Effects of convection and radiation on flame spread over solid fuel beds

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Effects of Convection and Radiation on Flame Spread Over Solid Fuel Beds by Linton Kaneki Honda A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (Aerospace and Mechanical Engineering) May 2001 Copyright 2001 Linton Kaneki Honda Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UM I Number: 3027724 Copyright 2001 by Honda, Linton Kaneki All rights reserved. _ _ ® UMI UMI Microform 3027724 Copyright 2001 by Bell & Howell Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. Bell & Howell Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNIVERSITY OF SOUTHERN CALIFORNIA The Graduate School University Park LOS ANGELES, CA 90089 This dissertation written by Linton Kaneki Honda Under the direction of his Dissertation Committee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of requirements fo r the degree of DOCTOR OF PHILOSOPHY Dean of Graduate Studies Date May 11, 2001_______ DISSERTATION COMMITTEE Chairperson Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Linton Kaneki Honda Professor Paul Ronney ABSTRACT EFFECTS OF CONVECTION AND RADIATION ON FLAME SPREAD OVER SOLID FUEL BEDS The effects of atmosphere composition and the convection environment (downward (opposed buoyant flow), upward (concurrent buoyant flow), or microgravity (negligible buoyant flow)) on the flame spread rate (Sf) over thin solid fuel beds were measured and compared to theoretical predictions. For downward and microgravity (pg) flame spread, two modifications to the standard air atmosphere were considered. First, the effect of diluent type on Sf was studied by comparing results using He, N2, Ar, C 02, and SF6 diluents. Like prior studies in N2 diluent, for He, N2, or Ar diluents it was found that downward Sf was larger than the pg Sf, however, for C 02 diluent, downward Sf was slightly lower than pg Sf and for SF6 diluent, the downward Sf was much lower than pg Sf. Moreover (unlike He, N2, or Ar), for C 02 and especially SF6 diluents the minimum 0 2 concentration required to support flame spread at pg was lower than the minimum concentration for downward spread at 1g (for SF6, the pg limit was even lower than the upward limit). This behavior is proposed to be a result of reabsorption of radiation emitted from the gases. Secondly, the effects of sub-flammability-limit concentrations of a gaseous fuel (CO or CH4) were measured and compared to an existing theoretical model that was extended to pg conditions. The agreement 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. between the model and experiment is reasonable. Notably, both model and experiment show that the effect of added gaseous fuel is greater at pg than for downward spread at 1g. For upward flame spread, steady spread was found under conditions where heat and momentum losses to the sides of the fuel sample or surface radiative losses were significant. These losses are argued to be unavoidable because the flame length grows until these losses balance the heat generation. By equating heat generation and losses, approximate predictions of spread rates were obtained. Experiments over a large range of Grashof number were performed and generally support the validity of the proposed mechanisms. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. DEDICATION This dissertation is dedicated to my children, who may one day read it and possibly be inspired to surpass me in the field of experimental research, design rockets, or find the cure for cancer. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGMENTS This investigation was supported largely by Grant NAG3-1611 and NCC3-671 from the NASA Lewis Research Center (LeRC), which has since been renamed Glenn Research Center (GRC) in Cleveland, Ohio. I would like to express my gratitude to the members of my Thesis Committee, Professors Fokion Egolfopoulos, Tony Maxworthy, Paul D. Ronney, and Satwindar S. Sadhal of the Aerospace and Mechanical Engineering Department and Professor Theodore T. Tsotsis of the Chemical Engineering Departments for their guidance and support. I would like to thank the entire Aerospace and Mechanical Engineering department for their confidence in me and allowing me to continue in my education. I would like to thank all the people at NASA-Lewis for their support throughout this investigation, especially grant monitors Sandra L. Olson and Suleyman A. Gokoglu, drop tower managers, Jack Lekan and Eric D. Baumann, and technicians Mike Johnston, Andrew J. Jenkins, and Jose Carrion. I would like to thank all the people from USC that helped me produce an operational drop- rig, namely Justin Fortemeyer, Quin Blackburn, Ken Conner, Dr. Liu, Mingshin Wu, Teodora Valdez, and Kevin Borer. Two very important students who worked with me on this project from USC were Alexander Ludorf (concurrent- flow) and Young-Jin Son (thick fuel opposed-flow). I would also like to thank my managers David Perry, Henry Rodriguez, Jeff Peterson, Tom Mueller, iii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Arturo Casillas, and Jacky Calvignac for allowing me to flex my work schedule to optimize writing my qualification proposal and dissertation. I would especially like to thank my loving wife Mi Honda for her patience and understanding through the entire process. I would like to thank my parents Glenn Honda and Lynne Stubenberg and my step-parents Lynn Honda and James Stubenberg for their support over the years. Most importantly, I would like to thank my advisor Paul D. Ronney, The Big Kahuna, again. His eloquence and style inspired me. He had faith in me from the day I walked into his office through all of my trials and tribulations and gave me the motivation to continue against all odds. Our countless conversations academic, experimental, and personal have only improved my regard for him. The greatest compliment I could receive would be to be told that I resemble PDR in teaching, presenting, experimenting or some other way. To this day, he continues to help me grow. For all of this, I thank him. Finally, I would like to thank God. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER TABLE OF CONTENTS Page DEDICATION ............. II ACKNOWLEDGMENTS...................................... ......Ill LIST OF TABLES..... ...... ¥ 1 1 1 LIST OF FIGURES ..... ..X LIST OF COMMONLY USED SYMBOLS AND ACRONYMS ....... XV ABSTRACT ..... XX 1 INTRODUCTION ........ 1 2 THEORETICAL BACKGROUND................. 7 2.1 Opposed-F l o w Fl a m e Sp r e a d .............. 7 2.1.1 General Theory.......................................................................................7 2.1.2 Radiation Dominated Flame Spread......................................................11 2.1.3 Partially premixed flame spread........................................................... 16 2.2 Co n c u rr en t-Fl o w Fl a m e Sp r e a d ..........................................................................22 2.2.1 Modeling predictions............................................................................ 25 2.2.1.1 Flame lengths........................................... 25 2.2.1.2 Spread ra tes........................................................................................................28 2.2.1.3 Transitions between regim es ........ 29 2.2.1.4 Comparison w ith previous results.................. 31 3 OBJECTIVES AND APPROACH ..... ..33 3.1 Opposed-Fl o w Fl a m e Spread ............................................................ 33 v Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.2 C o n c u rr en t-Flo w Fl a m e Sp r e a d ............................................... .36 4 EXPERIMENTAL APPARATUS AND PROCEDURES...........................38 4.1 PPM S ................................................................................. 39 4.2 Fl a m e Spread A p p ar atu s................... 44 4.2.1 Microgravity Apparatus.......................................................................44 4.2.1.1 Drop Frame........................................................ 47 4.2.1.2 C ham ber ........ 47 4.2.1.3 Im aging Systems..................... 48 4.2.1.4 Internal Apparatus .............................................. 52 4.2.2 Concurrent-flow Earth-gravity Apparatus............................................ 60 5 RESULTS ..... 66 5.1 T ests of H ypo th esis................................................................................................... 66 5.1.1 Steady Upward Spread.........................................................................66 5.1.2 Steady Downward Spread Over Thin Fuels.......................................... 68 5.2 D a t a A n a l y s is ...............................................................................................................70 5.2.1 Video...................................................................................... 70 5.2.2 Interferometry...................................................................................... 73 5.2.3 Thermocouples.....................................................................................74 5.3 Spread R a t e s ...................................... 77 5.3.1 Opposed-Flow Flame Spread............................................................... 77 5.3.1.1 Radiation Effects - Thin Fuels.................................................................... 77 5.3.1.3 Partially Premixed F u e l.................................................................................89 5.3.2 Concurrent-Flow Flame Spread........................................................... 94 vi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 CONCLUSIONS ..... 101 7 RECOMMENDATIONS FOR FUTURE WORK....................................... 104 7.1 T h ic k F u e l M o d e l s ............................................................................................ 104 7.1.1 Opposed-Flow Flame Spread..............................................................104 7.1.2 Concurrent-Flow Flame Spread.......................................................... 106 7.2 Fu tu r e O b je c t iv e s..................... 108 7.3 Pr e l im in a r y T h ic k Fu e l Resu lts, Opposed-Fl o w .........................................109 REFERENCES ................ 113 APPENDIX A FLOW TUNNEL ...... ............................................. ...118 A . 1 Ex p e r im e n t a l A pp ar atu s........................... ....118 APPENDIX B PROGRAMS ...... 120 B .l “ PPM S.BAS” .................................................... 120 B.2 “ BEFO R E.TTB” ............................................................................................................134 B.3 “ A F T E R .T T B ” ................................................................................................... 137 APPENDIX C DATA ...... ..138 C. 1 Opposed - Flo w Fl a m e Spr ead......................................................... 138 C.2 Co nc u rr en t - Fl o w Fl a m e Spr e a d ........................................ 146 vii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES Table Page 1 Predicted increase in spread rate (Sf) and flame length (L) with time (t) for laminar concurrent-flow fires................................................ 23 2 Predicted relations for the steady values of Nul and L/W for forced or buoyant convection and convective or surface radiation. Predictions are the same for thermally-thin and thermally-thick fuels. Note that since Rew ~W, Grw ~W3 and Plw ~W 1 , L is independent of W for radiatively-stabilized flames.................. 27 3 Predicted relations for steady values of SfiC on/Sf,0p p for thin fuels in forced and buoyant convection, and convective and surface radiative loss stabilization. Since Rew ~W, Grw ~W3 and Plw ~W 1, S f.co n is always independent of W for radiatively-stabilized flames. ...28 4 Comparisons of measured and predicted effects of added gaseous fuel on flame spread rates at 1g and pg................................ 94 5 Predicted increase in spread rate (Sf) and flame length (L) with time (t) for laminar concurrent-flow fires...............................................107 6 Predicted relations for steady values of Sf,con/Sfi0 p P for thick fuels, forced and buoyant convection, and convective and surface radiative loss stabilization...................................................................... 108 7 Opposed-flow flame spread data for thin fuels in argon..................... 138 8 Opposed-flow flame spread data for thin fuels in SF6.........................139 9 Opposed-flow flame spread data for helium.........................................140 10 Opposed-flow flame spread data for nitrogen...................................... 141 11 Opposed-flow flame spread data for CO2.............................................142 12 Partially-premixed flame spread data ...........................................144 13 Non-dimensional flame spread data for concurrent-flow flame spread (air). ............................................................................................ 146 14 Non dimensional flame spread data for concurrent-flow flame spread (N2 ). ..........................................................................................147 viii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 Non dimensional flame spread data for concurrent-flow flame spread (CO2, He, SFe).........................................................................148 16 L/W data for Nitrogen Diluents.........................................................................149 17 L/W data for SF6, CO2, and He Diluents........................................................ 150 ix Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES Figure Page 1 Schematic diagram of opposed-flow flame spread. Flame is fixed with fuel moving at Sf and uniform flow U which can be buoyancy induced (as in gravity) or forced............................................................ 2 2 Cartoon showing difference in convection-diffusion length, 8g, for 1g on the left and pg conditions on the right ................... 15 3 Illustration of difference radiation absorption and re-emittance effect (A effect).......................... 15 4 Numerical solution to non-dimensional thin fuel flame spread rate Sf in Equation 7 as a function of the radiation-only flame spread rate Sf,ra d at constant flow velocities, U................................................... 16 5 Model for non-merged partially-premixed flame spread over a solid fuel..................................................................................................... 18 6 Predicted regimes of concurrent-flow flame spread for buoyant convection, showing the type of flow (laminar or turbulent) and flame stabilization (convective or radiative). Also shown are lines corresponding to fixed atmosphere but varying fuel bed width (W) for air and the atmospheres yielding the lowest and highest Plw and Grw experimentally possible in this proposal, i.e., 0.25 atm 0 2 -He and 3 atm O2-SF6, respectively, for TV =618K, T-=300K and s=1.......................................................................................................31 7 Schematic of PPMS internal arrangement..............................................40 8 Schematic of PPMS external arrangement.............................................41 9 Schematic of “ The Rig” (MIDAS). A torturous path is needed to fold the interferometer beam to fit within the drop rig envelope. The dropping apparatus to requires its own power, computer, camera, etc. during pg. Images obtained are sent through a fiber optic line that connects the free-falling rig to VCRs. The rig lands on an airbag 8 stories below................................. 45 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 Cut-away view or 2.2 second drop tower................................................46 11 Flame spread across a thick solid fuel at 1g in 40% oxygen with C 0 2 diluent at 4 atm. Note buoyancy induced convective flow......... 49 12 Flame spread across a thick solid fuel at pg in 40% oxygen with C 0 2 diluent at 4 atm. Reaction products tend to remain over consumed fuel. No flow disturbance seen............................................49 13 Description of shearing plate....................................................... 51 14 Interferometer image. Side view of Kimwipe sample burning at 1 g in 42% O2- SF6 @ 4 atm. Very narrow convective-diffusive zone thickness. Field of view is approximately 4 cm by 4 cm............51 15 Interferometer image. Same sample burning as Fig. 17 in pg. Flame (same scale) is substantially expanded over 1g counterpart showing effect of reduced buoyancy induced flow velocity................. 51 16 Interior view of thin fuel sample holder. Kimwipe samples held in place by aluminum quenching plates on both sides. Interferometer path shown as red circle while radiometers and thermocouples are optional.............................. 55 17 Schematic of the concurrent-flow flame spread experimental apparatus. The test rack goes inside the pressure vessel after fuel sample loading. The PPMS is disconnected from the pressure vessel before ignition. The Video data is taken through the front while thermocouple data is send to the computer. The Kanthal wire is connected to external power source ..............61 18 Temperature profile as a function of time for upward spreading flame. ............................................. 64 19 Flame position as a function of time for concurrent-flow flame spread. Gaseous atmosphere is varied and measured position and times are taken toward the top of the fuel samples. Steady flame spread can be seen by drawing a straight line though the data points. The flame spread rate is the slope of this line............... 67 xi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 Flame position as a function of time for a thin fuel. Rig starts dropping at time = 0 seconds. For most conditions, data points after initial 0.5 second transition line up on a straight line showing a steady flame spread rate...................................................................... 69 21 Effect of number of fuel sheets on spread rate. Atmosphere: 33% 0 2 in N2at pg............... 70 22 Temperature profile comparison of upward and downward spreading flames in air at 1 atm. Downward flames are much smaller and the temperature changes more rapidly............................. 76 23 Flame spread rate vs oxygen concentration for thin fuels with nitrogen diluent at 1 atm.................................. ...79 24 Flame spread rate vs oxygen concentration for thin fuels with helium diluent at 1 atm............................................................................. 79 25 Flame spread rate vs oxygen concentration for thin fuels with argon diluent at 1 atm...............................................................................80 26 Flame spread rate vs oxygen concentration for thin fuels with carbon dioxide diluent at 1 atm............................................................... 80 27 Flame spread rate vs oxygen concentration for thin fuels with sulfur hexaflouride diluent at 1 atm....................................... 81 28 Interferometer image of spreading flames at 1g. Atmosphere: 30% 0 2 in N2 at 1 atm. Field of view is 4.2 cm x 3.2 cm.....................83 29 Interferometer image of spreading flames at pg. Atmosphere: 30% 0 2 in N2 at 1 atm. Field of view is 4.2 cm x 3.2 cm.....................83 31 Flame spread rate as a function of pressure for helium, nitrogen, sulfur hexaflouride, and carbon dioxide diluents at fixed oxygen mole fractions ........................................................... 88 32 Measured and predicted flame spread rates vs. equivalence ratio in 18% 0 2 with N2 diluent at 1 atm with CO as added gaseous fuel...............................................................................................................91 xii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 Measured and predicted flame spread rates vs. equivalence ratio in 18% O2 with N2 diluent at 1 atm with CH4 as added gaseous fuel......................................................................... 91 34 Measured and predicted flame spread rates vs. equivalence ratio in 30% 0 2 with N2 diluent at 1 atm with CO as added gaseous fuel.............................................................................................................. 92 35 Measured and predicted flame spread rates vs. equivalence ratio in 30% O2 with N2 diluent at 1 atm with CH4 as added gaseous fuel.............................................................................................................. 92 36 Effect of fuel bed width (W) on upward flame spread rate (Sf,c o n ) for thin fuel beds burning in ambient air. Predicted results are Sf ~ W3 for low Grw and Sf ~ W° for high Grw, with a transition Grw of 30,000.........................................................................................................95 37 Correlation of steady values of Sf,con/Sf,0 p p with Grw for all experimental data. "x2" indicates double-thickness fuel samples 97 38 Correlation of steady values of L/W with Grw for all experimental data. Legend is the same as Figure 37.................................................98 39 Ratio of measured L/W to right-hand sides of Equations 30 and 31, showing comparison of predicted and observed correlation of L to S f,c o n for convective-laminar, convective-turbulent, and radiative-turbulent spread regimes explained in Figure 6. Dashed horizontal line indicates ideal fit of prediction to experiments........... 100 40 Position of flame for a thick fuel burning at pg in 35% O2/CO2 at 4 atm. The rig drops at time = 0 seconds. The data to the right of the 1 second mark creates a straight line indicating steady state flame spread............................................................................................ 110 41 Image of Interferometer from the side of the fuel, and the upper black region represent the thick volume of the flame in pg............... 111 42 Image of Interferometer from the side of the fuel, and the upper right black region representing the convection-diffusion zone thickness at 1g is smaller than that of the pg...................................... 111 xiii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 Effect of oxygen concentration on spread rates over thick solid fuel beds at pg and 1g................ 112 44 Schematic of Olson’s flow-tunnel........................................................... 119 45 Example of spreadsheet used to graph results for thickness, oxygen concentration, and pressure effects........................................143 46 Example of partial output file 080299-5.dat. Time is in 1/100th second values with 00000 corresponding to time of starting program. Voltages measured are 1/13107 per volt input. These are usually pre-amplified by 200 times before the D-DACS receives them.......................................................................................... 145 xiv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF COMMONLY USED SYMBOLS AND ACRONYMS A Fuel surface area ap Planck mean absorption coefficient aR Rosseland mean absorption coefficients Ar Argon Bp Frequency factor of premixed fuel + oxygen chemical reaction c Speed of Light CL Convective-Laminar regime CO Carbon Monoxide CO2 Carbon Dioxide CT Convective Turbulent regime C p,g Specific heat capacity at constant pressure of gas Cp,s Specific heat capacity at constant pressure of solid Ep Activation energy of premixed fuel + ox. chemical reaction F Prescribed Function in deRis formulation F b e d Heat flux to the fuel bed fu Fuel GrL Length averaged Grashof number Grw Width averaged Grashof number Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. h Heat transfer coefficient He Helium k Conductivity L Length . Lv Latent heat of vaporization of the solid fuel Lf Length of convection-diffusion zone upstream of the flame front leading edge Le Lewis Number = a/D MIDAS Microgravity Interferometer Drop Apparatus System N2 Nitrogen Nul Length averaged Nusselt number N U lf Nusselt number based on length of the fuel surface Nuw Width averaged Nusselt number n Non-premixed (solid) fuel p Premixed (gaseous) fuel PI Planck number o x Oxidizer PACIFIC Premixed Atmosphere & Convection Influences on Flame Inhibition & Combustion PPMS Partial Pressure Mixing System xvi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Qs Heat flux to the fuel surface Qt Heat of formation Q Heating value qr Radiant heat flux per unit area Radiant heat flux per unit wavelength q” Heat release per unit area Ri Net radiative heat flux received by the fuel r 2 Constant of deRis formulation ReL Length averaged Reynolds number R©w Width averaged Reynolds number REEFS Radiation and Environmental Effects on Flame Spread RL Radiative-Laminar regime RT Radiative-Turbulent regime S Stoichiometric oxidant-to-fuel mass ratio s Characteristic strain rate sf Flame spread rate Sfto Radiation free flame spread rate Sf.con Concurrent-flow flame spread rate Sf.opp Opposed-flow flame spread rate Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. SF6 Sulfur Hexafluoride t Time T Temperature Tv Temperature of vaporization Tf Temperature of flame front T o o Temperature of ambient atmosphere U Induced flow velocity W Fuel surface width Y Mass Fraction Yp , o o Ambient mass fraction of premixed fuel Yo2, o o Ambient mass fraction of oxygen in the ambient atmosphere x Direction of flame propagation along fuel surface Xe Xenon y Direction perpendicular to flame propagation and fuel surface a Thermal diffusivity ttg Thermal diffusivity of the gas 6 Characteristic length scale for thermal transport s Bed emissivity r Thick fuel flame spread parameter Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. q Rate of Enthalpy increase Stoichiometric fuel / (oxidizer + inert) mass ratio k(X) Spectral absorption coefficient X Thermal Conductivity Xg Thermal Conductivity of the gas phase A Radiant heat emission rate per unit volume Equivalence ratio p Density Pg Density of the gas Ps Density of the solid fuel a Stephan-Boltzman Constant Is Thickness of the solid fuel Xp Penetration depth @p Reaction zone M9 Microgravity V Stoichiometric coefficient V Stoichiometric fuel/(oxygen + inert) mass ratio Vg Kinematic Viscosity ig Earth gravity Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT The effects of atmosphere composition and the convection environment (downward (opposed buoyant flow), upward (concurrent buoyant flow), or microgravity (negligible buoyant flow)) on the flame spread rate (Sf) over thin solid fuel beds were measured and compared to theoretical predictions. For downward and microgravity (pg) flame spread, two modifications to the standard air atmosphere were considered. First, the effect of diluent type on Sf was studied by comparing results using He, N2, Ar, C 0 2, and SF6 diluents. Like prior studies in N2 diluent, for He, N2, or Ar diluents it was found that downward Sf was larger than the pg Sf, however, for C 0 2 diluent, downward Sf was slightly lower than pg Sf and for SF6 diluent, the downward Sf was much lower than pg Sf. Moreover (unlike He, N2, or Ar), for C 0 2 and especially SF6 diluents the minimum 0 2 concentration required to support flame spread at pg was lower than the minimum concentration for downward spread at 1g (for SF6, the pg limit was even lower than the upward limit). This behavior is proposed to be a result of reabsorption of radiation emitted from the gases. Secondly, the effects of sub-flammability-limit concentrations of a gaseous fuel (CO or CH4 ) were measured and compared to an existing theoretical model that was extended to pg conditions. The agreement between the model and experiment is reasonable. Notably, both model and experiment show that the effect of added gaseous fuel is greater at pg than for downward spread at 1 g. xx Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. For upward flame spread, steady spread was found under conditions where heat and momentum losses to the sides of the fuel sample or surface radiative losses were significant. These losses are argued to be unavoidable because the flame length grows until these losses balance the heat generation. By equating heat generation and losses, approximate predictions of spread rates were obtained. Experiments over a large range of Grashof number were performed and generally support the validity of the proposed mechanisms. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 1 INTRODUCTION Flame spread over flat solid fuel beds is a useful paradigm for studying the behavior of more complex two-phase non-premixed flames such as building fires. For practical applications, the flame spread rate (Sf) can be affected dramatically by the ambient atmosphere (e.g., pressure and composition) and the flow environment. Being able to accurately predict the impact of these variables is a critical aspect of any flame spread model. Before proceeding to these effects, a summary of basic flame spread phenomena is useful. Flame spread can be defined as the advancement of a reaction zone, where a fuel and oxidizer react exothermically, over a solid or liquid fuel surface in an oxidizing environment. Heat that is released from this reaction zone vaporizes the fuel near it, and in turn, the fuel is free to react with more of the oxidizer. As the fuel and oxidizer are consumed (and the surrounding fuel is vaporized), the flame moves to a position where fresh fuel and oxidizer exist in stoichiometric proportions. This process is called a “ diffusion” or “non-premixed” flame spread and the speed of the moving flame is called the flame spread rate. When the flame spreads in the same direction as the environmental flow (i.e., fire spreading up the side of a building driven by upward-buoyant-flow) this behavior is known as upward or concurrent-flow flame spread. In contrast, when the flame spreads against the direction of the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. environmental flow (i.e., fire spreading down the side of a building) this behavior is known as opposed-flow flame spread as shown below in Figure 1. Oxidizer being converted and diffused into flame Fuel diffusing into flame heat transfer to ^ unbumed fuel bed heat transfer to vaporizing fuel bed _ u? W . — . . I gravity Figure 1. Schematic diagram of opposed-flow flame spread. Flame is fixed with fuel moving at Sf and uniform flow, U, which can be buoyancy induced (as in gravity) or forced. It is also useful, at this point, to define gravity and the induced buoyant flow in Figure 1. In general, the flow can be forced or buoyant-flow. If the flow were forced, as in through a duct by a fan, then an opposed-flow flame would propagate in the duct toward the fan. In many cases, the flow is created by the hot gas produced by the flame itself. This is known as buoyancy-induced flow. In microgravity (pg), this induced-buoyant-flow is significantly less than at earth gravity (1g). In Figure 1, this flow is shown to be uniform for analytical simplicity. However, in reality, the flow would be a boundary layer type of flow. 2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The important thing is that opposed-flow flame spread and concurrent- flow flame spread are both affected by the atmosphere (which includes the pressure, oxygen, and other chemical species concentrations) and the flow environment. However, atmosphere and flow affect opposed-flow flame spread and concurrent-flow flame spread in different ways. Therefore, it is useful break up this study into opposed-flow flame spread and concurrent-flow flame spread. Concerning ambient atmosphere effects, studies of flame spread in vitiated (partially burned) air and non-standard atmospheres such as those found in submarines and manned spacecraft are particularly important for the assessment of fire hazards in these enclosed environments as well as determination of the effectiveness of fire suppressants. However, few experiments have been conducted with diluents (He, Ar, N2, CO2, and SFe) and added gaseous fuels (CO and CH4) at 1g, and even less experiments have been conducted at pg. Concerning flow effects, convection may vary widely between different enclosures even when no forced flow is present because of buoyancy effects. Previous pg experiments (Bhattacharjee & Altenkirch, 1993; Olson e ta i, 1988) have provided considerable new insight into the effects of buoyant convection on the mechanisms of flame spread and flame extinction, however, such studies have not considered the effect of ambient atmosphere other than variations in the oxygen mole fraction and total pressure. Consequently, a goal of this work is to provide a broader 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. assessment of the effects of ambient atmosphere and flow effects on flame spread at pg than that available from previous studies. For opposed-flow (downward and microgravity) flame spread, two types of changes to the oxidizing atmosphere are considered in this work. One is the diluent type, which affects the radiative properties of the gas and thus may affect flame spread processes (Bhattacharjee & Altenkirch, 1993). It is of interest to study these diluent effects because in undersea and space-borne habitations it is sometimes desirable to use diluent gases other than nitrogen. The second change is the addition of sub-flammability-limit concentrations of a gaseous fuel (“partially-premixed” atmospheres). This is of interest because in fires in enclosures, combustion may occur under poorly ventilated conditions, so that oxygen is partially depleted from the air and is replaced by combustible gases such as fuel vapors, H2 or CO. Subsequent fire spread over the solid fuel could occur under conditions of varying oxygen and gaseous fuel content. The significance of flame spread under partially-premixed conditions has been noted previously (Beyler, 1984). Downward flame spread at earth gravity (1g) is chosen for comparison with the pg results to be obtained in this study because (in the absence of an imposed forced-flow) at pg the flame always propagates toward the fresh atmosphere and therefore is always characterized as opposed-flow. Downward flame spread at 1 g is also characterized as opposed-flow since the upward buoyant flow direction is opposite that of the flame propagation. Thus, 4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. a more meaningful comparison to pg results can be obtained with downward rather than upward flame spread at 1 g. For concurrent-flow (upward) flame spread, convective and diffusive transport are in the same direction. This is of interest when modeling more complex flame spread like up the side of buildings. Previous studies (Fernandez-Pello, 1984; Williams, 1976) have shown that the fuel surface area exposed to high-temperature combustion products increases with time, leading to accelerating spread. Using boundary-layer analyses, Fernandez- Pello (1984) predicted that flame length (L) and Sf for concurrent-flow (SfiC O n ) increase indefinitely with time (t). Delichatsios et al. (1996) also examined unsteady concurrent-flow spread. In contrast, some experiments (Fernandez- Pello & Hirano, 1983: Grayson et al., 1994) show steady L and Sf,co n - The analyses assumed adiabatic spread across infinitely wide samples, thus heat losses and lateral momentum losses were neglected. With such losses, the boundary layer thickness (8 ) could not grow substantially larger than the sample width (W). For example, a 10 cm thick boundary layer on a 1 cm wide fuel sample is usually not observed. If 8 is limited, then L and Sf are also limited. Even for infinitely wide samples, L could not grow indefinitely because surface radiative losses would eventually exceed heat generation rates. This is because (for boundary layer flows) the fuel bed heat flux (Q), thus fuel vapor generation rates and total heat generation rates, increase more weakly than linearly with L, whereas heat and lateral momentum losses increase roughly 5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. linearly with L. Also, Markstein and deRis (1972) suggested that for thermally- thin beds, fuel burnout could limit Sf. Therefore, in the presence of these losses, it is believed that the flame length must become steady when the heat generation and losses balance and thus the flame spread rate must also become steady. With a steady spread rate, the heat transfer to the fuel bed can be compared to the enthalpy needed to vaporize the fuel bed to determine a simple scaling model for the flame spread rate. Therefore, a goal of this work is to determine if these losses can lead to steady flame spread, and if so to predict the spread rate. All of the above considerations apply to thermally-thin fuel beds, where conduction through the fuel bed is negligible compared to conduction through the gas phase. This study focuses exclusively on thermally-thin conditions because the spread rates are high enough to reach steady-state in the short- durations needed at drop towers and in combustion chambers. This study can be extended to thick fuel beds with changes that are discussed in “Future Work,” Chapter 7. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 2 THEORETICAL BACKGROUND 2.1 Opposed-Flow Flame Spread 2.1.1 General Theory The most observed characteristic of flame spread over solid fuels is the flame spread rate (Sf). The flame spread rate is indicative of the mass burning and heat release rate. Therefore, it is one of the most important predictions of any theory on flame spread. It has been shown (Williams, 1976) that the flame spread rate over a solid fuel can be modeled by comparing the heat flux from the gas phase to the fuel surface (Qs ) to the rate of increase of in enthalpy of the solid fuel (here it will be called q for simplicity). Equation 1 shows the energy balance relation proposed by Williams (1976). Qs = hAAT = LFW(Tf -T v) = NuLpAgW(Tf -T v) (1) F In Equation 1, h is the heat transfer coefficient, N u lf is the Nusselt number based on length of the fuel surface exposed to this heat flux ( L f ) , A is the fuel surface area exposed to this heat flux, X the thermal conductivity, W the fuel surface width, T the temperature, and the subscripts g, f and v refer to the gas- phase, flame front and vaporization condition, respectively. It should be noted that for opposed-flow conditions, L f is the length of the convection-diffusion zone upstream of the flame front leading edge (6g ) which has been shown 7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (deRis, 1969) to be proportional to ag / U , thus the Reynolds and Nusselt numbers are of order unity. Also, a g = A .g / pg Cp ,g is the thermal diffusivity and U is the gravity induced flow velocity. In contrast, for concurrent-flow conditions, LF is the flame length (which can be much greater than 8 g since the entire flame is preheating the fuel surface) and must be determined as part of the problem. Consequently, values of N u lf and S f can be much higher for concurrent-flow than opposed-flow conditions. When the flame has a constant spread rate with complete combustion, the rate of enthalpy increase (q) is given below by Equation 2. ^ = / ?SC Ps'r s ( T V - T o o ) W S f ( 2 ) In this equation, it is assumed that the fuel is vaporized or pyrolyzed at a constant temperature, Tv, and the ambient temperature is T®. ps, Cp , and xs are the density, constant pressure specific heat, and fuel thickness, respectively. The subscripts s and qo refer to the solid fuel and ambient conditions, respectively. A relation for the flame spread rate is found by combining the right hand sides of Equations 1 and 2. Tf - Tv S f = Nu l ------- * - -------- f ------- -- (3) F o C t T - T r s p,s s v oo For “ thermally-thin” fuels (paper for example), where heat conduction through the solid fuel is negligible and the fuel is completely consumed as the flame 8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. passes, ts is the solid fuel bed half-thickness. For “ thermally-thick” (effectively semi-infinite) fuels, where heat conduction through the solid fuel is important, xs is replaced by the depth of thermal penetration into the solid fuel (xp ). A more detailed “ thermally-thick fuel” analysis is shown in “Future Work,” Section 7.1. Thermally-thin materials will always have a fuel half-bed thickness less than that of the depth of thermal penetration. Note that a given material like paper may behave as thermally-thin or thermally-thick depending the flow environment, and in turn, N uLf- It has been shown (Altenkirch et al., 1980; deRis, 1969) that the flame temperature (Tf), including nonlinear mass transfer effects, is essentially the same as the adiabatic temperature of a stoichiometric mixture of solid fuel at temperature Tv and oxidizing atmosphere at temperature T*. T j = x + s + t” s = Z m : (4> In this equation, Qf is the heat of formation, Lv is the latent heat of vaporization, Y is the mass fraction, M is the molecular weight, v is the stoichiometric coefficient, S is the stoichiometric oxidant-to-fuel mass ratio, and the subscripts fu and ox refer to solid fuel vapors, and oxidant, respectively. Note that the energy used to vaporize the solid fuel is included in the Lv term, which will affect Tf and thus Sf. It has been shown (deRis, 1969) that for opposed-flow conditions with thermally thin fuels the approximate value of N uLf is V2. Later, an exact 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. solution to the same problem was shown (Delichatsios, 1986) to be N u lf = tc / 4. However, both of these solutions require restrictive assumptions like uniform (plug-flow) opposed-flow velocity (U), infinitely fast chemical reaction when fuel and oxidant mix, constant thermodynamic and transport properties, and a Lewis number (ratio of the thermal diffusivity divided by the mass diffusivity) of one. Away from extinction conditions, where finite-rate chemistry effects can be neglected, these estimates are in reasonable agreement with experiments (Fernandez-Pello et al., 1981). Detailed numerical calculations (Bhattacharjee, 1993) have shown that these solutions are generally valid, though with some quantitative differences, particularly concerning the effect of the dimensionless group Lv / Cp ig Too. A key point to consider here is that without an imposed flow due to buoyancy (such as at pg) the flame spread rate is always an opposed-flow type where the flame spread rate (Sf) is equal to the self-induced flow velocity (U). At Earth gravity (1g), the self-induced flow velocity is typically an order of magnitude smaller than the buoyancy-induced flow velocity so the self-induced convection can be ignored. However in the absence of a buoyancy-induced flow velocity, self-induced convection cannot be ignored. For thin-fuels, the ideal spread rate shown in Equation 3 is independent of U so steady state flame spread is still possible. Previous pg experiments (Bhattacharjee & Altenkirch, 1993; Olson et al., 1988) have shown the importance of the effects of buoyant convection on the mechanisms of flame spread and flame 10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. extinction. However, these studies have only considered the effect of variations in the oxygen mole fraction and total pressure. 2.1.2 Radiation Dominated Flame Spread The discussion in Section 2.1.1 is appropriate when heat is transported to the fuel bed only by conduction or convection. However, it has been shown (deRis, 1968) that radiation terms can have a large impact on flame spread when added to the previously formulated problem for opposed-flow. In the analysis (deRis, 1968), the net radiative heat flux received by the fuel was prescribed as Riex /I upstream of the flame (x<0) and a constant R2 downstream of the flame (x>0 ). A.P.Cp,U _ _ ^gPgCp,s$f 'T f - T A 2RlF(21g / PgCpgSf l ) 2R I 2 . T - T V V 1 x J + ‘........................^ i (5 ) PgCpgSf (Tv- T J 7 v PgSf Cpg(Tv - T x) In this equation, F( 2 Xgl pg Cp ,g Sf I ) is a prescribed function. The middle and last terms on the right hand side of Equation 5 describe the effects of radiative heat transfer upstream and downstream of the flame front, respectively. Equation 5 is certainly analytically reasonable, but the radiation terms must be prescribed beforehand. This is rational when an external radiative flux is imposed on the flame, but does not apply when considering radiation generated by the flame itself. Also, radiative absorption and emission from the gases are not taken into account. However, in the absence of gravity where Sf is equal to U, the radiative properties of the gases can become very important. 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. It is proposed that this radiative absorption and emission from the gaseous atmosphere can have a substantial impact on the flame spread rate. This radiative impact can be dominant under some conditions as will be shown below. For radiatively-dominated flame spread, the heat flux to the solid fuel bed due to radiation can be estimated as qr8g W, where qr is the radiant heat flux per unit area. By equating this heat flux to the rate of increase in enthalpy of the solid fuel ps Cpst s (Tv - T«) W Sf as before, a relation for radiation- dominated flame spread is obtained as Equation 6 below. s / = r ( t h i n f u e l ) ( 6 ) P s ^ p ^ s V v *oa) For some 1g experiments (Brehob & Kulkarni, 1993) and space experiments (Olson, 1998), qr has been applied to the fuel bed externally (and thus not a function of flame spread). However, in some cases radiation generated by the flame itself (such as via gas-phase radiation) may be an important factor. Therefore, additional theory is needed to understand these phenomena. For the initial estimate of flame spread due to gaseous radiation, the flame front is assumed to be an isothermal volume of gas at temperature Tf with dimension Ax by Ay by W that radiates to the fuel surface at temperature T v. It is assumed that Ax = Ay = S g (as shown in Figure 2) because there is no length scale for reabsorption. Thus, the thermal thickness of the flame front for opposed-flow conditions is determined by the convective-diffusive zone thickness 8g« o c g / U = otg / Sf. When compared to flame spread at 1 g, this 12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. point becomes very important in pg because the convective-diffusive zone thickness becomes much larger at pg. Physically, the convective-diffusive zone thickness is smaller at 1 g because fresh oxygen is brought to the flame at a speed (U) over a distance (8g ) in time (t). In other words, the convection time goes like 8g / U. Whereas the diffusion time ~ 8g 2 / otg is independent of U. Balancing the convection-diffusion time results in 8g / U « 8g 2 / ag or 8g « ag / U as shown above. At 1g, the induced flow velocity becomes much larger so the oxygen can be brought to the flame much faster. Meanwhile, the fuel must diffuse from the surface at about the same rate so the location of stoichiometric mixing occurs closer to the fuel surface at 1g than at pg. That said, with the larger convective-diffusive zone thickness at pg, the radiation heat transfer (both to the surroundings and to the fuel bed) at pg is larger because the radiating volume is larger. It is still possible to transfer heat to the fuel bed at distances greater than S g, but this heat transfer is at a small angle to the fuel surface and thus is not a very efficient heat transfer mechanism. The heat flux per unit area to the fuel surface due to radiation can then be estimated as A8g, where A = 4 c t ap (Tf4 - Tv 4) is the radiant heat emission rate per unit volume, or is the Stefan-Boltzman constant and ap is the Planck mean absorption coefficient. The combined effect of gas-phase radiation and thermal conduction is then given by qr = A8g + A ,g(Tf - Tv ) / 8g. The effect of this radiation term (A), can be seen in Figure 3 as the ability of the nearby gaseous atmosphere to absorb the heat transfer from the flame and re-emit it both 13 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. back to the fuel and to the surroundings. Combining the expressions for qr, 8g, and Equations 5 and 6 for thin fuel case, leads to a cubic equation for Sf shown in Equation 7, which includes the effects of forced convection. 3 r + 2 U \ s,» - 1 + - u J,° v \ S, ] f,0 J u2 sl , 3 \ + f ,rad = 0 (7) In this equation, Sfi0 is the radiation-free spread rate given by Equation 3 with N u l f = 1 and Sf,ra d = (A a g2 / ps CpiS xs (Tv - Too))1 7 3 is the spread rate without conduction to the fuel bed. Upon inspection, Equation 7 (with numerical solution plotted in Figure 4) shows that increasing gas-phase radiation should increase Sf in thin fuel. Since the gas-phase radiation is a function of the properties to the gas mixture, changing the diluent from nitrogen (with a small A) to C 0 2 (with a larger A) or even SF6 (with a very large A) may increase the flame spread rate when dominated by this mechanism. Of course, the heat loss rate also increases, but the ratio of heat loss to heat generation will remain roughly constant or may even decrease due to radiative re-absorption. Interestingly, Equation 7 and Figure 4 also show that Sf is always higher with lower U if radiation is present. 14 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Width, W l Buoyancy induced velocity, U Width, W l Flame spread rate Towards fresh gas Figure 2. Cartoon showing difference in convection-diffusion length, 8 g, for 1 g on the left and |jg conditions on the right. Heat transferred to the surface Heat lost to the surroundings Optically thin - Radiation Heat Transfer Heat transferred to the surface Heat absorbed and re-emitted Hot Optically thick- Radiation heat transferred, Reabsorbed, and Re-emitted Figure 3. Illustration of difference radiation absorption and re-emittance effect (A effect). 15 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Flame Spread Solution U/Sfo = 0 U/Sfo = 1 U/Sfo = 3 U /Sfo = 5 3 — o S’ m 4 1 2 3 0 Sf.rad / Sf,o Figure 4. Numerical solution to non-dimensional thin fuel flame spread rate Sf in Equation 7 as a function of the radiation-only flame spread rate Sf,ra d at constant flow velocities, U. In summary, it is proposed that both radiative re-absorption effects and convective-diffusive zone thickness effects are very important when flame spread is dominated by radiation. Radiative re-absorption is strongly affected by the makeup of the ambient atmosphere (and thus the mean absorption length of the diluent). Meanwhile, the convective-diffusive zone thickness is strongly affected by the flow environment (and thus gravity). 2.1.3 Partially premixed flame spread The discussion in section 2.1.2 is appropriate when modeling opposed- flow flame spread in a pure oxidant / diluent atmosphere when radiative effects dominate. However, this is not always the case. When flame spread occurs in an enclosure, partially burned fuels become part of the atmosphere. When the atmosphere contains a gaseous fuel, it is considered premixed. However, 16 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. if the concentration of the premixed fuel is below the flammability limit, the flame will not propagate in the gaseous medium. This mixture is considered a sub-flammability limit gaseous fuel mixture. In the event that flame spread over a solid fuel occurs in an atmosphere with this added gaseous fuel and oxygen / diluent mixture, this is known as “partially-premixed flame spread” over a solid fuel. It has been found experimentally (Ronney et al,, 1995) that the addition of some combustible gases to an oxidizing atmosphere increases Sf at 1g substantially. The Sf may even increase by a factor of three with added gaseous fuel even if the total O2 mole fraction (including the CO and 0 2 parts) is fixed. The chemical reaction rates of the premixed fuel were found to have a substantial impact on Sf, even at higher 0 2 concentrations where finite-rate chemistry of the non-premixed fuel does not affect Sf significantly. Furthermore, a surprising result was found for added CO fuel, Sf is actually greater and the most dilute atmosphere capable of supporting combustion is weaker, when for a fixed total number of oxygen atoms in the gas phase, some of the oxygen atoms are in the form of CO rather than 0 2. For example, at 21% 0 2 - N2 with no CO, it was found that for a particular thin fuel S f« 0.21 cm/sec, whereas at 14% 0 2 - N2 with an equivalence ratio ( < j> ) of 0.5 (corresponding to 14% 0 2 and 14% CO, thus the same percentage of oxygen atoms), S f* 0.30 cm/sec. Also, the lowest value of 0 2 mole fraction that would support flame spread was about 0.17 at < | > = 0.5 and 0.10 at c f > = 0.5; 17 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the ratio of O atoms in these two atmospheres is 17 / (10 + 5) » 1.13. Moreover, these data did not even account for the fact that in practical fires the partially burned air will be hotter than ambient due to the heat release associated with the partial oxidation. It was shown (Ronney et al., 1995) that these results could be modeled with reasonable accuracy by treating the effect of the gaseous fuel as a thin planar semi-infinite premixed flame sheet located at x = xp < 0. Here, x is the spatial coordinate parallel to the fuel bed, upstream of the usual non-premixed flame as shown in Figure 5. This heat source was incorporated into the classical theory of flame spread over thermally-thin fuels developed by deRis (1969) and Delichatsios (1986) where a non-premixed flame sheet having mixing-limited heat release (i.e., infinitely fast reaction rate between the solid fuel vapors and oxygen) touches the fuel bed at x = 0 (i.e., xn = 0 ). Premixed Flame Front-Strength q” Ambient Environment F u e l/0 2/Diluent Non-premixed Flame Front Gaseous fuel Products of lean premixed flame Oxygen Heat Transfer From Flame Environment Flow Velocity, U Fuel vapor Solid Fuel Bed ► x Figure 5. solid fuel. ^ ^ ^ Flame Spread Rate S f x = x p x = x„ =0 Model for non-merged partially-premixed flame spread over a 18 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This analysis (Ronney et al., 1995) was divided into two configurations, non-merged flames (where the premixed flame and non-premixed flame were separated by some finite distance) and merged flames (where they were not). For non-merged flames, Ronney etal. (1995) used the equation for heat flux to the fuel bed (F) due to a premixed flame sheet, Fb e d = 2X gq” 1% pg Cp g U from Thomas (1982) and retraced the Delichatsios (1986) solution to obtain the relation for Sf with added heat flux due to the sheet. s _ 7t 4 T f ~Ty f 4 p sCp tS Ts Tv - Z ( \ \ 8 q" * * P g U C M { T f - X , \ (8) In this equation X refers to the conductivity, p the density, Cp the specific heat, x the fuel bed thickness, T the temperature, q” the heat release per unit area of the premixed flame front, U the opposed-flow velocity (which was buoyancy- driven since Ronney et al. did not impose any forced convection) and the subscripts g, s, v and o o refer to the gas mixture, solid fuel bed, vaporization condition and ambient condition, respectively. Note that the contribution of the non-premixed flame is represented by the first term inside the brackets while the contribution of the premixed flame is represented by the second term inside the brackets. From this equation, it can be seen that the added fuel increases q” and thus Sf beyond that of the non-premixed flame alone, which is consistent with experimental results (Ronney et al., 1995). Another prediction, which has not yet been previously tested, is that as U decreases (e.g., by decreasing buoyancy effects through the use of a low-gravity 19 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. environment) Sf should increase. Since U is not buoyancy driven in |jg, it becomes comparable in scale to the flame spread rate (Sf) of around 1 cm/sec. Therefore, the effect of raising the concentration of gaseous fuel should be greater at gg than at 1g. To estimate q”, the asymptotic analysis (Hamins et al., 1985) of partially-premixed flames in stagnation-point flow by was employed by Ronney et al. (1995). Rearranging and dimensionalizing Equations 5 and 20 to 23 in Hamins et al. (1985), Ronney et al. (1995) obtained a solution for q”. In Equation 9, % p is a scaled value of xp determined by solution of five equations for unknowns xp , % p, t p, yc p and m. This analysis will not be repeated here because Ronney et al. (1995) described the equations and solutions to the unknowns in detail. One point of interest, though, is that q” (and thus Sf) was strongly related to the reaction rate of the system and thus temperature of the premixed flame. Also, in Equation 9, the term s » L I / 5 « 1.2 (g2 / ag )1 /3 is the characteristic strain rate, induced by buoyant flow (deRis, 1969). At pg, s * U t o t a i / 5 * ( U b u o y a n c y + S f ) / 5 * ( 1.2(g2 / c c g )1 /3 ) + (Sf / S) * Sf / 5. In Hamins et al. (1985), s is presumed constant throughout the flow. This is not the case in the current experiments, but a more accurate model of 20 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. partially-premixed flame spread would require both detailed chemistry and hydrodynamics. Ronney et al. (1995) submitted that a merged-flame model occurs at (Tf - Tp ) / Tp < 0 P ‘1 , where 0 P = Ep / R Tp . In this case, the location of both flames occur at the point where x = xp (the same as x - xp =0 ) and consequently for merged flames we obtain a relation for Sf. s r = ' T t'' f ^ \ ( ( T f - r . ) l \ 4 j ^ P ps ^ s J I C c I 8 (10) For the merged flame, Tf in Equation 10 can be estimated. Tf = T« + cr In 1 - 1 + T _ In < / > \ - J (11) Here, r\ is the stoichiometric fuel / (oxygen + inert) mass ratio, Q is the heating value, Lv is the latent heat of vaporization of the solid fuel, YP i0 0 is the ambient mass fraction of premixed fuel, § = vp YP i0o/Y0 2,® is the equivalence ratio of premixed fuel in the ambient atmosphere, Yo2,° o is the ambient mass fraction of oxygen in the ambient atmosphere and the subscripts n and p refer to the non- premixed (solid) and premixed (gaseous) fuel, respectively. It can be verified from Equation 11 that the effect of gaseous fuel on Tf for the merged flame is minimal under conditions where there is a substantial amount of inert gas in the oxidizing atmosphere (i.e., when r\„ is small, as with hydrocarbons burning in air), therefore added gaseous fuel has little effect on Sf under most merged- flame conditions. In fact, for some cases Tf may decrease with increasing 21 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. gaseous fuel concentration if the Cp of the fuel is larger than that of the diluent. This is consistent with the experimental observations of Ronney et al. (1995). Also, since Equation 11 does not contain any U-dependent terms (under merged-flame conditions and with a low < j> ) , the effect of added gaseous fuel should be minimal for merged flames at pg just as it is at 1g. Using Equations 9 and 10, and connecting them at the merging point, Sf can be predicted for any value of « j > . In this case the curve will not be continuous whereas in reality, the transition from a merged flame to a non-merged flame would occur more gradually as increases. 2.2 Concurrent-Flow Flame Spread For concurrent-flow flame spread the convective and diffusive transport are in the same direction. It has been shown (Williams, 1976) that this causes more fuel surface area to be exposed to the hot combustion products as time passes, which results in an increasing rather than steady flame spread. It was also shown (Fernandez-Pello, 1984) that when using a boundary layer type analysis, the flame length and flame spread rate increase with time for both buoyant and forced convection as shown in Table 1. This is because as the heat is transferred from the flame to the fuel bed, more fuel vapor is generated leading to more heat release and thus a larger flame and thus a larger spread rate. These theoretical results are not entirely satisfactory because it is predicted that the flame length (Lf) and flame spread rate (Sf) will increase 22 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. indefinitely. In contrast, it is well known experimentally that even for concurrent-flow, both Lf and Sf can be steady under some conditions (Fernandez-Pello & Hirano, 1983; Grayson et al., 1994). Because the analyses of concurrent-flow flame spread described above assume adiabatic burning across an infinitely wide sample, heat and momentum losses were not considered. When considering these losses, it seems unlikely that, for a sample of finite width, the boundary layer thickness could grow to a value larger than the sample width. Also, it seems unlikely that the flame length could grow indefinitely, because the radiative loss from the fuel surface would eventually grow to the point where it would exceed the heat generation rate. Fuel Type Buoyant convection Forced convection Thermally thin Sf~tJ, L~t 4 Sf~tJ |, L ~ f Table 1. Predicted increase in spread rate (Sf) and flame length (L) with time (t) for laminar concurrent-flow fires. Using boundary-layer analyses, it has been predicted (Fernandez-Pello, 1983) that flame length (L) and Sf for concurrent-flow (SfiC O n ) increase indefinitely with time (t) as seen in Table 1. In contrast, some experiments using thermally-thin fuels (Delichatsios, 1989; Fernandez-Pello, 1983) show steady L and SfjC O n . The analyses assumed adiabatic spread across infinitely wide samples, thus heat losses and lateral momentum losses were neglected. With such losses, the boundary layer thickness (5) could not grow substantially larger than the sample width, W. If 8 is limited, then L and Sf are also limited. Even for infinitely wide samples, L could not grow indefinitely because surface 23 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. radiative losses would eventually exceed heat generation rates. Both assertions arise because (for boundary layer flows) the fuel bed heat flux (Q), and thus fuel vapor and total heat generation rates, increase more weakly than linearly with L. At the same time, heat and lateral momentum losses increase roughly linearly with L. Also, it has been suggested (Markstein & deRis, 1972) that for thermally-thin beds, fuel burnout could limit Sf, but not for practical sample dimensions. Therefore two hypotheses for steady, loss-limited flame concurrent-flow flame spread are proposed below. Hypothesis 1: For sufficiently narrow fuel beds, L grows until 8 » W , when transverse heat and momentum losses prevent further growth of L, which limits Q and thus Sf. Hypothesis 2: For sufficiently wide fuel beds, L grows until surface radiative loss (treated as a surface loss with no gas phase re absorption) is comparable to Q, when these losses prevent further growth of L, which limits Q and thus Sf. These are defined as convectively-stabilized and radiatively-stabilized flames, respectively. Although heat and momentum losses were considered, no finite- rate chemistry effects were considered. Consequently, these hypotheses apply only far from extinction conditions. It should be noted that there are no 24 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. adjustable parameters nor is it necessary for supplemental empirical quantities such as surface heat fluxes (Markstein 1972; Zhou, 1993) or pyrolysis times (Delichatsios, 1995). Pyrolysis here is defined as when heat (from the flame) causes molecules to cleave at their weakest points to produce smaller fragments (essentially vaporizing the solid just before it is burned). In flame spread over (an initially white colored) paper, the pyrolysis front or zone will be defined as the black or brown section paper. 2.2.1 Modeling predictions 2.2.1.1 Flame lengths If the assumptions in Section 2.2 above are true, then a scaling analysis should predict flame spread properties like flame length, spread rate, and stabilization mechanism for a wide range of conditions. It has been shown (Fernandez-Pello, 1984; Williams, 1976) that boundary-layer analyses are appropriate for concurrent-flow flame-spread analyses. Therefore, for forced-convection flame spread, it is assumed that 6 = LA R ef3 and N ul = B Rei_b (where A, B, a, and b are constants), where Nul is the length-averaged Nusselt number, Rei_= U L /v g the Reynolds number, U the forced convection velocity and vg the kinematic viscosity. For buoyant-convection dominated spread, it is assumed that 6 = L C G rf0 and Nul = D GrLd, where Gn_= g L3 / vg2 is the Grashof number. It is also assumed that the Prandtl number (Pr) is close to unity and that the thermal 25 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. expansion term generally present in G n_ is close to unity. This assumption is reasonable since the product density is 5 to 8 times smaller than the reactant density. For laminar flow, classical models yield (based on the momentum boundary layer thickness, for Pr=Q.72) A=0.664, a=1/2, 8=0.595, b=1/2 (Schlichting, 1960) and (defining 8 as the horizontal distance from the velocity maximum) C=1.37, c=1/4, D=0.476, and d = 1/4 (Gebhart, 1988). For turbulent flow A= 0.14, a=1/7, 8=0.0131 and b=6/7 (White, 1974), and (at GrL< 1 0 1 0, corresponding to all conditions within experimental parameters) C=0.030, c=0.10 (Cheesewright, 1968; Mason, 1974), 0=0.474, and d=0.25 (Churchill, 1972). Hypothesis I states 5«W, consequently L /^ r * A ~X-a R e w^ - “; Nu L * BA Re Re r v g (forced convection) (1 2 ) and (buoyant convection) (13) In these equations, all gas properties are temperature-averaged. Note that L / W and Nul are expressed through known experimental conditions Rew or Grw rather than unknown Rei. or GrL. 26 W c 4 - 3 c Gr w -3-d/ DC A~icGr w d X-3 c-G r v s Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Hypothesis II states that Q=(NuL kg / L)(Tf- Tv ) equals the radiative loss from the bed (H)=g s (Tv 4- Tm 4), where A .g, a, s, Tf, Tv, T o o , are the gas thermal conductivity, Stefan-Boltzman constant, bed emissivity, flame temperature (deRis, 1968), vaporization temperature, and ambient temperature, respectively. This leads to the relations below. (forced convection) (14) Y w » r / ^ P l / ^ G r / ' - ^ ^ N u L » E /l-idP lwM /^ “Grwd/'- ^ (buoyant convection) (15) In these equations, Plws kg (Tf - Tv) / W s a (Tv 4 - T*,4 ) is the Planck number. Equations 12 through 15 then yield predictions for L / Wand Nul. Buoyant convection Stabilization type Nul UW Convective - 3 d d DC *~3c Gr y f 3c -1 c C i-3cGr^-3c Radiative 1 3d d D l~3d PI y 3d Gr^~3d 1 1 d D l- 3d Pl^-3d Grjy3d Forced convection Stabilization type Nul L/W Convective - b b BA 1- a Re ] f a -1 a A l-° Re ffFa Radiative I b b Bl~ aP lyfa Re ]fa 1 1 b Bl~bP ljfb Re jf b Table 2. Predicted relations for the steady values of Nul and L/W for forced or buoyant convection and convective or surface radiation. Predictions are the same for thermally-thin and thermally-thick fuels. Note that since Rew ~W, Grw ~W3 and Plw ~ W 1 , L is independent of W for radiatively-stabilized flames. 27 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.2.1.2 Spread rates The concurrent-flow flame spread rate (SfiC O n) is estimated by equating Q to the rate of fuel bed enthalpy increase (which is equal to ps Cp,s xs (Tv - Too) W Sf, where ps, C P iS and xs are the fuel bed density, heat capacity and thickness, respectively) as shown in Section 2.2.1.1. Thus, for thermally-thin fuels, a non-dimensional flame spread rate can be expressed as s /,« » _ 4 „ X , ( T r -T S \ N u l . wf,ere S, S 7 1 W l,c,c f ’opp 4 n C t - T f ,opp rs p,s s ^ v o o J L . L i :. y u (16) where for compactness Sf,C O n is referenced to Sf for laminar, opposed-flow flame spread (Sf,o p p ) as discussed in Section 2.1.1. Combining Equation 16 with Nul from Table 2 yields predictions for Sf.con referenced to Sf,o p p shown in Table 3. For forced flow, U o pp = U is prescribed; for buoyant flow, Uo p p cannot be prescribed. An estimate was made (deRis, 1969) of Uopp* E1 /3 (g vg)1 /3 , where E * (0.72/Pr)(Tf- Tv) / T«. Stabilization type/fuel type Buoyant convection Forced convection Convective/thin . - 3 d d — D C 1_3c Gr^~3 c n - b b — BA l ~ a Re ] f a n Radiative/thin 4 ..1..........It - _it.. - D ^ dP l ^ dG r ^ d n A - J _ _ L . b - B 1 ~bP ljf b R e }fb n Table 3. Predicted relations for steady values of Sf,con/Sfi0pp for thin fuels in forced and buoyant convection, and convective and surface radiative loss stabilization. Since Rew ~W, Grw ~W3 and Plw ~W 1, S fiC O n is always independent of W for radiatively-stabilized flames. 28 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This analysis is readily extended to unsteady spread by neglecting loss mechanisms and setting Sf = dL / dt rather than Sf = constant. This leads to first-order differential Equations for L(t). For example, for thin fuels under buoyant flow Sf can be expressed as shown in Equation 17. Sfponif) d l A AC AC — * —NuL(f)Sf « GrL (t)cSf = - Gr, a t 71 71 7t \ W ) 3 c f,°pp (17) The solution to Equation 17 is shown in Equation 18. X -3c 1 Uc{l-3c) ( G r C c ^ o 1 - 3 c 71 V W J f,oP P J f,con For laminar flow (c= 1/4), 3c/ 'l~3c (18) $ fp o n ~ 4 ri4 W . 3 O/;0 p p I 0 -7 -3 ‘ (19) which has the form Sf ~ t3 proposed (Fernandez-Pello, 1984) in Table 2. The other relations in Table 2 can be derived similarly. Thus, the approach is considered quite general. 2.2.1.3 Transitions between regimes Transition between laminar and turbulent flame spread occurs when ReL or exceeds critical values, denoted ReL*»5x105 and GrL ’«4x108. By writing Rew = ReL /(L/W ) and Grw = G n_ / (L / W)3, with expressions for L / W 29 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. taken from Table 2, it is inferred that transition occurs, for convectively- stabilized flames, at the point shown in Equation 20. For radiatively-stabilized flames, transition occurs at the point shown in Equation 21. Transition between convective and radiative stabilization occurs when the predicted L are equal as shown in Equation 22. Each of these regions can be described as convective-laminar (CL), convective-turbulent (CT), radiative-laminar (RL), and radiative turbulent (RL). The regions and transitions for thin fuels with buoyant flow are plotted in Figure 6 for helium, nitrogen (air), and sulfur hexafluoride diluents at different oxygen concentrations and pressures. (forced); Grw = C 3(GrL^ (buoyant) Rer = 5 I(Rez) Pt^ (forced); Grw= D 3 (GrLj (buoyant) (21) m 1f V (forced); Grw ^ 3 c (buoyant) (22) 30 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. io r * i 1 - - - - - -1 —i — i — r~r ~ g 1 0 5 107 k, Radiative- ] : turbulent (RT):J^ 46% Q (D J2 10 E c 10 6 , 3 atm Convective - urbulent Direction of increasing fuel bed width m c o 104 ^ ^ .^ - H ^ O .a S a t m 0 10' 1 0 ' JflRadiative -| |laminar |RL| _ i i f ■ i iii" 4#flConvective - '"'Jaminar (CL) -f"v:vi:- ' f « ! 10 -1 10° Planck number (PI ) ' w Figure 6. Predicted regimes of concurrent-flow flame spread for buoyant convection, showing the type of flow (laminar or turbulent) and flame stabilization (convective or radiative). Also shown are lines corresponding to fixed atmosphere but varying fuel bed width (W) for air and the atmospheres yielding the lowest and highest Plw and Grw experimentally possible in this proposal, i.e., 0.25 atm (V H e and 3 atm O2-SF6 , respectively, for TV =618K, T«=300K and s=1. 2.2.1.4 Comparison with previous results Relatively few experimental or computational results are available for comparison with the predictions in Sections 2.2.1.1 through 2.2.1.3. It has been shown (Fernandez-Pello, 1983) that for thin fuel buoyant-flow experiments at low pressure (P) in 30% O2 / 70% N2 atmospheres with small W (~10 mm) that Sf,c o n ~ P1 '8. This is close to the prediction of Sf~ P2 for CL or RL spread (with c = d = 1/4, Sf ~ Grw1 (CL) or Sf ~ Grw1 Plw3 (RL), since 31 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Grw~ vg '2~ P2 and Plw- V ~ P°, S f~ P2 ). In contrast, it has also been shown (deRis, 1969; Zhang & Ronney, 1992) for downward (opposed-flow) flame spread that Sf ~ P°. It should be noted that it was also shown (Zhou, 1993) that when grid turbulence (with ReL < Ret* for all test conditions) and turbulence intensity were employed, there was little effect on Sf. Therefore, laminar values of a and b apply. Adiabatic analyses results (as shown in Table 2) also predict SfiC O n ~ U1 , but predict L ~ t1, whereas the non-adiabatic analysis predicts steady L ~ Rew1 . Unfortunately, no time-dependent data on L were reported (Loh, 1984; Zhou, 1993) to compare adiabatic and non- adiabatic models. When Ferkul and Tien (1994) modeled concurrent forced- flow flame spread over two-dimensional thermally-thin samples with surface radiative loss, they predicted steady spread with Sf> C O n ~ U1 (whereas Sfi0pp ~ U° shown by deRis, 1969) and L ~ U1. These predictions are consistent with Tables 2 and 3 for RL spread. In contrast, adiabatic analyses predict S f.co n ~ t1 for these assumptions as shown in Table 1. Jiang etal. (1996) found S f.con ~ g1 and L ~ g1 for concurrent buoyant spread, again consistent with RL predictions. 32 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 3 OBJECTIVES AND APPROACH As mentioned previously, a substantial number of flame spread studies (both analytical and experimental) has already been conducted by many other investigators. However, the effects of radiation of the gasses and partially- premixed atmospheres in opposed-flow flame spread and heat and momentum losses and radiative losses in concurrent-flow flame spread have not been discussed. This work will study these effects. 3.1 Opposed-Flow Flame Spread It was shown in Section 2.1.2 and 2.1.3 that opposed-flow flame spread could be affected by radiation and / or added gaseous fuel. To determine the importance of radiative re-absorption of the atmospheric gasses, we will measure the opposed-flow flame spread under 1 g and pg while varying the oxygen concentration, pressure, and diluent type (He, Ar, N2 , CO2, or SFe). To determine the importance of added gaseous fuels, we will measure the flame spread rate while adding sub-flammability limit gaseous fuels such as CO or methane. Section 2.1.2 showed that gas phase radiation can increase Sf across thin fuel beds, and that when such effects are important an imposed flow can affect Sf in unusual ways. These facts may have significant implications for spacecraft fire extinguishment systems using CO2 or other radiatively-active 33 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. diluents. Therefore, the flame spread rate and gas-phase radiation effects on Sf (through its effect on qr to the fuel bed at 1g and pg) will be studied by varying the mean absorption length of the diluent gas (He, Ar, N2, CO2, and SF6). Note that this effect should not be confused with heat loss via gas radiation. The spread rate should scale in a manner comparable to those predicted by Equation 7. Since the gas phase radiation flux is also related to the convection-diffusion zone thickness, which is affected by the induced buoyant flow velocity, experiments will be performed at both 1 g and pg. For opposed-flow flame spread with a partially-premixed gaseous fuel, the goal of a portion of this work is to determine whether the model for partially-premixed flame spread in Section 2.1.3 is applicable to pg as well. 1g experiments (Ronney etal., 1995) are in good agreement with Equations 8 and 10 of section 2.1.3. It was suggested that even small amounts of added gaseous fuels would greatly affect the q” term in Equation 11, thus raising the flame spread rate for non-merged flames. Lowering the induced opposed-flow velocity, U, would increase this effect even further. However, no testing has been performed at pg conditions (where the U would effectively be Sf) for partially-premixed flame spread. It is therefore an objective of this work to determine how well the theoretical models of Equations 8 and 10 for non merged flames and merged flames, respectively, compare with reality. To study the influences of finite rate gas-phase kinetics on partially- premixed flame spread, two different gaseous fuels having differing 34 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. characteristic chemical reaction rates will be used, namely CO and CH4. CO is chosen because of its practical importance to partialiy-premixed flame spread, which could occur in building fires. CH4 is chosen as a representative hydrocarbon fragment, which could result from pyrolysis of nearby solid fuel beds. Since radiation effects will be covered in another section, these tests will be performed in 0 2~ N2 atmospheres only (minimizing the radiative effects). Both 18% 0 2- N2 and 30% 0 2 - N2 atmospheres were employed. The former (18%) is close to the minimum 0 2 concentration and thus finite-rate chemistry effects may be significant. The latter (30%) is far from the limit and thus finite-rate chemistry effects are probably not significant. In summary, for the opposed-flow section of this study, the effects of gas phase radiation and added sub-flammability limit gaseous fuel on the opposed-flow flame spread rate and steadiness will be determined by varying the diluent (He, N2, Ar, C 02, or SF6 ), oxygen concentration (from the extinction limit up to around 50% 0 2 ), the pressure (0.25 up to 6 atm), and environment (1g and pg) for radiation effects. And the premixed gaseous fuel concentration (CO and CH4 ) and oxygen concentration (18% and 30%) will be changed for added gaseous fuel effects. These results will be compared to models in sections 2.1.2 and 2.1.3. 35 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.2 Concurrent-Flow Flame Spread It was shown in Section 2.2.1 that current models predict flame lengths and flame spread rates increase over time for most conditions. It was later shown that by adding transverse heat and momentum losses to current models, steady spread could be obtained because the flame lengths and thus spread rates were limited by these loses. These effects are not associated with finite-rate chemistry. Combining these loses with constant coefficients for turbulent or laminar flow results in CL, CT, RL, and RT stabilization regimes shown in Figure 6 . Another goal of this study was to determine in which stabilization was the limiting mechanism. This can be accomplished by comparing flame lengths and flame spread rates to the models proposed for each limiting mechanism in Section 2.2.1. For example, if the flame length and flame spread rate were independent of sample width (for some range of Grashof number), this would indicate the stabilization mechanism was RT or RL. Determination of RT or RL can be accomplished by comparing the left hand side (L/W) and right hand side of Equation 15. In this study, we will focus on models for buoyant convection where the flame spread rate is related to the Grashof number as shown in Table 3. Therefore, a set of experiments will be conducted for a large range of Grashof number by changing the gaseous atmosphere and width. Since Grashof number is equal to G rws = g W3 / v g2, changing the gaseous atmosphere (diluent or pressure) will affect vg. Gases with a very low vg such as SF6 will have larger Grashof 36 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. numbers while gases with a very high vg such as He will have a lower Grashof number. In addition, v g ~ P' 1 so a higher pressure will result in a lower vg and thus a larger Grashof number. The fuel sample width will have a cubic effect on the Grashof number. A very small width will have a lower Grashof number while a larger width will have a larger Grashof number. In summary, in the concurrent-flow section of this study, the effects of heat and momentum losses (for narrow widths) and radiation losses (for large widths) on the flame length, spread rate, and stabilization mode will be determined by varying the diluent (He, N2, Ar, CO2, or SFe), oxygen concentration (from the extinction limit up to 50% O2), the pressure (0.25 to 3 atm), the sample thickness, and width. 37 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 4 EXPERIMENTAL APPARATUS AND PROCEDURES Both the opposed-flow and concurrent-flow flame spread objectives require changing the gaseous atmosphere. Small errors in mixing may cause results to vary significantly (especially for partially-premixed flame spread where small changes in the gaseous fuel concentration can double or triple the flame spread rate). To minimize error, a very repeatable AND precise method of mixing is required. Therefore, all oxidizing atmospheres are mixed using a computer controlled Partial Pressure Mixing System (PPMS) described in Section 4.1. For opposed-flow flame spread, two properties were measured: flame length and steady or unsteady spread rate. To accomplish this a dedicated apparatus was developed. Since pg can only be accomplished in drop towers (NASA aircraft can fly low-gravity parabolas but the g-jitter degrades the results and sounding rockets, the space shuttle, and space station experiments are cost prohibitive due to the large number of data points required), a very robust system, capable of sustaining impact loads in the drop tower, with built in data acquisition was necessary. This apparatus, used for downward (opposed-buoyant-flow at 1 g) and microgravity (negligible-buoyant- flow), is described in Section 4.2.1. For concurrent-flow flame spread, three important properties were measured: flame length, flame spread rate, and flame spread steadiness. To 38 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. accomplish this a different dedicated apparatus was developed. Since the flame spread rate takes a finite time to become steady, a tall sample is required to reach steady state conditions. This apparatus, used only for upward (concurrent-buoyant-flow at 1g) flame spread, is described in Section 4.2.2. 4.1 PPMS A Partial Pressure Mixing System (PPMS) was built to prepare the oxidizing atmospheres used in the flame spread experiments. The PPMS used an Omega PX-623 Pressure Transducer with a range from 0 to 50 psi and an accuracy of 0.1% of full scale. ASCO U52257 electric 110 VAC unidirectional valves were used in series (with flow control in opposite directions) to control the flow. The flow was fine tuned using Whitey SS-ORS2 manual needle valves. The entire system used rather small 1/8” stainless steel tubing to reduce the amount of dead volume. A 22-bit Analog Devices RTI - 870 analog to digital (A / D) ISA card was used to control the electric valves and to read the voltage from the pressure transducer. An illustration of the PPMS is included in Figures 7 and 8 . The software, written in Turbo Basic (included in Appendix B) by Quin Blackburn and Paul Ronney, was compiled and executed using a PC to control the A / D card. 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. To external relay for Chamber (3 pin) T Analog Digital Output Input To Power 110 VAC (3 pin) Power Fuse Switch (3 amp) i r T T T r m r r r a SO • * , U S .■S'.* 1: 1 • S . t t i m z* To Cham ber a n d Gases Electrical Device Stainless Steel Gas Lines Gas Mixture Device Insulated Wires Figure 7. Schematic of PPMS internal arrangement. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Exhaust Gases Out Gas' Gas Tanks Gas Fuel Vacuum Power Power Valve Diluent Oxidizer Partial Pressure Mixing System Chamber Analog I/O. JDigital I/O Analog Devices RTI 870 Board Personal Computer Data Hard Drive o Figure 8 . Schematic of PPMS external arrangement. The PPMS uses the following method to mix the gases: I) Initial vacuum and diluent fill: 1) The software is informed of the desired atmosphere (diluent, oxidizer, fuel, pressure, etc). 2) The PPMS evacuates the chamber pressure to less than 0.01 atm. 3) Internal 110 VAC chamber fan is turned on. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4) The PPMS fills the container to 1 atm with the diluent (N2 i Ar, He, C 02, or SF6 ). 5) The PPMS again evacuates the chamber pressure to less than 0 . 0 1 atm. 6 ) The software estimates the chamber pressure after thermal expansion by taking pressure readings over time, curve fitting the data, and extrapolating the final pressure. 7) The PPMS fills the chamber to approximately 50% of the desired partial pressure of the appropriate diluent. 8 ) The software estimates the chamber pressure after thermal expansion. II) Oxidizer and/or catalyst fill: 1) The PPMS adds 80% of the desired partial pressure of oxidizer. 2) The software estimates the chamber pressure after thermal expansion. 3) The PPMS adds 80% the remaining desired oxidizer (this process is repeated until the entire 1 0 0 % of the desired oxidizer is added. 4) The software estimates the chamber pressure after thermal expansion. 42 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5) The PPMS adds the desired partial pressure of fuel (procedure is the same as the oxidizer steps 1 to 4). 6 ) The software estimates the chamber pressure after thermal expansion. 7) The PPMS adds the desired partial pressure of catalyst (procedure is the same as the oxidizer steps 1 to 4). 8 ) The software estimates the chamber pressure after thermal expansion. Ill) Final diluent fill: 1) The PPMS adds diluent to the final desired pressure (procedure is the same as the oxidizer steps 1 to 4). 2) The software estimates the chamber pressure after thermal expansion. 3) The software estimates the final constituent concentrations and displays them. 4) Internal fan is turned off. 43 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.2 Flame Spread Apparatus 4.2.1 Microgravity Apparatus The Microgravity Interferometer Drop Apparatus System (MIDAS) Rig was built as shown in Figure 9. This “MIDAS Rig” was a self contained experiment consisting of a frame, chamber, battery boxes, interferometer, video, computer, and other various components. Microgravity is obtained during a freefall at the 2.2 Second Drop Tower Drop tower at the NASA Glenn Research Center (GRC) in Cleveland, Ohio. The MIDAS Rig fell freely within a special drag shield that reduced the effects of air drag during the fall. The MIDAS Rig landed at the bottom of a 2.2 second drop tower (80 feet) onto an airbag shown in Figure 10. The resulting impact caused g-loads from 10 to 50 (usually 25) times that of 1g. Sometimes, steady state flame spread cannot be achieved in the limited time of the drop tower so a longer duration is necessary. For 5 seconds of freefall, the Zero-g facility at the NASA GRC was used. The MIDAS Rig freefell 500 feet in a vacuum chamber and landed in foam pellets (decelerating for 10 to 15 feet). The resulting impact ranged from 50 to 100 g’s and averaged about 60 g’s. These high loads required the entire apparatus to be over-designed for over 1 0 0 g’s (though many components failed upon impact requiring repair). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Drop Frame Shearing Plate Window Window Mirror Digital Image Processing System Test Chamber Beam Laser Expander Mirror < — 41 Fiber-optic Link Mirror Diffuser VCR Side View Figure 9. Schematic of “ The Rig” (MIDAS). A torturous path is needed to fold the interferometer beam to fit within the drop rig envelope. The dropping apparatus to requires its own power, computer, camera, etc. during pg. Images obtained are sent through a fiber optic line that connects the free- falling rig to VCRs. The rig lands on an airbag 8 stories below. 45 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 10. Experiment in Pre-drop position £ > Experiment preparation area Airbag Cut-away view or 2.2 second drop tower. 46 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ^09753 ^ 4.2.1.1 Drop Frame The high impact loads required that all components were over-designed for strength. The MIDAS Rig frame was 95 cm wide by 40 cm deep by 85 cm tall. It consisted of 4 - 0.5 cm thick sheet metal aluminum sections (top, bottom, and 2 sides) riveted together (supplied by NASA GRC). For strength, each corner had a triangular reinforcement that created a perpendicular connection. A schematic of the MIDAS Rig and its components can be seen in Figure 9. For rigidity, the bottom sheet was reinforced with another 1 cm thick sheet of aluminum on which the optics were mounted. Sandwiched between the two sheets of aluminum were two -1 5 cm by 5 cm by 75 cm long “C” beams. The two beams were mounted face down in the shape of a digital “M” with the top sheet of aluminum resting on top. 4.2.1.2 Chamber The chamber was made of 1 cm wall thickness aluminum pipe. It was 24 cm in diameter and 36 cm tall. It had a 10 cm wide by 15 cm tall by 2 cm thick Lexan window in front for video and human viewing and a 10 cm diameter window on each side for video and interferometer access. A window in the back of the chamber was available, but filled with an aluminum blank for fire safety and serviceability. The chamber had a 1.5 cm thick lid on the top held on by 8 - 5/8” diameter grade 8 bolts. An 1/8” o-ring sealed the lids and windows. Parker O-lube was used to on the o-ring to ensure a leak free seal. 47 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The window mounts and lid mounts were welded to the chamber. The square- bottom end, which also served as a mounting flange to the frame, was also held on by 8 bolts and was 1.5 cm thick. The chamber was also supported by another 1 cm thick plate to prevent leakage during frame flex and impact. This chamber was hydrostatically tested by U. S. Testing Labs in Los Angeles, CA to a pressure of 220 psi. At this pressure the chamber leaked but did not burst. At 150 psi, the chamber did not leak. All tests were conducted between 0 and 1 2 0 psi and a pressure relief fitting (with an opening pressure of 150 psi) was attached to the gas lines. The PPMS was used to leak check the chamber and found less than a 10‘7 atm per second leak rate. Several electrical feed-throughs were available for power and data. Each feed-through had a V z NPT connection. The feedthroughs were stainless steel %” swagelock connectors. The top chamber lid had no feedthroughs and thus could be opened without breaking any fluid lines. 4.2.1.3 Imaging Systems The MIDAS Rig had three imaging systems. System one was a high resolution color CCD camera mounted outside the front window. Example images are shown in Figures 11 and 12 for 1g and pg, respectively. Images were sent to an FM transmitter and sent via fiber optic cable coiled on top of the drag shield to an external SVHS recorder. Flame spread rate, color, shape, flame length, burnout zone, etc. were recorded. 48 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 11. Flame spread across a thick solid fuel at 1g in 40% oxygen with CO2 diluent at 4 atm. Note buoyancy induced convective flow. Figure 12. Flame spread across a thick solid fuel at pg in 40% oxygen with CO2 diluent at 4 atm. Reaction products tend to remain over consumed fuel. No flow disturbance seen. 49 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. System two was a shearing interferometer. A 15 mw laser beam is expanded from 0.1 cm to 4.0 cm then directed through the chamber. The change in density from a flame or other source then caused the light to bend slightly within the 4 cm diameter image. The laser exited the chamber, split into two beams, which followed almost the same path past a shearing plate. This plate reflected some of the light off the front of the plate while allowing some of the beam to pass into the plate and reflect off the back of the plate as shown in Figure 13. One laser traveled about 1 mm farther than the other laser and was thus just slightly out of phase. When both lasers finally reached the frosted window, the phase shift caused interference patterns or fringes. The fringes could be adjusted by changing the focal point of the expander, changing the shearing plate angle, or changing the distance of the laser path. The path taken by the expanded laser can be seen in Figure 9. The resultant image shows the side view of the flame with fringe distance inversely proportional to the change in density. Interferometer images can be seen in Figures 14 and 15 for 1g and pg, respectively. Notice particularly, that while the images may not even be visible using a standard video camera, the interferometer image clearly shows the convective-diffusive zone thickness discussed in section 2.1.2. The image was captured with a Sony XC-75 CCD color camera and sent via fiber optic cable to an external SVHS recorder. 50 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Shearing Plate Reflected beams Figure 13. Description of shearing plate. m m Figure 14. Interferometer image. Side view of Kimwipe sample burning at 1 g in 42% O2- SF6 @ 4 atm. Very narrow convective-diffusive zone thickness. Field of view is approximately 4 cm by 4 cm. Figure 15. Interferometer image. Same sample burning as Fig. 14 in pg. Flame (same scale) is substantially expanded over 1g counterpart showing effect of reduced buoyancy induced flow velocity. 51 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Source three was a black and white CCD mini camera mounted inside the chamber. It could be mounted with at an angle of 0 to 90 degrees, depending on the desired view. The main purpose of System three was to determine whether or not a fuel was burning on both sides. The MIDAS Rig used 2 power sources. The igniter and amplifiers used a 28 VDC battery box that consisted of seven 4-volt batteries. The maximum sustained output of the box was 1 0 amps (with bursts of 2 0 amps) for 1 hour. The typical consumption was 1 amp with 20 amp pulses (for ignition). The laser, cameras, fm transmitters, and computers were powered by a 12 VDC battery box consisting of 2 parallel-sets of three 4-volt batteries. The maximum sustained power output of this box was 20 amps. The typical consumption was about 3 amps. Due to the large power requirements of the MIDAS Rig, all components except for the computer were controlled via relay by the computer. The Rig was turned on at the top of the tower, dropped about 5 minutes later, and the computer (DDACS) turned all other components off 30 seconds after the drop to conserve power. 4.2.1.4 Internal Apparatus For thin fuels, laboratory Kimwipes were chosen as the fuel for three reasons. First, it produced minimal char, which could complicate interpretation of the experimental results. Second, it was thin enough (p x = 0.00179 g/cm2 ) that the fuel behaved as a thermally thin material (i.e., the conduction through 52 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the material is negligible compared to conduction through the gas phase). Third, because Sf was inversely proportional to px, for a given atmosphere Sf was higher for this fuel than most other materials commonly used in flame spread experiments. The sample was held in place using aluminum quenching plates on both sides. The resultant fuel sample size was between 0.5 and 10.0 cm wide by between 10.0 and 20.0 cm tall. Each sample was ignited from the top using a double “W” shaped wire as shown in Figure 16. The wire was approximately 16 cm long, therefore each section or “U” of the “W” is 4 cm. The igniter consisted of two high 28 VDC legs on each side with 1 ground leg in the middle of the “W.” 30 gauge kanthal wire was used because its resistance is almost constant as a function of temperature (while nichrome has a variable resistance). Since a large constant voltage across a short wire may vaporize it, a pulse mode was used for ignition. Usually, 3 or 4 pulses of 0.02 seconds (with 0.03 second intervals between pulses) were sufficient to ignite the sample. However, sometimes, 5 pulses of 0.05 seconds were needed. Due to the fact that the drop tower facility gives 2 seconds of drop time, a second more reliable igniter is used. A very small nitrocellulose membrane 0.5 cm wide and the thickness of paper is woven through the “W” and pulled taught against the paper. Since nitrocellulose does not need oxygen to ignite, it is less dependent on the atmosphere than tissue fuel alone. The nitrocellulose ignited after 1 or 2 pulses giving more time for the flame to reach a steady state. Also, due to the 53 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. rapid spread rate of the nitrocellulose, a suitable 2-D straight flame front was established. For most flames, the flow field developed by this rather violent ignition source was dwarfed by the propagating flame. However, for those very slow flames, the ignition source was of great concern because the initial plume of hot gas from the nitrocellulose would affect flame spread. The hot gas plume moved out around 1 inch in front of the advancing flame spread. This behavior would affect the flame spread in two ways. First, the gas temperature ahead of the flame was higher than the ambient temperature and, second, the flow field would then be in the direction of the flame (making it a concurrent-flow flame spread). Therefore, when gathering flame spread rates less than 1 or 2 cm per second, it was critical to minimize the amount of nitrocellulose used for ignition during a 2 second drop. It appeared that, for the slowest flame spread rates, the flame spread rate started at around 2 to 4 cm/sec and then slowed to 0.5 cm/sec within around 1 second (at which point the flame spread rate remained constant at 0.5 cm/sec for the remainder of the drop). This would suggest that the disadvantage of the ignition plume was minimized after 1 second. Still, an entire 1 second of pg data could be lost if the ignition source was too large. 54 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ground +24 VDC +24 VDC Ground Kanthal wire igniter +24 V Nitrocellulose membrane- Video camera << laser path Thermocouples Sample — Sample holder Cross Section Front View Figure 16. Interior view of thin fuel sample holder. Kimwipe samples held in place by aluminum quenching plates on both sides. Interferometer path shown as red circle while radiometers and thermocouples are optional. The procedure used to perform all thin fuel experiments had to be the same every time or pre-ignitions or camera inconsistencies would occur. A pre-ignition is when the apparatus ignites the sample before it is supposed to. This was due to the electrical relay used for ignition. When turning the power of the system on in the wrong sequence, the relay input is allowed to “ float” which allows a small current to leak through it. This can be thought of like an electrical fluid valve with no spring. If the power to the valve is turned on giving a solid “ off’ signal, then the valve shuts. However, if the fluid hits a valve with no signal, it may leak through. This leak was more than enough to 55 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ignite fuels in some environments. Camera inconsistencies may also occur when the battery power was low and the laser was turned on after the cameras. When the laser started, the resulting power surge caused the video cameras to loose sync. This was cured by keeping the batteries fully charged and always starting the laser before the cameras (which is written into the code, “before.ttb” shown in Appendix B). Thin fuel checklist and typical procedure: I) Fuel sample installation procedure: 1) Check Interferometer alignment. 2) Insert new fuel sample in chamber. 3) Align “ W ’ igniter. 4) Weave nitrocellulose membrane horizontally through “ W ” and clamp in quenching plates taught with kimwipe. 5) Test Resistance of “ W ” wire (should be between 1 and 10 ohms). 6 ) Align sample holder. 7) Close and secure lid. II) Atmosphere mixing procedure: 1) Plug in PPMS. 2) Start vacuum pump. 3) Ensure smoke and CO detectors are functional. 56 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4) Run PPMS to desired atmosphere (“PPMS.BAS” in Appendix B). 5) Wait at least % and hour. 6 ) Check battery status. 7) Recheck final pressure. 8 ) Disconnect PPMS and battery chargers. Ill) Pre-drop power-on procedure: 1) Load rig into drag shield. 2) Close drag shield and lift to top floor. 3) Open drag shield. 4) Turn on 12 VDC battery box ONLY. 5) Turn on D-DACS, and all other accessories. 6 ) Connect external computer and upload “before.ttb” 7) Ensure “ status” is displayed. 8 ) Run “before.ttb” (see Appendix B). 9) Press “ 4” for drop. 10) Ensure “please check connections” is displayed. 11) Flip drop tower switch to “off.” 12) Ensure video and interferometer are on and aligned. 13) Check output video for snow and adjust as necessary. 14) Start SVHS recorders. 15) Reset time code generators. 57 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16) Connect drop sensor pins. 17) Turn on 28 VDC battery box (for igniter and amps). 18) Ensure “please check connections” is still blinking. IV) Drop procedure: 1 ) Flip drop tower switch to “on.” 2 ) Ensure “ Thank you, Linton” is displayed. 3) Disconnect external computer and close rig. 4) Announce “stand by for a drop” on intercom. 5) Inform conductor that you are ready to drop. 6 ) Verbally count down “3, 2, 1, drop.” 7) Watch rig fall. V) Post-drop procedure: 1 ) After impact, wait 5 seconds and stop recorders. 2) Review video to ensure ignition as successful. 3) Estimate flame spread rate and record. 4) Return to 5th floor and wait for rig to be raised. 5) Open drag shield and remove rig. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. VI) Data download procedure: 1 ) DOWNLOAD DATA. 2 ) Connect external computer and type “new.” 3) Upload “after.ttb” (see Appendix B). 4) Type “ctrl-z.” 5) Type date and drop number as “082300-2.dat. 6 ) Type “run.” 7) Wait for all data to scroll across screen. 8 ) Type “new.” 9) Turn off both battery boxes. VII) Clean-up and storage procedure: 1 ) Reconnect PPMS and vent - vacuum - vent - vacuum chamber. 2) Disconnect PPMS. 3) Plug in battery chargers. 4) Thank operator. 5) Open chamber lid. 6 ) Ensure all surfaces are clean. 7) Prepare for next drop. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.2.2 Concurrent-flow Earth-gravity Apparatus Obtaining a steady flame spread rate, concurrent-flow (upward) flame spread at 1g required the use of a very tall chamber. One solution was to stack several identical 50 cm tall chambers end to end. However, the weight and cost prohibited this. Therefore, a slightly non-orthodox approach was used when assembling an experimental apparatus. A 4.0 meter tall, 25 cm diameter schedule 40 clear PVC tube, primarily used for transporting liquids, was acquired from Ryan Herco and cut in half. The tube had a maximum operating pressure of 130 psi. The final dimensions of the tube were 2 meters tall with a 25 cm diameter. The wall thickness was 1.0 cm. Flanges, each rated at 240 psi were thermally bonded to both ends of the tube. Feedthroughs were installed on the top for electrical and fluid transport as shown in Figure 17. The resulting pressure chamber was hydro tested to 100 psi. The PPMS was used to determine that the chamber had less than a 10'7 atm per second decay or rise rate. To reduce the possibility of the PVC chamber catching fire, a brass shim was rolled and inserted along the interior wall of the chamber. A 5 cm gap was intentionally left along the length of the tube to facilitate external viewing. An aluminum quenching plate was added to the interior base to reduce the chance of unintentional fires starting because of falling debris. 60 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Lid 2 - J * > o ■ o Electrical w iring to computer Rack/Vessel Umbilical Cable M ixing fan JWindow A lum inum clam ps Pressure Vessel Thermocouple ij-ik s a rack Test Rack Kanthal w ire ignition system I Oxidant Computer mumu Partial Pressure Gas M ixing System Figure 17. Schematic of the concurrent-flow flame spread experimental apparatus. The test rack goes inside the pressure vessel after fuel sample loading. The PPMS is disconnected from the pressure vessel before ignition. The Video data is taken through the front while thermocouple data is send to the computer. The Kanthal wire is connected to external power source. The fuel samples used in this section were also Kimwipe tissues from Kimberly Clark with an area density of 0.00179 gm/m2. Kimwipes were used because they do not leave any residue or char (similar to the opposed-flow series). They also burned readily in air and achieved a steady spread rate Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. very quickly. For the wider samples, steady spread did not occur instantly. 3 or 4 Kimwipes were used in series (ends butted together, not overlapping) to provide a sufficient length (usually 1 . 2 meters) and the width was cut appropriately. In some experiments, double thickness (i.e., 2 sheets placed one on top of the other) fuel samples were used. The fuel sample was held in place by 1.5 meter tall by 7 cm meter wide by 0.2 cm thick aluminum quenching plates on both sides. The quenching plates have length scale markings along their length and are held in place with a circular ring connected to the chamber. The center of the chamber had no structural members that might interfere with the flame progression. The sample was ignited using a 5 cm coiled 29 to 32 gauge kanthal wire with 28 to 40 VDC across it during ignition. The uncoiled length of the wire was approximately 20 cm. To facilitate ignition, extra tissue was wrapped around the coil. This extra tissue and fuel sample would simultaneously ignite. Flame images were collected through 3 sources. One camera was an external Panasonic VHS video camera was mounted 2 to 3 meters from the chamber approximately 1 meter above the floor. The camera was zoomed in 1 00% and focused. It was then zoomed out approximately 50% during the each experiment. However, for slower flame spread, the camera was sometimes zoomed in up to 90%. The flame shape, color, and length were also documented (more details can be found in Section 5.2.1). Additionally, 2 mini black and white cameras were mounted to the inside wall of the chamber. 62 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The approximate field of view of each was approximately 4 by 6 cm. One of the cameras was positioned in front of the flame approximately 1 meter from the floor. The other camera was positioned 10 cm above the first. Detailed flame shapes and a more accurate flame spread rate could be recorded using external VCRs. However, after only a few dozen experiments, both cameras caught fire and had to be removed. Due to the fact that the chamber was made of PVC, no more internal cameras were added in fear of a chamber fire. A set of 0.002 inch diameter type “S” (platinum/platinum-rhodium) thermocouples was used to measure the temperature profile of the passing flame. The thermocouple data was recorded by an Analog Devices RTI 800 A/D board at a rate of 100 hertz. The first thermocouple was mounted 2 mm behind the fuel sample approximately 1 meter above the floor and the fuel sample (x=2 mm, y=1 m). Due to the experimental setup, a minimum distance of 2 mm was required to reduce short circuits and thermocouple breaking. The next thermocouple was mounted 2 mm farther away from the fuel surface than the first (x=4 mm, y=1 m). The third thermocouple was mounted 2 mm farther away from the fuel surface than second (x= 6 mm, y=1 m). The final thermocouple was mounted 6 . 8 cm above the first (x= 2 mm, y=1 . 6 8 m). Temperatures, flame spread rates, and flame lengths were measured using an RTI 800 A/D board and recording thermocouple output voltages at 1/100 second intervals. An example of this measurement (already scaled to temperature) is included in Figure 18 . 63 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1400 L = (At at x = 2 mm, 900°C) x S ~ = 0.89 s x 99 mm/s \ = 88 mm 1200 1000 3 800 x = 2, y = 0 x = 4, y = 0 0> 600 Q_ x = 2, y = 68 400 200 Time (sec) Figure 18. Temperature profile as a function of time for upward spreading flame. Procedure: I) Sample Loading 1 ) Install new sample in sample holder. 2) Align coil igniter. 3) Attach internal feedthroughs. 4) Check power supply status. 5) Close and secure lid. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. II) Gas mixing procedure: 1 ) Switch valve to PPMS. 2) Run PPMS for desired gaseous atmosphere. 3) Wait at least 1/2 an hour. III) Run experiment: 1) Turn on video and camcorder. 2) Focus cameras and ensure they are recording. 3) Announce experiment. 4) Run software to collect data. 5) Connect power supply to external feedthrough. 6 ) Initiate burn. 7) Stop data collection software and VCR’s. 8 ) Disconnect power supply and external feedthrough. IV) Post run and cleanup: 1) Switch valve to bypass (vent). 2) Switch valve to PPMS and vacuum chamber. 3) Switch valve to bypass. 4) Remove lid. 5) Disconnect internal feedthroughs. 6 ) Remove sample holder. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 5 RESULTS Experiments showed that steady flame spread rates could generally be achieved even under conditions where steady spread was not expected based on Section 2.2.1. In addition, the flame length and shape did not change as a function of time. This result was found to hold under most conditions tested including pg, 1g, concurrent-flow, opposed-flow, radiatively-driven, and convectively-driven conditions. In addition, data was collected using video, thermocouples, and an interferometer. This data was analyzed using the techniques described below to determine flame length, flame spread rate, flame shape, temperature, convection-diffusion zone thickness, and type of flame. 5.1 Tests of Hypothesis 5.1.1 Steady Upward Spread It was predicted (Fernandez-Pello, 1983) that, in upward spreading flames, the flame length continually increases and thus the flame spread rate is unsteady. However, due to the limiting mechanisms proposed in Section 2 .2 , it was hypothesized that concurrent-flow flame spread could be steady. Experimentally, it has been observed that under most conditions, concurrent- flow flame spread is steady as shown in Figure 19. With the concurrent-flow apparatus, experiments were conducted under a variety of conditions. Larger 66 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. width fuel samples generally needed more length to achieve steady state. This required a taller sample and thus a taller chamber. Smaller width fuel samples required higher pressures, higher oxygen concentrations, and / or an ignition ramp (where the sample was ignited at a larger width and the width was tapered down to the correct width over some distance) to avoid quenching. Steady flame spread was obtainable with the concurrent-flow apparatus over a Grashof number range of 100 to 10 billion. Therefore, it is presumed that this setup was adequate. Position vs Time Plots for Various Mixtures 20 O 30% 02/N2 -10 mm, 1atm □ 30% 02/N2 -10 mm, 0.5 atm A 30% 02/C02 - 20 mm, 2 atm X A ir-60 mm, 1 atm 18 16 14 12 10 8 6 4 2 0 0 2 2.5 0.5 1 1.5 Time (sec) Figure 19. Flame position as a function of time for concurrent-flow flame spread. Gaseous atmosphere is varied and measured position and times are taken toward the top of the fuel samples. Steady flame spread can be seen by drawing a straight line though the data points. The flame spread rate is the slope of this line. 67 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.1.2 Steady Downward Spread Over Thin Fuels After the drop test began, during the initial transition phase, for a period of typically 0.5 seconds, the flame shape and spread rate changed as the mechanism changes from a buoyancy induced opposed-flow flame spread to a flame speed limited opposed-flow flame spread. After the transition phase, the flame position advances linearly with time indicating a steady flame spread rate as shown in Figure 20. At the 2 second mark, the drop ends. Most flames tested followed this pattern closely, however, for very fast and very slow flames, steadiness was uncertain. For very fast spreading flames, the sample size was to small to quantify a steady flame spread rate. Although the flame spread rate could be measured, less than 0.5 seconds would pass before the flame engulfed the entire sample. These tests need a larger sample and chamber, but limitations on size and weight in the drop towers prohibited this. For very slow spreading flames, the instrumentation and drop tower size became a problem. The video camera data became pixilated for very small spread rates and a flame did not appear to move over several frames. A possible solution was to zoom in on a smaller portion of the sample. This created another set of problems due to the limited size of the rig. If more space was allowed for the larger necessary lens, this would be a solution. Moving the camera closer also presented a solution. In some cases, this was done with limited success but the window or lens would become damaged. Another solution was the use of a larger tower like the zero-g facility, which 68 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. has 5 seconds of free fall compared to 2 at the smaller tower. This solution seemed to offer a better estimate of steadiness with the same instrumentation. However, due to the slow turnaround time (about 2 days per drop) this resource was only used for the lowest spread rate conditions. Flame spread rates were found to be the same in both drop towers except at the lowest 0 2 concentrations (where steady spread was not reached in the 2.2 second tower). Only steady flame spread data were reported in this work. Thin Fuel Flame Position in 34% 0 2 - C02,0.5atm 0 0.5 1 1.5 2 Time (sec) Figure 20. Flame position as a function of time for a thin fuel. Rig starts dropping at time = 0 seconds. For most conditions, data points after initial 0.5 second transition line up on a straight line showing a steady flame spread rate. To verify that the thin fuel used could be characterized as thermally- thin, that is no heat transfer occurs along the fuel bed and heat transferred to the fuel bed was used to vaporize the fuel, a set of tests was conducted at 1g and pg for one specific atmosphere (33% 0 2 in N2 ) with varying numbers of 69 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. fuel sheets. The results in Figure 21 show that, at least for this mixture, Sf is inversely proportional to the number of sheets (and thus psxs ) as is required for thermally-thin fuels according to Equation 6. Therefore, it was assumed that the results obtained in Section 5.3.1 and 5.3.2 can be characterized as applying to thin fuels with a substantial margin of thickness. o c n a a > ” 0 O j 5 - ! C u g o Slope = -1 jug 2 1 0.8 0.6 0.4 2 3 4 1 Number o f sheets Figure 21. Effect of number of fuel sheets on spread rate. Atmosphere: 33% O2 in N2 at pg. 5.2 Data Analysis 5.2.1 Video Video cameras were used to record the flame length, shape, color, and spread rate. The flame shape and color could be directly obtained from video images such as Figures 11 and 12, which show a flame spreading at 1g 70 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and jjg, respectively. This difference shows that the mechanisms that control the flame were not the same. The flame length could be measured from the video by comparing the length of the flame in pixels and comparing it to the number of pixels that each 1 cm marking on the sample holder corresponds (typically around 13). The number of pixels in 1 cm varied from run to run because the distance to the camera changed slightly so the pixels per cm must be recounted every time. The flame length was compared to the flame length of the next video frame and so on. For upward, concurrent-flow flame spread the flame length also varied with time {i.e., the flame flickered) so the longest repeatable flame length was used. This means that a flame length that oscillated between 5 cm and 6 cm continually over time (5.0 cm, 5.5 cm, 6.0 cm, 5.5 cm, 5.0 cm, 5.5 cm, 6.0 cm, 5.5 cm, and so on) will be given the length of 6.0 cm. However, a flame with a increasing flame length (5.0 cm, 5.5 cm, 6.0 cm, 5.5 cm, 6.5 cm, 6.0 cm, 6.5 cm, 7.0 cm, and so on) would not be given a flame length because a steady maximum could not be found. It should be noted that (within the experimental uncertainty) some flames with oscillating flame lengths appeared to have a steady flame spread rate when measured using the trailing edge (bottom) of the flame. These rates generally agreed with those obtained by averaging flame length and position of the whole flame, therefore, they were considered steady as long as the spread rates and lengths were not increasing. 71 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The video data could also be used to compute the flame spread rate. This was done by finding the position of the flame front as a function of time and taking the slope of this plot. For concurrent-flow flame spread, the trailing edge of the flame (seen as the point where the image changes from a bright pixel to a dark pixel) was used to calculate the flame spread rate. When a thin fuel was used, this also corresponded to the point where the paper had burned out. For opposed-flow flame spread, the leading edge of the flame (seen as the point where the image changes from a dark pixel to a bright pixel) was used to calculate the flame spread rate. Once the position vs time was established for either of these methods, a plot was made similar to Figure 20 (for thin fuels in pg at the 2.2 second drop tower). The slope of a least squares fit was used to determine the spread rate. The side camera was useful for determining whether or not a fuel was burning on both sides. If the interferometer was not used, the camera could easily be placed on the side window to observe flame spread. Major modifications (such as adding additional windows) could be made to the chamber that could make side cameras suitable for quantifying flame spread but were not necessary at this time with all of the other methods available. Although a side view of the flame was desirable, interferometer images captured the side view of the flame at a near parallel angle and also carried more information about the convection-diffusion zone thickness (not always visible to regular video). 72 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2.2 Interferometry A shearing interferometer (Section 4.2.1.3) was used to obtain information about the change in density of the gasses from the side view of the spreading flames. In this case, the MIDAS Rig used an interferometer to quantify the flame spread rate, flame length, and convection-diffusion zone thickness even when the flame was not visible in video. An example of an interferometer images in pg and in 1g are shown in Figures 14 and 15, respectively. Notice the larger flame convective-diffusive zone thickness in pg, which was predicted in Section 2.1.2. To determine the flame spread rate, two images like that of Figures 14 and 15 were used. An interferometer image with no flame was just horizontal fringes. The more the fringes are bent, the more severe the density difference. Therefore, the flame position was the point where most of the fringes have amassed. Notice that in Figure 14, the density change was so severe that the entire image seemed to be one black blob. However, when compared to the pg image in Figure 15, the flame location could be clearly identified where all the fringes have coalesced into more or less one line. This is the flame location. The flame location at x = 0.5 cm could be marked y = 0 cm. A few frames later the flame would be at location x = 0.5 cm and y = -0.1 cm. The flame had moved 0.1 cm in 3 frames would have a flame spread rate of 1 cm/sec. If the flame shape was steady, the same would be true when marking the flame at the center position (x = 0.0 cm) or farther out (x = 1.0 73 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. cm). Experiments show that interferometer data could be compared directly to video data by using the width of a 1 cm square piece of paper inserted into the expanded laser beam path as a length scale. When viewed on a monitor this 1 cm square is 64 pixels long. Using the same spread rate technique as regular video gave interferometer spread rates that were at most 5% different than regular video (regular video measures 1 cm per second while interferometer video may measure between 0.95 and 1.05 cm per second) and was within experimental uncertainty. The flame length was usually too tall to fit into an interferometer video image and was generally not measured with this method. 5.2.3 Thermocouples Temperature measurements were taken using thermocouples. The data can be plotted on a temperature versus time graph as in Figure 18 to show the flame spread rate, flame length, and temperature. To determine the flame spread rate, the trailing edge of the flame of was used. The time was taken as the trailing edge of the flame dropped past 900 °C for thermocouples 1 and 2 (thermocouple 2 is 6.8 cm above thermocouple 1 for concurrent-flow applications and 0.2 cm above thermocouple 1 for opposed-flow applications). The spread rate was calculated as the distance between the thermocouples divided by the time lapse (At). L was defined as Sf (At), where At is the time lapse between 74 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. leading and trailing edge passage at 900 °C, because flame length and position agreed well with the visible length and position at that temperature. The thermocouple closest to the surface (2 mm) was used because for small vg, 8 was very small. Consequently, more remote thermocouples exhibited no significant temperature rise. Note that temperature histories at two vertical locations (y=0, y=68) were very similar, indicating steady spread. For most experiments, the difference between regular video spread rates and thermocouple derived spread rates was less than 2%. To determine the temperature profile (shown in Figure 22), each thermocouple recorded the time it passed 50 °C, 200 °C, 400 °C, 600 °C, etc., as the temperature increased and again as the temperature decreased. A distance from a predetermined zero point was found by multiplying the time by the flame spread rate (the procedure shown above). Each temperature, 200 °C for example, would then have two vertical distances and one horizontal distance (the thermocouple position). The two profiles show a clear difference in the flame shape and temperature profile of concurrent-flow (where convective and diffusive transport are in the same direction) and opposed-flow flames (where convective and diffusive transport are in opposite directions). This is probably because for concurrent-flow flame spread, the flame length grew (in this case) until the radiative stabilization limited it (as discussed in Section 2.2.1.1). Meanwhile for the opposed-flow case, flame spread is limited by the ability to transfer heat to the fuel bed ahead of the flame. 75 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Upward Sf Temperature Profile Downward Sf Temperature Profile 25.0 25.0 50C 200C 400C 600C 800C 1200C 50C 200C 400C 600C 800C 1200C 20.0 20.0 — 15.0 f I § i 10.0 b f . 15.0 ra 0 '• £ 0 ) 1 I 10.0 m Q 5.0 5.0 0.0 0.0 i 0.5 1.0 Distance (horizontal) (cm) 0.0 1.5 1.5 0.0 0.5 1.0 Distance (horizontal) (cm) Figure 22. Temperature profile comparison of upward and downward spreading flames in air at 1 atm. Downward flames are much smaller and the temperature changes more rapidly. 76 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.3 Spread Rates 5.3.1 Opposed-Flow Flame Spread Actual flame spread data is presented in Tables in Appendix C.1. Results shown in the following sections have been grouped together where relevant and presented in graphical form in the figures that follow. 5.3.1.1 Radiation Effects - Thin Fuels In opposed-flow flame spread, one of the goals of this study was determine the importance of gas phase radiation effects by finding the flame spread rates and comparing them to theory. Testing was conducted in argon, helium, nitrogen, carbon dioxide, and sulfur hexafluoride diluents at 1g and pg. Figures 23 through 27 show the opposed-flow flame spread rate over a thin solid fuel at 1g and pg as a function of oxygen concentration using these diluents. Figures 23, 24, and 25 show that for helium, argon, and nitrogen (diluents with small radiation absorption and re-emittance) the 1g flame spread rate was always higher than the pg flame spread rate of the same oxygen concentration. This result agrees with the results of Olson etal. (1988) for N2. It was also found that the minimum oxygen concentrations that would support flame spread were lower at 1g than at pg for these diluents, again in agreement with prior studies. Thus, the results for helium, nitrogen and argon diluents are entirely consistent with prior work. The reason for the higher spread rates and wider flammable range at 1g was likely due to radiative heat 77 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. losses as discussed in Section 2.1.2. Figures 23 through 27 also show that the 3 methods of determining the flame spread rate (e.g., measuring the pyrolysis front, flame front, or interferometer image) were consistent with each other. Figure 26 shows that (in contrast to helium, argon, and nitrogen diluents) for carbon dioxide diluent there was very little difference between the values of Sf at 1g and pg for higher O2 concentrations and at lower O2 concentrations Sf was slightly higher at pg than 1g. The minimum 0 2 concentration was also slightly lower at pg (21%) than at 1g (24%). Finally, Figure 27 shows that for SF6 diluent, Sf was substantially higher at pg than at 1g for all 0 2 concentrations and the minimum 0 2 concentration was significantly lower at pg (31%) than at 1g (38%). For the 02 -S F 6 and 0 2- CO2 atmospheres with low O2 concentration, the spread rate required more than 2 seconds to reach steady-state, thus in this case 5-second drop facility test results were reported instead. The pg data at these low O2 concentrations showed a very gradual increase in Sf over time in the 2.2 second drop tower, whereas the Sf in the 5 second drop tower reached a steady Sf. Thus the data from the 5-second drop tower should be considered more reliable and was chosen over the 2.2-second drop tower data at low 0 2 concentrations. No similar behavior was observed for the other diluents because for SF6 mixtures at low O2 concentrations it was necessary to ignite and establish the flames at pg since they could not burn at 1g. In other cases, only the transition from 1g to pg spread needed to be accomplished during the drop test. 78 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Flame Spread Rate with Nitrogen Diluent 8 E 3 c o 3.5 3 2.5 2 1.5 1 0.5 0 ♦ ig Sf □ pg pyrolysis front A pg Sf o pg interferometer A A 15 20 25 30 35 0 2 Concentration (%) A 40 45 Figure 23. Flame spread rate vs oxygen concentration for thin fuels with nitrogen diluent at 1 atm. Flame Spread Rate with Helium Diluent o 0 ) C O ♦ 1g Sf □ pg pyrolysis front a pg Sf O pg interferometer O Concentration (%) 2 Figure 24. Flame spread rate vs oxygen concentration for thin fuels with helium diluent at 1 atm. 79 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 3.5 3 I f 2.5 1 2 1.5 1 0.5 0 10 Flame Spread Rate with Argon Diluent ♦ ig Sf □ M 9 Sf A pg pyrolysis front O pg interferometer *♦ * n □ D ♦ □ □ □ 15 20 0 2 Concentration (%) 25 □ 30 Figure 25. Flame spread rate vs oxygen concentration for thin fuels with argon diluent at 1 atm. Flame Spread Rate with Carbon Dioxide Diluent 2.5 3 1-5 u > ~ 1 0.5 0 ............i ' ..... ..... i ..........i............ i ♦ _ • © A 8 □ __ ♦ □ ♦ ig Sf § ♦ i — i □ pg pyrolysis front □ B # a pg Sf ♦ O pg interferometer a ® i i E E pg pyrolysis front 5 sec 20 25 30 35 40 O Concetration (%) 45 50 Figure 26. Flame spread rate vs oxygen concentration for thin fuels with carbon dioxide diluent at 1 atm. 80 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Flame Spread Rate with Sulfur Hexafluoride Diluent 2.5 $ 15 <fT 1 0.5 0 25 - ... . .I. " ... 1 ............... ♦ ig Sf □ pg pyrolysis front A pg Sf O pg interferometer EB pg pyrolysis front 5 sec v pg Sf 5 sec i i i O A O A < # ■ E B m E B m B B EEEB ^ ♦ i i 8 0 ♦ A ♦ ♦ 1 .....-..- I I ........... 30 35 40 45 C > 2 Concentration (%) 50 55 Figure 27. Flame spread rate vs oxygen concentration for thin fuels with sulfur hexafluoride diluent at 1 atm. These results give more insight into the effects of radiative transport on flame spread at pg. It should be noted here that the predictions in Section 2.1.1 were independent of U (which is affected by the g-level) while it is evident that U (g-level) affects the flame spread rate significantly. A mechanism for the observed diluent effects is suggested by the interferometer images of flames in the O2 - SF6 atmospheres. Figure 14 shows that at 1g, the thickness of the flame (convection-diffusion zone) in the 42% O2 - 58% SF6 atmosphere was very small (a few mm or less), in fact it was so thin that the density gradients were too large to be imaged properly 81 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. with our interferometer setup, whereas when this flame experiences pg conditions, the convection-diffusion zone thickness grew rapidly (within 1 sec) to several cm as shown in Figure 15. The increase in flame thickness (convection-diffusion zone) at pg was expected for the reasons mentioned in the Section 2.1.2, however, the increase for the O2 - SF6 atmosphere (compare Figures 14 and 15) is much greater than that for the O2 - N2 atmosphere (compare Figures 28 and 29). This difference cannot be explained based on the estimated 5 ~ a / U. This is because the ratio of the pg flame thickness (~ ag/ Sf) to the 1g flame thickness (~ [ ag 2/ g ]1 /3 ) should be smaller for the SF6 case since Sf is similar for the two cases but a is much smaller for the O2 - SF6 mixture. For the specific cases shown in Figures 32 and 33, representative values of 5 at pg were (1.4 cm2 / sec) / (1.7 cm / sec) ~ 0.8 cm for 0 2 - N2 and (0.22 cm2 / sec) / (1.3 cm / sec) ~ 0.17 cm for 0 2 - SF6 (using 1000K as a representative average temperature at which to evaluate a). For the O2 - N2 atmosphere, this estimate was comparable to the thermal thickness seen in the interferogram, however, for the O2 - SF6 atmosphere, 5 is underestimated by at least an order of magnitude. Therefore, an additional heat transport mechanism was probably active for the SF6 case that is unimportant for the N2 case. 82 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I Figure 28 Interferometer image of spreading flames at 1g. Atmosphere: 30% 0 2 in N2 at 1 atm. Field of view is 4.2 cm x 3.2 cm. Figure 29. Interferometer image of spreading flames at pg. Atmosphere: 30% 0 2 in N2 at 1 atm. Field of view is 4.2 cm x 3.2 cm. 83 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The most plausible source of an additional transport mechanism is absorption and re-emission of thermal radiation by the gases discussed in Section 2.1.2. Helium, argon and nitrogen do not emit thermal radiation and thus the only gaseous species producing significant radiation are the H20 and C 0 2 combustion products. At the conditions tested in this study, ap for the combustion products in these diluents are typically 1 m'1 and thus Lp is much larger than the characteristic size of the flame. Consequently, for these diluents the radiation can be considered optically thin (no reabsorption of emitted radiation) and therefore practically all emitted radiation is lost to the walls of the combustion vessel. However, for C 02 and especially SF6 diluents, Lp is much smaller. For example, at 300K and 1 atm, Lpfor a 42% 0 2 - 58% SF6 mixture is about 0.4 cm, and at a representative mean gas temperature of 1000K, Lp is about 6 cm. These values span the apparent thermal thickness seen at pg in Figure 15 but are much larger than that seen at 1g in Figure 14, which could explain why reabsorption would have a significant effect on Sf at pg but not 1g. The flame front view camera images (not shown) indicate that the visible flame front lies slightly inside (toward the fuel bed) from the region of high temperature gradient (closely spaced fringes) seen in Figure 14, thus the flame is characterized by a rapid temperature rise from the ambient atmosphere to the flame front and a much smaller gradient from the flame front to the fuel bed. This would be expected because it has been shown 84 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Lozinski et ai, 1994) that Lp is a very rapidly increasing function of temperature for SF6. Under conditions where reabsorption is significant, some of the radiation emitted near the zones of peak temperature would not be lost to the surroundings. This would not only decrease the net heat loss, which in itself would lead to an increase in Sf, but would augment conventional thermal conduction to the fuel bed and increase Sf above that expected in an adiabatic system with no radiative transfer. This effect had been demonstrated (Abbud- Madrid & Ronney, 1993) for the burning velocity of premixed flames at pg by seeding the gases with inert, radiating particles. For CO2 and SF6 diluents, the minimum O2 concentrations were lower at pg than at 1g for downward flame spread. Therefore, a set of tests on the 1g upward limits in these diluents was conducted. The upward limits were found to be 18.5% O2 (compared to 21% O2 in pg) in C 0 2 and 31.5% O2 (compared to 29% O2 in pg) in SF6. At these limits the flames propagated only upward with no lateral spread. Thus, for SF6 but not C 02 the pg environment is more hazardous than any 1g environment. In Section 2.1.2, Equation 7 proposed that Sf was always higher with a lower U when radiation was present. As a test of this prediction, experiments conducted in 40% oxygen with SF6 as the diluent at 1 atmosphere with an imposed flow using a low-speed pg flow tunnel (explained in further detail in Appendix A) show that at pg when the net opposed-flow velocity (= U + Sf) is 85 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. zero or moderate (not more than 3Sf), Sf is practically constant and higher than the 1g value as shown in Figure 30. These findings contrast behavior observed experimentally (Olson, 1991) in radiatively-inactive 0 2- N2 atmospheres, where any increase in U increases Sf until blowout occurs. Moreover, at 1g an increase in U actually decreases Sf. Spread rate vs Imposed Flow Velocity 8 C O E o, £ C O c r T J C O 2 Q. £ 0 C D E C O 2 1 0 10 8 6 0 2 4 -2 Opposed-Flow Velocity (cm/sec) Figure 30. Flame spread rate vs. Forced flow velocity (40% 0 2 -SF6, 1atm). The results above showed that the diluent (radiative effects) and U had a large impact on the flame spread rate as predicted in Section 2.1.2. As part of those radiative effects, Equation 7 also predicted that pressure could play a 86 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. role in S f.rad through its contribution to Lp which affects ap and thus A. As a test of the radiation hypothesis, a set of experiments was performed for each diluent except Ar at varying pressure (P) for one value of the 0 2 mole fraction, which is well above both the 1g and pg flammability limits for that diluent. As P increases, LP decreases and thus if the radiation reabsorption hypothesis were correct, the effects of diluent type seen at 1 atm should be stronger (weaker) at higher (lower) P. Results are shown in Figure 31. At 1g, for all diluents Sf is nearly constant or increases only modestly with P (by a maximum of 40% for He as P increases 8-fold), which is consistent with prior data (Zhang et ai, 1992). These results contradict the prediction in Equation 3 (deRis, 1969) without radiation effects where Sf for thin fuels is independent of P. At pg, for helium and nitrogen diluents, Sf increases with P, which is consistent with prior pg experiments (Bhattacharjee & Altenkirch, 1993) in optically-thin 0 2 - N2 atmospheres and is expected because the impact of optically-thin radiative losses is proportional to 4 c v aP T3o c g 2/ A ,g U2 ~ P'1 since ap ~ P and ag ~ P'1. Note that this impact was derived taking the ratio of residence time to radiation time. The residence time within the flame front is 8/U = a g/ U 2 and the radiative time scale is Tf / ( dT / d t) = Tf / ( Q l / p g Cp ,g ). Here QL is the radiant heat loss per unit volume = A - 4 <raP (Tf4 - T*,4 ) ~ 4 a aPT 4 (since T 4 » T*4 ). Therefore, the ratio = (ag / U2 ) / (pgC PiST f/ Ql) = 4 a aP Tf3 o c g 2 / Xg U2. Thus, for helium and nitrogen diluents the spread rates at 1g and pg tend to converge as P increases. For SF6 diluent, completely 87 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. different behavior is observed, suggesting a different mechanism is operative. As P increases, the pg spread rate diverges from the 1g value, which would be consistent with an increasing effect of radiation, leading to a reduction in the radiant heat loss and an augmentation of heat transport by radiation. Of course, if P were increased sufficiently, the effect may saturate since reabsorption would increase and little radiation would then escape. Flame Spread Rate as a Function of Pressure 6 27%02 w/N2 1 g 28%02 w/C02 1 g 27%02 w/ N2 pg ■ - 28%02 w/C02 pg - a - 30%O2 w/He 1 g - 50%O2 w/SF6 1 g 30%O2 w/He pg - - O - 50%O2 w/SF6 pg 5 4 3 O o - 2 1 0 2.0 1.0 1.5 0.0 0.5 Pressure Figure 31. Flame spread rate as a function of pressure for helium, nitrogen, sulfur hexafluoride, and carbon dioxide diluents at fixed oxygen mole fractions. 88 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Finally, we note that Zhang et at. (1992) reported cellular flame structures for downward-spreading flames at 1g in O2 - CO2 and O2 - SF6 atmospheres near the limiting O2 concentrations. As was later verified (Chen et a/., 1992), this was attributed to a mechanism similar to the diffusive-thermal instability of premixed flames. The same phenomenon was observed at 1g in this study. However, only a slight tendency for cellular structure was observed at pg, even for near-limit atmospheres, and the observed cell size was much larger at pg than at 1g (typically 2 cm vs. 0.3 cm). This is probably a result of the much greater flame thickness at pg than at 1g, which would have led to a much larger characteristic cell dimension at pg. 5.3.1.3 Partially Premixed Fuel The preceding results were obtained for opposed-flow flame spread without added gaseous fuels. In this work, it was found that the effect of adding gaseous fuel to the ambient atmosphere was qualitatively similar at 1g and pg in that Sf could be increased significantly by adding gaseous fuel as shown in Figures 32 through 35. For all cases tested, the effect of added gaseous fuel was significantly stronger at pg than 1g, as was hypothesized in Section 2.1.3. For example, at 18% 0 2 with CO fuel, Sf increased by a factor of about 2 at 1g as ( j) increased from 0 to 0.37, whereas at pg the increase was a factor of about 5. Also, at both 1g and pg, CO had a greater impact on Sf than CH4. This finding was consistent with the predictions of the model 89 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. described in Section 2.1.3, and was the result of the higher reaction rates (thus higher q” ) associated with CO - O2 chemistry as compared to CH4- O2 chemistry. Furthermore, for all but the case having the lowest values of Sf (18% O2 with CH4 fuel) case, Sf is actually higher at pg than 1g for large The experimental results shown in Figures 32 through 35 can be quantitatively compared with the simple theoretical model employed by Ronney et al. (1995) by substituting Sf itself for the opposed-flow velocity (U) at pg (in the prior 1g study U = 1.5 ( g og) 1 /3 was employed). Since there was no precise demarcation between non-merged flame and merged flame conditions, both solutions are shown in Figures 32 through 35. Note that in all cases, with gaseous fuel the predicted Sf was larger at pg than at 1g for non merged flame conditions and that for merged-flame conditions there was little effect of gaseous fuel on Sf. In fact, due to the specific heat of CH4, the predicted merged flame Sf actually went down with added CH4. Both of these features were consistent with the discussion in Section 2.1.3. 90 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Flame Spread Rate with CO in 18% Oxygen I Nitrogen 3 E ID Sf 1g non-merged model Sf |jg non-merged model - - - - Sf merged model ▼ ig sf □ M 9 pyrolysis front o pgsf A pg interferometer / / 0 0.1 0.2 0.3 0.4 Equivalence ratio of premixed fuel Figure 32. Measured and predicted flame spread rates vs. equivalence ratio in 18% O2 with N2 diluent at 1 atm with CO as added gaseous fuel. Flame Spread Rate with CH4 in 18% Oxygen / Nitrogen 4 i 11 1 1 1 r 1 ' ' 1 1" ' ' 1 ''— r~i—1 — 1 —! —| —1 —i —1 —1 —| 1 i 1 r 3 C O E o * * ■ (D □ O A Sf 1g non-merged model Sf gg non-merged model Sf merged model ig Sf Mg pyrolysis front pg Sf Mg interferometer 0 0.1 0.2 0.3 0.4 0.5 0.6 Equivalence ratio of premixed fuel Figure 33. Measured and predicted flame spread rates vs. equivalence ratio in 18% 0 2 with N2 diluent at 1 atm with CH4 as added gaseous fuel. 91 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Flame Spread Rate with CO in 30% Oxygen I Nitrogen 5 o _ _1 _ _! _ _ 1 ...1 Sf 1g non-merged model Sf pg non-merged model Sf merged model ▼ ig Sf □ pg pyrolysis front O pgS f A pg interferometer i — |-----1 -----1 -----r n 1 -----! -----r / A o _ ! i I L I I I 0.05 0.1 0.15 0.2 Equivalence ratio of premixed fuel 0.25 Figure 34. Measured and predicted flame spread rates vs. equivalence ratio in 30% O2 with N2 diluent at 1 atm with CO as added gaseous fuel. Flame Spread Rate with CH4 in 30% Oxygen / Nitrogen 5 4 'o' 0 > uT 2 1 0 0.1 0.2 0.3 0.4 Equivalence ratio of premixed fuel Figure 35. Measured and predicted flame spread rates vs. equivalence ratio in 30% O2 with N2 diluent at 1 atm with CH4 as added gaseous fuel. 92 “ I 1 j 1 i 1 i ------------ O A / Q / - 2 : o j r * □ 1------1 ------1 ------ 1 ------ 1 ------ 1 ------ 1 ------ r _____ Sf 1g non-merged model Sf pg non-merged model m a m g a m Sf merged model ▼ ig Sf □ pg pyrolysis front O pg Sf A pg interferometer Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figures 32 through 35 show that the model uniformly under-predicted the increase Sfat pg as < { > was increased. This can probably be attributed to radiative loss effects, which were not considered in the partially-premixed model. There was no corresponding change in the impact of radiative losses at 1 g since U ~ (gog)1 /3 which was much greater than Sf and was essentially constant. In fact, radiative losses were probably unimportant for all 1 g flames studied in this section but will play still play a role in pg. As a result, Sf can be expected to increase significantly more at pg than at 1g as < j > increases. As evidence of this hypothesis, note that the agreement between model and experiment improves as < | > increases. While the predictions and experiments were only fair agreement in terms of the actual values of Sf, the agreement was good in terms of the ratios of Sf at 1g and pg at large < ) > , as summarized in Table 4. For example, for 30% 0 2 in N2 with added CO fuel, at < j > = 0.2 the experimental spread rates are 3.00 and 3.75 at 1g and pg, respectively, whereas the corresponding theoretical predictions are 2.45 and 3.02; the ratio of Sf at pg to 1g is 1.25 for the experiments and 1.27 for the predictions. The agreement is not nearly as satisfactory where two factors not considered by the model (namely radiative loss and finite-rate chemistry of the non-premixed flame) are strongest (18% 0 2 with CH4 fuel). Radiative losses are strongest for this case since U = Sf is lowest. Finite rate chemistry is most important for this case because the 0 2 mole fraction, thus Tf, is lowest. 93 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Mole % O2 Gaseous fuel < t > s^gySfCig), Experiment Sf(pg)/Sf(lg), Theory 30 CO 0 . 2 0 1.25 1.23 30 CH4 0.35 1.33 1.27 18 CO 0.37 1.31 1.52 18 CH4 0.57 1 . 0 0 1.36 Table 4. Comparisons of measured and predicted effects of added gaseous fuel on flame spread rates at 1 g and pg. 5.3.2 Concurrent-Flow Flame Spread The preceding results were obtained for opposed-flow flame spread conditions. This section deals only with results obtained for concurrent-flow (upward) flame spread at 1g. All results were obtained for buoyancy induced flow (U) and will be compared to buoyant convection predictions directly (forced flow comparisons will not be made in this study). Figure 36 shows the effect of width (W) on Sf,C O n for ambient air. At low Grw, SfiC O n ~ W2 83, thus Sf.con/ Sfi0 P p ~ Grw0'94, close to the CL prediction shown in Table 3, Sf,conI S f,o p p ~ Grw • At Grw ■ > 30,000, S ^ co n ~ , thus S f^ c o n ! S^opp” " Grw , close to RL or RT predictions since SfiC O n / Sfi0 p p ~ Grw 1 Plw3 ~ W 3 W 3 ~ W°. The observed transition Grw is close to the CL-RL prediction Grw“ 30,000 as shown in Figure 6 . This should be followed by RL-RT transition at Grw * 90,000, but this cannot be discerned because the Grw range corresponding to RL behavior was narrow. Furthermore, there was little difference between RL and RT predictions for Sf since the constants D and d are only slightly different for laminar vs. turbulent flow. 94 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100 i n n i ’ "e r... ° Gr < 20,000 a B I O Best fit: Best fit: 0.58 S = 0.99 W f 3.30 10 1 102 Fuel bed width (mm) Figure 36. Effect of fuel bed width (W) on upward flame spread rate (Sf,C O n) for thin fuel beds burning in ambient air. Predicted results are Sf ~ W 3 for low Grw and Sf ~ W° for high Grw, with a transition Grw of 30,000. Figure 37 shows the correlation between SfiC o n / Sfi0 p p and Grw for all data. At low Grw , the proposed relation Sf,C O n / Sf,o p p ~ Grw 1 fits each data set for a given atmosphere well, although between different atmospheres a factor of 2.5 variation in SfiC O n / Sf,o p p was found at constant Grw . Nevertheless, the comparison was considered quite reasonable considering the wide range of experimental conditions tested. We believe much of the scatter results from varying degrees of dissociation for various atmospheres. At higher Grw , all Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. data bend towards horizontal, indicating Sf,C O n / Sfi0p p ~ Grw °, consistent with radiative stabilization in Table 3. The transition Grw varied from about 5,000 for the highest vg atmosphere tested (30% O2 - He, 0.25 atm) to 200,000 for the lowest vg tested (46% O2 - SF6, 3 atm). These transitions were in very good agreement with predictions shown in Figure 6 . SfiC o n / Sf,o p p predictions were in very good agreement with experiments for high and intermediate vg, though predictions were a little high for the lowest vg (3 atm O2 - SF6 predictions are slightly off the graph). Figure 37 shows the utility of the proposed scalings; wide ranges of Sf and Grw for varying Plw are correlated on one plot. Effects of Lewis number and other mixture properties were covered by referencing SfiC o n to Sf,o p p . Figure 22 shows a typical temperature profile for a spreading flame. It is based on temperature measurements at various times divided by the flame spread rate. Figure 38 shows the correlation of L / W with Grw- At low Grw, most data for a given atmosphere follow the predicted L / W ~ Grw 1 for CL spread shown in Table 2, although there was some scatter between different atmospheres. For large widths (W), L/W~ Grw *1/3 as required for width- independent L. 96 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Non-Dimensional Flame Spread vs Grashof Number 10 Q. Q. O C 8 (fT 1 01 102 103 104 105 106 107 108 109 10 Grashof Number (Gr ) Figure 37. Correlation of steady values of S f,con/S f,o p p with Grw for all experimental data. "x2 " indicates double-thickness fuel samples. " t t T tttttj Prediction (RT, Air) Prediction (CL, all) i u O o I O O n° - 0* 3 o ©is ■ o ▼ Q G □ / % < - . r A I O □ o A A/ A / + □ / / Prediction (RL, 30% O ^He, 0.25 atm) O 21% 02/N 2, 1 atm □ 30% 02/N 2 ,1 atm O 30% 02/N 2, 0.5 atm X 30% 02/N 2, 0.25 atm + 40% 02/N 2, 0.25 atm A 50% 02/N 2, 0.25 atm 50% 02/N 2, 0.25 atm (x2) 30% 02/N 2, 2 atm 35% 0 2 /C 0 2 , 1 atm 30% 0 2 /C 0 2 , 2 atm 50% 02/S F6, 1 atm O 45% 02/S F6, 2 atm S 46% 02/S F6, 3 atm ffl 25% 02/H e, 1 atm □ 30% 0 2/H e , 0.25 atm Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Flame Length over Width as a Function of Gr 1 0 0 o 21% 02/N 2, 1 atm ♦ 35% 0 2 /C 0 2 , 1 atm - □ 30% 02/N 2 ,1 atm A 30% 0 2 /C 0 2 , 2 atm - o 30% 02/N 2, 0.5 atm T 50% 02/S F6, 1 atm - X 30% 02/N 2, 0.25 atm O 45% 02/S F6, 2 atm + 40% 02/N 2, 0.25 atm H 46% 02/S F6, 3 atm A 50% 02/N 2, 0.25 atm E B 25% 02/H e, 1 atm • 50% 02/N 2, 0.25 atm (x2) □ 30% 0 2/H e , 0.25 atm ■ 30% 02/N 2, 2 atm 10 ■ ' ° Prediction ! o ^ o (CL, all) / » x o f ^ ■ X / * / # 0 £ ID E S ® □ i \ s o ( \ / \ □ / * °- o ro □ \ lA ho fflD O o O ffl mn o h i \ o 0 ▼ \ O Prediction \ (RT, air) fa o o x / - / -tp □ o i iiini i i i i - i iill - i i i i i m i . . j - i. . i i i m l i i i i i m l i i.jj.i, s o JJi] L_ w 102 103 104 105 106 107 10s 109 101 ° Grashof Number (GrJ Figure 38. Correlation of steady values of L/W with Grw for all experimental data. Legend is the same as Figure 37. 9 8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A critical aspect of our hypotheses is that Sf is determined by N ul, which in turn is related to L. From Table 2 and 3, the predicted relationships between Sf,C O n and L for buoyant flow are: Figure 39 shows the ratios of the left-hand to right-hand sides of Equations 23 and 24 with appropriate constants included as discussed in Section 2.2.1, based on measured Sf,C O n / Sf,o p p and L / W. The RT, CL, and CT relations are shown. However, the RL relation is not shown because only flame spread in a low pressure helium diluent would physically correspond to the CL regime and the results might clutter the figure. For large Grw, agreement with RT predictions was very good. In this case, for Grw> 200,000, the mean ratio was 1.63 with a standard deviation 37% of the mean. For smaller Grw, either CL or RT predictions were roughly consistent with experiments (though offset by a factor of about 3). However, only CL predictions were consistent with Sf data shown in Figure 38, as predicted by Figure 6. Figure 6 suggests that atmospheres with the smallest vg might exhibit CT behavior for marginal ranges of Grw- While no data in Figure 39 were consistent with CT predictions, spread rates with intermediate Grw ( 104 - 105) came closest, as (convective stabilization) (23) (radiative stabilization) (24) 99 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. expected based on Figure 6. Consequently, the relationships between measured L and Sf,C O n were generally consistent with our modeling hypotheses considering the transitions between regimes. o ( D J Z 10 c 0 ' ■ * - < m 1 0.1 CL X •S c o J S o 0 . 0 1 O o CL A CT X RT X V X XX , A O - A A A | 4 s . A 4 M &a a " A A A A A A A A A A A ° < * o ® o & 0 0 o ©% Oo cp, O o ° o A ‘ ’* J“ A a A O CXJ, ■ ^ 4 A a ^ A A- a A A Aa A A A A 8 9 I « il« l B 8 * ■ « » ■ ! .......ft P....% I " B ini, | lliiw l --J- m wl • a « « -■ 102 103 104 105 106 107 10s 109 101 ° Grashof Number Figure 39. Ratio of measured LAN to right-hand sides of Equations 23 and 24, showing comparison of predicted and observed correlation of L to Sf,co n for convective-laminar, convective-turbulent, and radiative-turbulent spread regimes explained in Figure 6. Dashed horizontal line indicates ideal fit of prediction to experiments. 100 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 6 CONCLUSIONS Experiments on flame spread over thin solid fuel beds in varying atmospheres were conducted at 1g and pg. The results can be divided into opposed-flow and concurrent-flow flame spread. For opposed-flow flame spread, the results were divided into two categories: radiation effects and partially-premixed fuel effects. For opposed-flow flame spread over thin fuels, experiments were conducted in a variety of 0 2-diluent atmospheres using He, N2, Ar, C 0 2 and SF6 diluents. He, N2 and Ar results were consistent with prior 1g and pg experiments in that the spread rates were higher and the minimum 0 2 concentrations were lower at 1g. In contrast, for C 0 2 diluent these properties were practically the same at 1g and pg and for SF6 the trends were completely reversed (compared with helium, argon, and nitrogen). These results are proposed to be an effect of reabsorption of thermal radiation emitted by the gases (which is important only for diluents with sufficiently large absorption coefficients) and an effect of the convection-diffusion zone thickness (which is related to U). These results are relevant to studies of fire safety in manned spacecraft, particularly the International Space Station that uses C 0 2 fire extinguishers. C 02 may not be as effective as an extinguishing agent at pg as it is at 1g in some conditions because of the differences in spread 101 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. mechanisms between the two cases. In particular, the difference between conduction-dominated heat transport to the fuel bed at 1g versus radiation- dominated heat transport at pg indicates that radiatively-inert diluents such as helium would be preferable in pg applications. For opposed-flow flame spread with added sub-flammability limit concentrations of gaseous fuels, experiments on flame spread over thin solid fuel beds in varying atmospheres were conducted at 1g and pg. It was found that when gaseous fuel (CO or CH4 ) was added to the premixed flame front, Sf increased at both 1g and pg, however, the increase was greater at pg. The observed Sf trends were close to those predicted by a simple model which considered the effect of the premixed fuel to be an increase in the heat flux to the fuel bed caused by a non-merged partially-premixed flame front upstream of the usual non-premixed flame front. These results are relevant to studies of fire safety in any enclosed environment such as submarines or manned spacecraft. In this case, partially burned combustion products may diffuse into the gaseous atmosphere and enhance spread. Since it was shown that at pg, this effect is even more important, it might be wise to purge affected compartments as soon as possible to reduce the spread of any unburned gaseous fuels to other parts of the vessel. For concurrent-flow flame spread, it was hypothesized that for narrow fuel beds, lateral heat and / or momentum losses limit flame length, and for 102 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. wide fuel beds, surface radiation losses limit flame length. These losses led to steady rather than accelerating spread for sufficiently tall beds. Spread rate predictions were developed for thermally-thin fuel beds and compared to thermally-thin fuel upward (concurrent-flow) flame spread experiments for varying width, thickness, pressure, diluent, and oxygen concentration. These data generally supported the proposed models. These results are relevant fire safety in buildings, up walls, etc. They are useful for developing improved models of concurrent-flow flame spread in more complex geometries, such as upward fire spread over walls. 103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 7 RECOMMENDATIONS FOR FUTURE WORK This section suggests future experiments that (while not vital to confirm the objectives of this study) would be complementary to this work. Although this study primarily focused on thin fuels, the same analysis can be extended to thick fuels. 7.1 Thick Fuel Models 7.1.1 Opposed-Flow Flame Spread In Sections 2.1.1 and 2.1.2, this study described the modeling technique for opposed-flow flame spread over thin fuels. This section will mirror the thin fuel analysis and briefly describe the thick fuel model. Beginning with the thin fuel results shown in Equation 3, for “ thermally thick” (effectively semi-infinite) fuels, where heat conduction through the solid fuel is important, xs is the depth of thermal penetration into the solid fuel. The heat flux within a thermally thick solid fuel can be estimated by Qs = Xs y (LfW) ((Tv- T o o ) / xs ), where the subscript y refers to the direction normal to the fuel surface. Rearranging this heat flux relation and substituting xs into Equation 3 yields an equation for the flame spread rate without the xs term for thick fuels. r _ A M Z - T J S NuL F Xg {Tf -T v) S f = N u L f K p £ v,s U - 104 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. It was shown (deRis, 1969) that an exact solution could be found with an effective Nulf = 1 and ag / I f = U leading to the relation for Sf. (26) ‘ k . p . C However, experimental results (Fernandez-Pello etai, 1981; Sibulkin, 1988) were found only to be in fair agreement. Notice that in Equation 26, Sf is proportional to U and thus indeterminate at pg. It was shown (West et al., 1996) through unsteady computations and space experiments that Sf — 1 / Vt (with no radiation), where t is the time after ignition. This means that according to this theory, steady state flame spread over thick fuels is not even possible at jug. Moreover, it states that even when flame spread occurs, its rate decreases until it extinguishes. At this point we shall add the radiation term, Qs is again estimated as qr§g W, where qr is the radiant heat flux per unit area. Sf is found by equating Qs to q as before. s / = „ r f i r _ r V (thkk ^ <27) P s C p ,A s V V T J For thick fuels, xs is the thermal penetration depth. t s = A ' s^Tv ~ (28) qr Previous theories (deRis, 1969; Tarifa & Torralbo, 1967) agree with the thick fuel results obtained above, so the present approach is considered valid. Note 105 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. that for thick fuels Sf ~ qr 2. Whereas for thin fuels Sf ~ qr1. Therefore, radiative effects should be much stronger for thick fuels. An analysis similar to that of thin fuels can be used to determine Sf for U=0. Sf = A a g 1/2 (29) a g P s C p s ( T v T j X g ( T f r v ) „ This result raises a number of interesting predictions, the most important of which are that for thick fuels without gas-phase radiation, no steady spread is possible (Sf = 0) and with gas-phase radiation, Sf ~ A1 /2 Thus, increasing gas- phase radiation should increase Sf for thick fuels. Just as for thin fuels, the gas-phase radiation is a function of the properties to the gas mixture diluent and U so changing these properties should have a large impact on Sf. The gas-phase radiation is also sensitive to pressure, an optimal pressure may exist to assist this flame spread mechanism. Equation 29 also shows that pressure (P) effects are important and could increase or decrease Sf since A ~ P and < x g ~ P'1 . Finally, Equation 29 shows that in a given atmosphere Sf can be much higher for fuels with low ps CP iS Xs. This is important when considering the short time durations of the drop towers that are used to study pg. 7.1.2 Concurrent-Flow Flame Spread For concurrent-flow flame spread, the same analysis as was used to develop the thin-fuel relations shown in Section 2.2 will be used. It was shown 106 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Fernandez-Pello, 1984) that the flame length and flame spread rate increase with time for both buoyant and forced convection and for thick fuels. Fuel Type Buoyant convection Forced convection Thermally thick Sf~ t 1 , L ~ t 2 Sf ~ t °, L ~ t 1 Table 5. Predicted increase in spread rate (Sf) and flame length (L) with time (t) for laminar concurrent-flow fires. For thick fuels, it has been shown (deRis, 1969) that SfiC o n can be estimated by substituting the solid thermal penetration depth (xp ) for xs in Equation 17. xp is estimated by equating Q to the heat flux from the fuel surface into the bed = Xs ( Tv - T«) / xp , where Xs is the solid thermal conductivity. L / I , Ty - T „ p N u l X g Tf - Ty With xp = xs, Equations 17 and 30 yield (30). / , i con = Nu\ f,opp fZ] -1 kW j Vs ) O — IT \^f,opp vopp PgC p , g \ Tf ~ T v P s^p,s\ I K ~ (31). Combining Equation 31 with NuL from Table 2 yields predictions for the concurrent-flow flame spread rate (Sf,C O n ) referenced to the opposed-flow flame spread rate (Sf,o p p ). 107 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Stabilization type / fuel type Buoyant convection Forced convection Convective / thick T 1 -6d -{1-60 _ f l - 3 c 3(1-3c ) c wrw E 1-2 b -(1-2 b) B1 A l~a R e / - 0 Radiative / thick -(i-6Q U p i 1 -\d C y 3 (1 -3 0 g r l W urw 1 -(1-2 b) -.(1-26) Bl~bPlwl~b R e^1 ^ Table 6. Predicted relations for steady values of Sf,C O n/Sf,opp for thick fuels, forced and buoyant convection, and convective and surface radiative loss stabilization. It has been shown (Loh, 1985; Zhou, 1993) that for concurrent laminar forced-flow experiments over wide, thermally-thick PMMA sheets that Sf~ U1 which is consistent with Table 6 for thick CL or RL spread since for a = b = 1/2, Sf, c o n ~ Rew° S f.o p p ~ L I . Additionally, this study showed that the thin fuel scaling model agreed reasonably with experiments. Therefore, this model is considered valid in principle. 7.2 Future Objectives According to Section 7.1.1, steady opposed-flow flow flame spread was possible when considering gas-phase radiation. With this radiation term, the effects of the gaseous atmosphere and flow environment should be more important to thick fuel flame spread than they are for thin fuel flame spread. Properties such as diluent (radiative properties), U, pressure, fuel sample thickness, density, conductivity, and specific heat were shown affect Sf. Therefore, the objective of future work should be to determine if steady 108 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. opposed-flow flame spread over solid thick fuels is possible, and if so, to determine the effect of the properties above. For concurrent-flow flame spread, the models in Section 7.1.2 suggested that the flame length would continue to grow with time as with thin fuels. However, Section 7.1.2 also showed that steady spread was possible due to the heat and momentum losses and radiative losses proposed in Section 2.2.1. Therefore, the objective of future work should be to determine if steady concurrent-flow flame spread over solid thick fuels is possible, and if so, to compare results to the scaling models in Section 7.1.2. 7.3 Preliminary Thick Fuel Results, Opposed-Flow Preliminary experiments were performed using thick fuels in opposed- flow to determine the if the models above were consistent with experimental results. It was found that the correct fuel choice was vital to producing reliable data. In addition the need for low density and conductivity for high spread rates limited the choices. Finally, a low sooting flame is desirable to decrease soot formation and increase instrumentation accuracy. A polyphenolic foam (similar to floral foam) was chosen for experiments. It is relative stiff with a low density and conductivity and does not melt. However, it still produced considerable soot. Figure 40 shows the progress of flame spread (flame position vs. time) at 1g and pg. The slope of these plots gives the spread rate; a straight line 109 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. indicates a constant spread rate and thus steady spread. From these tests, it can be seen that that in O2 - CO2 atmospheres, steady flame spread is possible over thick fuels at quiescent jjg conditions when gas-phase radiation effects are significant. Thick Fuel in 35% 0 2 / C 02 @ 4 atm 4.5 4.0 3.5 3.0 0.2.5 I 2.0 Q - 1.5 1.0 0.5 0.0 1.5 2.0 1.0 0.5 0.0 -0.5 Time (sec) Figure 40. Position of flame for a thick fuel burning at pg in 35% 0 2 /C 0 2 at 4 atm. The rig drops at time = 0 seconds. The data to the right of the 1 second mark creates a straight line indicating steady state flame spread. Figures 41 and 42 show that, like thin fuels, the convection-diffusion zone thickness can become significantly larger at microgravity than 1g. 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 41. Image of Interferometer from the side of the fuel, and the upper black region represent the thick volume of the flame in pg. la Figure 42. image of interferometer from the side of the fuel, and the upper right black region representing the convection-diffusion zone thickness at 1g is smaller than that of the pg. The scale of Figures 41 and 42 are the same. 111 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 43 shows that, as with thin fuels, the quiescent [jg Sf can be higher than its 1g (downward) counterpart for COa-diluted atmospheres in contrast to N2-diluted atmospheres have lower pg Sf. Also, like thin fuels, the radiative enhancement effect is highest at low Sf (low O2) and has less effect as Sf (U = Sf) goes up (0 2 concentration goes up). 10 ; Polyphenolic foam : 0.0267 g/cm 3 [ 4 atm total pressure ©--- pg, CO 1 g, CO 4 -J 1 TS Q. 1g, N 20 25 30 35 40 45 50 55 Mole percent O 2 Figure 43. Effect of oxygen concentration on spread rates over thick solid fuel beds at pg and 1g. These findings show that the results of the preliminary thick fuel experiments are consistent with the theory in Section 7.1.1. However, as with thin fuels, a larger range of the gas phase radiation term, A, can be determined using SF6 as the gaseous diluent. In addition, tests should be performed to assess the impact of pressure and fuel bed thickness. 112 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. REFERENCES Abbud-Madrid, A., & Ronney P. D. (1993). “Premixed Flame Propagation in an Optically Thick Gas.” AIAA Journal, 31: 2179. Altenkirch, R.A., Eichorn, R., & Shang, P. C. (1980). “Buoyancy Effects on Flames Spreading Down Thermally Thin Fuels.” Combustion and Flame, 37: 71. Bedir, H., Tien, J. S., & Lee, H. S. (1996). “Comparison of Different Radiation Treatments for One-Dimensional Diffusion Flames.” Fall Technical Meeting, The Combustion Institute, Eastern States Section. Hilton Head, S. C. Bhattacharjee, S., & Altenkirch, R. A. (1993). “ A Comparison of Theory and Experiment in Flame Spread Over Thin Condensed Fuels in Quiescent Microgravity Environment.” Twenty-Fourth Symposium (International) on Combustion, Combustion Institute, 24: 1669. Bhattacharjee, S., West, J., & Altenkirch, R. A. (1996a). Twenty-Sixth Symposium (International) on Combustion, Combustion Institute, 26: 1477. Bhattacharjee, S., West, J., & Dockter, S. (1996b). “ A Simplified Theory for deRis Flame Over Thin and Thick Fuels.” Combustion and Flame, 104: 66. Brehob, E. G., & Kulkarni, A. K. (1993). "Time-Dependent Mass Loss Rate Behavior of Wall Materials Under External Radiation." Fire and Materials, 17: 249-254. Cheesewright, R. (1968). “ Turbulent Natural Convection from a Vertical Plane Surface.” Journal of Heat Transfer, 90: 1-8. Chen, R. H., Mitchell G. B., & Ronney P. D. (1992). “Diffusive-Thermal Instability and Flame Extinction in Non-Premixed Combustion.” Twenty-Fourth Symposium (International) on Combustion, The Combustion Institute, 24: 213. 113 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Churchill, S. W., & Usagi, R. (1972). “ A General Expression for the Correlation of Rates of Transfer and Other Phenomena.” AlChE Journal, 18: 1121. Delichatsios, M. A. (1986). “Exact Solution for the Rate of Creeping Flame Spread Over Thermally Thin Materials.” Combustion Science and Technology, 44: 257-267. Delichatsios, M. A., Delichatsios, M., Chen, Y., & Hasemi, Y. (1995). “ Similarity Solutions and Applications to Turbulent Upward Flame Spread on Noncharring Materials.” Combustion and Flame, 102: 357- 370. DeRis, J. (1968). The Spread of a Diffusion Flame Over a Combustible Surface. Cambridge Massachusetts: Harvard University Press. DeRis. J. (1969). “ The Spread of a Laminar Diffusion Flame.” The Twelfth Symposium (International) on Combustion, The Combustion Institute, 23: 241-252. Ferkul, P. V., & Tien, J. S. (1994). “ A Model of Low-Speed Concurrent Flow Flame Spread Over a Thin Fuel.” Combustion Science and Technology, 99: 345-370. Fernandez-Pello, A. C. (1984). “Flame Spread Modeling.” Combustion Science and Technology, 39: 119. Fernandez-Pello, A. C., & Hirano, T. (1983). “Controlling Mechanisms of Flame Spread.” Combustion Science and Technology, 32: 1-31. Fernandez-Pello, A. C., & Mao, C. P. (1981). “ A Unified Analysis of Concurrent Modes of Flame Spread.” Combustion Science and Technology, 26: 147-156. Fernandez-Pello, A. C., Ray, S. R., & Glassman, I. (1981). “Flame Spread in an Opposed Flow: The Effect of Ambient Oxygen Concentration.” Eighteenth Symposium (International) on Combustion, The Combustion Institute, 18: 579-689. 114 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Gebhart, B., Jaluria, Y., Mahajan, R. L, & Sammakia, B. (1988). Buoyancy- Induced Flows and Transport. New York: Hemisphere Publishing Company. Grayson, G. D., Sacksteder, K. R., Ferkul, P., & T’ien, J. S. (1994). “Flame Spreading over a Thin Solid in Low-Speed Concurrent Flow- Drop Tower Experimental Results and Comparison with Theory.” Microgravity Science Tech., 7: 187-195. Greenberg, J. B., & Ronney, P. D. (1993). “ Analysis of Lewis Number Effects in Flame Spread.” International Journal of Heat Mass Transfer, 36: 315. Hamins, A., Thridandam, H., & Seshadri, K. (1985). “Structure and Extinction of a Counterflow Partially Premixed, Diffusion Flame.” Chemical Engineering Science, 40: 2027. Jiang, C. B., T’ien, J. S., & Shih H. Y. (1996). “Model Calculation of Steady Upward Flame Spread over a Thin Solid in Reduced Gravity.” Proceedings of the Combustion Institute, 26: 1353-1360. Ju, Y., Masuya, G., & Ronney, P. D. (1998). “Effects of Radiative Emission and Absorption on the Propagation and Extinction of Premixed Gas Flames.” Twenty-Seventh Symposium (International) on Combustion, The Combustion Institute, 27: 2619-2626. Law, C. K., & Chung, S. H. (1982). “Steady State Diffusion flame Structure with Lewis Number Variations.” Combustion Science and Technology, 29: 129-145. Loh, H. T., & Fernandez-Pello, A. C. (1984). “ A Study of the Controlling Mechanisms of Flow Assisted Flame Spread.” Proceedings of the Combustion Institute, 20: 1575-1582. Loh, H. T., & Fernandez-Pello, A. C. (1985). “Flow Assisted Flame Spread Over Thermally Thin Solids.” Proceedings of the First International Fire Safety Science Symposium, 1: 65. 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Lozinski, D., Buckmaster, J. D., & Ronney, P. D. (1994). “ Absolute Flammability Limits and Flame Balls.” Combustion and Flame, 97: 301. Markstein, G. H., & deRis, J. (1972). “Upward Fire Spread Over Textiles.” Proceedings of the Combustion Institute, 14: 1085-1097. Mason, H. B., & Seban, R. A. (1974). “Numerical Predictions for Turbulent Free Convection from Vertical Surfaces." International Journal of Heat Mass Transfer, 17:1329-1336. Olson, S. (1991). “Mechanisms of Microgravity Flame Spread over a Thin Solid Fuel: Oxygen and Opposed Flow Effects.” Combustion Science and Technology, 76: 160. Olson, S., Baum, H., & Kashiwagi, T. (1998). “Finger-Like Smoldering over Thin Cellulosic Sheets in Microgravity.” Twenty-Seventh Symposium (International) on Combustion, Combustion Institute, 27: 2525-2533. Olson, S., Ferkul, P. V., & Tien, J. S. (1988). “Near-Limit Flame Spread Over a Thin Solid Fuel in Microgravity.” Twenty-Second Symposium (International) on Combustion, The Combustion Institute, 22: 1213. Ronney, P. D., Greenberg, J. B., Zhang, Y., & Roegner E. V. (1995). “Flame Spread Over Thin Solid Fuels in Partially Premixed Atmospheres.” Combustion and Flame, 100: 474. Sacksteder, K. R., & Tien, J. S. (1994). “Buoyant Downward Diffusion Flame Spread and Extinction in Partial-Gravity Accelerations.” Twenty-Fifth Symposium (International) on Combustion, The Combustion Institute, 25: 1685. Schlichting, H. (1960). Boundary Layer Theory, 4th Edition. New York: McGraw-Hill. Sibulkin, M. (1988). “Free Convection Diffusion Flames from Burning Solid Fuels.” Progress in Energy and Combustion Science, 14:195. 116 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Tarifa, C. S., & Torralbo, A. M. (1987). “Flame Propagation Using the Interface Between a Gas and a Reacting Medium.” Eleventh Symposium (International) on Combustion, Combustion Institute, 11: 533-544. Thomas, P. H., & Webster, C. T. (1982). “Some Experiments on the Burning of Fabrics and the Height of Buoyant Diffusion Flames.” Combustion Science and Technology, 28: 173-175. West, J., Bhattacharjee, S., & Altenkirch, R. A. (1992a). “ A Comparison of the Roles Played by Natural and Forced Convection in Opposed-Flow Flame Spreading.” Combustion Science and Technology, 83: 233-244. West, J., Bhattacharjee, S., & Altenkirch, R. A. (1992b). “Investigation of Controlling Parameters in Transition Between Thermally Thin and Thermally Thick Flame Spread Over Solid Fuels in an Opposing Flow.” Fall Technical Meeting, Combustion Institute, Western States Section. October 1992. White, F. M. (1975). Viscous Fluid Flow. New York: McGraw-Hill. Williams, F. A. (1976). “Mechanisms of Fire Spread.” Sixteenth Symposium (International) on Combustion, The Combustion Institute, 16: 1281. Williams, F. A. (1985). Combustion Theory, 2nd Edition. Menlo Park, CA: Benjamin-Cummins. Zhang, Y., Ronney, P. D., Roegner, E., & Greenberg, J. B. (1992). “Lewis Number Effects on Flame Spreading Over Thin Solid Fuels.” Combustion and Flame, 90: 71-83. Zhou, L., & Fernandez-Pello, A. C. (1993). “ Turbulent, Concurrent, Ceiling Flame Spread: The Effect of Buoyancy.” Combustion and Flame, 92: 45-59. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX A FLOW TUNNEL A.1 Experimental Apparatus The flow-tunnel rig is similar in design and concept to the rig used in these experiments. The flow-tunnel was temporarily borrowed from Dr. Sandra Olson at the NASA GRC 2.2 second drop tower in Cleveland, Ohio. A full description of the flow-tunnel and experimental setup is shown by Olson (1991). Therefore, only a brief description will be given here. Figure 44 shows a very simplified schematic of the flow-tunnel. Kimwipe samples are held by steel quenching plates in the center of the cylindrical chamber. Ignition is performed using kanthal hotwires similar to those of our rig. A video camera views the sample through a window. A bottle of premixed oxidizer and diluent is regulated through an orifice and control valves. This gas is allowed to flow through a flow straightener, which allows a uniform flow field through the chamber tunnel. Gas is allowed to leave the chamber through vent holes at the top of the tunnel. A positive flow is established as shown in Figure 44 when the direction of flame spread and forced flow are in opposite directions (sample ignited on top). To achieve a negative flow at microgravity, the sample is ignited from the bottom. 118 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Gas Outlet Gas Bottle Test Chamber Drop Frame Window Camera Sample Holder Digital Image Processing System Internal Flow Flow Straightener Control valves Gas Inlet Fiber-optic Link VCR Side V ie w Figure 44. Schematic of Olson’s flow-tunnel. 119 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX B PROGRAMS B.1 “PPMS.BAS” ETIME = 0: ftime = 0: open "timel.dat" for output as #1 ’ STARTUP DEFINE VALVE SETTINGS %LOUT = 8: %DILUENT = 2 %DILUENTFINE= 1: %FUEL = 4 %OXIDIZER = 32: %OXIDIZERFINE = 64 %CATALYST = 16: %VACUUM = 128 %CHECKPRESS = 400: %CLOSEALL = 0 ' CP = PERCENT PRESSURES, CN = PARTIAL PRESSURES DIM CN(1:4): DIM CP(1:4) tstart = time: print #1, "start": print #1, time$ ' INITIALIZE I/O AND PRINT HEADER b = 768: out 782,0: wait 768, 128 wait 768, 128, 128: OUT 781,255: OUT 780,3: CLS: printprint: print “ Welcome to the partial pressure"; print “ gas mixing system.": print: print CALL TESTER ' GET GAS PARAMETERS FROM USER 1 PRINT “Please input the following:" PRINT:INPUT"FINAL TARGET PRESSURE: ",PFIN 120 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. INPUT'VACUUM PRESSURE: ",PVAC INPUT"% FUEL: ",CFUEL: 1NPUT"% OXIDIZER: ",COXID: INPUT'% CATALYST: ",CCATA: CDILU=100-(CCATA+COXID+CFUEL): PDIN=CDILU*PFIN/100: PRINT: INPUT "CORRECT(Y/N/Q)",A$ IF A$ = "Q" OR A$ = "q" THEN END: IF A$="N" OR A$="n" THEN GOTO 1 ' SORT LEAST TO GREATEST PRESSURE FOR 1=1 TO 3: FOR J=4 TO 2 STEP -1 IF CP(J)>CP(J-1) THEN GOTO 2: X=CP(J) CP(J)=CP(J-1): CP(J-1)=X 2 NEXT J: NEXT I ' CONVERSION TO PARTIAL PRESSURES FOR M=1 TO 4: CN(M)=PFIN*CP(M)/100: NEXT M CALL VALVE(%LOUT,ATM): CALL PRESS(4,ATM) ' EMPTY CHAMBER TO INITIAL VACUUM PRESSURE LOCATE 5,1: PRINT:PRINT"INITIAL VACUUM CYCLE" IF ATM<PVAC THEN LOCATE 6,1 : PRINT'SKIPPED DUE TO LOW INITIAL PRESSURE": SKIP=1 print #1, "vacuum cycle start": print #1, time$ 121 CP(1)=CFUEL CP(2)=COXID CP(4)=CCATA CP(3)=CDILU PRINT'% DILUENT: ",CDILU CLS Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ' FILL TO .6 ATM WITH OIL, THEN RE-VAC, TO MINIMIZE RESIDUAL GAS IF SKIP=1 THEN SKIP=0: GOTO 3 CALL VALVE(%LOUT + %VACUUM,ATM) DO WHILE ATM>PVAC*.8: CALL VALVE(%CHECKPRESS,ATM): LOOP CALL VALVE(%LOUT,ATM): delay 5 3 IF N=2 THEN GOTO 4 CALL FILLER(.6, .1,2, %LOUT + %DILUENT, ATM, etime, ftime) CALL VALVE(%LOUT,ATM): CALL PRESS(2,ATM) 4 CALL VALVE(%LOUT + %VACUUM,ATM) DO WHILE ATM>PVAC*.8 CALL VALVE(%CHECKPRESS,ATM): LOOP CALL VALVE(%LOUT,ATM): delay 2 A = TIMER: CALL WAITP(ATM, 0) B = TIMER: ETIME = ETIME + B - A print #1, "vacuum cycle end": print #1, time$ LOCATE 6,1: PRINT" ' CONSIDER REMAIN GAS IN CHAMB DIL, FINISH FILL TO 1/2 TOTAL DIL RESDILU = ATM: LOCATE 5,1 PRINT:PRINT USING"RESIDUAL DILUENT PRESSURE IS #.#### ATM";RESDILU PRINT:PRINT"FILLING OF GASES": 122 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PRINT:PRINT"PARTIAL DILUENT INPUT": T0T=PDIN/2 CALL FILLER(TOT,.05,1,%DILUENT + %LOUT,ATM, etime, ftime) A = TIMER: CALL WAITP(ATM, 0) B = TIMER: ETIME = ETIME + B - A print #1, "diluent cycle 1 end": print #1, time$ RDIL=ATM: CLS ' FILL WITH GASSES, IN LEAST TO GREATEST ORDER FOR S=1 TO 4 IF CP(S)=0 THEN GOTO 6 IF CP(S)=CFUEL AND FF=0 THEN GOSUB FUELER: GOTO 6 IF CP(S)=COXID AND OF=0 THEN GOSUB OXIDIZER: GOTO 6 IF CP(S)=CCATA AND CF=0 THEN GOSUB CATALYZER: GOTO 6 IF CP(S)=CDILU AND DF=0 THEN GOSUB DILUENTER: GOTO 6 6 print #1, "filling cycle # end": print #1 ,time$: NEXT S ' CLOSE ALL VALVES, DETERMINE RESULTANT GAS COMPOSITION CALL VALVE(%CLOSEALL,ATM) PERO=POXID/ATM*100: PERC=PCATA/ATM*100 PERF=PFUEL/ATM*100: PERD=PDILU/ATM*100 PERT=ATM/(PFIN)*100 ' DISPLAY RESULTS CLS: LOCATE 5,1: tend = timer IF POXID > 0 THEN OERR = 100* (POXID - COXID * ATM * .01) / POXID 123 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IF PCATA > 0 THEN CERR = 100 * (PCATA - CCATA * ATM * .01) / PCATA IF PFUEL > 0 THEN FERR = 100 * (PFUEL - CFUEL * ATM * .01) / PFUEL IF PDILU > O THEN DERR = 100* (PDILU - CDILU * ATM * .01) / PDILU PRINT:PRINT"FINAL RESULTS:" PRINT " GAS PRESSURE PERCENT DEVIATION":PRINT PRINT USING"OXIDIZER #.###ATM ###% ###%";POXID;PERO;OERR PRINT USING"CATALYST #.###ATM ###% ###%";PCATA;PERC;CERR PRINT USING" FUEL #.### ATM ###% ###%";PFUEL;PERF;FERR PRINT USING"DILUENT #.#### ATM ###% ###%";PDILU;PERD;DERR PRINT.PRINT USING"FINAL CHAMBER PRESSURE IS #.#### ATM";ATM PRINT USING'THIS IS ###.##% OF DESIRED FINAL PRESSURE";PERT PRINT USING"RESIDUAL GAS AFTER VACUUM: #.#### ATM";RESDILU print: print using "Total time elapsed: ###.# minutes"; (tend - tstart) / 60 print using "Time spent in extrapolation routine: ###.##"; etime / 60 print using "Time spent in filling: ###.## min"; ftime / 60: close #1 END ' VALVE - OPEN/CLOSE GIVEN VALVE(S) AND READ PRESSURE SUB VALVE(V,ATM) b = 768: IF V>255 THEN GOTO 20 'READ PRESSURE ONLY ' SET OPEN VALVES AND DISPLAY CURRENT SETTING VN=255-V: OUT 781 ,VN: LOCATE 22,1 124 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PRINT 'VALVE ENGAGED:": LOCATE 23, 1 IF V <> 0 THEN PRINT USING "###"; V: IF ATM = -1 THEN GOTO 15 ’ READ PRESSURE 20 OUT B,0 ’ wait b+ 15, 2 delay .1: L=INP(B+1): M=INP(B+2): H=INP(B+3) COUNTS=L+256*M+65536*H-4194304 VOLTS=COUNTS*2.3841857 * 10A (-6) IF VOLTS < 20 THEN GOTO 23 OUTb + 15, 1: out b, 0 1 WAIT 783, 2 delay .1: L=INP(B+1): M=INP(B+2): H=INP(B+3) COUNTS=L+256*M+65536*H-4194304 VOLTS=COUNTS*2.3841857 * 10A (-6) * 1.9386 OUTB + 15, 0 23 VOLTS=VOLTS-.61 ATM=VOLTS/2.94: LOCATE 20,1 PRINT "CURRENT PRESSURE" 1 DISPLAY PRESSURE PRINT USING "##.####"; ATM 15 END SUB I* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * ’ PRESS - DELAY FOR GIVEN TIME, READ PRESSURE DURING WAIT SUB PRESS(DEL.ATM) LOCAL J: MTIMER: J=MTIMER 125 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. DO WHILE J < DEL CALL VALVE(%CHECKPRESS,ATM) ' GET PRESSURE J= J + MTIMER/1000000:MTIMER ' FIND ELAPSED TIME LOOP END SUB ' FILLER - FILL TO TARGET PREESSURE (TP) WITHIN ERROR (ER) WITH ' VALVE SETTING VE SUB FILLER(TP,ER,RMSER,VE,ATM, etime, ftime) LOCAL FILTIM,TOTCH,E,RMSL,X, IP, sp, FILLRT flag = 0: FILLRT = 0: DELAY 2 E=3 ' INITIAL WAIT TIME CALL VALVE(%LOUT, IP) ' GET INITIAL PRESSURE IF IP > TP THEN GOTO 11: IF IP > TP - ER THEN E = 1 ' OPEN VALVES FOR E SECONDS a = timer 5 MTIMER FILTIM = 0: CALL VALVE(VE.ATM): LOCATE 13,1 PRINT USINGTILL TIME= ####.## SECONDS";E: count = e 16 if count < 1 then delay count else delay 1 count = count - 1: locate 13, 1: if count < 0 then count = 0 print using "FILL TIME= ####.## SECONDS”; count 126 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. if inkey$ = "A" then goto 17: if count > 0 then goto 16 17 b = timer CALL VALVE(%LOUT,ATM) FILT1M=MTIMER/1000000 ' AMOUNT OF TIME SPENT FILLING ftime = ftime + b - a: DELAY 5 CALL VALVE(%CHECKPRESS, ATM) sp = atm: A = TIMER IF RMSER < 1 AND ATM > TP - ER THEN CALL WAITP(ATM, TP - ER) B = TIMER: ETIME = ETIME + B -A IF ATM > TP - ER THEN GOTO 11 ' IF TARGET PRESSURE NOT REACHED, DETERMINE HOW LONG IT ' WILL TAKE AT LAST RATE OF FILL TOTCH = (TP - ER)-ATM IF FILTIM >= 5 OR FILLRT <= 0 THEN FILLRT = FILTIM / (SP - IP) ADTIM = TOTCH * FILLRT: sp = atm: E = ADTIM: E = E * .8 if inkey$ = "r" then e = 5: IF E<.05 THEN E=.05 PRINT #1, E, SP: IP = SP GOTO 5 11 END SUB I* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 1 FILL TO CURRENT PRESSURE + PARTIAL PRESSURE OF FUEL WITH ' FUEL THEN DETERMINE AMOUNT ADDED FUELER: 127 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ATMB=ATM:FF=1: LOCATE 5,1 PRINT:PRINT"FUEL INPUT": TAR=ATM+CN(S) PRINT:PRINT USING "TARGET PRESSURE IS #.#### ATM";TAR CALL FILLER(TAR,.005 * CN(S),.0009,%FUEL + % LOUT, ATM, etime, ftime) PFUEL=ATM-ATMB: LOCATE 6,1: PRINT RETURN I*************'*'******************'***********'******************** ' FILL TO CURRENT PRESSURE + PARTIAL PRESSURE OF OXIDIZER ’ WITH OXIDIZER THEN DETERINE AMOUNT ADDED OXIDIZER: ATMB=ATM:OF=1: LOCATE 5,1; PRINT:PRINT"OXIDIZER INPUT": TAR=ATM+CN(S) PRINT:PRINT USING'TARGET PRESSURE IS #.#### ATM";TAR ' DO MOST OF FILL WITH COARSE VALVE, FINISH WITH FINE CALL FILLER(TAR,.025 * CN(S),1,%OXIDIZER + %LOUT,ATM, etime, ftime) a = timer: CALL WAITP(ATM, 0) b = timer: etime = etime + b - a IF ATM>=.998*TAR THEN GOTO 7 CALL FILLER(TAR,.005 * CN(S),.0009, %OXIDIZERFINE + %LOUT, ATM, etime, ftime) 7 POXID=ATM-ATMB 128 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LOCATE 6,1: PRINT RETURN I**************************************************************** ' FILL TO CURRENT PRESSURE + PARTIAL PRESSURE OF CATALYZER ' WITH CATALYZER, DETERMINE AMOUNT ADDED CATALYZER: ATMB=ATM:CF=1: LOCATE 5,1 PRINT:PRINT"CATALYST INPUT": TAR=ATM+CN(S) PRINTPRINT USING'TARGET PRESSURE IS #.#### ATM" JAR CALL FILLER(TAR,.005 * CN(S),.Q009,%CATALYST + %LOUT,ATM, etime, ftime) PCATA=ATM-ATMB: LOCATE 6,1: PRINT RETURN I* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * ' ADD REMAINING HALF OF DILUENT GAS TO MIXTURE ' THEN DETERMINE TOTAL AMOUNT OF DILUENT DILUENTER: LOCATE 5,1: PRINT:PRINT"FINAL DILUENT INPUT" ATMB=ATM: TAR=ATM+CN(S)-RDIL PRINTPRINT USING'TARGET PRESSURE IS #.#### ATM";TAR ' DO MOST OF FILL WITH COARSE VALVE, FINISH WITH FINE CALL FILLER(TAR,.025 * CN(S),1,%DILUENT + %LOUT,ATM, etime, ftime) 129 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. a = timer: CALL WAITP(ATM, 0) b = timer: etime = etime + b - a IF ATM>= 998*TAR THEN GOTO 8 CALL FILLER (TAR,.0025 * CN(S),.0009, %DILUENTFINE + %LOUT,ATM, etime, ftime) 8 PDILU=ATM-ATMB+RDIL V *i c 'ir Jr , & c a t c *ic A . tflp *i cfc *i c i ^ c ^ c * fc *ic * $ ( 'it ' $ ( *ic 'id d t J p SUB WAITP(ATM, TP) 3010 CALL VALVE(%CHECKPRESS, A): DELAY 10 CALL VALVE(%CHECKPRESS, B) IF B < TP OR ABS(B - A) < .00005 THEN ATM = B : GOTO 3020 GOTO 3010 3020 'END OF SUBROUTINE END SUB !***************************************************************** LOCATE 6,1: PRINT RETURN SUB TESTER 1001 PRINT "Checking Valves... (F1 to skip)’ ON KEY(1) GOSUB 1002 KEY(1) ON: DONTREAD = -1: OKAY = 1 CALL VALVE(1 .DONTREAD): DELAY .1 130 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CALL VALVE(2,D0NTREAD): CALL VALVE(4, DONTREAD): CALL VALVE(8,D0NTREAD): CALL VALVE(16,DONTREAD): CALL VALVE(32,DONTREAD): CALL VALVE(64,DONTREAD): CALL VALVE(128,DONTREAD): DELAY .1 DELAY .1 DELAY .1 DELAY. 1 DELAY. 1 DELAY .1 DELAY .1 CALL VALVE(%DILUENT + %LOUT + %VACUUM, DONTREAD) DELAY 5 CALL VALVE(%LOUT + %VACUUM, DONTREAD): DELAY 3 CALL VALVE(%OXIDIZER + %LOUT + %VACUUM, DONTREAD) DELAY 5 CALL VALVE(%CLOSEALL, DONTREAD) ' %LOUT = 8 ' %DILUENTFINE = 1; ' %OXIDIZER = 32 ' %CATALYST =16 ' %DILUENT = 2 ' %FUEL = 4 ' %OXIDIZERFINE = 64 ' % VACUUM = 128 %CLOSEALL = 0 DELAY 2 ' %CHECKPRESS = 400 CALL VALVE(%LOUT, DONTREAD): CALL VALVE(%LOUT, ATM1) CALL VALVE(%LOUT + %VACUUM, DONTREAD): DELAY 4 CALL VALVE(%LOUT + %VACUUM, ATM2) 131 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LOCATE 12,1 : print "vacuum values:": print atml, atm2: print IF ATM2 > ATM1/1.5 THEN LOCATE 8,1 : PRINT "vacuum on?": OKAY = 0: DELAY 1 CALL VALVE(%VACUUM, DONTREAD) CALL VALVE(%CLOSEALL, DONTREAD): DELAY .5 CALL VALVE(%CLOSEALL, ATM1) CALL VALVE(%DILUENT, DONTREAD): DELAY 1 CALL VALVE(%CLOSEALL, DONTREAD): DELAY .5 CALL VALVE(%CLOSEALL, ATM2): LOCATE 13,1 : print "diluent values": print atml, atm2: print IF ATM2=<ATM1+.25 OR ATM2=<2 THEN LOCATE 9,1 : PRINT "Is diluent on?": OKAY = 0 CALL VALVE(%LOUT + %VACUUM, ATM2): DELAY 5 CALL VALVE(%VACUUM, DONTREAD) CALL VALVE(%CLOSEALL, DONTREAD): DELAY .5 CALL VALVE(%CLOSEALL, ATM1) CALL VALVE(%OXIDIZER, DONTREAD): DELAY 1 CALL VALVE(%CLOSEALL, DONTREAD): DELAY .5 CALL VALVE(%CLOSEALL, ATM2) LOCATE 14,1 : print "oxidizer values": print atml, atm2; print IF ATM2=<ATM1+.25 OR ATM2=<2 THEN LOCATE 10,1 : PRINT "Is oxidizer on?": OKAY = 0 132 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. locate 15,1: for i = 1 to 9; print" " next I; print" locate 16,1; PRINT 'Valve Check Done": KEY(1) OFF IF OKAY = 1 THEN goto 1 PRINT "EQUIPMENT FAULT DETECTED. PLEASE CHECK SETUP" PRINT CHR$(7) PRINT "Enter 'c' to cancel tests, or any other key to repeat" input a$: if a$ = "c" then goto 1: CLS GOTO 1001 1002 KEY(1) OFF CALL VALVE(%CLOSEALL, DONTREAD) GOTO 1 END SUB 133 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. B.2 “BEFORE.TTB” REM LINTON HONDA PACIFIC RIG CONTROL PROGRAM 2/28/99 PSET 14:PCLR 0,1,2:PSET 15:PCLR4:A=PIN(0,1,2,4,14,15) PCLR 2,3,4,5,6,7,8,9,10,11,12,13 SLEEP 50: DIM(1024,16384): XMIT- 10 SLEEP 0 REM DEFAULT WAIT TIME: 5 sec 30 W=500 45 X=0 50 V=0 REM MAIN MENU 120 PRINT '1. EXIT PROGRAM' 135 PRINT '2. RUN DROP' 140 INPUT 'SELECTION: ' A REM USER ENTRY: REM ENTERS: 1 250 IF A<>1 GOTO 1000 260 STOP REM ENTERS: 2 990 PRINT "TURN ON 28 VOLTS NOW' REM DOUBLE CHECK CONNECTIONS 1000 FOR 1=1 TO 54: PRINT\8;: NEXTI: SLEEP10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1010 PRINT "PLEASE CHECK CONNECTIONS AGAIN.."; 1020 SLEEP 25: IF PIN(O) = 0 GOTO 1000 1030 PRINT "THANK YOU, LINTON. CONNECTION VERIFIED.",\10: SLEEP 25 1040 PSET 2,9 REM DROP SEQUENCE 1310 IF PIN(0)=1 GOTO 1310 1331 PCLR2 REM RECORD INITIAL VALUES 1332 STORE X,#2,? 1333 STORE X,#2,CHAN(1) 1334 STORE X,#2,CHAN(2) 1335 STORE X,#2,CHAN(4) 1338 STORE X,#2,CHAN(7) REM FLASH IGNITOR 4 TIMES 1340 FOR K=1 TO 4 1350 PSET 3:SLEEP 3:PCLR 3 1360 SLEEP 3 1370 NEXT K 1380 PSET 2 REM RECORD START TIME 1440 S=? Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1450 PCLR2 REM STORE TIME AND DATA 1460 STORE X,#2,? 1461 STORE X,#2,CHAN(1) 1462 STORE X,#2,CHAN(2) 1463 STORE X,#2,CHAN(4) 1466 STORE X,#2,CHAN(7) REM LOOP FOR 5 SECONDS 1480 IF ?>S+10&V=0 PSET 2:V=1 1490 IF ?<S+1000 GOTO 1460 REM STORE FILE TRAILER 1500 FOR L=1 TO 30:STORE X,#2,1:NEXT L 1502 SLEEP 5 1505 PSET8 1506 SLEEP 10 1510 SLEEP 500 1520 PCLR 2,3,8,9 1530 STOP SLEEP70: XMIT+: SLEEP10: STATUS Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. B.3 “AFTER.TTB” REM LINTON HONDA PACIFIC GET DATA PROGRAM 2/28/99 PSET 14:PCLR 0,1,2:PSET 15:PCLR4:A=P1N(0,1,2,4,14,15) PCLR 2,3,4,5,6,7,8,9,10,11,12,13 SLEEP 50 5 X=0 6 Q=0 10 FOR L=1 TO 5 20 A=GET(X, #2) 30 IF A=1 Q=Q+1 35 IF Q=20 STOP 40 PRINT A,' 50 NEXTL 60 PRINT \13 70 GOTO 10 137 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX C DATA Actual flame spread rates are provided for reference below. Tables 7 through 11 show opposed-flow flame spread data at 1g and pg for argon, helium, nitrogen, carbon dioxide, and sulfur-hexafluoride respectively. Figure 45 shows opposed-flow flame spread data and preliminary plots for thick fuels. Table 12 shows opposed-flow flame spread data for partially-premixed fuel. Figure 46 shows a cropped output file from AFTER.TTB. Finally, Tables 13 through 17 show concurrent-flow flame spread and length data. C.1 Opposed - Flow Flame Spread Oxygen Cone. Pressure Widtfi... fc J t ig upward Ct 1g downward StOg m ~ ~ ......" (atm) (cm) (cm/sec) (cm/sec) (cm/sec) ........."27700 TDD 5.00 3.50 2.25 24.00 .........1700 5.C0 ........... 3700 2.00 21.00 1.00 ...... 5X 0 2.13 1.38 18.00 1.00 5.C0 1.38 0.75 15.00 TOO -5 X 0 0.88 0.38 14.00 1.00 5X0 0.65 0.25 13.00 TOO" 5.00 0.50 0.25 12.50 .........T.Q01 5.00 0.38 extinction 12.00 1.00 5 X 5 0.32 ^ j 11.75 1.00 ~ "" 5700 0.28 11.50 1.00 5.00 0.22 Table 7. Opposed-flow flame spread data for thin fuels in argon. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. oxygen uonc. Pressure Width (% )'" ... (atm) (cm) 50 1.00 5.00 46 1.00 5.00 42 1 0 0 5.00 40 1.00 " “ “5300 .......-......... “39 1.00 .............5300 38 1.00 ........ 5700 ............ 3 6 “1 .0 0 5.00 34 1.00 5 00 ................ ~ 32 ...........1 0 0 .............5.00 ......3T 1.00 5.00 30 1.00 5.00 29 1.00 5.00 28 1.00 5.00 50 0325 5.00 50 ‘ “ ““ 9350 5.00 50 1.00 5.00 50 ...... 1.50 “ 5300" " " 5 0 2.00 5.00 50 4.00 " ... " 5700* .........40 1.00 ............. 5300 .........“4 0 1.00 ......5.00 40 1.00 5.00 40 1 0 0 5.00 40 ..1 0 0 ......... 5 9 0 40 1.00 5300 ............ 40 1.00 5 9 0 40 1.00 5 9 0 ....... ...... 40 ..........T.OO 5.00 50 1.00 8.20 50 1.00 5.00 ........... 50 1.00 2.00 50 1.00 2.00 50 1.00 1 0 0 50 1.00 0.80 “5 0 1700 0.40 50 1.00 0.30 45 2300 8.30 45 2.00 6.68 ............... 4 5 2.00 4.00 45 2.00 2.00 45 2.00 1.00 45 “ ZOO 0.50 45 2.00 0.40 45 2.00 0.30 45 2300 0.25 4 5 2.00 ............0720 ..........." “45 7700 0.15 46 2.94 8 20 46 “2794 6 00 46 2.94 4 00 46 2.94 2.00 46 .... ““2394 1.00 46 2394 0.50 46 7 3 9 4 0.30 46 2,94 0.25 Forced Velocity |S t1 g upward (cm/sec) (cm/sec) St 1g downward (cm/sec) StOg (cm /sec) -4 00 1 00 0 60 0 00 1 00 2 00 4 00 8 00 8 00 extinction ext rction e x t i n c t i o n ” -7 5 1 5 0 25 00 “25130 19 7C O 1677G 10150 10150 “5320 TB30T 17760 16350 15300 13360 TT350 10300 7750 “ 3375 1307 203 00 “ 20 3 0 0 T 3 1 0 18 3C C 1 7 1 0 14.00 "7360 “ 1 0 ' lO 0fc4 0 44 0 32 0 27 130? 1 6 0 1 2 6 1 0 5 ex met on 036 4 “ 0347 “0.33 “ 0333 15330 Ixtinction 23001 1.40 1 0 1 30 Table 8. Opposed-flow flame spread data for thin fuels in SF6. 1 94 139 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Oxygen Cone. Pressure ' W i d t h ....... S t 1g upward S t 1g downward StOg (%) (atm) (cm) (cm/sec) (cm/sec) (cm/sec) 40150 " “ 1 70 0 5.0C 5.00 2.50 36700 1.00 5.0C ■ H 9 m h 9 B 3 | I M H H B M M ' V .S . ^ 1 % * , — H m P H I 4.50 2.40 30.00 1.00 ......... 570C 3.50 2.14 24.00 1.00 5.0C 2.50 1.50 2300 .....“ '1700' 5.0C 2.25 1.40 22.00 .... T706 5.00 2.10 1.20 21.00 '1700 5.00 1 9 0 ■ ”0.80 20706 - 1 T ( J 0 - ......... 5 0 0 1.75 extinction 20.06 1.00 5.00 1.88 19.00 ' 1700 5 00 1.67 18.00 1.00 5.00 1.25 18.00 1.00 5.00 1.43 17775 1700 5.00 1.30 17.50 1.00 5.00 7 ” ......... 1.30 .... 1 7 .2 5 1.00 5.00 1.20 17.00 1.00 ............5.00 1.03 17.00 1.00 .......... 5700 1.25 16750 TOO 5.00 1.03 16.00 1.00 S.Ofl extinction 25.00 1.00 ........... "8.00 T8.0Q 2.50 2 5 0 0 1.00 " ”5700 16.50 300 1 25.00 1.00 3.00 13.50 2 5 0 0 1700 ' “ ZOO 10.50 25.00 ...........1700 1755 8.00 25.00 1.00 1.20 5.50 '2 5 0 0 1.00 1.00 3.00 25.00 1.00 0.90 2.00 2500 1.00 0.80 extinction 30700 " "0725 7700 9.00 30.00 0.25 ......... 5700 8.00 30.00 0.25 3.00 5.00 30.00 0.25 2.00 3.00 30.00 70725 .........1.50 1.67 30.00 0.25 1.00 0.60 30.00 0.25 0.90 extinction 30700 _ u: 2S 5 0 0 8.00 3.00 1.25 30.00 0.50 1 5 7 0C 3.30 1.50 30.00 TOO 570( 3.50 2.14 30700 1.50 ...... .....”5 0 0 4.00 2.25 30700 2.00 ............. 5700 4.25 2.50 Table 9. Opposed-flow flame spread data for helium. 140 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. U xygen Cone. V% T Hressure {width |St ig u.pward | t 1 g downward (atm ) (cm ) (cm /sec) 5 .0 0 5 7 0 0 5750 5700 5.00 5”00 (crn/sec)" St Og (crnTse c) T079O 3 6 .0 0 3 3 .0 0 2 5 7 0 0 ” 27700 TTOT 21 .00 0750 8.00 tiw t t o o 57T 5 5 7 5 0 725" ion0 “ o r 2 7 .0 0 2 7 .0 0 2 7 7 0 0 ” 27700 2 7700 ""2T700 21700 2Too “ Z T7W 21 .00 21.00 ” 21 .00 21 T O O 21 .00 21 .00 " ttto o " 21 .00 "271700 2 1 .00 2TT00 "21700" 2T70O ”21 .00 21 .00 21 .00 2 1.00 TTOO 21 .00 21 .00 21 .00 21 .00 21 .00 21 .00 2 1 .00 21 .00 21 .00 TTOO 21 .00 “ 2TTOO TTTOO 2T7O0 2 1 .0 0 " 2 1 .C T 0 " TTTOO 21 .00 ”21 .00 21 .00 "TTTOO" 21.00 "2170 0 21.00 "TTTOO 21.00 2 1 0 0 TTTOO ” 2 1 . 0 0 1 .00 "TTOO" 1700 1 .00 1 .00 "TTOO 1 .00 1 .00 "TTOO "TTOO "TTOO 1.00 TTOO TTOO ” 1 7 0 0 "TTOO "TO T "0750 1 .00 1 .50 TTOO 1 .00 TTOO" TTOO TTOO 1 .00 TTOO TTOO ” 1 7 0 0 TTOO 1 .00 " T O O "TTOO 1700 TTOO 1 .00 T O O TTOO 1 .00 TTOO TTOO T O O T T O O TTOO TTOO T O O T T O O T O O T O O TTOO TTOO 1.00 T O O TTOO TTOO T O O T O O T O O T O O T O O T O O T O O TTOO TTOO TTOO TTOO T O O TTOO T O O TTOO T O O 3 .0 0 "2TOO 2 .5 0 T O O "2 700 "TO O T T O 1 .50 T O O T O O ” 0700 TTOO 0 0 7 "OTTO 0 .7 5 ” 0700 "0736" ” 070 0 " O T O O ” 0750" extin ctio n ” 074 u " O T exfTnOfToTT ” 8.20 " T T O O 6.5 0 0 7 0 0 ” 0700 6.00 _______ _ _ ” “5700 5.0 0 T E 00 T T O O ” 2 00” 2 1 4 .00 T O O “OTOO "OTOO "2770" "2700 ” 2720 "2700 "2700 "2 7 0 0 ” 2700 "2700 2.00 T 7 7 0 TTOO T O O " T 7 2 0 T O O T O O T O O TTOO TTOO T O O 0.9 0 "0700 0.8 0 0700 70700 "OTOO 0 .7 5 "OTTO 0 .7 0 076 4 0.62 0 . " 6 0 "0700 0 760 “OTOO ” 0750 1 .90 2 4 .0 0 2 0 70 0 27720 T5750 17.00 rsroo 0707 T 5 T 1 “ 2 07 0 "T3T2H "TB77 T 5 T 6 TTTT1 T T 7 T / T T 7 9 2 T3740 “T2T01 T 2 T 0 4 1 0.7 1 T 3 7 T 1 7704” T T T 01 9T7T "776 ’ T f j "779 ” 872' "7789 ” 8 T ” 0756 ” 07756 ”57? "5 75 / "576 ) ”672" 3 ” 075) "47 8 " T T O c < TT j 2 7 0 ) 2 7 7 ” 5 7 5 ' " 2 2 22 2 .6 5 i Table 10. Opposed-flow flame spread data for nitrogen. ■ 141 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Oxygen C o n e , j P re s s u r e !W iW (atm) (cm) bflg u p w a r d (cm/sec) Sl 1 g d o w n w a r d (cm/sec) S T D g ” (cmfsec) T070Q “D8OT’ "DDOT 1 ® ' OTDD' " 2 2 O T " TT.00 otoo otdd T O T ' TO T' TO T' T7OT T O T ' ■ ; j y 2 5 " T O T ' T O T ' TO T' TO T' TDD T DD" DD' 00" 00' DD' DD' D D D D TD D ' D D D D TDD TOT T T OT DD' " 5 7 0 C mm "O T ■ ■ "O T ~5OT - M 1 s m "5.C0 TOO "O T I V M r M * "500 im t ■ I gnF f l O T "B C D -SS 500 -O T 5 .U U 1 "O T extinction "O T "O T "O T " 2 7 4 0 "O T 2.50 T " 1 4 D O " 2 7 2 5 "2 7 1 0 TD D TO 2D " 0 7 8 0 1.75 extinction 1 78 5 T67 TOT TO T TDD TDD TOT TOT TOT TOT extinction D.90 extinction Table 11. Opposed-flow flame spread data for CO2. 142 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Thickness vs Sf at 40% 0 2 1 002 @ 4 atm S Normal q Open Both Sides 0.2 0.4 0.6 0.8 Thickness (in) 02 Concentration vs Sf @ 4 atm 9 8 . 7 . 25 30 35 40 45 02 Concentration (%) 50 55 5 0 . 5 . 0 . 5 . 0 - 5 . 0 . 5 . 0 . 5 - 0 - 5 . 0 . Pressure vs Sf at 40% 0 2 / C02 @ 4 atm 3 4 Pressure (atm) Normal Open Both Sides 02 0.03 1.39 0.04 3.08 0.08 4.33 0.10 4.72 0.24 4.56 0.38 4.56 0.50 4.83 0.88 4.72 0.08 Sf 0.75 0.56 t o o 0.94 2.00 1.67 3.00 3.33 4.00 4.83 4.40 5.28 5.00 2.78 5.50 5.83 6.00 5.67 5.00 5.94 Sf 27.50 0.63 30.00 2.50 35.00 4.17 40.00 4.83 50.00 8.75 5.00 Figure 45. Example of spreadsheet used to graph results for thickness, oxygen concentration, and pressure effects. 143 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. * ■ » ■ P % 1 » I M b 8— i ■gSf ucB B M M u£F '3 I K S uS r - ’f ........................... M M M M M M i-orn men & 5.0 30.0 65.0 0.33 3.00 3.00 first drop may be suspect to errors in phi 3.0 30.0 67.0 0.20 2.70 2.30 tall flame at ug 5.0 30.0 65.0 0.33 3.50 4.20 very tall flame thickness at ug 1.0 30.0 69.0 0.06 1.76 1.20 small flame at ug 4.0 30.0 66.0 0.26 2.75 3.00 tall flame ug 0.0 30.0 70.0 0.00 1.70 1.10 very small flame thickness at ug 6.0 30.0 64.0 0.40 60.00 phi is way to high, alm ost a com bustable mixture before drop 5.0 30.0 65.0 0.33 2.50 3.75 very tali ug flame 6.0 30.0 64.0 0.10 2.60 2.14 flame slows down just a bit 9.0 30.0 61.0 0.15 2,72 2.14 flame slow down 10.8 18.0 71.2 0.30 1.50 0.46 1.50 flame propagates much faster than pyrolysis front 14.4 18.0 67.6 0.40 1.70 normal flame spread in 1g, but com bustable gas in ug 7.2 18.0 74.8 0.20 1.36 1.00 1.00 flame propagates at sam e speed as pyrolisis front 5.4 18.0 76.6 0.15 1.20 0.75 0.75 flame spread not robust enough to survive impact 12.6 18.0 69.4 0.35 1.66 1.07 2.14 2.14 only CO mixture with separated flame front 3.6 18.0 78.4 0.10 1.15 0.63 0.75 flame spread not robust enough to survive impact 10.8 18.0 71.2 0.30 ignition during drop 1.8 18.0 80.2 0.20 0.91 0.30 0.51 flame spread not robust enough to survive impact 0.9 18.0 81.1 0.10 1.00 0.50 0.50 flame spread not robust enough to survive impact 3.2 18.0 78.9 0.35 1.25 0.50 0.88 flame spread not robust enough to survive impact 4.5 18.0 77.5 0.50 1.25 0.50 1.10 flame spread not robust enough to survive impact Table 12. Partially-premixed flame spread data 144 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I im e .... k i ...... .... ......'"-ppj ~ R 3 ............ K 4 10630 3008 4320 2592 3216 10657 2960 4928 4960 3280 10659 5040 5488 6208 4272 10662 5952 5120 7056 3824 10664 5728 4992 9136 4 3 04 10667 7120 5264 1 0048 3424 10669 8800 5312 10352 3472 10672 9520 5152 11984 4016 10674 11120 5584 12144 40 48 10677 11840 5408 12400 3328 10679 12176 5664 12704 3744 10682 12800 5344 12464 3760 10684 13024 5104 12400 43 6 8 10686 12880 5584 12064 3504 1 0689 12336 5408 12208 42 7 2 10691 12784 5424 11696 3280 Figure 46. Example of partial output file 080299-5.dat. Time is in 1/100t h second values with 00000 corresponding to time of starting program. Voltages measured are 1/13107 per volt input. These are usually pre-amplified by 200 times before the D-DACS receives them. 145 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. C.2 Concurrent - Flow Flame Spread Sarrple v u c S l i i n n m Sf GashofMirter N o n D m ensknalAr @1.00 atm Sarrple vydthirm Sf Qadxf Nurrber N b n Orrensional^ r @ 1.00 atm ffiBJ.. 25.9 0 ..... 2 8 7 8 ...... 600 6® 20513 7.29 70.0 0 24.0 0 ~ .." \ m m r te s t ...3 '® ". '""""£45"" ""20613.. 6 ® ' m m . 20.0 0 1 1 0 0 2 5 3 2 2222 800 5.57 2 0 5 1 3 619 eO® “'"17$. .. 8 6 5 3 7 9 0 1 9 9 1 ...W " ' 2 0 5 1 3 632 60.00 27.20 8 6 5 3 7 9 0 30.22 8 ® 6.23 2 0 5 1 3 6® 00.00 1 0 . 8 0 8 6 5 3 7 9 0 2080 ...I S O ' " " ' """W '"" 1 6 9 0 2 • 7.32 • m m " "'17300" 6 6 0 6 6 2 1 1389 7.00 4.85 1 3 7 4 2 5® s o . d o m o o 5 0 0 7 9 0 0 2000 7.00 4 ® 13742 5 .21 4 O .O 0 19.00 2 5 6 4 0 8 6 2 1 . 1 1 638 4.19 • 1 0 4 0 4 4 ® 4 O .O 0 1511 2 5 6 4 0 8 0 1379 6.10 9 5 0 2 4.32 35 .0 0 2007.. 1 7 1 7 7 3 7 2230 6.00 375 8 6 5 4 4.17 30.0 0 1328 ....1 0 6 1 7 2T™ 1 4 . 7 6 600 355“ 8 0 5 4 3.94 "WET .. 1525'"'" 8 3 3 2 0 3 1300 . . .8 ® .... 327 " " W " 3® 2500 1316 6 2 5 0 9 8 1 4 . 6 2 677 325 7 6 0 6 361 2250 1 1 .7 1 455352 1 3 0 1 550 2 ® £ X X X Z 000 0 294 . 2 D ;a r 14.77 3 2 0 5 1 1 1641 5.25 2® 5797 224 2D .00 11.92 3 2 0 5 1 1 1324 5.® — jg g - 5 0 0 8 ' ™ ' 2® 2 O 0 0 134) .......3 2 0 5 1 1 ...... 1 4 . 8 9 5 ® 2® "5 00 8 '" " 2® 1 2 9 1 3 2 0 5 1 1 14.34 5.® . . .2 ® . . . '50® 2® 20.00 1204 3 2 0 5 1 1 1338 5.® 1.® 5 0 0 6 2® 23.00 ..." 1 0 l 7 T ''3 2 0 5 1 1 ...... 11.00 5.® 1.® 50®' .....T87 17.50 1341 2 1 4 7 1 7 14.00 """5.® ""'' 1 . 7 8 .. .'s m ~ .." ...1 7 0 6 ....... '" is m " " i m . .. 1 3 5 2 1 5 8.49 4.75 '""'0 ® . . . 4 2 0 4 1.® 1250 11.33 7 8 2 5 0 1259 4.® 1.10 3 0 5 1 1 .2 2 ..123T 9 . 7 1 7 8 2 5 0 ..."""1679 4.® 1.38 "38BT" 1 S ...... .w a r 7.62 4 0 0 6 4 347 4.® 1.® 3 6 5 1 1 .4 8 1 0 . ( 3 0 349 4)084 943 ...4 ® .. 1 .1 6 3 6 5 1 1 . 2 0 1 0 . 0 0 7.90 4164 381 4.® Q88 2 5 6 4 9® 1000 328 ......40064..... 920 4.® 0.73 2 5 6 4 0 .8 1 1 Q O 0 ""7 ® """ 4 0 0 6 1 377.... 4.® 0 .7 S T - 2 5 0 4 988 1Q 00 313 4 0 0 6 4 9.03 4.® 0.® 2 5 0 4 ' 1.® 9.00 6.58 2 9 2 0 7 7 . 3 1 3® 0® 2 3 7 7 0.66 Table 13. Non-dimensional flame spread data for concurrent-flow flame spread (air). 146 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Sample w idth mm Sf Grashof Number 30% 02/N 2 @ 1.00 atm 30% 02/N 2 @ 0.50 atm 30% 02/N 2 @ 0.25 atm 40% 02/N 2 @ 0.25 atm 50% 02/N 2 @ 0.25 atm 50% 02/N 2 @ 0.25 atm t=2 30% 02/N 2 @ 2.00 atm S'2.00 33.00 "ZJOObTTS1 - " 2 0 T 2 T 4U.UI) 37.74 2554277 20.07 "30:00 34:00... 107/666 .......1'8.09 ... ~"zd: oo~ ...28:00'.. 3T5)255 15.00 18.05' 134698 a.6o 12.00.. " 1 T 6 4 " " 68965 . . . ^ "TO VO O ’12:85". 39911 ... ... 0:84 ....s :m /./4 20434 .........4:12' 6.UU " 5.00.... 8621 2.66 " W ' - 3:00 ■ ".... '4989""'.. ..'O'O 4.00"" 2.00 4554 1.06 ...ooi'oir- 25.80... "2 6 9 3 9 6 TSTOF 2 5 '.W " 21.10 156901... 14.31 20:00"“ 15.63 ....79821 ... .......9 12 16.00 ..T J T T " 338/5 '8.03 14.50 11.79 ...T W " " O T J ff 10.00 8.63 ■.... 9 9 7 8 ...... 5;U4 7.50 4:T5'"" """4 20 9"... 2.42 .... 5'.00 1.01 7 2 4 7 " " ...0:59' 4:00... U.6U 639 ” ' 0.35 40.00' ..l a w 158729 .... 9:50 3U.U0 ■ "Tgroo"- 66964 .......9.00 20.00 ...75:00... 19841 7.50 15.U0 9.00 83 /0 4.50 10:00 3.40... 2480 1 1.60 7.50 ....1.70'" ..... TO'4'6 ..... 0.85 5.80'' U.8U 484 0.40 20.00 ...t o' to'" ' 19864 3.57 16.00 " 8.80 8376 2.93 10.00... 4.80 4484 1.60 8.00.... 3.00 14/1 T;U0' ...CTO” ...T .5U " "...."570 0.5U 5.00'.... 0.90 310 0.30 4.40 0.65 184 .......0.18 16.00'" TU TJO "" "" 8360..... 2.86 lUOO” ' 6.82 ■ ■ - — 2-477 1.95 9.00"" '""5V56.... 1806 1.69 /.0 0 ... 3.49 850 .."”"i:uo 6.00 2.34 535 0.6/ 5.6U 2.07 435 ........'0.59 5.40 .... 1:76' - .....'"'390..... U.5U 5 .2 0 '" 1.50 348 0.43 5 .0 0 .... " ' 7.34” 310 0.38 4 .HO' 1.20 274 U7J4' 4.60 i:00..... 241 “ ' "0.29 4.20 0.90 184 ~1T26 4 .0 0 ... 0.71 159 ........0.20 1 5 0 0 5.45 8360 3.21 12.50 S '.O T J " " “ 4838-..... 2.56 lO O O " 3.85 24/7 1.9/ 8.5U ' TOO' ' 1521 1.54 7.00 4.1b 860 - —y T D - 6.00"" v |-4 - 5- - 635 .......'"0'.74' 5.00 0.85 310 U.44 bO.uO 39.00 34438/76 ’ '15.29 37'.0TT 36.00” ' 6076052 14.12 20.00.... " "O O VO O.. 1275510 11.76 i u .uO 17.70 " 159439.... 8.00.. 14.28 81633 6.60 6.00 T U .7 7 ' 34439 4.20 4.00..... 5.08 " "1020'4..... 1.99 BOO' " 2:88" " ”4305 1.13 2.00...." 1.05 .... 1276....... 0.41 Table 14. Non dimensional flame spread data for concurrent-flow flame spread (N2). 147 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Sam ple width mm Sf Grashof Number 35% 0 2 /C 0 2 @ 1.00 atm 30% 0 2 /C 0 2 @ 2.00 atm 50% 0 2 /S F 6 @ 1.00 atm 45% 0 2 /S F 6 @ 2.00 atm 46% 0 2 /S F 6 @ 2.94 atm 25% 0 2 /H e @ 1 atm 3o% 0 2 /H e @ .25 atm 82.00 25.00 48547035 16.67 50.00 22.00 11006042 14.67 30.00 'is :o 0 “ 2377305 '10.67 20.00 13.00 704387 0 7 ” " 10.oO 10.00 88048 6.67 6.00 0.40 19018 3.60 4.00 3.00 5635 2.00 82.00 25.40 220024697 23.74 60.00 21.00 86391185 19.63 40.00 17.80 " 25597388 16.64 20.00 13.80 3199674 12.00 10.00 11.10 399950 10.37 6.00 9.30 86301 8.69 5.00 8.30 49995 7.76 4.00 6.00 25597 5.61 3.60 3.50 10799 3.27 2.50 2.10 62'4'9 1.96 82.00 28.00 262153661 18.67 50.60 25.00 59432553 16.67 20. Off 10.00 3803683 12.67 20.00 16.70 3803683 11.13 10.00 13.00 475460 8.67 6.00 10.00 102699 6.67 4.00' 5.20 30429 3.47 83.00 18.80 1282826080 18.25 66.80 17.60 668748608' 17.00 40.00 16.60 143586406 16.12 20.00 15.00 179'48307 14.56 -lO.Oo 13.60 2243538 13.20 5.00 11.50 280442 11.17 4.00 10.00 143086 9.71 3.00 7.50 60576 7.28 2 .6 0 ........ 3.76 '35053 3.64 2.00 “T.07 17948 82.00 20.00 2605800414 15.38 60.00 20.00 1020833333 15.38 40.00 19.10 302460136 14.69 20.66 18.00 37808642 13.85 10.66 17.10 4726080 13.15 5.06' 14.00 590760 10.77 3.00 ..7.50 127604 ......... 5.77 80.00 18.00 2110973 7.20 60.66 16.50 517672 6.60 30.00 13.50 111795 5.40 20 .oO 10.50 33125 4.20 15.00 8.00 13074 ........... 3.20 12.00 5.60 7155 2.20 10.06 3.00 4141 1.20 9.00 2.00 3018 0.80 70.00 9.00 111960 2.70 50.00 8.00 40798 2.40 30.00 5.00 8812 1.50 20.00 3.00 2611 0.90 16.00 1.67 1102 0.50 10.00 0.60 326 0.18 able 15. Non dimensional flame spread data for concurrent-flow flame spread (C 02, He, SF6). 148 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. G ra sh of N u m b e r A ir @ 1 .0 0 atm 3 0 % 0 2 /N 2 ) @ 1.00 atm 1 — T0'% ” 0 2 /N 2 ) @ 0.50 a tm ......4 6%-------- O 2/N 2 ) @ 0.25 atm 5 u % 0 2/N 2 ) @ 0.25 atm a o % 0 2/N 2 ) @ 0.25 atm t= 2 3 o %..... 0 2/N 2 ) @ 2.00 atm 3 4 4 3 8 7 7 6 ,. 2 : g - j - 8 0 / 6 0 6 2 4.3 2 12 7 5610 7.0 6 159439 ...10TOT) 8"1 'STS" " ”f O T'O 'O 3 4 4 3 9 1 1 .67 1 0 2TT4 ~ ' 12.5"0 4 3 0 6 8.33 1 2 /6 5 :D D " .. 8360 10.00 .... 4838 8.8 0 2 4'/ / ' 8.00 1 5 7 T " 5 .8 8 8 5 0 " .... '" 2 .8 6..... 5 3 5 2.50 3 1 0 ... 2 .00 8 3 613' 8. 6/ 2 4"T7~ 4.00 18 0 6 3.89 8 50 2.14 fa 3 5" ' 1 .6 7 '" .... 4 8 5 ' 1 ,7"9 3 9 0 1.85 348 1 .92 3 10 1.50 2 / 4 1.56 24 1 ...... 1709........ 1 8 4 1.19 1 6 9........ 1 .00 1 8 4 1 . 1 9 3 1 0 1.00 6 10 1 .6 9 i 2 / 1 3TT3 " 2 4 8 2... "" .....'475 0 8 3 7 6 8 .6 7 1 9 8 5 4"' 8 ,0o 6 3 9 „ 1 25. 1 2 4 / 2.00 4'2"0"9"" 4.00 9 9 / 8 1 0 .00 7 9 8 2 1 6.5 0 10 4 8 8 8.40 6292 12 4.80 10 8 / 2 / 8 ......6.00 ..... 3 3 6 / 5 7 .3 3 2 2 2 0 3 3 4 6 2 .20 2 2 2 0 3 3'TB. 2.20 " 2 8 / / 2 .5 0 8 0 3 4 3.00 8698 6 . 6 7 2 0 6 1 8 7 .60 4 0 2 / 0 /.GO 1 3 6 y 1 u 6.67 69586'"" 6.67 322 166 6.00 10 8 7 2 7 8 S.oO 25 7 / 2 5 T " 3.75 8 U 3 4 .. .... 3109 1 1 .25 402 /0 13.60 7 8 6 5 T ..... 11.20 3 3 6 / 5 " 1 O.oo 2 I'bO TO ..... 9'. 3 7 ...3 2 2 1 5 6 " 7.00 10 8 / 2 / 8 7 .0 0 17 2 6 5 5 7 6 .43 2 8 / 7 2 5 2 r— T .1 '8 " ...... 3 6 6 9 b'6'4" 4 .6 7 5 0 3 3 6 0 6 '. “ 4 . 0 0 8 6 9822 8 2.50 Table 16. L/Wd ata for Nitrogen Diluents. 149 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. G rash of 30% 0 2 /C 0 2 45% 02/SF6 46% 02/SF6 25% 02/H e 30% 02/H e 35% 0 2 /C 0 2 50% 02/S F 6 Number @ 2.00 atm @ 2.00 atm @ 2.94 atm @ 1.00 atm @ 0.25 atm @ 1 atm @ 1.00 atm ' 220524697 2.56 86391183 3.50 25597388.. 5.25 3199674 5.00 369959 7,00 8639"! 5.83 49995 3.00 25897 2.50 10799 1.67 82T9' 2.00 1282826080 0.60 668748'SBS 1.50 143586456 1.75 17948307 3.00 2243538 6.00 280443 6.00 143585 5.00 60576’ 1.67 35055 1.00 i79'4§ 0.50 2 6 o 5 6 6 9 4 iT 0.61 1020833333 0.83 302469138 1.50 37808642 2.50 4726880 4.50 590760 6.00 127604 6.67 21-19673 2.63 517572".... 4.20 111795 5.33 33125 3.50 13974 3.33 7 - 1 5 5 ..." 3.33 4141 2.00 3018 1.11 '11-1350 4.14 40768 3.80 8812 3.33 2611 1.50 4102 1.00 328 0.50 48547035 3.66 11006042 3.20 2377305..... 4.67 704387 5.50 88048 8.00 19018 5.00 5835 5.00 262-15368-1 1.59 59432553 2.20 1283743-1 3.00 3803683 2.75 475460 4.50 102696 5.83 30429....... S'.O O Table 17. L/W data for SF6 i CO2, and He Diluents. 150 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Asset Metadata

Creator Honda, Linton Kaneki (author)

Core Title Effects of convection and radiation on flame spread over solid fuel beds

Contributor Digitized by ProQuest (provenance)

School Graduate School

Degree Doctor of Philosophy

Degree Program Aerospace and Mechanical Engineering

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag engineering, mechanical,OAI-PMH Harvest

Language English

Advisor Ronney, Paul D. (committee chair), [illegible] (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-98598

Unique identifier UC11337970

Identifier 3027724.pdf (filename),usctheses-c16-98598 (legacy record id)

Legacy Identifier 3027724.pdf

Dmrecord 98598

Document Type Dissertation

Rights Honda, Linton Kaneki

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA