Assessing the Hazards and Protection Schemes Related to IBC s in Operations Scenarios

Transcription

1 Assessing the Hazards and Protection Schemes Related to IBC s in Operations Scenarios Phase II Testing Final Report Prepared by: Christopher Mealy Joseph L. Scheffey Hughes Associates, Inc. The Fire Protection Research Foundation One Batterymarch Park Quincy, MA, USA foundation@nfpa.org Copyright Fire Protection Research Foundation September 2012

2 FOREWORD NFPA 30 Flammable and Combustible Liquids Code [1] provides specific guidance for both containment and fire protection of Listed IBCs containing flammable and combustible liquids in storage configurations. However, a common usage scenario involves the use of non-listed, composite IBCs containing flammable/combustible liquids in operations scenarios. The code does not provide specific fire protection criteria for these applications. In 2011, the Foundation initiated a research program to investigate the hazards of combustible liquids in composite intermediate bulk containers (IBCs) in operations scenarios. Both passive and active fire detection/suppression systems were considered when identifying potential mitigations strategies. The report concluded that the hazards associated with the use of composite IBCs in operations scenarios were severe and recommended that several different mitigations strategies be explored via full-scale testing. The use of a passive cellular glass insulation material, combined with a containment vessel, was identified as a potentially effective strategy. This project outlines the results of Phase II program to address recommendations related to this passive approach made in the Phase I report and build on the findings of the report through full-scale testing. The content, opinions and conclusions contained in this report are solely those of the authors.

3 Assessing the Hazards and Protection Schemes Related to IBC s in Operations Scenarios - Phase II Testing Project Technical Panel Christina Francis, Procter & Gamble Phil Hooker, Health and Safety Laboratory Brian Minnich, Schuetz Containers Systems Louis Nash, U.S. Coast Guard Jon Nisja, Minnesota State Fire Marshal Division Keith Olson, Ansul/Tyco Fire Products Tony Ordile, Haines Fire & Risk Consulting Bob Benedetti, NFPA staff liaison Property Insurance Research Group Sponsors CNA Insurance FM Global Liberty Mutual Tokio Marine Management, Inc. Travelers Insurance XL Group Zurich NA Project Contractor Hughes Associates, Inc.

4 Assessing the Hazards and Protection Schemes Related to IBCs in Operations Scenarios Phase II Testing Prepared for The Fire Protection Research Foundation Quincy, MA Prepared by Christopher Mealy Joseph L. Scheffey Hughes Associates, Inc Commerce Drive, Suite 817 Baltimore, MD Ph. (410) FAX (410) FINAL September 17, 2012

5 TABLE OF CONTENTS Page EXECUTIVE SUMMARY... iv 1.0 BACKGROUND AND OBJECTIVES TEST SERIES 1 LIQUID DISCHARGE CHARACTERIZATION TESTING Experimental Approach Test Configuration Test Procedure Test Results Analysis and Conclusions TEST SERIES 2 POOL FIRE SUPPRESSANT (PFS) EVALUATION Experimental Approach Experimental Materials, Setup, and Procedures Pool Fire Suppressant (PFS) Material Description Test Facility Test Configuration Test Fuels Ignition Scenarios Instrumentation and Data Acquisition Test Procedures Test Results Free-Burn Tests (Baseline Scenarios) Unignited Spill Tests Test 8 4-in. Cellular Glass Insulation System Test 10 2-in. Cellular Glass Insulation System Ignited Spill Tests ANALYSIS AND DISCUSSION REFERENCES APPENDIX A SUMMARY OF HEAT RELEASE AND HEAT FLUX DATA... A-1 ii

6 EXECUTIVE SUMMARY The companion Phase I report to these tests concluded that the hazards associated with the use of non-listed, composite IBCs in operations scenarios were severe and recommended that several different mitigations strategies be explored via full-scale testing. The use of a passive cellular glass insulation material, combined with a containment vessel, was identified as a potentially effective strategy. The first test series in this Phase II effort was intended to characterize the potential liquid discharge (leak) scenarios from various size puncture areas in the side of composite IBCs. The second series of tests was used to evaluate, at a reduced scale, the viability of a passive pool fire suppressant material, a cellular glass insulation, to significantly reduce the thermal threat resulting from an ignited spill or pool fire. For the leak tests, the predictions from Phase I were validated. Adequate containment against potential IBC leak scenarios would require that the IBC be located in the center of a 12 ft square containment area if a single IBC is being protected. In the passive material tests, two scenarios were investigated: an unignited spill (ignition occurring after the spill has drained into the containment vessel; and, ignited spill, where an ignition source is intimate with leaking fuel. This second scenario is of particular interest for the IBC operations scenario, where a leak may occur in the container liquid space from a leaking valve or container puncture, as characterized in the leak tests. A Class IB flammable liquid, heptane, was used in the tests. For unignited spills, the cellular glass insulation was very effective in reducing the overall heat threat. The rate of heat release was an order of magnitude less than the free burn. Heat flux to personnel and equipment were within acceptable limits. The cellular glass insulation was less effective for the ignited spill scenario. The qualitative reductions in heat release rate were not achieved; the resulting ignited spill fire had a heat release of at least 50% of the baseline burns. Personnel could not approach safely within 6 feet of the containment area. While nearby combustibles might not ignite, sprinklers in a 20 ft high building are likely to activate. Potential paths forward include investigation of higher flash point liquid scenarios. Other protection options include those identified in Phase I. Use of Listed/Approved IBCs remains an option. iii

7 1.0 BACKGROUND AND OBJECTIVES In 2011, the Fire Protection Research Foundation (FPRF) initiated a research program to investigate the hazards of combustible liquids in composite intermediate bulk containers (IBCs) in operations scenarios. An assessment of this hazard and potential mitigation strategies was performed and reported on by Hughes Associates Inc. [1]. The investigation was limited to the hazards associated with the use of both flammable and combustible liquids (i.e., Class IB IIIB) in non-listed/approved IBCs (i.e., Type 31HA1\Y). Hazard assessments were made based on thermal insult to the structure and thermal insult to neighboring combustibles. Both passive and active fire detection/suppression systems were considered when identifying potential mitigations strategies. The report concluded that the hazards associated with the use of composite IBCs in operations scenarios were severe and recommended that several different mitigations strategies be explored via full-scale testing. The use of a passive cellular glass insulation material, combined with a containment vessel, was identified as a potentially effective strategy. The intent of the insulation material would be to significantly reduce the thermal threat from a flammable or combustible liquid which might leak and ignite from a breached or punctured IBC. The purpose of this work was to address recommendations related to this passive approach made in the Phase I report [1] and build on the findings of the report through full-scale testing. Two different test series were conducted. The first test series was intended to characterize the potential liquid discharge scenarios from various size puncture areas in the side of composite IBCs. Theoretical calculations for the potential discharge distances were performed in the Phase I report, however, it was concluded that additional testing would be beneficial to confirm these results. The second series of tests was used to evaluate, at a reduced scale, the viability of a pool fire suppressant (PFS) material (FOAMGLAS ), identified as a passive fire protection scheme in the Phase I report. FOAMGLAS is a registered trademark of Pittsburgh Corning Corporation in the United States and other countries. The objective of this Phase II testing was two-fold. The first objective was to characterize the lateral discharge distance of liquid ejected from various size puncture areas created at different elevations on the composite IBC. The second objective was to document the performance of the FOAMGLAS material under various fire exposure and material configuration conditions. 2.0 TEST SERIES 1 LIQUID DISCHARGE CHARACTERIZATION TESTING In the event of a continuous liquid release from an opening in the sidewall of an IBC, it is likely that the liquid will have some lateral velocity due to the head pressure within the vessel. Under this assumption, the approach typically used for environmental protection concerns (i.e., raised sump), in which the containment area is only slightly larger than the IBC, is insufficient to capture the ejected liquid. It must have a larger footprint to capture all discharged liquid. This lateral discharge distance is dependent on the pressure in the ullage of the IBC, the head pressure due to the height of liquid above the opening, the height of the opening above the floor, and the fluid dynamics properties of the liquid. In Phase I, some assumptions were made to simplify the process. The pressure in the ullage of the IBC was assumed to be ambient (i.e., 14.7 psi). The impact of changes in liquid properties was assumed to be well represented by 1 HUGHES ASSOCIATES, INC.

8 changes in the discharge coefficients. The worst-case leak scenario was identified as an opening in the IBC at the mid-height of the container (i.e.23 in.). At points lower on the IBC, the discharged liquid reaches the ground sooner due to the reduced height from which it is released. At points higher, the head pressure (i.e., pressure resulting from the weight of liquid) is reduced and the liquid is not discharged with as much force. Based on these assumptions and the configurations, a series of calculations were performed to estimate distances under representative leak scenarios. In Series 1, the discharge distances from various size puncture holes in an actual IBC, at different heights, were measured to provide a quantitative basis for the containment area estimates from Phase I. 2.1 Experimental Approach Test Configuration IBC liquid discharge characterization testing was performed in the Hughes Associates Inc. (HAI) laboratory located in Baltimore, MD. The variables considered in this testing were puncture area, puncture height relative to the base of the IBC, and IBC elevation relative to the floor. A 275 gal Schuetz Model SX-EX composite IBC, shown in Figure 1, was used in all tests. Figure 1. Composite IBC with different hole scenarios shown. As shown in Figure 1, holes were manually created in the side wall of the IBC. These areas were created using a straight razor so that an exact hole cross-section resulted. The hole areas considered in this test were (0.25 and 4 in. 2 ). As shown in Figure 1, the smaller puncture area was circular in shape while the larger area was rectangular. The smaller area was selected to represent a piercing object with a cross-section comparable to a writing pen. The larger area was selected to represent a piercing object with a cross-section comparable to a forklift tine. The holes were created at elevations of (15, 24.5, and 34.5 in. from the bottom of the IBC frame and were offset 6 in. from the vertical centerline of the IBC. For the majority of the testing 2 HUGHES ASSOCIATES, INC.

9 conducted the IBC was located on the floor of the laboratory therefore the hole elevations were true elevations. In the final two tests, the IBC was elevated 1.2 m (46 in.) above grade. In these tests, hole elevations were 61, 70.5, and 80.5 in. Water was used as the discharging liquid in all tests to provide conservative estimates of discharge distances. Conservative estimates were expected because the density of water is generally higher than the majority of liquids stored in IBCs for use in operations. The trajectory (i.e., x and y travel distances) of the ejected liquid was documented using visual observations and video image analysis. Tests were documented by positioning the camera such that the entire path of travel (i.e., puncture orifice to ground) of the liquid stream was captured. The camera was positioned as close to the end of the discharged liquid stream as possible to reduce potential issues associated with parallax when analyzing video images. Liquid travel distances were measured against a gridded surface that was installed opposite the camera (i.e., on the other side of the liquid stream). The gridded surface was divided into 0.15 m (6 in.) squares in both the vertical and horizontal directions (see Figure 2). Figure 2. Experimental setup showing gridded measurements surface behind discharging liquid Test Procedure All tests were conducted using the same general test procedure. Prior to filling the IBC, all puncture areas sealed using a single layer of foil tape. Once sealed, the IBC was filled with approximately 275 gallon of water using the laboratory water supply. Consistent fill volumes were measured using the fill line provided on the top of the IBC. Once filled, video recording was started and the seal was manually removed. The seal was removed in a single motion such that the entire puncture area was instantaneously available for liquid flow. The trajectory of the liquid stream was noted and permitted to flow until the water level within the IBC reached the point of discharge and ceased flowing. 3 HUGHES ASSOCIATES, INC.

10 2.2 Test Results A total of fifteen tests were conducted to characterize the lateral discharge distance of liquid spilling from a composite IBC under various spill scenarios. With the exception of Test ID 1-1, each scenario was tested in duplicate. An additional test was performed for Test ID 1-1 to explore the impact of sealing the IBC while discharging. Further discussion of this additional test is provided below. A summary of the test variables and corresponding discharge distances are provided in Table 1. Photographs illustrating the discharge stream associated with each puncture scenario are provided in Figure 3 and Figure 4. Test ID 1-1A Table 1. Summary of Lateral Discharge Distances from IBC IBC Elevation (in.) Puncture Area (in. 2 ) Puncture Height as a Fraction of Total IBC Height(-) Measured Lateral Discharge Distance (in.) B C A B A B A B A B 46 Average Lateral Discharge Distance (in.) 1-6A A In general, duplicate testing showed repeatable results with differences in lateral discharge distance ranging from 1 3 inches. Whether or not the leaking IBC was sealed or open to ambient conditions was evaluated in Test ID 1-1 to determine if this condition had any impact on the lateral discharge distance of the ejected liquid. In tests 1-1A and 1-1C the IBC was not sealed during the liquid discharge, while in Test 1-1B, the IBC was kept air tight. As shown in Table 1, the difference in lateral discharge distance between the open and closed condition (i.e., 3 inches) was comparable to the natural repeatability between all tests with identical test parameters (i.e., variance of 1 3 inches) HUGHES ASSOCIATES, INC.

11 Scenario 1-1 Scenario 1-2 Scenario 1-3 Scenario 1-4 Scenario 1-5 Scenario 1-6 Figure 3. Representative photographs of liquid spills from each spill scenario at floor level. 5 HUGHES ASSOCIATES, INC.

12 Scenario 1-7 Scenario 1-8 Figure 4. Representative photographs of liquid spill from stacked configurations. 2.3 Analysis and Conclusions The lateral discharge distances associated with eight different composite IBC puncture scenarios were characterized. For IBCs located on the floor, the maximum discharge distance observed was 41 inches (3.4 ft). This distance was achieved using a 0.2 in. 2 hole located at the mid-height of the IBC. The average discharge distance for this area was 35 ± 6 inches (3.0 ± 0.5 ft). Discharge distances for the larger puncture area (i.e., 4 in. 2 ) were on average slightly greater with an average value of 37 ± 7 inches (3.0 ± 0.6 ft). In general, the results from both puncture scenarios were comparable despite the differences in opening area. Puncture elevation was found to have the greatest impact on lateral discharge distance. For both puncture areas, the greatest lateral discharge distance was measured at the mid-height of the 6 HUGHES ASSOCIATES, INC.

13 IBC. The next largest distance was observed for the 0.25 elevation and the shortest distances were consistently observed at the 0.75 elevation of the IBC. In the stacked configuration, the lateral discharge distance increased by 50 percent, to 60 inches, for the smaller puncture area. A discharge distance of 55 inches was measured for the 4.0 in. 2 puncture area, a 22 percent increase over the distance measured with the IBC on the floor. Based on these results, adequate containment against potential IBC puncture scenarios would require that the IBC be located in the center of a 12 ft square containment area if a single IBC is being protected. This is equal to the Phase I prediction. Stacked IBC configurations would require a slightly larger containment area with a footprint of approximately 15 ft square. These containment areas would capture all of the spills documented in this work and provide an additional 6 inches of added capture area. 3.0 TEST SERIES 2 POOL FIRE SUPPRESSANT (PFS) EVALUATION In Phase I, a passive fire suppression method was also explored. This approach potentially provides a means of mitigating the development of the liquid fuel fire without requiring active fire suppression such as sprinklers. This method consists of using FOAMGLAS material, an inorganic, closed-cell, cellular glass insulation material to effectively reduce the total area of fuel available for combustion. This approach was attractive since it potentially provides a relatively simple and reliable form of passive fire protection. When used in conjunction with a spill containment system, it might provide a means of reducing the burning area and burning rate of a pool fire which in turn reduces the thermal insult produced by the fire. The glass insulation is installed within the containment area. In the event of a liquid release, the material will float on top of the liquid surface to form a solid foam layer. This test series was used to evaluate the passive concept for the IBC operations scenario. Two spill scenarios were investigated: an unignited spill (ignition occurred after the spill had drained into the pan); and, ignited spill, where an ignition source was intimate with leaking fuel. This second scenario is of particular interest for the IBC operations scenario, where a leak may occur in the container liquid space from a leaking valve or container puncture (as characterized in the Series 1 discharge tests). Although expressly prohibited by NFPA 30, Class IB liquids were used as the source fuel, immiscible heptane was used, with the assumption that if the FOAMGLAS performed well with this volatile fuel, it would work with lower volatility fuels. Isopropyl alcohol was used as a miscible fuel, since alcohols are anecdotally known to be stored in IBCs. As in most combustible and flammable liquid tests, an ignition source was assumed to be present. No attempt was made to characterize the likelihood of ignition potential. 3.1 Experimental Approach The variables considered in this test series included fuel type, PFS material arrangement / installation, and the ignition/spill scenario. A matrix of the tests conducted with associated test variables is provided in Table 2. 7 HUGHES ASSOCIATES, INC.

14 Test ID Table 2. Summary of Series 2 FOAMGLAS Test Scenarios Purpose of Test Fuel Fuel Qty. (Gal.) Unignited Spill (U) Ignited Spill (I) FOAMGLAS Configuration FOAMGLAS Thickness (in.) 1 1 U N/A 2 Heptane Free-Burn 2 U None N/A 3 4 Evaluate FOAMGLAS in PE bags Evaluate FOAMGLAS loose configuration in Heptane 2 U 1.5-in. cubes (in bags) 2 U Cubes (loose) ~12 5 Alcohol Free-Burn IPA 1 U None N/A Repeat of Test 3 with 'tighter' installation of cubes Repeat of Tests 3 & 6 with partial submergence of material U 1.5-in. cubes 8 (in bags), extra U cubes added to ensure tight fit Repeat of Tests 3 & 6 with 4- in. thick FOAMGLAS Scoping test of potential IBC container with 4-in. thick FOAMGLAS U I 12 x 12 x 4-in. bricks Repeat of Test 8 with 2-in. thick FOAMGLAS Scoping test of potential IBC container with 2-in. thick FOAMGLAS and Suppression Heptane 2 U I 4 x 12 x 2-in. bricks 2 12 Heptane Free-Burn U None N/A Scoping test of potential IBC container with 2-in. thick FOAMGLAS with 0.25 in. uniform spacing Scoping test of potential IBC container with 2 layers of 2- in. thick FOAMGLAS with 0.25-in. uniform spacing and staggered seams I I 4 x 12 x 2-in. bricks with uniform 0.25-in. spacing 2 layers of 4 x 12 x 2-in. bricks with uniform 0.25-in. spacing with staggered seams HUGHES ASSOCIATES, INC.

15 3.2 Experimental Materials, Setup, and Procedures Pool Fire Suppressant (PFS) Material Description The material being evaluated in this testing was FOAMGLAS Pool Fire Suppressant (PFS) a product manufactured by Pittsburgh Corning, Pittsburgh, PA. FOAMGLAS, a registered trademark of Pittsburgh Corning Corporation in the United States and other countries, is a cellular glass insulation system designed to mitigate the hazard associated with liquid fuel fires by reducing the amount of surface area available for burning. The FOAMGLAS material, herein referred to as cellular glass insulation, is originally manufactured in large blocks (approximately 18 x 24 inches with thicknesses are large as 7 inches). These blocks can be cut to various sizes and thicknesses for various applications. A description of the various material configurations tested is provided in the individual test descriptions. Usually, the commercial cellular glass product is sold in nominally 8-in.cubed UV resistant plastic bags containing approximately 125 smaller cubes (nominal dimension of 1.25-in.cubed). These cubes are loosely packed within the bag (Figure 5). Figure 5. FOAMGLAS material in commercial packaging (left) and opened to show smaller cube material contained within (right). It was found during testing that the smaller cubes in the bag were not the best configuration for this application. During testing, solid blocks, measuring 12 x 12 x 4 inches thick were obtained. In some of the remaining tests, these blocks were then cut to create the desired dimensions Test Facility Fire testing to evaluate the performance of the PFS material was performed at the Naval Research Laboratory (NRL) fire test facility located in Chesapeake Beach, MD. All testing was performed inside the 50 ft square by 35 ft tall fire test facility. Test fires were centered beneath a 9 HUGHES ASSOCIATES, INC.

16 15 ft square hood calorimeter located inside the test facility. The calorimeter was instrumented and operated in general accordance with ASTM E2067 [2]. Exhaust for the calorimeter was provided by a variable speed exhaust fan capable of exhausting up to 30,000 cfm. The calorimeter was used in all testing to remove fire effluent from the test facility and measure the associated heat release Test Configuration All testing was conducted using a 5 ft square (25 ft 2 ) stainless steel pan with a 1 ft lip height. The pan was elevated approximately 8 inches above the floor of the test facility using concrete block. A one-inch water substrate was added to the pan prior to all heptane tests conducted. The water substrate provided a level substrate onto which both the fuel and PFS material could be floated. The water was also used to prevent thermal distortion of the pan during testing. No water substrate was used for the alcohol test. A steel grating was installed over the top of the pan in select tests (Tests 9, 11, 13, 14). This grating was used to suspend 5 gallon plastic containers above the FOAMGLAS material installed in the base of the pan. The grating was constructed from steel and consisted of 1 in. by in. bearing bars spaced nominally 1 in. on center running parallel to the length of grate with cross bars spaced every four inches. A photograph of the grating used is provided in Figure 6. Figure 6. Steel grating installed over top of test pan. Since these tests were considered scoping tests, containers simulating IBCs, but much smaller in size, were used to investigate a leaking IBC scenario. If the results with smaller containers were found to be acceptable, non-listed composite IBCs were on hand to perform a larger test. The plastic containers used were Nampac model M-4201, constructed from high density polyethylene (HDPE) having a capacity of 5 gallons. For all tests, the 10 in. x 10 in. x 14 in. tall containers were filled with 2.5 gallons of heptane and offset 6.5 inches from one another. The containers were oriented such that the pour mouths were facing one another. In all tests, the screw-fit cap for the container was twisted one full turn to allow the container to vent when exposed to the initiating fire. Baseline heat release rate data, as shown in Figure 7, was collected for the two plastic containers using the heptane pan ignition source. Baseline data showed that 10 HUGHES ASSOCIATES, INC.

17 the containers produced a peak fire size of approximately 160 kw with a burning duration of approximately 20 minutes. Heat Release Rate (kw) Time (s) Figure 7. Empty plastic container heat release rate Test Fuels The primary fuel used in this test series was commercial grade heptane. The heptane used was purchased from Tilley Chemical Company. The fuel has a reported flash point of 15ºF. Although not permitted to be used / stored in composite IBCs according to NFPA 30, heptane was selected for use in these tests to provide a conservative representation of the potential fire scenarios that could be developed by fuels traditionally allowed in these types of IBCs. Isopropyl alcohol (IPA) was used as a representative miscible fuel. The IPA used in these tests was 99% pure isopropanol, with a flash point of approximately 53 F Ignition Scenarios Two different ignition scenarios were used; unignited and ignited spills. The unignited scenario involved a two gallon heptane spill which was manually poured onto the cellular glass from an elevation of approximately 3 ft., generally at the north, center section of the pan. This spill was subsequently ignited using a small torch flame consisting of a small quantity of cotton rag wrapped around the end of a steel pipe and soaked in heptane. The ignited spill was intended to create a container failure scenario where leaking fuel was immediately ignited by a flame source. This scenario consisted of a 6 in. x 8 in. by 2 in. deep steel pan filled with approximately 0.29 gallons of heptane, position between two plastic 5 gallon containers as shown in Figure 6. The pan of fuel was ignited and permitted to burn freely for the duration of the test. This scenario produced an estimated 20 kw exposure fire that burned for 11 HUGHES ASSOCIATES, INC.

18 approximately 3 4 minutes, exposing the fuel-filled containers. The containers ultimately breached, and the discharging heptane stored in the containers ignited Instrumentation and Data Acquisition Instrumentation used in this test series consisted of the hood calorimeter performing oxygen consumption calorimetry and two heat flux transducers measuring incident heat flux from the pool fire scenarios. The hood calorimeter was instrumented to measure exhaust gas temperature, flow, and gas species data in general accordance with ASTM E2067 [2]. Gas temperature was measured using single, bare-bead, Type K, inconel sheathed thermocouple installed at the center of the crosssection of the duct. Gas flow conditions are recorded using a bi-directional probe installed at the center of the duct that was connected to a Setra Model 264 pressure transducer with a maximum range of ±5 in H 2 O and measurement accuracy of 0.25 percent full scale. Gas species concentrations were measured using a Rosemount Model NGA 2000 gas analyzer. The analyzer measured oxygen, carbon dioxide, and carbon monoxide concentrations. Incident heat flux measurements were collected using two horizontally-mounted, Schmidt- Boelter type, water-cooled heat flux transducers. The transducers were installed 3.3 and 6.6 ft from the edge of the test pan and offset 45 degrees from one another. The transducers were offset from one another to prevent the near transducer from obstructing the view of the far transducer. Both transducers had a range of kw/m 2 and were elevated 3 ft above the base of the pan (i.e., 2 ft above the lip of the pan). All collected during tests was collected at a frequency of once per second for the test duration using a National Instruments SCXI-1000 chassis with one SCXI-1303, 32-channel isothermal terminal block and one SCXI-1327, 8-channel high-voltage attenuator terminal block Test Procedures A general set of procedures were followed prior to, during, and after each test. Test-specific procedures are described for individual tests, but there was little variation except for the set up of the cellular glass. For all tests except the IPA test, a one in. deep water sub-layer was added to the test pan after installing the appropriate cellular glass insulation configuration. Steel grating and plastic containers were installed over top of the cellular glass insulation, when appropriate. Two minutes of background data was collected prior to each test. One minute prior to ignition, test video was started. Thirty seconds prior to ignition, two gallons of heptane was poured into the pan, on top of the cellular glass insulation. The heptane spill was ignited at time zero signifying the start of each test. For the tests where ignited spills were investigated, the exposure pan ignited simultaneously with the ignition of the two gallon spill. With the exception of Test 11, all test fires were permitted to burn uninhibited (i.e., no suppression) until they were either selfextinguished or had burned for a minimum of 20 minutes. After manual extinguishment, the majority of the scenarios evaluated were tested for re-flash potential using the torch flame described in Section The torch flame was held over top of the cellular glass insulation material to determine if fuel vapors were ignited. 12 HUGHES ASSOCIATES, INC.

19 3.3 Test Results A total of fourteen fire tests were conducted to evaluate the suppression performance of the cellular glass material. Four of the tests conducted were free-burning fires that were used to establish baseline fire size and thermal exposure conditions in an unprotected scenario. Of the remaining ten tests, six explored different configurations of the cellular glass system with the unignited spill, and four explored the impact of burning fuel containers located above the cellular glass system in an ignited spill scenario. The fire suppression capability of the FOAMGLAS material was evaluated based on several quantitative and qualitative criteria. The reduction in the overall heat release rate (HRR) of the fire and the reduction in incident heat flux to adjacent objects were quantitatively assessed. The objective in both cases was to reduce the thermal threat to the structure, and to personnel and equipment near the operation. For rate of heat release, free burns of the heptane pan fire provided the baseline for assessment. Qualitative measures of 50 and 75 % reductions in the HRR were established. Another measure is the reduction of HRR sufficient to prevent actuation of overhead automatic sprinklers. Using the calculation approach from Phase I, a limit of 2.2 MW was established for standard response sprinklers, with an activation temperature of 275 o F, a ceiling height of 20 feet, and an activation time of 60 seconds. The reduction in heat flux to adjacent objects was evaluated against two different thresholds as established in Phase I. A threshold associated with instantaneous human pain (2.5 kw/m 2 ) [3] and with the ignition of neighboring plastic materials (20 kw/m 2 ) were used for comparative purposes, at 3.3 and 6.6 ft ( 1 and 2 m) from the containment area. Flame height was also used as a qualitative measure of performance. A summary of the tests conducted and pertinent results from these tests is provided in Table 3. Plots of heat release and heat flux are provided in Appendix A Free-Burn Tests (Baseline Scenarios) Four free-burning heptane fires were conducted, Tests 1, 2, 5 (IPA), and 12. These fires were conducted to provide baseline fire hazard data characterizing the incident heat flux to adjacent objects when subjected to an unmitigated liquid fuel pool fire. To ensure that steadystate heat flux measurements were collected, spill volumes of 1 and 2 gallons were evaluated. For the heptane tests, one gallon of fuel was used in Test 1. The fire did not burn long enough so an additional gallon of fuel was used in Tests 2 and 12. Test 12 was conducted as a repeat of Test 2, and as a demonstration for visiting technical panel members and manufacturers on - site during this test. 13 HUGHES ASSOCIATES, INC.

20 Test ID Purpose of Test Table 3. Summary of Fire Suppressant (PFS) Material Testing Fuel Fuel Qty. (Gal.) FOAMGLAS Configuration Baseline Free Burns FOAMGLAS Thickness (in.) Peak HRR (kw) Peak Heat Flux at 1 m (kw/m 2 ) Peak Heat Flux at 2 m (kw/m 2 ) 1 1 N/A Heptane Free-Burn None 2 Heptane 2 N/A Heptane Free-Burn 2 None N/A Alcohol Free-Burn IPA 1 None N/A 770 < 1 < 1 Unignited Spill Tests 3 Evaluate FOAMGLAS in PE bags 1.5-in. cubes (in bags) ~ 1 < 1 4 Evaluate FOAMGLAS in loose configuration cubes (loose) ~ ~1 <1 Repeat of Test 3 with 'tighter' installation < 1 < 1 of cubes 1.5-in. cubes (in bags), extra Repeat of Tests 3 & 6 with partial Heptane 2 cubes added to ensure tight fit < 1 < 1 submergence of material 8 Repeat of Tests 3 & 6 with 4-in. thick FOAMGLAS 12 x 12 x 4-in. bricks < 1 < 1 10 Repeat of Test 8 with 2-in. thick FOAMGLAS 4 x 12 x 2-in. bricks < 1 < Scoping test of potential IBC container with 4-in. thick FOAMGLAS Scoping test of potential IBC container with 2-in. thick FOAMGLAS and Suppression Scoping test of potential IBC container with 2-in. thick FOAMGLAS with in. uniform spacing Scoping test of potential IBC container with two layers of 2-in. thick FOAMGLAS with 0.25-in. uniform spacing and staggered seams Heptane 2 Ignited Spill Tests 12 x 12 x 4-in. bricks x 12 x 2-in. bricks x 12 x 2-in. bricks with uniform 0.25-in. spacing 2 layers of 4 x 12 x 2-in. bricks with uniform 0.25-in. spacing with staggered seams HUGHES ASSOCIATES, INC.

21 Representative photographs depicting the progression of these free-burning fires are presented in Figure 8. Figure 8. Evolution of heptane free-burn fire (Test 1) at 3s, 30s, 55s, and 60s after ignition (left to right). The baseline heptane fires burned for between seconds before self-extinguishing. Peak fire sizes during this burning duration ranged from MW with average peak heat fluxes of 23 kw/m 2 and 10 kw/m 2 at distances of 3.3 and 6.6 ft, respectively. Flame height was feet. The fire sizes measured in these tests were approximately 50 percent of that predicted using the known pan area and the mass burning rate per unit area (55 g/s-m 2 ) for heptane available in the literature. This reduction is attributed to the relatively thin fuel layer (i.e., 2-3 mm) present in the pan during these tests. It has been shown that fuel layer thicknesses in this range produce smaller fire sizes than expected due to insufficient burning durations which limit the extent to which the fuel can reach a steady-state burning rate [4]. Test 5 was a free-burning isopropyl alcohol fire. It was conducted using only one gallon of IPA due to concerns of thermally damaging (i.e., warping) the steel pan. The miscibility of the fuel prevented the fuel from being poured onto a water substrate. The IPA was poured directly into the pan. Due to the uneven nature of the bottom of the test pan, this approach resulted in localized pooling of the fuel in the pan. The greater fuel depths in these localized areas resulted in the burning duration for this test being approximately 120 seconds. A series of photographs depicting the progression of the IPA free-burning fire is presented in Figure 9, were captured at 30 second intervals after ignition (top left to bottom right). The peak fire size in this test was 0.8 MW with peak heat fluxes of less than 1kW/m 2 at distances of 3.3 and 6.6 ft, respectively. Based on the burning duration, fire intensity, and inability to use a water substrate for the IPA fires, it was concluded that there would be no further testing using this fuel. There were sufficient issues to address with the heptane fires. 15 HUGHES ASSOCIATES, INC.

22 Figure 9. Evolution of IPA free-burn fire at 30s intervals from ignition to self-extinguishment Unignited Spill Tests Tests 3, 4,6,7,8, and 10 were conducted using the unignited spill scenario. The primary variable was the size and arrangement of the cellular glass insulation Test 3 Cellular Glass Insulation in Plastic Bags (Loose) Test 3, the first test conducted to evaluate the performance of the cellular glass insulation, was installed in the commercial packaging material as shown in Figure HUGHES ASSOCIATES, INC.

23 Figure 10. Pre-test configuration of cellular glass insulation system in commercial packaging. A 6 x 7 array (total of 42 bags) was installed such that a relatively tight fit was achieved and no additional bags, as provided by the manufacturer, could be added. The two gallon heptane spill was manually poured into the test pan and ignited. Once ignited, a flash fire burned for approximately seconds on top of the bags before decaying. After the decay of the flash fire, flaming was observed primarily in the void spaces between bags with some flame attachment to the plastic bags themselves. This type of burning was observed for the duration of the test with relatively no increase or decrease in fire intensity. A series of photographs showing the evolution of the heptane fire in this test are presented in Figure 11, taken 10 seconds, 2 minutes, and 22 minutes after ignition. Figure 11. Stages of burning observed in Test 3. This fire was permitted to burn for thirty minutes before it was manually extinguished. The initial flash fire was about 125kW. The maximum fire size was 0.3 MW, not associated with the flash fire. It occurred 8 minutes 37 seconds after ignition and was attributed to the burning of the heptane fuel and plastic bag material. An average fire size of 0.15 MW was measured over the duration of this test. With the exception of the flash fire, average flame heights in this test did 17 HUGHES ASSOCIATES, INC.

24 not exceed 3 ft. Maximum heat fluxes at the 3.3 and 6.6 ft offset distances did not exceed 2.5 kw/m 2 at any point during this test. Manual extinguishment was achieved using a hand-held garden hose line. Extinguishment was achieved in less than five seconds and the fire remained extinguished in the absence of a pilot. Re-flash testing was conducted approximately 30 seconds after manual extinguishment was achieved and the residual fuel locally ignited in the areas exposed to the torch flame Test 4 Cellular Glass Insulation (Mixed Configuration) After Test 3, the cellular glass insulation material that remained in the test pan was relatively loose with structured vertical and horizontal gaps present between the nominal 1.25 in. cubes (see Figure 12). Based on observations made during Test 3 regarding the presence of flaming at gap locations and the ability of the loose material to be mixed (as show in Figure 13), it was decided that Test 4 would be conducted with the cellular glass material in a loose formation, similar to stones used to suppress transformer oil spill fires. In an effort to limit the ability of the fuel sub-layer to burn within gaps created by the standard bags, the cellular glass cubes were manually turned over so no continuous gaps were present Figure 13. By limiting the availability of gaps within the insulation system, it was postulated that the ability of the fuel vapor to reach combustion air and the ability of the radiant heat from the fire to reach the fuel sub-layer would be inhibited. Figure 12. Configuration of cellular glass insulation system after Test HUGHES ASSOCIATES, INC.

25 Figure 13. Mixed configuration of cellular glass insulation used in Test 4. The loose installation was inches thick. The standard two gallon heptane spill fire ignition scenario was used to initiate this fire. The initial flash fire in this test (250 kw) was larger than that observed in Test 3. This increase in fire size was attributed to the roughness of the exposed cellular glass insulation material potentially retaining, or simply retarding, the extent of runoff of the spilled liquid. The location of the burning observed in this test differed from Test 3 in that the vast majority of the burning occurred along the perimeter of the pan. A series of photos depicting the progression of this fire test are presented in Figure 14, captured at the 10 seconds, 2 minutes, and 22 minutes after ignition (same sequence as Test 3, Figure 11). Figure 14. Stages of burning observed in Test 4. This fire was permitted to burn for thirty minutes before it was manually extinguished. A maximum fire size of 0.3 MW was achieved during this test. Similar to Test 3, this peak value was not associated with the flash fire; it was observed approximately 20 minutes after ignition and was associated with increased flaming along the perimeter of the pan. An average fire size 19 HUGHES ASSOCIATES, INC.

26 of 0.19 MW was measured over the duration of this test, similar to Test 3. During peak burning conditions, average flame heights of 3 4 ft were noted. Maximum heat fluxes at the 3.3 and 6.6 ft offset distances again did not exceed 2.5 kw/m 2 at any point during this test. The fire reflashed during the post-extinguishment torch test. The enhanced burning that was observed along the perimeter of the pan was attributed to the inability of the non-uniform configuration to consistently seal the air gap between the insulation material and the side wall of the pan. This air gap allowed fuel vapors to mix with fresh air and burn at these locations. In addition, several void spaces were identified over the course of the test where localized pockets of burning were observed Test 6 Cellular Glass Insulation (Tightly Packed) Observations from Test 3 were considered and an approach was developed to better fit the test pan with cellular glass insulation in Test 6. Additional material was added to ensure a relatively tight fit was achieved. In this test, the pan was outfitted with a 6 x 7 array of the nominal 8-in. thick commercially available cellular glass product, just as was used in Test 3. However, once installed, these units were manually pushed together against two sides of the test pan so there were no gaps. This resulted in a nominal 3.5 in. gap along the opposite pan walls. To eliminate this open are, these gaps were filled with loose cellular glass insulation cubes (Figure 15). Figure 15. Tightly packed cellular glass insulation material consisting of 6x7 array of prepackaged cubes bordered on two sides by loosely packed cellular glass insulation material. The standard two gallon heptane spill fire ignition scenario was used to initiate this fire. As expected, the initial flash fire was comparable to that observed in Test 3. After this initial flash 20 HUGHES ASSOCIATES, INC.

27 fire, the residual flaming observed was minimal for the duration of the test, with flame heights consistently less than 1 foot in height. The burning behavior observed in this test was the most reduced of any of the configurations tested up this point. Localized areas of flaming were observed but these areas remained relatively isolated. A series of photos depicting the progression of this fire test are presented in Figure 16, captured at 10 seconds, 5 minutes, and 25 minutes and 45 minutes (top left to bottom right) after ignition. Figure 16. Progression of fire in Test 6. To assess the degree of fuel burned and integrity of the cellular glass, this fire was permitted to burn for 47.5 minutes before it was manually extinguished. Extinguishment was achieved in less than five seconds using a hand-held water line. A maximum fire size of less than 0.2 MW and average fire size of less than 0.1 MW was measured over the duration of this test. Maximum heat fluxes at the 3.3 and 6.6 ft offset distances again did not exceed 2.5 kw/m 2 at any point during this test Test 7 Partially Submerged Cellular Glass Insulation This test was designed to evaluate the impact of partially submerging the PFS material. The collection of activated sprinkler water and gradual submergence of the material was identified as a likely event. This could reduce the effective depth of the cellular glass, potentially reducing its effectiveness. Test 7 was designed to determine the extent to which this would impact vapor suppression. An illustration of the experimental setup for Test 7 is provided in Figure HUGHES ASSOCIATES, INC.

28 Figure 17. Illustration of experimental setup for Test 7, with partially-submerged, single layer of cellular glass insulation material. The same cellular glass insulation configuration as in Test 6 was used in this test. With this configuration in place, a steel grating (described in Section 3.2.3) was installed over the top of the pan. The grating was used to restrict the vertical travel of the bags so that when the water level was increased, the bags were submerged. With the grating in place, approximately 80 gallons of water was added to the pan such that the bags were submerged approximately four inches. Once submerged, the two gallon heptane spill ignition scenario was used to start the test. The initial flash fire in this test was the largest of any of the tests conducted using the bagged material. Unlike in the previous two tests, the maximum heat release measured, 0.3 MW was associated with this flash fire and not the fire development later in the test. After the initial flash fire, the fire burned at a relatively steady-state with flames extending through the steel grating to height 1 2 ft above the grating. In general, the areas of burning were relatively localized. A series of photos documenting the progression of this fire is provided in Figure 18, captured at 1 minute, 8 minutes, and 15 minutes and 45 minutes after ignition. 22 HUGHES ASSOCIATES, INC.

29 Figure 18. Progression of fire in Test 7. This fire was permitted to burn for 30 minutes before being manually extinguished. Manual extinguishment was achieved in less than five seconds using a hand-held water line. A maximum fire size of 0.3 MW and an average fire size of less than 0.1 MW were measured during this test. Maximum heat fluxes at the 3.3 and 6.6 ft offset distances again did not exceed 2.5 kw/m 2 at any point during this test. Other than the initial flash fire being slightly larger, the overall suppression performance of the cellular glass material in this test, when submerged to half (4 inches) of the material original height, was generally comparable to that observed for tests in which the full height of the material (8 inches, Tests 3 and 6). From the results of this test, as well as Tests 3 and 6, a decision was made to evaluate solid (i.e., single piece), unwrapped (i.e., no plastic bag) sections of the cellular glass insulation at a thickness of four inches. It was postulated that this configuration would minimize the total gap area within the protection area (i.e., minimize potential for unsealed surface burning). Since the performance of the 4-in. layer of material evaluated in Test 7 was comparable to that measured in tests with eight inches of material, the 4- in. thickness was selected. This material, in 12-in. square blocks, was provided by the manufacturer. 23 HUGHES ASSOCIATES, INC.

30 3.3.3 Test 8 4-in. Cellular Glass Insulation System The configuration used in Test 8 consisted of a total of 25 solid blocks (5 x 5 array) of 4-in. thick, 12-in. square cellular glass material. This configuration provided the tightest fitting cellular glass insulation system to date. When installed on top of the water sublayer, the material remained buoyant, which indicated some degree of freedom between the blocks. A photograph of this configuration is provided in Figure 19. Figure in. thick, solid block cellular glass insulation configuration. The two gallon heptane spill ignition scenario was used to start the test. The flash fire in this test was the largest of any test conducted to date. This was attributed to the overall reduction in gap area and increase in cellular glass insulation surface area which resulted in a greater volume of fuel remaining on top of the material. However, this fire lasted for less than 10 seconds with only flamelets remaining after the decay. These small flamelets were observed in the majority of the gap areas for the duration of the test. A series of photographs illustrating the progression of this fire over the test duration is provided in Figure 20, were captured at 1 minute, 8 minutes, and 15 minutes and 45 minutes after ignition. 24 HUGHES ASSOCIATES, INC.

31 Figure 20. Progression of fire in Test 8. The fire was permitted to burn for 20 minutes before being manually extinguishment. Manual extinguishment was achieved in less than five seconds using a hand-held water line. A maximum fire size of 0.6 MW was measured during the initial seconds after ignition, associated with flash fire. After this, the maximum fire size did not exceed 0.1 MW. An average fire size of less than 0.1 MW was measured during this test. Maximum heat fluxes at the 3.3 and 6.6 ft offset distances again did not exceed 2.5 kw/m 2 at any point during this test. Other than the larger initial flash fire, the performance of the cellular glass insulation configuration evaluated in this test was better than any other configuration tested to date. The increase in the initial fire size was attributed to the inability of the material to provide adequate drainage for the fuel spill. This shortcoming was discussed and several alternative configurations were developed for future tests. However, based on the overall improved performance of the configuration used in Test 8, it was decided that a scoping test using surrogate plastic containers would be conducted using this configuration (see Section 3.3.3, Test 9). 25 HUGHES ASSOCIATES, INC.

32 3.3.4 Test 10 2-in. Cellular Glass Insulation System Test 10 was conducted determine if even a smaller thickness, 2 inches, could effectively suppress the fire. The availability of material and the dimensions to which the material was manufactured required that 4 x 12 in. blocks of two in. thickness material be used in this test. With this size block, a 5 x 13 array was installed within the test pan (Figure 21). The array was installed with the same general spacing between blocks that was used in Tests 8 and 9 (i.e., tight fitting but still buoyant). The gaps between individual blocks were slightly greater than in Test 8. Figure 21. Cellular glass installation used in Test 10. The two gallon heptane spill ignition scenario was used to start the test. The flash fire in this test was slightly smaller than in Test 8, reaching a peak value of 0.4 MW. The slight reduction in this test was attributed to the added gap area present in this test which provided additional means of drainage immediately following the 2 gallon heptane spill. The flash fire subsided within 10 seconds with very minimal flaming remaining. Several photographs showing the extent of this burning at the gaps between the cellular glass insulation are provided in Figure HUGHES ASSOCIATES, INC.

33 Figure 22. Representative photographs of burning observed in Test 10. The fire was permitted to burn uninterrupted for 15 minutes. After 15 minutes, several cellular glass insulation blocks were removed to explore the impact of void spaces in the system. When removed it was noted that flaming would occur in the absence of the block but remained at this location and did not spread to neighboring gaps. After an additional 5 minutes of burning the fire was manually extinguished. Manual extinguishment was achieved in less than five seconds using a hand-held water line. A maximum fire size of 0.4 MW was measured during the initial flash fire. The subsequent burning was less than 0.1 MW. Maximum heat fluxes at the 3.3 and 6.6 ft offset distances again did not exceed 2.5 kw/m 2 at any point during this test. The results of this test demonstrated that 1) a 2-in. layer of the cellular glass insulation material provided flame suppression that was comparable to that of both 4 and 8 in. material layers and 2) the addition of gaps (i.e., breaks in the cellular glass insulation layer) aids in the reduction of the surface burning by providing additional drainage for the spilled fuel Ignited Spill Tests Test 9 Surrogate IBC Containers over 4-in. Cellular Glass System This was the first test to evaluate the performance of the cellular glass insulation system when subjected to a ignited spill fire scenario. Two plastic containers were installed over top of the same cellular glass insulation system used in Test 8. The containers were placed onto the steel grating spanning the top of the test pan. Each container was filled with 2.5 gallons of heptane and offset 6.5 inches from one another. The containers were oriented such that the pour mouths were facing one another. The screw-fit cap was only twisted one full turn to allow for venting when exposed to the initiating fire. Prior to testing, a single layer of 1-foot square by 4-in. thick cellular glass insulation was installed in the test pan. The steel grating was then installed over top of the test pan. With the grating installed, the 2 5 gallon plastic containers were centered on the pan, located approximately nine inches above the top surface of the cellular glass insulation material. The fuel pan located between the plastic containers was then filled with approximately 0.29 gallons of heptane. 27 HUGHES ASSOCIATES, INC.

34 The standard two gallon heptane spill was added to the test pan and ignited. This resulted in the simultaneous ignition of the smaller pan of heptane located between the containers, Figure 22. Figure 22. Heptane pan fire located between plastic containers in Test 9. After ignition, a flash fire comparable to that of Test 8 was observed. This fire reached a peak value of approximately 0.4 MW but quickly decayed to a level of burning comparable to that shown beneath the steel grating in Figure 22. After this decay, the primary burning observed in this test was associated with pan fire located between the plastic containers. Small flamelets at gap locations beneath the grating were observed, but they generally did not reach above the grating. The pan fire, however, continued to burn and gradually involved/breached the plastic containers. The first container was breached 4 minutes 18 seconds after ignition. The fuel within the container began to spill onto the cellular glass insulation below the grating. Shortly after, the flaming was established on top of the cellular glass insulation, due to running fuel. Flames began to impinge on the plastic containers above. At this point in time, flame heights transitioned from 2 3 feet above the steel grating to 5 10 feet above the grating. The fire size increased due to additional spillage of fuel from the breached plastic containers. Although some drainage may have been occurred, the area of the fire on the cellular glass insulation as well as the flame heights associated with the fire gradually increased over the next 3 minutes. Approximately 7 minutes 8 seconds after ignition, 2 minutes 50 seconds after the initial breach of a plastic container, both containers failed catastrophically (i.e., all fuel released). This failure resulted in a dramatic increase in fire size with approximately 50 percent of the pan area becoming involved. 28 HUGHES ASSOCIATES, INC.

35 The fire gradually decayed over the next 4 6 minutes with both total coverage area and flame heights regressing to levels slightly larger than that observed in Test 8. A series of photographs illustrating the progression of this fire is presented in Figure 23, captured at 5 seconds, 4 minutes, 6 minutes, 8 minutes, 10 minutes, 12 minutes, 16 minutes, and 19 minutes (top left to bottom right) after ignition. The fire was permitted to burn for 19 minutes before being manually extinguished. Manual extinguishment was achieved in less than five seconds using a hand-held water line. A maximum fire size of 2.5 MW was measured during this test, approximately 8 minutes after ignition, coinciding with the catastrophic failure of the plastic containers. As a result of the increased fire size in this test, the maximum heat fluxes at the 3.3 and 6.6 ft offset distances reached values of 9 and 4 kw/m 2, respectively. The results of this test demonstrated the inability of this cellular glass insulation configuration to adequately drain the spilling fuel away from the ignition source. Consequently, the fuel was able to burn in place and continue to impinge on the plastic containers located above the spill. This prolonged exposure to flame eventually resulted in the catastrophic failure (instantaneous release) of all of the containers contents which produced a large liquid fuel fire that did not subside for several minutes. 29 HUGHES ASSOCIATES, INC.

36 Figure 23. Progression of fire in Test Test 11 Surrogate IBC Containers over 2 in. Cellular Glass System & Suppression The ignition scenario and general procedure used in Test 11 was identical to that used in Test 9 with the exception that a mock overhead suppression system was used. The cellular glass insulation configuration was the same as used in Test 10 (i.e., 5 x 13 array of 2-in. thick block). The objective of inserting a sprinkler was to determine if water suppression would retard overall fire growth, particularly melting and failure of the plastic containers. The sprinkler was a pendant type overhead sprinkler manufactured by Central having a k-factor of 5.6 and was operated at a flow rate of approximately 7.5 gallons per minute. The sprinkler head was centered four feet above the test pan. In this configuration and based on preliminary flow testing, the majority of the water spray fell into the test pan providing a nominal discharge density of 0.3 gpm/ft 2. The head was approximately three feet above the plastic fuel containers as shown in Figure 24. After ignition, a flash fire comparable to that of Test 10 was observed. This fire reached a peak value of approximately 0.4 MW but quickly decayed so that there were only small flamelets, as observed in Test 10. After this decay, the primary burning was associated with the pan fire located between the plastic containers until the containers breached. A series of photographs illustrating the progression of this fire is shown in Figure 25, captured at 5 seconds, 4 minutes, 6 minutes, 8 minutes, 12 minutes, and 20 minutes (top left to bottom right) after ignition. 30 HUGHES ASSOCIATES, INC.

37 Figure 24. Experimental setup for Test 11. The first container was breached approximately 2 minutes after ignition. The fuel within the container began to spill onto the cellular glass insulation below the grating. The failure mode of the container in this test was different than that observed in Test 9. In this test, the container failed in a manner that released a large volume of fuel rather quickly. This rapid release of fuel produced a relatively large spill fire beneath both containers, with flame heights reaching 8 10 feet above the steel grating. Approximately 30 seconds after the first container was breached the suppression system was manually activated. The suppression system had relatively no impact on the development of the fire. Shortly after activation (approximately 30 seconds), the second container breached and began to contribute the fire. Water was discharged for a total of 8 minutes and 20 seconds before it was secured. During this time several observations were made regarding the response of the cellular glass insulation system to the suppression water. Approximately half way through the eight minute discharge, burning at the fuel surface was observed around the perimeter of the pan. The increased burning in this area was attributed to the perimeter blocks being tilted up toward the center of the pan, and the lower edges at the side of the pan were observed to be partially submerged. This type of burning was observed for the remainder of the test. This tilting may have been a result of: thermal expansion of the pan, causing the perimeter blocks to become wedged; an unexpected side effect from the water discharge pattern of the sprinkler impacting the perimeter blocks; or the use of used blocks from Test 10. Several of these hypotheses were evaluated in Tests 13 and HUGHES ASSOCIATES, INC.

38 Figure 25. Progression of fire in Test 11. The fire was permitted to burn uninterrupted for 21 minutes at which point the fire was manually extinguished. Manual extinguishment was achieved in less than five seconds using a hand-held water line. A maximum fire size of 2.7 MW was measured 4 minutes 37 seconds) after ignition during this test. Maximum heat fluxes at the 3.3 and 6.6 ft offset distances were 8 and 3 kw/m 2, respectively Test 13 Surrogate IBC Containers over 2 in. Cellular Glass System (0.25 in. Gaps) Test 13 was conducted with a test configuration similar to that used in Test 11, only without water suppression and with larger, more uniform gap spacing between the cellular glass insulation. Suppression was not used in this test to evaluate whether or not it was the overhead suppression that caused the relatively poor performance of the material in Test 11. A wider, more uniform gap spacing was used to allow a larger volume of spilled fuel to be drained from the surface of cellular glass in order to prevent residual fuel burning on the surface of the cellular glass, which was contributing to relatively high heat release and flux. The gap spacing used in this test was 0.25-in. This spacing was maintained between each block using cardboard spacers (Figure 26). The two gallon heptane spill ignition scenario was used to start the test. In general, the fire development, from flash fire to container release to spill fire growth, was comparable to that observed in Test 11. This similarity indicated that very little change in fuel drainage between the two tests was achieved and that the impact of the overhead suppression system was minimal. A plot comparing the measured heat release rates for these two tests (Tests 11 and 13) is provided 32 HUGHES ASSOCIATES, INC.

39 in Figure 27. A series of photographs illustrating the progression of this fire is presented in Figure 28, captured at 3 seconds, 4 minutes, 6 minutes, 8 minutes, 12 minutes, and 20 minutes (top left to bottom right) after ignition. Figure 26. Cellular glass insulation system with 0.25 in. spacers Test 11-2in. PFS Layer with Suppression 2500 Test 13-2in. PFS Layer with 0.25in. Spacing 2250 Heat Release Rate (kw) Time (s) Figure 27. Heat release rate of Test 11 vs. Test HUGHES ASSOCIATES, INC.

40 Figure 28. Progression of fire in Test 13. The fire was permitted to burn uninterrupted for 20 minutes at which point it was manually extinguished. Manual extinguishment was achieved in less than five seconds using a hand-held water line. A peak fire size of 2.9 MW was measured at 5 minutes 4 seconds after ignition. Maximum heat flux values of 9 and 4 kw/m 2 were measured at off-set distances of 3.3 and 6.6 feet, respectively. The results of this test demonstrated that 1) the overhead water suppression system did not have a negative impact on the overall performance of the cellular glass insulation system and 2) the addition of slightly larger, uniform gaps (i.e., breaks in the cellular glass insulation layer) provided minimal, if any, improvement in the drainage performance of the material when exposed to a running fuel fire Test 14 Surrogate IBC Containers over 4 in. Cellular Glass System (Layered) The final test involved cellular glass installed using a layered material approach. Based on the prior test results, and discussion with the manufacturer, it was determined that added flame suppression performance could potentially be achieved using a system of two layers of 2 in. thick material. Both layers were installed using the same general configuration used in Tests 11 and 13. However, the second layer was rotated 90 degrees so that the seams between blocks of each layer were perpendicular (Figure 29). While be beneficial from a drainage stand point, parallel seams running the full depth were thought to be detrimental from a flame suppression standpoint. Gaps between blocks were approximately 0.25 inches. In order to maintain optimum drainage 34 HUGHES ASSOCIATES, INC.

41 but limit the extent to which flames could be established on the fuel layer, this perpendicular, two-layer arrangement was created. Figure 29. Lower (left) and upper (right) layers of cellular glass insulation system used in Test 14. The two gallon heptane spill ignition scenario was used to start the test. Fire development was generally comparable to Tests 11 and 13. The added layer of staggered material did not provide a substantial improvement in flame suppression or drainage compared to the previous two test configurations. A series of photographs illustrating the progression of this fire is presented in Figure 30, captured at 3 seconds, 4 minutes, 6 minutes, 8 minutes, 12 minutes, and 20 minutes (top left to bottom right) after ignition. Figure 30. Progression of fire in Test HUGHES ASSOCIATES, INC.