• A EFFECT OE UNBONDED JOINTS IN AN ALUMINUM HONEYCOMB-CORE MATERIAL FDD SANDWICH CONSTRUCTIONS May 1952 . ' ^. UbKUfal ■c This Report is One of a Series Issued in Cceperaticn with the AIR FORCE-NAVY-CIVIL SUBCOMMITTEE on AIRCRAFT DESIGN CRITERIA Under the Supervision of the AIRCRAFT COMMITTEE of the MUNITIONS BOARD No. 1835 UNITED STATES DEPARTMENT OF AGRICULTURE FOREST SERVICE FOREST PRODUCTS LABORATORY Madison 5, Wisconsin In Cooperation with the University of Wisconsin EFFECT OF UNBONDED JOINTS IN AN ALUMINUM HONEYC OMB-COR E ■ I m *'* ■»!■■■ I I - II ■ ■ I I ■■ ■ MH — ■— .^—i ■ I* — — ^— ^i^l» im~ ■■ ■■ ■■■■■^l» !■■■■■ f ■ W II ■ M 1^1 I MATERIAL FOR SANDWICH CONSTRUCTIONS^ By CHARLES B. NORRIS, Engineer p Forest Products Laboratory,- Forest Service U. S. Department of Agriculture Summary Static shear, fatigue shear, and compression tests were made on an aluminum honeycomb -core material. Bending and compression tests were made on sandwich panels having this core material and aluminum facings. The bonds between the honeycomb cells of some of the core material were completely removed. The modulus of rigidity and the shear strength of the unbonded material were found to be about 70 percent of those of the well -bonded material. Calculated values, taking into consideration the stress concen- trations in the neighborhood of the unbonded joints, lead to substantially the same value. Results of shear fatigue tests of the core material are substantially consistent with the shear-strength values obtained from the static shear tests. The compressive strength of the unbonded material is about 50 percent of that of the well-bonded material, which is consistent with the assumption that the compressive strength is proportional to the critical compression stress of the cell walls. Introduction The structure of a honeycomb-core material for sandwich construction is formed of sheets of material corrugated and bonded together as shown in figure 1, The core material is orientated in the sandwich construction so that the planes of the facings are perpendicular to the directions of the cells formed by the corrugated sheets. Thus the thickness of the sandwich construction is measured in the L direction of the core (the direction — - -This progress report is one of a series prepared and distributed by the Forest Products Laboratory under U. S. Navy, Bureau of Aeronautics, Order No. NAer 01237 and U. S. Air Force No, USAF-( 33-038) 51-^062E. Results here reported are preliminary and may be revised as additional data become available. 2 -Maintained at Madison, Wis., in cooperation with the University of Wisconsin. Report No. 1835 Agriculture -Madison perpendicular to the paper in figure 1) . The other two directions in the core material are indicated by the letters R and T as shown in the ficure. The shear B trains imposed on the core by the facings (when the sandwich construction is bent) lie, therefore, in the LR and LT planes. A question I as arisen regarding the necessary strength in the bends between the corrugated sheets of the core material. It is evident that If the shear stress in the core is so directed as to produce shear strains in the LR plane, the bonds should be capable of resisting shear stress; so t! at the shear strength of the bonds should be at least: m + n n B or about 3s for hexagonal core material: in which s is the shear strength of the core material that can be attained if the bonds do net fail, and m and n are the distances indicated in figure 1. It does not fcllcv, how- ever, that if the bonds have zero shear strength, the sandwich construction will have zero shear strength. If the bonds have zero shear strength, the shear stress will move from the core to the facings in the neighborhood of the unbonded joints in the core material, as described in Forest Products Laboratory Report No. 1505-A,- and shear stress concentrations will occur in the core material. If the shear stress in the core is so directed as to produce shear strains in the LT plane, the bond strength between the sheets in the core material should not be important unless failure in the core material involves buckling of the cell walls. The bond strength need be sufficient only to cause the double walls to act as a unit. Quite weak bonds should be suffi- cient for this purpose and, therefore, the required strength previously given for the LR plane is probably quite sufficient. The Glenn L. Martin Company has made a number of tests of an aluminum honeycomb -core material and of a sandwich construction embodying this core material and aluminum facings, to determine the effect on shear strengi of complete lack of bond in the core material. The following tests were made on both well-bonded and completely unbonded cere materialr : 1. Block shear tests of the core materials to determine modulus of rigidity and shear strength. the shear strength of the core materials. 3. Fatigue tests in shear on the core materials. h. Compressive tests of t.ie sandwich constructions and the core materials. 3 C. E. ilorris, W. S. hricksen, and W. J. Kommers. Fleimral Rigidity of a Rectangular Strip of Sandwich Construction. E u; pleinentary Mathematical Analysis and Comparison with the Results of Tests. Forest Products Laboratory Report V.o. 1505-A. Report No. IO35 -2- It is the purpose of this report to present the results of these tests and an analysis of them. Material The honeycomb material was made of aluminum foil 0.003 inch thick, per- forated in the usual way. The cells were substantially hexagonal, and the cell size (C in fig. l) was 3/8 inch. Two different foils were used. Foil A had a tensile strength of 33*^00 pounds per square inch, and foil B of 26,000 pounds per square inch. The core materials made from these foils had slightly different densities. The densities were: Core material A (cade from foil A), k.kQ pounds per cubic foot; and core material B, k.k5 pounds per cubic foot. The bond strength of these two core materials was roughly determined by applying tensile forces to the core material in the R direction. The tensile strengths (peel strengths) of the two materials were found to be 7*8 a ^d 11.0 pounds per square inch, respectively. Some of the core material of each kind was made with a bonding agent that could be leached out after the test specimens were made, so that zero bond strength was obtained; that is, the adjacent corrugated sheets of the core material were not bonded to each other at all in the completed sandwich panels . Sandwich panels were made from each of these four core materials (one of each kind of foil, both well-bonded and leached) by bonding aluminum (75S-T6) facings 0.02 inch thick to each side of a sheet of the core material l/2 inch thick. The adhesive used for bonding the core material to the facings or to the test apparatus was not affected by the liquid employed in the leaching process to remove the bonds in the core material. Block Shear Tests of Core Materials Description of Tests and Results The test method used was similar to that described in Forest Products Laboratory Report No. 1555,- paragraphs 36-^0. The specimens were 3 A inch thick, 3 inches wide and 9 inches long. The thickness of the shear plates was 3/8 inch. The loads and displacements during the early part of the test were obtained, and the test was continued until failure occurred, so that the modulus of rigidity and shear strength could both be computed. The results of these tests are given in table 1. The ratios given in table 1 are the ratios of the average values obtained from the unbonded cores to those obtained from the well-bonded cores. k -Methods of Test for Determining Strength Properties of Core Material for Sandwich Construction at Normal Temperatures. Forest Products Laboratory Report No. 1555 . (Revised Oct. 19^3.) Report No. 1835 -3- A nalysis of Kesults The values in table 1 show that the modulus of rigidity of the unbonded core material is less than that of the well-bonded core material. If the Bhear strains are in the LT plane, the ratio of the two moduli is 0.750; and if they are in the LR plane, the ratio is 0.653. The latter reduction is probably due, at least in part, to the reduction of the modulus of rigidity to zero at the locations of the discontinuities in the core material formed by the lack of bond. The former reduction, which is nearly as great as the latter, is not explained. An adequate analysis of the action of a honeycomb core within a sandwich panel has not been made. Table 1 shows that the average shear strength of the unbonded cores was less than that of the bonded cores. If the shear strains are in the LT plane, the ratio of the two strengtlis is 0.661; and if they are in the LR plane, the ratio is 0.732. The latter reduction occurs because of t shear stress concentration in the core due to the discontinuities in the core formed by the lack of bond, as subsequently set forth. The former reduction, which is slightly greater, is not explained. The shear stress concentration in the core in the neighborhood of the unbonded surfaces can be estimated by means of figure 5 of NACA Technical Note 2152.5 For use in this figure: t a 0.575 thickness of shear plate or skin w = 0.75 thickness of core or cap strip L = 0.375 distance between discontinuities — cell size or widtn of cap strip E = 10,000,000 modulus of elasticity of facing or skin G = 22,250 modulus of rigidity of core or cap strip (average from table 1 for the well-bonded cores) Thus * = 0.5 w L/nfe = 0.0331+ (Ewt By using these values in figure 5 of NACA Technical Note 2152,- the stress concentration is found to be 25.7 percent. The ratio of the two strength values is, therefore, — =— or O.79o, which compares favorably with the experimental value of 0.732. - -Shear Stress Distribution Along Glue Line Between Skin and Cap-strip of an Aircraft Wing. National Advisory Committee for Aeronautics Technical Note 2152. Report ho. 1335 -k* If the modulus of rigidity of the unbonded core (l4,500) is used in the same manner, a ratio of 0.798 is obtained. Beam Tests of Sandwich Panels Description of Tests and Results The test method used was similar to that described in Forest Products Laboratory Report No. 1556>- paragraphs 20-24. The specimens were 3 inches wide and tested over a span of 6 inches. All of them failed due to shear in the core. The maximum loads were read and the shear stress at failure was computed. The results of these computations are given in table 2, The ratios given in the table are the ratios of the average values obtained from the unbonded cores to those obtained from the well -bonded cores. Analysis of Results The values in table 2 show that the shear strengths obtained from the unbonded cores were less than those obtained from the bonded cores. The values of the ratios are similar to those obtained from the block shear tests, being 0.69^ when the shear strains are in the LT plane and 0.777 when they are in the LR plane. The shear stress concentration can be determined as before. For this calculation: t = 0.020 thickness of facing or skin w = 0.5 thickness of core or cap strip L = 0.375 distance between discontinuities — cell size or width of cap strip E = 10,000,000 modulus of elasticity of facing or skin G = 22,250 modulus of rigiditjr of ccra or cap strip (average from table 1 for well- bonded cores) Thus * = 0.04 w "Methods for Conducting Mechanical Tests of Sandwich Cons'lsmotion at T Iormal Temperatures. Forest Products Laboratory Report No. 155o\ (Revised Feb. 1950.) Report No. 1835 -5- By using these values in figure 5 of NACA Technical Note 2152,^ the stress concentration is found to be 29.5 percent. The ratio of the two strengths is, therefore, — ± — or O.772, which compares favorably with the experimental 1.295 value of 0.777. If the modulus of rigidity of the unbonded core (14,500) is used in the same manner, a ratio of O.778 is obtained. Fatigue Tests of Core Materials Description of Tests and Results The test specimens were identical to those used in the block shear tests. They were mounted in a fatigue machine similar to that described in Tores t Products Laboratory Report No. 1559.-*- Three groups of tests were made. In the first group the shear strain was in the LT plane and the stress level was 100 pounds per square inch. In the second and third groups the shear strain was in the LR plane. The stress levels in these two groups were 75 and 50 pounds per square inch, respectively. Each cycle consisted of raising the stress from 10 percent of the stress level to the stress level and back to 10 percent of the stress level. The numbers of cycles to failure for each specimen are given in table 5. In this table the averages given are the antilogarithms of the averages of the logarithms of the individual values. Thus these averages agree with the usual way of plotting stress level vs. cycles to failure curves. The stress levels given in percent in table 3 were determined by the use of figure 3 of Forest Products Laboratory Report No. 1559-H.- This curve was obtained from tests on an aluminum honeycomb -core material similar to that used in the present tests. This material has a 3/8-inch cell size and a wall thickness of 0.004 inch and was perforated in the usual way. The strengths given in table 3 were obtained by dividing the stress level in pounds per square inch by the stress level in percent and multiplying by 100. The ratios given are the ratios of the strengths, which, of course, are equal to the ratios of the stress levels in percent. Analysis of Results The strength ratios obtained from the fatigue tests agree reasonably well with those obtained from the block shear tests and the bending tests, as can be seen by comparing the ratios in table 3 with those in tables 1 and 2. 7 . Z. Lewis. Fatigue of Sandwich Constructions for Aircraft. Forest Products Laboratory Report No. 1559. o -Fred V/erren. Fatigue of Sandwich Constructions for Aircraft. Forest Products Laboratory Report No. 1559-H. Report No. 1835 -o- Better agreement is obtained from the groups of fatigue tests that contain four tests than is obtained from the groups having a lesser number. The fact that these ratios do agree indicates that the fatigue strength is not affected by the lack of bonding in the core material other than the effect that might be expected due to the reduction of the shear strength. Fatigue curves in which the stress level is plotted as a ratio of the shear stress to the shear strength may be used for unbonded as veil as for well- bonded honeycomb core materials. The strengths obtained from the fatigue tests are, in general, less than those obtained from the block shear and the bending tests. This may be seen by comparing the strength values in table 3 with those in tables 1 and 2. The difference may be due to the use, in computing the strengths, of the fatigue curve given in Forest Products Laboratory Report No. 1559-H,- which was obtained from a slightly different core material. Compressive Tests of Sandwich Constructions and of Core Materials Descriptions of Tests and Results The sandwich constructions were tested by a method similar to that described in Forest Products Laboratory Eeport No. 1556,2 paragraphs 12-14. The speci- mens were 3 inches square and the load was applied over their facings. The core materials were similarly tested. The specimens were 3 inches square and l/2-inch thick, with the lengths of the cells being orientated in the direction of the l/2-inch dimension. The strengths computed from the values of the maximum loads are given in table k. Analysis of Results VThen a honeycomb -core material is compressed in the direction of the length of the cells, the compression is resisted by the walls of the cells acting as plates subjected to edgewise compression. If the cell walls are thin with respect to their other dimension, their edgewise compressive strengths will be only slightly greater than their critical stresses. If the double walls are not bonded together, the critical load for a single unit of the structure (shown in figure 1) is given by kV where P is the critical load of a single wall. The critical load of a single wall is proportional to the cube of its thickness; thus if two walls are bonded together, the critical load of the double wall is 8P. If the double walls in a single unit of the structure (shown in figure 1) are all bonded together, each unit consists of two single walls and one double wall. Thus the critical load for a single unit of the structure is 2P plus 8P or 10P. The critical load of an unbonded cell is, therefore, roughly O.k of the critical load of a well-bonded cell. Each cell wall is subjected to the same compressive deformation, so that, if it is assumed that each does not greatly exceed its critical load as the deformation is increased, Report No, 1835 -7- the unbonded cells will have about O.k the compressive strengths of the well-bonded cells. This conclusion agrees approximately with the results of tests given in table k for the core materials . The cores of the sandwich panels were probably strengthened by the bonds between the core and facings. Conclusions 1. The tests indicate that the modulus of rigidity of the unbonded core material is approximately 75 percent of that of the well-bonded core material when the shear strain is in the LT plane, and about 65 percent when the shear strain is in the LR plane. This is due to the redistribution of shear strains in the neighborhood of the unbonded joints. The experi- mentally determined reduction of the modulus of rigidity in the LT plane due to the lack of bonding in the core material cannot readily be explained. It may be due to deflections of the unbonded cell walls. 2. The tests indicate that the shear strength of the unbonded core material is about 6Q percent of that of the well-bonded core material when the shear strain is in the LT plane, and about 75 percent when the shear strain is in the LR plane. The latter reduction can be quantitatively accounted for by consideration of the stress concentrations due to the dis- continuities in the core matsrial created by the unbonded joints. The former reduction is not understood. 5. The results of the shear fatigue tests of both the unbonded and the well-bonded core materials are substantially consistent with the results of the static shear-strength tests. k. The tests indicate that the compressive strength of the unbonded core material is about 55 percent of that of the well-bonded core material. This is substantially consistent with the assumption that the compressive strengths of the cell walls are proportional to their critical stresses. Report ; r o. 1855 -6- Table 1. — Moduli of rigidity and shear strengths obtained from block shear tests Core material A Core material B Well bonded Unbonded I7ell bonded Unbonded Modulus of rigidity Shear strength Modulus : Shear of : strength rigidity: Modulus : Shear of : strength rigidity: Modulus of rigidity Shear strength P.s.i. P.s.i. P.s.i. : P.s.i. P«B.i, : P.s.i. P.s.i. P.s.i. 68,700 258 69,400 273 70,000 269 Av...69,400 267 Shear strain in LT plane 46,1+00 174 61,1+00 200 1+7,700 11+1+ 1+8,700 172 68,1+00 221 1+6,1+00 • 11+8 58,600 165 63,100 210 52,600 139 51,200 170 61+, 300 210 1+8,900 11+1+ 0.738 0.636 • ••••••• ........ 0.761 0.686 She? ir strain in LR plane 25,000 : 131+ ii+,6oo 99 22,200 103 : 14,800 75 20,600 : 135 15,150 96 21,800 104 : 13,200 82 19,200 133 15,250 101 2l+,800 110 13,900 76 Av...21,600 : 131+ 15,000 99 22,900 106 : 14,000 77 0.691+ 1 0.739 0.612 0.726 Report No. 1835 Table 2. — Shear strength values of core material obtained from beam tests of sanav/lch construction Core material A Core material B Well bonded Unbonded Well bonded Unbonded P.s.i. P.s.i. P.s.i. P.s.i. Av Patios 289 300 293 ,294 Shear strain in LT plane 205 : 25^ 221 : 251 195 237 207 : 2i*7 Shear strain in LP plane 162 167 177 169 0.60+ 152 121+ 120 93 154 116 : 123 92 1^7 118 117 92 Av ..151 119 120 92 O.788 O.766 Report No. 1835 Table 3. — Cycles to failure obtained from shear fatigue tests of core material Core material A Core material B Well bonded : Unbonded Well bonded Unbonded Shear strain in LT plan e Stress level ICO pounds per square inch 1,311,000 789, 000 968,000 2, 107, 000 Av 1, 205, 000 Stress level (percent).... 35 Strength (p.s.i.) 286 Ratio 83, 000 70,000 328, 000 96,000 54,ooo 35^,000 219, 000 566,000 116,300 197,850 51 h9 196 204 0.686 Shear strain in LE plane Stress level 75 pounds per square inch 250,000 457,000 81, COO 174,000 Av 198,100 Stress level (percent) G6 Strength (p.s.i.) 113 Eatio 24,000 12,000 23, 000 16, 000 : 30,000 : 27,000 : 31, 000 : 12, 000 18, o4o 23,^30 82 8O.5 91.5 93.1 0.810 Shear strain in LE plan e Stress level 50 pounds per square inch 10,000,000 : 553.000 Stress level (percent) 38 : 57 Strength (p.s.i.) 132 : 87.7 Eatio i 0.664 116, 000 3,000 18, 660 69 145 0.711 2,000 2,000 90 85.4 O.896 1,335,000 : 87, 000 hi : 72 106 : 69.5 # • 0.656 Eeport No. I835 Table 4. -- Compressive strengths obtained from tests made in the L direction Core material A Core material E Well bonded Unbonded Well bonded Unbonded P.-s.i. P.s.i. P.s.i, P.s.i. Sandwich panels 582 330 kkl : 313 51+0 325 kko : 299 TA 350 1*7 : 269 ..552 337 kkl 295 0.611 Core ma : 0.666 ter: Lais iil8 187 357 128 393 : 205 355 Ikk H52 199 361 153 197 353 lAl 0.467 0.39k Eeport No. 1835 H OS •H U Q) I 1 O l>> V § o § •H -P o m 03 GO O U O o .a O •P .8 °? i •H oo CO UNIVERSITY OF FlORlDA MllllillWllllllWllllllill 3 1262 08928 5828