Openings in Sandwich Elements

Transcription

1 Openings in Sandwich Elements Warmuth, F. Institut für Stahlbau und Werkstoffmechanik, Technische Universität Darmstadt ( Lange, J. Institut für Stahlbau und Werkstoffmechanik, Technische Universität Darmstadt ( Abstract Sandwich panels with flat or lightly profiled faces often form the façade of a building. Due to different requirements like windows, doors or ducts, it is necessary to cut openings into the face of a building. At present, openings in sandwich panels require an additional substructure, which transfers the loads to the main structure. The aim of the project is to find a possibility to avoid this kind of substructure. Basically, there are two possibilities: For small openings in a panel, it seems to be obvious to look at the single element only. If the panel is not able to carry the load, it has to be strengthened. Therefore one solution could be a supporting frame, which is integrated in the opening and diverts the loads around it. Also additional profiles integrated in the joints - could be used to strengthen the element stiffness of the panel and to reduce the stress concentration in the corners of the openings. For larger openings it is necessary to include the neighbouring elements in the load transfer. Therefore, detailed information on the rigidity of the longitudinal joint and the torsional rigidity of the elements is needed. A calculation model shall be developed to describe the load transfer. First test results for a single element and an element formation with window openings are presented. These tests are part of the European research project EASIE ( In addition, comparative calculations with existing calculation proposals for openings without window frames are shown. The long-term objective is to generate a calculation model for sandwich elements with different kind of openings. This model should be only dependent on parameters which can be derived from some basic tests. The EASIE project has received financial support from the European Community s Seventh Framework Programme FP7/NMP2-SE-2008 under grant agreement No Keywords: sandwich, openings, load transfer, substructure, strengthening 61

2 1. Introduction Sandwich panels often form the outer shell of buildings. As façade elements they perform different tasks at the same time. Sandwich panels do not only seal the building, they also transfer the loads to any kind of substructure and form a very good thermal insulation.sandwich elements consist of two thin covering sheets, which enclose the core material. The sheets are usually made of galvanized steel. As core material, different materials with good insulating properties are used. The most common core materials are polyurethane (PUR) foam and mineral wool. Depending on the estimated use, the panels have thicknesses between 40 and 200 mm and a length up to 24 m. Due to different requirements like windows, doors or ducts, it is necessary to cut openings into the face of a building. These openings always result in an attenuation of the sandwich panel. At present, openings in sandwich panels require a substructure, which transfers the loads to the main structure. These replacements lead to an additional effort of erection and are visually not attractive. The aim of the presented project is to find a possibility to avoid this kind of substructure. In this paper, first test results of sandwich elements with windows are presented. Two different kinds of openings have been tested. In the first case, the opening has been cut in one panel and the window frame has been bonded in the opening. In the second case, the width of a complete panel has been replaced by a window. The load had to be transferred to the neighbouring elements. 2. State of the art / theory Basically, it can be distinguished between different kinds of openings. The most important differentiation is the size of an opening. We can distinguish between little openings like penetrations for supply lines and large openings for windows and doors. A second criterion is the location of the opening. From central openings in one panel to openings across longitudinal joints everything is possible. In recent years, some research projects on openings in sandwich panels were conducted. Especially for openings within one sandwich panel calculation proposals exist. In 1998 Courage/Toma (1998) published a possibility to calculate the stress peaks around openings and the consequences on the bearing capacity of the sandwich panel. Proposal Courage/Toma: If 0,1 0, 8 : In which σ K = wrinkling stress of the panel without opening σ N = maximum stress in the residual face sheet of the cross-section, referring to the complete panel β = b / B b = width of the opening B = width of the panel 62

3 Marc Böttcher (2005) proposed a more simple approach, in which he also included openings, which are not arranged at midspan. Proposal Böttcher: If 0,0 < β < 0,4 : σ N (1- β) σ K If 0,4 < β < 0,8 : σ N 0,6 σ K In both publications, the wrinkling stress is reduced depending on the relation between the width of the opening and the width of the panel. Depending on the size of the opening, the results of the two calculation models are more or less similar. For little openings, these calculation models make sense. For larger openings, the formulas given above lead to a very high loss of bearing capacity. Interaction with neighbouring panels For large openings it is not useful, to confine the load transfer only to the weakened element. Especially at openings with the width of one panel (complete cut-out) there has to be an interaction with the neighbouring panels. The loads have to be transferred over the longitudinal joint and the unweakened neighbouring panels can participate in the load transfer. The German association for light metal structures IFBS (2006) published a paper, how to calculate these kind of openings. In this calculation model it is assumed, that the load of the opening directly affects a torsional moment in the neighbouring element. For the verification of the system, the bearing capacity of the neighboured element and of the joint has to be proved.in all the publications mentioned above, only openings without any kind of strengthening were discussed. If windows are built into a sandwich façade, there exist always window frames, which almost certainly have a stiffening effect on the whole panel and especially on the covering sheet in the region of the window corners. That probably improves the load transfer. To check this assumption, panels with window frames have been tested. 3. Experimental investigations 3.1 Window-openings within one sandwich panel Test-set-up In a first test series sandwich panels with a window at midspan have been tested in a six point bending test. The width of the panels was 1 m, the window opening in each test 70 x 70 cm. Two different thicknesses of panels were tested. Four panels had a thickness of 60 mm and a total length of 4000 mm (test 1 to 4), two panels had a thickness of 120 mm and a total length of 6000 mm (test 5 and 6). In figure 1 the test set up is shown. Half of the panels were tested in a positive position (equates to wind pressure), the other half in a negative position (equates to wind suction). The important difference between these tests is the geometry of the window frame, which has some influence on the failure criteria. 63

4 Figure 1: Experimental set-up for test series 1 The deflections were measured at 5 different locations. Three displacement transducers were fixed at midspan (in the middle and left and right of the window frame), two at the corner of the window frame. In some of the tests, additional strain gauges have been applied around the window corner to get detailed information about the load transfer around the opening. Displacements Results Load-displacement diagrams registered only very little differences of the displacement at all measured points. One of the load-displacement diagrams is exemplary shown in figure 2. The uniform displacement shows the stiffening effect of the window frame, which avoid different deflections around midspan. The displacement was nearly linear until the failure of the panel. 3,5 load of the hydraulic cylinder in kn 3 2,5 2 1,5 1 0,5 0 midspan 1 midspan 2 w indow 1 w indow deflection at midspan ( x = 0 mm) and in the corner of the w indow ( x = 400 mm ) in mm Figure 2: Load-displacement-diagram of one test at midspan and in the corners of the window 64

5 Comparing the displacements of the test panels with the calculated displacements of an unweakened panel result in the following table: Table 1: Comparison of deflections with and without window frame Load level F = 1 kn x = 0 mm, deflection in mm Test Calculation Deviation without opening in % x = 420 mm, deflection in mm Test Calculation Deviation without opening in % 1 12,36 6,96 77,6 12,04 6,7 79,7 2 12,67 6,96 82,0 12,43 6,7 85,5 3 12,49 6,96 79,5 12,39 6,7 84,9 4 12,83 6,96 84,3 12,79 6,7 90,9 Load level max. load x = 0 mm, deflection in mm Test Calculation Deviation without opening in % x = 420 mm, deflection in mm Test Calculation Deviation without opening in % 1 35, ,0 35,1 19,3 81,9 2 33,65 18,7 79,9 33, ,7 3 37,81 21,1 79,2 37,54 20,3 84,9 4 33,82 18,9 78,9 33,52 18,2 84,2 Mode and level of failure: All the panels failed by wrinkling in the corner region of the opening. Depending on the geometry of the window frame the location of wrinkling was a bit different (see figure 3). The very stiff frame leads to wrinkling directly at the frame corner, where the panel is still complete over the whole width. In the case of the less stiff frame the face sheet wrinkles close to the corner of the opening in the panel. For the same panel thicknesses, the load levels by failure are quite similar for each panel. Figure 3: Different locations of wrinkling 65

6 Table 2: Comparison of failure loads with and without opening Test Thickness of the panel Failure load in kn Calculated failure load of the complete panel in kn Deviation in % ,44 10,81 68, ,25 10,81 69, ,6 7,94 54, ,27 7,94 58, ,80 18,99 64, ,85 13,06 62,83 Allocation of stresses: Strain gauges have been applied in the region around the window corner to get detailed information about the load transfer around the opening. Exemplary one test-set- up is shown in figure 5. Allocation of stress for different load levels (opening 700 mm) 0,00-50,00 Stress in N/mm² -100,00-150,00-200,00 1 kn 3 kn 5 kn 5,8 kn -250,00-300, distance from the longitudinal axis in mm Figure 4: Allocation of stress for different load levels Figure 5: Strain gauge position 66

7 In figure 4 the allocation of stresses along the width of the panel directly in front of the window frame is shown for different load levels. In the diagram, only half of the panel width is presented. Two interesting things can be pointed out. 1) The openings with a size of 70 cm and the window frame of 80 cm lead to explicit stress peaks in the region around the window corner. With increasing loads, the peaks become more definite. 2) In the middle of the panel the stress is increasing with higher loads, but has still a very low level. That means, in spite of the high stiffness of the window frame, the main part of the load is transferred through the covering sheet to the outer continuous part of the panel. 3.2 Window-opening in a panel interconnection system Test-set-up In a second test series an interconnection system with 3 sandwich panels and a window in the middle panel has been tested. The width of the panels was 1m, the window opening was a complete panel width (1 m) and 1 m length. Like in test series 1, two different thicknesses of panels were tested in a positive and a negative position. In figure 6 the test set up is shown. Figure 6: Experimental set-up for test series 2 Displacements Results Whole systems with windows lead to much higher deflections than sandwich panels without openings or windows. For the tests with 60 mm thick panels a comparison of the calculated displacements without window and the measured displacements with window is shown in figure 8. In some tests, 67

8 particularly the negative position, the window frame slipped out of the system. That explains the different deflections between neighbouring points of measuring in figure 7. Deflection in longitudinal direction ,1 0,2 0,3 0,4 0,5 Deflection in mm w ithout w indow 60_1 60_2 60_3 60_4 Length / Total length Deflection in lateral direction in x=0 (midspan) 0 Deflection in mm w ithout w indow 60_1 60_2 60_3 60_ Distance from the longitudinal axis in mm Figure 7: Deflections for the load level 4 kn, 60 mm panels and a span of 3900 mm Mode and level of failure: All the panels failed by an overstressing of the longitudinal joint and a following collapse of the complete system including wrinkling of the outer panels. In some cases the window frames slipped out of the joint. The slipping of the frames did not lead to an immediate collapse of the system, but certainly had an influence on the failure load. As the slipping is a question of an exact fabrication of 68

9 the window it can be avoided. Probably according to the inexactnesses of the windows sizes, the failure load levels were different for each system. Figure 7: Collapse of the system Table 3: Comparison of failure loads with and without window frame Test Thickness of the panel Failure load in kn Calculated failure load of the complete panel in kn Deviation in % Mode of failure ,30 15,94 41,66 Complete collapse ,5 15,94 46,68 Complete collapse ,7 12,91 48,10 Slipping out of the window, then collapse ,5 12,91 65,14 Complete collapse without load on the middle panel ,9 22,59 47,32 Complete collapse ,0 18,05 55,68 Slipping out of the window, then collapse 69

10 Allocation of stresses: Strain gauges have been applied to get detailed information about the load transfer around the window. Exemplary one test-set- up is shown in figure 8. Figure 8: strain gauge position Longitudinal stresses cross the panels Stresses in N/mm² _1 60_2 60_3 60_4-140 Distance from the longitudinal axis in mm Figure 9: Allocation of stress for load level 4 kn In figure 9 the allocation of stresses along the width of the panels directly in front of the window frame is shown for the load level 4 kn. In the diagram, only half of the test set-up width is presented. In each test, the characteristics of the stresses are similar. In the medial panel the stresses are nearly zero, all the load is transferred to the outer panels. There, a concentration of stress can be identified in the region close to the longitudinal joint. 70

11 4. Comparison with existing methods of calculation 4.1 Evaluation of the tests with one panel As described in chapter 2, different calculation proposals exist for single sandwich panels with openings. In table 4, the failure loads of the tests are compared to calculation results for openings without frames according to calculation proposals of Böttcher (2006) and Courage/Toma (1998). Table 4: Comparison of the failure load in kn between test and calculation Failure load in kn Tests Böttcher Deviation in % Courage/Toma Deviation in % 60_1 3,44 1, , _2 3,25 1, , _3 3,60 1, , _4 3,27 1, , _1 6,80 3, , _2 4,85 2, ,87 69 The comparison shows, that window frames have a positive effect on the load bearing capacity. The panels carry up to 155 % more load than the calculation results without window-frames suggest. 4.2 Evaluation of the tests with three panels In IFBS (2006) a proposal exists for calculating a 3 panel system with an opening. According to this proposal, there are three possible failure modes: Shear failure in the outer panel, wrinkling of the outer panel and failure of the longitudinal joint. By recalculating the tests according to the IFBS paper, one asserts that the longitudinal joint is always the critical failure mode. In table 5 the degree of capacity utilization for the different failure modes is shown for a load level of 4 kn per panel. Table 5: Degree of capacity utilization for different failure modes Load level: 4 kn Shear failure Wrinkling Failure of the joint 60 mm panel 0,31 0,38 0, mm panel 0,18 0,26 0,53 For the bearing strength of the joint, in IFBS (2006) only a rough estimate is given. That could be the reason for significant deviations between some of the test results and the recalculation, like shown in table 6. 71

12 Table 6: Comparison between tests and calculations Failure load per panel in kn 60_1 60_2 60_3 60_4 120_1 120_2 Test result 9,3 8,5 6,7 4,5 11,9 8 Calculation result 10,0 10,0 7,9 7,9 7,5 6,0 Deviation -7 % -15 % -16 % - 40 % + 59 % + 33 % 5. Conclusions Test results of sandwich elements with windows have been presented in this paper. Two different kinds of openings were tested. The results of test series 1 show, that window frames have a positive effect on the load bearing capacity of single sandwich elements with openings. The failure load can be clearly increased in comparison to panels with frameless openings. Nevertheless, for larger openings the loss of bearing capacity is very high. In the second series, the width of a complete panel has been replaced by a window. The entire load was transferred to the outer panels. The maximum loads in the second series varied within a wide range. The main reason for that is evidently the slipping of the window frames out of the joints. An interesting point is the bearing capacity of the longitudinal joints. Further tests will show, if a more exact knowledge about the bearing capacity of the joint bring better correlations between the tests and the calculations. At the end of the project, there shall exist a calculation model for sandwich elements with different kind of openings. This model should be only dependent on parameters which can be derived from some basic tests. References Böttcher M. (2005) Wand-Sandwichelemente mit Öffnungen, Publication of the institute for steel construction and material mechanics, TU Darmstadt Courage, Toma (1998) Structural detailing of opening in sandwich panels, Publication of the European Communities IFBS (2006) Berechnungsverfahren für Wand-Sandwichelmente mit Öffnungen, Publication 5.09 of the IFBS, Düsseldorf 72