PAPER-HCNEyCCMD CCRES FCR STRUCTURAL SANDWICH PANELS March I )t/ Of M ill II FU No. R1918 UNITED STATES DEPARTMENT OF AGRICULTURE FOREST SERVICE FOREST PRODUCTS LABORATORY Madison 5, Wisconsin In Cooperation with the University of Wisconsin Digitized by the Internet Archive in 2013 http://archive.org/details/honeoOOfore PAPER-HO" '-' ~ Bv . J. SLIDE, Chemical Engineer Forest Produe al 'oratory,- Forest Service U. S. Department ox Agriculture Sum mary Bonding of paper-honeycomb cores between thin and strong facing sheets results in sandwich panels that are becoming increasingly Important for structural applications. The panels are strong, stiff, light, and economical of raw materials. They can be made by processes that lend themselves to mass produc- tion from materials that are produced in large volume. This paper summarizes briefly some of the research at the U. S. Forest Products Laboratory on sandwich building panels having cores made from corrugated paper of low resin content. Extensive testing of these panels in the wet and dry condition has shown that satisfactory performance could be expected, and a considerable amount of data has been obtained to aid in the structural design of panels. For use in dwelling units, strength tests have indicated that a suitable sandwich wall panel is comparable to conventional house construction in bending strength and resistance to vertical loads. With proper treatment, the cores have reasonably good decay resistance. Thermal insulation values are relatively good and depend on the core density and specific construction; for some uses, the thickness of the panel may be determined by the thermal requirements. Insulation values can be further improved by filling cells with materials such as foams or loose insulation. Bowing of panels due to differential moisture and thermal conditions was found to be of approximately the same order as that of insulated, stressed- facing panels made of plywood glued to wood frames. The fire resistance of -Presented at a meeting of the Plastics Committee of the Technical Association of the Pulp and Paper Industry, Syracuse, "I. Y., Nov. 9, 1951. 2 Acknowledgment is made of information used in this report based on the work of various investigators at the Forest Products Laboratory and on work done by the Laboratory in coooeration with the Housing and Home Finance Agency. 3 ""Maintained at Madison, Wis., in cooperation with the University of Wisconsin. Rept. No. R1918 -1- Agriculture-Madison honeycomb panels is appreciably higher than nearly hollow panels with like facings. Filling of cells with foamed resin further increases fire resistance. Results of accelerated aging tests are favorable. Durability and bowing under outdoor conditions are under study in a sandwich- panel test unit, which was built in 19U7 and is still in very good condition. Introduction The principle of the sandwich panel was undoubtedly put to effective use for many years before it was defined by engineers and recognized as a separate type of construction dominated by certain mathematical principles. General recognition occurred during the accelerated search for high- strength, light- weight materials for aircraft in World "Jar II. Simply stated, a sandwich panel consists of an appreciably thick core of a low-density material bonded on each side to a thin sheet of strong, stiff material. A great variety of facing materials, such as wood, hardboard, asbestos board, aluminum, and plastic laminates, have been bonded to many lightweight core materials such as balsa wood, rubber, plastic foams, and formed sheets of cloth, metal, or paper. The new and ingenious materials, constructions, and ideas for use multiplied rapidly. Stringent demands for strength and special characteristics were set by designers to meet needs for wartime aircraft. As the development progressed, a distinction between highly specialized aircraft panels and other general building panels became evident. Obviously, a radical change in the outlook regarding materials, properties, and economics was indicated, although the principle remained unchanged. Demands for strength appeared less exacting and requirements for durability and thermal insulation were perhaps more critical for building panels than for aircraft panels. Problems of supply and cost were paramount for building panels. Cores made of sheet materials, such as cloth, nonwoven fabric, or paper formed into cellular configurations of one kind or another, became more important. Then careful consideration was given to economics, availability, and other properties, it became in- creasingly clear that paper would play a very important part in the large- scale development of honeycomb. It may indeed be only a slight overstatement to say that paper is almost the only material apparent at present for the large-scale development of high-strength honeycomb panels. In 19hl, following several years of research with lightweight, paper-core constructions and sandwich panels at the Forest Products Laboratory, the status of the work was reviewed from the standpoint of panel applications in building. Many of the most promising ideas were gathered together and crystallized in the form of a sandwich- panel test building unit for outdoor exposure (11).— K, nderlined numbers in parentheses refer to literature cited at end of report, Rept. No. R1918 -2- Since the erection of the test unit, a continuing interest in the structural sandwich panel has developed, as indicated by the large number of inquiries received from an unusual variety of sources. Kany are interested in only certain phases of the development, such as fabrication of the core or press- of the panels. Pulp and paper mills endeavoring to increase the number of products possible from their raw material consider the sandwich panel as an additional long-term outlet for paper. Some resin manufacturers consider th?t manufacture of sandwich panels provides a promising use for their products in core treatment and in plastic sandwich facings. A few boxboard mills have observed the similarity between their regular products an honeycomb and arc. following the development closely, thinking perhaps of the amplication of boxboard experience to honeycomb and, conversely, honeycomb principles to boxboards and containers. Plywood and veneer mills are interested in sandwich pan< Is as a means of increasing the monetary yield from a supply of logs. ! etal compan: ask about the feasibility of usin^ metal as facing v rial and see in the sandwich p-ncl an opportunity to use thin metal sheets for applications where they previously were not suitable. The advantages of sandwich con- struction in furniture and doors arc evident to of 1 'oducts. Architects and builders interested in n Lais inquire about strength, fire resistance, and thermal insulation afforded 'oy the hen ycomb core, feasibility of using sandwich construction in buildings, and possible sources of sandwich panels. hakers of hardboards end composition boards visual! n :w and int ] I Lag uses for their products. The continuous support that 4 fiberboards themselves suitable for items such as walls, partitions, and doors. 'hen fiberboards are used as facings, the possibility of maki high- strength structural panels from pulpwood and woods residues is evident. Basically, the problem of making honeycomb from paper can be viewed as one of "bl< -up" a certain volume of a dense paper to yiel i uch lar volume of a low-density cellular structure with certain properties. After the spacing between sandwich facings has been determined by design, the honeycomb must fill all of the volume between the facir . Density of core is therefore one of the key factors in economy, as the materials are rela- tively costly on a weight basis, and they must be made to provide the greatest possible volume consistent with adequate- property Paper can be "increased in volume" to produce core in numerous ways. Three of the more nrominent types used for sa: :h panels might be describee as "expanded," "figure 8," and "corrugated," In the expanded type, sheets of treated paper are laid up and interspaced with parallel and uniformly spaced strips of adhesive. Successive strips of adhesive are placed so that the centers of any one layer are positioned at midpoint between the strips of the preceding and the succeeding layer. The blanks after bonding are ex- panded to form hexagonal cell sections, as shown in figure 1. There has been an increasingly large interest in honeycomb structures of this typ (1, 2, h s 5, 7). Rept. No. R1918 -3- To produce the figure-8 core, a ribbon of paper is looped and bonded to form circular cells resembling a figure 8 in cross section, as shown in figure 2. For the production of either the figure-8 type or the expanded type, special machines are required. To produce the corrugated type of honeycomb, the cell or flute is made by hot forming paper between fluted rolls on equipment of the type used in making corrugated container board. The corrugated sheet, with or without uncorrugated interleaving sheets, can be assembled in many ways. The three types discussed chiefly in this report are shown in figures 3, h, and 5>. Each type of core has advantages and disadvantages peculiar to its construc- tion. The differences lie in equipment needed for production, limitations in resins and adhesives, relative difficulty of accurate cutting, strength, thermal insulating properties, ease of shipment, and other factors. As far as the base material and resin treatment are concerned, such problems as fiber, resin, wet strength, and durability are substantially common to all types. It was there- fore intended that the results of this work should apply to any honeycomb struc- ture made from paper treated with phenolic resin, although most of the data were obtained from sandwich panels having corrugated paper cores. This report summarizes some of the research that has been conducted at the Forest Products Laboratory on the structural type of corrugated-paper-honeycomb cores for use in building panels, as distinguished from use in aircraft. Consideration is given to the resin treatment of the base material and to the fabrication of the core and sandwich panels in the interest of low- cost building material. Strength properties, bowing, thermal properties, dura- bility, and fire resistance of the sandwich panels are discussed. Figure 6 shows a structural sandwich wall panel. Base Paoer For most experimental work on sandwich panels at the Laboratory, an unsized, neutral kraft paper was selected, because it was strong, potentially available in large quantities, and had favorable indications of permanence. The paper contained no sizing in order to permit better resin penetration and to favor permanence. A ^0-pound (weight per 3*000 square feet) paper was usually employed for cores having cells of A- flute size. Tests were made to explore the possibility of reducing the weight of paper without undue sacrifice of strength in the sand- wich panel. The use of lighter-weight papers results not only in an economy of base material but also in a reduction in the amount of saturating resin required to produce a given volume of core. Results showed that sandwich panels with surprisingly high strength properties could be made with paper weighing only 30 pounds per 3*000 square feet, which yielded core of a density of about 2.5 pounds per cubic foot. Thermal insulation data obtained on these core constructions showed that, for a particular core construction, an improvement in thermal values could be achieved by reducing the weight of the paper. The lower limit of paper weight was not determined, as it was felt Rept. No. R1918 -I- that paper of much lower weight than 30 pounds per 3,000 square feet would be difficult to handle in the corrugating and fabricating process. No information is available on the minimum strength requirements of the base sheet for satisfactory core structures. The relation between ordinary properties of paper and the properties of cores is obscure, and studies should be made to overcome this deficicn . ",\ r et strength can indicate the effectiveness of the resin treatment, ring compression of the sheet may have some s. sance, and accelerated agin; ma: ir - icatc permanence. Compres- sion, tensile, and shear tests on assembled core are informative but are costly and occur too late to assist paper makers in producing suitable papers. Until the requirements of paper for use in core can be defined in a more usable form, it will remain difficult to determine the best papers for par- ticular uses. At present, it is certainly safe to assume that any of the ordinary chemical pulps arc suitable. For many applications, especially those involving short time expectancy for which maximum permanence is not essential, a high- yield or scmichcmical pulp may be acceptable or even preferable for structural uses. Reclaimed paper may also provide a suitable fiber for cer- tain panels. Few data are available on the permanence of the high-yield or semichemical pulps. The permanence of the p-per may not, however, be critical, since the core mat rial in the sandwich panel is protected from ultraviolet rays, which contribute to the loss of strength of paper upon aging. Re sin- treated P aper One of the first problems common to any paper core is that of resin treatment of the paper. Although in some cases paper without resin might be acceptable, it is assumed that for general usage only resin-treated papers can be con- sidered. Since panels are likely to be subjected to damp or wet conditions, the presence of resin in the paper is necessary to yisld a product that is permanently strong and stiff in the wet condition. Even small amounts of cer- tain resins can be effective in this respect. In terms of tensile strength, it is not difficult to produce a paper that retains over 7^ percent of its strength when dry after water soaking that would cause untreated paper to lose almost all of its strength. The work reported relates to the use of phenol- formaldehyde resins of several types. Two of these are water-soluble and alcohol- soluble types and, although both are suitable, they have inherent differences that need to be considered when they are used with cellulose. With solutions of water-soluble resins, the swelling effect of the water on the fiber permits the resin to penetrate the fiber itself and through a "bulkina effect" (10) to maintain the fiber in a swollen state. This also causes a reduction in the equilibrium moisture content of the sheet with water vapor. iVhen alcohol- soluble resin is applied in the absence of a swelling solvent, the effect appears to be chiefly one of coating the fibers. The equilibrium moisture content of such paper, when based on the oven-dry weight of fiber alone, is nearly as high as that of untreated paper (3). Both may have high levels of strength in the wet condition. Rept. Mo. R1918 -5- Additional evidence that the water-soluble resin penetrates the fiber better than alcohol- soluble resin was indicated by decay tests made on paper treated with both types of resin (8). The results showed that a paper treated with 15 percent of water-soluble resin has considerable resistance to deca3>- fungi, as measured by strength, while the paper containing an equal amount of alcohol- soluble resin lost most of its strength under similar exposure conditions. I'Jhile the water-soluble phenolic resins are not especially good as fungicides, they can reduce the equilibrium moisture content of the paper and thus dis- courage the grox-rt-h o r fungi. Inherent brittleness may be a disadvantage of paper impregnated with water-soluble resin. A resin treatment of 1$ percent of water-soluble resin (based on weights of resin and fiber) was found to be adequate for providing paper of good strength when wet, decay resistance, and handling characteristics during corrugation and subsequent fabrication. Resin content in excess of about 15 percent does not seem to produce a gain in strength commensurate with the increased quantity of resin required. The tensile strength of a kraft paper treated with 15 percent of water-soluble phenolic resin ma;> be greater when wet after prolonged soaking in x^rater than the strength of the untreated paper x percent of resin was bonded to plywood and tested in tension and shear after exposure to a number of extreme temperature end moisture cycles. The following adhesiv produced sufficiently strong bonds between the core and facings to force a high percentage of failure in the core: (a) An acid-catalyzed, high- temperature-setting, phenol- resin adhesive. (b) An alkaline- catalyzed, intermediate- temperature- setting, phenol-resin adhesive. (c) An alkaline-catalyzed, room- temperature- setting resorcinol-resin adhesive. (d) An alkaline- catalyzed, high- temperature-setting, phenol-resin adhesive. The bonding of most of the experimental sandwich panels was done in hot-press equipment. For laboratory work, the short pressing cycle and the long assembly time favored the acid-catalyzed, high-temperature- setting, phenol-resin adhe- sive. Adhesive was applied to both the core and facings with a rubber roller or an ordinary paint roller at the rate of about 22 grams per square foot of surface, one-half to the core and one-half to the facings. It was also demonstrated that it may not be necessary to apply adhesive to the core in order to produce a satisfactory bond. Core and facings were allowed to stand after spread of adhesive to permit the evaporation of the solvent. Panel components were then assembled and placed in a hot press at a temperature of 230° F. (for the aoid-catalyzed phenolic adhesive) to cure the adhesive. Pressures ranging from 15 to 50 pounds per square inch were used, depending on density and type of core construction. Test panels up to lh feet in length and 6 inches in thickness were made in a h by U-foot laboratory press using step-pressinr techniques. Flat sandwich panels were produced with this method. Special presses are indicated for sandwich- panel manufacture. The pressures required are usually lower than can be obtained in the range of good pressure control on ordinary plywood or plastic presses. Because pressure require- ments are low, simple and perhaps less costly presses could be used. Contin- uous roller presses or bag-molding equipment may also be suitable. Certain special problems arise in the pressing of sandwich panels, but their manufac- ture is basically not complicated. Pressing of sandwich panels having dissimilar facings in hot presses has been difficult because of unequal dimensional movement of the facings due to mois- ture or thermal changes. In such cases, cold-pressing may be necessary. This of course creates its own problems with respect to adhesive limitations, assembly time, wat< r introduced with adhesive, long time under pressure and, in some cases, durability of the bond. Ordinary cold-setting adhesives, Rept. No. R1918 -I 1 - however, are undoubtedly adequate for a host of prospective sandwich- panel applications. One of the most persistent difficulties in the use of sandwich panels is in the problems caused by the necessity for edges, inserts, and connectors for panels. In some cases, the problem involves tying together thin fac- ing materials without severe stress concentrations and, in other cases, such as furniture, the problem is caused by u show-through" of core or inserts through decorative facings. These problems, probably as much as any other factor, have restricted the development and use of honeycomb core on a larger scale and should be studied from several standpoints. The differential dimensional movement between core and insert materials should be at a minimum, including the rate as well as the degree of move- ment. Adl'.esives and moisture introduced by adhesives would be a factor in this study. Examples of materials to consider for edges or inserts would be end-grain wood, plywood on edge, part honeycomb and part vood, metal, dense honeycomb, and mastics or fillers© Another approach would be to study engineering design factors for getting panels into the ultimate product without the use of molded-in inserts; this has certain ideal features from the standpoint of sandwich-panel manufacture, and it simplifies pressing. Properties of Sandwich Panels Strength Strength data were obtained on both large and small sandwich panels compris- ing paper-honeycomb core with facings of veneer, plywood, hardboard, asbestos board, aluminum, or other materials. These tests included static bending, impact bending, and column tests. The most common test conducted was the static bending test, which consisted of supporting the panels at the ends and applying an increasing load at two quarter-span points. The amount of deflection was recorded at various loads to the design load or until failure occurred. This test not only produced useful information on stresses developed in the facings and on stiffness of the assembly but gave infor- mation indicative of the shear strength of the cores and the quality of the bond between core and facings. In the impact bending test, a 10-inch-diameter sandbag, weighing 60 pounds, was dropped on the center of a large sand/rich panel, beginning at a height of 1 foot and increasing by increments of 1 foot to a height of 10 feet or until failure occurred. Panels were supported at each end, and both, instantaneous and permanent deflections were measured. The vertical load test was made to determine the ability of a panel to meet certain column structure requirements. The shortening of the panel in the vertical direction and its lateral deflections were measured. Strength tests conducted on large-size sandwich wall panels, 3 inches thick and having plyvood facings, indicated higher shear strengths developed in bending and greater resistance to vertical loads than found in conventional Rept. No. R191B -12- house construction. These panels withstood maximum loads in bending of at least 12 times the design loads of 20 pounds per square foot sometimes used for house panels. A considerable amount of design data has been obtained to make it possible to predict the strength of sandwich panels or to design them to meet certain strength requirements. The core of the sandwich may be considered as only a means of separating and stabilizing the facings to produce a member similar to an I-beam. The core simulates the web and the facing the flanges of an I-beam, wince the core in the sandwich is subjected to shear, as is the web of the I-beam, it is essential that the cor sufficiently strong to wi stand the shearing stresses imposed. It is assumed, however, that the core itself adds no stiffness to the sandwich construction. The following formulas are used for predicting the behavior of sandwich con- structions subjected to various loads. No consideration is given to concen- trations that may occur at joints and fastenings. Each type of fastening will have to be considered individually insofar as its effectiveness in trans- mitting loads and its strength are concerned. The constructions arc assumed to be well bonded with an adhesive and strong and durable enough to perform adequately for the conditions of use. The mean stresses in the facings of sandwich construction under flatwise bending loads are given by - ^ (i) 1,2 f x 2 (h+c)b where cr ^ ? = facing stresses in facings 1 or 2 f, , = thicknesses of facings 1 or 2 I: = moment h = total sandwich thickness c = core thickness b = width of sandwich The shear stress in the core of sandwich construction und - loads is given by 2V T = (FT^ (2) where r = core she:ir stress V = shear load Rept. No. R191B -13- The stiffness of sandwich construction having facings of the same material and equal in thickness is given by E r (h 3 -c 3 )b D = (3) 12X where D = stiffness E = modulus of elasticity of the facings F ^ =0.99 for wood or plywood facin-s ^ =0.91 for isotropic facings The maximum deflection of a sandwich construction having facings of the same material and equal in thickness under a uniformly applied load and simply supported at the ends is given by C Pa 3 I" -rr 2 cfE^ j 3oU D 2 2 \ a G L c where A = maximum deflection P = total load on the sandwich G = shear modulus of the core material c a = span Note: An approximation of the deflection may be obtained by neglecting the last term in the brackets in formula h. This term will be less than 10 percent for most constructions on long spans but may become appreciable for short spans. Strength data for sandwich panels of various thicknesses and comprising different facings are given in table 1. Most of the information on the 3- inch- thick panels was obtained by experimentation, while values for the 1- and 2- inch thicknesses were calculated mathematically with the above formulas. Values in the table show that reducing, the thickness of the panel from 3 inches to 1 inch would decrease its stiffness 10 to 1$ times and decrease the maximum load that it will suoport 3 to h times. To meet a span- deflect ion ratio of 270 or more under a uniform load of 20 pounds per square foot, a structural sandwich wall panel on a span of 96 inches would have to be more than 2 inches thick if its facings were of l/'i-inch Douglas-fir plywood. Certain properties of the honeycomb core may be varied considerably with only a mild deviation in stiffness of the resultant sandwich panel. Rept. No. R1918 -1.' - Dimensional stabi l ity and Moving of P anels In a structure such as a sandwich panel in which two facings are bonded to a core to form an integral panel, any Lonal mov< i one facing has an effect on the entire panel. A differential movement of facings causes bow- ing on an unrestrained panel. If dimensional change of both facings is equal, the length and width dimensions of the panel will increase or decreas< , but bowing will not result. This is important for many uses. The problem is chiefly related to the facings because the core does not have enough stiffness to cause ' of the panel or to cause it to remain flat, laboratory tests have demonstrated that the dimensional stability of a panel is not affected by the type of core. The magnitude of the bowing effect, however, depends on the thickness of core. The use of dissimilar facings is often desirable from an economic standpoint, yet dimensional instability of facings during panel manu- facture or exposure may rule out possible benefits, as indicated earlier in this report. For many applications, such as integral housing panels or what might be des- cribed as conventional prefabricated house panels, bow i : almost always a consideration. During exposures in housing when temperature and relative humidity conditions are approximately equal on both sides of the panel, the panel will remain reasonably stable. As col< Dps, the moisture content on the inner facing of roscopic mat rial decreases and that of the outer facing increases, causing an oui f ' ow. The effects of winter temperatures produce opposite curvature . r v the summertime, the bowing is less pronounced, since the moisture differential is less. This character- istic bowing pattern has been experienced by producers of stressed- skin, ply- wood prefabricated panels for many years. Although shrinkage on ansion in plywood with moisture changes is slight, the diff i . is enough to cause a detectable bow. In wood, hnrdboard, or other hygroscopic materials, warp- can be attributed to both moistur ai temperature differences, while in al- faced sandwich panels, only temperature differences cause dimensional changes, and these changes are due to thermal expansion or contraction. In conventional construction, it is desirable to install vapor barriers, usually of asphalt-impregnated paper or metal foil, to block the migration of vapor to the cold side of a wall. Various experiments were conducted or proposed to improve vapor resistance of sandwich panels, such as bonding of metal foil; blending aluminum flake with resin bonding adhesives; use of plastic vapor barriers between veneers, overlay papers, and special finishes; and, of course, metal or plastic facings. jause added cost is likely, some of these should not be resorted to unless their need has been demonstrated. Panel edges may also be sealed by various methods if this is indicated. As a generalization, the bowing of sandwich panels is probably neither greater nor less than stressed-skin housi aving similar facings. In one series of tests of bowing under laboratory conditions, six sandwich panels, each 3 inches thick, subjected to s 1 ire differences to determine their bowing characteristics. proximately 20 by 72 inches were placed in openings between two rooms, one of which was held at a tei - ture of -20 ?. and the other at +70' 7. The edges of the panels were sealed with aluminum fail to prevent the entrance of moisture. The facings Rept. Uo. rtl9l8 -1 - consisted of three-ply, l/U-inch Douglas- fir plywood with and without paper overlay* two-ply, 1/5;- inch plywood with paper overlay; and overlaid, l/8-inch veneer. One of the plywood- faced panels had two coats of aluminum paint on the warm side only. All panels deflected slightly toward the warm side immediately after being in- stalled, as the facing on the cold side underwent thermal contraction. As the test progressed, the deflection became less prominent, because the moisture content of the cold side increased. The overlaid, three- and two-ply construc- tions after 9 weeks' exposure became straight. At the end of the exposure, the overlaid veneer panel showed the greatest deflection, although the overlaid, three-ply panel showed the greatest initial change. After the panels were re- moved from the wall and exposed to 70° F. on both sides for 3 hours, the panels reversed themselves, bowing toward the cold wall. This was probably caused by the plywood on the cold side undergoing thermal expansion and possibly, in part by the absorption of moisture from the melting of the frost crystals that had formed inside the panels during the exposure. Although these test conditions are not exactly typical of ordinary service conditions, the amount of deflec- tion observed in these panels would not be considered objectionable. Accurate data on the dimensional change of sheet facing materials due to moisture or thermal changes is very important to a study of panel bowing. It is possible to calculate mathematically the bowing of a sandwich construc- tion if the percent expansion of each facing is known. The maximum deflection due to bowing caused by the expansion of one facing resulting from tempera- ture or moisture differential is given approximately by ^ = ka2 30 Oh where k = percent expansion of one facing as compared to the opposite facing a = length of panel h = total sandwich thickness Using this formula, the approximate bowing deflection was calculated for hardboard- and plywood- faced sandwich panels of various thicknesses. This information is given in table 1. The bowing deflection due to moisture or temperature differences is greatly dependent on thickness of the panel. As an example, a 96-inch-long, plywood- faced panel 3 inches thick bows about one-third as much as a similar panel 1 inch thick. Thermal Conductivity The basic values determined in establishing the thermal conductivity of a material or combination of materials used in a structure can be defined as follows: Rept. No. K1918 -1 '- k - thermal conductivity: The tine rate of heat flow through a homogeneous material or one of uniform structure under steady state conditions, expressed in British thermal units per hour per square foot per inch of thickness per degree difference in temperature between surface of the material. U - ove^-all coefficient of heat transmission, air to air: The time rate of heat flow expressed in British thermal units per hour per square foot per degree difference in temperature. The term U applies to the combination of materials used in a construction; for example, both facings and core of a sandwich, and includes standard values for fj_ and f for still air on the warm side and air moving at 15 miles per hour on the cold side. The usual method of comparing insulating values for wall construction is by comparison of heat transmission coefficients or U values. The U value for a construction is found from the relation 4 where the values of k are for the several constituents of the wall and t the thickness of these constituents. To compare insulating characteristics of various core constructions, tests were made on 1- inch- thick specimens using a guarded hot-plate apparatus. The data are summarized in table 2. The conductivity value of the core is affected by: cell size, density, resin content, construction, and type of material. Thermal insulation was significantly affected by the type of construction; values for k varied from as low as 0.30 to as high as 0.55 with variations in core structure. A k value of 0.U5 and 0,'t: British thermal units per hour per square foot per inch of thickness ocr °7. was obtained with cross-corrugated (XW) or the parallel- corrugated (P'l) structures having flutes perpendicular to facings. By placing these same structures in the test so that flutes were parallel instead of perpendicular to the plates, the value was reduced to 0.30 to 0.3!.. " n ut these structures have other disadvan- tages, as mentioned earlier in this report. For any particular core construc- tion, the density of the structure affects the insulation value. As an ex- ample, a PIJL core having a density of about 5.5 pounds per cubic foot had a k value of about 0.59, while a similarly assembled core, with a density of 3.3~> pounds per cubic foot, gave a k value of about 0.U7. An actual U value of 0.1 50 British thermal units per square foot per hour per °F. was obtained on a 3-inch- thick, large-size panel having the cross- corrugated (XM) type of core and l/h-inch, three-ply Douglas-fir plywood facings. This panel was exposed in a wall between two rooms, one controlled at 72° F. and U0 percent relative humidity and the other at -20° F. Th.is 0.150 value compares reasonably well with a calculated U value of O.lUl for the same panel. If the thickness of this oanel was reduced to 1 inch, the 1 value could be expected to increase to about 0.U2. - Rept. No. R1915 -17- An improvement in the insulation value of the sandwich construction can be realized by filling the honeycomb core with insulation or a foamed- in-place resin. A reduction in the k value of a corrugated core from 0,U6 to O.hO British thermal units per hour per square foot per inch of thickness per °F. was obtained when a phenolic resin was foamed into the core. A slightly lower value was obtained through the use of fill insulation. Foaming of resins into honeycomb appears very promising as a means of improving thermal insulation and fire resistance. F ire Resistance Sandwich panels were tested for fire resistance by two methods designed for housing materials: (1) Exposing one face of the panel to a standard flame that approximates conditions of fire in a house in which furnishings are being burned; and (2) introducing flame through a hole in the facing, as might occur at openings for electrical conduit or other house equipment. The first method gives fire-resistance values for sandwich construction that are comparable with those accepted for other types of house construction. The critical factor in this test is the ability of the facings and the bond between the facings and core to resist the high temperature without developing construction failures. Obviously when the facings or bond have failed, the construction is gravely weakened, since the facings are the principal load- carrying elements of the sandwich. At the Laboratory, this test was con- ducted in a gas- fired furnace according to the exposure conditions specified in American Society for Testing Materials (ASTH) Specification No. E-119-li7. The fire resistance of the wood- faced sandwich panels was appreciably higher than hollow panels faced with the same thickness of plywood. When aluminum- faced panels were exposed to the gas, a rather rapid buckling of the facing and failure of the bond vsually occurred. In most cases, the cores were badly charred after the test but retained their original form to a considerable extent. The fire resistance of the sandwich panel can be increased considerably by incorporating in the core foamed resin or an intumescent coating material, such as certain types of sodium silicates. It is conceivable that a material could be developed for deposition in the core that would serve the triple function of bonding the paper sheets, providing foam for fire resistance, and also improving, thermal insulation. By the second method, the likelihood of flame spread in the core was investi- gated. This was done by cutting a small hole in one facing, holding the panel vertically, and applying a gas flame to a small area for U minutes. In panels having flutes perpendicular to the facings, such as in the expanded figure-8 and the corrugated PN and PNL types of core, only slight flame spread occurred. Burning was restricted to the honeycomb material in contact with the flame. When the flame was removed, flaming stopped immediately. Some glow persisted for an additional 1 to 2 minutes. In the case of the cross- corrugated XN type of core in which one-half of the flutes are parallel to the length of the panel, the spread of flame occurred in the vertical direction due to the open channels. This could no doubt be readily improved by placing a Rept. Mo. R1918 -18- barrier sheet at the top of the panel or at intervals in the panel height, or perhaps by simply turning the length of the core blocks at 90 c to the vertical direction. The re sin- treated core in itself is not fire-resistant, but its use between sandwich facings does not seem to be hazardous. It is possible to add fire- resistant chemicals to the paper or to dip or spray the core assembly with such chemicals, but this is believed to be unnecessary when wood-based facings are employed in the panel construction. Such treatments could be effected, if necessary, but might create gluing and moisture- absorption problems, and their effects on long-time aging characteristics of paper are not well understood. Sand wi c h-panel Exterior Test Unit In 19U7* an experimental sandwich unit about 12 by U0 feet in size was built on the Laboratory grounds for long-time exposure tests. Incorporated in this unit are various types of sandwich wall, floor, partition, and roof panels. All have paper-honeycomb cores and are faced with veneer, plywood, overlaid plywood, hardboard, asbestos board, or aluminum for comparative purposes. The unit on the inside is equipped with heating coils and is controlled dur- ing the winter at a temperature of 72° F. and a relative humidity of U0 per- cent. The outside of the unit is exposed to the variable outdoor temperatures of the Madison, Wis., area. Before the wood-faced and aluminum- faced panels were installed in the test unit, they were tested to determine their deflection and span-deflection ratios at design loads. The 3- inch-thick, wood- faced wall panels met the requirement that the span-deflection ratio be not less than 270 under a design load of 20 pounds per square foot. Floor panels of 6- inch thickness with fac- ings of 3/8- inch, five-ply Douglas- fir plywood had a span-deflection ratio of about 800 under a load of k0 pounds per square foot (11). After 16 months' exposure, four wall panels were removed from the experimental unit for test. Two panels had facings of l/h-inch Douglas- fir plywood, and two had facings of 1/8- inch Douglas- fir veneer overlaid with paper. Cores were of the cross- corrugated XM type. On removal, the panels showed no visible signs of deterioration in either the core or the facings. The stiffness of the exposed panels was equal to or slightly greater than their stiffness prior to installation. This increase in stiffness may possibly be due to the further cure of the resin upon aging. Maximum loads obtained on these panels varied from 13 to 20 times the design load of 20 pounds per square foot. To obtain additional information on the effect of continuous weathering on appearance and strength, sandwich panels having 1- inch-thick, paper-honevcomb core and various wood facings with and without paper overlays were subjected to the following conditions in the Laboratory: 1. Immersed in water at 122° F. for 1 hour. 2. Sprayed with wet steam at 19h° to 200'- F. for 3 hours. 3. Stored at 10° F. for 20 hours. Ii. Heated in dry air at 212° F. for 3 hours. Rept. No. R1918 -19- 5. Sprayed with wet steam at 19U* to 200° F. for 3 hours. 6. Heated in dry air at 212° F. for 18 hours. This sequence of exposures was continued through six cycles, after which appearance of the specimens was noted, and bending tests were made to determine any change in strength properties, results of these tests were compared with those of tests made on control specimens not subjected to the aging tests. The reduction in shear stress developed in the cores of the aged specimens was about 20 to 30 percent as compared with that of the control specimens. Reduc- tion in stiffness was about 20 percent as obtained from a comparison of load- deflection ratios. The sandwich specimens were exceptionally straight, and no visual defects were apparent in the core. Although accelerated aging tests are never completely satisfactory, the performance of such specimens plus the observations made on the actual exposure unit indicate that good performance could be expected. One of the reasons for building the test unit was to obtain measurements of the actual bowing under outdoor conditions. Some data were obtained during the first year of exposure of these panels. Shortly after the unit was con- structed in June 19^7, deflection data showed a tendency of the panels with wood facings to bow slightly inward. In November when the heating system was turned on, the wood panels reversed their movement and bowed outward. The outward bowing was due to the lower outside temperature, which caused an increase in moisture content of the outer facing, and the shrinkage of the inner facings due to the heat. The bowing increased progressively in the wood-faced panels as the average outdoor temperature decreased and continued until late in March when the outdoor temperature began to rise again. The maximum bowing recorded in plywood- faced panels was about one-fourth inch. Aluminum- faced panels, not being affected by moisture, bowed toward the in- side as the temperature of the outside dropped. On a hot day with a high surface temperature on the outside, an outward bow could be noted. Results of the first year indicated that bowing of sandwich panels with facings of three-ply plywood is consistent with that of panels with three-ply stressed facings. Whether the inner surface was untreated, had aluminum paint, or had an overlay appeared to be unimportant in panels with three-ply facings. As the development of the sandwich structural panel grows, the need for more data on the behavior of such panels under ordinary exposure conditions will increase. It is hoped that this test unit will be helpful in resolving some of the problems regarding large-scale use of sandwich panels. Although there are many other uses for sandwich panels, the evaluation of these panels under housing conditions provides information that applies to most other products in which use of sandwich panels might be considered. No attempt is made in this report to detail actual or proposed uses for sand- wich panels, but these include partitions, doors, spandrell panels, and other constructions in houses, trailers, shelter buildings, warehouses, and farm buildings, lightweight shipping containers, and furniture. Because of the inherent structural strength of these panels, the greatest total benefit can probably be realized by using them to carry the principal loads in a construc- tion, not just to provide coverage. The general trend toward the use of sheet materials, both on the inner and outer surfaces of buildings, also points to long-term importance of sandwich panels as building materials, Rept. No. R1918 -20- Literature Cite d (1) AUTHOR UNKNOWN 19U6. EXPANDED STRUCTURAL PLASTICS. Bakelite Review, July. T55T. HONEYCOMB COKES OF AGE. Modern Plastics Vol. 28, No. 11. (3) BAIRD, P. K., SEIDL, R. J., AND FAHEY, D. J. 19U9. EFFECT OF PHENOLIC RESINS ON PHYSICAL PROPERTIES OF KRAFT PAPER. Forest Products Lab. Rept. R17SO, 10 pp., illus. (U) GRAHAM, P. H. 1951. HONEYCOMB CORE GIVES BEST WEIGHT-STRENGTH RATIO. Veneers & Plywood Vol. U5, No. 8. (5) LINCOLN, J. D. 19U6. PRODUCTION OF HONEYCOMB CORES. Modern Plastics Vol. 23, No. 9. (6) NORRIS, C. B. AND MACKIN, G. E. 19U8. AN INVESTIGATION OF ! ' fIC:\L PROPERTIES OF HONEYCOMB STRUC- TURES MADE OF RESIN- IMPREGNATED PAPER. Natl. Advisory Committee for Aeronautics. Tech. Note No. 1529. (7) RINGELSTETTER, L. A., VOSS, A. ''., A 'ORRIS, C. B. 1950. EFFECT OF CELL SHAPE ON COMPRESSIVE STRENGTH OF HEXAGONAL HONEYCOMB STRUCTURES. Natl. Advisory Committee for Aeronautics. Tech. Note No. 22U3. (8) SEIDL, R. J., KUENZI, E. W., FAHEY, D. J., AMD MOSES, C. S. 1951. PAPER-HONEYCOMB CORES FOR STRUCTURAL BUILDING PANELS: EFFECT OF RESINS, ADHESIVES, FUNGICIDES, AND 'EIGHT OF PAPER ON STRENGTH AND RESISTANCE TO DECAY. Forest Products Lab. Rept. No. R1796, 16 pp., illus. (9) ______ FAHEY, D. J., AND VOSS, A. W. 1~TT PROPERTIES OF HONEYCOMB CORES AS AFFECTED BY FIBER TYPE, FIBER ORIENTATION, RESIN TYPE AND AMOUNT. Natl. Advisory Committee for Aeronautics. Tech. Note No. 256U. (10) TARKOW, HAROLD 19U9. THE SWELLING AND SHRINKING OF WOOD, PAPER, AND COTTON TEXTILES AND THEIR CONTROL. Tappi Vol. 32, No. 5. (11) U. S. FOREST PRODUCTS LABORATORY 19U8. PHYSICAL PROPERTIES AND FABRICATION DETAILS OF EXPERIMENTAL HONEYCOMB-CORE SANDWICH HOUSE PANELS. Housing k Home Finance Agency Tech. Paper No. 7. Rept. No. R1918 -21- Table 1. — Stiffness and stre ngth of structural sandwich panels on a 96 -inch _ span- Facings : Panel • thick- ness : Center p : deflection^ : Span- : deflection ratio— Maximum : uniform : load : Approximate bowing _ deflection— In. 3 2 1 3 2 1 : 3 2 1 3 2 1 3 2 1 : In. : 0.18U J:53 : 2.390 : .207 : .U83 : 2.1).|0 : .169 : .395 : 1.810 : .202 : .U98 2.630 .313 : .735 3.350 521 : 212 Uo U6U : 199 : U5 553 : 2U2 53 1*75 : 193 : 36 : 306 : 130 : 29 Lb . per : In. sq. ft. : l/):-inch Douglas-fir plywood Two l/10-inch Douglas- fir veneers with 300 : 192 : 82 : 263 • 175 : 85 : 332 • 216 • : 101 : 3UU ! 218 : 9U : 269 • 176 : 82 : 0»38 58 i!i5 paper overlay One l/3-inch Douglas- fir veneer with paper overlay on each side l/U-inch tempered hardboard 1/8- inch tempered 1.00 i.5o 3.oo hardboard "All cores of the cross- corrugated XN type were made from 50-pound paper treated with 15 percent of resin. The core density was 2.5 pounds per cubic foot. 2 -Deflection under a uniform load of 20 pounds per square foot on a span of 96 inches. 3 -i.idspan deflection computed from the differential expansion of the two facings obtained by exposure of one facing to a relative humidit}' of 97 percent, the other to a relative humidity of 30 percent for 30 days. Rept. No. R1918 Table 2 . — Effect of core con struction, de nsity, and filler on thermal conductivity of core ma''e from re sin- treated paper Core : Filler Density k 1 construction in core k . per cu. ft. B.t.u. per hr. per sq, ft. per' inch . .... v. XN None 2,75 o.U5 : 2.22 do 2.9h M6 2.1 7 do 5.U7 3.35 .58 ,U7 1.73 : 2.13 PNL do <.$o .59 : 1. Figure 8 do 2,89 .53 1.89 pn Foamed resin : 5.36 J ■ 2. Figure C 1. .31 3.23 : Fill insulation m.72 .37 2.70 Rept. Mo. R1918 Figure 1 . - -Expanded type of core consisting of sheets of paper inter- spaced with parallel strips of adhesive and expanded to form hexa- gonal cell sections. Z M 87220 F JK JK ^»: JL ->: -J -*LJB -rff; -*> -<*• ■ J* Ajb-M -*: -* -dtf jil -*' -j* ^ &*l-tLgM ~M *# jl ^>: _*. ^ to«A jfiUkjKJL** ****** *iJ*-M JK. ^ ^* ^ K J* 4* ^Jk ->■ — *^^ -ai^ 1 -JK* -^T* -^»* '^^r* ,^ *. ^+: jtk J0- -#■ *+• JM ^: ^* &!&&§." Figure k. — Corrugated type of core designated as XN. It consists of corrugated sheets assembled with principal flute directions of adja- cent sheets at right angles . Z M 87222 F 2m 888-H F Figure 5. — Corrugated type of core designated as PN. It consists of corrugated sheets assembled parallel to each other and bonded at the crests . Figure 6. — Structural sandwich wall panel tested under vertical load. Z M 79770 F UNIVERSITY OF FLORIDA IlllllVilWIlllllllllll 3 1262 08866 6044