SONG, GUOWEN. Modeling Thermal Protection Outfits for Fire Exposures. (Under the

Transcription

1 ABSTRACT SONG, GUOWEN. Modeling Thermal Protection Outfits for Fire Exposures. (Under the direction of Dr. Roger L. Barker and Dr. Hechmi Hamouda) In spite of high performance fibers, fabrics, and advanced test methods, much remains to be learned to enhance the technical basis for improving thermal protective performance of materials and clothing to protect against burn injuries subject to intense heat exposures. While some analytical and numerical models have been developed about these materials, these models are based on some bench top tests. No general model exists that explains heat transfer in a configuration that realistically simulates the shape of the human body. In this research, a numerical model was developed that is capable of predicting heat transfer through two thermally protective clothing materials (Kevlar/PBI and Nomex ШA) and garments exposed to intense heat environments. The thermally induced thermophysical properties of the protective fabrics and distributions of air gaps between garments and manikin will be considered in the model which simulates heat transfer through single layer protective garment worn by manikin exposed in a flash fire. The integrated generalized model developed was validated using the Pyroman Thermal Protective Clothing Analysis System. A numerical fabric-air gap-skin model has been developed to calculate the heat transfer at 122 sensors locations over the manikin body. The flash fire generated in

2 Pyroman chamber is investigated by measuring the flame temperature over each sensor and its average heat flux. An estimated method is used to calculate the overall heat transfer coefficient at each sensor locations for a 4 second exposure to an average heat flux of 2.00 cal/cm 2 sec (84 kw/m 2 ). The thermal conductivity (k) and volumetric heat capacity (ρc p ) of the protective fabrics under high heating rate and high temperature are found not constant. A parameter estimation method is used to estimate heat induced changes in fabric thermophysical properties. The air gaps distributions (between garment and the manikin) of different garment (Kevlar/PBI and Nomex ШA ) and size (coverall size 40, 42 and 44) including a Nomex ШA garment size 42 that has undergone a 4 second exposure has been assessed using a three-dimensional body scanning technology. Nomex ШA coverall air gap sizes between the garment and manikin are considered as temperature dependent for a 4 second exposure as a result of thermal induced shrinkage. The multi-layer skin model and a burn evaluation method were used to predict second and third degree skin burn damage. The established numerical model was validated by Pyroman tests using thermally protective Kevlar/PBI and Nomex ШA coveralls. The manikin tests covered exposure time from 3 seconds without underwear, 4 seconds with and without underwear, and 5 seconds with underwear. A parametric study conducted using the developed numerical model indicates the influencing parameters on garment thermal protective performance in terms of skin burn damage for a 4 second flash fire exposure. The importance of these parameters was

3 analyzed and distinguished. These parameters includes fabric thermophysical properties, the flash fire characteristics in Pyroman chamber, garment shrinkage and fit factors, as well as garment temperature and test environment. Different skin models and their influence on garment thermal protective performance prediction were also investigated using the numerical model.

4 MODELING THERMAL PROTECTION OUTFITS FOR FIRE EXPOSURES by GUOWEN SONG A thesis submitted to the Graduate Faculty of North Carolina State University in partial fulfillment of the requirements for the Degree of Doctor of Philosophy FIBER AND POLYMER SCIENCE Raleigh, 2002 APPROVED BY: Co-Chair of Advisory Committee Co-Chair of Advisory Committee

5 "We gain strength, and courage, and confidence by each experience in which we really stop to look fear in the face... we must do that which we think we cannot." ~ Eleanor Roosevelt ii

6 BIOGRAPHY The author, Guowen Song, born December 14, 1965, received a Bachelor of Science degree in Textile Chemistry in Tian Jin Polytechnic University in He worked for a Chang Chun Textile Company for three years before returning to Tian Jin Polytechnic University and got a Master of Science degree in Textile Engineering and Chemistry in Following his graduation, he joined the faculty in Tian Jin Polytechnic University. In August, 1998, after one year work in T-PACC, he enrolled in the Ph.D. program in Fiber and Polymer Science at North Carolina State University in Raleigh, North Carolina. He is married to Tu Luan. iii

7 ACKNOWLEDGEMENTS The author wishes to express his deep gratitude to Dr. Roger L. Barker for his invaluable support, trust and guidance through the entire research. Dr. Barker s instruction and encouragement, which leads to author s graduation, will also benefit author in his future work. The author acknowledges with thanks Hechimi Hamouda, co-chairmen of his Advisory Committee, for providing support and advice, and also to other members of this committee: Andrey Kuzentsov, Peter J. Hauser and Robert C. Smart. Appreciation is given to Shawn Deaton and Jim Fowler, who help to manipulate Pyroman tests and some experimental work. Also thanks to Dr. Donald Thompson, Jon Porter and Dr. B. Scruggs for their support. The author is also appreciative to his fellow student Patirop Chitrphiromsri in Mechanical and Aerospace Engineering for help in programming. The author wishes to thank to David Bruner and Mike King of Textile/Clothing Technology Corporation (TC 2 ) for help in conducting the three dimensional body scanning and James Beck and R. McMaster in Michigan State University who helped in parameter estimation method and codes. Finally, the author would like to express his sincere thanks to his wife for coming with him enduring all the hardships; his parents and his wife s parents for their patience and support during the pursuit of his education. iv

8 TABLE OF CONTENTS LIST OF TABLES... IX LIST OF FIGURES...X CHAPTER 1 INTRODUCTION RESEARCH MOTIVATION GOALS AND OBJECTIVES APPROACH... 2 CHAPTER 2 A LITERATURE REVIEW PROTECTIVE CLOTHING AND TEST METHODS Protective Clothing Flame-resistant Fibers Suitable for Protective Clothing Test Methods Measuring the Thermal Protective Performance of Fabric Assessment of Protective Performance Using TPP Methods MANIKIN TESTING AND NCSU PYROMAN NCSU s Pyroman System Manikin Research on Garment Thermal Protective Performance FABRIC PROPERTIES AND FIRE CHARACTERISTICS Fabric Optical Properties in a Flash Fire Environment Fabric Thermal Properties in Flash Fire Condition Thermal Conductivity Heat Capacity and Thermal Decomposition Temperature Effects of High Heat Exposure on Fabric Dimensions Characterization of the Fire Environment MODLES FOR PREDICTING SKIN BURN INJURY Skin Burn Models Bioheat Transfer Models The Chen and Holmes Model The Weinbaum, Jiji and Lemos Model Skin Burn Models v

9 2.5 MODELING HEAT TRANSFER IN PROTECTIVE FABRICS Heat Transfer Models Torvi Model-- Fabric-Air Gap-Test Sensor Model Gibson Model -- Multiphase Heat and Mass Transfer Model Mell and Lawson Model Modeling Thermal Degradation in Fabrics CHAPTER 3 MODEL THE PYROMAN SYSTEM A NUMERICAL HEAT TRANSFER MODEL HEAT TRANSFER IN FABRIC, AIR GAP AND SKIN HEAT TRANSFER IN SKIN MODEL AND BURN EVALUATION FINITE DIFFERENCE METHOD CHAPTER 4 EXPERIMENTAL STUDIES CHARACTERIZING THE PYROMAN THERMAL ENVIRONMENT Heat Flux Distribution Heat Transfer Coefficient Determination CHARACTERIZING HEAT INDUCED CHANGE IN FABRIC PROPERTIES Parameter Estimation Method Estimation of Protective Fabrics Thermal Properties PROTECTIVE GARMENTS AIR GAP DISTRIBUTION IN PYROMAN BODY Three-Dimensional Body Scanning Technology Air Gaps Determination of Protective Garments in Pyroman Ease Measurement Method CHAPTER 5 NUMERICAL RESULTS AND MODEL EVALUATION GARMENTS USED IN THIS STUDY Garments Preparation Garments Fabric Thickness Garment Fabric Thermal Properties Summary of Garments and Fabric Properties Used in the Model FIRE BOUNDARY CONDITIONS GARMENTS AIR GAP SIZE DETERMINATION MODEL RESULTS AND PREDICTIONS MODEL EVALUATION vi

10 One Layer Garments without Underwear One Layer Garments with Underwear Model Evaluation Summary CHAPTER 6 PARAMETRIC STUDY INFLUENCE OF FABRIC THERMOPHYSICAL PROPERTIES Fabric Thickness Thermal Conductivity Volumetric Heat Capacity Emissivity Transmissivity INITIAL, AMBIENT AND FIRE DISTRIBUTION INFLUENCE Fabric Initial Temperature Ambient Temperature Same Garment and Ambient Temperature Fire Distribution influence GARMENT DESIGN AND FIT FACTORS Garment Components Shrinkage and Its Temperature Effect on Protective Performance Garment Size SKIN MODEL INFLUENCE Blood Perfusion Temperature Distribution in the skin Single Layer and Multi Layers Skin Model Comparison CHAPTER 7 CONCLUSIONS AND RECOMMENDATIONS SUMMARY RECOMMENDATIONS AUTHOR S NOTE REFERENCES APPENDICES APPENDIX 1 EXPERIMENTAL APPARATUS USED IN THIS RESEARCH TEMPERATURE MEASUREMENT DEVICE DATA ACQUISITION SYSTEM vii

11 HEAT FLUX SENSORS SENSOR CALIBRATION HEAT SOURCE APPENDIX 2 AVERAGE HEAT FLUX VALUES AS MODEL INPUT (HEAT FLUX: CAL/CM 2 SEC). 169 APPENDIX 4. NORMAL DISTRIBUTION APPENDIX 5. GARMENT FABRIC COMPRESSION TEST APPENDIX 6. 3D BODY SCANNING TECHNOLOGY APPENDIX 8. GARMENT EASE MEASUREMENTS APPENDIX 11 ESTIMATED HEAT TRANSFER COEFFICIENT APPENDIX 12 MODEL RESULTS OF KEVLAR/PBI COVERALL FOR 4 SECOND EXPOSURE APPENDIX 13 MODEL RESULTS OF NOMEX ШA COVERALL FOR 4 SECOND EXPOSURE206 APPENDIX 14 MODEL PREDICTIONS AND MANIKIN TESTS viii

12 List of Tables Table 2-1. Fabric properties [31] Table 2-2. Fabric optical properties [32] Table 2-3. Thermal Decomposition Temperature Table 2-4. Values for use in Equations for Nomex IIIA and Kevlar /PBI Fabrics Table 2-5. Values of P and E Used in Burn Integral Calculations Table 3-1. Human Skin Properties in Skin Model [91] Table 5-1. Test Garments Style and Fabric Property Table 5-2. Thermal Properties of Cotton Underwear Table 5-3. Calibrated Heat Flux Values for Pyroman Before and After Burns Table 5-4. Parameters Used in Model Predictions Table 6-1. Range of Values of Thermophysical Properties of Garment Fabric Table 6-2. Model Parameters for Garment Thickness Study Table 6-3. Model Parameters for Fabric Conductivity Study Table 6-4. Model Parameters for Fabric Volumetric Heat Capacity Study Table 6-5. Numerical Model Setup Table 6-6. Model Parameters for Fabric Transmissivity Study Table 6-7. Fabric Transmissivity Influence on Time to Body Burn Predicted by Model Table 6-8. Numerical Model Setup Table 6-9. Model Parameters for Heat Flux Distribution Study Table Numerical Model Setup Table Numerical Model Setup Table Model Parameters for Blood Perfusion Study Table Blood Perfusion Influence on Time to 2 nd and 3 rd Burn Table Numerical Model Parameters for Skin Model Study Table Numerical Model Parameters for Skin Model Study Table Thermal Properties of Skin Models ix

13 List of Figures Figure 2-1. Schematic Diagram of TPP Tester... 8 Figure 2-2. Close up View of TPP Tester in NCSU... 9 Figure 2-3. Temperature rise ( C ) in Skin Stimulant Figure 2-4. TPP Traces of the Test Fabrics [21] Figure 2-5. Pyroman Fire Chamber and Gas Delivery System Figure 2-6. Pyroman Manikin Figure 2-7. Nude Pyroman and Its Burning Figure 2-8. Pyroman Burn Prediction in terms of 2 nd Burn and 3 rd Burn Figure 2-9. Predicted Body Burn vs. Heat Exposure Intensity [25] Figure Predicted % Burn Injury after 5 s Exposure Figure Fabric Thermal Conductivity vs. Temperature Figure Three Phases Present in Hygroscopic Porous Media Figure 3-1. Elements of Heat Transfer and Burn Evaluation Figure 3-2. Elements of Fabric Air-Gap Skin Model Figure 3-3. Schematic for one-dimensional Heat Transfer Model Figure 3-4. Fabric-air gap-fabric-skin Model Figure 4-1. Specially Designed Sensor Used to Measure Flame Temperature and Heat Flux Figure 4-2. LabVIEW Data Acquisition System and Software Figure 4-3. Sensor Numbers and Distribution over Manikin Body Figure 4-4. Histogram and Cumulative Curve of Heat Fluxes of 122 Sensors Figure 4-5. The Scatterplot of Heat Fluxes vs. Normal Scores From a Calibration Figure 4-6. Heat Flux Distribution with Different Standard Deviation Figure 4-7. Heat Flux Distribution in a Manikin for a 4 second Exposure Figure 4-8. Heat Transfer Coefficient Estimation from Temperature Figure 4-9. Pyrocal Sensor Flame Temperatures and Copper Temperatures Figure Estimated Heat Transfer Coefficient Using Pyrocal Sensor Figure Flame Temperature and Heat Transfer Coefficient Figure Sensor Location Factor and Its Heat Transfer Coefficient x

14 Figure Schematic Diagram of the Transient Experiment for Parameter Estimation. 77 Figure Temperatures and Heat Flux Profiles during Exposure Experiment Figure Estimated Transient Thermal Properties of Kevlar/PBI during 4 Second Exposure Figure Air Gap Determinations by Superimposing Dressed and Nude Figure Dressed Pyroman 3D Body Scanning Image Figure Superimposed 3D Body Scanning Data Showing the Sensor Positions Figure Slicing the Body at Specific Sensor Position, Measuring Air Gap Figure Nude and Dressed Pyroman 3D Body Measurements Image Figure Air Gap Distribution between Garment and Manikin at the Typical Positions 84 Figure Different Garment Sizes and Their Different Air Gap in Pyroman Body Figure Different Garments with Same Size and Pattern Shows Different Air Gap Figure Comparisons of Air Gaps Before and After 4sec Exposure Figure Air Gap Distribution of Different Size Garments Dressed in Pyroman Figure 5-1. Kevlar/PBI Garment Fabric Thickness as a Function of Applied Load Figure 5-2. Nomex ШA Garment Fabric Thickness as a Function of Applied Load Figure 5-3. Sensor Flux Values on Different Calibration Days Figure 5-4. Temperature Distribution in Fabric Air-gap Skin Model during 4 Second Figure 5-5. Temperature Distribution in Fabric Air-gap Skin Model during 4 Second Figure 5-6. Temperature Profiles in Skin Model for a 4 Second Exposure Figure 5-7. Omega Integral Value ( Sensor #60) in Pyroman for a 4 Second Exposure Figure 5-8. Omega Integral Value after (Sensor #56) in Pyroman for a 4 Second Exposure 98 Figure 5-9. Temperature History in Skin with Underwear in 4 second Exposure Figure Omega Value with and without Underwear in 4 second Exposure Figure Comparison between Kevlar/PBI Garment Manikin Test and Figure Second and Third Degree Burn Location of Pyroman Test for Figure Second and Third Degree Burn Location Predicted by Numerical Model for 102 Figure Second and Third Burn Location of Pyroman Test for Figure Second and Third Burn Location Predicted by Numerical Model for Figure Comparison between Nomex ШA Garment Manikin Test and Figure Second and Third Burn Location of Pyroman Test for Nomex ШA Coverall Figure Second and Third Burn Location Predicted by Numerical Model for Figure Second and Third Burn Location of Pyroman Test for xi

15 Figure Second and Third Burn Location Predicted by Numerical Model for Figure Comparison between Kevlar/PBI Garment Manikin Test and Figure Comparison between Nomex ШA Garment Manikin Test and Figure Manikin Tests and Numerical Model Results for Kevlar/PBI Figure Manikin Tests and Numerical Model Results for Nomex ШA Figure 6-1. Relationship between Burn Damage Predicted by Figure 6-2. Garment Fabric Thermal Conductivity and Predicted Figure 6-3. Effect of Fabric Volumetric Heat Capacity and Predicted Thermal Protection116 Figure 6-4. Effect of Fabric Emissivity on Garment Protective Predictions Figure 6-5. Effect of Garment Initial Temperature on Body Burn Predictions Figure 6-6. Ambient Temperature Effect on Garment Thermal Protective Prediction Figure 6-7. Garments and Ambient Temperature Influence on Body Burn Prediction Figure 6-8. Pyroman Heat Flux Standard Deviation Effect on Burn Prediction Figure 6-9. Nomex Deluxe Protective Coverall Figure Effects of Garments Components and Body Burn Prediction Figure Influence of Shrinkage during Exposure on Burn Prediction Figure Fabric Shrinkage Temperatures and Protective Prediction Figure Effect of Garment Size and Thermal Protective Prediction Figure Effects of Skin Model Initial Temperature Distribution and Figure Single Layer and Multi-layer Skin Model and Burn Prediction xii

16 CHAPTER 1 INTRODUCTION 1.1 Research Motivation There is inadequate fundamental understanding of the thermal mechanisms associated with the thermal protective performance of materials and garments in flash fire environments. Furthermore, no generally applicable model exists to explain heat transfer and thermal degradation processes and the influence of flash fire conditions. Nor is there a basic understanding of the effects of garment design and fit on thermal protection, especially in configurations that realistically simulate the shape of the human body. Therefore, a reliable model is needed to account for these factors in predicting the thermal protective performance of garments in realistic conditions. A considerable amount of research has been conducted on modeling heat transfer and predicting skin burn injury as a result of heat transfer through fabric layers. Intense heat exposures induce fundamental changes in fabric thermal, optical and spatial properties through pyrolysis, char formation, and shrinkage. These exposures produce nonlinear transient thermal physical properties and fabric optical characteristics [1, 2]. Consequently, heat transfer models cannot accurately predict protective clothing performance without considering the complex dynamics that significantly alter the properties of clothing materials exposed to high thermal exposures. No generalized model for heat transfer through protective garment layers, exposed in a realistic fire configuration, can yield meaningful results without including these data. This research developed numerical model capable of predicting heat transfer through clothing materials and garments exposed to intense heat environments. The primary objective is to construct a model to simulate heat transfer through single layer 1

17 protective garments worn by manikin exposed to flash fire. This approach is intended to provide foundation for a model that will consider the effects of heat exposure on the transient thermophysical properties of protective materials, as well as the air gap distribution between the fabric and skin, and characteristics of the heat source. The integrated generalized model developed was validated using the Pyroman Thermal Protective Clothing Analysis System [3]. 1.2 Goals and Objectives The goal of this research is to develop an understanding of heat transfer in protective garments exposed to intense fire environment, and to establish systematic basis for engineering materials and garments for optimum thermal protective performance. Specific objectives are to characterize the flash fire environment produced in the Pyroman test; to generate a database of protective fabric transient thermophysical properties exposed in intense heat as well as flash fire; to develop a model to predict performance of protective garments under intense thermal environments in terms of burn injury; to develop a fundamental understanding of mechanisms of transient heat transfer through protective garments exposed extreme thermal environments. The research advances the science of thermal modeling and contributes to the development of improved thermal protective materials and clothing. The generalized model can be applied to improvement the safety and comfort performance of protective materials, fabrics and ensembles. 1.3 Approach This research developed a heretofore unavailable database of protective fabric thermophysical properties and high temperature heat source characteristics. Effects of fabric moisture content and shrinkage will be considered by this model. This model will be used to predict burn injury in human body by calculating of heat transfer in protective garments. 2

18 The characteristics flash fire generated in the Pyroman test chamber will be examined by measuring the flame temperature above each of 122 manikin sensors and corresponding heat flux history. The heat transfer coefficient for each of 122 sensors will be determined by these measurements. The heat flux distribution of 122 sensors during 4 second exposure is statistically investigated. All the heat flux histories of 122 sensors of calibration nude burn before the garment test are used as one of the input of the numerical model. This research will measure the thermophysical properties of commonly used protective fabrics including their thermal conductivity and volumetric heat capacity; these unique data will be used in the heat transfer model to improve the accuracy of skin burn prediction. A parameter estimation approach will be used to estimate fabric thermophysical properties from dynamic thermal exposure experiments. This parameter estimation method used will treat measurement and model errors in a statistical context in order to provide means for estimating the temperature dependent thermophysical properties of fabrics. An important consideration will be the effect of heat induced shrinkage. This is important because shrinkage reduces the air space between the fabric and skin, and increases layer-to-layer contact, therefore producing changes in heat transfer efficiency. Since the air gap between the fabric and skin is important to garment protective performance. This research will use three-dimensional (3D) body scanning technology to measure the actual air gap distribution between garments and the Pyroman instrumented manikin body. The air gap distributions of different size garments including one after exposed to 4 second burning will be analyzed using three-dimensional body scanning technology. A one dimensional finite difference formulation will be developed and used to provide a numerical solution for the coupled system of differential equations that control heat transfer by radiation, convection, and conduction and the thermal response of skin. 3

19 An iteration technique will be utilized to accommodate nonlinearities in the governing equations that occur as a result of the dependence of fabric thermal conductivity and volumetric heat capacity on temperature and moisture content. The convergence of iterative relative to controlling equations will be insured by purposely linearizing source terms and by using an appropriate overrelaxation or underrelaxation parameters. 4

20 CHAPTER 2 A LITERATURE REVIEW 2.1 Protective Clothing and Test Methods Protective Clothing The purpose of thermal protective clothing is to reduce the rate of heat build up in human skin in order to provide time for the wearer to react, and avoid or minimize skin burn injury [4]. This is accomplished by using a garment which is both flame resistant and thermally insulating. The protective garment must also meet other requirements: it must maintain integrity during the heat exposure, and be liquid-repellent. Fibers used in the fabric or composite should be non-melting and flame resistant. Toxic gases should not be emitted at high temperature. The fibers should resist shrinkage and maintain strength and flexibility at high temperatures. Low thermal conductivity is needed to reduce heat transfer to underlying skin. Fabrics should not break or split open when exposed to flames. They should have low air permeability to minimize convective heat transfer. The protective garments themselves should be designed with closures and waistbands to reduce chimney effects possible during exposures. Ease of maintenance and proper fit must be considered. One of the most important functions for protective garments is to limit the amount of heat stress to the wearer, and not hinder normal work. This is can be challenging since severe changes meant to increase protection often come at the expense of garment comfort Flame-resistant Fibers Suitable for Protective Clothing Flame-resistant fibers can be divided into two classes: inherently flame-retardant fibers and chemically modified fibers and fabrics. Inherently flame resistant fibers are those in which the flame resistant properties are built into the polymer or fiber structure, such as aramid, modacrylic, polybenzimidazole (PBI), semi-carbon, and phenolic fibers. 5

21 The molecule chains of heat resistant fibers have a stiff backbone due to the aromatic groups which limit bond rotation, thus resulting in high decomposition and melting temperatures and low thermal shrinkage [8]. Nomex is the best-known meta-aramid fiber, developed for protective clothing used by military personnel, astronauts, and others working in specialized industrial applications [6]. The Nomex was introduced by DuPont in A similar meta-aramid fiber was introduced by Teijin in 1972 as Conex. While aramid fibers contain no FR chemical elements such as phosphorus or halogen, their chemical structure is such that they do not easily break down into combustible molecular fragments. They produce relatively little smoke when heated. However Nomex may shrink and break open under intense heat. For this reason, para-aramid fibers, Kevlar, are often blended with Nomex to reduce high heat shrinkage. An example of Nomex III, a blend of 95% regular Nomex with 5% Kevlar, is widely used commercial available protective fabric that uses this percentage. Polybenzimidazole (PBI) fiber, produced by Hoechst-Celanese, resists high temperatures and chemicals and is reported to have excellent textile characteristics [7]. These fabrics generally provide better fire protection than aramid fibers while remaining flexible, integrity, with no afterglow and shrinkage [8]. Normally flammable fibers, such as cotton and wool, can be treated to make them to suitable for use in thermal protective garments. Proban 210, which is phosphorus- and nitrogen-containing flame-retardant, is one of the most successful cellulose fire retardant finishes [9]. It self-polymerizes to form a three-dimensional-network polymer within the microvoids found in cotton fibers. Curing is followed by a post-oxidative treatment, which raises the phosphorus to the more stable oxidation state. Wool has a relatively high LOI (Limiting Oxygen Index) and high combustion temperature. Titanium and zirconium complexes are effective flame-retardants for wool. 6

22 Test Methods Fabric flammability is often characterized by ease of ignition, rate and extent of flame spread, and by the amount of heat evolved in burning. Other factors often considered are the duration of flaming, temperature of the burning fabric, and ease of flame extinction. Quantification of these properties depends on the nature of the ignition source, orientation of the test fabric, location of the ignition (top, bottom, edge, or face), and environmental conditions. Moisture and ambient temperature are important, as well as the air movement in the test sample [10]. The 45 0 test method (ASTM D [11]): This test uses a 50 mm by 165 mm specimen held in place at 45 0 to the horizontal. A standard burner flame is applied to the upper surface near the lower end for 1 sec. The time for the flame to travel 127 mm is then recorded. This test can only eliminate dangerously flammable fabrics as it fails to ignite most fabrics. However, it is very reliable in its ability to eliminate these dangerous flammable textiles. Vertical tests (ASTM D 3659 [12]): This test method is a more stringent measure of fabric flammability than the 45 0 test method. In this test a sample is exposed to a vertical flame for a given length of time. Then the char length, afterflame, and afterglow times are measured. The reliability of individual tests varies considerably. The Limiting Oxygen Index (LOI) (ASTM D 2863 [13]): The Limiting Oxygen Index is the lowest concentration of oxygen necessary to support combustion in fabric samples. This test is conducted by igniting the top of a vertically oriented sample with a hydrogen flame. In this test, the flame itself does not assist in burning the test fabric; therefore, repeatable results are easily obtained. Following fabric igniting, the ratio of oxygen and nitrogen are adjusted until the sample is completely consumed at a slow and steady rate. The LOI entrusted as the minimum fraction of oxygen in the chamber respond to maintain slow and steady combustion. 7

23 Measuring the Thermal Protective Performance of Fabric Several methods have been developed to measure the insulative properties of fabrics against high-intensity thermal energy. Heat sources range from radiant panels to gas burners or a combination of two, depending on the type of fire simulated. Heat sensors range from copper disk to skin simulant sensors which the thermal inertia of the sensor material is almost the same of human skin. Two thermal protective performance (TPP) test methods have received wide application by different associations and standards organizations. One method is ASTM D , which uses a single laboratory gas burner as the heat source. The other procedure is a more versatile method that combines two gas burners and quartzs heaters to provide different mixtures of radiant and convective heat. Typical TPP experimental arrangements use a methane gas flame in combination with a bank of quartz tubes to provide a convective and radiant heat source. Sensor block To data acquisition system Sensor Spacer Fabric specimen Water cooled shutter Meker burner Quartz Tube Bank Figure 2-1. Schematic Diagram of TPP Tester 8

24 Figure 2-2. Close up View of TPP Tester in NCSU The fabric specimen is mounted 50 mm above the burner top (Figure 2-1 and Figure 2-2). A pneumatic and water cooled shutter controls the duration of exposure and comes front to end the exposure. The heat transferred through the fabric is measured by a copper calorimeter sensor painted with flat black. The distance between the fabric and the calorimeter heat sensor can be varied using spacer plates of different thickness. The TPP test is a convenient, precise and a relatively inexpensive means for comparative the thermal protective performance of fabrics. However, TPP tests are limited to a certain heat exposures, and to the configuration of test fabric. They provide no information about the spatial effects which may be important in predicting the protective performance of clothing worn on the human body. They also provide no information on the effects of garment design and construction, and the role of seams, closures, pockets, or vents to thermal protective performance by clothing in actual wear. 9

25 Assessment of Protective Performance Using TPP Methods Barker and Shalev [14, 15] have reported extensively on the use of TPP test to understand the thermal response of fabrics in high heat exposures. In these studies, the fiber type is shown to have a significant effect on heat transfer, contrary to the behavior observed in low heat flux condition where fabric structure dominates. They point out that thermal physical properties change greatly during a TPP exposure. Therefore, retention of thermal properties, not the initial values of these properties, is the key to heat transfer in TPP tests. They found that the air/fiber ratio of the fabric and the maintenance of air volume in fabric structure are important to thermal protective performance. Air and fiber conduction dominate in intense exposures; direct radiation transmission, as the fabrics were relatively opaque, is not as important. Barker and Shalev show fabric air permeability does not correlate with TPP test results; hot gases do not appear to blow through the fabric. Their work indicates that the ability of a fabric to maintain a profusion of surface fiber is important in convective exposures, as fibrils on the surface of the fabric function as baffles, holding still air and extending the boundary layer on the fabric, hence increasing heat transfer resistance. Their research also shows that moisture plays an important role in determining thermal transfer in TPP tests. Moisture in fibers can increase the volumetric heat capacity of and provide an ablative effect, thus increasing thermal protective performance. Barker and Shalev s experiments used a 50% convection and 50 % radiation heat source. Barker and Lee [16, 17] analyzed the TPP of single layer fabrics using different heat source combinations and intensity levels. They show that, for all the fabrics, the times to exceed the Stoll criterion are lowest in 100% radiation exposures. Except for the 0.48kW/m2 exposures, fabrics insulate best against a 50% radiation and 50% convection. This finding is attributed to the effect of protruding fibers on thermal insulation. They assume that stagnant air layers entrapped by surface fibers play an important part in heat transfer especially in convective heat exposure. On the other hand surface fibers are expected to play a less important role in purely radiant exposures [18]. Thermal properties change of different fabrics was presented with time for different exposures. 10

26 The effects of different test conditions and fabric properties on fabric properties on test results were also discussed. Stoll et al [19] studied the effect of heat exposure duration and the thickness of the air space between the thermal sensor and the test fabric. She found that, for a given exposure and duration, the air gap increases the insulation offered by the fabric air assembly. The increased insulation of the fabric air gap assembly slows heat transfer through it. For a fabric with weight of 3oz/yd 2, the optimum air gap thickness for 1 to 3 seconds exposures is approximated 4 mm. Further increase in air thickness results in rapid heat transfer to the sensor by heat convection through the entrapped air layer Cotton Nomex PBI 15 Temperature rise (deg C) mm 1 mm 2 mm Air space thickness (mm) Figure 2-3. Temperature rise ( C ) in Skin Stimulant of Naval Materials Laboratory after 3- second Flame Exposure [19] Freeston [20] used a skin simulant heat sensor in his investigation of the effect an air gap between the two layers of fabrics. His results show that heat transfer declined as the air gap between two fabric layers increased from 0 to 2 mm for each of three fabric specimens viz. cotton, Nomex and PBI (Figure 2-3). 11

27 FR COTTON # 34 FR COTTON # 68 NOMEX KEVLAR PBI P/K Temperature Rise (deg. C) Time (s) Figure 2-4. TPP Traces of the Test Fabrics [21] The heat transfer response of different types of fabrics exposed 2.0 cal/cm 2 sec is shown in Figure 2-4 [21]. Most fabrics exhibit an exponential response as the rate of heat transfer increases with fabric pyrolyzes in intense heat. Cotton fabrics, treated with different FR finishes show fundamentally calorimetric curves. In the case of cottons treated with a phosphonium FR finish, a relatively rapid deposition of non combustible condensate on the calorimeter occurs when the critical volatilization temperature for the applied flame retardant is reached. For cottons treated with halogenated FR finish, flame regression due to radical scavenging reaction steadily interferes with burning as antimony halide species evolve. Catastrophic break down is observed in cotton materials treated with halogenated finish. FR wool produces heat transfer response characterized by constant slope, similar to that observed in inorganic fibers. 12

28 2.2 Manikin Testing and NCSU Pyroman Garments burning behavior and thermal protective performance cannot be fully predicted from bench-scale fabric testing. Manikin tests provide more realistic evaluation of protective performance evaluation since they measure the influence, not only of the thermal physical properties of fiber and fabrics, but also of design and fit of garments themselves. The development of full scale manikin tests was a significant step toward the realistic evaluation of garment protective performance since they provide a realistic simulation of real fire exposures. In 1940 s Baker and Smith used the first manikin to compare the burning rate of shirts. Colebrook used non-instrumented manikins to test garment in a wire body form. Non-instrumented manikins continued to be used extensively in assessing garment flammability [22]. One of the first instrumented manikins was used in 1962, when Stoll conducted tests for the United States Navy that analyzed a leather-covered manikin quipped with temperature detector paper and melting point indicators [23]. By 1972, these efforts had involved into a full-scale instrumented manikin tests that used burning aircraft jet fuel as a heat source. DuPont developed this technology and built an instrumented manikin called Thermo-man NCSU s Pyroman System Pyroman, located in College of Textile, North Carolina State University, is one of few manikins in the world actively involved in garment protective performance research testing. This system was recently upgraded to include a new sensor technology and data acquisition system, as well as a new software package. The Pyroman Thermal Protective Clothing Analysis System [24] consists of a number of integrated components, designed to work together to measure the performance of protective clothing under full scale, flash fire exposure conditions. Figure 5 shows a diagram of the system. 13

29 Gas Supply System: Propane gas is supplied to the burner system from a buried tank through a series of valves and reducers. Pressure sensitive switches monitor the system to maintain safe operating conditions. Electrically controlled valves prevent supply of high-pressure gas for the test exposure unless all of the safety devices are satisfied and the test is ready to be run. The gas supply line is insulated and electrically heated to provide for a constant supply of fuel throughout changes in the weather. The gas supply line is also vented through solenoid valves, which are open, when the system is not in use. Figure 2-5. Pyroman Fire Chamber and Gas Delivery System Fire Chamber: The instrumented manikin and the exposure system are housed in a flame resistant room (Figure 2-5) with large viewing windows on one wall and double entrance doors on the opposite wall. The fire chamber is provided with supply and exhaust ducts and fans, which are automatically controlled to provide safe startup of the 14

30 system and rapid removal of the products of combustion and degradation after a test exposure. The speed of these fans is controlled to permit testing under wind conditions with velocities up to about 5 miles per hour. Flash Fire Exposure System - Burners and Control Panel: The most important requirements of the flash fire system are safe operation and reproducibility. Eight industrial burners, which have been modified, produce the flash fire and are carefully positioned to create a large volume of fire, which fully engulfs the manikin. Each burner has a pilot flame which is lighted and proven before the gas is supplied to the torch. The gas control panel monitors the state of each pilot flame and prevents opening of the exposure torch valve if there is no pilot flame present. This feature provides both safety and control over the position and number of torches used in each test. The gas control panel also monitors the condition of the gas supply line and safety devices and will shut the system down and vent the gas in the supply line in case of a malfunction. Figure 2-6. Pyroman Manikin 15

31 Manikin: The test manikin is a size 40 regular male, made from a flame resistant polyester resin reinforced with fiber glass (Figure 2-6 and Figure 2-7). There are sockets for 122 heat sensors, which are uniformly distributed on the surface. Leads from each sensor are taken to the data acquisition unit through a guarded, heat shielded cable. The manikin is suspended from the ceiling of the burn chamber on an adjustable fixture. Computer System: A sophisticated computer system is used to control the National Instruments data acquisition system, acquire data from the sensor system and to calculate and display the results of the numerical model used to estimate skin damage. The output from the 122 sensors is fed into conditioned analog amplifier multiplexer. This amplified output is then fed to a 12 bit resolution DAQ board, which transfers the analog signal to digital signal with sampling rate of 1.25MS/s. At this speed, each of the 122 sensors can be read more 10 times per second. The code width for the range 0-10V input is 2.4 mv. Figure 2-7. Nude Pyroman and Its Burning The garment testing protocol sequence includes dressing the manikin with the garment, interacting with the computer to assure safe conditions, lighting the pilot flames, 16

32 exposing the garment to the flash fire, acquiring data, and running the fans to vent the chamber. The data acquired by the system is used to calculate the incident heat flux and to predict burn injury for each sensor. The computer control system will compute the received temperature profiles of each sensor, translated into heat flux history of total data acquisition time. These heat fluxes apply to skin model to calculate temperature change at the certain depths in skin model. Figure 2-8. Pyroman Burn Prediction in terms of 2 nd Burn and 3 rd Burn The calculated incident heat flux is used to calculate the temperature of human tissue at two depths below the surface of the skin, one representing second degree and the other representing third degree burn injury. The prediction results can be obtained using burn evaluation model which programmed in the computer. Figure 2-8 shows a burn prediction in terms of second and third degree burn. 17

33 Manikin Research on Garment Thermal Protective Performance Dale [25] compares protective garments at different heat exposure levels. The results of his experiments, shown in Figure 2-9, indicates that intrinsically heat resistant materials like Nomex IIIA and blend of Kevlar and PBI show a steady rise in Nom ex III (42R-210 g/m2) Kevlar/PBI (44R-187 g/m2) Proban FR7-A (42R-307 g/m2) Proban Indura (42R-334 g/m2) % Body Burn Thermal Energy Level (cal/sq.cm) Figure 2-9. Predicted Body Burn vs. Heat Exposure Intensity [25] predicted body burn injuries with increasing level of input thermal energy, while FR cotton materials show a sudden rise in predicted body burn injuries at an exposure level of thermal energy in between cal/cm 2 ( kj/m 2 ), depending upon fabric mass per unit area. The performance of FR cotton results from lower thermal degradation temperature, and thermal chemical reactions that occur when heat exposure energies produce this degradation in these materials. Behnke et al [26] investigated the protective performance of garments made of fabrics with different thermophysical properties. They selected Kevlar, Nomex IIIA, Proban cotton and Zirpro wool for testing on "Thermo-man". These materials represent the range of anticipated fabric reactions to intense thermal exposures. Proban Cotton 18

34 and Kevlar were selected for their dimensional stability while Nomex IIIA and Zirpro wool were selected due to their relatively higher shrinkage with high temperature exposure. All of these test specimens were woven fabrics with densities in the range of g/m2 ( oz/yd 2 ). These fabrics were tested as size 42 regular single coveralls. In these studies the data from unprotected heat sensors, located in the head of the manikin were not included since the coverall did not cover the manikin head. Figure 2-10 shows the predicted burn injury after 5 sec. exposures on "Thermo-Man". Crown [27] studied different garment systems with different underwear and compared with the TPP test. These tests show that much less burn occurred with FR Aramid underwear comparing with non FR cotton. Barker et al, using the Pyroman Burning Evaluation System, investigated the effects of protective garment size fitted in the manikin. They tested three different size coverall made of Nomex and Kevlar/PBI [28]. These sizes are 40, 42R and 44 (about 3% size differences). For Nomex III, where the thermal average shrinkage is 10 to 15%, size 42R offers maximum protections in term of burn injure percentage. Data on burn injury percentage in case of coveralls made of Kevlar/PBI show that the garment size does not affect the level of protection. Pawar and Barker [1] related manikin-burning prediction with TPP tests in both contact and spaced configurations. They notes that some sensor positions in manikin test are similar to TPP contact configuration, which is comparable to TPP results, and other sensor positions show the results that are comparable to TPP results with spaced configuration. 19

35 nd degree burn (%) 3rd degree burn (%) Total (%) Nomex III 100% Kevlar Proban cotton Zirpro wool Figure Predicted % Burn Injury after 5 s Exposure and heat flux of 2.0 cal/cm 2 sec on "Thermo-Man " Behnke and Barker compared stationary manikin tests results with a dynamic Thermal-Leg evaluation system [26]. The Thermal-leg evaluation system assess the ability of clothing materials to protect the wearer in realistic simulations of running motions similar to a victim escaping a flash fire accident. These novel experiments show that, in a dynamic configuration, the ability to retain strength and structural integrity of the protective in prolonged exposure to flames is much more important in terms of protection. In this research, Nomex III and Kevlar 100 fabrics appear to be superior to FR cotton and FR wool samples. 20

36 2.3 Fabric Properties and Fire Characteristics Fabric Optical Properties in a Flash Fire Environment Fabric optical properties play an important role in determining garment protective performance, especially when exposed to intense fire environments. The flash fire generated in Pyroman that is created by burning liquid propane through eight burners, thus providing turbulent jet flame. The impinging fire is usually considered a convective heat source, but at high flame temperatures, a significant amount of the thermal energy is transferred by thermal radiation. Since a substantial part of the thermal energy is radiant heat, the optical properties of the fabric can affect the heat absorption characteristics of the protective garment. Proper treatment of the optical properties during the burning process is, therefore, crucial to developing a model capable of accurately predicting protective performance of clothing. Polymeric materials are highly absorbing at wavelengths greater than 3µm. Spectral absorbance is lower at wavelength range shorter than 3µm. Dark color fabrics are 60%--70% reflective at wavelength in the near infra-red [21]. Almost all fabrics, regardless of their color, reflect poorly in the far infra-red and the ultraviolet. Light colored fabrics reflect well in the visible range of the electromagnetic spectrum. On the other hand, optical transmittance is largely related to fabric porosity [48]. Hence, in order to calculate how much energy is transmitted and absorbed by any given fabric system, it is necessary to know the spectral distribution of the energy source and the spectral reflectance and transmittance of the fabric [29]. Morse et al. [31] measured spectral reflectance and transmittance using a Gier- Dunkle integrating sphere reflectometer with a Beckman DK-2A spectrophotometer between 0.5 µm and 2.5 µm. This method enabled measurement of energy in all directions and yields integrated values for each wavelength. Morse also used high and low reflectance backing to the fabric and a reflectometer to determine the directional 21

37 reflectance and transmittance. Reflectance beyond 2.5 µm was found to be low for all samples. Table 2-1 shows that fabrics optical properties change after exposure to heat. Table 2-1. Fabric properties [31] Property Nomex S-PBI 121 Weight (gm/cm 2 ) Thickness (cm) Moisture regain Specific Heat (cal/gm. 0 C) (5 0 C to C) ρ row α τ Optical Properties ρ ( C B.B.) Charred α τ % max from 40% max from 2% at C Linear Shrinkage 300 to C 400 to C to 19% at C ρ - Reflectivity, α - Absorptivity, τ - Transmissivity Table 2-2. Fabric optical properties [32] Fabric Transmittance % Reflectance % Absorptance % NomexШ PBI S-PBI Ross [32] measured the optical properties of several fabrics using the Beckman DK-2A reflectometer with a xenon source (Table 2-2). 22

38 Backer et al. [33] also measured the optical properties of fabrics, both before exposure to heat and in charred states using the integrating sphere and opaque fabric backing. Other treatments of radiant transmission through fabric consider the absorption of incident radiation [37]. These models use Beer s law with an extinction coefficient and measure using transmissivity measurements in the infrared region. An extinction coefficient for fabrics of interest can be calculated as: γ = ln(τ ) L fab Torvi uses a Nicolet Fourier Transform Infrared (FTIR) spectrometer to measure transmissivity in Nomex IIIA and Kevlar /PBI fabrics samples before and after 10 sec exposure to a heat flux of approximately 80 KW/m 2 using. Torvi shows no significant changes to the transmissivities of these fabrics before and after burning. Aluminization is a common method used to change fabric optical properties. Aluminization can increase the reflectance of protective fabric up to 90 percent, a benefit in purely radiant thermal exposures [34]. Aluminization, however, is detrimental in a flame environment due to improved conductive transfer and ignition of the laminate [35, 36]. Furthermore, Aluminized fabrics are otherwise impervious to moisture and air, stiff, costly and sensitive to soiling and soot. 23

39 Fabric Thermal Properties in Flash Fire Condition Thermal Conductivity The literature describes a variety of methods to measure fabric thermal conductivity [38]. In comfort conditions, a guarded hot plate can be used. Other methods include a thin heater apparatus or a heat flow meter. The Thermal Properties Test Fixture (TPTF), developed by the Ktech Corporation, uses a skin simulant sensor to estimate the thermal properties of fire fighting clothing materials at low heat flux exposure levels [39]. This apparatus evaluates clothing materials with relatively little compression loads, while allowing for evaluation of wet materials. A computer program is used to determine thermal properties based on the heat flux and measured temperature rise measured in different layers in the test fixture. The Ktech method is similar to the method used by Stoll and Chianta [40] and Baker, et al. [41]. They fitted experimental data to results using analytical models to determine the thermal conductivity of protective fabrics and charring ablators, under high heat flux conditions. Differential Scanning Calorimeter (DSC) techniques can also be used to determine thermal conductivity values for protective fabrics, as described by Shalev [42]. Other than DSC method, all these methods assume either constant thermal properties or operate in a temperature range far below the intense heat flux condition expected in protective clothing. These techniques are dependent on the choice of the analytical or numerical model used to fit the data. It is known that the initial thermal properties and physical dimension of protective fabrics change during high heat exposures due to the shrinkage and degradation [36]. Morse et al. [43] calculate fabric thermal conductivity using the flowing equation: K = x( V f k f + V a k a ) + y V a k k f f k a + V f k a, where V, V = volume fraction of fiber and air, and f a 24

40 k f, k a = conductivity of fiber and air. In this equation, x +y = 1 And V V = 1, a + f Torvi used a simplified model to weight contributions from the solid fibers and the air, as well as the contribution of radiation heat transfer between fibers [44]. Such that k = ( k + k ) + k. eff gas solid rad Since heat transfer in fibrous materials is a combination of conduction/convection in the air between fibers, conduction in solid fibers, and radiation heat transfer between fibers, effective conduction can be represented as: k ( T ) = ( ν k ( T ) + (1 ν ) k ( T )) k ( T ), eff air air air fiber + rad where ν air is the volume fraction of air in the fibrous material. For Nomex fiber [45, 46], k fiber ( T ) = ( T ( K) 300k) T 700k = 1.0 T > 700k. The thermal conductivity of air has been represented by a linear relationship [47], k air ( T ) = ( T ( K) 300K) T 700K = ( T ( K) 700K) T > 700K. 25

41 In these models, the fibers in the fabric are assumed to function as infinite plates acting as radiation shields. Hence, the portion of the thermal conductivity due to thermal radiation between the fibers is assumed to equal k rad σε = fiber x T + T 2 ε fiber )( T1 + T ), 2 2 ( where ε fiber = emissivity of the fibers, and x = width of the particular finite element. The radiation portion of thermal conductivity is known to be very small. For a 100K temperature difference across on finite fabric element in the fabric, the contribution due to thermal radiation is about 5% of the total thermal conductivity [44]. This model can be used to calculate the thermal conductivity in a wide range of temperature of interest. Figure 2-11 shows the relationship between thermal conductivity and temperature predicted for Nomex IIIA. Thermal Conductivity v.s. Temperature Theramal Conductivity (W/m C) K Fibre 0.6 K Air K Fabric Temperature (K) Figure Fabric Thermal Conductivity vs. Temperature 26

42 The difference between the thermal conductivities of the Nomex IIIA and Kevlar/PBI are expected to be very small. Consequently, these relationships can be used to represent the effective thermal conductivity of both Nomex IIIA and Kevlar /PBI protective clothing materials Heat Capacity and Thermal Decomposition Temperature The specific heat of polymeric materials ranges from cal/g/k at 20 0 C [20]. The heat capacity of most fabrics changes about 50% when temperature rises from 500 K to 1000K. Schoppee [48] et al. advance an empirical formula to calculate change in C P with temperature for average polymeric materials: C p = 2.22T + 629, where T is in Kelvin and C p is given in J/kg.K. Backer et al. [49] introduced an exacting method of measuring fabric heat capacity by dropping a tightly rolled fabric into a water-containing calorimeter. For many different fabrics he found heat capacities in the range of cal/g/k, measured over temperatures range of C. Two thermal analysis methods can be used to calculate fabric heat capacity at high temperatures: Thermal Gravimetric Analysis (TGA) and the Differential Scanning Calorimeter (DSC). Detailed information of the application of these methods can be found in Torvi [50]. From TGA curves found in the literature [51, 52], the mass of Nomex and Kevlar /PBI fabrics remains fairly constant until the onset of thermal decomposition. The approximate temperature ranges over which the majority of the thermal decomposition occurs for these materials in helium and oxygen environment is noted as [51]. 27

43 Table 2-3. Thermal Decomposition Temperature Material Helium Oxygen Nomex IIIA C C Kevlar /PBI C C During burning, the heating rates can be at the order of about C per minute. The heating rate in TGA is 20 0 C per minute. Some investigators have found that heating rate affects TGA results, shifting the TGA graph to higher temperature as the heating rate increases. Efforts have been made to develop equipment to measure at higher heating rates, including Herderson and O Brien [53], Bingham and Hill [54], and Shlensky, et al. [55]. These experiments show that, while TGA curves do not shift indefinitely as the heating rate increases, the ultimate thermogravimetric curve no longer tends to depend on heating rate. The advantage of using DSC to measure the specific heat of fabrics includes accuracy, rapid data acquisition, and relatively small sample sizes [56]. The only problem associated with the use of DSC for this purpose relates to the temperature limitation. Torvi attempted to use differential thermal analysis (DTA) to obtain high temperatures, but encountered difficulty due to the accuracy of DTA. Torvi expressed the apparent specific heat equation for Nomex and Kevlar /PBI fabrics using information obtained from TGA and DSC as follows: c A ( T ) slope( T 300K) = T < T wt 1 hwtr ( moist) Twtrslope T 2 = Twt1 T Twtr2 wtr = slope( T 300K) T wtr2 < T < Trxl h T slope( T T 300) + slope 2 rx rx = rxl Trx1 T Trx2 rx slope( T rx 1 300K) = T rx 2 T >, 28

44 where: hwtr is the latent heat of vaporization of water, h rx is the energy associated with thermal decomposition, moist is the initial mass fraction of moisture in the fabric, and slope is the slope of specific heat-temperature curve for each fabric. Values used in above equations for Nomex IIIA and Kevlar/PBI are shown in Table 2-4 below: Table 2-4. Values for use in Equations for Nomex IIIA and Kevlar /PBI Fabrics Constant Nomex IIIA Kevlar /PBI ( KJ / Kg 0 C) h wtr moist T wtr1 ( 0 C) T wtr2 ( 0 C) slope (J/kg 0 C 2 ) h rx (kj/kg 0 C) T rx1 ( 0 C) T rx2 ( 0 C) These calculations are for a heating rate of 20 0 C per minute, slower than heating rate in actual burns. It has also been assumed that the thermal decomposition reactions of the fabrics are endothermic. Schoppee, et al. [48] have stated that, in tests which use a radiant heat source rather than a flame, exothermic reactions may be possible due to the significantly higher amount of oxygen available to the fabric. 29

45 Effects of High Heat Exposure on Fabric Dimensions Fabric mass per unit area, thickness and bulk are important to thermal protective performance, especially in intense heat flux flame conditions. Manikin tests show that a Nomex III 6.2 oz/yd 2 coverall protects 59.5% of the body from burn, while Nomex IIIA 7.4 oz/yard 2 protects 72.3% in 3 sec. duration with exposure of 84kW/m 2 [57]. This is because the increased mass per unit area changes fabric density, emissivity, heat capacity and thickness. In transient heat transfer processes, these changes increase time to the second and third degree burn. Therefore, the ability of a fabric to maintain its original weight will directly affect its performance. Since thermogravimetric analysis (TGA) is often used to determine mass loss as a function of temperature and time, we need to compare the heating rate in TGA tests with the actual conditions of high intensity exposures. The initial thickness of protective fabrics correlates well with thermal protective performance [58]. This is due to the overwhelming importance of the thickness of the still air maintained by the fabric in a conduction dominated heat transfer process. Fabric thickness is less important if air gaps are introduced in the protective system. Fabric thickness can change in intense heat exposures due to melting, charring and ablation. This process has been modeled for space vehicle reentry heat shields [59]. Change in fabric density during thermal exposures may arise from shrinkage, or from charring. This can be modeled by considering the fabric to be composed of varying proportions of pure char and pure plastic. Shrinkage may dramatically reduce the air gap between the fabric and skin or fabric to fabric, hence, considered to be the potential burn injury. Fabric thickness can vary considerably depending on the pressure at which they are measured. Some investigators, therefore, treat thickness by including it in the thermal conductance, which is thermal conductivity divided by the thickness, rather than by measuring it separately [60]. 30

46 While the density of the protective fabrics (Nomex and Kevlar /PBI) is expected to change during the exposure, these changes have been shown to be relative small when the fabric temperature does not exceed C. This can be further justified by NASA database of properties of thermal protective materials [61]. However, when the temperature produced by the thermal exposure exceeds this range, fabric densities may not remain constant, especially for fabrics that shrunk, such as Nomex Characterization of the Fire Environment Real fires produce a thermal environment characterized by turbulent buoyant diffusion flames [62]. The radiation characteristics of the fire depend on the degradation products of the combustion process. For this reason, fires have different characteristics depending on type of fuel that is burned. Common combustible degradation products from polymers are methane, ethane, ethylene, formaldehyde, acetone and carbon monoxide [63]. Noncombustible products can include carbon dioxide, hydrogen chloride and water, or in the case of FR cotton, laevoglucose [64]. Water has strong absorption bands at 2.7 and 6.3 µm, CO 2 at 2.7 and 4.3 µm. Radiant intensity, which is wavelength dependent, determines the potential hazard of the fire exposure. Heat flux from a fire is similar to a black body at the fire temperature, usually in the range of C [65]. Wavelengths range 1-6 microns at heat flux levels above 84kW/m 2, with a peak at about 2 microns. Holcombe an Hoschke[66] measured heat fluxes of kw/m 2 from simulated mine explosions, and kw/m 2 for JP-4 fuel fires. Krasiy, et al. [67] reported estimates of 180kW/m 2 in seven room fires from just below flashover to flashover and severe postflashover fires. Thermal Protection Performance (TPP) test methods combine the effects of flame impingement and spectral and transfer modes simulate exposure conditions. Schoppee et 31

47 al [68] compare the behavior of quartz panels to a blackbody that, at maximum output, wavelengths coincide with temperatures above 1023K. At lower temperatures, the emissive power of a quartz panel falls within the waveband containing 75 percent of the total emissive power of a blackbody at the same temperature. David [69] compares quartz lamps and a cone heater having different spectral distribution of radiant energy. He heated fire fighter jacked materials, whose reflectivity and absorbtivity curve depend on the wavelength of the incident radiant energy, with the same thermal flux. These experiments show that the different temperature history occurred on the fabric and the different prediction time to get 2 nd degree burn was found. For these reason, he concluded that a cone heater may be more representative of actual fires than quartz lamps. Fire is generated in the Pyroman chamber with liquid propane gas burned in eight gas burners. If propane is assumed to react with stoichiometric air, then the chemical reaction for complete combustion can be written as [71] C 5 + N 3H8 + O N2 3 CO2 + 4H 2O Adiabatic flame temperatures of about 2400K and 2270K were calculated using STANJAN with and without dissociation, respectively. Adiabatic flame temperature is the maximum possible temperature for this flame. Actual flames are cooler due to heat transfer from the flame and incomplete combustion. Siegel and Howell [70] report values of about 2200K for flame temperature in their experiments. Maximum experimental values for laboratory burners using methane show flame temperature the order of 2000K to 2100K [65]. Holcombe and Hoschke [66] report that approximately 25% of the heat energy released by a Meker burner is thermal radiation. Shalev [42] found that a propane burning Meker burner produced a heat flux which was approximately 70% convective and 30% radiative in nature. 32

48 2.4 Modles for Predicting Skin Burn Injury Skin Burn Models Burns, the result of thermal attack to human skin, are some of the worst injuries that can happen to human beings. Burn injuries require a long time to heal and are sometimes difficult to treat clinically. Burn injuries, which are time and temperature dependent, have been classified as first, second, third, or fourth degree burns [72]. In a first degree burn, the major tissue response of first degree burns is vasodilation of the subpapillary vessels which results in redness to the burned region of the skin. No systemic effects occur and discomfort is temporary. Healing is normally quick with no permanent scarring or discoloration. Second degree burns involve damage to epidermis and dermis layers of skin. Second degree burns are characterized by capillary damage which produces tissue edema and blisters. In these burns cell structure can be damaged, and blood vessels may be distorted and partially blocked. There is an associated loss of fluids and which leads to systemic effects. Plasma volume may also be lost, a major factor causing shock in untreated burn patients. Second degree burns may be further divided into superficial or deep, depending on the penetration depth in the injured zone. Superficial second degree burns are those in which a significant fraction of the cells at the base of the dermis are not destroyed. In this case, healing is normally prompt and without scarring since the majority of the cells at the dermal base are not injured. Deep second degree burns result in loss of much of the dermal base. Certain elements, such as hair follicles and glands, may remain and there is widespread stasis and destruction of cells in the subpapillary plexus. Third degree burns involve destruction of all epidermal elements and supporting dermal structure, including damage of blood vessels in the burn region. With no blood flow, the cells in the region of full thickness burn eventually die. Large volumes of extravascular fluid are lost die to injury to underlying tissue and surrounding the area of full thickness injury. 33

49 Fourth degree burns involve incineration of the skin tissue. Muscle, bone, and other structures beneath subcutaneous tissue may be damaged. Healing is not significantly different from the with third degree burns; although greater complications due to the injuries to underlying tissue may occur. In addition to burn injury, systemic heat trauma due to the thermal stress and the inflammatory mediators occur within the body and are released to the circulatory system. Most of these traumas result from the altered condition of skin due to intense heat exposure. Traumatic effects include the shock of fluid loss, decrease in cardial output, and injuries to the respiratory system. An increase in body metabolic rate can occur to compensate for the large losses from outer evaporation from injured areas, as well as complications related to nutritional defects and altered immune function Bioheat Transfer Models Pennes [73] proposed a transfer equation to describe heat transfer in human tissues: T ρ c t = k 2 T G ( ρc) ( T b T c ) Pennes model assumes that skin tissue above an isothermal core is maintained at a constant body temperature. The resulting simplified bioheat equation is based on following specific assumptions [74, 75]: heat is linearly conducted within tissues; tissue thermal properties are constant in each layer, but may vary from layer to layer; blood temperature is constant and equal to body core temperature; negligible between the large blood vessels (arteries and veins) and the tissue; the local blood flow rate is constant. In long duration, low intensity heat exposure, the rate of metabolic energy production is included in the above equation. In a case of high intensity exposures (84 kw/m 2 from flash fires), metabolic energy production can be assumed to be negligible. 34

50 Several investigators have questioned the validity of the assumptions underlying Pennes model. Wulff [76] claim that the blood flow contribution to heat transfer in tissue must be modeled as a directional term of the form ( ) u T ρ c b, rather than the scalar perfusion term suggested by Pennes. Klinger [77] points out that Pennes equation does not include heat transfer in the vicinity of large blood vessels. Deficiencies in Pennes model result from the fact that thermal equilibrium process occur, not in blood capillaries, as he assumes, but rather in pre- and post-capillary vessels. Nor does Pennes' model account for convective heat transfer due to the blood flow, or for heat exchange between the small and closely-spaced vessels [77] The Chen and Holmes Model Chen and Holmes model [78] group blood vessels into two categories: large vessels, each treated separately, and small vessels, treated as part of the continuum that includes the skin tissue. In their model, heat transfer between small blood vessels and tissue is separated into three modes. The first, perfusion mode, considers equilibration of blood and tissue temperature. The thermal contribution is described by a term, q, similar to the perfusion term in Pennes' equation: p q p ( ρc) ( T T ) = ω, b a where ρ, c, and T are defined as in Pennes' equation, ω is the perfusion parameter that reflects blood flow within vessels in the control volume, and temperature of the blood within the largest vessel in the control volume. T a represents the The second, convective mode, deals with blood vessels that are already thermally equilibrated. This model represents the part of heat transfer that occurs when the flowing blood convects heat against a tissue temperature gradient. For this mode of heat transfer, 35

51 blood is assumed to be in thermal equilibrium with the tissue at a temperature T; the heat transfer contribution can be estimated as: q c ( c) u T = ρ, b where u is the net volume flux vector permeating a unit area of the control surface. A third mode describes thermal conduction due to small temperature fluctuations that occurs in the blood along the tissue temperature gradient: q pc = k T, p where k p is a perfusion conductivity tensor that depends on local blood flow velocity within the vessel, the relative angle between the directions of the blood vessel, the local tissue temperature gradient and the number of vessels in the control volume. The Chen and Holmes model accounts for all three modes of heat transfer between the blood and the tissue, so that T ρc = b b + t ( k T ) + ( ρc) ω ( Ta T ) ( ρc) u T + k p T qm. The Chen and Holmes model has been applied to different biothermal situations in Xu [79] and in Xu et al. [80] The Weinbaum, Jiji and Lemos Model Based on anatomical observations in peripheral tissue, Weinbaum et al. [81] and Jiji et al. [82] conclude that the main contribution of local blood perfusion to heat transfer in tissue is associated with incomplete countercurrent heat exchange between pairs of arteries and veins, not with heat exchange at the capillary level. They propose a model 36

52 that consists of three coupled thermal energy equations to describe heat transfer involving arterial and venous blood and skin tissue. Their models are stated as: ( ρc) b ( ρc) 2 dta π rb V ds 2 dtv π rb V ds T ρc t b = q a = q = v 2 ( k T ) + n g( ρc) ( T T ) nπ r ( ρc) b a v b b V d ( T T ) a ds v + q m, where g is the volumetric rate of the bleed-off blood flow (the flow out of or into the blood vessel via the connecting capillaries), n is the vessel number density, r b is the vessel radius, and V is the blood velocity within the vessel. Applications of this model to a variety of biothermal situations are presented in Dagan et al. [83], and in Song et al. [84, 85] Skin Burn Models Heriques and Moritz [86], working at the Harvard Medical School, were among the first to publish a skin burn model. They claim that skin burn damage can be represented as a chemical rate process, so that a first order Arrhenius rate equation can be used to estimate the rate of tissue damage as: dω E = p exp( ), dt RT where Ω - a quantitative measure of burn damage at the basal layer or at any depth in the dermis, P frequency factor, S -1, E the activation energy for skin, J/mol, R the universal gas constant, J/kmol.K T the absolute temperature at the basal layer or at any depth in the dermis, K, and T total time for which T is above 440C (317.15K). 37

53 Integrate this equation yields: t E Ω = p exp( ) dt. RT 0 Integration is performed for a time when the temperature of basal layer of the skin, T, exceeds or equals to 44 0 C. Bench scale tests of protective fabrics typically use data from the work of Stoll and Chianta [90] to estimate of the time for second degree burn. Simplicity is the main advantage of this method. Stoll criteria assume a rectangular heat pulse exposure. Any variation from rectangular heat pulse invalidates the use of Stoll criteria to predict skin burn injury. Butter [87] also uses the Henriques s burn integral. However, much of his work involves determining the threshold of unbearable pain when non-penetrating infrared radiation was used to heat the skin of human volunteers. Stoll [18] published extensively on determining the skin pain threshold temperatures, and a constants used in the Henriques burn integral and the thermal protection of fabrics. Mehta and Wong [74] used the Henriques burn integral to predict skin burns. However, they conclude that the Henriques equation is valid only for superficial (epidermal) burns. They point out that the temperature used to calculate the pre-exponential factor and activation energy had not been accurately measured. They express doubt as to whether data from low intensity, long duration tests can be used in high intensity, short duration tests. Mehta and Wong model skin as a finite solid with different layers, each with different properties. They alter the upper time limit in the Henriques integral to include cooling time. Takata [89] used a large number of anaesthetized pigs exposed JP-4 liquid fuel fires to analyze the skin data. Data observed by these researchers are summarized in table

54 Table 2-5. Values of P and E Used in Burn Integral Calculations Source P (Hz) for epidermis at temperature E (J/kmole) for dermis at temperature 50 C < 50 C 50 C < 50 C Weaver & Stoll [1969] 2.18 x x x x 108 Takata [1974] 4.32 x x x x 108 Mehta & Wong [1973] 1.43 x x x x Modeling Heat Transfer in Protective Fabrics Predicting thermal protective performance requires an ability to model heat transfer through protective clothing materials, through interfacial air gaps between the skin and clothing, and, finally, through the skin itself. Protective garments exposed to a fire hazards undergo three heating phases. During an initial warm up phase, the temperature of the fibers in the fabric and the moisture retained within the fabric increases at a rate dictated by the system s thermal properties and by the intensity of the incident heat. Consequently, the amount of retained moisture and its thermal properties varies. For most fabric systems, the fiber content and their thermal properties remain constant during the initial heating phase. The second heating phase is marked by the onset of changes in the thermal properties of fabric. Changes in the amount of retained moisture and its thermal properties continue to occur during this phase. Initially, most of these changes are due to the react of surface chemical treatments and finishes to heating, or to slight degradation of fiber surfaces. Most of these changes initiate focusing the exposed side of the fabric system and propagate toward the skin side of the protective material. If the fiber does not melt or transition temperature is not exceeded, the structural integrity of fabric system is maintained during this phase of heating. The end of the second heating phase is the temperature criteria below which thermally protective fabrics are designed to function. 39

55 Protective fabrics exposed beyond the second phase loose its protective properties and, in some instances, become a source of harm to the wearer. In practice, the occurrence of the subsequent third phase occurs only when the protective clothing system is used beyond its intended limits of application. A third and final phase of the exposure is marked by chemical and structural degradation of the protective fabric. At this point, no moisture is retained by the fabric. This phase is followed by rapid fabric decomposition or combustion. At this point, the fabric itself becomes a source of off-gassing heat and flame Heat Transfer Models Heat transfer models have been developed to characterize the behavior of protective fabrics in short duration high heat flux exposures. Some models focus on specific mechanisms of heat transfer, while others provide a predictive model for a particular thermal test. Three models offer the most promising foundation for development into a complete generalized model. These models are Torvi Model [91], the Gibson Model [92], and the Mell and Lawson Model [93] Torvi Model-- Fabric-Air Gap-Test Sensor Model Torvi [91] introduced a model to describe an experimental apparatus consisting of a fabric held horizontally over a Meker burner, with a copper calorimeter held over the fabric. The model treats heat transfer in the vertical dimension only. It accounts for convection and radiation in the air gap between the burner and the fabric; conduction, absorbed radiation, and thermochemical reaction within the fabric, and conduction, convection, and radiation in the air gap between the fabric and the sensor. Torvi s model accounts for the most significant contributions to heat transfer from the burner, through the fabric to the skin. It can be extended to treat heat transfer in 40

56 multiple dimensions, and through multiple layers of fabric. Extensions of this model treat convective heat transfer in the fabric and heat conveyed by moisture within the fabric or in air gaps. Torvi accounts for convection and radiation on the outside of the material, exposed to the burner, and conduction/convection and radiation in the air gaps between fabric and skin. Radiation heat flux is exposed as the sum of blackbody components from hot gases, from the fabric to ambient air, and from the burner head to the fabric. Therefore, q rad = σε T g 4 g σε F f a (1 ε )( T g 4 f T 4 a 4 σfb (1 ε g )( Tb T ) + 1 ε f A 1+ F b (1 ε g ) + ε f A 4 f f b ), 1 ε b ε b where σ is the Stefan-Boltzmann constant, ε g, ε f, and ε b are emissivities of the hot gases, the fabric, and the burner head, respectively, T g, T f, T a, and T b are the temperatures of the hot gases, the outside of the fabric, the ambient air, and the burner head, respectively, F a and F b are view factors accounting for the geometry of the fabric with respect to the ambient air and to the burner, respectively, and A f and A b are the surface areas of the fabric and the burner head, respectively. Radiation heat flux on the inside is q rad = 1 ε ε s s σ ( T + A A s f 4 f T 1 Fs 4 s ) 1 ε f + ε f, where T s, ε s, and A s are the temperature, emissivity, and surface area, respectively, of a test sensor taking the place of skin, and F s accounts for the geometry of the fabric with respect to the sensor. 41

57 Torvi accounts for conduction, thermochemical reaction, and absorption of incident radiation that occurs with the fabric. The resulting energy balance equation is written as C T t = x T x A γ x ( T ) k( T ) + γ qrade, where T is the temperature, C A is a temperature-dependent "apparent" specific heat, which incorporates latent heat associated with thermochemical reaction, k is a temperature-dependent thermal conductivity, γ is the extinction coefficient of the fabric, and q rad is the incident radiation heat flux Gibson Model -- Multiphase Heat and Mass Transfer Model Gibson [92] built on Whitaker s theory of coupled heat and mass transfer through porous media [94] to derive a set of equations modeling heat and mass transfer through textile materials as hygroscopic porous media. Gibson applies continuity, linear momentum conservation, and energy conservation equations to the fabric as a three-phase system consisting of a solid phase with a concentration of bound water, a free liquid water phase, and a gas phase of water vapor in air. This model treats heat transfer in three dimensions, accounting for conduction by all phases, convection by the gas and liquid phases, and transformations among the phases. Gibson s model thoroughly treats convection heat transfer in the fabric. Gibson and Charmchi [95] extended the model to include heat transfer to skin in contact with the fabric. Further extensions can be made to include air and moisture mass transfer in multiple layers and air gaps, as well as radiative transfer. Gibson has applied Whitaker s theory, coupling heat and mass transfer through porous media [94], to derive a set of equations that model heat and mass transfer through 42

58 textile materials as hygroscopic porous media [92]. The material modeled a mixture of a solid phase, consisting of solid (e.g., polymer) material plus water absorbed into the polymer matrix, a liquid phase consisting of free liquid water solid, and a gaseous phase consisting of water vapor and inert air (Figure 2-12). Gibson applies continuity, linear momentum conservation, and energy conservation equations to the fabric as a three-phase system consisting of a solid phase with a concentration of bound water, a free liquid water phase, and a gas phase of water vapor in air. The model treats heat transfer in three dimensions. It accounts for conduction by all phases, convection by the gas and liquid phases, and all transformations among the phases. Gibson writes the thermal energy balance equation as: ρ c p T t + j ( c p ) j ρ v j j + ρ ( c + h β vap p m& ) lv β v β + Q l + m& sl i ( c p ) i + ( Q l ρivi T + h ) m& vap sv = ( K T eff T ). Figure Three Phases Present in Hygroscopic Porous Media Gibson applies continuity, linear momentum and energy conservation equations to the fabric, treating heat transfer in three dimensions. The model accounts for conduction 43

59 by all phases, convection by the gas and liquid phases, and transformations among the phases. If we number each distinct species, (1) for water, (2) for dry solid, and (3) for inert air, the thermal energy balance equation can be written as: ρ C p T t ( c + p ) 1 σ γ ( ρ β v β + εσ ρ1v1 + εγ ρ1v1 ) + ( c p ) 2 + h ε vap σ ρ v m& lv 2 2 σ + Q l + ( c m& sl p ) ε 3 γ + ( Q l ρ v h γ vap T ) m& sv = ( K T eff T ), where ρ C = ρ +. p 1 ( c p ) 1 + ρ 2 ( c p ) 2 ρ3 ( c p ) 3 The bracketed terms denote a volume averages over all phases, or over the single phase as indicated by a superscript. The term ε denotes the volume fraction of a single phase, v denotes velocity, K eff is the effective thermal conductivity tensor, h vap is the heat of vaporization of the liquid phase, Q l is the heat of desorption from the solid phase, and m& lv, m& sl, and m& sv denote the mass flux desorbing from the solid to the liquid, desorbing from the solid to the gas, and evaporating from the liquid, respectively. Gibson and Charmchi extended the model to include heat transfer to skin in contact with the fabric [5]. Their model thoroughly treats convection heat transfer in fabrics, provided the heat flux is not too high. It further treats heat transfer in three dimensions and can be extended to treat high heat fluxes, the absorption of radiation, and to model thermochemical processes in fabrics Mell and Lawson Model Mell and Lawson [93] use a model similar to Torvi's to treat heat transfer in a fire fighter's turnout coat, or a multi-layer composite consisting of a shell layer, including trim material, moisture barrier, and thermal liner. Like Torvi, Mell and Lawson account 44

60 for conduction and absorption of incident radiation in the material layers. In addition, Mell and Lawson advance a forward-reverse radiation model, defining incident fluxes on both sides of each layer of material to account for interlayer flux due to reflected radiation between fabric layers. Their model is an example of extending the Torvi model to treat a multi-layer fabric assembles. It could be extended to treat heat transfer in three dimensions, to model convective heat transfer and heat conveyed by moisture within the air gaps Modeling Thermal Degradation in Fabrics The combustion and thermal degradation of polymeric fabrics are complicated processes involving physical and chemical phenomena that are only partially understood. A number of different approaches for modeling this problem have been suggested in the literature. Ricci [96], Whiting et al. [97], Delichatsios and Chen [98], and Staggs [99] suggest modeling thermal degradation of polymers as a Stefan problem, where the degradation of the solid is assumed to occur infinitely rapidly once a critical temperature reached. Other researchers model solid-phase degradation using limited global in-depth reactions [100][101][102]. Kashiwagi [103] reviews physical and chemical phenomena involved in polymer combustion and highlights the complexity of this process. Staggs [104] suggest a heat transfer model that incorporated a general single-step solid-phase reaction for thermal degradation of polymer material. Staggs model does not account for heat transport by gaseous products escaping from the solid. The critical-temperature approach has been utilized by some researchers to derive simplified models of thermal degradation. This approach is overly simplified, because it is assume that the polymer can volatilize only at surface exposed to heat. Staggs [99] model s thermal degradation utilizing the following kinetic rate law: Dµ = f ( µ, T ), Dt 45

61 where µ is a scalar quantity representing the progress of the reaction (the ratio of the mass of the material element to its initial mass), t is the time, and f is a function determined by the rate of degradation. Different forms for the f function can be used to model specific decomposition mechanisms characteristic of different types of polymers or other chemicals used in thermally protective clothing. For the most polymeric fabrics, the use of an n th order Arrenius reaction provides sufficient agreement with experimental data; so that: f ( T / T ) n = Aµ exp, A where T A is the activation temperature, A is the empirically derived pre-exponential factor, and n is the order of the degradation reaction. Accounting for degradation in the solid material produces an additional term in the energy conservation equation. If m 0 is the initial mass of the material element, then its mass at time t is µ () t m0. The rate of heat consumption for the vaporization of Dµ polymer material in this element during its degradation can be expressed as H m0, Dt where H is the heat of vaporization. Recasting this relation for a unit volume results in the following term for the energy equation for the fabric layer: ρ H Dµ ρ H n q = = Aµ exp µ Dt µ ( T / T ) A. Staggs shows that the n th order Arrenius reaction is used to model thermal degradation, the predicted surface temperature increases slowly during the mass loss period. Surface temperature is not determined solely by the properties of the polymer material, such as the specific chemical reaction that describes its degradation, but by interaction between reaction kinetics and the rate of heat loss [99]. This finding explains 46

62 why the critical-temperature ablation models of polymer degradation do not provide sufficiently accurate to predict heat transfer through fabrics exposed to intense heat. As previously noted, Staggs model does not consider transport of gaseous products from the degrading polymer material. This model can be supplemented with a model that describes transport of these products through the fabric layer, modeled as a thermally degrading porous medium. One of the issues to be addressed is the change of permeability of the porous medium as a result of its thermal degradation. It is expected that degradation will produce a decrease of permeability in the fabric and thus worsening its mass transport characteristics. This will hinder not only transport of the gaseous products of polymer degradation but also moisture transport through the degrading fabric. This, in turn, can lead to accumulation of heat moisture in the fabric because the moisture that results from sweating will not be removed. Moisture accumulation in fabrics has been proven to have a profound influence on heat transfer and thermal protective performance. 47

63 CHAPTER 3 MODEL THE PYROMAN SYSTEM The NCSU Pyroman is an instrumented, six-foot, one-inch tall, high-temperature manikin embedded with 122 heat sensors. It is used to test and measure the protective performance of a variety of garments and clothing systems under realistic flash fire conditions. The Pyroman dressed with protective garments and engulfed in flames so that factors like garment construction, fabric weight, material type, style, fit and the impact of outerwear and undergarments can be taken into account. Results of these tests are then analyzed to determine the skin damage in terms of second and third degree burn. It is the most advanced life-size thermal burn injury evaluation system in the world today. In this chapter, two configurations are modeled to represent heat transfer in one layer protective garment dressed in a manikin with and without underwear. The developed two models are fabric-air gap-skin model and fabric-air gap-fabric-skin model used to calculate heat transfer at one hundred-twenty two sensors locations in Pyroman body. The heat transfer in fabric, air gap and human skin as well as corresponding boundary conditions are described in this chapter. Fabric shrinkage, and the effect of this on air gap size, is also included. The skin model and burn evaluation model used in present model are outlined. 3.1 A Numerical Heat Transfer Model Fundamental elements of the Pyroman fire test and burn evaluation processes are illustrated in Figure 3-1. Flash fire is generated in the Pyroman burn chamber by eight propane burning torches. The intense heat from the flash fire transfers through the fabric and air gaps between the garment and manikin body. One hundred twenty-two sensors embedded in Pyroman body record the temperature change and feed this information to computer. The temperature profiles recorded by these sensors are translated into 48

64 corresponding heat flux profiles and then applied to a skin burn model. The skin model calculates the temperature profile at basal layer and dermal layer according to the thermal properties of the skin. Second and third degree burn skin damages are made using temperature histories by computing the Omega (Ω) value in the Pyroman burn evaluation model. Flash fire: Air gap Sensor Heat Flux Profile Skin Model Burn Evaluation Model t E Ω = P exp( ) RT 0 Protective garments system Pyroman Body Basal layer T-t history, 2 nd degree burn data Dermal layer T-t history, 3 rd degree burn data Figure 3-1. Elements of Heat Transfer and Burn Evaluation in the Pyroman System The numerical model developed in this research model the entire burning process. It first calculates heat generated by the flash fire generated in Pyroman chamber from the estimated general heat transfer coefficients and measured heat fluxes in calibration burns. The heat transfer in the fabric and air gaps is calculated in conduction, radiation and convection modes. A bioheat transfer equation is subsequently applied in conjunction with a multi-layer skin model to estimate temperature profile in basal and dermal skin layer. Skin damage is predicted using burn evaluation model. The model is illustrated in Figure 3-2. In order to perform the analysis of heat transfer in the model, several assumptions are made. A one-dimensional heat transfer process is assumed in the model; no mass 49

65 transfer occurs in fabric and air gap; the fabric is considered as grey body for radiation. As the special protective garment is chosen here, the thermal-chemical reaction and degradation of fabric are neglected for short time exposure in this research. The thermal properties of skin are assumed to be constant. For the present model, the thermal conductivity and volumetric heat capacity of fabrics under intense exposure are not assumed constant; these properties are estimated using parameter estimation method. The heat flux at each sensor location is assumed uniform due to one-dimensional heat transfer. An in-depth radiation in fabric is involved in the heat transfer in fabric as introduced in Torvi s model [91]. Flash fire Air gap Skin Model Burn Evaluation Model t E Ω = P exp( ) RT 0 Protective garment system Basal layer T-t history, 2 nd degree burn data Dermal layer T-t history, 3 rd degree burn data Figure 3-2. Elements of Fabric Air-Gap Skin Model 50

66 3.2 Heat Transfer in Fabric, Air gap and Skin The schematic of this model is illustrated in Figure 3-3. The model is assumed that convective heat transfer occurs only as far as the surface of fabric. Radiative heat flux is assumed to penetrate the fabric to a certain depth. Based these assumptions, the energy balance in the infinitesimal element of fabric can be described by the following equation: ρ T t x T x γx fab ( T ) Cp fab ( T ) = k fab ( T ) γ q rad e (3-1) Figure 3-3. Schematic for one-dimensional Heat Transfer Model where ρ fab is the density of the fabric; Cp fab is the specific heat of the fabric; k fab is the thermal conductivity of the fabric; γ is the extinction coefficient of the fabric which can be determined from the transmissivity τ and the thickness L fab, 51

67 γ = ln(τ ) /, (3-2) L fab q rad is the incident radiation heat flux on which can be expressed as follows: q rad = σε ( T T ) σε F (1 ε )( T T ), (3-3) g g fab fab fab amb g fab amb where σ is the Stefan-Boltzmann constant, ε g, and ε fab are the emissivity of the hot gases and the fabric, respectively, T g, T fab, and T amb are the temperatures of the hot gases, the outside surface of the fabric, and the ambient air, respectively, F fab-amb is the view factor accounting for the geometry of the fabric in relative to the ambient air. The boundary conditions for the outside surface of the fabric are (x = 0), for t > 0, T k, (3-4) ( qconv + qrad ) x 0 fab ( T ) = = x x= 0 where q rad is the incident radiation heat flux, and q conv is the convective heat flux between the fabric and hot gas given as follows: q conv = h T T ), (3-5) fl ( g fab The subscript fl, g, and fab refer to the burner flame, the hot gases from the burner, and the outside surface of fabric. At the inside surface of the fabric (x = L fab ), for t > 0: T (, (3-6) ( qair, rad + qair, cond conv ) x L fab k fab T ) = / = x x= L fab 52

68 where q air, cond/conv is the thermal energy transfer by conduction/convection from fabric to the human skin across the air gap given as: q = h ( T T air, cond / conv x= L air, gap fab skin fab ). (3-7) In the equation 3-7, h air, gap is the heat transfer coefficient of air due to conduction and natural convection in air gap given by: h air, gap k L ( T ) air =, (3-8) Nu air, gap where Nu is the Nusselt number, k air (T) is the thermal conductivity of the air and L air,gap is the thickness of the air gap. In some specific fabric, the air gap is a function of fabric temperature. This is because the heat induced shrinkage of garment during exposure reduces air gap size. In the equation 3-6, q air, rad is the energy transfer by radiation from fabric to the human skin across the air gap given as: q air 4 4 ( T T ) σ fab skin, rad =, (3-9) A 1 ε skin fab 1 1 ε skin + + A fab ε fab F fab skin ε skin where σ is the Stefan-Boltzmann constant, ε fab, and ε skin are the fabric and human skin emissivities, T fab, and T skin are the temperatures of the inside surface of the fabric, and the human skin, F fab-skin is the view factor accounting for the geometry of the fabric relative to the human skin, A fab and A skin are the surface areas of the fabric and the human skin, respectively. In order to consider the heat transfer by natural convection occurred in between the air gap of fabric and skin when the air gap size and temperature difference are large 53

69 enough, a corrected Nusselt number was selected used in equation 3-8 which represents natural convection in a vertical enclosure heated from one side. Catton [107] using a relationship based on Denny and Clever s work [108], expresses the Nusselt number correlation for air in a long vertical enclosure as: Nu = Ra, (3-10) Ra 3 = g β T δ / αν, (3-11) where Nu = the Nusselt number (hδ/k) Ra = Rayleigh number (gβ Τδ 3 /αν) g = the gravity acceleration (9.81 m/s 2 ) β = the thermal expansion coefficient of the air (k -1 ) Τ = the temperature difference across the air gap α = the thermal diffusivity of the air (m 2 /s) ν = the kinematic viscosity of air (m 2 /s) Similarly, Hollands, et al. [109] gives the correlation for any chosen reference tiled enclosure as Nu = Ra cos τ Ra cos τ 1 / 3 1, (sin 1.8τ ) 1 Ra cos τ (3-12) where τ = angle of inclination. The notation [ ] in above equation indicates that if the argument in the square brackets is negative, the quantity should be taken as zero. The correlation for vertical enclosure can be calculated by combining with the scaling suggested by Ayyaswamy and Catton [110] as: 54

70 o 1 / 4 Nu ( τ ) = Nu ( τ = 90 )(sin τ ). (3-13) Another configuration modeled in this research is one layer protective garment with underwear. In this configuration an extra cotton fabric layer is added with assumption that no air gaps exist between skin and underwear cotton fabric. The elements in this configuration are illustrated in Figure Flash fire Underwear fabric Skin Model Burn Evaluation Model t E Ω = P exp( ) RT 0 Protective garment system Air gap Basal layer T-t history, 2 nd degree burn data Dermal layer T-t history, 3 rd degree burn data Figure 3-4. Fabric-air gap-fabric-skin Model The underwear fabric under short time exposure is experiencing relatively low temperatures; therefore, the thermal properties of underwear fabric assume constant. The heat transfer in underwear fabric can be described as: 2 T T ρ underwear Cp underwear = k underwear 2, (3-14) t x where ρ underwear = density of underwear fabric Cp underwear = specific heat of underwear fabric k underwear = thermal conductivity of underwear fabric 55

71 The boundary conditions at the outside surface of the underwear fabric are (x = L fab + L gap ), for t > 0, T ( q air rad + q air, cond conv ) x = L fab L gap k underwear = / + x x = L fab + L gap,, (3-15) where q air,rad = the energy transfer by radiation from the protective fabric to the underwear fabric across the air gap given as: q air 4 4 ( T T ) σ fab underwear, rad =, (3-16) x = L fab + L gap A 1 ε underwear fab 1 1 ε underwear + + A fab ε fab F fab underwear ε underwear where σ = the Stefan-Boltzmann constant ( W/m 2 K -4 ) ε fab = the emissivity of the protective fabric ε underwear = the emissivity of the underwear fabric T fab = the temperature of the inside surface of the protective fabric T underwear = the temperature of the inside surface of the underwear fabric F fab-underwear = the view factor accounting for the geometry of the protective fabric with respect to the underwear fabric A fab = the surface areas of the protective fabric A underwear = the surface areas of the underwear fabric q air,cond/conv = the energy transfer by conduction/convection from the protective fabric to the underwear fabric across the air gap given as follows: q = h ( T T air, cond / conv x= L + L air, gap fab underwer fab gap ), (3-17) h air, gap = the heat transfer coefficient of air due to conduction and natural convection in air gap given by: 56

72 h air, gap k L ( T ) air = Nu, (3-18) air, gap where Nu = the Nusselt number k air (T) = the thermal conductivity of the air L air,gap = the thickness of the air gap. At the inside surface of the fabric (x = L fab + L gap + L underwear ), the underwear fabric is assumed to contact directly to the human skin. There is no air gap between the underwear fabric and the human skin. The conductive heat transfer only occurs at the interface of fabric and epidermis skin layer. L underwear, is the thickness of underwear fabric, for t > 0 k underwear T x = k T x skin x = L fab + L gap + L underwear x = L fab + L gap + L underwear. (3-19) The initial condition is a given temperature distribution at t = 0. In addition, the underwear fabric temperature is assumed as initially uniform. T(x, t = 0) = T i (x) 3.3 Heat Transfer in Skin Model and Burn Evaluation The present model incorporates Pennes Model to describe the heat transfer in skin. Pennes model assumes the energy exchange between the blood vessels in the skin and the surrounding tissue. According to this Model, the total energy exchanged by the flowing blood is proportional to volumetric heat flow and the temperature difference between the blood and skin tissue. The bio-heat transfer equation is written as: T ρ skincp skin = ( kskin T ) + ( ρcp) bloodωb( Ta T ) + qm, (3-14) t 57

73 where ρ skin and Cp skin are the density and the specific heat of human skin. k skin is the thermal conductivity of the human skin; ρ blood and Cp blood is the density and the specific heat of the blood; ω b is blood perfusion; T a is the arterial temperature; and q m is the metabolic volumetric heat. Table 3-1. Human Skin Properties in Skin Model [91] Human Skin Symbol Value Epidermis Dermis Sub-cutaneous Thermal conductivity (W/m. C) k s Density (kg/m 3 ) ρ s 1200 Specific heat (J/kg. C) C p,s 3600 Thickness (m) Thk s 8.0X10-5 Emissivity of human skin ε s 0.94 Initial surface temperature (K) T s Thermal conductivity (W/m. C) k s Density (kg/m 3 ) ρ s 1200 Specific heat (J/kg. C) C p,s 3400 Thickness (m) Thk s 2.0X10-3 Initial surface temperature (K) T s Thermal conductivity (W/m. C) k s Density (kg/m 3 ) ρ s 1000 Specific heat (J/kg. C) C p,s 3060 Thickness (m) Thk s 1.0X10-2 Initial surface temperature (K) T s The boundary conditions for the model at the surface of the skin are (x = L fab + L gap ), for t > 0, T ( ) =. (3-15) fab gap ( qair, rad + qair, cond conv ) x= L fab Lgap k skin T / + x x= L + L 58

74 The boundary condition at the blood vessel is (x = L fab + L gap + L tissue ), for t > 0, T = T a. The initial condition is a given temperature distribution at t = 0. In addition, fabric temperature is assumed to be initially uniform. T(x, t = 0) = T i (x) Multi-layer skin model is used in present model. These layers are epidermis, dermis and subcutaneous with different thickness and thermal properties [91] which is shown in Table 3-1. The numerical approach utilizes the finite difference method to predict the temperature and heat flux occurred under simulated flash fire conditions. The thermal properties of fabric are assumed to change while skin layers properties are assumed constant. The blood perfusion is included only in the latter two regions. The tissue burn injury model is based on work by Henriques and Moritz [86]. Thermal damage begins when the temperature at the basal layer (the interface between the epidermis and dermis in human skin) rises above 44 ºC. The destruction rate of the growing layer can be modeled by a first order chemical reaction. Arrhenius rate equation can be used for the rate of tissue damage as: dω dt = E P exp, (3-16) RT where Ω = a quantitative measure of burn damage at the basal layer or at any depth in the dermis P = the frequency factor or pre-exponential factor, s -1 E = the activation energy for skin, J/mol R = the universal gas constant, J/kmol K 59

75 T = the absolute temperature at the basal or at any depth in the dermis, K t = total time for which T is above 44 ºC ( K) The above equation can be integrated over the time interval that the temperature at the basal layer is above 44 ºC. Ω = t P 0 E exp dt, (3-17) RT For predicting first and second degree burns, T is the temperature of the basal layer. First degree burn occurs when the value of the burn integral, represented by Ω, reaches 0.53 at the basal layer, while second degree burn happens when Ω = 1.0 at the same location. For predicting third degree burns, T is the temperature of the dermal base (the interface between the dermis and the sub-cutaneous layer). Third degree burn occurs when Ω = 1.0 at this location. These tissue burn damage criteria can be applied with providing the appropriate values of P and E. These values were suggested by Weaver and Stoll [11] for the basal layer and by Takata and et al. [12] for the dermal base. The values of P and E are: Epidermis for T < 50ºC P = s -1 E/R = 93,534.9 K for T 50ºC P = s -1 E/R = 39,109.8 K Dermis for T < 50ºC P = s -1 E/R = 50,000 K for T 50ºC P = s -1 E/R = 80,000 K 60

76 3.4 Finite Difference Method For the present model, a finite difference method was used to solve the differential equations that describe heat transfer through the fabric, air gap, and skin layers [111, 112]. Due to nonlinear term of absorption of incident radiation, the Gauss- Seidel point-by-point iterative scheme was used to solve these equations. In order to avoid divergence that is usually found in the iterative scheme, the underrelaxation process was added to the Gauss-Seidel method. In addition, the Crank-Nicholson implicit scheme was used to solve the resulting ordinary differential equations in time. The program was written in Microsoft FORTRAN PowerStation. 61

77 CHAPTER 4 EXPERIMENTAL STUDIES The numerical model developed by this research models each of the one hundred twenty-two thermal sensors embedded in the Pyroman body (Figure 4-3). Application and validation of the model required several experimental studies. The first study involved characterizing the flash fire generated the Pyroman chamber. The second study required measurement of the changes in the thermal conductivity and volumetric heat capacity of the selected protective fabrics as result of the intense heat exposure. Third analysis examined the existing air gap distribution and sizes between protective garment and the manikin body, and the effect of heat shrinkage on the distribution of the air gaps on manikin body. 4.1 Characterizing the Pyroman Thermal Environment The Pyroman flash fire simulation is produced by eight propane burning industrial torches. If propane is assumed to react with stoichiometric air, then the chemical reaction can be written as [71] C 5 + N 3H8 + O N2 3 CO2 + 4H 2O If complete combustion of the propane fuel is consumed, adiabatic flame temperatures varying between 2270K and 2400K are presented [71]. Adiabatic flame temperature is the maximum possible temperature for this flame. In actual conditions, the flame temperatures are expected to be cooler due to heat losses to environment and incomplete combustion. The Pyroman flash fire exposure was characterized by measuring the flame temperature and corresponding heat flux alone each of the one hundred twenty-two 62

78 sensors distributed over the manikin body. These measured temperature and corresponding heat flux history are used to determine overall heat transfer coefficient. Specially designed thermal sensors were used to measure flame temperature alone the manikin surface (Figure 4-1). A Pyrocal sensor was adopted to include type B or type R thermocouple with diameter inch, a type T thermocouple was used to measure copper disc temperature of Pyrocal sensor (Details of these devices can found in Appendix 1). Data from these thermocouples were recorded using a LabVIEW Data Acquisition System (Figure 4-2) Figure 4-1. Specially Designed Sensor Used to Measure Flame Temperature and Heat Flux Using these flame measurements methods, the temperature of the flame incident on the manikin was determined to range between 1100K and 1800K in a 4 second exposure. The flash fire generated in the manikin chamber was adjusted to produce an average heat flux 2.00 cal/cm 2 sec all over the manikin body. Figure 4-4 shows the distribution of heat flux values of the one hundred twenty-two sensors in Pyroman. Variations in heat fluxes are expected due to the three dimensional shape of the manikin surface, and because of the complex and dynamic nature of the flame column surrounded the manikin. Therefore, depending on the location on the manikin body (e.g. arms, legs, shoulders), sensors located in different positions on manikin body will have different heat flux values with respect to the flash fire. 63

79 Figure 4-2. LabVIEW Data Acquisition System and Software Figure 4-3. Sensor Numbers and Distribution over Manikin Body Heat Flux Distribution An example of calibration burn values before garment test in manikin is given in Appendix 3 shows the heat flux of 122 sensors and their normal scores. Figure 4-4 shows a heat flux distribution observed in the manikin flash fire is a normal or bell shaped 64

80 distribution. In order to further access this distribution normality [113], a normal scores plot is made as shown in Figure 4-5 which demonstrated that the heat flux distribution of 122 sensors with an average of 2.00 cal/cm 2 sec during 4 second exposure exhibits a approximate normal distribution. A Kolmogorov-Smirnov test was used to conform distribution normality. Details of this statistical analysis can be found in Appendix % Sensor Percentage (%) Frequency Cumulative % 100.0% 80.0% 60.0% 40.0% 20.0% More Heat Flux (cal/cm 2 sec).0% Figure 4-4. Histogram and Cumulative Curve of Heat Fluxes of 122 Sensors around Pyroman in 4 second Exposure In the manikin test, an average heat flux of 2.00 cal/cm 2.sec measured from 122 sensors of the manikin is used to simulate flash fire conditions, and the standard deviation of these heat fluxes distribution is between 0.25 and 0.5. Figure 4-6 shows the heat flux distributions with different standard deviation for a 4 second exposures. The larger standard deviation indicates more extreme high and low flux values occurred during the exposure. A heat flux distribution in a manikin with average of 2.00 cal/cm 2 sec is illustrated in Figure 4-7. Red color represents extremely high values and blue represents extremely low. A different standard deviation of heat flux distribution will impact the burn predictions. A parametric study in Chapter 6 will discuss this in detail. 65

81 The Scatterplot of Heat Flux vs. Normal Scores for Pyroman 122 Sensors Figure 4-5. The Scatterplot of Heat Fluxes vs. Normal Scores From a Calibration 4 Second Exposure Normal Distribution with Different Standard Deviation, Average Heat Flux 2.00 cal/cm2 sec SD=0.25 SD=0.38 SD=0.5 Nomal Density (f(x)) Heat Flux Figure 4-6. Heat Flux Distribution with Different Standard Deviation 66

82 Front N010626H ARM 199 BACK 196 FRONT 202 HEAD 233 LEG 195 Rear Legend Flux Levels: 170 Avg 200 S.D %CV No In. Figure 4-7. Heat Flux Distribution in a Manikin for a 4 second Exposure with Average 2.00 cal/cm 2.sec 67

83 4.1.2 Heat Transfer Coefficient Determination In order to calculate the heat transfer from the generated flash fire to the manikin or dressed manikin in the burn chamber, an overall heat transfer coefficient is needed for each of 122 sensors. The estimation of the heat transfer coefficient, h, from transient temperature measurements and heat flux calculation has aspects of both the inverse heat conduction problem and parameter estimation [114]. Thermocouple measures surface temperature T OM q M surface heat flux Calculated from temp. profile Thermocouple measures flame temp. T f (t) Flame Sensor Figure 4-8. Heat Transfer Coefficient Estimation from Temperature Measurement and Heat Flux Calculation In this section, two kinds of sensors are used in Pyroman to examine the heat transfer coefficient. One is Pyrocal sensor and the other is skin simulant sensor. The heat transfer coefficient is estimated by sensor surface temperature measurement T om and its flame temperature T f. (Figure 4-8). For these experiments, Pyroman was equipped with specially designed sensor to measure the flame temperature above the sensor in a flash fire exposure (2.00 cal/cm 2. sec). Type B and R thermocouples with 0.002" diameter were used to measure the flame temperature. A type T thermocouple was used to measure the sensor surface temperature (Figure 4-8). For the Pyrocal sensor it is treated as a lumped body that is, one in which the temperature is a function of time only. A function specification procedure with constant h 68

84 is chosen to estimate the heat transfer coefficient [114, 115]. The estimation equation is as following. ˆ qˆ M h M = T ( t ) 0. 5 ( Tˆ + Tˆ (4-1) ) f OM OM 1 where h M estimated heat transfer coefficient q M calculated heat flux T f Flame temperature T OM estimated surface temperature at time t M The Pyrocal sensor is basically made of thin copper disc. This sensor can be treated as a lumped body, which the temperature is uniform but varies with time only. The differential energy equation for a copper disc sensor can be written as: dt ρ c p L s + K L ( T T 0 ) = h ( T T ) (4-2) dt where ρ = the density of a copper disc sensor ( kg/m 3 ) c p = the specific heat of a copper disc sensor ( J/kg ºC) L s = thickness of a copper disc sensor (1.524 mm) T = the temperature of a copper disc sensor (ºC) K L = the total heat loss coefficient of a copper disc sensor (200 W/m 2 ºC) T 0 = the initial temperature of a copper disc (ºC) h = the heat transfer coefficient between flame and sensor (W/m 2 ) T = the flame temperature (ºC) The method is based on the function specification procedure with the h = constant assumption. The temporary constant, h M, is assumed for direct sequential estimation procedure for h. The sum of squares, S, that is defined as follows is minimized with respect to h M. 69

85 70 = + + = r i i M i Y M T S ) ( (4-3) where Y = the measured temperature from experiment T = the calculated temperature from theoretical analysis r = the number of future times over which h M is temporarily constant. By taking partial derivative of S with respect to h M, and setting the equation equal to zero, the equation can be rewritten as: 0 ) ( = + = + + M i M r i i M i M h T T Y (4-4) The analytical expression for temperature is approximately given by: = + + ) ( ) ( exp M i M s p L M L M M L M M M L M M L M M i M t t L c K h K h T K T h T K h T K T h T ρ (4-5) The sensitivity coefficient, Z, is defined as: M i M i M h T Z = From the temperature expression, the sensitivity is rewritten as:

86 = ) ( ) ( exp ) ( ) ( ) ( ) ( ) ( M i M s p L M s p M i M L M M L M M M L M M M L L M M M L i M t t L c K h L c t t K h T K T h T K h T T K K h T T K Z ρ ρ (4-6) Due to nonlinearity of the problem, the Gauss iterative method is needed. First, we assume that the estimation of h M (ν-1) is known at the previous iteration. Then we search for new h M (ν) at the current iteration. By solving for h M (ν), the approximation of the heat transfer coefficient is: = + = = r i i M r i i M i M i M M M Z Z T Y h h 1 2 1) ( 1 1 1) ( 1 1) ( 1 1 1) ( ) ( ) ( ) ( ν ν ν ν ν (4-7) By using two-term Taylor series expansion and neglecting higher order terms, the approximation of temperature is explicitly calculated by: ) ( 1) ( ) ( 1) ( 1 1) ( 1 ) ( = ν ν ν ν ν M M i M i M i M h h Z T T The iteration keeps repeating until the changes in h M (ν) are less than some small amount or criterion, such as: 6 ) ( 1) ( ) ( 10 < ν ν ν M M M h h h After the value of h M (ν) converges, M is increased by one, then the procedure is repeated for the new h M (ν). Finally, the transient heat transfer coefficients between the flame and sensor will be obtained.

87 An example of three Pyrocal sensors flame temperature profiles and their sensor surface temperature responses is shown in Figure 4-9. The dynamic fire situation is observed from these flame temperature profiles. These data are used to estimate heat transfer coefficient of each sensor. Figure 4-10 shows an example of an estimated heat transfer coefficient from the Pyrocal temperature measurements during 4 second exposure. Flame Temperature ( 0 C) Sensor Flame Temperature and Sensor Temperature During 4 Second Pyroman Exposure Flame Temp. of Sensor 75 Flame Temp. of Sensor 70 Flame Temp. of Sensor 91 Sensor 75 Temp. Sensor 70 Temp. Sensor 91 Temp Sensor Temperature ( 0 C) Time (Sec) Figure 4-9. Pyrocal Sensor Flame Temperatures and Copper Temperatures During 4 second Exposure 72

88 Pyroman Sensor Heat Transfer Coefficient Determination ( Pyrocal Sensor) Heat Transfer Coefficient (W/m 2. 0 C) Estimated H. T. Coefficient Sensor Temperature Flame Temperature Temperature ( 0 C) Time (sec) Figure Estimated Heat Transfer Coefficient Using Pyrocal Sensor The flash fire generated in Pyroman chamber is turbulent jet flames generated using eight propane torches around manikin body. In a 4 second exposure, a dynamic burning process is observed as shown in previous fire temperature measurements. From the heat transfer coefficient determination, it was found that the heat transfer coefficient is temperature and location dependent because of the dynamic fire and the complicated manikin body geometry. The location factor was used to represent the influence of human body, that the embedded sensor surface is not always parallel to flame direction. Figure 4-11 and Figure 4-12 demonstrated the relation of heat transfer coefficient with temperature and location factor, respectively. The higher temperature tends to produce large heat transfer coefficient. The location factor influence on heat transfer coefficient mainly attribute to how well the sensor surface parallel to vertical direction, which is also the flame direction. 73

89 Sensor Flame Temperature and H.T. Coefficient Flame Temperature ( 0 C) Temp. H.T.Coefficient Heat Transfer Coefficient (w/m 2.0 C) 800 Sen68 Sen77 Sen45 Sensor Number 0 Figure Flame Temperature and Heat Transfer Coefficient Sensor Location and H.T. Coefficient H.T.Coefficient (W/m 2.0 C) H.T. Coefficient Location Factor Sen1 Sen97 Sen115 Sen119 Sen6 Sensor Number Location Factor (sensor surface direction) Figure Sensor Location Factor and Its Heat Transfer Coefficient 74

90 4.2 Characterizing Heat Induced Change in Fabric Properties Fabric optical properties play an important role in the garment protective performance, especially in exposures to intense fire environments. The propane burning torches used to produce a flash fire in the manikin chamber is turbulent jets of flame. Although these flames usually constitute a convective heat, a significant amount of thermal energy is transferred by radiation. Because of the significant radiant heat capacity, the optical properties of the fabric can markedly affect heat absorption. Treatment of fabric optical properties during the burning process is crucial to modeling garment thermal protective performance in the Pyroman system. Fabric thermal conductivity and volumetric heat capacity is the main factors governing heat transfer in fabrics. Thermal Conductivity can be determined by using a variety of methods. This can be divided into two regions. One is normal condition in which the thermal conductivity is constant while temperature changes; the other is high temperature range in which the thermal conductivity is a function of temperature. Guarded hot plate and the Thermal Properties Test Fixture (TPTF) are the methods used in normal conditions (comfort conditions). In high temperature conditions, the differential scanning calorimeter (DSC) and model calculation are often used to determine these fabric properties. Thermal conductivity in absolute dry conditions can be calculated by a simplified model which uses a weighted sum of the contributions from the solid fibers and the air, as well as the contribution of radiation heat transfer between fibers. The calculation results of this model introduced in Chapter 2. Advancing the accuracy of the model used to predict heat transfer in fabric exposed to intense heat required a knowledge of heat induced change in fabric thermal conductivity and volumetric heat capacity. This research used a parameter estimation approach to quantify fabric thermal properties in dynamic heat exposure. 75

91 4.2.1 Parameter Estimation Method Although fabric thermal properties can be measured using DSC and TGA [91], parameter estimation theory is considered to be the best way to estimate thermophysical properties from dynamic experiments [116]. The parameter estimation approach uses experimental measurement and model errors in a statistical context and provides useful information to optimize the experiment. Dynamic methods present attractive choice because the experiments required to generate needed data are reasonably short while producing a significant amount of information on the thermal behavior of the material. However, dynamic techniques require more complex modeling of physical phenomenon and a sophisticated capacity to process signal. Since estimation process embodied by the technical typically use inverse solution (analytical or numerical), the unavoidable presence of errors in the measured data may have a detrimental effect in the final estimates, owing to the intrinsic illposedness of any inverse problem. Therefore, in addition to precision measurements, the key to a reliable estimation of fabric thermophysical properties is the choice of the algorithm used. Measured data must be analyzed in a statistical context in order to estimate not only the thermophysical properties, but also the related variances, which are as important as the unknown properties. To optimize the estimation process, the design of the dynamic experiment and of the estimation algorithm must be integrated, and continuously adjusted to improve the accuracy of the estimates [116]. This study conducted experiments to estimate heat induced change in thermal conductivity and volumetric heat capacity of Kevlar/PBI and Nomex ШA fabrics in short duration (4 second) and high flux exposures. In these experiments, the skin simulant sensor was used to measure heat flux coming through the test fabric and fabric surface temperature. Thermocouple is employed to measure fabric surface temperature change during short exposure. These temperature and heat flux were an input to a parameter estimation code [117] (Figure 4-13). 76

92 Figure Schematic Diagram of the Transient Experiment for Parameter Estimation The parameter estimation technique involves minimizing a weighted sum of squares criterion, S, that involves the measured and the calculated from the model. T (β ) ji is a function of the estimated thermal parameters, the thermal conductivity and volumetric heat capacity. The vector β contains the true parameter values, and the estimated values of the parameters are found by minimizing S through the use of a modified Gauss method. S jt = ( Y ji T ( β ) ji ) j1 n i= 1 2 J T is the number of the thermocouples, Y ji represents the measured temperature at location X j. 77

93 4.2.2 Estimation of Protective Fabrics Thermal Properties In the experiment, the heat intensity is set at the order of kw/m 2 and the duration is around 4-6 second. The transient heat flux behind the fabric is measured by skin simulant sensor using Duhamel s theorem. Starting from a uniform temperature, normally at 25 0 C, 65% RH, the desired thermal exposure is imposed, and from the measured response of the specimen in temperature and heat flux, the effective thermal conductivity and volumetric heat capacity of the fabric are simultaneously reconstructed, as a function of time, by solving the associated inverse non-linear heat conduction problem. An example of this estimation of fabric transient thermal conductivity and volumetric heat capacity is shown in Figure The Fabric is Kevlar/PBI with a weight of 254g/m 2. The experiments measured data are presented in Figure The results show a tendency of decrease for both thermal conductivity and volumetric heat capacity during exposure. This agrees with the research work of Barker and Shalev [21] about weight loss and density change during bench top exposures. The estimated transient thermal properties of these two protective fabrics are used as one of the input of the model. 400 Fabric Temperature and Skin Simulant Sensor Flux and Temperature vs. Time 50,000 Temp ( 0 C) Fabric Surface Temp Skin Simulant Sensor Temp Heat Flux 45,000 40,000 35,000 30,000 25,000 20,000 15,000 10,000 5,000 Heat Flux (w/m 2 ) Time (sec) Figure Temperatures and Heat Flux Profiles during Exposure Experiment 78

94 Following considerations are suggested by practical experience in transient measurements on fabric material. It is known that the fabrics contain moisture and large amount of trapped air. During sudden exposure to intense heat, the residual moisture causes sudden process of evaporation, migration and condensation. These changes along with the weight loss contribute to change of thermal properties. The estimated values here during high heating rate and temperature are not the real thermal conductivity and volumetric heat capacity based on assumed pure conductive model. These effective estimated values only represent the heat transfer process in this specific exposure condition including all the phenomena occurred Thermal Conductivity (W/m.K) Thermal Conductivity Volumetric Heat Capacity Power (Volumetric Heat Capacity) Power (Thermal Conductivity) Volumetric Heat Capacity (J/m 3.K*10 6 ) Temperature ( 0 C) Figure Estimated Transient Thermal Properties of Kevlar/PBI during 4 Second Exposure The difference between the thermal conductivities of the Nomex IIIA and Kevlar/PBI are expected to be very small [37]. So the estimated properties can be used to represent the effective thermal conductivity of both materials. Different weight fabrics have different volumetric heat capacity due to their different densities. Different weight 79

95 of fabrics volumetric heat capacity can be obtained by applying density to the results shown in Figure Protective Garments Air Gap Distribution in Pyroman Body Air gaps entrapped between protective garments and human body are considered as one the major factors that slow down the heat transfer to skin when exposed to intense heat conditions. Different air gaps between the fabric and skin will impact the burn predictions as shown in section 4.2. The determination of air gap s distribution and quantification of air gap size at each sensor location of different garments is very crucial to model Pyroman. These air gap sizes are one of input of the model. To understand its sizes and distribution is also very important to improve garment thermal protective performance Three-Dimensional Body Scanning Technology Three-dimensional body scanning technology [118] developed by [TC]² includes a white light-based scanner and proprietary measurement extraction software. The scanner captures hundreds of thousands of data points of an individual's image, and the software automatically extracts dozens of measurements. Detailed information about this scanning system is attached in Appendix Air Gaps Determination of Protective Garments in Pyroman The principal of this air gap determination is superimposing the extracted data from the nude and dressed Pyroman (Figure 4-16). First data collected from the nude Pyroman which all the 122 sensors are highlighted by colored dots. The arms are 80

96 carefully positioned to make maximum surface exposure. This nude manikin scanning data will used as a base data. Then the dressed Pyroman scanning is performed with the exactly same manikin position. The next step is superimposing the dressed manikin data to the base data (nude scanning data) using Geomagic software. A data alignment and color management are carried out for air gap measurement both local (specific sensor) and global (contour along specific position horizontally or vertically). These whole measuring processes are illustrated in Figure 4-17, Figure 4-18 and Figure Figure Air Gap Determinations by Superimposing Dressed and Nude Body Scanning Data A nude and dressed Pyroman automatically extracted measurements are shown in Figure The differences between the nude and dressed at some typical measurement points are given in Figure The size of dressed garments is 42 (chest measurement in inches) coverall in the VF Corporation Deluxe style. The Deluxe style is a typical, high quality industrial coverall with standard pocketing. 81

97 Figure Dressed Pyroman 3D Body Scanning Image Figure Superimposed 3D Body Scanning Data Showing the Sensor Positions 82

98 Figure Slicing the Body at Specific Sensor Position, Measuring Air Gap Figure Nude and Dressed Pyroman 3D Body Measurements Image 83

99 The data of Figure 4-20 show that the air gaps of the coveralls dressed in manikin do not evenly distributed. Some areas like shoulder, knee and upper back do not hold much air gap, while other areas such as waist, thigh show a lot of air gaps between. This is also can be seen in Figure 4-22 which shows the average of arm, leg, front and back of different sizes of Kevlar/PBI coverall. 3D Image Measurements of Pyroman Body v.s. Dressed NeckCirc ChestCircumference WaisCir HipsCir ThighMax maxbicep maxwrist Backneck10 Shoulder14 AcrossChest AcrossBack Ftwaist18 Bkwaist19 AbdomenFT FtNecktoWaist FtShoulderSlope BackofNecktoWaist BkShoulderSlope waistheight hipheight KneeHeight OutseamLt bkwaistheight39 VerticalRiseRight InseamLt KneeMax CrotchLengthFull ShoulderLengthRt shirtsleevelength collar Measurements (inches) Dressed Nude Typical Positons Figure Air Gap Distribution between Garment and Manikin at the Typical Positions In order to determine the air gap distributions of different garment and size, the coverall Kevlar/PBI with size 44, size 42, and size 40 are scanned in Pyroman body. The air gaps are not evenly distributed along the Pyroman body. The Figure 4-22 shows the average of air gaps in arm, legs front and back area of these different garment sizes. It is as expected that among all these three sizes legs hold the largest air gaps and arms hold the least. In garment size 44 large air gaps are observed in the back. It does not show much difference in garment size 40 among arms, front and back. In garment size 42 which fits the Pyroman body shows the front area holds more air gaps than back and arm area. These different air gap sizes of different sized garment could change the skin burn damage pattern in the exposures because of the role of air gap in heat transfer process in protective system. 84

100 Figure Different Sized Garments and Their Air Gaps in Pyroman Body The different coverall (Nomex and Kevlar/PBI) with the same garment size 42 and pattern are examined and their air gap distributions are illustrated in Figure It is shown that Kevlar/PBI protective coverall holds larger average air gap than Nomex ШA protective coverall. This is mainly attributable to the different fabric drapability and stiffness. It is known that Nomex garment shrinks when fabric temperature reaches C degrees during 4 second exposure [48]. The shrinkage during exposure reduces the air gap and increases heat transfer rate. In order to examine the air gap change, a 4 second exposed Nomex ШA coverall with size 42 is scanned using three-dimensional body scanning system and the measurement results are given in Figure A sharp decrease in average air gap is observed, especially in legs, which more than 90% air gap is reduced. The overall garment shrinkage for 4 second exposure is more than 50% on average. The reduced air gap will increase the heat transfer and a worse skin damage is anticipated. 85

101 Figure Different Garments with Same Size and Pattern Shows Different Air Gap Figure Comparisons of Air Gaps Before and After 4sec Exposure To summarize the air gap distribution of the different sized protective coveralls between the manikin in this study using the three-dimensional body scanning technology, 86

102 a 3D chart is drawn shown in Figure These air gap measurements include Kevlar/PBI coverall with three sizes (size 44, size 42, and size 40) and Nomex 42 coverall before and after 4 second heat exposure. The knowledge gained through these results is very crucial in building this model. First, different garment size shows different air gap size distribution over the Pyroman body; second, in all the garments leg area hold the largest average air gaps and arm area hold least; third, the garment fabric drapability and stiffness change the air gap distribution between the coverall and Pyroman body; and finally Nomex ШA (6.5oz/yd 2 ) garment shrinkage after a 4 second exposure reduces more than 50% air gaps on average and as much as 90% in the manikin leg Arm Front Back Leg Air Gap Size (mm) Leg 2.92 Back 0.00 Front Nomex 42 Burned Kevlar /PBI 40 Nomex 42 Kevlar /PBI 42 Kevlar /PBI 44 Arm Figure Air Gap Distribution of Different Sized Coveralls Dressed in Pyroman The air gaps of these coveralls at the sensor locations are listed in Appendix 7. 87

103 4.3.3 Ease Measurement Method In this research, ease measurements were performed before and after manikin burning test to measure the changes before and after exposure as a supplement for air gap determination. This method measures the excess fabric at specific points on garment from dressed manikin. Detailed information can be found in Appendix 8. 88

104 CHAPTER 5 NUMERICAL RESULTS AND MODEL EVALUATION In this section, numerical model predictions are given at several sensor locations to show temperature distribution in the protective fabric, air gaps between the coverall and the manikin, and also in human skins. Skin burn damages and integral value (Ω) change during exposure and cooling period are illustrated at these specific sensor locations on the manikin. In order to validate the established numerical model, more than 40 manikin burn tests ( including calibration burn) were performed to cover varying exposure times (from 3 second to 5 second), configurations (with and without underwear) and different kinds of coveralls (Kevlar/PBI and Nomex ШA) Garments Used in This Study Two protective garments are selected in this research: Kevlar/PBI coverall and Nomex ШA coverall. These coveralls are typical, high quality industrial coverall with standard pocketing made by VF cooperation in Deluxe style [Figure 6-9] Garments Preparation All garments used in this research are laundering 5 times by Cintas Laboratory using their standard industrial laundering procedure before manikin test. This will be similar to ASTM but temperature and detergents will be modified to AATCC 135. The garments are conditioned at 21 ± 3 0 C, 65 ± 5% relative humidity Garments Fabric Thickness Fabric thickness is a complex textile property since it is known to depend on the pressure on local applied in the measurement [18]. Pressure sensitive thermal effective 89

105 thickness must be carefully considered to assess the thermal protective performance of fabric. Figure 5-1 and Figure 5-2 show how the fabrics of Kevlar/PBI garment (4.5 oz/yd 2 ) and Nomex ШA garment thickness varies as a function of the pressure of the measurement. These data show that the thickness of these fabrics (laundering 5 times ) varies significantly as shown in the figures: from approximately 1.2 mm to 0.6 mm from low to high pressure. Most of the thickness change occurs at relatively low applied loads (< 10g/cm2). This is an important consideration since fabric thickness is known to be clearly correlative with entrapped air. Therefore, fabric thickness, especially the effective thermal thickness, are expected to be important determinants of thermal protective performance Garment Fabric Kevlar/PBI 4.5 oz/yd 2 KES-FB3 Compression Test 50 EMC = 59.41% LC = RC = 26.93% WC = Pressure Pressure (gf/cm 2 ) Garment Fabric thickness (mm) Figure 5-1. Kevlar/PBI Garment Fabric Thickness as a Function of Applied Load (Measured on KES compression test) 90

106 Nomex 6.5 oz/yd 2 Garment Fabric KES-FB3 Compression Test EMC = 53.33% LC = RC = 28.82% WC = Pressure Pressure (gf/cm 2 ) Fabric Thickness (mm) Figure 5-2. Nomex ШA Garment Fabric Thickness as a Function of Applied Load (Measured on KES compression test) Garment Fabric Thermal Properties The two kinds of protective fabrics thermal conductivity and volumetric heat capacity were estimated using parameter estimation method (see 4.2.2). The emissivity of 0.9 and transmissivity of 0.01 of these two fabrics were chosen for each of these two fabrics [91] Summary of Garments and Fabric Properties Used in the Model The garments type and fabric thermal properties used in present model are summarized in Table 5-1. Table 5-2 shows the cotton underwear thermal properties used in this research. 91

107 Table 5-1. Test Garments Style and Fabric Property Garment Kevlar/PBI Coverall Nomex ШA Coverall Garment Fabric 93% meta-aramid, 60% para-aramid 5% para-aramid and 40% PBI and 2% anti-static fiber Fabric thickness: 0.7 mm 0.8mm Weight 4.5 oz/yd 2 (153g/m 2 ) 6.0oz/yd 2 (203g/m 2 ) Fabric Thermal Conductivity Transient From (W/m.K) Transient From (W/m.K) Fabric Volumetric Heat Capacity Transient See Figure 4-16 Range (J/m 3.K*10 6 ) Transient See Figure 4-16 Range (J/m 3.K*10 6 ) Extinction Coefficient γ 0.8 Deluxe style coverall Deluxe style coverall Garment type and size 42 Size 42 size (see Figure 6-12) (see Figure 6-12) Manufacturer VF Corporation VF Corporation Table 5-2. Thermal Properties of Cotton Underwear Property Symbol Value Weight W(g/m2) 146 Thermal conductivity k underwear 0.07 Volumetric heat capacity ρ Cp underwea (J/ m 3 K) 2.73X10 5 Thickness L underwear (mm) 0.95 Emissivity ε underwear

108 5.2. Fire Boundary Conditions Usually several calibration burns are required to perform before and after manikin garment tests to confirm the average heat flux of one hundred twenty-two sensors and their standard deviation are within the range. Table 5-3 shows the average heat flux of the 11 days before and after garment manikin tests. An average heat flux values of each individual sensor among these calibrations are used as the fire boundary conditions for the model. The average values are listed in Appendix 2. Not only good consistency on the average heat flux is observed among these calibration burns, but also the individual sensors in Pyroman show excellent consistency (Figure 5-3). Flame temperature profiles at some selected sensor positions are listed in Appendix 10. Table 5-3. Calibrated Heat Flux Values for Pyroman Before and After Burns Day Before (cal/cm 2 sec) After (cal/cm 2 sec) Day * Day Day Day * Day Day Day * Day Day * Day Day * No calibration data collected after day of exposure 93

109 Sensor Heat Flux vs Different Days Sensor (Arm)4 Sensor (Back)99 Sensor (Front) 89 Sensor (Leg) 53 Heat Flux (cal/cm 2 sec *10-2 ) Day Figure 5-3. Sensor Flux Values on Different Calibration Days 5.3. Garments Air Gap Size Determination The air gap sizes between the garment and the manikin at each sensor location are determined using 3D Body Scanning Technology. The values of Kevlar/PBI coverall size 42, Nomex ШA coverall size 42, and burned Nomex ШA coverall size 42 (exposed in 4 sec exposure manikin test) are given in Appendix Model Results and Predictions In this section, two sensor positions (sensor #60 and #sensor 56) in the Pyroman body are selected with measured air gap and flame boundary conditions. The predictions were made in 4 second exposure with average flux 2.00 cal/cm 2 sec. Some other parameters used in model prediction are given in Table

110 The temperature distributions in fabric, air gap, and skin in different time for a 4 second exposure for sensor #60 (air gap: 3.64mm) and sensor #56 (air gap 1.2mm) are illustrated in Figure 5-4, Figure 5-5 respectively. Comparing the temperature of the front and back of the fabric in 4 second exposure, it shows almost C temperature differences and the temperature distribution in the fabric is not linear because of the fabric transient thermal conductivity and volumetric heat capacity, as well as the in-depth radiation. About C temperature drop in the air gap is also demonstrated in these two cases. Table 5-4. Parameters Used in Model Predictions Conditions Symbol Value Temp. of ambient air T amb K Emissivity of hot gases ε g 0.02 View factor between fabric and ambient F fabric-ambient 0.89 View factor between fabric and human skin F fabric-skin 1.0 Total time 4.0 sec Time step 0.05 sec Error 1.0x10-6 Relaxation factor α 1.0 Maximum iteration

111 Fabric 0.7mm Time 1 sec Time 2.0 sec Time 3.0 sec Time 4.0 Sec Time 1.5 sec Time 2.5 sec Time 3.5 Sec Temperature ( 0 C) Air gap: 3.64mm Skin layers Distance (mm) Figure 5-4. Temperature Distribution in Fabric Air-gap Skin Model during 4 Second Exposure with 3.64mm Air-gap (Sensor #60 location) Fabric 0.7mm Time 1 sec Time 2.0 sec Time 3.0 sec Time 4.0 Sec Time 1.5 sec Time 2.5 sec Time 3.5 Sec Temperature ( 0 C) Air Gap 1.2mm Skin Layers Distance (mm) Figure 5-5. Temperature Distribution in Fabric Air-gap Skin Model during 4 Second Exposure with 1.2mm Air-gap (Sensor #56 location) 96

112 The temperature histories in the skin model for these two sensor locations are shown in Figure 5-6 and the corresponding integral values (Ω) are given in Figure 5-7 and Figure 5-8, respectively. The peaks of these temperature profiles for second and third degree burn all appear behind 4 second. This indicates that the energy stored energy in fabric goes on transferring to skin after exposure. For the case of sensor 60, the temperatures for second degree burn rises above 50 0 C while the temperature for third degree burn never reaches 50 0 C. In this case the second burn time occurs at second and no third degree burn which is shown by their Omega values in Figure 5-7. In the case of sensor 56, however, the temperature for second degree burn rises above 65 0 C and 50 0 C for the third degree burn. Consequently, as shown by their Omega values (Figure 5-8), the second degree burn occurs at 4.58 second and third degree burn occurred at relatively longer time, second Skin 2nd for Sensor 56 Skin 3rd for Sensor 56 Skin 2nd for Sensor 60 Skin 3rd for Sensor 60 Temperature ( 0 C) second exposure Time (sec) Figure 5-6. Temperature Profiles in Skin Model for a 4 Second Exposure 97

113 Henriques' Integral Value for 2nd burn second exposure Integral value for 2nd burn Integral value for 3rd burn For second degree burn, Temperature = 44 C at time = 3.17 sec Second degree burn occurs at time = sec For Third degree burn, Temperature = 44 C at time = 7.60 sec No third degree burn occurs Henriques' Integral Value for 3rd Burn Time (sec) 0.00 Figure 5-7. Omega Integral Value ( Sensor #60) in Pyroman for a 4 Second Exposure Henriques Integral Value for 2nd Degree Burn second exposure Second degree burn occurs at time 4.58 sec Third degree burn occurs at time sec Henriques' Integral Value for 2nd Burn Henriques' Integral Value for 3rd Burn Henriques Integra Value for 3rd Burn Time (sec) Figure 5-8. Omega Integral Value after (Sensor #56) in Pyroman for a 4 Second Exposure 0 98

114 The added one layer cotton underwear reduces the heat transfer to skin and lowers the temperature rising in skin layers. In order to compare the temperature change in the skin model, a comparison is made by adding one layer to previous sensor case, sensor 56. The results are given in Figure 5-9. Without underwear the time to second degree burn is 4.58 second, while with underwear the time moves to 7.35 second and no third degree burn occurs. Figure 5-10 shows the lowered Henriques integral value for third degree burn which indicates no burn happened. This is mainly attributed to the extra insulation layer slowing down heat transfer to skin and reduced temperature rising in skin. Skin 2nd for Sensor 56 Skin 3rd for Sensor 56 Skin 2nd for Sensor 56 with underwear Skin 3rd for Sensor 56 with underwear Without Underwear: 2nd degree burn occurs at 4.58 sec 3rd degree burn occurs at sec Temperature ( 0 C) With Underwear: 2nd degree burn occurs at 7.35 sec No 3rd degree burn Time (sec) Figure 5-9. Temperature History in Skin with Underwear in 4 second Exposure 99

115 Integral Value for 2nd Burn Integral Value for 3rd Burn Time (sec) Henriques' Integral Value for 2nd Burn Henriques' Integral Value for 2nd Burn with underwear Henriques' Integral Value for 3rd Burn Henriques' Integral Value for 3rd with underwear Figure Omega Value with and without Underwear in 4 second Exposure 5.5. Model Evaluation The numerical model developed by this research was used to estimate the burn injuries expected to the Pyroman manikin clothed with one layer protective coveralls made with Kevlar/PBI and Nomex ШA, with or without cotton underwear. The garment type and fabric thermal properties are shown in Table 5-1. Cotton underwear thermal properties are given in Table 5-2. In order to validate the established numerical model, more than 40 manikin burn tests ( including calibration burn) are performed to cover varying exposure time (from 3 second to 5 second), configuration (with and without underwear) and different kinds garments (Kevlar/PBI and Nomex ШA). These garment manikin tests results are compared with the numerical model in terms of predictions of second degree burn, third degree burn and total burn percentage. Three replicates are conducted for each garment test conditions. 100

116 These garments manikin tests covered 11 days. Table 5-2 shows the average heat flux of the 11 days before and after tests. An average of each individual sensor of these calibrations is used as input of the model for fire boundary conditions One Layer Garments without Underwear Figure 5-11 shows the comparison of manikin test results and numerical model predictions of Kevlar/PBI coverall after three and four second exposure. The burn location distribution of manikin test and model prediction are also compared in Figure 5-12 and Figure 5-13 for three second and Figure 5-14 and Figure 5-15 for four second exposure. From these data and location distributions it is clear that this numerical mode did a good job in predicting burn injuries of protective coveralls exposed in three and four second intense heat condition rd % 2nd % Total % Burn Prediction (%) Total % Mankikin Three Second Model Three Second Mankikin Four Second Model Four Second 2nd % 3rd % Figure Comparison between Kevlar/PBI Garment Manikin Test and Numerical Model Prediction without Underwear 101

117 Figure Second and Third Degree Burn Location of Pyroman Test for Kevlar/PBI Coverall in 3 sec Exposure Figure Second and Third Degree Burn Location Predicted by Numerical Model for Kevlar/PBI Coverall in 3sec Exposure 102

118 Figure Second and Third Burn Location of Pyroman Test for Kevlar/PBI Coverall in 4 sec Exposure Figure Second and Third Burn Location Predicted by Numerical Model for Kevlar/PBI Coverall in 4sec Exposure 103

119 For Nomex ШA coverall, the comparison of manikin test results and model predictions are shown in Figure The test data and model predictions are very close for four second exposure; for three second exposure, however, it shows a little difference in both 2 nd degree burn and 3 rd degree burn. This is expected because the fire boundary conditions used in this model are from four second calibration burns and the difference in heat transfer coefficients for three second duration may be a factor to this deviation. The burn location for three second is listed in Figure 5-17 (manikin) and Figure 5-18 (model). For four second exposure it is given in Figure 5-19 (manikin) and Figure 5-20 (model). These burn locations of manikin test and model prediction show a similar pattern rd % 2nd % Total % Burn Prediction (%) Mankikin Three Second Model Three Second Mankikin Four Second Model Four Second 3rd % 2nd % Total % Figure Comparison between Nomex ШA Garment Manikin Test and Numerical Model Prediction without Underwear 104

120 Figure Second and Third Burn Location of Pyroman Test for Nomex ШA Coverall in 3 sec Exposure Figure Second and Third Burn Location Predicted by Numerical Model for Nomex ШA Coverall in 3 sec Exposure 105

121 Figure Second and Third Burn Location of Pyroman Test for Nomex ШA Coverall in 4 sec Exposure Figure Second and Third Burn Location Predicted by Numerical Model for Nomex ШA Coverall in 4sec Exposure 106

122 One Layer Garments with Underwear The comparison of manikin test and model prediction for Kevlar/PBI and Nomex ШA coverall with cotton underwear are illustrated in Figure 5-21 and Figure These comparisons indicate that the numerical model is able to predict burn damage under one layer protective garments with underwear. Recalling the assumptions we made for this additional cotton underwear in chapter 3 that no air gap between the underwear and skin, this may be the main factor cause the difference in five second exposure. The burned sensor locations of these with underwear are listed in Appendix rd % 2nd % Total % Burn Prediction (%) Mankikin Four Second Model Four Second Mankikin Five Second Model Five Second 3rd % 2nd % Total % Figure Comparison between Kevlar/PBI Garment Manikin Test and Numerical Model Prediction with Underwear 107

123 rd % 2nd % Total % Burn Prediction (%) Mankikin Four Second Model Four Second Mankikin Five Second Model Five Second 3rd % 2nd % Total % Figure Comparison between Nomex ШA Garment Manikin Test and Numerical Model Prediction with Underwear Model Evaluation Summary Figure 5-23 and Figure 5-24 show the summary results of these manikin test results and model predictions. These results lead us to believe that this numerical 1D model successfully predict thermal protective performance in terms of burn injury of Kevlar/PBI and Nomex ШA protective coveralls exposed to intense flash fire conditions (2.00 cal/cm 2 sec ) at varying exposure time. This model can differentiate varying exposure time and the clothing configuration of with and without cotton underwear. Garment shrinkage during exposure can be taken into account by considering the changes of air gap size between garment and manikin body. Other shrinkage parameters such as shrinkage time and rate can also be involved into model. The success of modeling these Nomex ШA and Kevlar/PBI protective garments at varying exposure time provides us a very useful tool to understand and explore protective garment systems and manikin thermal protective evaluation system. 108

124 Figure Manikin Tests and Numerical Model Results for Kevlar/PBI Figure Manikin Tests and Numerical Model Results for Nomex ШA 109

125 CHAPTER 6 PARAMETRIC STUDY The primary motivation for developing the model is to engineer better protective garments. This can be accomplished by using the model to perceive the effects of fabric material and manikin test in predicting a potential burn injury. It can even be used to understand complicated skin and burn evaluation models. These manikin test parameters can be categorized into four groups: fabric thermophysical properties, garment factors, factors associated with manikin test itself and skin model. 6.1 Influence of Fabric Thermophysical Properties The thermophysical properties of the garment are of importance to thermal protective performance. The main thermophysical properties govern heat transfer through the fabrics are thickness, thermal conductivity (k) and volumetric heat capacity (ρc p ). Fabric optical properties including emissivity (є) and transmissivity (τ) also influence heat transfer, especially exposed in intense flash fire environments. A serious of parametric studies were performed to determine the effects of varying fabric thermal properties on the thermal protective performance predicted by this model, a series of parametric studies are performed. The range of values for each of the properties used in the parametric studies is given in Table 6-1, while the results of the model numerical results are discussed below. Table 6-1. Range of Values of Thermophysical Properties of Garment Fabric 110

126 Fabric Thickness Fabric Thermophysical Properties Property Range Thickness (mm) Thermal Conductivity (W/m 0 C) Volumetric Heat Capacity (J/m 3 K * 10 6 ) Emissivity Transmissivity As mentioned in 5.1.2, fabric thickness is simple nominal values and will change after laundering and during exposure. In order to examine the effect of thickness on garment protective performance, the thickness values varied from 0.3 to 2.4mm with the input into numerical model. The commonly used protective fabric thickness is about mm. Figure 6-1 shows predicted relationship between the fabric thickness and garment protective performance. The main parameters for this prediction are given in Table 6-2. Skin model, burn evaluation model and others relate to this model are the same as used in chapter 5. Table 6-2. Model Parameters for Garment Thickness Study Model Parameters Average Heat Flux 2.00 cal/cm 2 sec S.D Fabric Weight 203g/m 2 Transient, see Figure 4-16 Fabric Thermal Properties And Table 5-1 Garment Coverall Deluxe Style The same as Kevlar/PBI Garment Air Gap Sizes Coverall size 42 Garment Size 42 Burning Time (sec) 4 111

127 90 80 Burn Prediction (%) nd Burn % 3rd Burn % Total Burn Prediction% Fabric Thickness (mm) Figure 6-1. Relationship between Burn Damage Predicted by Model and Fabric Thickness Figure 6-1 shows the expected decrease in total burn with fabric thickness. Three regimes can be distinguished. When fabric thickness is below 0.6 mm and lager than 0.9, the total burn does not show much change. In the range from 0.6 to 0.9mm, a sharp drop is observed. Increasing the fabric thickness increases fabric thermal resistance, it can be expected to increase the temperature difference of back side of garment material and the side exposed to the flash fire, thus leads to lower fabric back side temperature and a decrease in the rate of energy transfer across the layers of the fabric and the skin. This is expected since the radiation heat transmission is the 4th power of absolute temperature at the back of the fabric. This may explain the sharp drop in predicted burns between 0.6 and 0.9mm. Below 0.6 mm the second degree burn trades off with third degree burns. In such a case some second degree burn gets worse and change to third degree burn although total burn does not change significantly. Little addition burn injury protection is gained with single layer fabrics greater than 0.9mm at this exposure condition. Commercial available single layer protective fabric thickness typically varies between 0.5mm-0.8mm. It is interesting to note that the analytical model permits the 112

128 greatest influence of changes with this thickness change. Therefore, Increases in fabric thickness in this range can be expected to increase the thermal protective performance. Additional garment components, such as pockets and bands that increase thickness, will also change the protective performance Thermal Conductivity Thermal conductivity and volumetric heat capacity are the main parameters governing heat transfer in fabrics. Thermal conductivity is a heat transport property, while volumetric heat capacity is considered as thermodynamic properties. Moisture content in fabric changes both the thermal conductivity and volumetric heat capacity. For the parametric study, the thermal conductivity is allowed to vary between 0.02 and 0.24 W/m. 0 C. The predicted second, third and total body burn are shown in Figure 6-2. The main model parameters are given in Table 6-3. Skin model, burn evaluation model and others relate to this model are same as used in chapter 5. Table 6-3. Model Parameters for Fabric Conductivity Study Model Parameters Average Heat Flux 2.00 cal/cm 2 sec S.D Fabric Weight 203g/m 2 Fabric Thickness 0.7mm Transient, see Figure 4-16 and Fabric Thermal Properties Table 5-1 Garment Coverall Deluxe Style The same as Kevlar/PBI Garment Air Gap Sizes Coverall size 42 Garment Size 42 Burning Time (sec) 4 113

129 90 80 Burn Prediction (%) nd Burn % 3rd Burn % Total Burn Prediction% Thermal Conductivity (W/m 0 C) Figure 6-2. Garment Fabric Thermal Conductivity and Predicted Pyroman Burn Estimate Fabric thermal conductivity, which is referred to as a transport property, provides an indication of the rate at which energy is transferred by the diffusion process. The larger thermal conductivity, the more chances to increase the rate of heat transfer. At specific exposures, this will cause the temperature of the back of the garment rising faster, hence increasing the energy transfer between the fabric and skin. The influence of fabric thermal conductivity on manikin estimated burn injury is shown in Figure 6-2. The model predicts a sharp increase in total burn as well as in second and third degree burn when the fabric thermal conductivity ranges from W/m 0 C. Beyond 0.8 W/m 0 C, the model predicts the total burn relatively constant, although third degree burn is predicted to increase as the fabric conductivity increase Volumetric Heat Capacity Volumetric heat capacity is the product of fabric density (ρ) and specific heat (C p ) which it is a measure of the ability of a material to store thermal energy. For this parametric study, the fabric volumetric heat capacity is allowed to vary between 0.2 x

130 - 2 x 10 6 J/m 3 K. The main parameters for this prediction are listed in Table 6-4. Skin model, burn evaluation model and others relate to this model are the same as used in chapter 5. Table 6-4. Model Parameters for Fabric Volumetric Heat Capacity Study Model Parameters Average Heat Flux 2.00 cal/cm 2 sec S.D Fabric Weight 203g/m 2 Fabric Thickness 0.7mm Transient, see Fabric Thermal conductivity Figure 4-16 Garment Coverall Deluxe Style The same as Kevlar/PBI Garment Air Gap Sizes Coverall size 42 Garment Size 42 Burning Time (sec) 4 The analytical model prediction indicates and inverse connection between fabric volumetric heat capacity and the estimated body burn (Figure 6-3). Total burn prediction is predicted to hold relatively constant below volumetric heat capacity of 0.7 x 10 6 J/m 3 K, while second and third degree burn trade off each other. Between 0.8 x 10 6 and 1.4 x 10 6 J/m 3 K, a markedly decrease in estimated body burn is observed. As the capacity of the fabric to store energy increases, backside temperature of fabric decrease as the result of less energy transferred through the fabric. Moisture content in fabrics can significantly change volumetric heat capacity; hence it can significantly influence thermal protective performance. 115

131 90 80 Burn Prediction (%) nd Burn % 3rd Burn % Total Burn Prediction% Volumetric Heat Capacity (J/m 3 K * 10 6 ) Figure 6-3. Effect of Fabric Volumetric Heat Capacity and Predicted Thermal Protection Emissivity Emissivity is the ratio of the radiant energy emitted by a surface to the radiant energy emitted by a blackbody at the same temperature. Fabric emissivity depends strongly on the nature of surface, which is influence by the method of fabrication, finishing, thermal cycling, and chemical reactions with the environment [120]. This optical parameter is very important in heat transfer in air gaps between fabric surface and skin, especially in intense thermal environments where the dominate mode of heat transfer is by radiation in the air gap. In this model for fabric Kevlar/PBI and Nomex, an emissivity of 0.9 is chosen. This agrees with Morse, et al. [31] who measured values between 0.88 and 0.91 for virgin and charred Nomex and Kevlar/PBI fabrics. In figure 6-4 an emissivity change from 0.2 to 1 is studied in term of burn prediction by this model. The main parameters in the model are given in Table

132 Table 6-5. Numerical Model Setup Model Parameters Average Heat Flux 2.00 cal/cm 2 sec S.D Fabric Weight 203g/m 2 Fabric Thickness 0.7mm see Fabric Thermal Properties Table 5-1 in Kevlar/PBI Garment Coverall Deluxe Style The same as Kevlar/PBI Garment Air Gap Sizes Coverall size 42 Garment Size 42 Burning Time (sec) 4 Fabric Emissivity vs. Garment Thermal Protective Performance nd Burn % 3rd Burn % Total Burn Prediction% Burn Prediction (%) Emissivity Figure 6-4. Effect of Fabric Emissivity on Garment Protective Predictions Fabric emissivity plays an important role in garment protective performance in flash fire environments. The reason for this is that the dominated heat transfer mode in the air gap between the fabric and skin is radiation. The energy transferred through this mode is proportional to the fabric emissivity. Lower emissivity value will decrease the rate of heat transfer between fabric and skin. Fabric finishing, dyeing process may 117

133 dramatically change its emissivity, for example, coating with aluminized or some metal material may reduce fabric emissivity from 0.9 to 0.5. As illustrated in Figure 6-4, the fabric emissivity change mainly increase the second burn prediction Transmissivity In this model, an extinction coefficient is introduced to represent in-depth absorption of incident radiation as discussed in chapter 3. Transmissivity measured from these two kinds of protective fabrics are used to calculate the extinction coefficient. The fabric transmissivity here represents the ability of absorption of incident radiation. It is a property of fabric structure dependent. For this model a transmissivity of 0.01 is chosen for each of the two fabrics. A range from to 0.5 is used to study its influence on garment protective performance predicted using this model. The main parameters associated with the model are given in Table 6-6. The results indicate that fabric transmissivity does not obviously change the burn prediction. It does, however, influence the time to second and third degree burn (Table 6-7) Table 6-6. Model Parameters for Fabric Transmissivity Study Model Parameters Average Heat Flux 2.00 cal/cm 2 sec S.D Fabric Weight 203g/m 2 Fabric Thickness 0.7mm see Fabric Thermal Properties Table 5-1 in Kevlar/PBI Garment Coverall Deluxe Style The same as Kevlar/PBI Garment Air Gap Sizes Coverall size 42 Garment Size 42 Burning Time (sec) 4 118

134 Table 6-7. Fabric Transmissivity Influence on Time to Body Burn Predicted by Model Fabric Transmissivity Influence on Time to 2nd and 3rd Degree Burn Time to 2nd degree burn Time to 3rd degree burn Fabric Tran. Transmissivity Transmissivity Sensor Number Sensor no No no Sensor no No no Sensor no No no Sensor Sensor Initial, Ambient and Fire Distribution Influence Garment initial temperatures, ambient temperature of the chamber and generated flash fire heat flux distribution along the manikin body have large effect on thermal protective performance prediction. In ASTM F , The standard test method for evaluation of flame resistant clothing for protective against flash fire simulations using an instrumented manikin, the garment should be conditioned at least 24 hours in 20 0 C and 65% relative humidity and the calculated heat flux standard deviation is less or equal to 0.5 cal/cm 2 sec of average 2.00 cal/cm 2 sec. In this parametric study, all these parameters are examined using this model. The main parameters for this model are in Table 6-8. Table 6-8. Numerical Model Setup Model Parameters Average Heat Flux 2.00 cal/cm 2 sec S.D Fabric Weight 203g/m 2 Fabric Thickness 0.7mm see Fabric Thermal Properties Table 5-1 in Kevlar/PBI Garment Coverall Deluxe Style The same as Kevlar/PBI Garment Air Gap Sizes Coverall size 42 Garment Size 42 Burning Time (sec) 4 119

135 Fabric Initial Temperature The garment temperature range is varied between 0 0 C and 45 0 C to simulate in winter situation (temperature at 0 0 C) and warm-up conditions at which temperature is about C. The prediction results are shown in Figure 6-5. The data from model prediction indicate that garment initial temperature shows a significant influence on thermal protective performance predictions. Large fabric volumetric heat capacity is considered to be the main factor for this influence. The temperature of the back of the garment with a lower initial temperature is rising slower than the higher one during the same exposure. This decreased temperature rising in the back of the fabric slows down the heat transfer to skin. The fabric initial temperature primary leads to second degree burn prediction changes while the third degree prediction keeps relatively unchanged (Figure 6-5). Garment Initial Temperature vs Burn Prediction Burn Prediction (%) nd Burn % 3rd Burn % Total Burn Prediction% Temperature ( 0 C) Figure 6-5. Effect of Garment Initial Temperature on Body Burn Predictions 120

136 Ambient Temperature A range of 0 0 C C ambient temperature is used in model parametric study. Normally the ambient temperature of the test chamber is around 25 0 C for conditioned chamber. The results from model predictions show that the ambient temperature has a relative small effect on protection prediction offered by the garments (Figure 6-6). It should be noted that this effect is mainly on third degree burn prediction, especially when ambient temperature exceeds 35 0 C. This is mainly because the higher ambient temperature slows down the cooling rate after burn, and consequently, the slower cooling process increase the heat transfer to skin. Ambient Temperature vs Burn Prediction Burn Prediction (%) nd Burn % Total Burn Prediction% 3rd Burn % Burn Prediction (3rd %) Ambient Temperature ( 0 C) Figure 6-6. Ambient Temperature Effect on Garment Thermal Protective Prediction Same Garment and Ambient Temperature This is a case that combines garment initial temperature and ambient temperature effect. The low garment and ambient temperature is selected to simulate the severe winter conditions, while the relative high temperature is to simulate the warmed up garment in 121

137 hot environments. The ranges of these temperatures are between C 45 0 C. The main parameters used in numerical model are listed in Table 6-8. The model predictions indicate that the effects of initial garment and ambient temperature on burn prediction are significant (Figure 6-7). This influence is primary on second degree burn prediction. Lower fabric and initial temperatures slow down the back side temperature rise of the fabric, therefore, it decreases the heat transfer between the fabric and skin. Garment Fabric and Ambient Temperature vs. Burn Prediction nd Burn % 3rd Burn % Total Burn Prediction% Burn Prediction (%) Temperature ( 0 C) Figure 6-7. Garments and Ambient Temperature Influence on Body Burn Prediction Fire Distribution influence In chapter 5, some graphical methods are used to demonstrate the heat flux of 122 sensors in 4 second exposure exhibits a bell-shaped distribution normal distribution. Only two factors involve in normal distribution: the mean and standard deviation. Before the Pyroman tests, a series of calibration burns are performed to examine its average heat flux and heat flux standard deviation. The total average of 122 sensors should be 122

138 within 5% of 2.00 cal/cm 2 sec and standard deviation is required to be lower than 0.5. In this parametric study, a series of different standard deviation normal distribution data with same mean are generated to simulate different fire distribution. The range of standard deviation is between 0.14 and 0.65 and the average value is 2.00 cal/cm 2 sec. The main parameters for numerical model are given in Table 6-9. The skin model and burn evaluation model is given in chapter 5. Table 6-9. Model Parameters for Heat Flux Distribution Study Model Parameters Average Heat Flux 2.00 cal/cm 2 sec S.D Fabric Weight 203g/m 2 Fabric Thickness 0.7mm see Fabric Thermal Properties Table 5-1 in Kevlar/PBI Garment Coverall Deluxe Style The same as Kevlar/PBI Garment Air Gap Sizes Coverall size 42 Garment Size 42 Burning Time (sec) 4 123

139 Pyroman Sensor Heat Flux Standard Deviation vs. Burn Prediction nd Burn % 3rd Burn % Total Burn Prediction% 50 Burn Prediction (%) Standard Deviation Figure 6-8. Pyroman Heat Flux Standard Deviation Effect on Burn Prediction The standard deviation of total 122 sensors heat flux of Pyroman predicted by this model does not show a significant influence on third degree burn prediction. The second degree burn prediction, however, decrease as the standard deviation increasing as illustrated in Figure Garment Design and fit Factors Protective garment design involves a lot of consideration in many respects such as maintenance, air ventilation, visibility and reinforcement in some parts. Some components are added to the garment to fulfill these functions, and also these components add extra thermal protection to the garments. The air gaps between garment and skin plays an important role in protective performance. Different garment size changes the air gap size distribution (relative to the same body size), and therefore changes the protective 124

140 performance in specific exposures. Garment shrinkage during the exposure changes the air gap size distribution along the body, and damage the protective performance. In this chapter, these parameters are examined using this analytical model Garment Components The garment components include pockets, zipper, collar and waist inserts etc. The coveralls used in this research are typical, high quality industrial coverall with standard pocketing made by VF cooperation in Deluxe style. The components of these coveralls are listed in following (Figure 6-9). Concealed, two-way NOMEX taped brass zipper, snap at top of zipper One-piece, topstitched, lay flat collar Two-piece cuff, sleeve vent, concealed snap closure Two set-in front pockets, two breast pockets Two patch hip pockets (left has concealed snap), double tool pocket Elastic waist inserts Bi-swing action back The coveralls with and without these components are modeled. The results predicted using this model are shown in Figure As these components add extra layer or layers to the local area and decrease the heat transfer, hence the odd of the local area to get burn is reduced. 125

141 Bi-swing action back Large spade-style breast pockets, double needle lockstitch set, Left pocket has pen/pencil slot Generously cut sleeve with forearm pleat Double-needle lockstitch set tool pocket on right leg Elastic waist inserts Two set-in front pockets, two breast pockets Two patch hip pockets (left has concealed snap), double tool pocket Front fly closure covers zipper with hidden snap closures at neck and waist so no metal exposed inside or outside Figure 6-9. Nomex Deluxe Protective Coverall 126

142 The garment with these components mainly reduced the second degree burn prediction for both Nomex and Kevlar/PBI compared with garment without these components as shown in Figure Garment Components Effect on Thermal Protective Performance Burn Prediction (%) PBI with Components PBI without Components Different Garment Nomex with Components Nomex without Components 2nd Burn % rd Burn % Figure Effects of Garments Components and Body Burn Prediction Shrinkage and Its Temperature Effect on Protective Performance Fabrics made of Nomex fibers shrink about 50% when exposed to flash-fire temperature [48]. Abbott s data show that Nomex shrinks dramatically when heated to temperature above C. This can be improved by blending with amounts of Kevlar. Garment shrinkage during exposure reduces the air gap and increases the heat transfer rate, hence increase the burn prediction. In order to examine garment shrinkage influence, a Nomex ШA Coverall without and with shrinkage is modeled. The main parameters associated with this numerical model are shown in Table

143 Table Numerical Model Setup Model Parameters Average Heat Flux 2.00 cal/cm 2 sec S.D Fabric Weight 203g/m 2 Fabric Thickness 0.7mm see Fabric Thermal Properties Table 5-1 in Nomex ШA Garment Garment Air Gap Sizes Coverall Deluxe Style Nomex ШA Coverall size 42 in Appendix 7 Garment Size 42 Burning Time (sec) 4 The model prediction results about effects of garment shrinkage on burn prediction during 4 second exposure are demonstrated in Figure Garment Shrinkage Influence on Burn Prediction Burn Prediction (%) Nomex with Components without shrinkage Nomex with Components with shrinkage Nomex without Components without shrinkage Nomex without Components with shrinkage Total Burn Prediction% 3rd Burn % 3rd Burn % 2nd Burn % Total Burn Prediction% Figure Influence of Shrinkage during Exposure on Burn Prediction The garment shrinkage during exposure decreases the air gaps which function as a thermal insulation layer. The decreased air gap increases the heat transfer between the fabric and skin. This air gap change mainly impact three degree burn prediction. In this 128

144 case some second degree burns are getting even worse while garment shrinks and transformed to third degree burn. If we can raise protective fabric shrinkage temperature, this can improve the garment thermal protective performance. In figure 6-12, garment fabric shrinkage temperatures are plotted against burn predictions using this model, which shows a significant improvement in protective performance. Fabric Shrinkage Temperature vs. Protective Prediction Total and Second Burn Prediction (%) nd Burn % Total Burn Prediction% 3rd Burn % Third Degree Burn Prediction (%) Shrinkage Temperature ( 0 C) Figure Fabric Shrinkage Temperatures and Protective Prediction Garment Size Different garment size changes the air gap distribution. The air gap distributions of different garment size (size 40, size 42 and size 44 of Kevlar/PBI Coverall) are determined using 3D Body Scanning technology (see Appendix 7). In order to study garment size effect on protective performance prediction, these air gap distributions of 129

145 different garment size are modeled in this research. The Table 6-11 shows main parameters used in the model for prediction. Table Numerical Model Setup Model Parameters Average Heat Flux 2.00 cal/cm 2 sec S.D Fabric Weight 203g/m 2 Fabric Thickness 0.7mm see Fabric Thermal Properties Table 5-1 in Kevlar/PBI Garment Coverall Deluxe Style Kevlar/PBI Coverall size Garment Air Gap Sizes 40, 42, 44 Garment Size 40, 42, 44 Burning Time (sec) 4 Garment Size Effect on Protective Performance nd Burn % 3rd Burn % Burn Prediction (%) PBI Size 40 PBI Size 42 PBI Size 44 2nd Burn % rd Burn % Garment Size Figure Effect of Garment Size and Thermal Protective Prediction The numerical model results of different garment size effect demonstrate a significant influence on third degree burn, especially between size 42 and 44 (Figure 6-130

146 15). The differences in air gap size of different garment size, when exposed in intense heat conditions, provide different insulation layer thickness and therefore, are expected to change heat transfer mode which subsequently increases the heat transfer rate. The increased heat transfer rate causes some 2 nd body burn to 3 rd body burn Skin Model Influence From literature review, a lot of skin models proposed to model human skin exposed to thermal hazard. They are different in layers, in layer thermal properties, initial temperature distribution and blood perfusion. The primary goal of this skin model parametric study is to investigate the influence of different skin models and its parameters on predicted skin damage Blood Perfusion The blood perfusion is important in longer thermal exposures of low heat fluxes and it should be included in skin model because of the ability of the body to react. In order to study its influence to burn prediction in flash fire conditions, a skin model with and without blood perfusion is used to predict burn damage. The model parameters used in prediction are shown in Table 6-12 and the model prediction results of some selected sensor are listed in Table The model results lead us to believe that the effect of blood perfusion on burn prediction is minor. It does not practically influence the burn prediction in terms of second and third body burn. From shown in Table 6-13, it changes the time to second and third degree burn, especially when this time is longer than 10 second. This is anticipated, because the longer time allows the blood an opportunity to carry away some of the energy before damage is sustained. 131

147 Table Model Parameters for Blood Perfusion Study Model Parameters Average Heat Flux 2.00 cal/cm 2 sec S.D Fabric Weight 203g/m 2 Fabric Thickness 0.7mm see Fabric Thermal Properties Table 5-1 in Kevlar/PBI Garment Coverall Deluxe Style The same as Kevlar/PBI Garment Air Gap Sizes Coverall size 42 Garment Size 42 Burning Time (sec) 4 Table Blood Perfusion Influence on Time to 2 nd and 3 rd Burn With Blood Perfusion Without Blood Perfusion Sensor No. Time to 2nd (Sec) Time to 3rd (Sec) Time to 2nd (Sec) Time to 3rd (Sec) no no no no Temperature Distribution in the skin One of the functions of human skin is to help regulate body s core temperature. The core temperature of the body must be maintained within a small range around 37 0 C in order to keep biochemical reactions proceeding at required rates. Under normal ambient conditions, the skin surface temperature is about C to 34 0 C. Some skin models proposed a constant initial skin temperatures of, C [88], 34 0 C [119], or 37 0 C [86]. Some other models suggest linear initial or higher order initial temperature distribution. In this parametric study, three initial temperature distributions are used. The first is constant at 37 0 C from surface to subcutaneous base, the other is linear distribution 132

148 from C surface to 37 0 C subcutaneous and the third is quadratic distribution from C surface to subcutaneous base. The main parameters associated with these model predictions are shown in Table 6-14 and the 2 nd, 3 rd and the total burns predicted by the model are illustrated in Figure Table Numerical Model Parameters for Skin Model Study Model Parameters Average Heat Flux 2.00 cal/cm 2 sec S.D Fabric Weight 203g/m 2 Fabric Thickness 0.7mm see Fabric Thermal Properties Table 5-1 in Kevlar/PBI Garment Coverall Deluxe Style The same as Kevlar/PBI Garment Air Gap Sizes Coverall size 42 Garment Size 42 Burning Time (sec) 4 The linear and quadratic temperature distributions show little effect on burn predictions, while constant temperature (37 0 C) distribution comparing to the linear and quadratic indicates a large influence on second body burn prediction ( Figure 6-14). The linear and quadratic distributions give a lower initial temperature than constant distribution and therefore at the same exposure it needs more energy to get second degree burn. The changes of these temperature distribution show little influence on third degree burn. This is expected because the temperature difference of these three distributions is very small at interface of dermis and subcutaneous. 133

149 Skin Model Initial Temperature Distribution Influence rd Burn % 2nd Burn % Burn Prediction (%) Constant Linear Quadratic Temperature Distribution Figure Effects of Skin Model Initial Temperature Distribution and Body Burn Prediction Single Layer and Multi Layers Skin Model Comparison Skin models used to predict burn damage can be divided into single layer skin model and three-layer skin model in which the thermal properties of each layer are different. In this study, different skin models are used to compare the burn prediction. The thermal properties of these models are outlined in Table 6-16 and the main parameters used in this model are listed in Table

150 Table Numerical Model Parameters for Skin Model Study Model Parameters Average Heat Flux 2.00 cal/cm 2 sec S.D Fabric Weight 203g/m 2 Fabric Thickness 0.7mm see Fabric Thermal Properties Table 5-1 in Kevlar/PBI Garment Coverall Deluxe Style The same as Kevlar/PBI Garment Air Gap Sizes Coverall size 42 Garment Size 42 Burning Time (sec) 4 Table Thermal Properties of Skin Models Single Layer Models Multi-layer Model Properties Tissue Epidermis Dermis Subcutaneous Thermal Conductivity (W/m 0 C) Volumetric Heat Capacity (J/m 3.K *10 6 )

151 Skin Models vs Brun Prediction Burn Prediction (%) Single Layer 0 Multi-layer 2nd Burn % 3rd Burn % Total Burn Prediction% 2nd Burn % 3rd Burn % Total Burn Prediction% Multi-layer Single Layer Figure Single Layer and Multi-layer Skin Model and Burn Predictions The single layer skin model indicates a higher prediction in second degree burn and lower third degree burn, while a higher total burn prediction (Figure 6-15). This is expected because in multi-layer skin model the lower thermal conductivity and higher volumetric heat capacity of epidermis than the one in single layer skin model. These difference in thermal properties leads to a slower temperature rising in the base of epidermis. Hence, multi-layer skin model is given a lower second degree burn prediction under the same exposures. However, the dermis layer in multi-layer model has a higher thermal conductivity than single layer model, this cause the heat transfer in dermis with a relatively higher rate, which leads to a higher third degree burn in multi-layer model. 136

152 CHAPTER 7 CONCLUSIONS AND RECOMMENDATIONS This research developed a numerical model for the Pyroman Instrumented Manikin Protective Garment Evaluation System. The model has been demonstrated using actual Pyroman tests with different garment materials, clothing configuration and flash fire exposure conditions. This model successfully predicted burn injures. A parametric study was performed using this model to study the effects of garment fabric thermophysical properties, garment design and fit, garment shrinkage, and initial temperature of the test garment and ambient environment. The parametric study also explored the effects of different skin models on burn injury predictions Summary The simulated flash fires generated in Pyroman chamber produce a dynamic, turbulent flame exposure. The heat fluxes measured by 122 sensors embedded in Pyroman body during 4 second exposure were found to be normally distributed with average heat intensity of 2.00 cal/cm 2 sec. Flame temperatures measured in the Pyroman chamber range between 800 and C. The estimated overall heat transfer coefficients of 122 sensors for the manikin range of W/m 2 0 C during 4 second exposure. The values of the manikin heat transfer coefficient at 122 sensor locations depend on flame temperature and sensor location. Higher flame temperature above each sensor is expected to generate large heat transfer coefficient. The complex geometry of Pyroman manikin create different conditions over the manikin, depending also on the directions with which the flame impingement. This is largely determined by the alignment of the eight fire producing torches. Manikin body locations, such as chest, back, and middle of leg, are expected to show a high heat transfer coefficients; while shoulder, thigh, and upper arm locations always exhibit a lower heat transfer coefficients. 137

153 This research demonstrated that thermophysical properties of Kevlar/PBI and Nomex ШA protective fabrics are not constant in these flash fire conditions. This research shows the specific manner in which fabric thermal conductivity and volumetric heat capacity change in these exposures decreases. Three-dimensional Body Scanning Technology was used to determine the local and overall air gap of the clothing-skin and variations on the manikin torso. The clothing-manikin air gaps were found not evenly distributed in protective coveralls dressed in Pyroman manikin. Some positions including shoulder, knee and upper back locations are close to the manikin body; other locations, such as the waist and thigh indicate a large air gap area. Leg holds the largest air gap. Arms and back location retain the least space between the manikin and coverall. Using different garment size changes the size of air gap, but not the distribution. The garment drapability and stiffness also affect the size of air gap. Nomex ШA Coverall (size 42, deluxe style, 6.0oz/yd 2 ) shrinks significantly in a 4 second exposure to an average heat flux of 2.00 cal/cm 2 sec. This fabric shrinkage lowers the air gap by more than 50% on average; while shrinkage in leg area was shown to reduce the air layer as much as 90%. Fabric thickness plays an important role in garment protective performance. Because of the inherent compressibility of the protective fabrics, it is necessary to measure the effective thickness or the thickness measured at low applied pressure, in order to predict thermal protective performance. Higher pressure thickness measurements override the effect of fibers extruding form the surface of the fabric which can affect thermal protective insulation. For a 4 second exposure to heat intensity of 2.00 cal/cm 2 sec. Garment protective performance was shown to be particularly affected by fabric thickness variations in a range of mm. This can be partially attributed to the 138

154 dominating mode of radiant in air gaps existing between fabric and skin. Therefore careful choose protective fabric thickness is an important considering in providing thermal protection. Fabric thermal conductivity and volumetric heat capacity are the most important factors governing heat transfer in protective garments. Changing these thermophysical properties can dramatically alter protective performance: Lowering thermal conductivity and increasing volumetric heat capacity improves protective performance of the garment. The effects of change in these parameters are especially critical in the range of W/m 0 C, for thermal conductivity and 0.8 x x 10 6 J/m 3 K, for volumetric heat capacity. Significantly, fabric moisture content has a large influence on both thermal conductivity and volumetric heat capacity. Increasing the air volume fraction in fabric while lowering moisture content reduces the fabric thermal conductivity and slows down rate of heat transfer. At the same time, moisture in fabric also reduces volumetric heat capacity, thus reducing the capability to store energy, improves the heat transfer rate. Which thermophysical property dominates the heat transfer in the fabrics is shown to depend on the specific conditions of thermal exposures. Fabric emissivity is shown to have large influence on thermal protective performance. Fabric emissivity is the factors solely relative to the radiation mode of heat transfer. Lowering the fabric emissivity will obviously improve thermal protection performance, especially in short time exposure with intense heat conditions. The influence of fabric transmissivity on thermal prediction performance in this study are minor, only change a fraction of second time to second degree burn and a few second to third degree burn. This research showed that, for single layer garments exposed to intense fire conditions, fabric thickness is the major factor influencing thermal protective performance. However, the fabric thermal conductivity and volumetric heat capacity are important parameters controlling heat transfer. Both of these properties are affected by 139

155 the moisture content in the fabric. Fabric emissivity can not be neglected since the radiant heat transfer occurs across the air gap existing between the skin side of heated fabric. The initial temperature of the test garment as well as test ambient temperature shows influence on thermal protective predictions, especially on the effect of predicted second burn. Lower the initial garment temperature and ambient temperature, a lower predicted burn. The influences are more pronounced for the test conditions where garment and ambient are at same temperature and exposure duration is short. This can be attributed to large volumetric heat capacity of fabric. A Bell-shape distribution is found in the 122 heat flux values measured by 122 sensors in Pyroman in 4 second exposure. This has been statistically proved to be normal distribution. With average heat flux 2.00 cal/cm 2 sec, increasing standard deviation are expected to reduce the burn prediction. This effect mainly influences the second degree burn. The research shows that garment components originally designed for other respects of considerations significantly enhances the garment thermal protective performance. For same size garments, garment made with more flexible fabrics show a small air gap between the clothing fabric and the surface of manikin body. Different sizes of the same style garment develop the same air gap distribution, but with different sizes of air gap. Air gap between the garment and manikin body have a significant effect on burn protection in flash fire exposure, especially on third body burns. It should be noted that, however, too large air gap sizes may increase the possibility of forming natural convection which will increase heat transfer rate between the fabric and skin. 140

156 Garment shrinkage during exposure reduces air gap and thereby increases the heat transfer rate. The garment shrinkage leads to high prediction level of third degree burn in 4 second flash fire exposures. Therefore, choosing of the garment proper size can significantly improve garment protective performance. Maintain dimensional stable during flash fire exposure and add functional components provide additional protection. The assumption of blood perfusion in skin model has minimal effect on predicted body burns in these exposures. With blood perfusion, it slightly reduces the time to 2 nd or 3 rd degree burn, especially the time to get 2 nd or 3 rd degree burn exceeds 20 second. Different temperature distributions in skin model demonstrate a large effect on burn predictions. The effect is pronounced in constant temperature distribution comparing to linear and quadratic. A large second degree burn prediction and low third degree burn prediction are observed in one layer skin model relative to the multi-layer skin model. This study suggests that the use of different skin models and their temperature can greatly affect burn predictions of garment testing in instrumented manikin. Therefore, a precise skin model selection and its standardization would be beneficial for manikin testing for thermal protective performance Recommendations A two or three dimensional model will be required to more precisely model manikin fire testing, and for an even greater understanding of thermal protective mechanisms. Such models can account the heat transfer in all directions. Hence powerful computation facilities are required. This research focused on short time intense flash fire conditions. Other research may be extended to long time, medium or low level heat conditions. 141

157 The successful modeling of one layer protective garment system providing evidence that extension to multi layer fire fighter suits is feasible. For multi layer protective garments, the modeling will be even more complex since the air gaps existing in between different layers which function differently. The thermal properties of multi layer protective garments systems required to be determined. Varying exposure time at different energy level can be modeled to understand heat transfer and moisture transfer in multi-layer protective system. Sensors used in manikin testing play an important role in precisely predicting burn injury. An additional modeling study of different types of sensors and analysis of the responses of different sensors exposed to a variety of heat levels are required on manikin scale. The manner in which thermal sensor cools following heat exposure is another area for further investigation. The effect of moisture on thermal protective performance of one layer and multi layer can be further explored using the model. This research indicates that different skin models and their temperature distributions can greatly affect predicted garments thermal protective performance. Standardization is required of the burn translation algorithm and its parameters Author s Note Caution should be taken in drawing conclusions about the safety benefits from these results. These data described in the thesis are from laboratory tests and controlled exposures. It is only focused on two common protective fabrics. The numerical model established in this research solely considers heat transfer in these materials based on the assumptions that no mass transfer and thermal degradation between specific temperatures range. The parametric study described in this thesis independently varied individual parameters, while there will actually be some of interdependence of the individual 142

158 parameters. The test conditions may not represent actual field conditions, which can be physically complicated and unqualified. Therefore, the results which are obtained by using this numerical model should not be used as an estimation of the protection which these or other material can provide in a real flash fire or other accident. In addition, this model is only intended to gain an appreciation of manikin testing and garment thermal protective mechanism. The intention of this work is not to recommend either of the fabrics, or any other particular fabric, for use in thermal protective garments. The conclusions drawn from this research are based on Pyroman Thermal Evaluation System located in NCSU and the limited quantities of fabrics and garments available to the author. 143

159 References 1. Shalev, I., and Barker, R. L., Predicting the Thermal Protective Performance of Heat-Protective Fabrics from Basic Properties, Special Technical Testing Publication 900, American Society for Testing and Materials, Barker, R. L. and Lee, Y. M., Analyzing the Transient Thermophysical Properties of Heat-Resistant Fabrics in TPP Exposures, Textile Research Journal, Vol. 57, No. 6, June Pawar, M., Barker, R. L., Shalev, I., Hamouda, H., Behnke, W.P. and Bender, M., Analyzing the Performance of Protective Garments in Manikin Fire Tests, College of Textiles, Center for Research on Textile 4. Bajaj, P. and Sengupta, A.K., Protective Clothing, Textile Progress, Vol.? Number? Soroka, A. J., Thermal Protective Clothing, presented at DuPont Nomex Protective Apparel Seminar, Edmonton, Alberta, May, Behnke, W. P. and Miner, L. H., Industr. Fabr. Prod. Rev., Buyers Guide, 1986, P McIntyre, J. E. in Proceedings of Industrial, Technical, and Engineering Textiles Group conference, May, 1988, The Textile Institute, Manchester, paper 7; Textile Horizons, 1988, 8, No. 10,

160 8. Jackson, R. H., PBI Fiber and Fabric Properties and Performance, Textile Research Journal, Vol. 48, 1978, pp Tomasino, Charles, Chemistry & Technology of Fabric Preparation & Finishing, North Carolina State University, Conn, J. J., and Grant, G. A., Review of Test Methods for Material Flammability, Department of National Defense Research Establishment Ottawa Contractor Report, March American Society for Testing and Materials, ASTM D : Standard Test Method for Flammability of Apparel Textiles, West Conshohocken, PA American Society for Testing and Materials, ASTM D 3659 Standard Test Method for Flammability of Apparel Textiles by semi-restraint Method, West Conshohocken, PA 13. American Society for Testing and Materials, ASTM D 2863 Measuring the Minimum Oxygen Concentration to Support Candle-like Combustion of Plastics (Oxygen Index), West Conshohocken, PA 14. Shalev, I., Barker, R.L., "A Comparison of Laboratory Methods for Evaluating Thermal Protective Performance in Convective/Radiant Exposures, Textile Research Journal, Vol. 54, Oct., , Shalev, I., Barker, R.L., "Analysis of Heat Transfer Characteristics of Fabrics in an Open Flame Exposure, Textile Research Journal, Vol. 53, Aug., , Lee, Y. M., and Barker, R. L., Thermal Protective Performance of Heat-Resistant Fabrics in Various High Intensity Heat Exposures, Textile Research Journal, Vol. 57, 1987, pp

161 17. Lee, Y. M., and Barker, R. L., Effect of Moisture on the Thermal Protective Performance of Heat-Resistant Fabrics, Journal of Fire Science, Vol. 4, 1986, pp Barker, R. L. and Lee, Y. M., "Evaluating the Thermal Protective Insulation Properties of Advanced Heat Resistant Fabrics, 32nd International SAMPE Symposium, April 6-9, pp , Stoll, A. M., Chianta M. A., and Munroe, L. R., "Flame contact studies, Journal of Heat Transfer, Transaction of the ASME, Vol. **, Aug., pp , Freeston, W.D., "Flammability and heat Transfer Characteristics of Cotton, Nomex and PBI Fabric," J. Fire & Flammability, Vol. 2, Jan., 57-76, Shalev, I., "Transient Thermo-Physical Properties Thermally Degrading Fabrics and Their Effect on Thermal Protection, Doctoral Dissertation, North Carolina State University, Norton, M. J. T., Kadolph, S. J., Johnson, R. F., and Jordan, K. A., Design, Construction, and Use of Minnesota Women, A Thermal Instrumented Mannequin, Textile Research Journal, Vol. 55, 1985, pp Stoll, A. M., and Chianta, M. A., Heat Transfer Through Fabrics as Related to Thermal Injury, Transactions of the New York Academy of Sciences, Vol. 33, 1971, pp Barker, R.L., Hamouda, H. and Behnke, W. P. Applying Instrumented Manikin Technology for Testing the Fire Resistance of Protective Clothing, College of Textiles, Center for Research on Textile Protection & Comfort, North Carolina State University. 146

162 25. Dale, J.D., Crown, E.M., Ackerman, M.Y., Leung, E. and Rigakis, K.B. "Instrumented Manikin Evaluation of Thermal Protective Clothing," Performance of Protective Clothing: 4th Volume, ASTM STP 1133, J. P. McBriarty and N.W. Henry, Eds., American Society for Testing and Materials, Philadelphia, pp , Behnke, W.P., Amotz, J. G. and Barker, R.L., "Thermo-Man and Thermo-Leg: Large Scale Test Methods for Evaluating Thermal Protective Performance," Performance of Protective Clothing: 4th Volume, ASTM STP 1133, J. P. McBriarty and N.W. Henry, Eds., American Society for Testing and Materials, Philadelphia, Crown, E.M., Ackerman, M.Y., Dale, J.D. and Rigakis, K.B., "Thermal Protective Performance and Instrumented Mannequin Evaluation of Multi-Layer Garment Systems," Performance of Protective Clothing: 5th Volume, Pawar, M. Analyzing the Thermal Protective Performance of Single Layer Garment Materials in Bench Scale and Manikin Tests, PhD Thesis, College of Textiles, North Carolina State University, Stoll, A. M. et al, Heat Transfer Through Fabrics, Naval Air development Center, Johsville, Pennsylvania, Coskren, R. J., and Kaswell, E. R., Development of PBI Fabric for Flight Suit Wear Test, Technical Report, AFML-TR , Air Force Materials Laboratory, Morse, H L., Thompson, J.G., Clark, K.J., Green, K.A., and Moyer, C.B., Analysis of the thermal Response of Protective Fabrics, Technical report, AFML-RT-73-17, Air Force Materials Information Service, Springfield, VA. 147

163 32. Ross, J. H., Thermal Conductivity of Fabrics as Related to Skin Burn Damage, Journal of Applied Polymer Science: Applied Polymer Symposia, Vol. 31, 1977, pp Backer, S., Tesoro, G. C., and Toong, T.Y., Textile Fabric Flammability, MIT Press, Cambridge, MA Ordinanz, W., Personal Adaptation and Protective Measures, Work in Hot Environments and Protection Against Heat, The Iron and Steel Institute, London, England, Shalev, I., and Barker, R. L., Protective Fabrics: A Comparison of Laboratory Methods for Evaluating the Thermal Protective Performance in Convective/Radiant Exposures, Textile Research Journal, Vol. 54, Shalev, I., and Barker, R. L., Analysis of the Heat Transfer Characteristics of Fabrics in an Open Flame Exposure, Textile Research Journal, Vol. 53, 1983, pp David A. Torvi, J. Douglas Dale, Heat Transfer in Protective Fabrics Under Flash Fire Conditions, Second International Conference on Fire Research and Engineering, Bomberg, M., Application of Three ASTM Test Methods to Measure Thermal Resistance of Clothing, Performance of Protective Clothing: Fourth Volume, ASTM STP 1133, J.P. McBriarty and N. W. Henry, eds., American Society for Testing and Materials, West Conshohocken, PA, 1992, pp Gagnon, B. D., Evaluation of New Test Methods for Fire Fighting Clothing, A thesis of Master of Science of Worcester Polytechnic Institute,

164 40. Stoll, A. M., Chianta, M. A., and Munroe, L.R., Flame Contact Studies, Transactions of the ASME, Journal of Heat Transfer, Vol 86, 1964, pp Baker, D.L., Wool, M.R., and Schaefer, J.W., A Dynamic Technique for Determining the Thermal Conductivity of Charring Materials, Thermal Conductivity-Proceedings of the Eighth Conference-1968, C.Y. Ho, and R.E. Taylor, eds., Plenum Press, New York, 1969, pp Shalev, I. Transient Thermophysical properties of Thermally Degrading Fabrics and Their Effect on Thermal Protection, Ph.D. thesis, North Carolina State University, Raleigh, NC, Morse, H L., Thompson, J.G., Clark, K.J., Green, K.A., and Moyer, C.B., Analysis of the thermal Response of Protective Fabrics, Technical report, AFML-RT-73-17, Air Force Materials Information Service, Springfield, VA 44. Torvi, David Andrew, Heat Transfer in Thin Fibrous Materials under High Heat Flux Conditions, Doctor Thesis, University of Alberta. Canada, Willians, S. D., and Curry, D. M., Thermal Protection Materials: Thermophysical Property Data, NASA Reference Publication 1289, National Aeronautics and Space Administration, Scientific and Technical Information Program, Futschik, M. W., and Witte, L. C., Effective Thermal Conductivity of fibrous Materials, General Papers in Heat and Mass Transfer, Insulation, and Turbonmachinery, HTD-Vol. 271, American Society of Mechanical Engineers, New York, 1994, pp Holman, J.P., Heat Transfer, Six Edition, McGraw-Hill Book Company, New York,

165 48. Schoppee, M.M., Seklton, J., Abbott, N. J., and Donovan, J. G., The Transient thermomechanical response of Protective Fabrics to Radiant Heat, Technical Report, AFML TR-77-72, Air Force Materials Laboratory (1976) 49. Backer, S., Tesoro, G. C., and Toong, T. Y., Textile Fabric Flammability, MIT Press, Cambridge, MA, David Andrew Torvi, Heat Transfer in Thin Fibrous Materials under High Heat Flux Conditions, Doctor Thesis, University of Alberta. Canada, Brown, J. R., and Ennis, B. C., Thermal Analysis of Nomex and Kevlar Fibers, Textile Research Journal, Vol. 47, 1977, pp Properties of Polybenzimidazole High-Performance Fiber, PBI Products Division Technical Literature, Hoechst Celanese, Charlotte, Henderson, J. B. and O Brien, E. F., Measurement of the Decomposition Kinetics of a Polymer Composite Using a High-Temperature, High-Heating-rate Simultaneous Thermal Analyzer, High Temperature-High Pressures, Vol. 19, 1987, pp Bingham, M. A., and Hill, B. J., Comparison of Thermal Characteristics of Materials Examined by Thermogravimetry and Flammability Tests. Journal of Thermal Analysis, Vol. 9, 1976, pp Shlensky, O. f., Aksenov, L.N., and Shashkov, A. G., Thermal Decomposition of Materials: Effect of Highly Intensive Heating, Elsevier Science Publishers, Amsterdam, Mraw, S. C., Differential Scanning Calorimetry, Chapter 11 in Specific Heat of Solids, CINDAS Data Series on Material Properties Vol. I-2, C. Y. Ho, Ed., Hemisphere Publishing Corporation, New York,

166 57. Crown, E. M., Dale, J. D., Evaluation of Flash Fire Protective Clothing Using an Instrumented Mannequin, Final Research Project Report for Alberta Occupational Health and Safety Heritage Grant Program, March 30, Shalev, I., and Barker, R. L., Analysis of the Heat Transfer Characteristics of fabrics in an Open Flame Exposure, Textile Research Journal, Vol. 53, 1983, pp Castle, G. K., Alexander, J. G., Tempesta, F. L., and Belason, E. B., Analytical Predictions of the Thermal Response of Decomposing Materials in Fire Environments, Journal of Testing and Evaluation, Vol. 1, 1973, pp Chen, N. Y., Transient Heat and Moisture Transfer Through Thermally Irradiated Cloth, Ph.D. Thesis, Massachusetts Institute of Technology, Cambridge, Massachusetts, Williams, S. D., and Curry, D. M., Thermal Protection Materials: Thermophysical Property Data, NASA Reference Publication 1289, National Aeronautics and Space Administration, Scientific and Technical Information Program, Afgan, N. H., and Reer, J. M., Heat Transfer in Flames, Wiley and Sons, New York, NY Hilado, C. J., Flammability Handbook for Plastics, Technomic, Westport, RI Remy, D. E., Sousa, J. A., Caldarella, G. J., and Levasseur, L. A., Thermal Degradation Products of THPOH-NH 3 Treated Cottons, Technical Report, Natick/TR-80/003, CEMEL Lewis, B., and von Elbe, G., Combustion, Flames, and Explosions of Gases, Third Edition, Academic Press, Inc., Orlando,

167 66. Holcombe, B. V., and Hoschle, B. N., Do Test Methods Yield Meaningful Performance Specifications?, Performance of Protective Clothing: First Volume ASTM STP 900, R. L. Barker and G.C. Coletta, eds., American Society for Testing and Materials, West Conshohocken, PA, 1986, pp Krasny, J., Rockett, J. A., and Huang, D., Protecting Fire Fighters Exposed in Room Fires: Comparison of Results of Bench Scale Test for Thermal Protection and Conditions During Room Flashover, Fire Technology, Vol. 24, 1998, pp Schoppee, M.M., Seklton, J., Abbott, N. J., and Donovan, J. G., The Transient thermomechanical response of Protective Fabrics to Radiant Heat, Technical Report, AFML TR-77-72, Air Force Materials Laboratory (1976) 69. Brian David Gagnon, Evaluation of New Test Methods for Fire Fighting Clothing, A thesis of Master of Science of Worcester Polytechnic Institute, Stoll, A. M., Chianta, M. A., and Munroe, L. R., Flame Contact Studies, Transactions of the ASME, Journal of Heat Transfer, Vol. 86, 1964, pp Drysdale, D., An Introduction to Fire Dynamics, John Wiley and Sons, Chichester, Anthony, C.P., and Thibodeau, G. A., Textbook of Anatomy and Physiology, 10 th ed., The C. V. Mosby Co., St. Louis, 1979, pp , , Pennes, H. H., Analysis of Tissue and Arterial Blood Temperatures in Resting Human Forearm, Journal of Applied Physiology, Vol. 1, 1948, pp

168 74. Mehta, A.K, and Wong, F., Measurement of Flammability and Burn Potential of Fabrics, Full Report from Fuels Research Laboratory, Massachusetts Institute of Technology, Cambridge, Massachusetts, Hodson, D. A., Eason, G., Barbennel, J.C., Modeling Transient Heat Transfer Through the Skin and Superficial Tissues I: Surface Insulation, Transaction of the ASME, Journal of Biomechanical Engineering, Vol. 108, 1986, pp Wulff, W., The Energy Conservation Equation for Living Tissue, IEEE Transactions of Biomedical Engineering, Vol. BME-21, pp , Klinger, H.G., Heat Transfer in Perfused Biological Tissue I: General Theory, Bulletin of Mathematical Biology, Vol. 36, pp , Chen, M.M., and Holmes, K.R., Microvascular Contributions in Tissue Heat Transfer, Annals of the New York Academy of Sciences, Vol. 335, pp , Xu, L.X., The Evaluation of Heat Transfer Equations in the Pig Renal Cortex, Ph.D. Dissertation, Univ. of Illinois at Urbana-Champaign, August, Xu, L.X., Chen, M.M., Holmes, K.R, and Arkin, H., The Evaluation of the Pennes, the Chen-Holmes, the Weinbaum-Jiji Bioheat Transfer Models in the Pig Cortex, ASME WAM, HTD-Vol. 189, pp , Weinbaum, S., Jiji, L.M., and Lemos, D.E., Theory and Experiment for the Effect of Vascular Temperature on Surface Tissue Heat Transfer Part I: Anatomical Foundation and Model Conceptualization, ASME Journal of Biomedical Engineering, Vol. 106, pp ,

169 82. Jiji, L.M., Weinbaum, S., and Lemos, D.E., Theory and Experiment for the Effect of Vascular Temperature on Surface Tissue Heat Transfer Part II: Model Formulation and Solution, ASME Journal of Biomedical Engineering, Vol. 106, pp , Dagan, Z., Weinbaum, S., and Jiji, L.M., Parametric Study of the Three-Layer Microcirculatory Model for Surface Tissue Energy Exchange, ASME Journal of Biomedical Engineering, Vol. 108, pp , Song, W.J., Weinbaum, S., and Jiji, L.M., "A Theoretical Model for Peripheral Heat Transfer Using the Bioheat Equation of Weinbaum and Jiji, ASME Journal of Biomedical Engineering, Vol. 109, pp , Song, W.J., Weinbaum, S., and Jiji, L.M., A Combined Macro and Microvascular Model for Whole Limb Heat Transfer, ASME Journal of Biomedical Engineering, Vol. 110, pp , Henriques, F. C., Jr., Studies of Thermal Injuries V. The Predictability and the Significance of Thermally Induced Rate Processes Leading to Irreversible Epidermal Injury, Archives of Pathology, Vol. 43, 1947, pp Buetter, K., Effects of Extreme Heat and Cold on Human Skin II. Surface Temperature, Pain and Heat Conductivity in Experiments with Radiant Heat, Journal of Applied Physiology, Vol. 3, 1951, pp Stoll, A. M., and Greene, L. C., Relationship Between Pain and Tissue Damage Due to Thermal Radiation, Journal of Applied Physiology, Vol. 14, 1959, pp Takata, A. N., Rouse, J., and Stanley, T., Thermal Analysis Program, I.I.T. Research Institute Report, IITRI-J6286, Chicago,

170 90. Stoll, A, M., and Chianta, M. A., Method and Rating System for Evaluation of Thermal Protection, Aerospace Medicine, Vol , pp David A. Torvi, Ph.D. Thesis, University of Alberta, Gibson, P. Governing Equations for Multiphase Heat and Mass Transfer in Hygroscopic Porous Media with Applications to Clothing Materials, Technical Report Natick/TR-95/004, Mell, W. E. and Lawson, J. R., A Heat Transfer Model for Fire Fighter's Protective Clothing, National Institute of Standards and Technology, NISTIR 6299, January Whitaker, S. A Theory of Drying in Porous Media, Advances in Heat Transfer 13, Academic Press, New York, Gibson, P. W. and Charmchi, M., Integration of a Human Thermal Physiology Control model with a Numerical Model for Coupled Heat and Mass Transfer through Hygroscopic Porous Textiles, presented at the 1996 ASME International Mechanical Engineering Congress & Exhibition, Atlanta, GA, Nov , Ricci, R., Traveling Wave Solutions of the Stefan and the Ablation Problems, SIAM J. Math. Anal., Vol. 21, pp , Whiting, P., Dowden, J.M., Kapadia, P.D., and Davis, M.P., A One-Dimensional Mathematical Model of Laser Induced Thermal Ablation of Biological Tissue, Lasers in Med. Sci., Vol. 7, pp , Delichatsios, M.A. and Chen, Y., Asymptotic, Approximate and Numerical Solutions for the Heatup and Pyrolisis of Materials Including Reradiation Losses, Comb. and Flame, Vol. 92, pp ,

171 99. Staggs, J.E.J., A Discussion of Modeling Idealized Ablative Materials with Particular Reference to Fire Testing, Fire Safety Journal, vol. 28, pp , 1997a 100. Vovelle, C., Delfau, J., and Reuillon, M., Experimental and Numerical Study of Thermal Degradation of PMMA, Comb. Sci. and Tech., Vol. 53, pp , Wichman, I.S. and Atreya, A., A Simplified Model for the Pyrolisis of Charring Materials, Comb. and Flame, Vol. 68, pp , Di Blasi, C. and Wichman, I.S., Effects of Solid-Phase Properties on Flames Spreading over Composite Materials, Comb. and Flame, Vol. 102, pp , Kashiwagi, T., Polymer Combustion and Flammability- Role of the Condensed Phase, Proc. 25th Int. Symp. on Combustion, The Combustion Institute, pp , Staggs, J.E.J., A Theoretical Investigation into Modeling Thermal Degradation of Solids Incorporating Finite-Rate Kinetics, Combust. Sci. and Tech., Vol. 123, pp , 1997b 105. Grimes, R., Mulligan, J.C., Hamouda, H., and Barker, R. The Design of a Surface Heat Flux Transducer for Use in Fabric Thermal Protection Testing; Performance of Protective Clothing: Fifth Volume, ASTM STP James S.Johnson and S.Z Mansdorf, Eds. American Society for Testing and Materials.West Conshohocken, PA. 1966, pp

172 106. Grimes, R., The Design and Calibration of a Surface Heat Flux Transducer for Use in Fabric Thermal Protection Testing; MS Thesis. North Carolina State University. Raleigh, NC Catton, I., Natural Convection in Enclosures, Proceedings of the 6 th International Heat Transfer Conference, Toronto, Vol. 6, 1978, pp Denny, V. E., and Clever, R. M., Journal of Computational Physics, Vol. 16, 1974, pp Hollands, K. G. T., Unny, T. E., Raithby, G. D., and Konicek, L., Journal of Heat Transfer, Vol. 98, No. 2, 1976, pp Ayyaswamy, P. S., and Catton, I., Journal of Heat Transfer, Vol. 95, 1973, pp Patankar, S. V., Numerical Heat Transfer and Fluid Flow, Taylor & Francis, Tannehill J. C., Anderson D.A., and Pletcher R. H., Computational Fluid Mechanics and Heat Transfer, 2 nd Ed, Taylor & Francis, Bailey, E. D., Probability and Statistics Models for Research, John Wiley & Sons Inc., Beck, J. V., Blackwell, B., and St. Clair, C. R., Jr., Inverse Heat Conduction: Ill- Posed Problems, John Wiley & Sons, Inc., Özisik, M. N., and Orlande, H. R. B., Inverse Heat Transfer: Fundamentals and Applications, Taylor & Francis,

173 116. Beck, J. V. and Arnold, K., Parameter Estimation in Engineering and Science, published by Wiley, Beck, J. V., and McMasters, Users Manual for Prop1D Program for Estimating Thermal Properties from Transient Temperature and Heat Flux Measurements, Version 7.0, Beck Engineering Consultants Company, Okemos, MI Bruner, David, Application of Body Scanning Technology, Presentation in College of Textiles, North Carolina State University, May Diller, K. R., and Hayes, l. J., A finite Element Model of Burn Injury in Blood Perfused Skin, Transactions of the ASME, Journal of Biomechanical Engineering, Vol. 105, 1983, pp Incropera, P. F. and Dewitt. P. D., Fundamentals of Heat and Mass Transfer, John Wiley & Sons, Fourth Edition,

174 APPENDICES 159

175 Appendix 1 Experimental Apparatus Used in This Research The main objective of these experimental apparatus is to get thermal response data from fabric or heat source of the interest. In some setups the responses are crucial to the quality of estimation and determination. In this chapter, the experimental work is described. These are including the type of thermocouples and their gauge selection in order to precisely measure flame temperature in both bench top test and Pyroman test, as well as fabric surface temperature exposed to intense heat conditions. The major two kinds of sensors, the copper disk sensor and skin simulant sensor, used in protective testing field and their heat flux calculation algorithm are introduced. The Medtherm sensor is used in this research to calibrate all the bench top test heat sources and to determine calibration factor of the sensors. The apparatus in heat transfer coefficient determination as well as fabric thermal conductivity and volumetric heat capacity estimation are discussed. The data acquisition system and software used in all the experimental apparatus are also discussed. Temperature Measurement Device Temperature is one of the most widely measured physical quantities. Its accurate measurement is essential in calculate or estimate other parameters. The transient flame and fabric surface temperature were measured using thermocouples and infrared thermometer. Depending on the measured temperature range and the necessary responding time, the thermocouple gauge and type need to be selected. In order to measure the yarns temperature on the front and back surface of the fabric, fine gauge thermocouples are used. Larger wires will be more rugged and is easy to handle, but this would have a slower time response. However, finer gauge wires will be more difficult to use. In order to determine how fine a gauge of thermocouple wire should be used, some specific tests are designed to compare the influence of thermocouple wire gauge, 160

176 including comparing with the infrared thermometer to measure fabric surface temperature. The diameters of the yarns are of the order of 250 to 400 µm. As the junction of the thermocouple where the temperature is measured will be at least twice the diameter of the separate thermocouple wires, so thermocouples with 36 and 40 A.W.G. are chosen to measure fabric surface temperature. The maximum temperature of fabric surface in this research is among C. Therefore, chromel-alume (type K) thermocouple is used, as this type K can be used to measure temperature up to C for short duration (3-10 seconds). The flame temperature is in a range of C. Therefore Type R and Type B with diameter inch are chosen to measure the flame temperature. The temperature range of type R is C and type B is C. The thermocouple response time is important in heat transfer coefficient estimation and fabric thermophysical property estimation. The following figure 1 shows different response and measurement of the same flame. Flame Temperature Measurement with Different Thermocoupl Gauge Diameter:0.15 Diameter: Temperature (0C) Time (sec) Figure 1. The Flame Measurement with Different Thermocouple Gauge (type R) 161

177 Data Acquisition System A laptop computer is used in conjunction with LabVIEW data acquisition system to record the response of the sensors, thermocouples, and infrared thermometer. LabVIEW programming system is used to control the data acquisition system. The outputs from the thermocouples and sensors is fed into signal condition 5B47 module and amplified. The DAQ board is used to control and operate these voltage signals. The LabVIEW program in the computer converted these voltages into temperature data according to different type thermocouples. These inputs are recorded 1000 reading per second. The data then saved in a datafile for data analysis (Figure 2). The capability of the data acquisition to receive data is crucial in determination of heat transfer coefficient and in fabric thermal conductivity and volumetric heat capacity estimation. The specifications for data acquisition system: Computer: Laptop CPU 1GB with RAM 512 DAQ board: DAQCard-AL-16E-4 16 inputs, 500ks/s, 12-bit Multifunction I/O Signal conditioning Module: 5B47 (from B, R, K, T), Isolated Linearized Figure 2. Data Acquisition System 162

178 Heat Flux Sensors A heat flux sensor typically consists of a thermopile or sometimes just a pair of thermocouples in which the elements are separated by a thin layer of thermal resistance material. Under a temperature gradient, the two thermopile junction layers will be at different temperatures and will therefore register a voltage. The heat flux is proportional to this differential voltage. The three modes of heat transfer, radiation, conduction and convection can be measured by heat flux sensors. Most heat flux sensors are calibrated using radiation heat source, which are the most consistently repeatable sources for this purpose. So emissivity is an essay here in order to get near 100% incident heat flux. Sensors are typically coated black to improve emissivity so the absorbed radiation is nearly equal to the incident radiation. For the conductive heat flux, the sensor is in direct contact with a heated material. A good thermal contact is necessary. If the contact is poor, there will effectively be a high thermal resistance between the sensor and the material interest, which can seriously alter the sensor reading. For convective heat flux, the heat transfer coefficient is a function the flow s thermal conductivity and its characteristics. The heat transfer coefficient is usually determined by measuring the surface heat flux. This procedure assumes that the heat transfer coefficient for the heat flux sensor and the surrounding system are the same. Basically, two kinds of sensor are used in thermal protective testing field. One is slug type copper sensor; the other is skin simulant sensor. The TPP sensor and Pyrocal sensor are slug copper sensor; and Thermoman sensor and Alberta sensor are skin simulant sensor TPP Sensor The TPP sensor (Figure 3) is a copper disk sensor which is widely used for bench top testing of thermally protective clothing materials. The thermal sensor itself consists of a 1.57x 0.06 inch copper disk. Four J-type (iron constantan) thermocouples are secured in the disk, positioned at 120 degree intervals and at the center. Heat flux is calculated from 163

179 the temperature rise, indicated by the thermocouple output, and from the mass and specific heat capacity of the copper disk [105]. This sensor is highly reliable and rugged. There is no loss management in its heat flux calculation. When exposed to a constant heat flux, the copper disk temperature rise is linear to time. q = mc A cu ε dt ( t dt p cu ) cu Insulating Block Copper Disk Lock Nut Thermocouple Tube Thermocouples Figure 3. TPP sensor Pyrocal Sensor This sensor is developed by North Carolina State University for use in instrumented manikin fire testing systems[105]. It is insulated slug type sensor with 0.50x 0.06 inch copper disk, surrounded radially by a thin copper ring thermal guard. Both the disk and the ring are supported by an insulating holder to minimize heat transfer to and from the body of the calorimeter thus approximating one-dimensional heat flow. Beneath the surface of the copper disk, an insulating air cavity is maintained and a T-type (copper-constantan) thermocouple is attached to the lower side of the disk (Figure 4). The entire assembly is encapsulated within a protective shell. Heat flux is calculated from temperature rise and the known properties of the copper slug using a procedure that increases the accuracy of the heat flux estimate by compensating for heat losses [106]. 164

180 165 Figure 4. Pyrocal Sensor Used for Fabric Thermal Protective Testing [107] Alberta Sensor Alberta sensor is a skin simulant sensor developed by the University of Aberta for use in their instrumented fire test manikin (Figure 5). The sensor is made of colorceran, a mixture of inorganic materials including calcium, aluminum, and silicate with asbestos fibers and a binder. This sensor material is reported to have thermal properties such that heat transfer will be similar to human skin which suddenly exposed to heat flux. A T- type thermocouple is mounted on the surface of the sensor. The thermocouple wire runs through a hole drilled inside the sensor. Details of this sensor and procedures used to calculate heat flux can be found in reference [91]. The surface heat flux is calculated using Duhamel s theorem. ( ) i L cu p cu L T t T K dt t dt d C C q cu + = ) ( ) ( ρ ( ) ( ) ( ) ( ) ( ) = = ) ( ) ( ) ( ) ( ) ( 2 ) ( ) ( ) ( ) ( ) ( t t T t T t t T t t t t t T t T t t t T t T t t t T t T C k t q n s n s n n s n i i n i n i s i s i n i s n s i n i s n s p n π ρ

181 Figure 5. Alberta Sensor Thermoman Sensor This sensor is a polymeric skin model sensor developed by DuPont for use in the Thermo-Man fire test manikin. This sensor employs a thin-skin calorimeter that incorporates a Type T thermocouple buried below the exposed surface of a cast resin plug. The resin plug is made of a thermoset polymer that reportedly exhibits a thermal inertia similar to undamaged human skin [106]. Heat transfer is calculated using an inverse heat transfer model that relies on an accurate location of the thermocouple bead. Sensor Calibration A MEDTHERM sensor is used as referee to calibrate the above sensors. First the radiation heat source is adjusted to 84kW/m 2 using MEDTHERM sensor, then other sensors exposed to this known heat source for certain length of time and record their reading. Calibration factor can be calculated based on these calibration data for each specific sensor. The figure 6 is one of calibration examples to get calibration factor for the different sensors. 166

182 Senser Calibriation Chart W/m Medtherm Pyrocal TPP Dupont Alberta Time (sec) Figure 6. Sensor Calibration Using Medtherm Sensor Heat Source Two heat sources are used in this research. One is pure radiation heat source using eight quartz lamps. This apparatus is also called RPP tests (Figure 8 and Figure 9). Most heat flux sensors are calibrated using radiation heat source, which are the most consistently repeatable sources for this purpose (Figure 7). Insulating Block Sensors Sensor Mounting Configuration 1 2 Spacer 3 LabView/ Data Acquisition Analog Backplane Device System Quartz Lamps Figure 7. Radiation Heat Source 167

183 Figure 8. Radiation Heat Source The other heat source used in this research is a typical TPP heat source using two Meker burner and eight quartzs heaters (Figure 9). Two Meker burners using a methane gas flame in combination with a bank of quartz tubes to provide convective and radiant heat source. This heat source is used in fabric parameter estimation setup, to estimate fabric thermal conductivity and volumetric heat capacity. Figure 9. Gas Burner and Radiation Heat Source 168

184 Appendix 2 Average Heat Flux Values as Model Input (heat flux: cal/cm 2 sec) Sensor H.Flux Sensor H.Flux Sensor H.Flux Avg S.D

185 Appendix 3 Sorted Fluxes and Their Normal Scores of Pyroman 4sec Exposure Sensor Number Heat Flux Normal Scores Sensor Number Heat Flux Normal Scores Sensor Number Heat Flux Normal Scores

186 Appendix 4. Normal Distribution For a normal distribution, only two parameters are used to determine the distribution: the mean (µ) and the standard deviation (σ) as shown in following normal density equation. Y i = σ 2 ( X i µ ) 1 2 σ 2 π e 2 Where Y i is normal density, σ is standard deviation and µ is sample mean. 171

187 Appendix 5. Garment Fabric Compression Test Nomex ШA Coverall Compression Test Garment ( 6.0oz/yd 2 ) Fabric Compression Report Pure Compression Test Request#: Garment N Size 42 Company Name: Mr Song Date Tested: 03/19/02 Technician: Jon Porter SAMPLE: GN 42 SPEED (mm/sec): 1.0 Compressive force (g/cm^2): 50.0 Stroke Sensitivity Switch (MM) = 5 Sensitivity: 2 X 5 Fabric Width: 10 CM. Fabric Weight (g/cm^2): 2.00 Gap Dial SActual Gap Distance: 1.75 WC (Work) WC' RC LC Delta-T Thick Tm N Compr. Recov %Resil Lnrty T EMC% (mm) (mm) Avg / Kevlar/PBI Coverall Compression Test Garment (4.5oz/yd2) Kevlar/PBI Compression Report Pure Compression Test Request#: G 42 Company Name: Mr Song Date Tested: 03/19/02 Technician: Jon Porter SAMPLE: G 42 SPEED (mm/sec): 1.0 Compressive force (g/cm^2): 50.0 Stroke Sensitivity Switch (MM) = 5 Sensitivity: 2 X 5 Fabric Width: 10 CM. Fabric Weight (g/cm^2): 2.00 Gap Dial SActual Gap Distance: 1.75 WC (Work) WC' RC LC Delta-T Thick Tm N Compr. Recov %Resil Lnrty T EMC% (mm) (mm) Avg /

188 Appendix 6. 3D Body Scanning Technology 3D Body Measurement System technology [118] developed by [TC]² includes a white light-based scanner and proprietary measurement extraction software. The scanner captures hundreds of thousands of data points of an individual's image, and the software automatically extracts dozens of measurements. This measurement information can be electronically compared to garment specifications and other data in order to recommend the size an individual should purchase or used as a basis for made to measure clothing. [TC]² chose the white light phase measurement profilometry (PMP) approach. The structured light and PMP application is well suited for body measurement because of the short acquisition time, accuracy and relatively low cost. The system design uses four surface sensors. The sensors are stationary, so each must capture an area segment of the surface. The area segments from the sensors are combined to form an integrated surface which covers the critical areas of the body that are needed for making apparel. The software program was developed in the Microsoft Developer Studio Visual C++ under the Windows NT platform. OpenGL was utilized to create the graphics display tools. The software performs the following functions: graphical user interface, controlling the acquisition sequence, acquiring and storing image buffers, processing acquired images and calculating resulting data points, and displaying graphical output. The PMP method involves shifting the grating preset distances in the direction of the varying phase and capturing images at each position. A total of four images are taken from each sensor, each with the same amount of phase shift of the projected sinusoidal pattern. Using the four images of the scene, the phase at each pixel can be determined. The phase is then used to calculate the three-dimensional data points. With the raw scan data acquired, a wealth of 3D geometric information is available which can be used to predict best fit ready-to-wear clothing sizes or to make clothing for the scanned individual with a level of fit that would be difficult to achieve with a manual 173

189 measurement process. However, a data extraction step is necessary to get the key measurements. This process is a fully automated computer process. The automation is desirable because the time required for a computer operator to extract the information using an interactive data analysis tool can be as great as or greater than the manual process using measurement tapes. The automated process developed by [TC]² is beginning with the raw scan data (Figure 2). Figure 2. Actual Scan Raw Data The raw scan data is further processed into a proprietary format that has several advantages over the raw form of the scan data. This proprietary format (graphically shown below in Figure 3) results from a sequence of processes including: 1.Data Filtering, which removes any stray points. 2.Segmentation of the body into individual limbs (arms, legs, torso) 3.Smoothing, which removes low level noise in the scan data. 4.Filling, which closes any small gaps in the scan data. 5.Compression, on the order of 100:1, to achieve a very "light" yet fully defining data set 174

190 Figure 3. Processed Scan Data The primary advantage of using this "processed" scan data is that it allows for the creation of measurement extraction algorithms which are relatively more robust, repeatable, and accurate as compared to algorithms which operate on the data in its raw form. This in turn allows for the measurement extraction process to be automated, that is, to occur without operator intervention. 175