THE UNIVERSITY OF MICHIGAN INDUSTRY PROGRAM OF THE COLLEGE OF ENGINEERING A STUDY OF THE RATE OF THERMAL RADIATION THROUGH POROUS INSULATING MATERIALS Bert K. Larkin A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the University of Michigan 1957 June, 1957 IP-223

Doctoral Committee: Associate Professor Stuart W. Churchill, Chairman Assistant Professor Kenneth F. Gordon Associate Professor Robert L. Hess Assistant Professor C. Wilbur Peters Associate Professor J. Louis York

ACKNOWLEDGEMENTS I wish to express my appreciation to all those who have helped me during the course of this undertaking. The encouragement and advi ce of Dr. Fred Kurata are largely responsible for my attempting to do doctorate work. Much of the aid and inspiration needed to complete this thesis were furnished by my committee chairman, Dr. S. W. Churchill. I am also indebted to the other members of my committee, Drs. K. F. Gordon, J. L. York, R. L. Hess, and C. W. Peters. The interest and aid of Mr. Marion Hollingsworth of the Owens-Corning Fiberglas Corporation were most helpful. The glass fiber specimens and much of the thermal conductivity data were obtained from Mr. Hollingsworth. My laboratory associates, Dr. Jin H. Chin, Mr. George C. Clark, and Mr. Philip H. Scott were also very helpful. Mrs. Barbara Ray aid much of my correspondence work. I also wish to express my thanks to the DuPont Company for their fellowship grant during the year lyj51956. Finally, I wish to express my appreciation to the Industry Program of the College of Engineering for their work in reproducing this thesis. iii

TABLE OF CONTENTS Page ACKNOWLEDGEMENTS iii LIST OF FIGURES vi LIST OF TABLES ix ABSTRACT xi INTRODUCTION 1 PART I THEORETICAL WORK 5 A. Derivation of the T4o Flux Equatirns 6 B. Treatment of Simultaneous Radiation, Conduction and Convection 14 C. Solution of the Flux Equations 17 D. Estimation of the Radiation Parameters from Elc ztromagnetii Thcory 22 PART II EXPERIMENTAL PROGRAM 35 A. General Description of the Method 36 B. Experimental Conditions 37 C. Experimental Equipment 41 D. OjerS ting Procedure 48 E. ExperimentQl Data 50 PART III ESTIMATION OF THE RADIATION PARAMETERS FROM TRANSMISSION DATA 62 A. The MethDd 63 B. The Radiation Parsaseters 65 C. Comparison of Experimental and Theoretical Values for tne Radiation Parameters 75 D. Grouping the Parameters into Dimensioniless Form 82 iv

PART IV ESTIMATION OF THE RADIANT HEAT TRANSFER RATES 86 A. The Method 87 B. Effects of Temperature 88 C. Effects of Boundary Emisi',ity 92 D. Effects of Fiber Size 95 E. Effects of Bulk Density 17 F. The Temperature Distribution 99 G. The Importance of Radiation Compared with Other Heat Transfer Mechanisms 101 H. Comparison of Radiation Rates o106 I. Radiation Through Large Black Fibers 109 PART V SUMMARY 110 PART VI CONCLUSIONS 113 PART VII RECOMMENlDATIONIS FOR IMPROVING INSULATIONS 116 APPENDIX A EQUIPMENT CALIBRATIONS 119 A. The Amplifier 119 B. The Detector 119 C. The Source 126 APPENDIX B SAMPLE CALCULATIONS 128 APPENDIX C THE SCATTERING OF ELECTROMAGNETIC RADIATION BY AN ISOLATED INFINITE CYLINDER 134 APPENDIX D CYLINDRICAL SCATTERING FUNCTIONS 145 APFENDIX E EXPERIMENTAL TRANSMISSION DATA AND CALCULATED RADIATION PARAMETERS 152 APPENDIX F INSULATION FIBER SIZE DISTRIBU'IONS 160 NOMENCLATURE 165 BIBLIOGRAPHY 169 v

LIST OF FIGURES Figure Pge 1 A Porous Insulation 7 2 Representation of the Angluar Distribution of Scattered Power 9 3 Scattering Functions for a Sphere of Refractive Index 0.90 24 4 The Scattering Coefficient for an Infinite Cylinder 30 5 The Back Scattering Fraction for an Infinite Cylinder 31 6 The Scattering Parameter BKS for an Infinite Cylinder 32 7 A Schematic Diagram of the Apparatus Used in Obtaining Transmission Data 42 8 A Photograph of the Apparatus Used in Obtaining Transmission Data 43 9 Relative Radiation Flux vs. Source Distance 51 10 The Transmission of Black Body Radiation Through Foamglas 54 11 The Transmission of Black Body Rtdiation Through Styrofoam 55 12 The Transmission of Black Body Radiation Through Fiberglas A 56 13 The Transmission of Bleck Body Radiation Through Fiberglas B 57 14 The Transmission of 4000F Black Body Radiation Through Glass Fibers 58 15 The Transmission of 400~F Black Body Radiation Through Glass Fibers 59 16 The Radiation Parameters for Foamglas 66 17 The Radiation Parameters for Styrofoam 22 67 18 The interception Parameters for Some Bonded FiLrus Insulations 65 vi

Figure Page 19 The Absorption Parameters for Some Bonded Fibrous Insulations 69 20 The Interception Parameters for Glass Fibers 70 21 The Absorption Parameters for Glass Fibers 71 22 A Comparison of Theoretical and Experimental Values for the Scattering Parameter for Glass Fibers 76 23 A Comparison of Theoretical and Experimental Values for the Scattering Parameter for Glass Fibers 77 24 A Comparison of Theoretical and Experimental Values for the Scattering Parameter for Glass Fibers 78 25 A Comparison of Theoretical and Experimental Values for the Scattering Parameter for Glass Fibers 79 26 The Dimensionless Intercepti n Parameter for Glass Fibers 83 27 The Di:!.ensioriess Absorp'i.on Paraieneter for Glass Fibers 84 28 The Radiant Conductivity at';he Boundary for Styrofoam and Foama;las 89 29 The Radiant Conductivity at the Boundary for Fiberglas 90 30 The Radiant Conductivity at the Boundary for Glass Fibers 91 31 The Effect of Boundary Emuissivity on the Radiant Conductivity at the Boundary 93 32 The Effect of Boundary Emissivity on the Radiant Conductivity in Fiberglas B 94 33 The Effect of Fiber Size on the Radiant Conductivity at the Boundary for Glass Fibers 96 34 The Effect of Bulk Density on the Radiant Conductivity at the Boundary for Fiberglas B 98 35 A Computed Temperature Distribution 100 36 The Contribution of Rladiation to Heat Transfer Through Styrofoam and Foamglas 102 37 The Contribution of Radiatiioll to Heat Transfer Through Fiberglas A and B 103 vii

Figure Page 38 The Contribution of Radiation to Heat Transfer Through Glass Fibers 104 A-1 The Frequency Response fLr the Modified Ten Cycle Amplifier I 120 A-2 A Linearity Test of the RadiatiJn Detector 122 A-3 An Illustration of the Detector Position 123 A-4 The Angular Response of the Detector 124 A-5 A Test of the Source Strength by the Stefan-Boltzmann LaW 127 B-1 Transmission Data 130 C-1 Radiation Impinging Upon an Infinite Cylinder 135 viii

LIST OF TA3LES Table I The Physical Properties of the Insulations 38 II The Chemical Cornposi;ion of thE Cless in the Fibrous Insulations 39 III The PhysiE&al Properties of' Styrofoamr 81 IV Thermal Conauctivity Data 105 V R&adiant Conductivities for Unbonded Glass Fibers 106 VI Radiant Conductivities for Styr.foam and Foamglas 108 B-1 Trans!,issiols &hrouh CGl:ss Fioers B-2 Values t ~f for 35 Mizron Glass Fi ers 131 B-1 Scattering Functions for rr = 1.5 14( D-2 Scuatoering Functions for m - 2.0 148 D-3 Scattering Functions for m = 2.5 150 E-1 TTrainszmissions Through oty:r-foam 152 E-2 Tranursmi$isions Through Foamsglas 15 E-3 Transniisions Through Fiberglas A. 153 E-4 Transmissions Through Fiber,,.las B. 153 E-5 Tr'nsmissions Thr-ugh 0.17 Micron Diameter Glass Fibers at a Density of 0.5 lo./cu.ft. 154 E-6 Transmissions Through 0.0O Micron Diarieter Glass Fibers at a Density of 0.5 lb./cu.ft. 154 E-7 Transmissions Through 2.5 Micron Diameter Glass Fibers at a Density oi'.5 lo./:u.f. 155 E-8 Trwismissisins Through 5j Micron Diwumntcr GlaJs Fibers at a Density of 0.3) lb./2u.ft. 135 E-9 Transmissions Through 10 Micron Diameter Glass Fibers at (a Density of 0.) lh./;u.ft.. ix

Table P E-10 Transmissions Through 20 Micron Diameter Glass Fibers at a DenJity of 0.) lu./ u. ft. 156 E-ll Transri.;sions Through 35 Micron Diuaneter Glass Fibers at a Density of'10.5 lb./cu.ft. 157 E-12 Radiation Prareters for Styrofoam 157 E-13 Radiation Parameters for Foiglas 157 E-14 Radiation Parameters for Fiberglas A 158 E-15 Radiation Parameters for Fiberglas B 158 E-16 Radiation Paruameters for Glt-ss Fibers at a Density of 0.5 lU./ j u.t. 139 F-1 The Fiber Size Distribution for Glass Fibers of 0.17 Micron Mean Diameters 161 F-2 The Fiber Size Distribution for 5Glass Fibers of 0.20 Micron Mear Diameters 163 x

Abstract In this study, estimates were made of the rate of heat transfer by thermal radiation through porous insulations. Two types of materials were considered, foamed insulations and fibrous insulations. In an effort to simplify the problem, the insulations were assumed to be infinite in their lateral dimensions and of finite thickness, and the heat transfer process was assumed to occur at steady state. By postulating that radiation can only be propagated forward or backward along the direction of finite insulation thickness, it is possible to describe the radiation process by two differential equations. Completely describing the heat transfer process requires one other equation, the equation of continuity. Because two of these equations were non-linear, an approximate solution was employed. Evaluation of the radiant heat transfer rate from she apprcximate solution requires numerical values for two parameters. These parameters are the scattering cross section and the absorption cross section of a unit volume of insulation. The determination of these parameters was the objective of an experimental study. The experimental data consisted of a set of transmission measurements for black body radiation through varying thicknesses of an insulation. From these transmission measurements it is possible to estimate the scattering and absorbing parameters. The radiation properties of 11 different insulations were investigated. Four of these insulations are manufactured commercially; two of which were fibrous, and two were of the foamed type. The other xi

insulations were made of closely sized glass fibers and were utilized in an attempt to demonstrate the effect of fiber size on the radiation parameters. Electromagnetic theory shows that an optimum pore or fiber size exists for blocking radiation by a given mass of insilation. From the results of measurements on the fibrous insulations and some crude estimates from electromagnetic theory, this size was estimated to be in the range 2-5 microns. Other variables which were treated were insulation bulk density, temperature, and emissivity of the surfaces bounding the insulation. From the results of this study it was estimated that radiation comprises from 3 to 10 percent of the heat transferred through porous insulations at room temperatures. At higher temperatures, radiation can account for 20 to 30% of the heat transfer. xli

INTRODUCTION With the introduction of low density insulating material considerable saving of both space anu vriight Iay be realized in many applications. These materials are being used extensively in consumer appliances, homes, clothing, industrial process equipment, automobiles, and aircraft. Heat transfer problems x.rislng from the a,;li[ation of these materials are generally solved sy Freitting the heat transfer process as if it were wholly a conduction process. There is one property of these insulations that cannot be explained solely on the basis oi a conduction mechanism. This property is the variation in the thermal conductivity with changing bulk deensity. As the bulk density is decreased, the concuctivity first decreases, passes through a minizium, then finally in.reases. The conducition mechanism predicts that the thermal conductivity would continuully decrease with decreasing density, finally approaching the conductivity of the gas fillixld the void spaces. This defect in the conduction theory suggests that the other heat transfer processes, radiation and conve Lion, must be conk'iu uting to the thermal conductivity. The purpose of this study is to obtain data from whic;h predictions may be nmade of the rate of radiant heat transfer through pjru: insulations. This information would place a stronger theor, tical fourdation under the present thermal conductivity values for the se insulations. It would enable one to predict more accurately the effects of temperature and insulation physical properties on heat transfer rates. -1

-2Finally, the results of this study might suggest methods for making a more efficient insulating material. The lightweight insulations possess a relatively large amount of void space; usually more than 95 percent of their volume is occupied by gas. Typical solid materials now being employed in these insulations are polystyrene, polyurethane, wood fibers, and glass. From a geometrical point of view there are two types of insulations. In one type the solid is the continuous medium, the void space being dispersed gas bubbles; the other type of insulation consists of a matrix of fibers held together by a suitable bonding agent. The method which will be used to solve the problem applies to both types of insulation. However, the physical significance of some of the parameters used to describe the heat transfer processes will necessarily be somewhat different in each case. The amount of quantitative work on this subject is rather limited. Some of the authorities in the field of heat transfer have mentioned that radiation is a possible contributor in heat transfer through porous insulations. Assuming the insulation consists of a series of' opaque plates, Jakobl0 concludes that the heat transfer rate by radiation varies in proportion to the temperature gradient, the cube of the absolute temperature, and the space between the plates. Similar results were obtained by Bosworth.3 By applying diffusion theory to photon transport through gases, he obtained a relation showing the radiant conductivity to be proportional to the absolute temperature cubed and inversely proportional to the mean free path. The treatment of radiation through a heterogeneous catalyst bed led to a similar result.

Apparently the first quantitative estimate of the radiant contribution was contained in the work of Allcut. He reported the differences in conductivity of gl.ss wool resulting from first facing the insulation with paper and then with a.luminum foil. From the results, one might infer that radiation compnrised from 1 to 10 percent of the heat transfer rate. The insulation densitics ranged from 1 to 3 lb/cu ft and temperatures averaged 80 to 900F. One of the most extensive works on the role of thermal radiation was that of Verschour and Greebler.20 By computing the radiant heat interchanged between fibers, they obtained a series solution which, in the limit, converged to give a result nearly identical to that of Jakob. From measurements Df the absorptivity of glass fibers theoretical radiation rates were computed. The3e compared favorably with experimentally observed rates for evacuated insulations. Unfortunately, orly two fiber sizes were employed and these sizes were reported as the mean of sizes extending over an unspecified range. Hence, little could be concluded as to the effect of fiber size. Topper19 treated the case of thermal radiation in an opaque foamed insulation from a theoretical point of view. Again the result showed the rate to be proportional to the temperature gradient and the cube of the absolute temperature. EHniker hias presentcd a rather detailed theoretical analysis of simultaneous radiation and heat conduction. By assuming that radiation travels in only two directions, he was able to describe the radiation process by two differential equetions. If' one solves these equations and the equation of contirnuity, then the radiant heat transfer rate and temperc.ture distribution are obtained.

-4. The method of treating the radiation problem which will be employed in this study is nearly identical to the one Hamaker employed. Although considerable work has been done in this field, there are many points which need clarification. Most of the previous works conclude that heat transfer by radiation depends upon the cube of the absolute temperature and upon the temperature gradient. Effects of temperature on the various radiation parameters are unknown. The manner in which the fiber or pore size changes the rate of heat transfer has not been presented clearly. The importance of the emissive properties of the materials at the extremities of the insulations has been overlooked in much of the previous work. Another point frequently overlooked in dealing -with porous insulations is that the size of the pores or fibers may be comparable to that of the wavelength of the radiant energy. In this case the usual methods of solving radiation problems are apt to lead to erroneous results since the effective areas of bodies obstructing radiant energy may vary from zero to several times their geometrical areas. The problem of the transport of thermal radiation through insulations is very similar to pr )blems encountered in other fields. Meteorogists have considered the problem in computing the amount of sunlight which is transmitted by our atmosphere. The transmission of sunlight through turbid water greatly affects the plant and animal life of the oceans. The ability of radar and infrared energy to penetrate the atmosphere and artificial fogs is of corsiderible rmilit, ry importance.

PART I THEORETICAL WORK

DERIVATION OF THE TWO FLUX EQUATIONS In this analysis it will be assumed that heat is being transferred under steady state conditions through a piece of insulation of finite thickness and infinite lateral dimensions. It is also assumed that the material is sufficiently thick to contain a large number of fibers, or gas bubbles, and that the insulations are homogeneous. Homogeneity here implies that the smallest volume of insulation which will be considered is that which contains a number of fibers or pores per unit volume equal to the average of the number of figers or pores per unit volume taken over the entire insulation. In describing the radiation process it will be assumed that radiation may be propagated in only two directions, forward or backwards along the x axis, see Figure 1. The vector i2 designates the monochromatic radiant flux in the positive x direction, and the vector i1 designates the flux in the reverse direction. A monochromatic flux is defined as the amount of power between wavelengths X and X + dX passing through a unit area normal to the direction of propagation. The assumption of the two flux method is based upon the knowledge that the net flow of radiant energy can occur only along the x axis. A more tenable assumption would be the six flux model developed by Chu and Churchill.5 In this model, one flux vector is oriented along each axis of an orthogonal coordinate system. However, for this particular geometry, the four vectors perpendicular to the direction of heat flow are identical, so the only vectors worthy of consideration are those along the x axis. Thus the six flux model degenerates into -6

SMALL INSULATION VOLUME FACE OF T2 | 0 UNIT AREA i2. 1.ON2 0 x x + Ax I Figure 1. A Porous Insulation

-8a two flux one. The diffusion equation gives essentially the same result as the two flux method, but it is less convenient for fitting boundary conditions. The mechanisms contributing to changes in the rate of radiant transfer include absorption, reradiation and scattering. Scattering refers to the reflection or refraction of electromagnetic waves. The scattering process is a rather complicated one. It is necessary to consider not only the amount of energy intercepted by a given body, but also the manner in which this energy i3 redistributed about the body. Figure 2 shows a typical distribution of scattered energy about a body whose dimensions are of the same order of magnitude as the wavelength of the incident beam. Using the actual scattered power distribution in the proposed two flux method is an impossible task. For this reason the distribution of scattered power is arbitrarily resolved into two components, F and B, see Figure 2. The forward component, F, represent:- that fraction of the scattered power which continues to travel i-L the direction of the incident beam. The backward component, B, is that fraction of the scattered power which is reflected backward. By virtue of this assumed distribution F + B 1 (1) For the purpose of describing the amount of power scattered by a body, the scattering cross section, Ss, is employed. The scattering cross section is the fraction of the power scattered from a beam carrying one unit of power per unit area normal to the direction or propagation. If n is the number of scattering bodies per unit volume

g9ACTUAL DISTRIBUTION REPRESENTATION BY THE OF SCATTERED POWER TWO FLUX MODEL F SCATTER ING B PARTICLE DIRECTION OF THE INCIDENT BEAM Figure 2. Repreeentation of the Angular Distribution of Scattered Power

-10of a dispersed mediun, then l/nSs is the mean free path for scattering of a photon. The probability that a photon will bc scattered by traveling a distance dx in the medium is proportional to nSsdx. Treatment of the process of absorption leads to the use of the absorption cross section, Sa. This quantity is defined to be that fraction of the power absorbed from a beam carrying one unit of power per unit area normal to the direction of propagation. This cross section may be expressed in terms of the more conventional parameters used for solving radiation problems. For large bodies the absorption cross section is the product of the absorptivity and the area of the body projected onto a plane perpendicular to the incident beam. Since absorption is being considered as a possible radiation mechanism, the first law of thermodynamics then requires that emission must also be considered. In addition, Kirchoff's radiation law must hold. Thus a body of any size must be capable of emitting as much power as it absorbs. The bodies producing scattering and absorption are obviously different in the various types of insulation. Scattering occurs whenever electromagnetiu waves enc,-unter a discontinuity in refractive index. For fibrous insulations this condition exists at the surface of the fibers. In the foamed insulations scattering will occur at the surface of the gas bubbles. Absorption occurs by the interaction of radiation with the atoms and molecules of the insulation. Obviously absorption will be greatest in the solid materials comprising the insulation. In fibrous insulations this would be in the fibers. In foamed materials this would be in the solid which entrains the bubbles.

-11Now that the various radiation mechanisms have been mentioned, one may now proceed to describe how they affect the radiant heat transfer rate. The first step in this task is to examine the change in each flux as it passes through a small volume of insulation, such as the one shown in Figure 1. The face of this volubne is one unit in area, parallel to the boundaries of the insulation and perpendicular to the x axis. The difference in flux i2 at x + Ax from that at x is given by Ai2 = -n (S., + Sav) i2Ax + nF Ss i2Ax + nB S i lAx + ne A g(k,T)Ax (2) where n is the number of bodies scattering energy per unit volume, 8sX is the monochromatic scattering cross section per scattering body in the insulation, Sa. is the monochromatic absorption cross section per scattering body in the insulation, Ax is the volume of the insulation, A is the ercitting area per body in the insulation, and ek is the emissivity of the insulation. The function g(X,T) is Planck's radiation function which gives the energy radiated at each wave length for a black body. By dividing the above equation by Ax and taking the limit as ax vanishes, one obtains the differential equation di2 dx= -n (S, + S ia) i2 + nFXSski2 + nBSsil + nekA g(X,T) (5) Substituting equation 1 into equation 5 gives din' = -n (BXS X + Se,) i2 + nBXSsxil + ne (X, T)

-12If one defines the spectrslly integrated flux by 00 I2 = i2(x) dX (>) 0 then 12 is the radiant power per unit area traveling in the x direction. Introducing the integrated flux into equation 4 gives dI2 4 dx -n (BSs + Sa) I2 + nBSsIl + neA T (6) where the values of B, Ss, Sa' and e are now averages over the entire spectrum, said a T4 is the integral of Planck's function. 00 00 a T4= f g(k,T) dx = (7) C eCXT -(7) O 0 e where c1 is 3.74' (1k)5 erg cm /sec and c2 is 1.439 cm OK. The parameters in equation 6 are averaged such that 00 BS5 _ / BSX i2 dX (8-a) 2 s k 0 a 12 Sa i2 dX (8-b) 20 e = Ex g(x,T) dX (8-c) dr0 The use of the above definitions in eqaation 6 presumes that fluxes and 12 have similar spectral di.)ribujicrns.

Kirchof's law: requires that for radiati.n correspondirng to an absolute temper-.ture, T sa (T) =- (T) A Tnus dI2 4 kx -n (BSs + Sa) I2 + nBSSI1 + n Sa T (lo0) As a matter of convenience, the following groupings o' parameters are defined M = n(BSs + Sa) N = nBS (1) P = M-N = nSa The parameters M, N, and P are the interception, the scattering, and the absorption cross section, respectively, for a unit volume of insulation. Since the three parameters are related, only two are needed to:ie>cribe the radiation properties of an insulation. Employing these paramlxetcrs in equation 10 gives dI2 = - MI2 + NI1 + P aT4 (12) In a similar manner an equation may be written for the other coanpcier't -dI 4 d. - MI +NI + P aT L) dx 1 2

TREATMENT' OF SIMULTANEOUS RADIATIOI, COINDUCTION AND CONVECTION In the preceding section the rate of radiant transfer has been described by a pair of differential equations, 12 and 13. Coxnpletion of the problem requires establishing values for parameters M and N, solving the flux equation, and fitting the solution to boundary conditions. Unfortunately the flux equations have introduced temperature as another variable, thereby requiring another equation to solve the problem. This equation may be derived from an energy balance. At steady state the rate of heat transfer must be constant, and may be represented as follows qt -~ c + qv + qr (14) where qc is the rate of heat transfer by conduction, qv is the rate of heat transfer by convection, and qr is the rate of heat transfer by radiation. The conduction rate may be expressed by the Fourier law qc k d (15) where kc is the thermal conductivity. It should be noted that this conductivity has been defined in such a manner that it includes contributions from both gas and solid conduction. Convection may occur within the gas bubbles of a fosme e- insulation or in the space between fibers in a fibrous insulation.

-15Although the convection process may be a discrete one, it is expected that convection effects can be represented by,, the following rela-tion. =e k di (16) The above a-proximnation consists of using a continuous function to describe the behavior of a large number of discrete processes. Such approximations are commonly used to describe the macroscopic behavior of molecular processes, such as the conduction of heat through gases, and the rate of chemical reactions. The relations expressed by equations 15 and 16 serve as definitions of the conductivities, kc and kv. The deviations of experimental data from these assumed relations may be compensated by expressing the conductivities as functions of temperature or location. Now if one defines a gross conductivity as Kc = kc + kv (17) then the heat transfer rates by conduction and convection may be combined as dx qc + qv = - Kc Ax (18) Because of the approximations introduced i:nto the preceding derivation, little insight into the nature of the conduction and convection mechanisms can be gained by using the gross conductivity. However, it does suggest that both processes can be treated as if they were conduction, i.e., the rate is the product of a conductivity and the negative temperature gradient.

-16The rate of heat transfer by radiation will be the net flux qr 12 2 1 (Ia) Substituting equation 19 into equation 14 gives qt =q qqc +v I (2 2) 9t"9,'9,+ I2- 1 ~(20) Assauiing equation 18 to be valid, then equation 20 nmay be written as dT qt K2 d x

SOLUTION OF THE FLUX EQUATIONS Unfortunately, the flux equations, 12 and 13, are non-linear and an exact solution has not been devised. The use of an analog computer was considered as a possible means of solution. However, the computer was not capable of accuracy greater than that of an approximate analytical solution. The basis for the approximate solution comes from assuming that the temperature gradient across the insulation is constant. From equation 21 it is evident that the assumllption is valid if the rate of heat transfer by radiation is small, or is constant. If the temperature gradient is constant, then the absolute temperature may be expressed as a linear function of the insulation depth, x. By substituting this relation into the non-linear terms of equations 12 and 13, a pair of linear equations are obtained. Solving the linearized equations by the usual methods gives, (Cl N M+ (CfN2+M Il(x) = M( M) clx + 2 C2 N +MC sillh XT +.a T, + 1... + 2 12 (M+N) 2e 2 (M2 - N2) 24 a (T1 - T2)3T 24 a (T1- T)4 + _ + _. (M+N)(M 2_N2) 13 4 (M2-N2)2 -17

I(x) = C1 sinh x M2N+ C2 cosh x M2N2+ a T4 4 a (T1- T2)T3 12 a(T1- T2)2T2 24 a (T1- T2) T - (M + N) + 2 (M2 _ N2) f- (M2-N2)(M+N) 24 a (T1 - T9) + 4 4 (M2 N)2 (1 M) - M)C2 -M2 -- N2 qx+ -.. -...... sih x N N2 N 8 a (T1 - T2)T5 48 a (T1 - T2)3T (M + N) I (M+N)(MW-N2) e3 where C1 and C2 are arbitrary constants, ~ is the insulation thlic'ness, T1 is the temperature at x =, and T2 is the temperature at x = 0O The value of the arbitrary constants are fixed by boundary conditions at each side of the insulation. These conditions are: 4 I2 () = T + ( 1 - e)) I (D) (25) I () = E T +( - e) 2 (i) (26) where co is the ermissivity of the surf ace emitting radiation i,. o the insulation at x = 0, and To is the absolute temper.u.u1e of tilat surface. Equation 25 states that the flux I; entering the insulation at x = 0 is the sum of the flux emitted by material outside the insulation and

the flux I1 which is reflected back into the insulation. These equations assume that the material outside the insulation is opaque. For the case in which the radiation entering each side comes from a source which is at the temperature of each side, the values of To and TA in equations 25 and 26 become T2 and T1, respectively. If the insulation is absorbing and several inches thick, 22 40(T -T2)T I -q:'M2 - N2 + M + N rM -N + M N (1 (M 2 N) % (o)= % ~ —N + M N() a-(M +N) 24 a (T1 - T2) T ~~~2 2 l4 a (Tl - ~2) T2 ] N 0 (M2 2 In many cases of i:,terest t:.e only "rin. of i;nportance in equation 27 is that containing T,3 sr (~) =, eo - N M + 4 o ( T1 T()1 -NaM QN1-a (M + N) (27-a) Equation 27 mnill be a valid approxination if ~IYY2 is larger than 5. The siinplification introduced by equation 27-a will cause an error PAT of less than 10 percent providing that P 2 is greater than 10. 3 (T1 - T2)

If the insulation is non-absorbing, then the radiant heat transfer rate may be coCnputed exactly. The result is 0 A 0o In comparing.adiation with other heat transfer mechanisms, it is convenient to employ the radiant conductivity, kr, which is defined as kr d(29) Again assuiing that the temperature varies linearly with x k -T1r (29-a) For thick )absorbing'insulatione, khO) may be computed from equation 27-a kr(~ "eO4 N4 +R94 a T2 / M2, 11 + M + 1 2- ) For wno-ebporbing insulations a (T3 +T2T +TT 2 +T3) kr - -.,... 1 o o (U 1) 1. - 1 + N eo ++

Nearly all of the materials used in porous insulations are absorbing for:noderate t`e-mperaturef say le&:s than 10000F. Thus for most applications equat-ion 30 will be em ployed. It should be pointed out that the rcadiant conductivity iS no-; a constant in the OwEae sense as the thermal c.onductivity, for in many cases it varies approximnately as the cube of the absolute temperature. Now that the radiant heat transfer rates have been obtained, one may exairhe the eeffect of rad iL tion on thk teirperature distribution. Integration:. of equ&aio-n'_l;ives T (x dx 7' K: Kc2) For the case, in ni ih equatio)n 27-a is valid and both sides of the insulation arc bounded:'y a mO:tCri.l. of the sane emissivity T T 2 Kc N ( - 2O) Kc (M + N) (53) If data were aJva'il'IL~ on ti Wi, rl rta7 of hIeat transfci, tiic temperatures and the radiation parameters, then the gross condOucti-ity K and the temperatui'e diiti-ioutsiDn could bc corr.-uted. A temperature distribution Coiiputed from equ.ttion 33 is sh-o;m in Figure 3j.

ESTIMATION OF THE RADIATION PARAMETURS PROM ELECTROMAGNETIC THEORY For some idealized cases the radiation parameters can be computed. Because of the tremendous amount of numerical work for every particular case, much of the information will be qualitative. If one assumes that a foamed insulation can be treated as a random dispersion of uniformly sized spheres, it is possible to estimate the scattering parameter N. Rewriting the definition of N N = nBS5 (11) The value of n, the number of spheres per unit volume, is given by 3 Ps' P - --- ~ (34) 4rj3 Ps' Pg where R is the sphere radius, p is the density of the solid material, pg is the density of the gas, and p is the insulation bulk density. The scattering cross section is frequently given in terms of the scattering coefficient Ss Ks =,it2 (35) ~2 Substituting these relations into equation 11 gives 3 BKs P - P N =.-,,- (56) R Ps P The values of BKs could be estimated from the results of the Mie Theory.l1 Mie's calculations are a solution to the Maxwell Equations for a plane parallel monochromatic wave impinging upon an isolated sphere. A summary of this work has been presented by Stratton.17 The 9 ~~~~~~-~~~~ 3,~~~~

-23results of these calculations give the scsttering coefficient and scattering distribution as functions of wavelength, particle size, and refractive index. For two reasons no quantitative calculations will be made from Mie Theory. First, she refractive index is not known for any of the materials used in foamed insulations over the spectral region in which radiation occurs. The second reason is that a very large number of hours of computation by a high speed machine would be necessary. However, some data are available which illustrate the manner in which the scattering functions B and K. vary with sphere size. Figure 3 illustrates the behavior of these functions. The data for Fibure 3 were taken from the work of Boll2:ad Chu, et al.4 The values of B were calcul4aed by the method outlined by Clark, et al. These curves show scv-wr1 i:rportrat properties of the scattering functions. For large Opheres, say those greater than 100 microi.s in diameter, the scattering coefficinat approaches a value of 2. TDie value of B also approaches some constant. Then for large spheres the value of N varies inversely in -proportion to the radius. On the basis of a fixed mass of insulation, small bubbles would be nmore efficient for scattering radiation than la.rge ones. If one now conr~iders the case in which the spheres are quite small, say C.1 nmicron iu: dimueter, a different conclusion is reached. For such small soheres the value of D is constant at 1/2. The function Ks varies as the fourth power of' the radius, and hence N varies in proportion to the radius cubed. Since N is a measutre oi' the ability ox' an insulation to scatter radiati.,:, the larger the value of N, the smaller rate of heat

o~ SCATTERING FUNCTIONS B, Ks, BKs o~0 g g 0 0 o b II O 0

transl'er by raCir.tion. By the larg-e sphere approxirmation one sees that small spheres are mnore effective, and the sm<all sphere approxi.aation states that large ones are more effective. Obviously there must be an optimum sphere size for scattering radiation. The irregular geometry of the solid in a foamed insulation makes calculation of the absorption cross section more difficult. Repeating the definition given in the previous section P = nS (11) Assuming each bubble is enclosed in a cube of solid which is large in comparison with tihe radiant wavelengths S n-2/) (1 - e-u) (37) a where 6 is the expcnentiiLl absoripion coefficient and u is the mean thickness of solid in each cube enclosing a bubble. From a geometrical consideration of a sphere of rai las R in a cube of volxume n i, th value of u may!)e estimiated as u= X(; i, P- 4r(p./3 Ps- Pg/ PS P According to he above a-p)roximationls P 4 (1 e- (y) A similar treatment will be made for a fibrous insulation. This type of insulation is assumed to be a ranldoml matrix of inf'inite cylinders whose axis are in parallel planes ala oerpeneiiular to the

direction of heat flow. In this instance both scattering and absorption occur in the fibers. The fiber density, n, is given by 4 P- P n = 2. (40) fDf L P, - Pg where Df is the fiber diameter, and L is the mean fiber length in an insulation cube of unit volume. The scattering cross section per unit length of an infinite cylinder may be expressed in terms of a scattering coefficient. Ss = Ks Df (41) Then the scattering cross section of a fiber of length L is K Dp. Combining this with equation 40 leads to the following expression for the radiation parameter N, 4 BKs p - p N = - ~, (42) x Df Ps - Pg Electromagnetic theory can provide some information on the behavior of the parameter N as the fiber size is changed. Unfortunately, the amount of numerical work is equally great as for the case of the sphere. There is even one additional difficulty; the theory has not been fully developed. Previous solutions to the problem15 have termi14 nated with calculation of the field strengths. Montroll and Hart have calculated scattering cross sections for materials of refractive index less than 1.5. Because of the incomplete condition of published work on the theory, a derivation is presented in Appendix C. The results are listed below.

1 2 2 2 2 K5' a jAj2 + IP.1 + 2E'An'2 + IPn1I (43) nal where of is the cylinder circumference measured in wavelengths. = x Df (44) The values of An and Pn above are computed from the following relations. m*Jn(aC) Jn' (Alf) Jn(m4) Jn' (C) Jn(= Har) (c) nJn' (ma)' ( ) Hn(a) m*Jn(m*) in' (C)- Jn(a) Jn, (mC) (46) m*Jn(m*a) n' (a) Hn (a) Jn' (m*C) In the above expressions m* is the refractive index. The functions J and H are the first Bessel Function and first Hankel Function, respectively. The primed notation signifies differentiation with respect to the argument, thus -(x)-. j' (x) (47) dx Values for An and Pn are listed in Appendix D for m = 1.5, 2.0, and 2.5. The distribution of scattered energy can be computed from the coefficients A. and Pn. However, the manner in which one distributes the energy into the forward and backward components is arbitrary. One possible method would be to assiegn to F all the energy leaving the cylinder in the forward direction.

n/2 F = Sr(G) de / f Sr() de (48) -x/2 0 where Sr(9) is the radial Poynting vector at angle G, measured from the forward direction of propagation. The radial Poynting vector gives the energy flux passing through a unit area on the surface of a cylinder surrounding the scattering fiber. 00 o n+l R 1 24 IOF 2OPIn2rI l - + K(PnPk) 2n-1 + 2 Ij n=l n=l (49) C4 Z ~0 ( f sin(- k ini )1 K(PnPk) Re (PnP) n k + + n=l k=n+l where Re(z) signifies the real part of complex number z, and z is the complex conjugate of z. For parallel polarized radiation F, =F (A) (50) Most thermal radiation will be randomly polarized, so F = (F1 F,, ) (+F)

A few calculations were made of the functions Ks and F for refractive indices of 1.5, 2.0, and 2.5. The results are given in Figures 4, 5, and 6. Examination of the equations and the numerical computations show a striking similarity between scattering by spheres and cylinders. 14 Montroll's work demonstrates that Ks approaches a value of 2 for large fibers, say 100 microns in diameter. The back scattering coefficient also approaches a constant value. Naturally, the value of B and the mar-ner of approach to a constant value will depend upon the refractive index. On the basis of these assumptions, the value of the parameter N will vary inversely with fiber diameter. The case of the small fibers, say 0.1 micron in diameter, is similar to that for small spheres. The back scattering fraction, B, is constant at 1/2 while Ks varies as the cube of the diameter. Thus N varies as the square of the diameter. Again one concludes that an optimum fiber size exists for scattering radiation. The values of B and Ks in Figures 3, 4, 5, and 6 are expressed for the case of monochromatic radiation. Using the functions for black body radiation requires averaging them over the spectrum. One method would be to use Planck's function as a weighting function, as suggested by equations 8-a and 8-c. 00 I BK5(x) g(xT) dX BKB = 0 ~ ( 52) J g(xT) dX 0o

SCATTERING COEFFICIENT Ks.0~~~~~~~~~~~~~.~ I I II I II I 1 II II II II b 0 __ C1-3 3II I i *~~~~~~~~~~~~~~ CD 0 I-I.~~~~~~~~~~~~~~~~~~i~ — 41' 0F -j~ CD9

.5 z 0 H4 I —.4 3 z.3 oI I I I l nr'c~~~~~~J~~rDf a: rn - 1.5, --—. -- { m= 2.0,.... m= 2.5,... / a.I 0 7. Df Fi rire 5 The Back, Scattering Fraction for a.1 Infinite Cylinder

SCATTERING PARAMETER BKs o D 3 3 l 19'~~~ I I jI I I It~ I ~3I I I II IC+ rUI Cii ~cFn~I! cF 0d FIIi~

-33The function L(X,T) was defined by equation 7. This procedure gives each wavelength a weight in proportion to the energy it carries. The absorption paramueter P was defined by P- nSa (11) In terms of the insulation properties 4 K p-p gDf Ps' Pg where K, is the ratio of' the &~bsorotion cross section to the geometrical cross section. From. ele trouiagnetic theory K Re(AO + Po) - oA01 12 2 R(An + Pn) + IAn12 n=l + Pr 2 (-4) The nuierical computation of Ka is especially tedious. If the complex refractive index mn* is real, then the material in nonabsorbing and Ka vanishes. For large fibers, Ka will be equal to the absorptivity of the material. Then for this case the absorption parameter P varies inversely with fiber diaA:eter. This result is icde.Ltical to that for the scattering parameter N. Absorption for very small fibers may be obtained fromt K a - ak 1 + 2 log ( ) ()

-34where a is the real part of the refractive index iad k is the i.aiginary part. For non conduct.lng dialectsi: mar'erials k is zero and im is the square root of -he wield tric ox..tan. C nductors, such as rnetals, may have Large vauLues of mn and k. A qbuntitative definition of m and k in terms of thte dit lectric coneti;n; and the electrical conductivity may be found in Stratton. 7 Substituting equation 55 into equation 53 shows that P varies approximately as the square of the diameter. Again, as in the case for the scattering parameter N, an optimum size exists for blocking radiation. The exact values for these optimum diameters would vary with the refractive index of the insulation and the spectral distribution of the radiant energy. It would be fortuitious if the optimum size for both scattering and absorption would coincide.

PART II EXPERIMENTAL PROGRAM

GE2NERAL DESCRIPTION OF THE METHOD The experimental work consisted of measuring the transmission of radiiant energy through various thicknesses of the porous insulations. Transmission here refers to that fraction of the incident energy which passes through the insulation. A heated tube served as the source of radiation and was located far from the insulation. By using a blackened thermocouple as a radiation detector, one can easily measure the flux incident upon the insulation and that which is transmitted through the insulation. The ratio of these quantities is the fraction transmission. A plot of fraction transmission against insulation thickness yields a curve which may be termied a transmission curve. From this curve it is possible to estimate the radiation parameters M and N defined by equation 11. Substitution of these values into equation 24, along with the proper boundary conditions, determines the radiant heat transfer rate. -36

EXPERIMENTAL CONDITIONS Many insulations were tested, and eleven were sufficiently homogeneous and transparent to yield a reproducible set of transmission data. Two insulations were of the foamed type; the others were fibrous. One of the foamed insulations was Styrofoam 22. This material is expanded polystyrene and appears to be white. The other foamed insulation was Foamglas. This insulation is expanded glass and contains carbon particles, thus making it appear black. The commercial fibrous insulations were designated as Fiberglas A and B. In addition to fibers, these 1n1iulations contained some glass in the form of spheres and were bonded by a phenolic material. These bonding mzteriuls comprised about 20 percent of the mass of the insulations. Seven of the fibrous insulations were especially prepared by the Owens-Corning Fiberglas Corporation and contained no binding materials. Single filEaent glass of diameters 2.5, 5, 10, 20, and 35 microns was utilized in preparing five of the specimens. The other two insulations contained small amounts of shot and were composed of fibers of varying size with mean diameters of 0.17 and 0.20 microns. The mean diameters were assumed to be the arithmetic average of a series of 200 fiber diameter measurements taken by an electron microscope. The results of these measurements are tabulated in Appendix F. All1 of the fibers in the insulations were oriented in parallel planes. Table I lists the properties of the insulations. The glasses employed in making the insulations had a specific gravity of about 2.5. Their nominal composition is listed in Table II. -37

-38"TABLE I. THE PHYSICAL PROPERTIES OF THE INSULATIONS Mean Fiber or Insulation Manufacturer Pore Size Bulk Density Styrofoam The Dow.05 inch 1.8 lb/cu ft Chemical Co. Foamglas Pittsburgh.05 inch 9.0 lb/cu ft Corning Corp. Fiberglas A Oens Corning 1.0 micron 0.6 lb/cu ft Fiberglas Corp. Fiberglas B Owens Corning 5.0 microns 0.6 lb/cu ft Fiberglas Corp. Unbonded Glass Owens Corning 0.17 micron 0..50 lb/cu ft Fibers Fiberglas Corp. Unbonded Glass Owens Corning 0.20 micron 0.50 lb/cu ft Fibers Fiberglas Cor). Unbonded Glass Owens Corning 2.5 microns 0. O0 lb/cu ft Fibers Fiberglas Corp. Unbonded Glass Owtns Corning j.O microns 0.50 lt/cu I Fibers Fiberglas Corp. Unbonded Glass Owens Corning 10.0 microns 0.o0 lb/I u tc Fibers Fiberglas Corp. Unbonded Glass Owens Corning 20.0 microns 0.50 lb/cu ft Fibers Fiberglas Corp. Unbonded Glass Owens Corning 55.0 microns 0.)0 lb/cu ft Fibers Fiberglas Corp.

-3;'T'IYLE II. THE CHEMICAL COi)POSITION OF THE GLASS IN THE FIBROUS INSULATIONS Component Percent by Weight Si.o2 54.1.tC~a'_j ~14. 6 CaO MgO 21.8 B205 8.4 Na2O o.6 Fe205 0.2 F2 0.5 100.0 Temperature affects the parameters M and N in two ways. First, it changes the insulation physical properties such as density and refractive index. Second, a change in temperature changes the spectral distribution of the raciant energy. For the mattrials used in these insulations the spectral distrloution effects are more important than the others. Because of this, the data were taken with the insulation at room! tenmperature and correlated as a function of source temperature. Si.,ce the mnaterii ls were at rc om t(eaper&ture as thle measurements were taken, obviously no transmission dcata could be ob-l.ined for source temperatures lower thoLn room temperat ire. Actually, the noise in the thermocouple syjte.i, and the difficldty in o rtaining very thii homnogeneous insul.tion s&.ples', fixed the wir.iniu.l source te:mperature considerably above room temiperature. For the fibrous material this

temperafture was 200OF. The high absorptivity of 5he blai k Famcd las eliminated the possibility of t:ki:lg do:ta below;OC)~F. St)yrofoam was more transparent t.nd c,\uld be (;ut inv very thin pieces; so data were recorded as low as 110~F. In general the highest source temperatures studied were those corresponding to the upper limit of commercial utility of the insulations. For Fiberglas A and B this was 60CCF. The data for Styrofoam extended to almost 5000F which was 1000F in exc ess of its maximtum serviceable value. Data on the cth+ r materitals extcnded to 800~F. The minimuam measureable value of transmission was fixed by the noise level in the detector syz:tern and thie amount of stray radiation. For most cases data were.., raker when the transmission signal was lower than the noise. In eneral this lower signal limit was -4 about 10 of the signal from the incident radiation.

EXPJi.MENrAL EQUIPMENT The main components of the experimental equipment are shown in the schematic diagram, Figure 7, and the phutograph, Figure 8. These components are: 1. The black body source 2. The chopper 3. The insulation 4. The detector 5. The mpliA:ier 6. The recording system The black body source ( Jnsi,ted of a copper pi;e wrapped with heating wires and covered with 2 incaes of KaylD instatiin. The pipe was 36 inches long and 4 inches nominal diumeter. kn aluminum plug was used to seal one end of the pipe. This plug was cylindrical in shape externally, with a conical cavity. The purpose oi' the cavity was to increase the anrougt of radia;ing surface farcing the mouth of the source. A coating of SauereisEn ceiment on the outside of the pipe served as electrical insulation for the heating wires. These wires were size 18 Chromel-A, 58 feet in length. To facilitate'ranpiing the wire around the pipe, the wire was divided into three segments, with each piece extending over one ftoot of the pipe length. I RMount of wire in each segment was adjusted to give a uniform temperature along the entire pipe length. The length of wire in the segment nearest the mouth was greatest; the length of the other two segments were one-half and one-fourth the length of the first. Temperatures along the pipe -41

1 0 CYCLE BLACK BODY SOURCE RECORDER IO CYCLE INSULATION CHOPPER - AMPLIFIER DETECTOR Figure 7. A Schematic Diagram of the Apparatus Used in Obtaining Transmission Data

P j\I ~ O:'- - F g e 8P gr h o t e p a t s ec i O t i n T an Ht *~~~~~~~~~~~~~~ 8m Figure 8 A Pllotogaph of the Apparatus Ueed in Obtaining Transmiss on Datew

-44were determnined from four iron-constwtitan thermot. ):ul!:3 fastened to the inside of the pipe at interv:,ls of one foot. At 500~F the greatest variation in temperature along the pi,;e wuas lQF. The emitting area was a 3-1/2 in. diameter circle. The remainig front area of the source was covered with aluminuLn foil to reduce stray radiation. The chopper consisted of a two bladed propeller, 18 inches in diamietcr driven by a synchronous mot,.r. The propeller was cum from a 1/16 in. sheet of plexi-las, covered with aluminumn foil, and then painted black. As a result of the chopping, the radiation falling upon the insulation varied periodically in time as a square wave. The frequency of this radiation was 10 cycles per second. Using modulated radiation had three distinct advantages. If one measured only the 10 cps radiation, then there was no need to correct the thermocouple signal for changes in room temperature. Second, modulation should eliminate reradiation of absorbed energy. Finally, it permits the use of an a.c. amplification system which is more stable than a d.c. system. This greater stability means that greater amplifications are possiLle, and lower signal levels may be utilized. The insulation specimens were held in plywood frames 12 in. square and 4 in. deep. These frames were supported by wooden brackets with the insulations normal to the incident radiation. One frame was fitted with screen wire facings, so that the insulations could be compressed. The four commercial insulations were 11 in. squares and varied in thickness from 1/8 in. to 1 in. The fibrous insulations containing no bonding agents were mounted in the shape of 6 in. diameter circles and were less than 1/a in. thick. By masking the front of these samples to reduce their area, and observing no change in traunslission,

-45it was possible to show that edge effects were negligible. Thus the insulation samples could be considered as infinite slabs. Probably the most important component was the thermocouple detector. The instrument is very similar to those employed in infrared spectrcmeters. It was constructed in Detroit by Mr. Charles M. Reeder and Company from Junction materials containing bismuth, antimony, and tellurium. The face of the receiver was darkened with an evaporated film of gold black and covered with a potassium bromide window. The interior of the detector was evacuated to reduce its noise level. By showing the detector signal to vary as the inverse square of distance from a point source, one can demonstrate the linearity of the detector. Such a test did show this instrument to be linear, see Figure A-2. Since the transmitted radiation strikes the receiver at high angles, it is possible that reflection from the detector window could cause the detector to give low signals. By measuring the angular response of the detector, and assuming the transmitted energy to be diffuse, it is possible to estimate the maximum error. The results of this estimate are given in Appendix A and show the maximum error to be 15 percent. Actually the error was less than this value. This fact was demonstrated by measuring transmissions with a detector not fitted within a window and comparing the results with those of the other detector. In all cases the reflection error was less than 2 percent. The detector was mounted on a brass block. A brass screw 1/2 in. in diameter passed through a tapped hole in the center of the brass block. By turning the screw it was possible to move the detector horizontally. This detector traverse system was driven by a

-46reversible motir so the detector could be moved forwe..rd or backward along a 3 in. p'th parallel';o the. i:;sullti_.n. With this mechanism! it was possible to obtain point by pirita trar'smission data which then could be averaged to yield tr.rnsmris';ions for non-hom~oigeneous materials. A series of measurements were taken of the unattenuated flux at several source temperatures. This data indicated that the detector signal was proportional to.he difference of the fourth power of the absolute temperatures of tile souirce and detector, see Appendix A. Since the deviation of individual points from this relation is small, both source and detector must be nearly blackA. Since the signal from the detectr was often as low as a few hundredths of a microvolt, signal anplification had to be employed. The instrunent used for this purpose was a high-gain 1U cycle amplifier, model 33-A, made by Electro-Mechanical Research, Inc. of Ridgefiela, Connecticut. A modification was made on the amplifier since it was designed for use with high impedance receivers such as photo.tubes. As a result of the modification, the input signal was fed to the pri:..ary of the transformer, and the secondary of the transformer was connected to the grid of the first ampli~:'ication sta-ge. The reason for the modification was to increase the signal before it was mixed with then electrical noise in the first stage. The transformer employed was type HG-3 of the Thermador Corporation in Los Angeles. The maximum gain of the amplifier was about 58 million, and the noise was about 0.005 microvolt for a 10 ohm input impedance. Sharp tuning in the ithird stage of amplification eliminated most of the signal at frequencies other than 10 cycles, see Figure A-l,

Output signals from the amplifier were connected to a network of precision resistors. By proper selection of resistors, voltages corresponding to 0.001, 0.0025, 0.005, 0.01, and 0.025 of tlie output signal could be obtained. These voltages could be measured either by a Leeds and Northrup porta1le potentiometer or a Brown Electronik recorder. The recorder was especially useful for measurements near the noise level or for measurements on non-uniform insulations where averaging of the- data was necessary.

OPERATLNG PROCEDURE First, the source heater was turned on and the voltage adjusted to give the desired steady state source temperature. Attainment of steady state conditions usually required several hours. The next step was to remove the detector from the storage dessicator and install it in position in the traverse mechanism. Because the humidity of the atmosphere was likely to fog the potassium bromide detector window, it was stored in a dessicator. The chopper, amplifier, and recorder were then turned on. By proper selection of network resistor and amplifier gain, the detector output was indicated on the recorder scale. Next the traverse motor was started and the recorder traced the unattenuated radiation field over the path of the detector. From this trace and a knowledge of the variation of flux with source distance, the value of the flux incident upon the insulation could be computed. The insulation was then placed between the source and the detector. By increasing the amplifier gain, the detector signal again fell upon the recorder scale. Another traverse was made, giving a value for the transmitted flux. The average of this flux divided by the average incident flux was the fraction transmission for the insulation. If noise was appreciable, an additional traverse was made. In this instance the radiation was blocked by a brass plate inserted in front of the insulation. The time of the traverse was about 30 seconds, and the recorder trace gave a measure of the noise over this -4b

-49time interval. The mean noise was estirmated by finding the area under the trace and dividing by its length. This value, when corrected to the proper gain, was subtracted from the transmitted and incident flux measurements.

EXPERItMENTAL DATA The calculation of the fraction transmission frora a set of measurements will be illustrated, Because the radiation was not collimated, the flux decreased with increasing source distance. Since the detector was behind the front of the insulation, ihe incident flux measured by the detector must be corrected to that at the front of the sample. Figure 9 illustrates the flux variation with source distance. For calculation of the transmission of 285~F radiation through a piece of Fiber6las B the following data were needed: Distance from detector to source 14 in. Distance from insulation to source 15-1/2 in. Incident flux at detector 3.32 millivolts Gain 1.00 Transmitted fiux 1.35 m.v. Gain 10 Noise 0 Incident flux at insulation 3 d - 5. 5.60 m.v. (Using Figure i) 5.00 Fraction Transmission x = 0.037 10 i.6o The manner in "which the ~:tsullation thickness was determined will now be described. "or the thickest specimens of each of the four commercial insulaiLons, the mass, thi 2kness, and area of each specimen were measured. These data determined the bulk density of the insulations. The thinn,r Specimens were assumed to have the same bulk density, and their thickness was fbxed by mreasuring their mass and area. -50

RELATIVE DETECTOR SIGNAL ~ o I C) —.I o ICJ Z 0 z rr_ _ _ _ _ __l_ _ _ <~.0

-52-. For the unbonded fibrous insulations the bulk density was assumed to be 0.50 lb/cu ft. By simply measuring the mass and area of each specimen, the thickness was then fixed. In describing the transmission properties of a given specimen, it is necessary to specify both the bulk density and thickness. The arbitrary assumption of bulk density merely provided a convenient basis for presenting the data and does not limit the generality of the experimaental results. The influence of bulk density will be discussed later, so that the results of this study may be applied to fibrous insulations of density other than 0.50 lb/cu ft. In terms of the flux variables defined in the theoretical section, the flux incident on the insulation is I1 (2), and the transmitted flux is I1 (o). The flux 12 (D) is zero. Writing these as boundary conditions: I (e) - 1 1'~ Ce) =1 ~(56) I2 (o) = o The reradiation of energy should be very small. First, the amount of power passing through the insulation is small, so that the insulation is essentially at the same temperature as the detector. Second, the detector will respond to only the ten cps component of the reradiated energy. By omitting the reradiation term from equations 12 and 13 dI2 __ = -MI +NI (12-a) -dI1 = - MI + NI (13-a) dx

-53Solving these equations for the transmission, gives JM - N2 1WG _N2 cosh - N + M sinh I M2 NC Expanding equation 57 in an infinite series 2 r - c I1 (O) = 2ry N -j\I- e J2M -N N+M r 1 n ~~ ~n 2 I 2In _-M2n, JM2 N( M2 N2 + M If the value of JM2 _ N2 is large, then the series portion of equation 58 reduces to unity for thick insulations. (o) 2 jM. - N2 -. M2N2 1 (0) ~-~Ze (59) 1 - N +M For cases in which the assumptions of equation 59 are valid, then a semi-log plot of the transmission data will be a straight line. Typical sets of the transmission measurements are presented in Figures 10-15. A tabulation of all the transmission data Is given in Appendix E. From the plots of the data, it is evident that the logarithm of the transmission varies linearly with thickness for large thickne'ies, as predicted by equation 59. However, at low thicknesses the data follow a curve, for example, see Figure 15. As the thickness vanishes the fraction transmission must approach unity.

-540.1 0.01 or\ z 0.001 0 C) i880 F SOURCE 0 400~F SOURCE 0.0001 0 0.1 0.2 0.3 0.4 INSULATION THICKNESS, IN. Figure 10, T2he 2raoaz~icsion of.la'k::udry Radimi., I. t ruuh Foamz.l.as

-55H100F SOURCE 0 170~F SOURCE O 285~F SOURCE o.I 0.001 0.001 00001 O 0.2 0.4 0.6 0.8 1.0 INSULATION THICKNESS, IN. Figure 11. The Transmission of Black Body Radiation Through Styooa

-561.0 210~ F SOURCE 0 285 F SOURCE 468'F SOURCE O 5800F SOURCE V 0 I co () - 0.01 ODOI.... l.0001 0.2 0.4 0.6 0.8 1.0 INSULATION THICKNESS, IN. Figure 12. The Transmission of Black Body RIadaetion Through Fiberglas A

-571.0 2100F SOURCE O 285~F SOURCE 400~F SOURCE O 580~F SOURCE v 0.1 o, I I -' —-' —-_-' -' 0. 0.001 O... 0 0.2 0.4 0.6 0.8 1.0 1.2 INSULATION THICKNESS, IN. Figure 13. The Tranmismision of Black Body Radiation Through Fiberg.ea B

-581.0 BULK DENSITY 0.5 LB/CU FT z 0.1I 0 0.2 0_4 0.6 0.8 1.0 _o 0L 0.01 0.17 MICRON DIA. FIBERS 0 0.20 MICRON DIA. FIBERS 4 2.5 MICRON DIA. FIBERS o 0.001...... __.... 0 0.2 0.4 0.6 0.8 1.0 INSULATION THICKNESS, IN. Figure 14. The transrission o~ iOO0F Black B0dy Radiation Through GLasS Fibers

-591.0 BULK DENSITY 0.5 LB/CU FT 5 MICRON DIA. FIBERS 0 10 MICRON DIA. FIBERS A 20 MICRON DIA. FIBERS O 35 MICRON DIA. FIBERS 7 01.0 01)0) 0 2 3 4 INSULATION THICKNESS, IN. F'tgI\ 15

-60A series of measurements were taken in an attempt to establish the effect of bulk density upon the radiation parameters for a fibrous insulation. In these mreasurc.,ent; the bulk density ofi an insulation specimen was varied by comi.ressing the mintErial between screens. At each particular bulk density, the transmission was measured. The densiti:-s varied from 0.5 l1/cu f't to 5.0 lb/cu ft. The results of this test demonstrated that the transmission through a aiece of fibrous insulation was independent of the bulk density. From equations 42 and 53 in the previous section the parameters N, P, and M should vary with density according to the factor P - Pg/Ps p Pg for a fibrous insulation. If the bulk density has any influence upon transmission, then equation 57 requires that the quantity -M2- N2 must vary with density. Under the experimental conditions described above, the thickness may be described by B _ w P Pg (60) Ps P - Pg where w is the weight of the solid in a unit area of insulation. From equation 60 and the density factors in equations 42 and 53 it is evident that N2is independent of bulk density. Hence, the theoretical work predicts that the transmission through a piece of insulation is independent of its density. Since experiment and theory both agree on this pont, it is assumed that equations 42 and 53 describe the influence of density upon

-61the radiation parameters. It is expected that this relation might fail in two cases: 1. The fiber orientation changes with density. 2. The fibers become compressed sufficiently so that their scattering properties change with density.

PART III ESTIMATION OF THE RADIATION PARAMETERS FROM TRANSMISSION DATA

THE METHOD The values of both the scattering parameter N and the interception parameter M may be estimated from trans:.ission curves, Figures 10-15. From any two points on a curve these values could be computed using equa.tion 57. However, the data scatter such that the answers vary considerably, depending upon which two points are selected. The random nature of the deviation of individual points in Figures 10-15 fror the curves seemns to) indicate that the errors are due mainly to non-unifor.niJy in the samples rather than a failure of the two flux theory to predict the behavior. For those data whi h f llow straight lines on a semi-log plot, it is possible to compute the parameters which best fit the data by the method of least squcare. This method gives the slope and intercept of the line, and the parameters M and N are then computed from equation 59. It may be noted in Figure 15 that the data fail to follow the straight line relation. This behavior was interpreted to mean that the insulations wrere not sufficiently thick for equation 59 to be valid. In this case the radiation parameters must be evaluated by fitting equation 57 or 58 to the dat4a. Since equation 57 is poorly suited for least squares analysis, the series expansion, equation 58, was utilized. By empl)ying a seri-_s Jf;u tessi.re approximartions it was possible to obtain a least squart s fit to the data. The first step -63

in the pro".:edure:.as to estimate the sutra. i ht line regi,:,i!f'h: T1-ta. From. this line the approximate vcalues of M and N vere ut':va lsing equation 59. Nca.t, 5ha series;t;icn <)' eqaOutio0:'i,?'ed from th esti,:nated vt;lucs of M ard N. Frnm the oriji; d t,. e' s-eries portion of equation 58, the radic' tion parauleters may be ob tanieo fromn e(uflation %8 using least squares. IK th~ latter values of M and N differ ap.)receirtly from the ass;srm-ed oia;, the operetion;:.us.t be ra e.?atcd. A &t:tail~u dr exa:,216. t1of ihi prci ee ii pr,;,ented in Apen!( ix B. Thle curves drawn tr th,ru hh h cldata ir. Fi6ures 10-ij art those obtatinec.',y fit-irL the two flaX tlau sixs.i n equatitin to;-he data,

TIHE D^n.T.T. raTO'PAR?.T R. The radiation constalnts obtained fromn the tranisismsioni measurenments are pr sented in Figur0_s 16-21. The first four -;lotis.m'ow t;he ~nch-ier in -hi h radiotioin tel. icrture inflaenaces the radiati in parameters fur -he cor:.mercial i-isul.tions. On tht next two plots, one -Ua.y- ob.;erve t.ie inlfluence of radiation -teriperat-ure and fiber.ize upon the par.net_ rs for ut lndted 613ass fisers. Frl. ti;e definitions i.n the th-..reti al sec-vion, is is recalled oh t -,bilc para..ctcr M is she intercepti Dn cross sestionr. per uni' vsolunele of in.ulation. It is a measure of the at'lity of amn i.:sulcsion to sca -.r acn sos;-orb radiation. Obviously, the larger the value of M, the s.-i.l er.wiill be the radiant!lea transfer rate. Another,arameter wa; I, w -n,. -cattering cross section per unit insulation vclonie. The par=,-i-,,c:-,. - P b:m b trmIed the absorption Criss sction per univolume o Fro,..he -a: ta of Figure 16 for Foamnlas, one notes that the in srceptn f oai mr.-~s-,.i~e- is m'liuch lari er -than the absorptioLn paraneteer. This appears t:. i. i;- tlat.th. c.t;eri-g process is -,redomriinEmt over absorptJicn. -fo.r. 1iocking radiation. The factors causing the do casoe in the. iiL coei'ffient';ith increasing temcrir.ture are diff'i.1z1a- tu p;ectit'y';;ii );;t addi-tixnal dat. Bec::.use the s-oectral distribution is:.n-'.;i-'Q with te:;pr'.u.e, bc -h the refrac'tive i:-Z:n and,ffect<ie,"!.re ri' e.ri Chn-:mgi., Eithcr effect cuuld ce.use the observed bEah v.or. 1 Ti, - 1,bs'or- ti n, para.ie.'r ifor Foa.:m- I'ns apea rs tor be rl — Itively independent of temperature. One poss ible explanation for this -65

100 90 80 w. 50 z 0 240 60 200 400 600 200 TEMPERATURE, OF FIGURE 16 THE RADIATION PARAMETERS FOR F OAMGLAS

-6740 30 V) I0 0.0 UJ 20 0" 0 0 00 200 300 400 TEMPERATURE, O F FIG. 17 THE RADIATION PARAMETERS FOR STYROFOAM 22

1 10 - - TV _ 100 90 80 I. 70 Iw 2 60 0 1, 5C,1. II n, 1z 40 FIBERGLAS A 30 (FIBERGLAS B 20 10 0 200 400 600 TEMPERATURE OF FIGURE 18 THE INTERCEPTION PARAMETERS FOR SOME BONDED FIBROUS INSULATIONS

-69FIBERGLAS A.9.8.7 i.6 JC.l.5 C2.1.2 0 200 400 600 TEMPERATURE, OF Figure 19 The Absorption Parameters for scme Bonded Fibroue Insulations

60 50 Insulation bulk density 0.5 Ib/cu ft 200~F source temperature 0 300 F source temperature V 400 ~ F source temperature 0 Z 6000F source temperature do.~ ~8000F source temperature X 40 I:: I-,aJ 0,_ w_ 20 I0 0 I0 20 30 40 FIBER DIAMETER, MICRONS Figure 20 The Interception Parameters for CGlass Fibers

Insulation bulk density 0.5 Ib/cu ft 2000~ F 0 3000 F V 4000 F 0l 6000 F A 8000~ F X 0.3,V LL) IVV\ IL 0( mn 0.1 0 10 20 30 40 FIBER DIAMETER, MICRONS Figure 21 The Absorption Pararleters for Glass Fliers

-72behavior may be derived from tUic a-bsor: Lng iproperties of two of tile constiuenlts, glass aind carbon. Materials made of glass absorb very little in the visible region, absorb moderately in the near infrared, and absorb strongly in the region of wav,.lengths frotn 6 to 20 microns. As the source temiperature decrc ses,:s ore a.ld more of the radiant energy is carried at longer wavelengths, wlhere glass absorbs more strongly. Thus onle would anticipate that the absorption cross section of insulations made of glass would increase with decreasi::g temperature. The behavior of carbon is quite different. In the visible spectrum carbon absorbs strongly, but becomes relatively transparent in the infrared. From these properties, one would expect the absorption cross sections of insulations contain ing carbon to decrease with decreasing temperatures. It may be that the absorption properties of carbon and glass are complementing one another so that the absorption cross section for the insulation varies only slightly with temperature. From the data for Styrofoam in Figure 17 one notes that the interception parameter decreases with decreasing temperature while the absorption parameter increases with decreasing temperature. The absorbing characteristics of the polystyrene are qualitatively similar to those of glass. Thus the behavior of the absorption cross section is what one would anticipate. The decrease in scattering cross section with decreasing temperature is difficult to explain in the absence of data on the refractive index. It is interesting to note how the relative importance of the absorption and scattering processes change with temperature. At 200~F the dominant mechanism for reducing radiation is scattering, while at 100~F the dominant mechanism is absorption.

-73The interception parameters for Fiberglas A and B are presented in Figure 18. Both sets of' values decrease with increasing temperature with the greatest change being exhibited by insulation A which was composed of small fibers. Absorption parameters for these insulations are given in Figure 19. The data for insulation B follow the predicted behavior for glass. However, the data for insulation A show a maximum. Since this behavior is not exhibited by any of the other insulations, it is probably caused by the absorbing properties of the bonding material in the insulation. It should be pointed out that the interception parameter is much larger than the absorption parameter. This indicates that scattering is probably the primary process for limiting radiant heat transfer in these materials. The radiation parameters for unbonded glass fibers of 0.50 lb/cu ft bilk density are presented in Figures 20 and 21. From Figure 20 one may observe the effects of temperature and fiber size upon the interception parameter M. As in the case of the commercial insulations these quantities decrease with increasing source temperature. The influence of fiber size is great. For large fiber sizes the parameter M follows a hyperbolically shaped curve. In the range of fiber diameters from 2 to 5 microns, the parameter M passes through a maximum. The location of this maximum appears to shift to slightly smaller fiber sizes as the temperature increases. For fibers less than a micron in diameter, the values of M decrease with decreasing fiber size. The behavior of M with changing fiber size was predicted qualitatively in the theoretical section. For large fibers the value

of M should follow a hyperbola, asymptotically approaiching zero with increasing fiber size. For very small fibers the values of M should vary in proportion to the square of the fiber size. In Figure 21 the absorption parameters P are presented for unbonded glass fibers. In general, increasing temperature decreases the absorption parameter. However, the influence of temrperature change is less than that for the interception parameter. To avo id crowding, only one curve is presented on Figure 21. This does not imply thatz all She data should fall on one curve. Varying fiber size produces effects srililar to those observed in the case of the interception parameter. At low fiber sizes, the absorption par:nieter vanishes. For fibers in the range of 2-5 microns in diameter the value of P is a maxiLlum, and for large fibers P follows the hyperbolic relation.

COMPARISON OF TEA L AND THEORETICAL VALUES FOR TEE RADIATION PARAWMTRS In an attempt to check some of the experimental values, a few estimates have been made of the radiation parameters from electromagnetic theory. As mentioned previously, the refractive indices for materials used in fibrous insulations have not been established in the spectral region of interest. Even if these values were available, the amount of numerical work necessary is such that a digital computer would be required. There is otne additional difficulty in the calculation of the scattering parameter which should be mentioned. In fitting the distribution of scattered energy to the two flux model it is necessary to divide the scattered energy into two components. The method of distributing the scattered energy is arbitrary. In this respect radiation parameters calculated from theory are only approximate. From the calculations presented in the theoretical section, the values of N may be estimated for glass fibers. In these cases the refractive index is presumed to be real and have values of 1.5, 2.0, and 2.5. The first step in the calculation procedure was to solve equation 52 using the data presented in Figure 6. This gives the mean value of BKs over the spectrum at each source temperature. Substituting this into equation 42 with the appropriate densities, yields the value of the scattering parameter N. In this case the insulation was considered to have a bulk density of 0.5 lb/cu ft and the glass a density of 156 lb/cu ft. The results of these computations are presented in Figures 22-25. In the case of the data for the 200~F source, Figure 22, the experimental values exceed all of the theoretical ones. This appears -75

-7T80 Source teimperature 200~F Bulk density 0.5 Ib/cu ft Experimental values Theoretical values, m= 1.5 TI. m=2.0 - z~~~~~ m- 2.5 -- -60 Ldw cn 40 I'20 0/ 0 10 20 30 FIBER DIAMETER, MICRONS Figure 22 A Comparison of Theoretical andl Experimnental Values

-T7I00 Source temperature 400~F Bulk density 0.5 Ib/cu ft Experimental values ---- Theoretical values, m=1.5 m =2.0.80 8 lm=2.5 ----- 80 z LJo _ 60 0. w 40' (I) 20 O I 0 20 30 F IBER DIAMETER, MICRONS Figure 23 A Comparison of Theoretical and Experimental' Values for the Scattering Parameter for Glass Fibers

-7&I00 Source temperature 600~F Bulk density, 0.5 Ib/cu ft Experimental voalues --- Theoretical values, m=l.5 80 m=2.0 -mm z 80 _ 4 0 1 Z 40AM 20 Fu0 IO 20 30 FIBER DI AM ETER, MICRONS Figure 24 A Comparison of Theoretical and Experirental Values for the Scattering Parameter for Glass Fibers

-79~ 140.__ _ Source temperature 800~F Bulk density 0.5 Ib/cu ft i Experimental values ---- 120 Theoretical values, o 10 2m 1.5 ~ m =2.O 80 60 Ik w40 0 10 20 30 FIBER DIAMETER, MICRONS Figure 25 A Comparison of Theoretical and Experimental Values for the Scatteromg Parameter for Glass Fibers

-80to indicate that either the function BKs canmnot be represented by a single real refractive index between 1.5 and 2.) over the spectrum, or the method of obtaining B is very inaccurate. In successive plots, one observes that the theoretical values increase with temperature while the experimental ones decrease. Finally, in Figure 25 many of the theoretical values exceed the experimental ones. While no quantitative agreement between theory and experiment has been attained, the plots do demonstrate that both theory and experiment show that an optimum fiber size exists for blocking radiation. The fact that the theoretical calculations for real refractive indices of 2.0 and 2.5 fit the experimental data better than those for refractive index 1.5 does not necessarily imply that the real refractive index of glass is higher than 1.5. The imaginary part of the refractive index of glass is not zero as assumed in the calculations. It seems possible that compensation for this erroneous assumption requires the use of a real refractive index which is larger than the actual one. This explanation is suggested by the data in Figures 22-25. The imaginary part of the refractive index decreases witih increasing radiation temperature. Thus a high value of the real refractive index would be required to fit the data for low temperatures. At high temperatures the data would be best described by a lower refractive index. Another check may be made upon the experimental results. In this case an estimate will be presented for the value of the absorption parameter P for Styrofoam. Using equation 39, one may compute P for a foamed insulation. The necessary data were taken from the work of McIntirc and Kennedy2 and are tabulated in Table III.

TABLE III. THE PEYSICAL PROPIES OF STYROFOAM Property: Bubble radius, R 0.050 in. Insulation density, p 1.8 lb/cu ft Polystyrene density, Ps 66.0 lb/cu ft Absorption coefficient, 0 3.73 mm-1 From these properties the value of P was found to be 4 in.. Referring to Figure 17, one observes that the experimentally determined values of P varied with source temperature between 2 and 8 in.'1. Since the above absorption coefficient was not identified with a specific wavelength region, it is impossible to make a more precise comparison.

GROUPING THE PARAMETERS INTO DIMENSIONLESS FORM It is possible to present the data for the fibrous insulations in a more compact form. Earlier work has shown that the parameters vary inversely with fiber diameter, for large fibers. Thus the product of the radiation parameter and the fiber size should be independent of fiber size for large fibers. For very small fibers this function vanishes. In the experimental section it was concluded that the density influences the parameters as p - P sPs - Pg' By dividing the radiation parameter for a given bulk density by the above density factor, a quantity is obtained which is independent of density. The product of fiber size and radiation parameter divided by the density factor furnishes a compact and sensitive function for correlating the data. This method of correlating the data is presented in Figures 26 and 27. Por fibers. larger than 5 microns the points fall on approximately horizontal lines. The deviation of individual points from the, lines is of the order of 10 to 20 percent which is probably the amount of experimental error. The portion of the curves displaying deviation from the large fiber relation, designates that area in which the results of geometric optics no longer describe the radiation process. From these plots one may predict the radiation parameters for glass fibers of size and densities other than those examined in this work, Since the curves presented in Figures 26 and 27 have been drawn averaging the data, they are probably more accurate representations of the parameters than those presented in Figures 20 and 21. -82

6 0 2000 F source 4 V 300 0 F_ c3 ________ 400 OF 0 2 _____ ____ 600 0F_ _ _ 6800 F x __________ 8000F 0 10 20 30 40 FIBER DIAMETER, MICRONS Figure 26. The Dimenaicz ars Interception Paramnter for Glass Fibers

200OF source.03 4000F 600 0 800 OF.01 Q. 220 30 40 FIBER DIAMETER) MICRONS

A simlilar correlation maiht be made based upon the small fiber relation. However, the quality of the insulation samples which were made of small fibers does not justify their use as a means of extending the data.

PART IV ESTIMATION OF TE RADIANT HEAT TRANSFER RATES

aTH METHOD As a result of the work presented in previous sections, it is now possible to predict the rate of thermal radiation through several porous insulations. Computation of this rate requires numerical values for the radiation parameters M and N, insulation thickness, boundary temperatures, and boundary emissivities. The radiation rate may be estimated from either equation 24 or one of its simpler forms. Examples will be presented to illustrate the influence of temperature, emissivity, fiber size, and density upon the radiation rate. -87

EFFECTS OF TEMPERATURE In general the rate of energy transport by thermal radiation is greatly influenced by temperature. If one considers thick absorbing insulations, such that equation -27-a is valid, then the rate of radiation varies as the cube of the absolute temperature, assuming the radiation constants do not vary with temperature. This result is in agreement with the earlier treatments of this subject which were mentioned in the introduction. However, the radiation parameters of most of the insulations which were examined decreased with temperature. Thus the rate of heat transfer by radiation may increase with temperature more rapidly than the cube of the absolute temperature. A quantitative demonstration of the influence of temperature is given in Figures 28-30. The rates have been expressed in terms of the radiant conductivity evaluated at the insulation boundary. These values were computed from equation 30.

-eg1.0 0.8 0.6 0.4 ~STYROFOA M 22 0 FOAMGLAS " 0.2 0.1 > 0.08 100 200 400 600 800 1000 TEMPERATURE OF FIGURE 28. THE RADIANT CONDUCTIVITY AT THE BOUNDARY FOR STYROFOAM AND FOAMGL As

1.0 0.8 0.6 0.4 FIBERGLAS A O FIBERGLAS B A 0.2 0.I 0.08 z o 0.06 Q: 0.04 / BOUNDARY EMISSIVITY E a 0.5 100 200 400 600 800 1000 TEMPERATURE OF FIGURE 29. THE RADIANT CONDUCTIVITY AT THE BOUNDARY FOR FI BERGLAS

o10.O BOUNDARY EMISSIVITY = 1.0 INSULATION BULK DENSITY= 0.5 Ib/cu.ft FIBER DIA.'35p. 0 20p. L —.: _, 2.5A/ m0, I. 0.17p2 ~. I I 7 I' -I 0 ~II 02 200 300 400 800 600 700 800 TEMPERATURE, OF FIGURE 30. THE RADIANT CONDUCTIVITY AT THE BOUNDARY FOR GLASS FIBERS

EFFECTS OF BOUNDARY EMISSIVITY The influence of the emissive properties of boundary surfaces on the heat transfer rate has been frequently overlooked. A rigorous definition of a heat transfer problem involving porous insulations requires a specification of the insulation boundary emissivities. In Figure 31 one may observe the effect of emissivity upon the radiant conductivity at the boundary. All such curves must pass through the origin and have a maximum value when the emissivity is one. The amount of curvature in such plots varies considerably for the various insulations. For example the date for Styrofoam exhibit considerably less curvature than does that for Fiberglas A. From Figure 32 one may observe the effects of boundary emissivity on the radiant conductivity throughout the insulation. In this example the emissivity of the boundary at one side of the insulation is assumed to be one. For one case the conductivity is computed assuming the emissivity at the other boundary is also one. Under these conditions the conductivity increases in proportion to the cube of the absolute temperature and thus increases with depth into the insulation. In the second instance the emissivity at the hot side was taken to be zero. Now the conductivity kr is reduced everywhere and passes through a maximum value in the insulation interior. -92

A8V(Nnos 31iL iL ALIALLOflONOCO iNVItV83 3HI NO A.LIAISSIWN3 AVNn08 s 0 1 331-3 3HI'12'3Wn9U AllAISSIIN3 ANVCNnO',1 Z'1 0'1 80O 9'0 t'o "0 0 -' /0 -I 9'3.-Io 00~ V SV19&.0

0.3 OT |SIDE EMISSIVITY 1.0 l~~~~~~~~~tO _$S I 0 o 0.1 | _ THICHNESS I in. --- F | ~~~HOT SIDE T E MP. 600 ~ F \ z | COLD I500 N F o | ~~~~COLD "EMISSIVITY 1.0\ 0 II. O 0.2 0.4 0.6 0.8 1.0 COT SIDE TEP 0 EPTH I NSULATION IN Z<g~F T COLD,, -^,* 500%Fk -o_* v_. & a ~~~~~COLD " EMISSIVITY 1.0 0 0.2 0.4 0.6 0.810 DEPTH INTO THE INSULATION,IN

EFFECTS OF FIBER SIZE By using the values of the radiation parameters for glass fibers of various sizes to compute the radiant conductivity from equation 30, one may observe the effect of fiber size upon the radiation rate. These calculations are presented in Figure 33. The minimum in the curve corresponds to the optimum fiber size for reducing radiation. This size is the samle as that which makes the parameter M a maximum. -95

-960.5 0.4 INSULATION BULK DENSITY 0.5 Ib./cu.ft. c0.3 U. 0 z0.2 0.1 0 10 20 30 40 FIBER DIAMETER, MICRONS FIGURE 33. THE EFFECT OF FIBER SIZE ON THE RADIANT CONDUCTIVITY AT THE BOUNDARY FOR GLASS FIBERS

EFFECTS OF BUIX DENSITY Figure 34 demonstrates the effects of insulation bulk density upon the radiant conductivity. This curve follows the general shape of an equilateral hyperbola. The curve terminates at a density of 0.07 lb/cu ft corresponding to insulation containing no fibers. At this point the radiation is that exchanged by two infinite plates. The effect of bulk density upon the radiation properties of foamed insulations is more complicated. If one considers the size of the bubbles to be fixed, then increasing the bulk density increases the absorption parameter P and decreases the scattering parameter N. The net effect of this change depends upon the relative importance of scattering and absorption. If absorption is more effective in retarding radiation, then increasing the bulk density will decrease the radiation rate. On the other hand, if scattering is more important, then increasing the bulk density will increase the radiation rate. -97

-98CONDITIONS LOW SIDE TEMPERATURE 2000 F l *l I | ] I IHIGH,,, 500O~E d- I~ 1 1 iEMISSIVITY AT THE BOUNDARYS 1.0 INSULATION THICKNESS 2.0 in. 2 I 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 BULK DENSITY, Ib./cu. ft. FIGURE 34 THE EFFECT OF BULK DENSITY ON THE RADIANT CONDUCTIVITY AT THE BOUNDARY FOR FIBERGLAS B

THE TEMPERATURE DISTRIBUTION If the values for the radiation rate across the insulation are known, then the temperature distribution may be computed from equation 32. From the example presented in Figure 35 it appears that the temperature varies almost linearly with distance. This computation was greatly sinplified by the fact that the Styrofoam had high absorbing properties and equation 53 could be utilized., c39

200 v 150 COMPUTED DISTRIBUTION CL WI 0 0 }~ 2~~CONDITIONS: STYROFOAM 22 BOUNDARY EMISSIVITY 1.0 THICKNESS 2.0 in. 0 I 2 DEPTH INTO THE INSULATION, in FIGURE 35 A COMPUTED TEMPERATURE DISTRIBUTION

THE IMPORTANCE OF RADIATION COMPARED WITH 0OTIE HEAT TRANSFER MECHANISMS Now that estimates have been made for the radiation rates it is desired to examine the relative importance of radiation. One method of expressing the radiant contribution is the percent of heat transferred by radiation. This quantity was assumed to be 100 times the ratio of k. to Kc + kr. Values for the radiant conductivity at the boundary were computed from equation 30. Experimentally measured thermal conductivities were obtained from several sources.7,918 Figures 36-38 present the results of this group of calculations. In all cases the percentage of heat transferred by radiation increases rapidly with temperature. With the exception of Styrofoam, the contribution from radiation is only a few percent at room temperatures. At high temperatures, radiation may account for about 20 percent of the heat transfer. For Styrofoam the contribution by radiation may be as high as 20 percent even at room temperature. It should be pointed out that the above figures are only estimates. The amount of data on the thermal conductivities of polous insulations is rather limited. Consequently many of the values for the thermal conductivities were obtained by extrapolation of low temperature data. The following table lists the thermal conductivity data utilized in the calculations presented in Figures 36-38. -101

-10240 CONDI TIONS I. THE RATES ARE EVALUATED AT THE BOUNDARY 2. THE BOUNDARY EMISSIVITY = 0.5 30 W I cc z 20~ ft-,STYRO FOAM 22AS FOAGLAS O 200 400 600 800 TEMPERATURE, ~F FIGURE 36. THE CONTRIBUTION OF RADIATION TO HEAT TRANSFER THROUGH STYROFOAM a FOAMGLAS

-10320 z 15 FIBERGLAS B 0 -. cr'L i I | V a\FIBERGLAS A IJ _5 | CONDITIONS A I. THE RATES ARE EVALUATED AT THE BOUNDARY 2. THE BOUNDARY EMISSIVITY = 0.5 0 200 400 600 TEMPERATURE, F FIGURE 37. THE CONTRIBUTION OF RADIATION TO HEAT TRANSFER THROUGH FIBERGLAS A 8 B

-10450 40 CONDITIONS I. THE RATES ARE EVALUATED AT THE BOUNDARY 2. THE BOUNDARY EMISSIVITY - 1.0 z 0 (1 0 30 0 co.4 i.- 20 / FIBERS $ 20 —- l -------—?- 5 Ib./cu.ft. /5, FIBERS _ 3.4 Ib./cu ft. Z1 a. 10 10 FIBERS 5 lb. /cu. ft. 0 200 400 600 800 TEMPERATURE FIG.38 THE CONTRIBUTION OF RADIATION TO HEAT TRANSFER THROUGH GLASS FIBERS

-105TABLE IV. THERMAL CONDUCTIVITY DATA Conductivity Insulation pOF Btu, l'. Reference 0 F bhr, sq ft OF Foamglas 100 0. 43 7 200 0. 50 300. 55 400 0.62 500 0.62 Styrofoam 22 50 0.26 18 100 0.295 150 0.335 Fiberglas A 100 0.24 9 200 0.52 300 0.42 400 0.58 Fiberglas B 100 0.51 9 200 0.43 300 0.595 4oo o.83 Unbonded 70 0.21 9 Glass Fibers 100 0.22 5 microns diameter 150 0.238 3.4 lb/cu ft 250 0.275 Unbonded 70 0. 216 Glass Fibers 100 0.23 10 microns diameter 150 0.255 5 lb/cu ft 250 0.307 Unbonded 70 0.223 Glass Fibers 100 0.238 20 microns diameter 150 0.263 5 lb/cu ft 250 0.512

COMPARISON OF RADIATION RATES Several methods have been employed by other investigators in an effort to predict the rate of radiation through porous insulations. An attempt will be made to check some of their values. By measuring the thermal conductivity of a highly evacuated fibrous insulation, one may evwua+te the contributions of fiber conduction and radiation to the heat transfer process. The term high vacuum here requires that the mean free path:'or collisions between gas molecules-be much greater thaon the distance between adjacent fibers. Under these conditions conduction and corxvecti on do not contribute significantly to the heat transfer process. The following table pre9 sents the results of such measurementsJ and affords a comparison with the result's of this study. TABLE V. RADIANT COiJDUCTIVITIES FOR UNBONDED GLASS FIBERS Fiber diameter 5 microns 10 microns Bulk density 3.4 lb/, u f't 5 lb/:u ft Temperature 16( ~F 16c~F Conductivity of the Btu in. Bt in. 0 07 Bt.,, 0_ evacuated insulati. n Ir, uiq it, F hr, q, RadiarL (:onJdut:tivity fr,)n thLis study, 0.0 tu rl. I) O(4 -Btl E 0. hr, q ft hr, sq t, F -106

-107Since the experimental values are about 10 timzes higher than the radiant conductivities, one concludes either the values obtained in this study are low or the fibers conduct heat at appreciable rates. McIntire and Kennedyl2 have devised a method for estimating the radiant conductivity for foamed insulations. Their method is as follows: 1. The void fraction is computed from the insulation bulk density and the density of the insulation components. 2. The thermal conductivities of the insulation components are multiplied by their respective void fractions and added. Presumably this defines the contribution by conduction. 3. A small value for convection effects is assumed. 4. The difference between the experimentally measured conductivity and the conductivities from conduction and convection is attributed to radiation. The following table presents the results of a similar analysis and compares it with the results of this study. In this case the agreement between the two sets of values is relatively good.

TABLE VI. RADIAdNT CONDUCTIVITIES FORi STYROFOAM AND FOAMGLAS Iziulat~ii~l: Styrcf rua Fomrl;. Temperature 70OF 400 ~F Void fraction 0.975 0.~4 Solid component Polystyrene Glass Gaseous component Air Carbon dioxide Btukin. Btu,__n. Solid conductivity 0.8 h q int 6.0 hry Sq ftl or hr, sq ft~ ~F Gas conductivity O.l1 " O.12 " Estimated conductivity 0.19 Ty o. )48 Experimental conductivity 0.27' 0.62 Estimated radiation and convection conductivity o.ob " o72 Radiant conductivi!;y from this study: e = 0.5 o.o48 0.038 6 0 1.0 o.o80 "0.050 I

RADIATION TMiROUGH LARGE BLACK FIBERS In several treatments of this subject the fibers have been assumed to be large in comparison with the wavelength of the radiation. To illustrate the deviation of small fibers front the usual optical laws, a calculation will be made of the radiant conductivity for black fibers, 5 microns in diameter. In this case the absorption coefficient Ka is unity and the parameter P is computed from equation 553. Assuming the fibers are as dense as glass and the insulation has a bulk density of 0.5 lb/cu ft, P was computed to be 17.6 in. 1. For a boundary emissivity of 1, equation 50 shows chat kr has a value of 0.112 Btu, in. The conductivity based upon the parameters taken from hr. sq ft, ~F transmission measurements was 0.056 Btu, in. Thus the glass hr, sq ft, ~F fibers appear to be three times more effective in retarding radiation than what one would estimate upon the basis of the usual laws for black fibers. As mentioned in previous sections these effects vary with fiber diameter. If one considered the 0.2 micron diameter fibers, the fibers would obstruct less radiation than the large black fiber relation predicts. -109

PART V SUMMARY

SUmmARY A two flux model has been utilized to describe the process of radiation through a slab of porous insulation. Applying this model to an analysis of the change in the radiant fluxes over an infinitesiml volume of insulation yields a pair of simultaneous non-linear differential equations. By assuming that the steady state temperatures of the insulation vary linearly with insulation thickness, one may obtain an approximate solution to the differential equations. A complete description of the heat transfer process requires a consideration of the effects of conduction and convection as well as radiation. It was suggested that the conduction and convection rates might be grouped and expressed in terms of a gross thermal conductivity and the temperature gradient. At steady state the rate of heat transfer is a constant. From this relation and the approximate solution to the two flux equations, one may estimate the steady state temperature distribution within the insulation. Such a distribution was computed for a particular case, and the results indicated that the temperature varied almost linearly with insulation depth. It was demonstrated that the transmission properties of an insulation could be expressed in terms of two parameters. These parameters are the absorption and scattering cross section of a unit volume of insulation. The evaluation of these properties was the purpose of an experimental study. The experimental data consisted of a set of measurements of the transmission of black body radiation through varying thicknesses of the insulations. By fitting the two flux transmission equation to this data, it was possible to evaluate the parameters. -li

-112In this manner the radiation properties of eleven insulations were determined for temperatures between 200~F and 800~F. An attempt was:nade to estimate the radiation parameters from electromagnetic theory. Because the amount of numerical work required was excessive, and the values of the refractive indices for glass have not been established in the infrared, no information of quantitative value was obtained by this method. Equations and data have been presented for estimating the radiation properties of glass fibers of sizes and bulk densities other than those examined in this study. From the approximate two flux solutions it is possible to estimate the rate of radiant heat transfer through a porous insulation from the following information: 1. The radiation parameters for the insulation 2. The temperatures at the insulation boundaries 3. The emissivities of the surfaces bounding the insulation 4. The insulation thickness.

PART VI CONCLUSIONS

CONCLUSIONS On the basis of the experimental data and the theoretical work it is possible to make some generalizations regarding the effects of several variables on radiation through porous insulations. 1. If the radiation parameters are independent of temperature, then the rate of thermal radiation varies approximately as the cube of the absolute temperature. In most cases the parameters vary with temperature such that radiation increases with temperature more rapidly than the cube of the absolute temperature. 2. The fiber or pore size of a porous insulation has a great effect upon the radiation properties of the insulation. It was demonstrated that an optimum fiber diameter exists for obstructing radiation by a given mass of fibrous insulation. This diameter was found to range between 2 and 5 microns. It is certain that a similar behavior is exhibited by foamed insulations. Although no data were taken to establish the optimum size of foam bubbles, this size is probably of the same order of magnitude as that for fibers. 3. Assuming that a fibrous insulation may be considered to be an infinite matrix of parallel fibers, one may predict the influence of bulk density upon the radiation rate. Under these conditions the radiation rate varies inversely in proportion to the difference between the bulk density and the density of the gas in the insulation. Thus for large bulk densities the rate varies inversely with bulk density. The case of foamed insulations is more complicated. Increasing bulk density may either increase or decrease radiation depending upon whether scattering or absorption is more important in retarding radiation. -114

4. Reducing the emiss$vfity of the surfaces bounding the slab of insulation reduces the radiation rate within the insulation. Naturally these effects will be greatest at the boundary and diminish at large distances from the insulation boundary.

PART VII RECOMMEIDATIONS FOR IMPROVING INSULATIONS

RElCOmOEINDATIONS The results of this study suggest several methods for possible improvement of porous iasulations. Although no investigation has been made of the effects of glass composition upon the radiation parameters, it seems possible that improvements could be realized by altering glass composition. The work of McIntire and Kennedy12 has shown that materials may be added to Styrofoam to make it more opaque in the infrared. Such a chaucb cuuld improve its insulation properties considerably. It has been shown that a fiber or pore size exists for minimum radiant heat transfer through a given mass of insulation. Since processes other than radiation contribute to heat transfer through porous insulations, it is very doubtful that the optimum fiber or pore size for radiation would correspond to the optimum size for the sum of the three heat transfer processes. Collision diameters for gaseous diffusion are quite small compared to the optinum fiber or pore size with respect to the radiation process. Thus reduction in fiber or pore size below the optimum for- ~radiation would probably reduce the rate of heat transfer by conduction. On this babib it seems likely that the optimum size for the gross heat transfer would be less than that for radiation. The use of materials of low emissivity at the boundaries of an insulation to reduce the radiation rate has become relatively common. These boundary materials may also retard the diffusion of mass through the insulations. -117

AAPPENDICES

APPENDICES

APPENDIX A, EQUIPMENT CAIBRMATIONS The Amplifier The detector signals were amplified by a high-gain ten-cycle amplifier manufactured by Electro-Mechanical Research, Inc. Although this model, 33A, was intended for high impedance sources, it could be used for amplifying the detector signals if one added an input transformer. The transformer was a Thermador type HG-3 and was connected to the first stage of amplification. The maximum gain was 38 million. By measuring the amplifier output for signals of various frequencies, one may determine the frequency response of the amplifier. The resuIts of such a test are shown in Figure A-1. The Detector Probably the most important piece of equipment was the thermocouple radiation detector. Before it could be used, it was necessary to demonstrate that it was linear and estimate its angular response. The signal voltage from a linear detector varies directly in proportion to the amount bf radiant power impinging upon the detector. If such a detector were placed in the radiation field of a point source, its signal would vary inversely with the square of the source distance. Writing this algebraically e a (A-l) where e is the detector signal, r is the source distance, and d is a proportionality constant. By solving this equation for the radius, one obtains r -d/e (A-2) -119

-120I0 9 w 8 0 7 ii6 5Y,.J 3", 0 2 4 6 8 10 12 14 16 18 20 FREQUENCY, cps Figure A-1. The Frequency iOfspoMnse for the Modified Ten Cycle Amplifier

To test this relation, a set of data were taken of the detector signal at various distances from an auto headlight bulb. From the plot of this data, Figure A-2, it is evident that the relation does hold. Thus the detector is linear over the range of energy fluxes employed. Because much of the energy from the insulation which strikes the detector is incident at high angles, it is desirable to know the angular response of the detector. The angle of incidence of a beam is the angle the beam makes with the normal to the receiver face. Figure A-3 illustrates the position of the receiver and insulation. The angular response function, J(Q), is the ratio of the detector signal for energy incident at any angle G to the signal for energy incident normal to the detector. j(a) (A-3) e(o) Experimentally measured values of J(9) are plotted in Figure A-4. If the receiver responds uniformly to energy in any direction, then the response function will be the cosine. Such a receiver may be termed a cosine or two pi receiver. From Figure A-4, one notes that at low anles the response is slightly higher than the cosine, while at high angles it is too low. The low results are probably due to a shielding of the receiver face by the mount and reflection losses from the KBr window. Since the detector is used to measure both parallel and diffuse radiation, some error could be introduced by this low response to energy at high angles of incidence. The manner in which the detector response function influences the transmission measurements will now be illustrated. In the first

RECIPROCAL OF THE SQUARE ROOT OF THE DETECTOR SIGNAL 0 _( P US.0 o 0 C) C I0 rnl.~ ~~,. m ____ ___ -(, 00 r- of U1 rn -4 m - m -44i t 0 FI~~~~: — I C.. ]> C; b,~~~~~~~~~~~

23L3INSULATION _.... I DETECTOR Figue A3 An3 Illtrusttion of the Detector Position

-1241.0 Experimental data.9.8 Co.7 Fz.6 U,) -a3 o — A cosine receiver.2 00 20 40 60 80 90 ANGLE OF INCIDENCE OF THE RADIATION, DEGREES Figure A 4 The Angular Response of the Detector

case assume the flux impinging upon the receiver is normal to its face and produces a signal of unity: e(O) = 1 (A-4) This is the case for measuring the incident radiation. Now assume that this flux falls on the insulation and a fraction, t, is transported to the other side. By this definition t is the fraction transmission. If this flux struck the receiver normal to its surface then the signal would be t, the fraction transmission. Unfortunately, the energy is not collimated so one must know the angular response of the detector. In the extreme case, this energy would be diffuse, that is, its inten, sity follows the coaine law. The detector signal would be e t - j() cos e da (A-5) where a is the solid angle subtended by the insulation about the receiver. In terms of the coordinates in Figure A-3 e t J(e6) cos (2x)x (A-6) e - ~ (x2 + ~2) (A-6) (2 + 52) and cos (2 82)1/2 then e ot (e) 2d x e (x2 +,2) 3/2

-126In the case where j(G) is the cosine, then 2 2x dx e = t 6 > 22=t (A-8) 0 (x + 62)2 Thus the cosine detector still indicates the fraction transmission. In the case for the actual detector, the integral A-8 is evaluated graphically using the data of Figure A-4. The answer is e = 0.84 t Thus, in the extreme case, the detector would give transmissions which were 16 percent low. However, actual comparison of transmissions measured by a cosine detector and the one mentioned above show there is no appreciable error introduced by thc highly scattered energy. The Source Although the source aparac~rd to be black, it was desired to have a more rigorous check. For this purpose a series of measurements were taken of the detector signal for various source temperatures. The Stefan-Boltznaann L rea d;r&ictc that the signal would vawry as e b (T4 _T ) (A-9) where b is a proportionality constant, T is the source temperature, and To is the detector temperature. Since the detector temperature was e.1.~.1 5 3:S~:i~. C,~.-,,, iTR',:i.....'~:~.:. A'-':'i.:'k X'%v:Wary in proportion,So tite i7ourth;*,_,>.':S'1t Seas'<.f>'r,,i,sti" 4>> En;::'"of, li.:ate. t haut`he data.........v,,+ff,,. i:';:':',:ace ad deteti,) are::;;sentially tblack,

awI t au= 1o-ula 94S a t4 q qrl;u3aS aaanog etq jo 4sal V 5 V ay C -J o01 x'to'3nlvu3dW3l 3Dounos 3H iO 83MOd H.lUIOJ 091 Otl OZI 001 08 09 Ob0 O0 6 /11 1 II

APPENDIX B. S. rPI3 C; LCULATIONS A set of experimental measurements were taken from which the transmission of radiant energy could be computed. The method of treating this data is outlined in Part III. By plotting the transmission against insulation thickness, and drawing a smooth curve through the points, one obtains a transmission curve for the insulation. Assuming that the two flux theory presented in Part I is correct, then equation 57 must be the smooth curve passing through the points. One may compute the constants M and N which fit equation 57 to the data. Since the data scatter, the constants were evaluated by making a least squares fit to equation 57. The series expansion of this equation is xo) 2 M2 - eif~~F~ - (-l)g [M2 j n KN 2+M n J 2_ + M e (58) If the insulation is'absorbing and very thick, then the series portion of equation 58 approaches a value of unity. In this region the traisiaission varies expn.entially with thickness. Thus a semi-log plot of transmission against thickness would be a straight line for large thicknesses. A survey of the data on Figures 10-15 shows that this is frequently the case. The following set of data will be employed to illustrates the calculation of the constants M and N. -128

-129TABLE B-1. TRANSMISSIQNS TEROUGH GLASS FIBERS Insulation: 35 pi diameter unbonded glass fibers aource Temperature: 800~F Thickness, in. 0 0.60 0.71 1.39 3.22 Transmission 1.00 0.311 0.272 0.171 0.056 Figure B-1 illustrates the data. The first step in the calculation procedure is to fit the data to a straight line. This line is shown on Figure B-1. It was drawn below the points at high transmissions in an effort to estimate the straight line region of the transmission curve. By equation 59, the intercept of this line is 0.385 The slope of the line is f T 2 I3 113 log 385 =0.60 in. 1.0635 Solving for the approximate constant M = 2.52 in.1 N = 2.44 =N o.60 Now the value of the series in equation 58 may be estimated. Define f to be the series

-1301I.0 Actual transmission curve C,) (/) C,) z First approximation 0.1 LL 0.53 Ib/cu ft bulk density 35/pdia glass fibers 800oF source.02 -... 0 I 2 3 INSULATION THICKNESS. IN. 1'i' 1 rux e IdI'LL.Ilo 1 Oll D tDn

-131f~= Z(-1)[ M-e2JnMe N2 (B-1) n=O Ni + P The quantity in the brackets is J.... -M.60 - 2.52 0.615 For2 the +M 0.60 f 2.52 For the 0.60 in. thick specimen, the value of f is estimated as follows: f a 1 + 0.615 e''120(0'60) + 0.6152 e-2.4(.60) + = 1.390 Following this procedure, one obtains the following values for f. TABE B-2. VALUES OF f FOR 35 MICRON GLASS FIBERS Thickness, in. 0.60 0.71 1.39 3.22 f 1.39 1.53 1.13 1.01 By using the definition of f and transforming equation 5, one may obtain log EL log'-| N (B-2) The relation B-2 states that a plot of log T(f against insulation thickness would be a straight line. From the slope E.nd intercept of this line one may compute the radiation constants. Assuming the experimental data deviate in a random fashion from this relation, the method of least squares will give the best straight line.

-132If p and q are the slope and intercept of this line, then equation B-2 requires that p= — M2=-_N (B-3), lo~g 2/M2: - N2 (B-4) 1M2 _ N2 + M The log functions employed above are those to the base e, 2.71828' The method of least squares requires that p and q be calculated from the relations log nq + p n (B-5) ja1 ~ j-1 n n n ~ log q fj + p j (B-6) j=l j=l J=1 where the index j is used to designate each measurement. The summations extend from the first measurement, j = 1, to the last, j = n. By summing the data in Tables B-1 and B-2 according to equations B-5 and B-6, one obtains -7.88 = 4 q + 5.92 p -13.995 = 5.92 q + 13.165 p Solving for p and q -1 p = 0-0.53 in. q = -1.185 From these values the radiation parameters are found to be

-133M = 2.92 in. N = 2.87 in.'l 2 - 0.53 in.'1 The above values represent the second order approximations for the radiation parameters. Higher order approximations may now be attempted using the above values to compute f. The results for the third order approximation are M = 3.00 in.'i N - 2.95 in.'1 J, - 2- 0.53 in. l Using these values in equation 58, one may construct a curve through the data, see Figure B-1.

APPENDIX C. TEE SCATTERING OF ELECTROMAGNETIC RADIATION BY AN ISOLATED, IINFINITE CYLINDER The cylinder is illustrated in Figure C-1. Both the rectangular coordinates x, y, and z, and the cylindrical coordinates r, z, and g will be employed. It will be assumed that the cylinder axis coincides with the z axis and that the cylinder has a radius a, and a real refractive index m. The solution of the problem requires solving a set of partial differential equations and fitting the solution to certain boundary conditions. The equations are commonly called the Maxwell Equations and may be expressed as curl E 1 (C-l) c t curl H 4 E + 1 D (C-) c c( div D = 4gp (C-3) div H = (C-4i) where E is the electric field strength vector, H is the magnetic field strength vector, D is the electric displacement vector, c is the velocity of light, t is tinue, a is the electrical conductivity, and p is the electric charge density. -134

*135z Ey r Hz - Sx 8 sx_ PERPENDICULARLY POLARIZED INCIDENT RADIATION VECTOR a CYLINDER RADIUS a Ez.......S-.. Sx PARALLEL POLARIZED INCIDENT,_ o RADIATION VECTOR igunr C4l. E4tito Ipi~gig Upo a Ia Xnfite Cylinder

-136Parallel Polarization In order to facilitate solution, the problem is divided into two parts. The first part consists of obtaining a solution for parallel polarized radiation. The incident ray is traveling in the x direction and is described by the vectors Ez and Hy. Assuming these vectors may be described by the usual wave relations E eip(tx) (C-5) where p =, c is the velocity of light, and t is time. For the case in which the magnetic permeability is unity, equation C-1 gives Hy as eip(ct-x) (C-6) The radiant flux is given by the Poynting Vector S S = c Ex H (c-7) 4n The timne average radiant flux in the x direction is Sx = c Ez Hy = c (c-8) where Hy is the complex conjugate of Hy. Now that the incident flux is known, one next wishes to compute the scattered field. This requires the expressing of the incident wave in cylindrical coordinates, E - eip(ct-x) eip('t-rcosg) (c-9)

-137Since the above function is continuous and periodic in 9, it may be expanded in a Fourier Series E = e Pt an cos n + bn sin nG (C-10) (10 n-O The function is even, therefore its Fourier expansion will contain no odd functions, so bn is zero. The value of ao is given by 0 22 a - 1_ f e-iprcosQ d Jo (pr) (C-ll) 0 where Jo is the Bessel function of the first kind, order zero. For n greater than zero, an may be expressed as 2v =_iprcosg n anm f cos nQ e ip c0 de = 2 i Jn(Pr) (C-12) an i 0 Substituting the values for the coefficients into C-10 gives Ez = eict [J(Pr) + 2 in Jn(pr) cos n@ (C-13) n=l It will now be presumed that the solution for the region outside the cylinder consists of two parts. One part is the original unscattered wave, the other part is the scattered wave. The scattered energy must travel radially from the cylinder and the only cylindrical function which will produce such a wave is the Hankel function n(X) - Jn(x) + iNn(x) (C-14) where Nn is the second kind of Bessel function of order n.

-138Then the solution for the region outside the cylinder may be expressed by Ez eipct {Jo(pr) + AHo(pr) + 2 in cos nG [Jn(pr) + AnHn(pr)]} n=l (C-l5) The Hankel function has a logarithmic singularity at the origin. The region inside the cylinder obviously must contain the origin, hence the solution for this region can contain no Hankel functions. Then the solution inside the cylinder will be a series of Bessel functions with the argument modified by the refractive index m. Ez = e ipct [BJ(mpr) + 2jin cos nG BnJn(mpr)] (C-16) n=l Solutions C-15 and C-16 must satisfy boundary conditions at the surface of the cylinder. These conditions are that Ez and He must be continuous at the cylinder surface. Satisfying this condition requires that Jn(a) + AnHI,(a) = BnJn(mC) (C-17) Jn' () + AnHn'(cr) = mBnJn'(U) (c-8s) where C = 2-ia and the primed notation denotes differentiation. Solving these relations for An gives m J(a) Jl'(rrx) - J,(n(') Jn'(c) An ('( - -(:.(. (c-19)

-139The next step in the problem consists of finding the function He. From equation C-1 1 He aE c 3tH ar (C-la) Then the magnetic field strength of the scattered wave is He - -ieipct [AoHo'(pr) + 2 in AnHn'(pr) cos nG (C-20) n=l The scattered power is given by the time average of the complex Poynting vector C (E,) (c-a) rs 8 ZHe s where the subscript s indicates that only the scattered fields are being considered. The integral of rSrs over the range of 9 from 0 to 2n gives the total scattered energy from the cylinder of unit length 2n r 8rs d9 = d [ IAo2 H o' + 2 IAnI2 H nn'] (C-22) 0 n-l For very large arguments, Then,, frrs d = 2p [ Ao + 2 IAnl ] (C-24) O n=l

-140The scattered energy at distance r is given as frSrs de =- J[ AQ + I ZAn2 I (C-25) - nn=l The scattering cross section is the energy scattered per unit flux in the incident beam. From. equation C-25 and equation C-8 S 2X I [AoI + 2 lA |I (C-26) n=l The evaluation of the forward scattering fraction, F, is arbitrary. The method which will be used is 3/2 JSr d F - - (c-27) JSr d@ 0 I II 2 2 4 (-1)n+lRe(AoA2n-1) JA)I + 2 IA 4 (.1"i) ~ in ( - ) n (n + -I n..l~n"k n+k

-141Perondicular Poarization The problem is now half solved; the other type of polarization has to be treated. Fortunately, the results are very similar. The scattering cross section is 8 2XE IPO~ + 2 P 2 (C-29) nal and the forward scattering fraction is F1. F,, (P) (C-3o) The values of Pn are fixed by boundary conditions similar to those of equations 17 and 18. m Jn(mnx) Jn'(a) -J (a) J'(mC) p.......... (c151)' n = m Jn(mo) n' (d)'- l(a) Jn'(mo) (-) Since thermal radiation is likely to be randomly polarized, the averages of the values Ss and F for each type of polarization were employed. Sma~l Fiber Approximtion If the fiber diameter is much smaller than the wavelength of the energy, the calculations are greatly simplified. Using the series approximation for the Bessel function aa2 Jo(ar) = 1 - ) (C-32) gives mJO(a} - Jo(~~) Jo'(a) = 24 (2 -l). (C-35)

-142The expressions for the Hankel functions are HO(a) _ 2i log 2 (C-34) Ho t(aCZ) = A (C-35) to0 =_ai Then Jo(m~) Ho'(a) - m Jo'(r) 1Ho(a) = 2 (C-36) From equations C-36 and C-33, one obtains A0 as 2 t24 (c2_7) IAo 1 J a16..-l (C-37) By the present order of approximation all other terms in the series are zero. Thus xA's 4 (m21 (c-38) Equation C-28 gives the value for the scattering fraction, F F = 1/2 (0-39) Absorbing Cylinders In the previous work it has been assumed that the cylinder was a dielectric, that is, the refractive index was real. This assumption may not be valid for many materials, particularly for glass in the infrared where absorbing properties are displuyed. The complex refractive index m* m* a m - 1k (C- 4O) it used to describe t;hL;e nmater ill.. The r,'al I:Ilrt of mn* i genr:riLly termed the refractive index. The imoaglrnary part, k, ia vrellteld t the exponential absorption coefficient, $, by

-1432nk (C-41) Fortunately the results of the scattering functions from equations C-26, C-28, C-29, and C-30 are still valid if one uses m* to compute the values of An and Pn from equations C-19 and C-31. Introducing the absorption property requires calculation of another quantity, the absorption cross section 2a 1/2 Srrd] s~. ~i 0 - (C-42) For parallel polarization Z, ip(ct- r cos e) + [AOHO(pr) + 2Zin AHn(pr) cos z] eipct n=l (c-43) H -cos G e i [(AoHo(pr) + 2i AH'(pr) cos ne] Zn' n=i eipct (C-44) Sr 4n z E c-7) Performing the integration in equation C-42, Sa - 2 [(Re(Ao) + 1A01 + 21ReAn + 1~12 ] (C-45) n=l

-144Considering the other rlode of polarization gives ~~~~2X 2 ~~T-'I0 S 3 2k [RePo + I PoI + 2RePn + I Pn (C-46) n=l For most cases the absorption cross section is the mean of that for perpendicular anld parallel polarizations. Smell Absorbing Cylinders Using the first terms in the series expansions of the Bessel function leads to ReA)24 2 2 (C-47) e(Ao) = 16 [(m2 - k2 -1) + k (1 + 2 log -)] (C-47) Im(AO) = a m2 - k2 1) (C-48) PO e O (c-49) Thus, for randomly polarized light, Sa - i C' a mk(l + 2 log 2) (C-50) 16 ya where 7 = 1.781. From this work, the absorbing parameter P will vary approximately as the cube of the diameter of the cylinder.

APPENDIX D. CYLINDRICAL SCATTERING FUNCTIONS The scattering functions An and Pn were computed from equations C-19 and C-31 for a few cases. In each computation the refractive index was assumed to be real. The following tables list the values obtained. -145

TABLE D-1. SCATTERING FUNCTIONS FOR m = 1.5 n Re An Im An Re Pn Im Pn a = o.8 o -0.2977 0.4572 -0.0029 0.0535 1 -0.0029 0.0535 -0.0302 0.1711 2 -0.0000 0.0014 -0.0002 0.0146 a = 1.0 0 -0.4334 0.4957 -0.0185 0.1348 1 -0.0185 0.1346 -0.0595 o.2365 2 -0.0000 0.0050 -0.0011 0.0344 3 -0.0000 0.0001 -0.0000 0.0015 a = 1.5 0 -0.5362 o0.4986 -0.3966 0,.4892 1 -0.3965. 4893 -0.1572 0.3641 2 -0.0033 0.0580 -0.0242 0.1537 5 -0.0000 0.0027 -0.0001 0.0147 a = 2.0 0 -0.5644 0.4957 -o.8399 o.56656 1 -o.8399 0.3665 -0.3528 0.4777 2 -0.1292 -0.3356 -0.1462 0.3533 3 -0.0172 - 0.1299 -0.01)6 -0.1413 4 -0.0000 0.0011 -0.0000 0.0061

-147TABLE D-1 (CONT'D) n Re An Im An Re Pn Im Pn a = 3.0 0 -0. 0176 -0.1315 -0.9133 0.2824 1 -0.9128 0.2825 -0.9992 -0.0258 2 -0.9959 0.0620 -0.7766 o.4176 3 -0.4882 0.4993 -0.2948 0.4562 4 -0.0035 0.0593 -0.0136 0.1159 5 -0.0000 o.o0039 -0.0002 0.0131 6 -0.0000 0.0000 -0.0000 0.0011 a = 4.0 0 -0.9165 -0.2780 -o.6831 -0.4664 1 -0.6831 -o. 4664 -0. 9715 -0.1699 2 -0.9984 -o.o0418 -o0.8668 -0.33591 3 -0.9587 -0.2004 -0.9988 -0.0447 4 -0.9685 0.1746 -0.5482 0.4975 5 -0.0151 0.1218 -0.0295 0.1693 6 -0.0001 0.0094 -0.0005 0.0223 7 -0.0000 o.o00oo6 -0.0000 0.0023

TABLE D-2. SCATTERING FUNCTIONS FOR m z 2.0 n Re An Im A Re Pn Im Pn cx = 0.5 0 -0.4383 -0.4962 -o0.00ooo4 -0.0206 1 -o.ooo4 -0.0206 -o.0142 -0o.118 cx = 1.0 0 -0.8892 0.3139 -0.2708 0.4449 1 -0.2703 -0.4441 -0.154 o.5361 2 -0.0000 0.0002 -0.003 0.056 cX = 1.5 0 -o.9128 0.2821 -0.9678 -0.1787 1 -0.9678 -0.1787 -0.5187 o.4998 2 -o0.065 0.2465 -0.0774 0. 2b72 3 -0.000. 0092 -0.000 0. 000 c _ 2.0 0 -o.992 0o.96 -0.989 -o0.10 1 -O.933 -0.250 -0.797 -0.403 2 -0.823 -0.588 -0.567.495 3 -oo0,014 -0.117 -0.021 -0.144 4 -0.000 0.004 -0.000 0.010

-149TABLE D-2 (CONT'D) ln Re An Im An Re Pn Im Pn a = 3.0 0 -0.115 -0.319 -0.022 0o.146 1 -0.022 0.146 -0.502 -0.495 2 -0.553 -0.497 -0.117 -0.322 3 -0.351 -0.477 -0.765 -0.427 4 -0.954 0.209 -0o.o81 0.273 5 -0.000 0.016 -0.000 0.023 a =4.0 0 -o.841 0.367 -0.201 0.401 1 -0.201. 401 -0.698 o.459 2 -0.683 0.493 -0.095 0.293 3 -0.c89 0.105 -0. 346 0.475 4 -0.321 -o.467 -o.607 -o. 5 -0.092 0.267 -0.575 0.1)6 6 -0.00oo8 0.0o89 -0.00 o0.0o56

-150TABLE D-3. SCATTERING FUNCTIONS FOR m = 2.5 n Re An Im An Re Pn Im Pn a = 0.5 o -0.888 -o.316 -0.002 -0.041 1 -0.002 0.041 -0.022 -0.145 2 -0.000 0.001 -0.000 0.026 a 1.0 0 -0.996 0.074 -0.990 -0.115 1 -0o. 990 0. 01l -0.250 O. 434 2 -0.001 0.051 -o0.00 0.072 a = 1.5 0 -0.806 -o.403 -0. 675 -o. 469. -o.675 -o.469 -o.816 -0.589 2 -0.856 0.372 -0.162 o.369 5 -0.000 0.018d -0.001 0.025 a = 2.0 0 -0.0 53 0.179 -o. 420 -0.495 1 -0.42o -0.495 -0.058 -o.2j5 2 o.277. -0.448 -0. 480 -o. 4>9 3 -0.000 -0.005 -0.011 0. 105

-151TABLE D-3 (CONT'D) n Re An Im An Re Pn Im Pn (X= 3.0 0 -0.951 -0.218 -0.714 0.453 1 -0.714 0.453 -0. 998 -o.o - 2 -0.993 0.0o86 -0.122 -0.5 7 3 -0.037 -0.188 -0.054 0.227 4 -0.180 -0.133 -0.009 -0.096 5 -0.055 -0.228 -o.oo~ o.059 6 -0.000 0.002 -0.000 0.002 = =4.0 0 -0.276 -0. 449 -o. —. 429 1 -0.054 -0.225 -0.307 -0. 500 2 -0.482 -0.504 -0.410 -0.491 3 -0.795 -0.404 -0. 50o8 -0. 501 4 -0.596 -0.491 -o.616 0.487 5 -0.000 0.025 -0.020 0.140 6 -0.001 -0.051 -0.000 0.025 7 -0.000 -0.024 -0.000 -0.007

APPENDIX E. EXPRIMEiTTAL 7TL2ASMISSION DATA AND CAIXI3IATED RADIATION PARAMETERS The experimentally measurec values of the transmissicon through various thickneez of the insulztiors c.re tabulated in the following. TABLE E-1. JRANSMISSIONS THROUGH STYROFOAM Source Temperature, OF 110 145 170 285 Insulation Thickness, in. 0.12. 0.295 0.295 11. 0.362 0.0245 0.0298 0.0327 o0.038 0.482 o.oo09 0.0121 0.0119 0.0149 0.724 0.00074 0.0017 0.0020 0.0029 o.844 -- --- 0.oo014 TABLE E-2. TRA1NSMISSIONS THROUGH FOAMGLAS Source Temperature, ~F 400 468 580 860 Insulation Thicknless, in. 0.106 0.0302 0.0375 o.o40 0.052 0.118 0.01)3 0.0273 0.0296 0.038 0.224 0.00126 0.00192 0.0020 0.0028 0. 342 o.oooo4. 0.00010 u. 0001 U.00u17 -152

-153TABiLE E- 3. TRANSMISSIONS THROUGH FIBERGLAS A... -- - --.-:u --- -'4.'...!'-" j' ~ "' -:.- --'- " -" -..,,-,-'- "'-k:.'......-...'.:~..,.. Source Temperature, OF 210 285 468 580 Insulation Thickness, in. 0.233 0.0385 0.0494 0.0674 0.0724 0.326 0.0139 0.0184 0.0262 0.0324 0.500 0.00283 0.o038 O.0061 o.oo0066 0.733 0.00041 0.00073 0.00o114 0.0017 0.966....0.00026 AB E-4. TRANSMISSIONS THROUGH FIBERGLAS B Source Temperature, OF 210 285 400 580 Insulation Thickness, in. 0.408 0.0289 0.0377 o.o46 0.064 0.520 0.0117 0.0177 0.028 0.032 0.592 0.0095 0.0112 0.127 0.021 0.928 0.00123 0.0023 0.0031 0.0060 1.00 0.00077 o.0315 0.0029 0o.o0041

-154TABLE E-jR. TRAi SlISINS TiRQUGh O.17 iECRO; DkLAkTi GLOLS FiBRS AT A DENSITY OF 0.5 LB/CU FT Source Temperature, ~F 200 306,;OO 6o00 oo Insulation Thickne ss, in. 0.0o66 0.304 0.34 0.405 0.45 o.48 0.104.25 O. 31 0.357 0.396 0.44 0.292 0.098 0.12 0.22 O.l1( 0. 0 O.452 0.057 0.070 0.12 u.14 0.13: 0.744 0.022 0.029 0.050 0.065 0.070 TABLE E-6. TRANSMISSIONS THROUGH 0.20 MICRON DIAMETER GLASS FIBERS AT A DEN4SITY OF 0.5 LB/CU FT Source Temperature, OF 200 300 400 600 000 Insulation Thiclkess, in. 0.135 o.168 0.184 0.25 0.275 0.i>Y 0.03 0.120 0.1360 0.20 0. 3 0.31 0.210 0.090 013 0.200 0.23 0. 2 0.345 0.056 0.0o 0.105 0.11 C.14 0.548 0.025 0.032 (0.030oo,')4.0 o86

-155TABLE E-7. TANSMISSIONS THROUGH 2.5 MICRON DIAMEbT GLASS FIBERS AT A DENSITY OF 0.5 LB/CU FT Source Temperature, OF 200 300 400 600 800 Insulation Thickness, in. 0.254 0.065 0.072 0.083 0.095 0.11 0.316 0.0305 0.050 0.054 0.063 0.073 0.570 0.0088 0.016 0.020 0.025 0.032 o0.916 0.0012 0.0029 o.oo4 0.0058 0.0078 TA&tD z-8. TRANS SIO NS THROUGH 5 MICRON DIATE GLASS FIBERS AT A DENSITY OF 0. 5 LB/CU FT Source Temperature, ~F 200 300 400 600 800 Insulation Thickness, in,. 0.32s 0.0345 0.0457 o0.o60 0.0822 0.119 0.357 0.0283 0.040 o0.0o48 0.074 0.095 0.678 0.00635 0.0,75 0.0123 0.021 0.039 1.00 0.00132 0.0017 0.0035 0.0061 0.011

-156TABLE E-9. TRA0SMISSIOINS TmOUGI 10'ICRON DIAMITER GLASS FIBERS AT A DENSITY OF 0.5 LB/CU FT Source Temperature, lF 200 300 400 600 o00 Insulation Thickness, in. 0.495 0.0383 u.057 0.063 0.097 0.145 o.68o 0.019 0.024 0.035 0.063 o.o86 0.820 0.0091 0.0154 0.020 0.046 0.070 1.13 0.0040 o0.00oo65 0.0090 0.022 0.036 TABLE E-10. TRANSMISSIONS ThROUGH 20 MICRON DIAMETER GLASS FIBERS AT A DENSITY OF 0.5 LB/CU iFL Source Termperature, OF 200 300 400 600 80o Insulation Thic knuts, in. 0.404 0.122 0.10 0 203 0.63 0.321 0.495 0.095 0.120 0.175 0.230 0.302 1.32 0, 027 0.03>' 0.061 a. 1'. 1.57 0.0,14 0.023 0.034 0.065 0. i 0

-157TABIB E-ll. TR.AISMSSIONS THROUGH 35 MICRON DUiTER GASS FIBE AT A DENSITY OF 0.5 LB/CU FP Source Temperature, O~F 200 300 400 600 800 Insulation Thickness, in. 0.655 0.097 0.115 0.157 0.251 0.311 0.77 0.083 0.107 0.14j 0.22 0.272 1.51 0.043 0.058 0.089 0.125 0.171 3.50 0.0074 0.0083 0.016 0.034 0.056................,,~, —,,,,,~,, - l,.- - ~_~, l, _- -— ~......... _.... The values of the radiation parameters M and P which were computed from the previous transmission data are tabulated in the following. TABIE E-12. RADIATION PARATES FOR YST R0F!OAk Temperature 1100F 145 ~F 170~F 285~F M, in.'l 10.1 13.7 13.9 15.1 P, in.1l 7.7 3.1 2.8 2.0 TABLE E-13. RADIATION PARAME TERS FOR FOAMGLAS Temperature 400~F 468QF 580 F 880~F M, in.l1 77.1 69.0 72.0 53.0 P, in7-1 4.8 4.6 4..3 --— ~~~~~~~~~~~~ 6';; =I — --— " —-~I- ------ ~ -

-158TABLE E-14. RADIATION PARA!"TERS FOR FIBERGLAS A Teimperature 210 285 F 2568 F 3 ~F M, inA-1 62.0 47.5 33.1 32.9 P, in.-l 0.62 0.73 0.98 O.V, TABLE E-l5. RAD'i.TIOw P —RiiMETERS FOFR FrIBERGLAS n Teanperat';re 2100F O-I ~F 400 ~F 5o00 ~F M, in.a- 34.0 31.1 30.2 t. P, in.-1 0.52 0.i5. o.

-159TABLE E-1,6. RADIATIOrT PmRAaEMRS FOR GuLSS FIBERS AT A DENSITY OF 0.5 LB/CU FT Tti.tper. Iure: 200~F 30U0 F OO ~F 600 F 8, O~F O (.li!4i73'on Fibers M, in.'l 25.0 14.0 11.9 13.0 P, in.-l 0.17 o.164 0.15j6 0.158 0.159 0.20 Micron Fibers M, in.'l 34.2 28.6 19.0 16.6 14.0 P, in.1 0.23 0.212 0.273 0.262 0.233 2.5 Micron Fibers M, in.l 45.5 37.6 35.3 312 265 P, in.,i 0.357 o. 294 0.262 0.252 o. 264 5 Micron Fibers M, in.4l 52.6 38.8 31.6 23.7 18.9 P, in.1 0.25 0.263 o. 88 0.31 0.306 10 Mt[icron Fibers M, in.'1 28.5 2i. 3 19.7 1.4 10. P, in.' 0.22 0.204 0.187 0.174 0.180 20 Micron Fibers M, in.'l 13.2. 6j 7.4 4.84 P, in.1 o.114 0.12 0.112 0.09 o.og96 M, in. 4 7.3 5.5 3.53 4.64 P, in.-l 0.062 0.058 0.059 0.050 0.045

APPENDIX F. INSULATION FIBER SIZE DISTRIBUTIONS Of the unbonded fibrous insulations, two specimens contained fibers of a variety of sizes. The arithmetic mean diameter of each specimen was 0.17 and 0.20 microns. Tabulated in the following are distributions of fiber sizes obtained by the examination of a sample of 200 fibers from each insulation under an electron microscope.9 -160

-161TABLE F-1. THE FIBER SIZE DISTRIBUTION FOR GLASS FIBERS OF 0.17 MICRON MEAN DIAMETER Fiber Diameter, Hundred Thousandths Number of Fibers of an Inch 0.125 5 0.150 4 0.175 3 0.200 5 0.225 8 0.250 11 0.275 8 0.300 4 0.325 4 0.350 5 0.375 14 o.400 6 0.425 3 0.450 6 0.475 3 0.500 12 0.525 7 0.550 3 0.575 2 o.600 6 0.625 4 o.650 5 0.675 2 0.700 4 0.725 2 0.750 3 0.775 4 0.800 1 o.825 2 o.850 1 0.875 6 0.900 2 0.925 2 o. 950 1 0.975 2 1.o000 5 1.025 6 1.075 1 1.100 1 1.125 5 1.250 2 1.375 1 1.400 2

-102. TABLE F-1 (CONT'D) Fiber Diameter, Hundred Thousandths Number of Fibers of an Inch 1.425 1 1.45o 1 1.625 1 1.679 1 1.900 1 1.925 1 2.025 1 2.125 1 2.175 1 2.250 2 2.350 1 2.-375 1 3.250 1 3.975 1

-163TABLE F-2. TEE FIBER SIZE DISTRIBUT1ION FOR GLASS FIBERS OF 0.20 MI(CRON WMEAN DIANMEER Fiber Diameter, Hundred Thousandths Number of Fibers of an Inch 0.14 5 o.16 4 0.18 1 0.20 12 0.22 8 0.24 6 0.26 8 0.28 8 0. 30 10 0.32 9 0.34 5 0.36 10 0.38 8 0.40 6 0.42 6 0.44 5 o.46 o o.48 3 0.50 9 0.52 2 0.54 1 o.56 2 0.58 4 0.60 3 0.62 5 0.64 4 o.66 3 0.68 2 0.70 2 0.72 1 0.74 1 0.76 2 0.78 1 0.80 4 0.82 1 o.84 0 o.86 o 0.88 3 0.9o 1 o 092 o o.g4 2 0.9g6 o 0.98 2 1.00 1

-164TABLE F-2 (C'OiT'L) Fiber Diameter, Hundred Thousandths Number of Fibers of an Inch 1.06 1 1.08 1 1.10 1 1.26 1 1.28 1 1.48 1 1.50 1 1.j6 1 1.64 1 1.o0 1 1.90 2 2.20 1 2.30 1 2.74 1 2.84 1 2.86 1 2.94 1 3.50 1 3.66 1 3.90 1 4.oo00 2 4.20 1 3.46 1 4.68 1 4.72 1 5.04 1 5.06 1 5.16 1

-165NOMENCLATURE Latin Letters A The area for emission of a body, sq.in., An A scattering function, B The backward scattering component, b A proportionality constant, c The velocity of light in vacuum, D The electric displacement vector, d A proportionality constant, Df The fiber diameter, microns, E The electric field strength, e The detector response, millivolts, F Thne forward scattering component, f A series expansion of the radiation parameters M & N, g(X,T) Planck's radiation function, hr ft H The magnetic field strength, Hn The Hankel function of the first kind, Btu hv The convection coefficient, hr, ft, u' I1 The spectrally integrated flux vector in the negative x direction, hr.sq.ft' il The monochromatic flux vector in the negative x direction Btu hr.,sq.ft.,micron I2 The spectrally integrated flux vector in the positive x Btu direction, hr.sq.ftu. i2 The monochromatic flux vector in the positive x Btu dire tion, hr.,sq.ft.,micron

-166Jn The Bessel function of the first kind, j(@) The detector/angular response function, k The imaginary'part of the refractive index, Ka The absorption coefficient, Btu in Kc The conductivity for both conduction and convection hrt.,sq ft.,F kc The thermal conductivity,.ft n~F' kr The radiant conductivity, Btu, in. hr.,sq.ft., F Ks The scattering coefficient, Btu in. kv The convection conductivity, hr.sq.ft F' L The mean fiber length, I The insulation thickness, in., -i M The interception parameter, in., m* The refractive index, m The real part of the refractive index, N The scattering parameter, in.-l n The number of scattering bodies per unit volume of insulation, cu.in. P The absorption Parameter, in. -1 Pn A scattering of function, p The slope of the straight line portion of the tran.,mni ssio n curve, 2n p The propagation constant, L, iL A;,pel(-ix C olnly, q The intercept of the straight litle portion of the trari.s;tsnion curve, qc'The rate of heat trarIF:fer by eorl i,'tlor, [er wixi ta.rea[x[. - it The rate of h-at -traritle'rc peer;:1t,,Jr.....,.ra

-167Btu qr The rate of heat transfer by radiation per unit area, h rBq ft qv The rate of heat transfer by convection per unit area, Btu hr.,sq.ft. R The radius of a reflecti:;g sphere, r A radial distance variable, Sa The absorption cross section per body, sq.in., Sr The radial Poynting vector, Ss The scattering cross section per body, sq.in., T The insulation temperature, degrees Rankine, t The independent variable measuring time, in Appendix C only, t The fraction transmission through the insulation, u The thickness of solid associated with each bubble in a foamed insulation, w The weight of solid in a square foot of insulation, x The independent variable measuring the penetration into the insulation, in., y The independent variable describing the distance across the face of the insulation, z The independent variable measuring the distance along the cylindrical axis.

-168Greek Letters a The sphere or cylinder circumference measured in wavelengths, B The expotential absorption coefficient, 7 Euler's constant, 1.781, 8 The distance between the detector and the insulation, e The emissivity, @ The angle a scattered beam makes with the forward direction of the incident beam, radians, X The wavelength of radiant energy, microns, P The bulk density of the insulation, lb./cu.ft., P The electric charge density, in Appendix C only, pg The density of the gas filling the void space in the insulation, lb./cu.ft., os The density of solid material in the insulation, lb./cu.ft., a The Stefan-Boltzmann constant, 0.1713 hr.,.ftu, O4 c' The electrical conductivity, in Appendix C only, n The solid angle subtended by the insulation about the receiver.

-169BIBLIOGRAPHY 1. Allcut, E. A., "An Analysis of Heat Transfer Through Thermal Insulating Materials," in the Proceedings of the General Discussion on Heat Transfer, Institution of Mechanical Engineers, London, 232-235, (1951). 2. Boll, R. H., Gumprecht, R. 0., and Sliepcevich, C. M., J. Opt. Soc. Am., 44., 18-21, (1954). 3. Bosworth, R. L. G., Heat Transfer Phenomena, John Wiley and Sons, New York, 59-61, (1952-). 4. Chu, C.. M., Clark, G. C., and Churlhill, S. W., "Angular Distribution Coefficients," University of Michigan Engineering Research Institute, Ann Arbor, Michigan (1957). 5. Chu, C. M., and Churchill, S. W., IRE Trans., AP-4,2, 142-148, (1956). 6. Clark, G. C., Chu, C. M., and Churchill, S. W., J. Opt. Soc. Am., 47, 81-84, (1957). 7. "Foamglas Insulation for Piping & Process Equipment," Pittsburgh Corning Corp., Pittsburgh, Pennsylvania, (1951). 8. Hamaker, H. C., Phillips Res. Rep. 2, 55-67, 103-125, (1947). 9. Hollingsworth, M., Private Commu ilcation. 10. Jakob, Max, Heat Transfer, Vol. I, John Wiley and Sons, New York, 86-87, (194497).? 11. McAdams, W. H., Heat Trans:nission, McGraw-Hill, New York, 10-11, (1954). 12. McIntire, 0. R., and Kennedy, R. N., Chem. Eng. Prog., 44, 727, (1948). 13. Mie, G., Ann. der Physik, 25, 377 (1908). 14. Montroll, E. W., and Hart, R. W., J. Appl. Phys., 22, 1278, (1951). 15. Schaefer, C., Ann. der Physik, 31, 455 (1910). 16. Scott, P. H., Clark, G. C., Sliepcevich, C. M., J. Phys. Chel_.., 59, 849, (1955). 17. Stratton, J. A., Electromagnetic Theory, McGraw-Hill, New York, 563, (1941). 18. "Styrofoam," The Dow Chemical Company, Plastics Technical Service, Midland, Michigan, (1955). 19. Topper, Leonard, Ind. Eng. Chem., 47, 1377-1379, (1955). 20. Verschoor, J. D. and Greebler, Paul, A.S.M.E. ITrans. 74, 961-968, (1952).

UNIVERSITY OF MICHIGAN I3 9lllIIIIIIII01I5 03023 222011l 3 9015 03023 2220