1. INTRODUCTION Sandia National Labs is using a leak calculation tool coupled with computational fluid dynamics (CFD) to model leak scenarios in a maintenance garage to help the codes and standards community understand the consequences and outcomes of these leaks. In this report we describe the worst case scenario of a complete venting of a CNG cylinder due to a faulty pressure relief device (PRD). This can then be compared to more likely but less severe cases. As previously reported in [1], three leak scenarios were identified by a HAZOP study to be simulated using computation fluid dynamics to gain a better understanding of the conditions in a maintenance garage during the leaks. The first two, a dormant LNG blow-off and a crack in a CNG or LNG fuel system line, went smoothly and the results in the previous report are complete. However, for the third scenario, a mechanical failure of a thermally activated pressure relief device (PRD), there were some problems. The boundary conditions for the leak went into a low temperature regime where there was a previously undiscovered error in the software. The first two scenarios did not reach this temperature, so there is no issue with the results. In re-examining the case in more detail, the error was found and corrected. We present the differences in the leak conditions after the correction as well as the new CFD results that incorporate the corrections. The maintenance garage is of typical size, has a pitched roof with cross beams, and ventilation system. This case involves the complete venting, or blowdown, of a cylinder that is 150% of the size of the maximum cylinders used on vehicles. We are not addressing the question of whether this is a likely scenario, and in fact with new safeguards there are reason to believe that it that it might not be, as discussed in SAND [1]. It is presented here for completeness of that report as a worst-case hazard used by code development committees. Page 37 of 129

38 2. METHOD 2.1 Mechanical Failure of a Thermally Activated PRD In the event a CNG cylinder becomes engulfed in a flame, the onboard storage cylinders are protected against excessive pressure buildup by a thermally triggered PRD, designed to fully open without the possibility of reseating in the event of activation. Accordingly, inadvertent actuation due to some mechanical failure would result in a rapid and uncontrollable decompression of all cylinder contents. Advances such as the use of dual activated valves have been implemented to reduce the likelihood of unintended release, although there remains some nominal risk. The Standards Development Organizations view such a release as a bounding event for hazard potential. For this scenario, the entire contents of a 700 L, fully pressurized (250 bar) CNG cylinder at room temperature (294 K) was released into the NGV maintenance facility. The specified release point was identical to the LNG blow-off scenario [1], which is located at the top of the vehicle. The PRD orifice diameter was set to 6.2 mm (0.24 ) based on the flow rate specifications of typical commercially available PRDs. At the start of the release, the valve was assumed to immediately transition to a fully open position and remain that way for the duration of the release. While keeping the mass flow rate consistent, the initial leak area was increased to10 cm 2 (1.55 in 2 ) due to the gridding constraints, and Netflow was used to model the transient blow-down. In the initial report [1] accurate results for this scenario were hampered by a mistake in the Netflow code (described in detail below). That mistake has been corrected for this report. 2.2 PRD Leak Calculation Netflow is a network flow simulator developed by Sandia. It was originally developed to model low speed airflows and contaminant transport in buildings. It has since been adapted to model high Mach number fully compressible transonic flows in piping networks [2]. A typographical error was found in the thermal conductivity value for methane in Netflow [2, 3]. At standard temperature and pressure methane has a thermal conductivity of 3.5e-2 W/m-k. The value used in Netflow was three orders of magnitude smaller which caused the heat transfer coefficient between the gas and the tank wall to be three orders of magnitude too small. The Netflow analysis was rerun with the corrected thermal conductivity using the ideal gas equation of state (EOS). Once a correct velocity and temperature profile of the leak was established, the entire leak scenario simulation was rerun. 2.3 Garage Description The maintenance garage was modeled as a pitched roof building (1:6 pitch) that was 30.5 m long (100 ), 15.2 m wide (50 ) and 6.1 m tall (20 ), with the roof peak located at the center and 127 cm (50 ) higher than the corresponding eaves (see schematic in Figure 1). Note that although the roof and main building are shown with different colors to emphasize the pitch, the enclosure was treated as a single volume. A roof layout with horizontally orientated support beams was investigated to see if the supports would cause the accumulation of flammable mixture in discrete pockets. Nine beams that were 15.2 cm wide (6 ) and 107 cm tall (42 ) were spaced 3.05 m apart (10 ) and ran parallel to the roof pitch. The garage contained two vents that were used for air circulation; one near the floor along one of the smaller building side-walls, with the second placed on the opposite side wall near the roof. Each vent was m tall (25 ) and 3.42 m wide (131 ). The NGV was modeled as a cuboid with a height and width of 2.44 m (8 ) and a length of 7.31 m (24 ). The vehicle was centered on the building floor with the major axis Page 38 of 129

39 aligned to the building minor axis. There was no fluid flow through the volume representing the vehicle. Figure 1. Schematic of the NGV maintenance facility used for the simulations. The roof had a 1:6 pitch and had layouts with and without 9 evenly spaced, horizontal supports. Two circulation vents were located on the smaller building side-walls (shown in yellow), with one placed low and the other high to maximize room currents. 2.4 Flow Solver To perform the analyses, a numerical modeling, previously validated for large-scale indoor hydrogen releases scenarios [4, 5], was adopted. The CFD solver, Fuego [6], was used to perform the natural gas release simulations from a representative NGV inside the scaled warehouse. Fuego is a Sandia National Laboratories developed code designed to simulate turbulent reacting flow and heat transfer [6] on massively parallel computers, with a primary focus on heat transfer to objects in pool fires. The code was adapted for compressible flow and combustion, and is well suited for low Mach number flows. The discretization scheme used in Fuego is based on the control volume finite element method [7], where the partial differential equations of mass, momentum, and energy are integrated over unstructured control volumes. The turbulence model was a standard two equation (k-ε) turbulence model [8] with transport equations solved for the mass fractions each chemical species, except for nitrogen which was modeled as the balance. For the calculations reported here, the first order upwind scheme was used for the convective terms. Note that methane was used as a proxy for natural gas in this simulation. The Fuego code solved the conservation equations in a time-dependent manner with both gravity and buoyancy effects accounted for. A slip wall boundary condition with a constant temperature (294 K) was used for all surfaces. The simulations were performed with mechanical ventilation with a uniform air flow velocity of 2.0 m/s (6.56 ft/s) which was forced through the floor vent into the enclosure, producing 5 air changes per hour (ACH) for the enclosure. The upper enclosure exhaust vent was assigned an open boundary condition with a total pressure of 1 atm and a temperature of 294 K. For all scenarios, the initial turbulence was negligible (k = 0.11 cm 2 /s, ε = cm 2 /s 3 ). For the mechanical ventilation, air was forced into the enclosure at the prescribed flow rate for 720 seconds prior to the start of the release to ensure the enclosure airflow was nominally steady. Page 39 of 129

40 3. RESULTS 3.1 Corrected Netflow Results To velocity and temperature of a leak caused by the failure of a PRD was calculated by simulating the entire contents of a 700 L, fully pressurized (250 bar) CNG cylinder at room temperature (294 K) being released into normal atmospheric conditions. The PRD orifice diameter was set to 6.2 mm (0.24 ) based on the flow rate specifications of typical commercially available PRDs. At the start of the release, the valve was assumed to immediately transition to a fully open position and remain that way for the duration of the release. The figure below shows a comparison of the exit velocity profiles for the original and new simulation with the corrected thermal conductivity. There is a small deviation between 80 and 300 seconds of the tank blowdown. Figure 2. Leak exit velocity for simulation with incorrect thermal conductivity, 3.5e-5 W/m-K, and correct thermal conductivity, 3.5e-2 W/m-k. While the velocity profile is relatively unchanged with the correction (Figure 2), a larger difference between the two simulations can be seen below in the tank temperature history (Figure 3). With the corrected thermal conductivity, the minimum temperature reached in the tank was approximately 240 K, while a value of 100 K was obtained in the original simulation. This higher, more consistent temperature is also within the valid temperature range of the software used for the CFD simulation, while the previous lower, incorrect temperature was not. This results in much more trustworthy results from the simulated leak scenario. Page 40 of 129

41 Figure 3. CNG tank temperature history comparison between incorrect and corrected thermal conductivity. Figure 4. The CNG tank mass inventory remains fairly consistent with the correction of the thermal conductivity. 3.3 CFD Results For this scenario, a plume of flammable gas formed between the leak location at the top of the vehicle and the ceiling of the garage. The region with flammable gas concentrations then spread outward across the ceiling and filled a region up to approximately 80 thick while the leak was occurring. It should be noted that during the initial phases of the blowdown, flow patterns allowed the flammable mass region to completely fill the space between the vehicle and the ceiling (see Figure 5 through Figure 7). As can be seen in Figure 2, the entire blowdown lasts approximately 10 minutes, and most of the mass has emptied the tank in less than 5 minutes (see Page 41 of 129

42 Figure 4). The flammable mass dissipates from the ceiling within 15 minutes of the start of the blowdown. A more complete time lapse of the leak is presented in Appendix A. Flammable volumes in the figures are in units of cm 3. Figure 5. The ventilation was run for 720 seconds before the leak was started, so this image shows what happens within the first second of the leak. The boundary of flammable mass is shown in white. A plume has already reached between the leak location at the top of the vehicle and the ceiling of the garage. Page 42 of 129

43 Figure 6. The maximum flammable volume of natural gas in the garage is reached ~220 seconds after the start of the leak. At this point, most of the mass has left the tank. The gas dispersing out of the ventilation and throughout the garage starts to bring the concentration back below the flammable limit of 15%. The flammable mass is still in the plume above the leak and has spread to cover most of the ceiling. In places it is more than 80 inches thick. Figure 7. As the tank empties, the flammable mass clears from the garage and has completely dispersed by 680 seconds. Page 43 of 129

44 4. OVERPRESSURE CALCULATIONS Using a simple calculation [9] that accounts for the maximum flammable mass in the building, we can estimate the overpressure that would result if the leak were to ignite at that point in time. pp = VV TT + VV NNNN VV γγ TT VV ssssssssssh (σσ 1) 1 VV TT VV TT Where VV TT is the volume of the facility, VV NNNN is the volume of flammable methane, VV ssssssssssh is the stoichiometric consumed methane volume, σ is the stoichiometric methane expansion ratio (7.561), and γ is the specific heat ratio of air (1.4). This overpressure correlation as developed only considers the sudden combustion of all flammable contents, which is unlikely to happen for a volume of flammable gas that is as large as seen in this case. The presence of ventilation, wall heat transfer, and the fact that the mixtures will continually lean out will mean that the actual overpressure will be much lower than is calculated. (If the enclosure was perfectly sealed and there was no heat transfer out of the box, then the Δp calculated would be the same, assuming the flammable volume stayed constant throughout the entire burn.) On the other hand, the flame front might become increasingly turbulent due to obstacles such as the beams, perturbing the flame-front making and making it even more turbulent, which would result in an increase in the turbulent flame speed. It is possible that the burn velocity could become fast enough that it could transition into a detonation, in which case the overpressures will be much greater. This is brought to the attention of the reader so that the assumptions in the calculation are clear, and it is known that the result should be taken as an estimate only. During this simulation, the maximum flammable volume of m 3 occurred at seconds from the start of the leak blowdown (942.5 seconds into the simulation). The volume of the garage is 3122 m 3, and the stoichiometric consumed methane volume is 590 m 3. These conditions are estimated to produce a change in pressure, or overpressure, of about 220 kpa. As stated above, as long as there is not enough turbulence to produce a detonation, this is most likely an overestimation of the actual overpressure that would occur for this scenario in this garage. According to [10], this is large enough to collapse unreinforced concrete walls (see Table 1). Even if the calculated overpressure were as much as 50% off, it would still have this same consequence. Table 1. Consequences of overpressures in an enclosed space [10]. Overpressure (kpa) Consequence 6.9 Injuries due to projected missiles 13.8 Fatality from projection against obstacles 13.8 Eardrum rupture Unreinforced concrete wall collapse Page 44 of 129

45 5. CONCLUSIONS AND FUTURE WORK With the corrected algorithms for the Netflow calculation of a blowdown of a CNG tank, we were able to successfully simulate the leak flow of a NGV inside of a garage using CFD. For this extremely large but unlikely event, the flammable mass region was able to completely fill the ceiling area of a typical garage. While this is a somewhat short lived event, with ventilation all of the flammable gas dissipates within 15 minutes, the possible consequences if there is an ignition source in contact with the flammable mass region are quite severe. The maximum overpressure, which would occur around 200 seconds into the blowdown, would be most likely be large enough to collapse unreinforced concrete walls. In the Netflow analysis of the storage tank, the ideal gas equation of state was used to model the thermodynamic properties of the fuel. The main assumption here is that the compressibility effects are small. To help qualify this assumption, REFPROP [11] real gas equations were used to calculate the thermodynamic properties in the storage tank during a transient blowdown. The result of this analysis was a higher predicted mass flow rate. However, for the full CNG tank Feugo analysis, a severe overpressure was already predicted when using the ideal gas equation and using the higher mass flow rates will not change this outcome. The use of the real versus ideal gas equations should, however, be studied further for future transient analysis. We should also note that all the simulations performed used Methane as a proxy for the Natural Gas mixture that is found in these types of systems. It is currently unknown how much of an effect using a proxy for the fuel has on the mass flow rates and is worth investigating in the future. Page 45 of 129

46 6. RECOMMENDATIONS We recommend that any codes and standards committees using this report as a reference consider whether or not they think that this is a credible release (none have occurred since 2002) which needs to be considered when updating fire code restrictions. If it is considered to be a credible release, it is clear that sever results could occur if this large of a release were to ignite, so precautions to prevent that ignition should be put in place. If not, then the modeling done for the Phase I report [1] shows that the most likely release scenarios do not result in hazardous concentrations in the beam pockets, within 18 inches of the ceiling, with or without ventilation. Page 46 of 129

47 7. REFERENCES 1. Ekoto, I.W., et al., Analyses in Support of Risk-Informed Natural Gas Vehicle Maintenance Facility Codes and Standards: Phase I SAND (SAND ). 2. Bozinoski, R. and W.S. Winters, Netflow Theory Manual. 2016, Sandia National Laboratories. 3. Winters, W.S., A new approach to modeling fluid/gas flows in networks. 2001: Sandia National Laboratories. 4. Houf, W.G., et al. Simulation of Hydrogen Releases from Fuel-Cell Vehicles in Tunnels. in World Hydrogen Energy Conf Essen, Germany. 5. Houf, W.G., et al., Hydrogen fuel-cell forklift vehicle releases in enclosed spaces. International Journal of Hydrogen Energy, (19): p Moen, C.D., et al. A Multi-Mechanics Approach to Computational Heat Transfer. in ASME Int. Mech. Eng. Cong. and Exhibition New Orleans. 7. Minkowycz, W.J., et al., Elliptic systems: finite element method 1, handbook of numerical heat transfer. 1988, New York: J. Wiley and Sons, Inc. 8. Papageorgakis, G.C. and D.N. Assanis, Comparison of linear and nonlinear RNG-based k-epsilon models for incompressible turbulent flows. Numerical Heat Transfer Part B- Fundamentals, (1): p Bauwens, C., J. Chaffee, and S. Dorofeev, Effect of ignition location, vent size, and obstacles on vented explosion overpressures in propane-air mixtures. Combust Sci Technol, : p Jeffries, R.M., S.J. Hunt, and L. Gould, Derivation of Fatality of Probability Function for Occupant Buildings Subject to Blast Loads. Health & Safety Executive, Lemmon, E.W., M.L. Huber, and M.O. McLinden, NIST Reference Fluid Thermodynamic and Transport Properties - REFPROP Version , National Institute of Standards and Technology, Standard Reference Data Program: Gaithersburg. Page 47 of 129

48 APPENDIX A: TIME LAPS OF LEAK Page 48 of 129

49 Page 49 of 129

50 Page 50 of 129

51 DISTRIBUTION 1 Douglas B. Horne DBHORNE LLC Clean Vehicle Education Foundation 6011 Fords Lake Ct. Acworth, GA (electronic copy) 1 Dan Bowerson NGVAmerica 400 North Capitol St NW Washington, DC (electronic copy) 1 MS9957 Amanda Dodd 1914 (electronic copy) 1 MS9957 Myra L. Blaylock 8253 (electronic copy) 1 MS9957 Radoslav Bozinoski 8253 (electronic copy) 1 MS9957 Patricia Gharagozloo 8253 (electronic copy) 1 MS9052 Isaac W. Ekoto 8367 (electronic copy) 1 MS0899 Technical Library 9536 (electronic copy) Page 51 of 129

52 Page 52 of 129

53 SANDIA REPORT SAND Unlimited Release March 2014 Analyses in Support of Risk-Informed Natural Gas Vehicle Maintenance Facility Codes and Standards: Phase I Isaac W. Ekoto 1, Myra L. Blaylock 1, Christine A. LaFleur 1, Jeffery L. LaChance 1, Douglas B. Horne 2 1 Sandia National Laboratories 2 Clean Vehicle Education Foundation Prepared by Sandia National Laboratories Albuquerque, New Mexico and Livermore, California Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL Approved for public release; further dissemination unlimited. Page 53 of 129

54 Issued by Sandia National Laboratories, operated for the United States Department of Energy by Sandia Corporation. 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 Telephone: (865) Facsimile: (865) Online ordering: Available to the public from U.S. Department of Commerce National Technical Information Service 5285 Port Royal Rd. Springfield, VA Telephone: (800) Facsimile: (703) Online order: 2 Page 54 of 129

55 SAND Unlimited Release March 2014 Analyses in Support of Risk-Informed Natural Gas Vehicle Maintenance Facility Codes and Standards: Phase I Isaac W. Ekoto Hydrogen and Combustion Technologies Sandia National Laboratories 7011 East Avenue, MS 9052 Livermore, CA Myra L. Blaylock Thermal/Fluid Science & Engineering Sandia National Laboratories 7011 East Avenue, MS 9957 Livermore, CA A. Christine LaFleur and Jeffery L. LaChance Risk and Reliability Analysis Sandia National Laboratories P.O. Box 5800, MS 0748 Albuquerque, NM Douglas B. Horne DBHORNE LLC Clean Vehicle Education Foundation 6011 Fords Lake Ct. Acworth, GA Page 55 of 129

56 Abstract Safety standards development for maintenance facilities of liquid and compressed gas fueled large-scale vehicles is required to ensure proper facility design and operation envelopes. Standard development organizations are utilizing risk-informed concepts to develop natural gas vehicle (NGV) codes and standards so that maintenance facilities meet acceptable risk levels. The present report summarizes Phase I work for existing NGV repair facility code requirements and highlights inconsistencies that need quantitative analysis into their effectiveness. A Hazardous and Operability study was performed to identify key scenarios of interest. Finally, scenario analyses were performed using detailed simulations and modeling to estimate the overpressure hazards from HAZOP defined scenarios. The results from Phase I will be used to identify significant risk contributors at NGV maintenance facilities, and are expected to form the basis for follow-on quantitative risk analysis work to address specific code requirements and identify effective accident prevention and mitigation strategies. 4 Page 56 of 129

57 Acknowledgments The authors gratefully acknowledge Bill Houf (SNL Retired) for his assistance with the set-up and post-processing of the numerical simulations. 5 Page 57 of 129

58 6 Page 58 of 129

59 Contents 1. Introduction Historical Code Development Process Proposed Code Development Research Objectives and Scope Existing Code Requirements Ventilation Pit Ventilation Gas Detection Ignition Sources Electrical Classification Preparing a Vehicle for Repair Maintenance and Decommissioning of Vehicle Fuel Containers Conventional NGV Repair Facility HAZOP Background Assumptions HAZOP Methodology HAZOP Results Scenario Analysis Maintenance Garage Simulation Boundary Conditions Dormant LNG Blow-off Scenario CNG and LNG Fuel System Line Cracking Mechanical Failure of a Thermally Activated PRD CFD Scenario Results Dormant LNG Blow-off Scenario Results CNG and LNG Fuel System Line Cracking Results Full-Scale Tank Blow-Down due to a Mechanical Failure of the PRD Summary and Conclusions 35 References 37 Appendix A: HAZOP Data Sheets 41 Appendix B: Supplemental CFD Simulation Data 45 Distribution 49 7 Page 59 of 129

60 Figures Figure 1: Typical large-duty LNG vehicle fuel system schematic major components highlighted. Note that system isolation and overpressure protection is after the heat exchanger so that all fuel system natural gas is gas phase Figure 2: Typical large-duty CNG vehicle fuel system schematic with most major components. Note the fail-close solenoid valves on all storage tanks isolate natural gas from the fuel system during regular maintenance Figure 3: Schematic of the NGV maintenance facility used for the simulations. The roof had a 1:6 pitch and had layouts with and without 9 evenly spaced, horizontal supports. Two circulation vents were located on the smaller building side-walls, with one placed low and the other high to maximize room currents Figure 4: NGV Maintenance facility natural gas mole fraction contours at 10, 60, and 306 seconds into the release for the facility layouts without (top) and with (bottom) roof supports for the LNG blow-off scenario. Velocity maps are also shown along the facility centerline to illustrate the impact of room currents on flow dispersion Figure 5: Time-history of the total natural gas flammable mass and volume for the LNG blow-off scenario. Note that the simulation for the scenario without ventilation in the facility without support beams was terminated 100 seconds into the release once steady flammable concentrations had been firmly established Figure 6: Grid convergence test that used the coarse (195,000 nodes) and fine (2.5 million nodes) grids for the LNG blow-off scenario with roof rafters to ensure repeatable results Figure 7: Mass flow rate time-history plot for the CNG line cracking scenario calculated from NETFLOW Figure 8: Maintenance facility natural gas LFL iso-contours at 2.5 (top), 10 (center), and 30.0 (bottom) seconds into the release for the layouts without roof supports for the CNG line cracking scenario Figure 9: Time-histories of total natural gas flammable mass and volume for the CNG blowdown scenario Figure 10: Mass flow rate time-history for the CNG tank blow-down scenario calculated from NETFLOW for a 700 liter tank pressurized with natural gas to 248 bar and released through an a 6.2 mm diameter orifice. Note that the tank volume was 50% greater than normal to simulate a worst case scenario Figure 11: Maintenance facility natural gas LFL iso-contours for the CNG tank blow-down scenario from a 700 liter tank pressurized to 248 bar for the facility layout with active ventilation, roof support beams, and a vertical release into the enclosure. Time histories of flammable mass and volume are also included Page 60 of 129

61 Figure 12: NGV Maintenance facility natural gas mole fraction contours at 10, 60, and 306 seconds into the release for the facility layouts without roof supports for the LNG blow-off scenario Figure 13: NGV Maintenance facility natural gas mole fraction contours at 10, 60, and 306 seconds into the release for the facility layouts with roof supports for the LNG blow-off scenario Figure 14: Maintenance facility natural gas mole fraction contours at 2.5 (top) and 30.5 (bottom) seconds into the release for the layouts without roof supports for the CNG line cracking scenario Tables Table 1: Typical service and maintenance activities Table 2: Operation States of CNG and LNG-Fueled Vehicles Table 3: HAZOP Results selected for modeling analyses Page 61 of 129

62 Nomenclature ACH AHJ API BLEVE CDF CNG CVEF DOE EIGA HAZOP IBC ICC IFC IMC ISO LFL LNG NFPA NGV NIST OEM PRD PRV QRA SDO SNL Air Changes per Hour Authority Having Jurisdiction American Petroleum Institute Boiling Liquid Expanding Vapor Explosion Cumulative Distribution Function Compressed Natural Gas Clean Vehicle Energy Foundation Department of Energy European Industrial Gas Association Hazardous and Operability Study International Building Code International Code Council International Fire Code International Mechanical Code International Standards Organization Lower Flammability Limit Liquid Natural Gas National Fire Protection Agency Natural Gas Vehicle National Institute of Science and Technology Original Equipment Manufacturer Pressure Relief Device Pressure Relieve Valve Quantitative Risk Assessment Standards Development Organization Sandia National Laboratories 10 Page 62 of 129

63 1. Introduction The growth of natural gas vehicle (NGV) fleets in recent years, especially for interstate commerce, has increased the need for additional gaseous fuel friendly maintenance facilities across the country. The NGV industry has largely focused its efforts on development of vehicles and fueling infrastructure, while issues with maintenance facility design and operation have been left to fleet owners. Sometimes in conjunction with paid consultants, fleet owners have had to use their own internal staff to interpret the intent of applicable codes to develop a facility design for liquefied natural gas (LNG) and/or compressed natural gas (CNG) applications that will be approved by the authority having jurisdiction (AHJ). The process can be difficult since the codes allow performance-based designs but provide little actual design guidance and have requirements that necessitate expert evaluation of expected hazardous conditions. Guidance that provides a better understanding of the code committee intent when the language was drafted is needed in order to apply those requirements across a diverse (e.g., ceiling height, layout, roof construction, heating, ventilation electrical) suite of maintenance facilities Historical Code Development Process Relevant codes for NGV maintenance facility operations have been developed over a number of years beginning in the late 1990s after a series of unintended releases from first generation pressure relief devices (PRDs) installed on CNG storage cylinders. The codes were initially written as prescriptive requirements and are now moving towards performance documents with the requirements based on assumed hazards determined from the cumulative expert knowledge and field experience of standards development organization (SDO) code committee members. Code requirements for CNG and LNG vehicles have key distinctions based on historical user experience with the respective technologies. The initial wave of PRD failures were either the result of models improperly selected for the design working pressure or some sort of design flaw. As a result of these incidents, the basic hazard for CNG systems was identified as the unintended release and subsequent ignition of natural gas while the vehicle is in the repair garage. The code committees assumed that the reasonable release amount was 150% of the total contents from the largest cylinder on the vehicle, with the extra 50% considered to be a safety factor. Since CNG cylinder PRDs are designed to only relieve during a fire, and not due to spurious in-cylinder pressure increases, PRD design standards were quickly revised. Since then, PRDs have performed as expected to protect the cylinder during a fire with few recorded failures; however, fire protection codes have not revised the assumed release amount based on the largest cylinder size. Such a requirement could promote unintended consequences, such as the use of a larger number of smaller cylinders, which can paradoxically increase overall system risk due to the increased number of failure points. The quantification of the hazard level for CNG vehicles is part of an ongoing study and will be submitted to the relevant code committees for reconsideration of existing requirements. It should be noted that the Clean Vehicle Education Foundation (CVEF) is currently investigating a series of PRD releases from imported cylinder valves rated at 260 bar that include both thermal and rupture disc PRDs. Since the rupture disc PRDs and valves are not suitable for the US 300 bar working pressures, CVEF has prepared a safety bulletin to stop the use of these valves [1]. 11 Page 63 of 129

64 For LNG vehicles, existing codes do not define a specific release scenario but instead assume two release types. The basic hazard is the possible ignition of gas released from the LNG tank relief valve due to pressure building as the contents warm over a period of time. Vacuum insulated LNG tanks are designed to have a hold time of up to several days before the pressure builds to the relief setting. Typically the LNG tank pressure would build at a rate of about 103 kpa (15 psig) per day giving a hold-time of about seven days, which is a normal operating parameter of LNG tanks. There are operating procedures that can greatly reduce the probability of a LNG tank PRD release during planned maintenance/repair operations, such as operating the vehicle to reduce the pressure in the tank, and monitoring the pressure and rate of pressure rise in the tank before entering the repair garage. The codes also have requirements that address possible liquid-phase LNG spills in the maintenance facilities that can subsequently flash-boil; however, there are no reported incidents within the historical records Proposed Code Development Research To develop a comprehensive analysis into existing regulatory issues regarding NGV maintenance facility operations discussed in greater detail in the following section the CVEF has partnered with Sandia National Laboratories (SNL) to take advantage of Sandia s extensive experience performing similar analyses in support of hydrogen refueling infrastructure [2]. The collective expertise in code interpretation, CFD modeling, sensitivity studies, hazard analysis, NGV fuel systems and facility operations are leveraged to develop guidelines for modification and construction of maintenance facilities. The scope of work has been split into two phases. The current report discusses the results and conclusions from Phase I, which involves a detailed survey of existing codes and regulations and quantification of the risk to personnel and property from any credible hazards. Phase II will be a follow-on study where the understanding generated in Phase I is leveraged to develop best practices to mitigate the identified hazards and design guidance based on facility configurations, along with a proposal for recommended changes to existing fire protection codes. Note that much of the existing code language was developed from rule of thumb based on user experience, without risk-informed analysis of potential hazards as recommended by the Fire Protection Research Foundation [3]. A risk-informed process, or quantitative risk assessment (QRA), leverages insights obtained from qualitative hazardous and operability study (HAZOP) combined with more quantitative metrics to establish code requirements. For NGV maintenance facility operations these metrics include the results of deterministic analyses for select accident scenarios, leakage frequency events, and safety margins to account for uncertainties. The QRAs enable identification of high-risk scenarios for NGV maintenance facility operations along with the dominant causal factors. Furthermore, QRAs can be used to evaluate the effectiveness of accident prevention and mitigation strategies so that risk can be reduced to acceptable levels. The impact of physical or engineered mitigation solutions for specific hazards must be balanced against procedural techniques that, while cheaper and easier to implement, also introduce the additional risk of human error [4]. 12 Page 64 of 129

65 1.3. Objectives and Scope The Phase I work described in the current report has been separated into two activities: (1) A HAZOP based on SDO expert advice was developed, which included a comprehensive review of NGV onboard fuel system components and an analysis of recorded historical incidents; and (2) Leverage Sandia s validated computational modeling capabilities [5, 6] to evaluate credible release scenarios based on the HAZOP analysis. Although the justification is laid out later in the text, scenario details are summarized below: A fully fueled LNG vehicle assumed to be left dormant in a NGV maintenance facility for a duration that exceeds the onboard storage hold time (~7 days). The resulting pressure buildup causes the pressure relief valve (PRV) to relieve, which leads to a controlled release of cool gas phase natural gas (~160 K) through a vertically orientated vent stack until the tank pressure falls below to the PRV seat pressure. Pressurized residual natural gas downstream of the system isolation and heat exchanger of an LNG vehicle is released into the facility when the fuel system is purged by a maintenance technician. Pressurized residual natural gas downstream of the system isolation of a CNG vehicle is released into the facility when the fuel system is purged by a maintenance technician. This scenario is identical to the previous scenario except that the fuel systems for CNG systems have roughly double the volume and pressures that are roughly an order of magnitude larger. A fourth scenario was also performed where the entire contents of a 700 L, fully pressurized (250 bar) CNG cylinder were released into the NGV maintenance facility due to the activation of a thermally triggered PRD. New safeguards such as the use of dual activated PRD valves that use parallel but independently activated PRDs, should make inadvertent activation unlikely. Nonetheless, the recent unintended PRD releases described earlier highlight the possibility for human error, and accordingly this event is deemed to be a basic worst-case hazard used by code development committees. The present report summarizes existing code requirements for NGV repair facilities to highlight inconsistencies from competing codes and identify code requirements that need quantitative analysis into their effectiveness. The HAZOP analysis is summarized in Section 3 and is expected to form the basis for follow-on QRA work on specific code requirements highlighted in Section 2 Scenario analysis based on the computational modeling results are discussed in Section 4. It should be noted that the consequence analysis does not extend beyond the scenarios described above and that without the follow-on QRA work there is no way to establish whether these are the most impactful possible scenarios. Finally, a summary of all results along with conclusions based on the data are given in Section 5. These results are meant to inform SDOs on the technical requirements for safe repair shop facility and design, with the hope for improved code harmonization and the implementation of scientifically defensible codes and standards. 13 Page 65 of 129

66 14 Page 66 of 129

67 2. Existing Code Requirements Existing code requirements have been thoroughly documented by CVEF [7], and are summarized here. The dominant US and international codes that cover vehicle maintenance facilities are the International Code Council s Fire (IFC), Mechanical (IMC), and Building (IBC) codes [8-10], along with NFPA codes 30A, 52, and 88A [11-13]. It is important to note that these codes are voluntarily adopted by states on a case-by-case basis and enforced by the local Authority Having Jurisdiction (AHJ). Since the local AHJ has the ability to enforce additional requirements beyond the national codes, they should be consulted early as part of the initial evaluation. The codes discussed below only apply to major repair facilities, with both NFPA 30A and the IFC exempting minor repair facilities from all code requirements specific to CNG and LNG. IFC exempts garages that do not work on the vehicle fuel system or use open flames (i.e., welding) from all additional requirements. NFPA 30A exempts garages that do not perform engine overhauls, painting, body and fender work, and any repairs requiring draining vehicle fuel tanks from all additional requirements. The maintenance work that can be done without any modifications to the facility include lubrication, inspection, engine tune-ups, replacement of parts, fluid changes, brake system repairs, tire rotation, and similar routine maintenance. When a maintenance facility considers adding NGVs to their operations, an analysis of maintenance tasks by type as a percentage of the overall activities should be performed, which can help determine if the facility could be divided into major and minor repair areas. With proper physical separation, the codes require only that those facility areas designated as major repair areas to be subject to the additional NGV requirements Ventilation Table of the IMC [9] requires all vehicle repair garages, regardless of fuel type or maintenance performed, to have a ventilation rate of 229 lpm per square meter of floor area (0.75 cfm/ft 2 ). However, NFPA 88A has a more stringent requirement of 305 lpm/m 2 (1.0 cfm/ft 2 ) for enclosed parking garages that house liquid and gaseous-fueled vehicles that should be considered as the base rate since even for minor repair garages since vehicles could be parked while awaiting repair. Where mechanical ventilation is required by IFC , it must operate continuously except when it is either interlocked with a gas detection system for LNG or electrically interlocked with the lighting circuit for CNG applications. There is a discrepancy between NFPA 30A and IFC in that NFPA 30A only requires ventilation for fuel dispensing areas within the maintenance facility. However, IFC uses similar language for CNG repair facilities that assumes indoor fueling will always be part of the repair facility there is a requirement that the system shall shut down the fueling system if the ventilation fails. The mechanical ventilation requirement is 5 air changes per hour (ACH), with two exceptions: (1) work is not to exchange parts and the maintenance requires no open flame or welding and (2) repair garages with AHJ approved natural ventilation. Clarification is needed on these IFC requirements note that NFPA separates indoor dispensing from repair facility 15 Page 67 of 129

68 requirements. With regards to LNG fuels NFPA 30A has an additional requirement that repair garages have a gas detection system interlocked to the mechanical ventilation Pit Ventilation Ventilation requirements for pits, below grade, and subfloor work areas are part of the basic requirements for liquid fuels where flammable vapors may accumulate. For existing facilities, this requirement should already be met. However, the IFC requires ventilation flow rates of 457 lpm/m 2 (1.5 cfm/ft 2 ) while NFPA 30A requires 305 lpm/m 2 (1.0 cfm/ft 2 ), with neither code containing specific requirements to CNG or LNG. Until the codes are harmonized, the local AHJ must specify the applicable rate for each facility. While experience has shown that there is a very low probability of a release of LNG liquid, the cold vapor release may initially be heavier than air and persist in a subgrade area before eventually warming up and rising due to buoyancy. The existing ventilation requirement for liquid fuels should be adequate for the addition of LNG to major repair facilities with approval of the local AHJ. Note that pit requirements were not considered for the present analysis, but the potential for accumulation of cool LNG within a pit is something that should be considered for future work Gas Detection There is no requirement for gas detection in either major or minor repair garages where odorized CNG vehicles are maintained. However, both IFC and NFPA 30A require approved gas detection systems for major repair garages servicing LNG vehicles. Specific requirements under these codes for gas detection installation and operation are similar and may require the expertise of a gas detection design engineer for optimal performance Ignition Sources The IFC does not have any specific requirements for CNG and LNG repair garages with respect to ignition sources although for liquid fuels IFC does require that ignition sources be restricted from the space within 0.46 m (18 ) of the floor. The liquid fuel ignition source requirement is likewise the standard requirement in the IBC, IMC and NFPA 70. Nonetheless, it is doubtful these requirements should likewise be applicable to CNG/LNG due to the differences in dispersion characteristics. In NFPA 30A, the restrictions on heating equipment in major repair garages only apply to areas where ignitable mixtures may be present. At the moment, the only way to quantify where these flammable mixtures exist is to perform computational fluid dynamic (CFD) modeling of credible CNG and LNG releases requires within representative facility geometries. There is a need to develop and validate reduced order methods that are expedient and accessible to a wide range of users, but still provide a sufficient level of accuracy Electrical Classification While the IFC does not have specific requirements for electrical classifications of NGV repair garages, NFPA 30A Chapter 8 does include electrical classification area requirements for liquid fuel vehicles for pits and the space within 0.46 m (18 ) from the repair garage floor. At the moment, there is a requirement for major CNG vehicle repair garages in NFPA 30A that 16 Page 68 of 129

69 classifies the area 0.46 m (18 ) from the ceiling as Class 1, Division 2 unless the area below the ceiling has ventilation of at least 4 ACH. While NFPA 30A is silent on classified areas for LNG in major repair garages, in practice LNG would generally be subject to the same requirements as liquid fuels in pits and as CNG in the 0.46 m (18 ) space below the ceiling. When considering what constitutes a credible release, it was noted earlier that existing CNG code requirements were based on the release of 150% of the contents of the largest cylinder in the repair facility in response to a series of PRD failures in the 1990s. The PRDs have been through several design revisions since then and the last few cases of premature release were over ten years ago. A proposal has been submitted by CVEF to review these requirements in NFPA 30A based on a QRA that considers the likelihood of different CNG/LNG releases and the configuration of representative maintenance facilities Preparing a Vehicle for Repair The only code requirement that addresses mitigation of the assumed hazards from releases of natural gas is IFC by: (1) Valve closures prior to maintenance to isolate CNG cylinders and LNG tanks from the fuel system balance to limit the potential fuel quantity that could be released due to damage or error during maintenance operations. (2) Operating the NGV until it stalls due to low fuel pressure in the system to further reduce the possible release volume. (3) Require the NGV fuel system be leakage tested by appropriate methods if there is a concern that the fuel system has experienced any damage. If damage is suspected the vehicle may need to be de-fueled prior to any maintenance Maintenance and Decommissioning of Vehicle Fuel Containers Code requirements for vehicle fuel containers are part of the maintenance requirements for vehicle mounted fuel storage containers; hence, NFPA 52 [12] should be consulted for specific requirements. Note that the latest edition (2013) incorporates several critical safety related changes for CNG cylinder maintenance based on lessons learned from incidents during maintenance operations. Also CVEF has published the document Safety Advice for Defueling CNG Vehicles and Decommissioning and Disposal of CNG Cylinders [14]These include requirements that repair facilities create specific written procedures for inspection and decommissioning of CNG cylinders and incorporate approved defueling capabilities; although no specific requirements for maintaining or venting LNG fuel tanks are given, these are considered best practice. Modifications to the maintenance facility are needed to accommodate fuel container defueling or fuel system maintenance and end of life decommissioning of CNG cylinders. 17 Page 69 of 129

70 18 Page 70 of 129

71 3. Conventional NGV Repair Facility HAZOP The HAZOP purpose study is to identify and characterize potential hazards through a structured and systematic examination of a specific system [15, 16]. For the current study, however, the HAZOP was performed on the operational activities that take place for a heavy-duty NGV maintenance facility. A detailed analysis of generic, system components was performed to identify hazards that could be encountered in a representative facility. The HAZOP focused on failures that were the result of an unexpected or uncontrolled release of natural gas (liquid or gaseous phase), with specific hazards identified in order to characterize the associated consequences. The ultimate goal is to leverage these findings to develop industry best practices and propose improvements to existing codes. Other hazards associated with heavy-duty vehicle maintenance activities (e.g., mechanical, electrical, ergonomic, and noise) were not considered as these hazards are not unique to NGV maintenance activities. HAZOP studies are usually performed on discrete industrial processes, with defined inputs and outputs from each process step or system component. Hazard scenarios are developed using a system of guidewords indicating relevant deviations from system design intents. For the present HAZOP to be most useful, an application-specific method was used that combined aspects of a failure mode and effects analysis with a HAZOP study, which is described further in this section. Spreadsheets that contain all identified hazard scenarios are included in Appendix A Background Assumptions Table 1 identifies typical activities associated with the NGV maintenance, which were used to categorize the operations into Operation States based on where they are typically conducted (Indoor or Outdoor) and the fuel system state during the maintenance activities (see Table 2). Operation State 3 (Dead vehicle storage) could occur either indoors or outdoors, so this operation state was broken up into 3in and 3out. Operation States 6 and 7 are differentiated based on the fuel system state. Operation State 6 represents fuel system services that require the entire fuel system to be evacuated and rendered inert (e.g., replacement of the solenoid valve on a CNG cylinder), while Operation State 7 is characterized by repair activities that can be performed with the isolation valve closed between the bulk tanks and the remainder of the fuel system. Table 1: Typical service and maintenance activities Service Maintenance and Repair Activities Inspection of fuel storage and delivery piping, components (including PRD) Inspection of fuel safety systems Troubleshoot/ Testing Exchange filters Drain and replace fluids (non-fuel system) Replace non fuel system component (brakes, tires, transmission, etc.) Repair leaking fuel system Replace fuel system components (e.g., tank, PRD, valve, plug, pressure gauge, economizer, fuel gauge coaxial cable) Leak Testing 19 Page 71 of 129

72 3.2. HAZOP Methodology The HAZOP procedure involved an examination of each system component and identification of scenarios, conditions or failure modes that could lead to a release of natural gas. Typical largeduty LNG and CNG vehicle fuel systems that were analyzed are respectively depicted in Figure 1 and Figure 2. For each scenario identified, the component targeted as the source of the release is recorded in the Component column of the HAZOP datasheets by referring to the system and component number used in these schematic diagrams. For example, releases of LNG from the storage tank are labeled LNG-4 and releases associated with the CNG manifold are labeled CNG-5. Additionally, the relevant Operation States when the Hazard Scenario is applicable are indicated in the datasheets as well, indicated by the Operation State number from Table 2. The relevant Operation States assigned to each Hazard Scenario were based on the state of the fuel system. If no natural gas is expected to be in the manifold (CNG-5) because the isolation valve (CNG-4) is expected to be closed, then a release from the manifold is not deemed feasible for this analysis. Situations where a release is possible due to human error or failure to close the isolation valve are dealt with in the Hazard Scenarios associated with the isolation valve itself and in Hazard Scenario 37. Table 2: Operation States of CNG and LNG-Fueled Vehicles Outdoor Indoor Preparation for Service Service Operation State 1 Defueling 2 Cracking of fuel system (FMM only) 3out 3in 4 Dead vehicle storage Dead vehicle storage Engine operation/idling (during testing, fuel run down, inspection and troubleshooting activities) 5 Service on non-fuel systems 6 Service on fuel system [Group 1] 7 Service on fuel system [Group 2] Fuel System State Entire fuel system (FMM and tanks) being evacuated Tank valve off, FMM being evacuated Fuel system charged but idle, key-off Fuel system charged but idle, key-off Key-on operation Tanks valve off, FMM evacuated (Run Down) Entire fuel system evacuated Tanks valve off, FMM Run Down then cracked Restart 8 Fuel line refilling, connection of a small pony tank OR valve opening followed by restart Fuel system recharging 20 Page 72 of 129

73 Figure 1: Typical large-duty LNG vehicle fuel system schematic major components highlighted. Note that system isolation and overpressure protection is after the heat exchanger so that all fuel system natural gas is gas phase. Figure 2: Typical large-duty CNG vehicle fuel system schematic with most major components. Note the fail-close solenoid valves on all storage tanks isolate natural gas from the fuel system during regular maintenance. The potential Causes and Consequences for each Hazard Scenario are noted in the datasheets in the respective columns. Additional columns are included in the datasheets where prevention 21 Page 73 of 129

74 features, detection methods, and mitigation features information can be recorded. These fields were not completed at this point, except for a couple of samples, because this data can be different for the various different Operation States applicable to each Scenario. It is intended that these scenarios will be split out individually as needed and populated as part of Phase II of this project. These measures will be used as the basis for identifying best practices and codes and standards improvements HAZOP Results The HAZOP resulted in the identification of 41 Hazard Scenarios, although many were applicable to multiple Operation States. Three Hazard Scenarios (HAZOP numbers 7, 14, and 19) were selected for further characterization by modeling due to the intentional release of natural gas indoors. Two of these (HAZOP numbers 14 and 19) can be combined into one modeling scenario as they result in the exact same release situation: the venting of the entire CNG tank contents. HAZOP information for these scenarios is shown in Table 3. Two additional situations where natural gas is intentionally vented indoors were also selected for modeling but were not identified in the HAZOP because they are controlled releases. These situations involve the venting of residual natural gas pressure in the fuel system downstream of the isolation valve. Venting for CNG and LNG fuel systems were separately considered. Table 3: HAZOP Results selected for modeling analyses Modeling Scenario NA 3 NA 4 HAZOP Number Component LNG-4 (LNG tank) LNG Bleed Valve CNG -7 Bleed Valve CNG-1 (Cylinders) CNG-3 (Pressure Relief Device) Operation State Hazard Scenario Causes Consequences Overpressure of 3in, 4, 5, tank and proper Excessive hold time, Minor release 7, 8 operation of relief insulation failure of GNG valve 5, 7 5, 7 3in, 4, 5, 7, 8 3in, 4, 5, 7, 8 Residual pressure is vented from fuel system downstream of isolation valve Residual pressure is vented from fuel system downstream of isolation valve Overpressure of Cylinder Failure of PRD to hold pressures below activation pressure Intentional Intentional External fire AND successful operation of PRD Mechanical defect, material defect, installation error, maintenance error Minor release of GNG Minor release of GNG Potential catastrophic release of CNG Potential catastrophic release of CNG 22 Page 74 of 129

75 4. Scenario Analysis To perform analyses of the identified HAZOP scenarios, a numerical modeling approach, previously validated for large-scale indoor hydrogen releases scenarios [5, 6], was adopted. The CFD solver, Fuego [17], was used to perform the natural gas release simulations from a representative NGV inside the maintenance facility. Fuego is a SNL developed code designed to simulate turbulent reacting flow and heat transfer [17] on massively parallel computers, with a primary focus on heat transfer to objects in pool fires. The code was adapted for compressible flow and combustion, and is well suited for low Mach number flows. The discretization scheme used in Fuego is based on the control volume finite element method [18], where the partial differential equations of mass, momentum, and energy are integrated over unstructured control volumes. The turbulence model was a standard two equation (k-ε) turbulence model [19] with transport equations solved for the mass fractions of each chemical species, except for nitrogen which was modeled as the balance. For the calculations reported here, the first order upwind scheme was used for the convective terms. Note that methane was used as a proxy for natural gas in all simulations. For releases that involved transient blow-downs, the isentropic expansion was modeled using the NETFLOW compressible network flow analysis code [20]. Time-histories of the flammable mass and volume, along with calculations for the maximum flammable extent i.e., the distance from the release point where flammable mixture is present are provided for each scenario. These plots are complemented by iso-contour images of the flammable boundary for each release at select time intervals to better illustrate the development of flammable clouds. Finally, maximum possible overpressures from an ignition event are calculated to help determine the harm posed for an unintended ignition event. The overpressure results will help identify scenarios where further mitigation efforts for release and ignition events are needed Maintenance Garage The maintenance garage was modeled as a pitched roof building (1:6 pitch) that was 30.5 m long (100 ), 15.2 m wide (50 ) and 6.1 m tall (20 ), with the roof peak located at the center and 127 cm (50 ) higher than the corresponding eaves (see schematic in Figure 3). Note that although the roof and main building are shown with different colors to emphasize the pitch, the enclosure was treated as a single volume. A roof layout both with and without horizontally orientated support beams was investigated to determine if the supports would cause the accumulation of flammable mixture in discrete pockets. For the condition with supports, 9 beams that were 15.2 cm wide (6 ) and 107 cm tall (42 ) were spaced 3.05 m apart (10 ) and ran parallel to the roof pitch. The garage contained two vents that were used for air circulation; one near the floor along one of the smaller building side-walls, and a second placed on the opposite side wall near the roof. Each vent was m tall (25 ) and 3.42 m wide (131 ). The NGV was modeled as a cuboid with a height and width of 2.44 m (8 ) and a length of 7.31 m (24 ). The vehicle was centered on the building floor with the major axis aligned to the building minor axis. There was no fluid flow through this volume. 23 Page 75 of 129

76 Figure 3: Schematic of the NGV maintenance facility used for the simulations. The roof had a 1:6 pitch and had layouts with and without 9 evenly spaced, horizontal supports. Two circulation vents were located on the smaller building side-walls, with one placed low and the other high to maximize room currents Simulation Boundary Conditions The Fuego code solved the conservation equations in a time-dependent manner with gravity and buoyancy effects accounted for. A slip wall boundary condition with a constant ambient temperature (294 K) was used for all surfaces. The simulations were performed with and without mechanical ventilation to determine the impact on the development of flammable volumes in the garage. For the conditions with ventilation, a uniform air flow velocity of 2.0 m/s (6.56 ft/s) was forced through the floor vent into the enclosure, to produce 5 ACH for the enclosure. The upper enclosure exhaust vent was assigned an open boundary condition with a total pressure of 1 atm and a temperature of 294 K. A relatively coarse grid was used with 195,000 node points. For the tank blow-down simulation with higher Reynolds number exit conditions, a fine grid was used that had 2.5 million grid points and spacing that was a least half of what was used for the original grid. For example, node spacing values around the leak and near the vents were 5 cm and 15 cm for the reference coarse grid, while these values were 2 cm and 6 cm respectively for the fine mesh. For all scenarios, initial turbulence was negligible (k = 0.11 cm 2 /s, ε = cm 2 /s 3 ). For conditions with mechanical ventilation, air was forced into the enclosure at the prescribed 5 ACH flow rate for 750 seconds prior to the start of the release to ensure the enclosure airflow was nominally steady Dormant LNG Blow-off Scenario A schematic of major LNG vehicle supply system components such as the tank, heat exchanger, fuel shutoff valve, and flow regulator are provided in Figure 1. These components are designed to limit natural gas content within the downstream fuel system. Instead, a more serious threat was deemed to be a fully fueled LNG vehicle that was left dormant in the NGV maintenance facility for a period longer than the LNG tank hold time (~7 days). As a result, the pressure buildup would cause a PRV to relieve and release a controlled amount of cool gas phase natural gas (~160 K) through a vertically orientated vent stack until the tank pressure fell below to the PRV seat pressure. Based on industry input, the release was expected to be about 1.7% of the 24 Page 76 of 129

77 cylinder contents before the PRV seats. Rather than rapidly discharging, the PRV was expected to weep for several minutes with a nearly constant flow rate until the tank pressure reaches the seat pressure. Once reseated, the PRV likely would not relieve again for up to a day or more. Code requirements dictate the release points be from a 'safe location', which has typically been interpreted as a point that is above head height and roughly vertical. Relief vents are normally 3/8" stainless steel tubing with a plastic slip on cap to protect from rain water. For the current scenario, saturated methane vapor was released through a vertically orientated 3/8 vent stack, whose exit was 2.44 m (8 ) above the floor; note that the saturated vapor exit temperature (160 K) and density (1.23 kg/m 3 ) at atmospheric pressure were taken from the online NIST calculator [21]. The fully fueled large tank had a volume of 700 liters, and the release of 1.7% of the cylinder contents corresponded with roughly 2.3 kg (5.1 lbs) of fuel. The nominal expected flow rate was 7.58 g/s (l.0 lbs/min), which resulted in a leak duration of 306 seconds. Due to gridding constraints, the leak area was modeled as a 10 cm 2 (1.55 in 2 ) square hole with an exit velocity of 61.5 cm/s (2.02 ft/s). Although the leak greatly exceeded tubing area, the plastic rain cap would result in a much larger effective leakage area; thus the 10 cm 2 exit area was deemed reasonable CNG and LNG Fuel System Line Cracking From the HAZOP there were concerns that a natural gas release may occur during the purge of a vehicle fuel system as part of regular operational maintenance. Current NGV fuel systems are equipped with fail-closed solenoid valves located either at the tank or fuel supply manifold. The solenoid valves can only be actuated open when the engine is running, which effectively isolates onboard storage from the fuel system when the engine is off there is no recorded instance of the valves failing open. For the identified scenarios, it was assumed that maintenance is to be performed on a CNG or LNG fueled vehicle where cylinder or manifold valves were used to isolate the fuel storage from the remainder of the fuel system where the work will be performed. However, room temperature (294 K) residual natural gas downstream of the onboard storage isolation (and heat exchanger for LNG vehicles) remains in the fuel system. Prior to the start of maintenance, a technician purges the remaining natural gas by cracking a ½ tube fitting on the fuel system at the control panel in the engine compartment both are assumed to be on the vehicle side at a height of 1.0 meters from the floor. For LNG vehicles, original equipment manufacturer (OEM) specifications indicate downstream line and filter volumes are around 1 to 2 liters with a maximum pressure of 8.62 bar (125 psia). Accordingly, for this scenario the fuel system storage volume was set to 1.8 liters (110 in 3 ) with an overall natural gas storage mass of 10.4 g. Following LaChance et al. [2], the release area was assumed to be 3% of the overall tube area, which corresponded to a 3.8 mm 2 hole size. For CNG vehicles, the fuel system volumes are roughly double those for LNG vehicle, and the storage pressure can equal the tank pressure. Hence, the CNG line cracking scenario was identical except that the storage volume was increased to 3.3 liters (201 in 3 ) and the storage pressure was increased to 248 bar (3600 psia), which corresponded to an overall natural gas fuel system mass of 630 g. Note that for both scenarios it was presumed that the shutoff valve was engaged, which prevented the contents downstream of the storage isolation to escape once the line was cracked. Transient blow-downs were modeled as an isentropic expansion using NETFLOW [20]. Once again, gridding constraints limited the leak area to a 10 cm 2 (1.55 in 2 ) square hole, but was 25 Page 77 of 129

78 considered reasonable since the released gas was expected to first accumulate in the control panel or engine compartment before escaping into the maintenance facility Mechanical Failure of a Thermally Activated PRD In the event a CNG cylinder becomes engulfed in a flame, onboard storage cylinders are protected against excessive pressure buildup by a thermally triggered PRD designed to fully open without the possibility for reseat in the event of activation. Accordingly, inadvertent actuation due to some mechanical failure would result in a rapid and uncontrollable decompression of all cylinder contents. Advances such as the use of dual activated valves have been implemented to reduce the likelihood of unintended release, although there remains some nominal risk due to the potential for human error. The SDOs view such a release as a bounding event for hazard potential. For the final scenario, the entire contents of a 700 L, fully pressurized (248 bar) CNG cylinder at room temperature (294 K) was released into the NGV maintenance facility. Note that the tank volume was 50% greater than normal to simulate a worst case scenario. For convenience, the specified release point was identical to the LNG blow-off scenario. The PRD orifice diameter was set to 6.2 mm (0.24 ) based on the flow rate specifications of typical commercially available PRDs. At the start of the release, the valve was assumed to fully open and remain that way for the duration. Once again gridding constraints limited the initial leak to 10 cm2, and NETFLOW was used to model the transient blow-down CFD Scenario Results The primary hazards associated with unintended natural gas releases are the maximum overpressure above ambient and the associated integrated pressure time-history or pressure impulse after the combustible gas mixes with air and ignites. Confinement, particularly with obstacles, can exacerbate overpressure and pressure impulse hazards for sufficiently small enclosures due to the volumetric expansion of gases [22], and can introduce new threats such as flying debris or building collapse [23]. Probit models for individual harm criteria are generally given a function of the expected maximum overpressure and the integrated pressure time-history or pressure impulse, along with any relevant structural details. Analytic methods to evaluate overpressure hazards from confined and vented deflagrations within enclosures generally only consider uniform air-fuel mixture compositions [22, 24-27], and not stratified environments with combustible clouds expected from the scenarios described. Recently, Bauwens and Dorofeev [28] developed an analytic model that only considers the flammable mass quantities and enclosure volumes, without any regard to amount of mixing. Model results yielded good agreement with peak overpressure measurements from large-scale hydrogen release and deflagration experiments by Ekoto et al. [29]. Accordingly, the model was used here to estimate peak overpressure hazards based on the flammable mass prediction from the CFD simulations; pressure impulse was not considered. Note that the model assumes no instability enhancement of the flame front (e.g., acoustic) and that local blast waves were relatively minor; reasonable assumptions for leaks with small flammable volumes. Equation 1 describes how the adiabatic increase in pressure depends on the mass of hydrogen consumed: Eq. 1. {[ ] 26 Page 78 of 129 }

79 where p 0 was the ambient pressure, V T and V NG were the total facility volume and expanded volume of pure methane following the release respectively, χ stoich was the natural gas-air stoichiometric mole fraction, σ was the expansion ratio for stoichiometric natural gas-air combustion, and γ was the air specific heat ratio. Note that it was convenient to define V NG as the ratio of total flammable natural gas mass which was a ready output from the FUEGO CFD simulations to the known ambient density of pure natural gas. It was thus important to accurately predict the flammable mixture across a range of characteristic leaks. The lower (LFL) and upper flammability limits (UFL) for methane mixed with air at atmospheric conditions is 5.0 and 15.0% methane volume fraction respectively [30], while mixtures outside of this range present no possibility for combustion Dormant LNG Blow-off Scenario Results The first scenario involved a PRV release of cool natural gas through a vent stack for a fully fueled LNG vehicle that was left dormant in a maintenance facility beyond the prescribed holdtime. Natural gas mole fraction maps from the maintenance facility central plane for conditions with mechanical ventilation are illustrated in Figure seconds after the start of the release for facility layouts with and without roof supports. Velocity maps from the maintenance facility central plane for the conditions with and without roof supports in illustrate the influence of the strong inlet flows needed to sustain the 5 ACH ventilation rate. When ventilation currents reached the vehicle side, they were deflected upward and formed a low-pressure recirculation region that was capable of bending a vertical natural gas plume toward the vent inlet. For the facility layout with roof supports, there was no substantial shape change in the flammable region. Figure 4: NGV Maintenance facility natural gas mole fraction contours at 10, 60, and 306 seconds into the release for the facility layouts without (top) and with (bottom) roof supports for the LNG blow-off scenario. Velocity maps are also shown along the facility centerline to illustrate the impact of room currents on flow dispersion. 27 Page 79 of 129

80 For both scenarios, flammable natural gas was confined to a small region near the source; areas shaded in blue are too lean to combust. To more clearly illustrate this point, time-histories of the total mass and volume of flammable natural gas within the enclosure (i.e., mixture between the LFL and UFL) for each scenario is plotted in Figure 5. For the facility configuration without beams, the flammable volume and mass initially spiked to a peak value ~10 seconds after the release before assuming a nominally constant value, whereas for the facility with flammable beams the values were nominally steady throughout the release duration. Interestingly, the condition with support beams had a lower flammable mass and volume for most of the release as vortical structures induced by the support beams were able to more rapidly mix air into the release plume. Over time it appears that both the flammable mass and volume steadily increased as the cloud within the center of the maintenance facility steadily grew, although the release duration was too short for this to become a significant hazard. Note that for the conditions without ventilation the maximum for the layouts with and without support beams were 158 and 169 cm respectively. When ventilation was included, the respective flammable extents for the layouts with and without beams were reduced to 85 and 115 cm. A maximum flammable mass of 28 g occurred for the no support beams facility layout without ventilation, which corresponded to a max possible overpressure potential of 125 Pa from equation 1. According to probit models from [31] the lowest potential overpressure harm threshold is the threat of broken glass, which has a lower limit of 1 kpa. Hence, no substantial hazard is expected from this scenario. Figure 5: Time-history of the total natural gas flammable mass and volume for the LNG blow-off scenario. Note that the simulation for the scenario without ventilation in the facility without support beams was terminated 100 seconds into the release once steady flammable concentrations had been firmly established. To ensure the simulation results were not from an artifact of the coarse grid geometry, a gridconvergence study was performed for the scenario with roof supports that was believed to be more sensitive to grid sizing. The fine grid described earlier was used to repeat the simulation and the flammable mass time-history from both simulations, and as can be seen in Figure 6 produced identical results to the simulation with the coarse grid out to just past 200 seconds into the release. From these results it is clear that simulation outputs are independent of grid sizing. 28 Page 80 of 129

81 Figure 6: Grid convergence test that used the coarse (195,000 nodes) and fine (2.5 million nodes) grids for the LNG blow-off scenario with roof rafters to ensure repeatable results CNG and LNG Fuel System Line Cracking Results For the second scenario, the impact of a fuel system ½ line cracked prior to the start of maintenance operations for CNG fueled vehicles was analyzed since the total fuel within LNG fuel systems is much lower than for CNG vehicles, only the CNG release was considered here. Moreover, only the facility layout without roof supports was considered since the plume from the side-release was not expected to be influenced by the centrally located circulation region above the vehicle. The transient blow-down was modeled via NETFLOW, with the release rate timehistory provided in Figure 7. Figure 7: Mass flow rate time-history plot for the CNG line cracking scenario calculated from NETFLOW. Center plane LFL iso-contour maps for the facility without support beams are provided at select times in Figure 8. Complementary time-history plots of the total flammable mass and volume are 29 Page 81 of 129

82 included in Figure 9. By 2.9 seconds into the release when the flammable extent was greatest at 265 cm, the exit plume near the vehicle contained the peak flammable mass values (up to 100 g) due to a combination of high initial mass flow rates and limited mixing. Nonetheless, the peak flammable mass and volume values were small, which limited the possible overpressure to 0.43 kpa; well below the lowest harm threshold. Moreover, the duration of flammable mixture within the enclosure was very short, with all flammable regions diffused away by 23 seconds into the release (see Appendix B for further details). Figure 8: Maintenance facility natural gas LFL iso-contours at 2.5 (top), 10 (center), and 30.0 (bottom) seconds into the release for the layouts without roof supports for the CNG line cracking scenario. 30 Page 82 of 129

83 Figure 9: Time-histories of total natural gas flammable mass and volume for the CNG blow-down scenario Full-Scale Tank Blow-Down due to a Mechanical Failure of the PRD In the final scenario, the transient blow-down was modeled of a fully fueled CNG cylinder with a 700 liter volume and pressurized to 248 bar that released all contents due to the mechanical failure of a thermally activated PRD through a 6.2 mm diameter orifice. Once again the transient blow-down was modeled via NETFLOW, with the blow-down curve plotted in Figure 10. Note that higher flow rates and longer release durations meant these simulations were far more computationally expensive. Accordingly only a single configuration could be evaluated within the current project scope. To ensure the worst-case-scenario, the facility layout with roof supports and active mechanical ventilation was selected since vortical flow structures above the plume were thought to aid in the accumulation of flammable mixture near the release point (see Appendix B for further details). The fine mesh was used to ensure convergence of all conservation equations for the higher Reynolds number flow from the larger release. Figure 10: Mass flow rate time-history for the CNG tank blow-down scenario calculated from NETFLOW for a 700 liter tank pressurized with natural gas to 248 bar and released through an a 6.2 mm diameter orifice. Note that the tank volume was 50% greater than normal to simulate a worst case scenario. 31 Page 83 of 129

84 Images of LFL iso-contours from the release plume at discrete times are provided Figure 11, along plots of the flammable mass and volume for each time selected. It should be noted that the rapid expansion forced temperatures within the tank to quickly drop, which likewise lowered the leak exit temperature. By 220 seconds into the release the temperatures at the leak exit plane had dropped below the condensation point (i.e., 160 K at ambient pressure), which was expected to result in two-phase flow behavior in the exit stream. Liquid parcel velocities develop at different rates relative to the vapor phase due to density differences. The difference in phase velocity, often referred to as the slip velocity, can significantly impact cryogenic releases dispersion results [32]. Velocity slip modeling is beyond the current simulation capabilities, which means dispersion data beyond 220 seconds into the release cannot be trusted. However, by this point kg or about 87.6% of the original tank contents had been evacuated. Thus, it seems likely that flammable mass values within the enclosure had reached or were near their peak values by this time. 32 Page 84 of 129

85 Figure 11: Maintenance facility natural gas LFL iso-contours for the CNG tank blow-down scenario from a 700 liter tank pressurized to 248 bar for the facility layout with active ventilation, roof support beams, and a vertical release into the enclosure. Time histories of flammable mass and volume are also included. 33 Page 85 of 129

86 From Figure 11, it can be observed that the release plume rapidly reached the ceiling located 4.9 m above the vehicle release point, and retained flammable concentrations from the vehicle to the ceiling for the duration of the release. Two distinct peaks in both flammable mass and volume time histories were observed. The first occurred 68 seconds into the release as the flammable mixture steadily accumulated into the roof rafters and began to spread horizontally across the ceiling. The peak flammable mass at this point was 473 g, which for the present facility corresponded to a peak estimated overpressure of 2.1 kpa from equation 1. A second peak at 501 g, which corresponds to a peak estimated overpressure of 2.2 kpa, occurred 220 seconds into the release as the cooler release plume became denser with slower mixing rates within the release plume. As mentioned earlier, the simulation accuracy is questionable beyond this point in the release. Nonetheless, it appears that the flammable mass is steadily increasing and has not yet hit an asymptote. If as a worst case scenario, the flammable mass were to triple to around 1.5 kg which seems extremely conservative given the small amount of natural gas remaining in the tank and the relatively low flow rates by this point peak overpressures would increase to around 6.6 kpa. According to [31, 33], even this conservative overpressure estimate is still below the threshold needed for injuries due to projected missiles (6.9 kpa), eardrum rupture (13.8 kpa), or the collapse of unreinforced concrete walls (15 kpa). Note that most of the flammable volume exists in the plume, which itself is mostly located below the 0.46 m threshold for protection from electrical ignition sources stipulated in NFPA 30A. It is also important to note that the overpressure calculation should be linearly proportional to the facility volume. Hence, if the facility volume were to be halved, the expected overpressure from the volumetric expansion of hot gases would roughly double above the reported values, which could introduce potentially hazardous scenarios. 34 Page 86 of 129

87 5. Summary and Conclusions CVEF and SNL have partnered to analyze current regulatory issues regarding NGV maintenance facility operations. The goal has been to leverage their collective experience with code interpretation, hazard analysis, NGV fuel system design, and facility operations, along with welldeveloped modeling capabilities to inform code development for NGV facility construction and maintenance. While existing code language has been developed from user experience, it is recognized by SDOs that risk-informed approaches that identify high-risk scenarios along with dominant causal factors and that quantify the effectiveness of accident prevention/mitigation strategies are needed. The scope of work has been split into two phases with the current report summarizing the results from Phase I. Phase I work involved a detailed survey of existing regulations, a HAZOP to identify critical hazards from operational activities, and an analysis of potential consequences for credible hazards. These measures will be used as the basis for identifying best practices and codes and standards improvements. The HAZOP analysis included additional columns where prevention features, detection methods, and mitigation features information can be recorded. These fields were not completed for the Phase I work since these data can be different for the various Operation States applicable to each scenario. These scenarios will be split out individually as needed and populated as part of Phase II work. Phase II work is also expected to use the Phase I generated information to develop best practices, suggest hazard mitigation strategies, and recommend changes to existing fire protection codes. For the hazard analysis work, detailed CFD simulations were performed at Sandia to examine the 3 release scenarios identified from the HAZOP: (1) a dormant LNG blow-off, (2) indoor CNG fuel system purge downstream of the storage isolation valves, and (3) a full-scale CNG tank blow-down due to a failure of the PRD. Methane was used as a proxy for natural gas in the simulations. The reference NGV facility had dimensions of 30.5 m long, 15.2 m wide and 6.1 m tall, with pitched roof. Geometries with and without evenly spaced roof rafters were examined. The impact of active ventilation at the commonly prescribed rate of 5 ACH versus a facility with passive ventilation was also considered for the dormant LNG blow-off scenario. For conditions with mechanical ventilation, air was forced into the enclosure 720 seconds before the start of the release to ensure internal steady flows. The vehicle was modeled as a cuboid and placed in the center of the NGV maintenance facility. Harm potential from peak overpressure was estimated using an model developed by FM Global for transient leaks and validated against previous Sandia data for hydrogen indoor refueling scenarios. For the overpressure model inputs, the time-history of the flammable mass and volume (i.e., natural gas/air mixture within the flammable bounds) was extracted from the CFD simulation results. From velocity maps within the NGV maintenance facility, ventilation currents were observed to form recirculation regions when they interacted with the vehicle or roof rafters, which could distort the release plumes and generate flammable mixture accumulation regions. However, for the scenarios investigated, little sensitivity was observed for ventilation or roof supports due to the short durations of the releases relative to the ventilation rates and the propensity of the support structures to enhance mixing. Accordingly, for the low-flow release scenarios that involved a dormant LNG blow-off or a CNG fuel system purge, the flammable masses, volumes, and extents were low, and the flammable regions disappeared shortly after the conclusion of the 35 Page 87 of 129

88 leaks. Moreover, predicted peak overpressures indicated there was no significant hazard expected. For the larger release, low leak-exit temperatures late into the release resulted in natural gas state conditions that could not be modeled FUEGO simulation package, with results beyond this point were rejected, although over 85% of the cylinder contents had evacuated into the enclosure by this point. Nonetheless, the release plume quickly achieved a nearly steady flammable volume that extended from the release point at the vehicle up to the ceiling located 4.9 meters above the release, before spreading slightly across the ceiling. Two peaks were observed in the flammable mixture time-histories. The first peak occurred 68 seconds into the release where vessel flow rates were still relatively high and previously expelled mixture accumulated in flammable concentrations along the ceiling. The second peak occurred at the end of the accepted simulation results and was attributed to increasingly cool and dense exit plumes that had slower mixing rates. For both peaks, there was roughly 0.5 kg of natural gas predicted to exist in flammable regions, which for the facility examined could produce an overpressure of around 2.2 kpa enough to break glass, but not much else. It was noted that flammable mass values would likely further increase beyond if the leak dispersion characteristics were properly modeled. However, even a conservative estimate for the expanded overpressure potential is still below the threshold required for significant harm. It should be cautioned that no attempt to calculate local blast-wave pressures was performed, which could result in additional overpressures above those described here. However, the relatively small volumes of the flammable regions mean that there is little opportunity for flame acceleration needed for blast-wave development. For Phase II work, additional layout configurations should be evaluated with the tank blow-down scenario, since this is the only scenario capable of generating harmful overpressure effects. Furthermore, since the current simulations require several weeks to run, there is a need simplified tools development to enable parametric investigations of multiple facility configurations and leak conditions. Current work in this regard for the hydrogen safety programs could be leveraged for use with natural gas. 36 Page 88 of 129

89 References [1] Horne DB, Technical Bulletin: Improper Use of Underrated European CNG Valves and Rupture Disc PRDs on US Vehicles. Clean Vehicle Technology Foundation, [2] LaChance J, Houf W, Middleton B, Fluer L, Analyses to Support Development of Risk- Informed Separation Distances for Hydrogen Codes and Standards. SAND , Sandia National Laboratories, March, [3] Rose SE, Flamberg S, Leverenz F, Guidance Document for Incorporating Risk Concepts into NFPA Codes and Standards. Fire Protection Research Foundation, March, [4] Groth KM, Mosleh A, A data-informed model of performance shaping factors and their interdependencies for use in human reliability analysis. Reliability, Risk and Safety: Theory and Applications Vols 1-3, 2010: [5] Houf WG, Evans GH, James SC, Merilo E, Groethe M, Simulation of Hydrogen Releases from Fuel-Cell Vehicles in Tunnels. Proc World Hydrogen Energy Conf, Essen, Germany, May 16-21, [6] Houf WG, Evans GH, Ekoto IW, Merilo EG, Groethe MA, Hydrogen fuel-cell forklift vehicle releases in enclosed spaces. Int J Hydrogen Energy, 2013;38: [7] Horne DB, Guideline for Determining the Modifications Required for Adding Compressed Natural Gas and Liquefied Natural Gas Vehicles To Existing Maintenance Facilities. Clean Vehicle Education Foundation, August 31, [8] International Fire Code, International Code Council, Country Club Hills, IL, [9] International Mechanical Code, International Code Council, Country Club Hills, IL, [10] International Building Code, International Code Council, Country Club Hills, IL, [11] Code for Motor Fuel Dispensing Facilities and Repair Garages, National Fire Protection Association, NFPA 30A, Quincy, MA, [12] Vehicular Gaseous Fuel Systems Code, National Fire Protection Association, NFPA 52, Quincy, MA, [13] Standards for Parking Structures, National Fire Protection Association, NFPA 88A, Quincy, MA, Page 89 of 129

90 [14] Horne DB, Safety Advice for Defueling CNG Vehicles and Decommissioning and Disposal of CNG Cylinders. Clean Vehicle Technology Foundation, [15] "Risk Assessment and Risk Management for the Chemical Process Industry." Edited by Greenberg HR, Cramer JJ, New York: Stone & Webster Engineering Corporation, Nostrand Reinhold, [16] Hazard and Operability Studies (HAZOP Studies) Application Guide, British Standard IEC, 21882:2001, Aug 28, [17] Moen CD, Evans GH, Domino SP, Burns SP, A Multi-Mechanics Approach to Computational Heat Transfer. Proc ASME Int Mech Eng Cong and Exhibition, New Orleans, IMECE , Nov , [18] Minkowycz WJ, Sparrow EM, Schneider GE, Pletcher RH, "Elliptic systems: finite element method 1, handbook of numerical heat transfer." New York: J. Wiley and Sons, Inc., [19] Papageorgakis GC, Assanis DN, Comparison of linear and nonlinear RNG-based k- epsilon models for incompressible turbulent flows. Numer Heat Tr B-Fund, 1999;35:1-22. [20] Winters WS, A New Approach to Modeling Fluid/Gas Flows in Networks. SAND , Sandia National Laboratories, July, [21] National Institute of Science and Technology Available from: [22] Bauwens CR, Chaffee J, Dorofeev S, Effect of Ignition Location, Vent Size, and Obstacles on Vented Explosion Overpressures in Propane-Air Mixtures. Combust Sci Technol, 2010;182: [23] LaChance J, Tchouvelev A, Engebo A, Development of uniform harm criteria for use in quantitative risk analysis of the hydrogen infrastructure. Int J Hydrogen Energy, 2011;36: [24] Bradley D, Mitcheson A, Venting of Gaseous Explosions in Spherical Vessels.1. Theory. Combust Flame, 1978;32: [25] Tamanini F, Dust explosion vent sizing - Current methods and future developments. J Phys Iv, 2002;12: [26] Baker QA, Tang MJ, Scheler EA, Silva GJ, Vapor cloud explosion analysis. Process Saf Prog, 1996;15: Page 90 of 129

91 [27] van den Berg AC, Versloot NHA, The multi-energy critical separation distance. J Loss Prevent Proc, 2003;16: [28] Bauwens CR, Dorofeev S, CFD Modeling and Consequence Analysis of an Accidental Hydrogen Release in a Large Scale Facility. Proc International Conference on Hydrogen Safety 5, Brussels, Belgium, [29] Ekoto IW, Houf WG, Evans GH, Merilo EG, Groethe MA, Experimental investigation of hydrogen release and ignition from fuel cell powered forklifts in enclosed spaces. Int J Hydrogen Energy, 2012;37: [30] Zabetakis MG, Flammability Characteristics of Combustible Gases and Vapors. Washington D.C.: Bulletin 627, U.S. Department of Interior, Bureau of Mines, [31] "Guidelines for Evaluating the Characteristics of Vapor Cloud Explosions, Flash Fires, and BLEVEs." Center for Chemical Process Safety, Wiley-AIChE, [32] Giannissi SG, Venetsanos AG, Markatos N, Willoughby DB, Royle M, Simulation of Hydrogen Dispersion Under Cryogenic Release Conditions. Proc nternational Conference on Hydrogen Safety 5, Brussels, Belgium, [33] Jeffries RM, Hunt SJ, Gould L, Derivation of fatality of probability function for occupant buildings subject to blast loads. Health and Safety Executive, Page 91 of 129

92 40 Page 92 of 129

93 Appendix A: HAZOP Data Sheets 41 Page 93 of 129

94 42 Page 94 of 129

95 43 Page 95 of 129

96 44 Page 96 of 129

97 Appendix B: Supplemental CFD Simulation Data In this Appendix, supplemental CFD simulation data that could not easily fit into the body of the text is included. For the LNG blow-off scenario, concentration maps are provided in Figure 12 and Figure 13 for the conditions with and without roof supports respectively. From these images, it can be observed that ventilation induced low pressure regions led to substantial distortion of the release plume near the release where flammable concentrations were highest. For the scenario without roof supports, the plume impinged on the ceiling and formed a wall jet that spread along the ceiling. The spread direction was biased towards the exit vent due to the room currents from the ventilation system. Figure 12: NGV Maintenance facility natural gas mole fraction contours at 10, 60, and 306 seconds into the release for the facility layouts without roof supports for the LNG blow-off scenario. 45 Page 97 of 129

98 For the facility layout that included roof supports, recirculation vortices formed by the interaction between the room currents and the beams resulted in a localized accumulation region of lean natural gas near the release plume. Over time, the concentration of plume became richer as very little natural gas was able to escape through the exit vent. However, as was seen in Figure 5, the impact on flammable concentrations within the enclosure was negligible since the accumulation rates were slow relative the release duration. It was thought that the accumulation region could have a bigger impact for longer duration releases, which is why this facility configuration was selected for the CNG tank blow-down scenario. Figure 13: NGV Maintenance facility natural gas mole fraction contours at 10, 60, and 306 seconds into the release for the facility layouts with roof supports for the LNG blow-off scenario. 46 Page 98 of 129

99 Natural gas concentration maps from the maintenance facility center plane at 2.5 and 30.5 seconds into the release for the NGV facility configuration without support beams are provided in Figure 14. Despite flammable concentrations initially concentrated near the release, the rapid decay in mass flow rates coupled with strong diffusion that quickly mixed the plume with ambient air led to very short durations for flammable mixtures in the facility. Figure 14: Maintenance facility natural gas mole fraction contours at 2.5 (top) and 30.5 (bottom) seconds into the release for the layouts without roof supports for the CNG line cracking scenario. 47 Page 99 of 129

100 48 Page 100 of 129

101 Distribution 1. Douglas B. Horne DBHORNE LLC Clean Vehicle Education Foundation 6011 Fords Lake Ct. Acworth, GA MS0748 A. Christine LaFleur MS0748 Jeffery L. LaChance MS9052 Isaac W. Ekoto MS9052 Daniel E. Dedrick MS9054 Robert Q. Hwang MS9054 Arthur E. Pontau MS9957 Myra L. Blaylock MS0899 Technical Library 9536 (electronic copy) 10. MS0115 OFA/NFE Agreements Page 101 of 129

102 Page 102 of 129

103 National Fire Protection Association Report 6 of 42 10/26/2016 3:45 PM maintenance facility. From velocity maps within the NGV maintenance facility, ventilation currents were observed to form recirculation regions when they interacted with the vehicle or roof structures, which could distort the release plumes and generate accumulation regions for flammable mixture. However, for the scenarios investigated, little sensitivity in the development of flammable regions was observed for simulations with or without active ventilation. This was due to the small duration of the release relative to the ventilation rate. Similarly, the sensitivity of flammable mixture development with facility layouts with or without ceiling beam supports was likewise weak as the mixtures were generally already lean by the time they reached the beams. The LNG blow-off scenario was modeled as a constant leak for five minutes. During that time the amount of flammable mass of the release quickly reached a steady state, and dissipated within seconds of the leak stopping. For the case of the indoor CNG fuel system purge downstream of the storage isolation valves, the gas was purged from the line within 30 seconds, and the release cloud was within the flammable concentration limits only during the time of this blowdown. For both of these cases, the released gas was in the flammable range only within an area of several feet from the leak source: less than two feet for case (1) and less than 10 feet for case (2). In neither case did the area of flammable mass reach the ceiling. For case (1) simulations with and without ventilation were compared. While the ventilation did slightly lower the amount of flammable mass, it did not alleviate it completely. For the low-flow release scenarios that involved a dormant LNG blow-off or a CNG fuel system purge, peak overpressures predicted by the FM Global overpressure model with input flammable mass values from the CFD simulations were well below 1 kpa no significant hazard is expected for such a low overpressure. A worst case scenario, which has a much lower likelihood of occurring, was also modeled as a venting of a full CNG 700 L cylinder. This size is 50% larger than a normal cylinder. This scenario (described in Analysis of a Full Scale Blowdown Due to a Mechanical Failure of a Pressure Relief Device in a Natural Gas Vehicle Maintenance Facility by Blaylock, et al.) shows that a flammable region will form from the release point at the vehicle up to the ceiling, and will then pool with a thickness up to 80 for the size of garage modeled. The release will last around 15 minutes before dissipating. The Technical Committee should consider whether the potential for this scenario to occur (none have occurred since 2002) should be the basis for the code requirements. If not, then the modeling shows that the most likely release scenarios do not result in hazardous concentrations in the beam pockets, within 18 inches of the ceiling, with or without ventilation. (SAND O) Related Item Public Input No. 25-NFPA 30A-2015 [Section No ] Submitter Information Verification Submitter Full Name: Myra Blaylock Organization: Sandia National Labs Street Address: City: State: Zip: Submittal Date: Mon May 16 00:49:24 EDT 2016 Committee Statement Page 103 of 129

104 National Fire Protection Association Report 7 of 42 10/26/2016 3:45 PM Committee Action: Resolution: Rejected The modeling performed does not address all credible release scenarios and is therefore too limited to make a change to the current requirements in the code. Page 104 of 129

105 National Fire Protection Association Report 8 of 42 10/26/2016 3:45 PM Public Comment No. 31-NFPA 30A-2016 [ Section No ] * In major repair garages where CNG vehicles are repaired or stored, the area within 455 mm (18 in.) of the ceiling shall be designated a Class I, Division 2 hazardous (classified) location. Exception: In major repair garages, where ventilation equal to not less than four air changes per hour is provided, this requirement shall not apply. Statement of Problem and Substantiation for Public Comment RE: Support for Public Input No. 25-NFPA 30A-2015 (Sect ) Dear Sir/Madam, Waste Management (WM) is writing to support Public Input No. 25-NFPA 30A-2015 (Sect. No ) and the deletion of section WM began using compressed natural gas (CNG) in its collection vehicles as early as 1994 and then began significant implementation of CNG in We now operate the largest fleet of Class 8 natural gas vehicles (NGVs) in North America approaching 6,000 collection vehicles. In addition to our fleet vehicles, we also have nearly 100 facilities outfitted for indoor NGV maintenance across the U.S. and Canada. While we are familiar with the failures of the original thermally activated PRDs in the 1990s that were the impetus for the original NFPA 30A Section 8.2.1, we have never experienced a premature PRD failure in any of our fleet vehicles. We are also not familiar with any premature PRD failures since We believe the reason for this success rate over the past 14 years is that the NGV industry worked diligently through the 1990s and early 2000s to phase out older PRD technology and to develop the newer, safer PRDs that we have been using on all of our CNG cylinders since. Given recent Sandia National Laboratory research showing that the more likely releases from CNG and LNG vehicles would be a much lower volume (than the original PRD release scenarios) and that those smaller releases do not allow for increased natural gas concentrations in the ceiling area of maintenance facilities, we question whether NFPA 30A needs to contain Section any longer. Sincerely, Chip Wertz National Director Fleet Facility Infrastructure cbwertz@wm.com Related Item Public Input No. 25-NFPA 30A-2015 [Section No ] Submitter Information Verification Page 105 of 129

106 National Fire Protection Association Report 9 of 42 10/26/2016 3:45 PM Submitter Full Name: Chip Wertz Organization: Waste Management Street Address: City: State: Zip: Submittal Date: Mon May 16 14:14:08 EDT 2016 Committee Statement Committee Action: Resolution: Rejected The modeling performed does not address all credible release scenarios and is therefore too limited to make a change to the current requirements in the code. Page 106 of 129

107 National Fire Protection Association Report 0 of 42 10/26/2016 3:45 PM Public Comment No. 30-NFPA 30A-2016 [ Section No ] Page 107 of 129

108 National Fire Protection Association Report 1 of 42 10/26/2016 3:45 PM * Page 108 of 129

109 National Fire Protection Association Report 2 of 42 10/26/2016 3:45 PM Table shall be used to delineate and classify areas for the purposes of installing electrical wiring and electrical utilization equipment where Class I liquids are stored, handled, or dispensed. [See also Figure 8.3.2(a) and Figure 8.3.2(b).] Table Class I Locations Motor Fuel Dispensing Facilities Class I Location Division (Group D) Zone (Group IIA) Extent of Classified Location a Dispensing device (except overhead type) b,c Under dispenser containment 1 1 Entire space within and under dispenser pit or containment Dispenser 2 2 Within 450 mm (18 in.) of dispenser enclosure or that portion of dispenser enclosure containing liquid handling components, extending horizontally in all directions and down to grade level Outdoor 2 2 Up to 450 mm (18 in.) above grade level, extending 6 m (20 ft) horizontally in all directions from dispenser enclosure Indoor With mechanical ventilation 2 2 Up to 450 mm (18 in.) above floor level, extending 6 m (20 ft) horizontally in all directions from dispenser enclosure With gravity ventilation 2 2 Up to 450 mm (18 in.) above floor level, extending 7.5 m (25 ft) horizontally in all directions from dispenser enclosure Dispensing device (overhead type d ) Repair garage, major e (where Class I liquids or gaseous fuels are transferred or dispensed f ) (see and 8.3.1) 1 1 Space within dispenser enclosure and all electrical equipment integral with dispensing hose or nozzle 2 2 Within 450 mm (18 in.) of dispenser enclosure, extending horizontally in all directions and down to grade level 2 2 Up to 450 mm (18 in.) above grade level, extending 6 m (20 ft) horizontally in all directions from a point vertically below edge of dispenser enclosure 1 1 Entire space within any pit, belowgrade work area, or subfloor work area that is not ventilated 2 2 Entire space within any pit, belowgrade work area, or subfloor work area that is provided with ventilation of at least 0.3 Page 109 of 129 m 3 /min/m 2 (1 ft 3 /min/ft 2 ) of floor

110 National Fire Protection Association Report 3 of 42 10/26/2016 3:45 PM Class I Location Specific areas adjacent to classified locations Repair garage, minor e (where Class I liquids or gaseous fuels are not transferred or dispensed f ) (see and 8.3.1) Specific areas adjacent to classified locations Division (Group D) Zone (Group IIA) Extent of Classified Location a area, with suction taken from a point within 300 mm (12 in.) of floor level (see ) 2 2 Up to 450 mm (18 in.) above floor level of the room, except as noted below, for entire floor area Unclassified Unclassified Up to 450 mm (18 in.) above floor level of the room where room is provided with ventilation of at least 0.3 m 3 /min/m 2 (1 ft 3 /min/ft 2 ) of floor area, with suction taken from a point within 300 mm (12 in.) of floor level 2 2 Within 0.9 m (3 ft) of any fill or dispensing point, extending in all directions Unclassified Unclassified Areas adjacent to classified locations where flammable vapors are not likely to be released, such as stock rooms, switchboard rooms, and other similar locations, where mechanically ventilated at a rate of four or more air changes per hour or designed with positive air pressure or where effectively cut off by walls or partitions 2 2 Entire space within any pit, belowgrade work area, or subfloor work area that is not ventilated 2 2 Up to 450 mm (18 in.) above floor level, extending 0.9 m (3 ft) horizontally in all directions from opening to any pit, belowgrade work area, or subfloor work area that is not ventilated Unclassified Unclassified Entire space within any pit, belowgrade work area, or subfloor work area that is provided with ventilation of at least 0.3 m 3 /min/m 2 (1 ft 3 /min/ft 2 ) of floor area, with suction taken from a point within 300 mm (12 in.) of floor level (see ) Unclassified Unclassified Areas adjacent to classified locations where flammable vapors are not likely to be released, such as stock rooms, switchboard rooms, and other similar locations, where mechanically ventilated Page 110 of 129

111 National Fire Protection Association Report 4 of 42 10/26/2016 3:45 PM Class I Location Repair garage, major e (where lighter-than-air gaseous fueled g vehicles are repaired or stored) (see ) Specific areas adjacent to classified locations Remote pump Division (Group D) Zone (Group IIA) Extent of Classified Location a at a rate of four or more air changes per hour or designed with positive air pressure, or where effectively cut off by walls or partitions 2 2 Within 450 mm (18 in.) of ceiling, except as noted below Unclassified Unclassified Within 450 mm (18 in.) of ceiling where ventilation of at least 0.3 m 3 /min/m 2 (1 ft 3 /min/ft 2 ) of floor area, with suction taken from a point within 450 mm (18 in.) of the highest point in the ceiling Unclassified Unclassified Areas adjacent to classified locations where flammable vapors are not likely to be released, such as stock rooms, switchboard rooms, and other similar locations, where mechanically ventilated at a rate of four or more air changes per hour or designed with positive air pressure, or where effectively cut off by walls or partitions Outdoor 1 1 Entire space within any pit or box below grade level, any part of which is within 3 m (10 ft) horizontally from any edge of pump 2 2 Within 900 mm (3 ft) of any edge of pump, extending horizontally in all directions 2 2 Up to 450 m (18 in.) above grade level, extending 3 m (10 ft) horizontally in all directions from any edge of pump Indoor 1 1 Entire space within any pit Sales, storage, rest rooms (including structures [such as the attendant s kiosk] on or adjacent to dispensers) 2 2 Within 1.5 m (5 ft) of any edge of pump, extending in all directions 2 2 Up to 900 mm (3 ft) above floor level, extending 7.5 m (25 ft) horizontally in all directions from any edge of pump Unclassified Unclassified Except as noted below Page 111 of 129

112 National Fire Protection Association Report 5 of 42 10/26/2016 3:45 PM Class I Location Tank, aboveground Division (Group D) Zone (Group IIA) Extent of Classified Location a 1 1 Entire volume, if there is any opening to room within the extent of a Division 1 or Zone 1 location 2 2 Entire volume, if there is any opening to room within the extent of a Division 2 or Zone 2 location Inside tank 1 0 Entire inside volume Shell, ends, roof, dike area Tank, aboveground, shop-fabricated, secondary containment tank used for the storage of Class I motor fuels. Shell, ends, roof 1 1 Entire space within dike, where dike height exceeds distance from tank shell to inside of dike wall for more than 50 percent of tank circumference 2 2 Entire space within dike, where dike height does not exceed distance from tank shell to inside of dike wall for more than 50 percent of tank circumference Within 3 m (10 ft) of shell, ends, or roof of tank Entire inside volume Within 450 mm (18 in) of shell, ends, or roof of tank Vent 1 1 Within 1.5 m (5 ft) of open end of vent, extending in all directions Tank, underground 2 2 Between 1.5 m and 3 m (5 ft and 10 ft) from open end of vent, extending in all directions Inside tank 1 0 Entire inside volume Fill opening 1 1 Entire space within any pit or box below grade level, any part of which is within a Division 1 or Division 2 classified location or within a Zone 1 or Zone 2 classified location 2 2 Up to 450 mm (18 in.) above grade level, extending 1.5 m (5 ft) horizontally in all directions from any tight-fill connection and extending 3 m (10 ft) horizontally in all directions from any loose-fill connection Vent 1 1 Within 1.5 m (5 ft) of open end of vent, extending in all directions Page 112 of 129

113 National Fire Protection Association Report 6 of 42 10/26/2016 3:45 PM Location Vapor processing system Division (Group D) Class I Zone (Group IIA) Extent of Classified Location a 2 2 Between 1.5 m and 3 m (5 ft and 10 ft) from open end of vent, extending in all directions Pits 1 1 Entire space within any pit or box below grade level, any part of which: (1) is within a Division 1 or Division 2 classified location; (2) is within a Zone 1 or Zone 2 classified location; (3) houses any equipment used to transfer or process vapors Equipment in protective enclosures Equipment not within protective enclosure 2 2 Entire space within enclosure 2 2 Within 450 mm (18 in.) of equipment containing flammable vapors or liquid, extending horizontally in all directions and down to grade level 2 2 Up to 450 m (18 in.) above grade level within 3 m (10 ft) horizontally of the vapor processing equipment Equipment enclosure 1 1 Entire space within enclosure, if flammable vapor or liquid is present under normal operating conditions 2 2 Entire space within enclosure, if flammable vapor or liquid is not present under normal operating conditions Vacuum assist blower 2 2 Within 450 mm (18 in.) of blower, extending horizontally in all directions and down to grade level 2 2 Up to 450 mm (18 in.) above grade level, extending 3 m (10 ft) horizontally in all directions Vault 1 1 Entire interior space, if Class I liquids are stored within a For marine application, grade level means the surface of a pier, extending down to water level. b Refer to Figures 8.3.2(a) and 8.3.2(b) for illustrations of classified locations around dispensing devices. c Area classification inside the dispenser enclosure is covered in UL 87, Standard for Power- Operated Dispensing Devices for Petroleum Products. d Ceiling-mounted hose reel. e The terms repair garage, major and repair garage, minor are intended to correlate with Article of NFPA 70. For the purposes of application of this table, these terms do not include Page 113 of 129

114 National Fire Protection Association Report 7 of 42 10/26/2016 3:45 PM associated floor space used for offices, parking, or showrooms. f Includes draining of Class I liquids from vehicles. g Includes fuels such as hydrogen and natural gas, but not LPG. Figure 8.3.2(a) Classified Areas Adjacent to Dispensers. Figure 8.3.2(b) Classified Areas Adjacent to Dispenser Mounted on Aboveground Storage Tank. Exception 1 : The extent of the classified area around a vacuum-assist blower shall be permitted to be reduced if the blower is specifically listed for such reduced distances. Exception 2: For shop-fabricated, secondary containment tanks used for the storing of Class I motor fuels, the extent of the Class I Division 2 location shall be limited to 18 inches from the tank shell. Statement of Problem and Substantiation for Public Comment The Committee was tasked with looking further at this issue which was originally submitted as a public comment. At the direction of the Chair, Randy Moses and RB Laurence met to discuss resolution of the issue. This proposed change acknowledges that small, shop built secondary containment tanks, Page 114 of 129