- Email: [email protected]
New peel stopper concept for sandwich structures
Available online at www.sciencedirect.com
COMPOSITES SCIENCE AND TECHNOLOGY Composites Science and Technology 67 (2007) 3378–3385 www.elsevier.com/locate/compscitech
New peel stopper concept for sandwich structures J. Jakobsen *, E. Bozhevolnaya, O.T. Thomsen Department of Mechanical Engineering, Aalborg University, Pontoppidanstraede 101, 9220 Aalborg East, Denmark Received 3 January 2007; received in revised form 6 March 2007; accepted 18 March 2007 Available online 12 April 2007
Abstract The paper addresses the damage tolerance of sandwich structures, where the prevention and limitation of delamination failure are highly important design issues. Due to the layered composition of sandwich structures, face–core interface delamination is a commonly observed failure mode, often referred to as peeling failure. Peeling between the sandwich face sheets and the core material drastically diminishes the structural integrity of the structure. This paper presents a new peel stopper concept for sandwich structures. Its purpose is to effectively stop the development of debonding/delamination by rerouting the delamination, and to confine it to a predefined zone in the sandwich structure. The suggested design was experimentally tested for different material compositions of sandwich beams subjected to three-point bending loading. For all the tested sandwich configurations the suggested peel stopper was able to stop face–core delamination and to limit the delamination damage to restricted zones. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: B. Fracture; B. Interface; C. Crack; C. Delamination; C. Sandwich
1. Introduction Sandwich materials are layered structural components composed of thin strong face layers separated and bonded to light weight core materials. This particular layered composition creates a structural element with a very high bending stiffness to weight ratio as well as bending strength to weight ratio. Sandwich structures are often utilized in the marine, aerospace, train and automotive industries, where low weight is a critical design parameter. Furthermore, large parts of wind turbine blades are made using composite sandwich materials. The general concept of sandwich structures has been investigated and developed by many researchers over the past 50 years, see for example Zenkert [1,2], Allen [3] and Gay [4]. It is well known that sandwich structures may suffer from sudden failure if the allowable design loads are exceeded. When structures are made of ductile metallic materials, e.g., steel or aluminium, usually they do not fail *
Corresponding author. Tel.: +45 9635 9322; fax +45 9815 1675. E-mail address: [email protected] (J. Jakobsen).
0266-3538/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.compscitech.2007.03.033
catastrophically in a sudden way, due to their ability to redistribute the loads by plastic yielding. As opposed to this, composite and sandwich structures often exhibit a more brittle behaviour, and this, together with their inherent layered composition may lead to a sudden and fast failure by delamination (peeling), which occurs without any prior warning. Therefore, the study of initiation, propagation and arrest of delamination failure are issues of extreme importance for modern sandwich structures. Failure phenomena related to sandwich structures have been studied intensely by various researchers. Zenkert et al. have studied failure in foam cored sandwich panels with and without initial debonds [5–8]. They examined sandwich beams subjected to three-point bending loading under quasi-static and fatigue loading conditions. For the majority of the tested configurations face–core delamination occurred as failure mode. Moreover, in some cases failure initiated in the centre of the core and then kinked toward the face sheets and continued as a delamination along the face–core interfaces. Failure phenomena in general, and crack tip propagation in particular, in sandwich structures under in-plane
J. Jakobsen et al. / Composites Science and Technology 67 (2007) 3378–3385
compression loading have been studied by Carlsson et al. [9–11] among others. Sandwich elements with initial debond imperfections were experimentally studied, and it was concluded that the initial debonds would grow when the load applied to the test specimens exceeded the buckling load of the face sheet (i.e., wrinkling of the sandwich faces). Furthermore, their results stated that larger initial debonds would lower the total buckling strength of the sandwich panel, and that the geometrical shape of the debonded area seems also to have a significant influence on load bearing ability of the sandwich panel. Bozhevolnaya et al. [12–16] investigated the influence of core junctions on the static and fatigue strength of sandwich beams. It was found that modifications of the geometric shape of a core junction can have a considerable effect on the fatigue life of sandwich structures. Additionally, similar to the studies [5–8] for the case of static and fatigue loading, it was observed that face–core delamination very often was the dominating failure mode, which usually followed shear cracking of the compliant core in the sandwich beam. Core junction failure was further studied for sandwich structures subjected to in-plane tensile loading [17], and it was concluded that failure generally initiated in the vicinity of the core junction, and that final failure occurred as tensile face failure. According to the previous works cited above, a commonly observed failure phenomenon for sandwich structures is face–core delamination, which usually follows various types of locally induced damages in the sandwich core and/or in the vicinity of sandwich sub-structures like core joints, inserts or edge stiffeners. This failure mode is also referred to as face sheet peeling. Several techniques for improving the peeling strength of face-core interfaces are known today. For example, Grenestedt [18] suggested a new peel stopping manufacturing technique. The basic principle of this technique is that the debonded face sheets are able to be separated from the structure in order to arrest delamination growth beyond the implemented peel stoppers [19]. This method also implies that the sandwich structure looses a huge part of its overall bending stiffness and in-plane tensile strength, as part of the face sheets simply peels of. Accordingly, the advantages of the structural sandwich concept no longer exist or are significantly diminished. Another approach to arrest face sheet delamination is to stitch the face sheets together [20–22]. This method creates a sandwich panel with extraordinary high transverse stiffness and strength. Moreover, the shear properties can also be improved if the stitching angle is inclined 45° with respect to the normal plane of the sandwich panel. The method is very effective when applying it to monolithic composite laminates, but rather tedious and design restrictive in sandwich applications. Moreover, manufacturing complexities and costs are often increased with this method. However, the commercially available sandwich systems X-corTM and K-corTM of Albany Engineered Composites [29] belong to this type.
3379
The research presented in this paper concerns the proposal of a new peel stopper design, which effectively prevents peeling of the faces, and limits debonding/ delamination to a priori restricted areas of the sandwich component. The design of the new peel stopper is described, and the choice of appropriate peel stopper material is substantiated. The efficiency of the new peel stopper concept is validated by experimental tests with sandwich beams subjected to three-point bending up to the failure. 2. Peel stopper concept The basic design of the suggested peel stopper is illustrated in Fig. 1. The peel stopper is a sub-structural component embedded into the sandwich panel (like an insert or edge stiffener), and its main purpose is to arrest face–core interface crack propagation by rerouting the crack path into a closed/restricted area of the sandwich panel, thus preventing the spreading of debonding/delamination into the remaining parts of the sandwich structure. For the present study, the peel stoppers were manufactured from an elasto-plastic material, with elastic properties close to the sandwich core properties. Generally, it is recommended that the material of the peel stopper is chosen to be compliant and with large straining capability (i.e., ductile), and the elastic stiffness of the peel stopper is recommended to be of the same order as the elastic properties of the main sandwich core (or somewhat higher). Good adhesion properties with respect to both the core and the faces are required as well. The peel stoppers may be mounted into a sandwich panel (e.g., a sandwich beam, plate or shell) together with other sub-structural components like e.g., structural inserts as shown in Fig. 1. Essentially, there are no or only minor
Fig. 1. Proposed design of the peel stopper (a). The case shown displays a crack rerouting angle of 10°. A suggested implementation of the proposed peel stoppers in a sandwich plate is shown in (b). The grid type pattern will confine damage to the grid mesh.
3380
J. Jakobsen et al. / Composites Science and Technology 67 (2007) 3378–3385
Fig. 2. The basic idea of the peel stopper is to force the crack to propagate along the stopper–core interface (internal curve – arrowed line) and not along the face–core interface (dashed line).
manufacturing difficulties associated with the introduction of the peel stoppers, since they are similar to insert types already widely used in sandwich structures. Failure is often initiated by crack formation in the interior core parts of the sandwich structure due to fatigue load conditions, impact/shock (dynamic) loads or from manufacturing imperfections. Under such circumstances the core crack will often propagate towards a core–face interface, from where it usually proceeds as a delamination along the face–core interface. The principle idea of the new peel stopper concept is that propagation of delamination/debonding is prevented to spread beyond the boundaries of the proposed peel stoppers. This inhibition is due to a rerouting of the crack along the internal curve of the peel stopper (arrowed line) instead of propagation along the interface (dashed line), as shown schematically in Fig. 2. This technique will confine the debonding/delamination to a limited area of a single grid of the peel stoppers (cf. Fig. 1), and thereby prevent the spreading of damage further into the plate, thus averting its total collapse. Furthermore, the proposed peel stopper will allow the debonded sandwich face be kept attached to the sandwich component (contrary to the method described in [18,19]). This will retain some structural load carrying capability of the debonded structure after partial delamination/debonding under all types of loadings. 3. Test specimens Verification and functionality tests of the proposed peel stopper design were performed by means of comparing three sandwich beam test configurations, denoted as (a)– (c) in Fig. 3. These beams were manufactured with PMI RohacellÒ cores. Furthermore, an additional test configuration with PVC DIABÒ cores (denoted as (d) in Fig. 3) was made in order to validate the concept with respect to other material compositions. For each configuration (a–c), two specimens were manufactured and tested quasi-statically in three-point bending loading, as illustrated schematically in Fig. 3. The main purpose of the experiments was to induce a shear failure in the softer core of the sandwich beam, followed by crack propagation and crack-kinking towards the face–core interfaces and finally delamination along the interfaces. This allowed a detailed
Fig. 3. Three test configurations (a, b, and c) with RohacellÒ foams and carbon fibre reinforced composite faces for the quasi-static validation of the peel stopper concept. Test configuration (d) with DIABÒfoams and aluminium faces for the quasi-static validation of the peel stopper concept.
study of the influence of the presence of the peel stoppers on the propagation of completely developed delaminations. Each test specimen had a total length of 500 mm and a beam span between the supports of 460 mm. The beam configurations (a)–(c) were manufactured with a 1 mm thick carbon fibre laminate face sheets. The lay up was (0°, 90°, 0°) of both top and bottom faces. The core of these specimens consisted of two 25 mm thick PMI foam core parts from RohacellÒ with different densities (51WF and 200WF). The stiffer core, 200WF, was located at the edges of the beams, and an araldite diaphragm was placed in the beam centre to avoid indentation failure due to the external loading. In beam configurations (b) and (c), polyurethane inserts were embedded between the two cores as shown in Fig. 3. A conventional butt insert was used in configuration (b), and the proposed peel stopper was used in configuration (c). The material data are specified in Table 2. Configurations (b) and (c) were designed to have equal mass and identical material composition. Configuration (a), which represents a realistic design configuration, was considered as a reference to evaluate the two other configurations against. The discrepancy between the average core mass of configurations (b) and (c) is less than 4% of the total mass of the beam (cf. Table 1). This difference may Table 1 Mass measures of the tested configurations (Fig. 3) Specimen
Avg. core mass [g]
Avg. of total mass [g]
b c
143 152
223.5 232.5
J. Jakobsen et al. / Composites Science and Technology 67 (2007) 3378–3385
3381
Table 2 Mechanical properties of the tested beams Materials
E-modulus [MPa]
Tensile strength [MPa]
Compress. strength [MPa]
Elongation at failure [%]
Test configurations (a)–(c) T700 UD/SE 84LV [23] – face RohacellÒ 200WF [24] – edge core RohacellÒ 51WF [24] – main core
129200 350 75
2844 6.8 1.6
1187 9.0 0.8
–
Test configurations (d) AL 7075-T6 [17] – face DivinycellÒ H200 [28] – edge core DivinycellÒ H60 [28] – main core
71000 250 75
503 7.1 1.8
503 4.8 0.9
Test configurations (b)–(d) PERMAlock 40496 (PU) [25] – peel stopper
100
All test configurations Araldite 2011 [27] – adhesive Araldite B30 [26] – diaphragm
– 3700
10
–
Shear lap strength 26 MPa 60 100
be caused by a larger adhesive surface area in configuration (c). Every test configuration shown in Fig. 3 was manufactured by assembling and bonding the core prior to prepreg/face lamination using a vacuum bagging technique. The sandwich panel was cured for 6 h at 100 °C, and afterwards post-cured for 48 h at room temperature. Finally, the sandwich panel was cut into separate beams with a final width of 58 mm. A fourth configuration (d) (Fig. 3) was manufactured in order to examine if the concept with a polyurethane peel stopper exhibits the same effect with other face and core material compositions. Configuration (d) consisted of 1 mm thick Aluminium (AL 7075-T6) face sheets and 25 mm thick DivinycellÒ foam cores (H200 and H60). Moreover, as for configurations (a)–(c), a diaphragm was placed in the centre of the beam to introduce the central local load. Configuration (d) was not manufactured using the vacuum bagging technique. Instead the sandwich core components were glued together with an AralditeÒ 2011 adhesive and cured at room temperature for 24 h. After this the aluminium face sheets were bonded to the assembled core using the same adhesive under controlled pressure condition followed by curing for another 24 h. All the tested configurations (a)–(d) were geometrically similar, and the only difference was the choice of material composition. The mechanical properties of the sandwich beams constituents are given in Table 2.
3.5 3.0 11 4.5 [17] 5.5 [17] 25
5–6
tal testing of the sandwich beams, since it provided a controlled shear cracking of the core in the bulk of the weaker foams. The shear cracks propagated towards the face sheets, where crack kinking occurred followed by crack propagation in a delamination mode along the face–core interfaces. An experimental set-up was designed and manufactured on the basis of a 100 kN servo hydraulic Schenk HydropulsÒ testing machine. The testing machine has four test regions, where the lowest region was used for this particular test setup. The upper load limit in this test region is 12.5 kN, which gave a discrepancy between the input and output load signal of less than 3%. A load controlled mode was used during loading, which was performed with a load rate of 0.02 kN/s. The central deflection of all specimens was recorded via the displacement of the cross head. High-speed video recording of specimen failure was enabled with a frame rate of 6000 frames/s. The observed flexural load vs. central deflection responses of configurations (a)–(c) shown in Fig. 4 were very similar as expected. This is an indication that the overall structural stiffness of the sandwich beam was not affected by the introduction of the peel stoppers.
4. Test results It is known that sandwich structures are particularly vulnerable to transverse shear loading. At the same time this type of loading occurs very often in practice, where it may (and often does) provoke an initial shear failure of the sandwich core, which is then inevitably followed by a delamination of the sandwich core and faces. Therefore the three-point bending scheme was chosen for experimen-
Fig. 4. Applied force vs. central displacement for the specimens of configurations (a)–(d). Two specimens of each configuration were tested.
3382
J. Jakobsen et al. / Composites Science and Technology 67 (2007) 3378–3385
The failure characteristics of all six test specimens are summarised in Table 3. The recorded maximum loads and maximum central deflections of the beams at failure were quite close for three different configurations. The average failure load measured for the specimens of configuration (c) was 2216 N. This load level is within 5% of the average failure loads of configurations (a) and (b). Additionally, the central deflection at failure for configuration (c) is observed to be between the central deflections observed for configurations (a) and (b). This difference is estimated to be around 9%. In this connection it should
be mentioned, that the short length of the weak core compared to the total length of the edge stiffeners and peel stoppers is the reason for the difference in central deflections observed for the improved (c) and conventional (a), (b) edge stiffeners. If the length of the pure core part of the sandwich structure was larger, compared to the total length of the embedded sub-structures, which would be the case for realistic design configurations, the difference in the ultimate displacements, and thus the influence of the peel stoppers on the overall structural stiffness of the sandwich, would be much smaller.
Table 3 Failure characteristics of the test specimens Specimen and configuration
Failure load [N]
Avg. failure load [N]
Cross head displacement at failure [mm]
Avg. cross head displacement at failure [mm]
Location of failure initiation
Completed delamination
a1
2307
2320
11.58
11.78
core
Yes
a2
2332
core
Yes
b1
2094
core
Yes
b2
2106
core
Yes
c1
2185
core
No
c2
2246
core
No
d
1868
Compliant (51WF) Compliant (51WF) Compliant (51WF) Compliant (51WF) Compliant (51WF) Compliant (51WF) Compliant (H60)
core
No
11.97 2100
10.11
9.96
9.80 2216
10.43
10.83
11.22 –
18.77
–
Fig. 5. Failure of test specimen a1 and b2 (cf. Fig. 3 and Table 3).
J. Jakobsen et al. / Composites Science and Technology 67 (2007) 3378–3385
Regarding configuration (d), one specimen was tested under the same load conditions as specimens (a)–(c). The load vs. cross head displacement curve is shown in Fig. 4. Notice that the maximum critical load for configuration (d) is close to those previously measured and shown in Table 3. This is explained by the fact that the strength of the weaker core is a limiting factor in the beam design, and the strength of RohacellÒ 51WF is close to the strength of the DivinycellÒ H60 (cf. Table 2). Damage initiation and development of failure occurred according to the predicted scenario, as shown in Figs. 5 and 6. High speed video recordings were used to identify the location of failure initiation and its progression in the sandwich specimens. In all cases, failure initiated as a shear crack in the centre of the weak core, the crack tip kinked towards the faces and continued as a delamination along face–core interfaces. Notice that a full delamination of the face occurs for the cases of conventional edge stiffeners (Fig. 5), while the peel stopper in Fig. 6 clearly confines the crack inside the weak core, and, moreover, lets the sandwich face still be attached to the sandwich beam edge. The sandwich beams with embedded peel stoppers were subsequently loaded in the three-point bending fixture in order to inspect the damage zone in the vicinity of the peel stoppers as seen in Fig. 7. All beams of configurations (a) and (b) ended up with completely delaminated face sheets, while peel stoppers in configuration (c) effectively stopped delamination. A failure scenario similar to the observations for configuration (c) was observed for configuration (d), as illustrated by the high-speed recording shown in Fig. 6. Again, the peel stopper confirmed its efficiency in rerouting
3383
Fig. 7. Post-mortem inspection of the cracked part of test specimen c2 (cf. Fig. 6).
the crack propagation, and by stopping the crack propagation beyond the bounds of the area marked by the peel stoppers (cf. Fig. 8).
Fig. 6. Failure of test specimen c2 and d (cf. Fig. 3 and Table 3).
3384
J. Jakobsen et al. / Composites Science and Technology 67 (2007) 3378–3385
three-point bending. To validate the functionality of the peel stopper concept, three test configurations were examined: one without any peel stopping reinforcement, another one with a conventional butt-reinforcement and the third configuration with the proposed peel stoppers. Different material constituents were used in the test specimens with embedded peel stoppers. Only the configurations with the embedded peel stoppers were able to arrest the development of debonding/delamination beyond the bounds of the peel stoppers, and thereby prevent total delamination of the sandwich faces. Acknowledgements
Fig. 8. Post-mortem inspection of the cracked part of test specimen d (cf. Fig. 6).
A post-mortem inspection of the damaged beams has demonstrated that in all cases the beams containing the peel stoppers retained at least 10% of their initial bending stiffness, while the conventional beams retained none. The use of peel stoppers or other crack stopping devices are usually expected to lead to a higher ultimate load carrying capability (strength) of sandwich structures. However, the ‘‘one-dimensional’’ nature of the investigated sandwich beams did not allow to fully verify the ability of peel stoppers to increase the post-damage capacity of these structural elements. Moreover, in the present investigation it is important to realise that for ‘‘one-dimensional’’ sandwich beams/panels it is unrealistic to expect a significant increase of the strength capacity of a partly damaged (delaminated) sandwich, as there is very limited (none) capacity for redistribution of loads. Thus, only experiments with and modelling of ‘‘two-dimensional’’ structural sandwich elements, i.e., sandwich panels, plates or shells, with peel stoppers can be expected to show an increased strength. This will be further examined and clarified in experiments with ‘‘two-dimensional’’ sandwich structures like panels and plates. 5. Conclusions A new peel stopper design, which effectively prevents face sheet peeling and confines debonding/delamination to the restricted areas in sandwich components, is introduced. The proposed peel stopper design has been tested on sandwich beam configurations loaded quasi-statically in
The work presented was supported by the Danish Research Council for Technology and Production Sciences; Grant N 26-04-0160, ‘‘Structural Grading – a novel concept for design of sandwich sub-structures’’, and the Innovation Consortium ‘‘Integrated Design and Processing of Lightweight Composite and Sandwich Structures’’ (abbreviated ‘‘KOMPOSAND’’) funded by the Danish Ministry of Science, Technology and Development. The support received is gratefully acknowledged. The authors also acknowledge DIAB group (Sweden) and Degussa Ro¨hm GmbH (Germany) for supplying the sandwich core materials used in this investigation. References [1] Zenkert D. An introduction to sandwich constructions. London: EMAS; 1995. [2] Zenkert D. The handbook of sandwich construction. London: EMAS; 1997. [3] Allen HG. Analysis and design of structural sandwich plates. Franklin Book Co; 1969. [4] Gay D, Hoa SV, Tsai SN. Composite materials: design and application. New York: CRC; 2003. [5] Burman M, Zenkert D. Fatigue of foam core sandwich beams-1: undamaged specimens. Int J Fatigue 1997;19(7):551–61. [6] Burman M, Zenkert D. Fatigue of foam core sandwich beams-2: effect of initial damage. Int J Fatigue 1997;19(7):563–78. [7] Zenkert D, Schubert O, Burman M. Fracture initiation in foam–core sandwich structures due to singular stresses at corners of flawed butt joints. Mech Compos Mater Struct 1997;4(1):1–21. [8] Shipsha A, Burman M, Zenkert D. Interfacial fatigue crack growth in foam core sandwich structures. Fatigue Fract Eng Mater Struct 1999;22(2):123–31. [9] Vadakke V, Carlsson L. Experimental investigation of compression failure mechanisms of composite faced foam core sandwich specimens. J Sandwich Struct Mater 2004;6(4):327–42. [10] Carlsson L, Matteson R, Aviles F, Loup D. Crack path in foam cored DCB sandwich fracture specimens. Compos Sci Technol 2005;65(15– 16, special issue):2612–21. [11] Avile´s F, Carlsson L. Experimental study of debonded sandwich panels under compressive loading. J Sandwich Struct Mater 2006;8(1):7–31. [12] Thomsen OT, Bozhevolnaya E, Lyckegaard A. Structurally graded core junctions in sandwich elements. Compos Part A 2005;36(10, special issue):1397–411. [13] Bozhevolnaya E, Thomsen OT. Structurally graded core junctions in sandwich beams: fatigue loading conditions. Compos Struct 2005;70(1):12–23.
J. Jakobsen et al. / Composites Science and Technology 67 (2007) 3378–3385 [14] Bozhevolnaya E, Thomsen OT. Structurally graded core junctions in sandwich beams: quasi-static loading conditions. Compos Struct 2005;70(1):1–11. [15] Bozhevolnaya E, Lyckegaard A. Local effects at core junctions of sandwich structures under different types of loads. Compos Struct 2006;73(1):24–32. [16] Lyckegaard A, Bozhevolnaya E, Thomsen OT. Parametric study of structurally graded core junctions. J Sandwich Struct Mater 2006;8(5):423–35. [17] Jakobsen J, Johannes M, Bozhevolnaya E. Failure prediction of inplane loaded sandwich structures. In: NAFEMs conference: prediction and modelling of failure using FEA, Denmark: Copenhagen/ Roskilde; 2006. [18] Grenestedt J. Development of a new peel-stopper for sandwich structures. Compos Sci Tech 2001;61(11):1555–9. [19] Wonderly C, Grenestedt J. Dynamic performance of a peel stopper for composite sandwich ship structures. J Compos Mater 2004;38(10):805–31.
3385
[20] Kim JH, Lee YS, Park BJ, Kim DH. Evaluation of durability and strength of stitched foam-cored sandwich structures. Compos Struct 1999;47(1–4):543–50. [21] Lascoup B, Aboura Z, Khellil K, Benzeggagh M. On the mechanical effect of stitch addition in sandwich panel. Compos Sci Tech 2006;66(10):1385–98. [22] Potluri P, Kusak E, Reddy T. Novel stitch-bonded sandwich composite structures. Compos Struct 2003;59(2):251–9. [23] Gurit Product Manual:. [24] Degussa Product Data Sheet: . [25] Hojstrup Product Manual:. [26] CIBA–GEIGY data sheet. [27] Huntsman, Araldite 2011 adhesive data sheet: . [28] DIAB group, Divinycell H-grade data sheet: . [29] Albany Engineered Composites:.
COMPOSITES SCIENCE AND TECHNOLOGY Composites Science and Technology 67 (2007) 3378–3385 www.elsevier.com/locate/compscitech
New peel stopper concept for sandwich structures J. Jakobsen *, E. Bozhevolnaya, O.T. Thomsen Department of Mechanical Engineering, Aalborg University, Pontoppidanstraede 101, 9220 Aalborg East, Denmark Received 3 January 2007; received in revised form 6 March 2007; accepted 18 March 2007 Available online 12 April 2007
Abstract The paper addresses the damage tolerance of sandwich structures, where the prevention and limitation of delamination failure are highly important design issues. Due to the layered composition of sandwich structures, face–core interface delamination is a commonly observed failure mode, often referred to as peeling failure. Peeling between the sandwich face sheets and the core material drastically diminishes the structural integrity of the structure. This paper presents a new peel stopper concept for sandwich structures. Its purpose is to effectively stop the development of debonding/delamination by rerouting the delamination, and to confine it to a predefined zone in the sandwich structure. The suggested design was experimentally tested for different material compositions of sandwich beams subjected to three-point bending loading. For all the tested sandwich configurations the suggested peel stopper was able to stop face–core delamination and to limit the delamination damage to restricted zones. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: B. Fracture; B. Interface; C. Crack; C. Delamination; C. Sandwich
1. Introduction Sandwich materials are layered structural components composed of thin strong face layers separated and bonded to light weight core materials. This particular layered composition creates a structural element with a very high bending stiffness to weight ratio as well as bending strength to weight ratio. Sandwich structures are often utilized in the marine, aerospace, train and automotive industries, where low weight is a critical design parameter. Furthermore, large parts of wind turbine blades are made using composite sandwich materials. The general concept of sandwich structures has been investigated and developed by many researchers over the past 50 years, see for example Zenkert [1,2], Allen [3] and Gay [4]. It is well known that sandwich structures may suffer from sudden failure if the allowable design loads are exceeded. When structures are made of ductile metallic materials, e.g., steel or aluminium, usually they do not fail *
Corresponding author. Tel.: +45 9635 9322; fax +45 9815 1675. E-mail address: [email protected] (J. Jakobsen).
0266-3538/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.compscitech.2007.03.033
catastrophically in a sudden way, due to their ability to redistribute the loads by plastic yielding. As opposed to this, composite and sandwich structures often exhibit a more brittle behaviour, and this, together with their inherent layered composition may lead to a sudden and fast failure by delamination (peeling), which occurs without any prior warning. Therefore, the study of initiation, propagation and arrest of delamination failure are issues of extreme importance for modern sandwich structures. Failure phenomena related to sandwich structures have been studied intensely by various researchers. Zenkert et al. have studied failure in foam cored sandwich panels with and without initial debonds [5–8]. They examined sandwich beams subjected to three-point bending loading under quasi-static and fatigue loading conditions. For the majority of the tested configurations face–core delamination occurred as failure mode. Moreover, in some cases failure initiated in the centre of the core and then kinked toward the face sheets and continued as a delamination along the face–core interfaces. Failure phenomena in general, and crack tip propagation in particular, in sandwich structures under in-plane
J. Jakobsen et al. / Composites Science and Technology 67 (2007) 3378–3385
compression loading have been studied by Carlsson et al. [9–11] among others. Sandwich elements with initial debond imperfections were experimentally studied, and it was concluded that the initial debonds would grow when the load applied to the test specimens exceeded the buckling load of the face sheet (i.e., wrinkling of the sandwich faces). Furthermore, their results stated that larger initial debonds would lower the total buckling strength of the sandwich panel, and that the geometrical shape of the debonded area seems also to have a significant influence on load bearing ability of the sandwich panel. Bozhevolnaya et al. [12–16] investigated the influence of core junctions on the static and fatigue strength of sandwich beams. It was found that modifications of the geometric shape of a core junction can have a considerable effect on the fatigue life of sandwich structures. Additionally, similar to the studies [5–8] for the case of static and fatigue loading, it was observed that face–core delamination very often was the dominating failure mode, which usually followed shear cracking of the compliant core in the sandwich beam. Core junction failure was further studied for sandwich structures subjected to in-plane tensile loading [17], and it was concluded that failure generally initiated in the vicinity of the core junction, and that final failure occurred as tensile face failure. According to the previous works cited above, a commonly observed failure phenomenon for sandwich structures is face–core delamination, which usually follows various types of locally induced damages in the sandwich core and/or in the vicinity of sandwich sub-structures like core joints, inserts or edge stiffeners. This failure mode is also referred to as face sheet peeling. Several techniques for improving the peeling strength of face-core interfaces are known today. For example, Grenestedt [18] suggested a new peel stopping manufacturing technique. The basic principle of this technique is that the debonded face sheets are able to be separated from the structure in order to arrest delamination growth beyond the implemented peel stoppers [19]. This method also implies that the sandwich structure looses a huge part of its overall bending stiffness and in-plane tensile strength, as part of the face sheets simply peels of. Accordingly, the advantages of the structural sandwich concept no longer exist or are significantly diminished. Another approach to arrest face sheet delamination is to stitch the face sheets together [20–22]. This method creates a sandwich panel with extraordinary high transverse stiffness and strength. Moreover, the shear properties can also be improved if the stitching angle is inclined 45° with respect to the normal plane of the sandwich panel. The method is very effective when applying it to monolithic composite laminates, but rather tedious and design restrictive in sandwich applications. Moreover, manufacturing complexities and costs are often increased with this method. However, the commercially available sandwich systems X-corTM and K-corTM of Albany Engineered Composites [29] belong to this type.
3379
The research presented in this paper concerns the proposal of a new peel stopper design, which effectively prevents peeling of the faces, and limits debonding/ delamination to a priori restricted areas of the sandwich component. The design of the new peel stopper is described, and the choice of appropriate peel stopper material is substantiated. The efficiency of the new peel stopper concept is validated by experimental tests with sandwich beams subjected to three-point bending up to the failure. 2. Peel stopper concept The basic design of the suggested peel stopper is illustrated in Fig. 1. The peel stopper is a sub-structural component embedded into the sandwich panel (like an insert or edge stiffener), and its main purpose is to arrest face–core interface crack propagation by rerouting the crack path into a closed/restricted area of the sandwich panel, thus preventing the spreading of debonding/delamination into the remaining parts of the sandwich structure. For the present study, the peel stoppers were manufactured from an elasto-plastic material, with elastic properties close to the sandwich core properties. Generally, it is recommended that the material of the peel stopper is chosen to be compliant and with large straining capability (i.e., ductile), and the elastic stiffness of the peel stopper is recommended to be of the same order as the elastic properties of the main sandwich core (or somewhat higher). Good adhesion properties with respect to both the core and the faces are required as well. The peel stoppers may be mounted into a sandwich panel (e.g., a sandwich beam, plate or shell) together with other sub-structural components like e.g., structural inserts as shown in Fig. 1. Essentially, there are no or only minor
Fig. 1. Proposed design of the peel stopper (a). The case shown displays a crack rerouting angle of 10°. A suggested implementation of the proposed peel stoppers in a sandwich plate is shown in (b). The grid type pattern will confine damage to the grid mesh.
3380
J. Jakobsen et al. / Composites Science and Technology 67 (2007) 3378–3385
Fig. 2. The basic idea of the peel stopper is to force the crack to propagate along the stopper–core interface (internal curve – arrowed line) and not along the face–core interface (dashed line).
manufacturing difficulties associated with the introduction of the peel stoppers, since they are similar to insert types already widely used in sandwich structures. Failure is often initiated by crack formation in the interior core parts of the sandwich structure due to fatigue load conditions, impact/shock (dynamic) loads or from manufacturing imperfections. Under such circumstances the core crack will often propagate towards a core–face interface, from where it usually proceeds as a delamination along the face–core interface. The principle idea of the new peel stopper concept is that propagation of delamination/debonding is prevented to spread beyond the boundaries of the proposed peel stoppers. This inhibition is due to a rerouting of the crack along the internal curve of the peel stopper (arrowed line) instead of propagation along the interface (dashed line), as shown schematically in Fig. 2. This technique will confine the debonding/delamination to a limited area of a single grid of the peel stoppers (cf. Fig. 1), and thereby prevent the spreading of damage further into the plate, thus averting its total collapse. Furthermore, the proposed peel stopper will allow the debonded sandwich face be kept attached to the sandwich component (contrary to the method described in [18,19]). This will retain some structural load carrying capability of the debonded structure after partial delamination/debonding under all types of loadings. 3. Test specimens Verification and functionality tests of the proposed peel stopper design were performed by means of comparing three sandwich beam test configurations, denoted as (a)– (c) in Fig. 3. These beams were manufactured with PMI RohacellÒ cores. Furthermore, an additional test configuration with PVC DIABÒ cores (denoted as (d) in Fig. 3) was made in order to validate the concept with respect to other material compositions. For each configuration (a–c), two specimens were manufactured and tested quasi-statically in three-point bending loading, as illustrated schematically in Fig. 3. The main purpose of the experiments was to induce a shear failure in the softer core of the sandwich beam, followed by crack propagation and crack-kinking towards the face–core interfaces and finally delamination along the interfaces. This allowed a detailed
Fig. 3. Three test configurations (a, b, and c) with RohacellÒ foams and carbon fibre reinforced composite faces for the quasi-static validation of the peel stopper concept. Test configuration (d) with DIABÒfoams and aluminium faces for the quasi-static validation of the peel stopper concept.
study of the influence of the presence of the peel stoppers on the propagation of completely developed delaminations. Each test specimen had a total length of 500 mm and a beam span between the supports of 460 mm. The beam configurations (a)–(c) were manufactured with a 1 mm thick carbon fibre laminate face sheets. The lay up was (0°, 90°, 0°) of both top and bottom faces. The core of these specimens consisted of two 25 mm thick PMI foam core parts from RohacellÒ with different densities (51WF and 200WF). The stiffer core, 200WF, was located at the edges of the beams, and an araldite diaphragm was placed in the beam centre to avoid indentation failure due to the external loading. In beam configurations (b) and (c), polyurethane inserts were embedded between the two cores as shown in Fig. 3. A conventional butt insert was used in configuration (b), and the proposed peel stopper was used in configuration (c). The material data are specified in Table 2. Configurations (b) and (c) were designed to have equal mass and identical material composition. Configuration (a), which represents a realistic design configuration, was considered as a reference to evaluate the two other configurations against. The discrepancy between the average core mass of configurations (b) and (c) is less than 4% of the total mass of the beam (cf. Table 1). This difference may Table 1 Mass measures of the tested configurations (Fig. 3) Specimen
Avg. core mass [g]
Avg. of total mass [g]
b c
143 152
223.5 232.5
J. Jakobsen et al. / Composites Science and Technology 67 (2007) 3378–3385
3381
Table 2 Mechanical properties of the tested beams Materials
E-modulus [MPa]
Tensile strength [MPa]
Compress. strength [MPa]
Elongation at failure [%]
Test configurations (a)–(c) T700 UD/SE 84LV [23] – face RohacellÒ 200WF [24] – edge core RohacellÒ 51WF [24] – main core
129200 350 75
2844 6.8 1.6
1187 9.0 0.8
–
Test configurations (d) AL 7075-T6 [17] – face DivinycellÒ H200 [28] – edge core DivinycellÒ H60 [28] – main core
71000 250 75
503 7.1 1.8
503 4.8 0.9
Test configurations (b)–(d) PERMAlock 40496 (PU) [25] – peel stopper
100
All test configurations Araldite 2011 [27] – adhesive Araldite B30 [26] – diaphragm
– 3700
10
–
Shear lap strength 26 MPa 60 100
be caused by a larger adhesive surface area in configuration (c). Every test configuration shown in Fig. 3 was manufactured by assembling and bonding the core prior to prepreg/face lamination using a vacuum bagging technique. The sandwich panel was cured for 6 h at 100 °C, and afterwards post-cured for 48 h at room temperature. Finally, the sandwich panel was cut into separate beams with a final width of 58 mm. A fourth configuration (d) (Fig. 3) was manufactured in order to examine if the concept with a polyurethane peel stopper exhibits the same effect with other face and core material compositions. Configuration (d) consisted of 1 mm thick Aluminium (AL 7075-T6) face sheets and 25 mm thick DivinycellÒ foam cores (H200 and H60). Moreover, as for configurations (a)–(c), a diaphragm was placed in the centre of the beam to introduce the central local load. Configuration (d) was not manufactured using the vacuum bagging technique. Instead the sandwich core components were glued together with an AralditeÒ 2011 adhesive and cured at room temperature for 24 h. After this the aluminium face sheets were bonded to the assembled core using the same adhesive under controlled pressure condition followed by curing for another 24 h. All the tested configurations (a)–(d) were geometrically similar, and the only difference was the choice of material composition. The mechanical properties of the sandwich beams constituents are given in Table 2.
3.5 3.0 11 4.5 [17] 5.5 [17] 25
5–6
tal testing of the sandwich beams, since it provided a controlled shear cracking of the core in the bulk of the weaker foams. The shear cracks propagated towards the face sheets, where crack kinking occurred followed by crack propagation in a delamination mode along the face–core interfaces. An experimental set-up was designed and manufactured on the basis of a 100 kN servo hydraulic Schenk HydropulsÒ testing machine. The testing machine has four test regions, where the lowest region was used for this particular test setup. The upper load limit in this test region is 12.5 kN, which gave a discrepancy between the input and output load signal of less than 3%. A load controlled mode was used during loading, which was performed with a load rate of 0.02 kN/s. The central deflection of all specimens was recorded via the displacement of the cross head. High-speed video recording of specimen failure was enabled with a frame rate of 6000 frames/s. The observed flexural load vs. central deflection responses of configurations (a)–(c) shown in Fig. 4 were very similar as expected. This is an indication that the overall structural stiffness of the sandwich beam was not affected by the introduction of the peel stoppers.
4. Test results It is known that sandwich structures are particularly vulnerable to transverse shear loading. At the same time this type of loading occurs very often in practice, where it may (and often does) provoke an initial shear failure of the sandwich core, which is then inevitably followed by a delamination of the sandwich core and faces. Therefore the three-point bending scheme was chosen for experimen-
Fig. 4. Applied force vs. central displacement for the specimens of configurations (a)–(d). Two specimens of each configuration were tested.
3382
J. Jakobsen et al. / Composites Science and Technology 67 (2007) 3378–3385
The failure characteristics of all six test specimens are summarised in Table 3. The recorded maximum loads and maximum central deflections of the beams at failure were quite close for three different configurations. The average failure load measured for the specimens of configuration (c) was 2216 N. This load level is within 5% of the average failure loads of configurations (a) and (b). Additionally, the central deflection at failure for configuration (c) is observed to be between the central deflections observed for configurations (a) and (b). This difference is estimated to be around 9%. In this connection it should
be mentioned, that the short length of the weak core compared to the total length of the edge stiffeners and peel stoppers is the reason for the difference in central deflections observed for the improved (c) and conventional (a), (b) edge stiffeners. If the length of the pure core part of the sandwich structure was larger, compared to the total length of the embedded sub-structures, which would be the case for realistic design configurations, the difference in the ultimate displacements, and thus the influence of the peel stoppers on the overall structural stiffness of the sandwich, would be much smaller.
Table 3 Failure characteristics of the test specimens Specimen and configuration
Failure load [N]
Avg. failure load [N]
Cross head displacement at failure [mm]
Avg. cross head displacement at failure [mm]
Location of failure initiation
Completed delamination
a1
2307
2320
11.58
11.78
core
Yes
a2
2332
core
Yes
b1
2094
core
Yes
b2
2106
core
Yes
c1
2185
core
No
c2
2246
core
No
d
1868
Compliant (51WF) Compliant (51WF) Compliant (51WF) Compliant (51WF) Compliant (51WF) Compliant (51WF) Compliant (H60)
core
No
11.97 2100
10.11
9.96
9.80 2216
10.43
10.83
11.22 –
18.77
–
Fig. 5. Failure of test specimen a1 and b2 (cf. Fig. 3 and Table 3).
J. Jakobsen et al. / Composites Science and Technology 67 (2007) 3378–3385
Regarding configuration (d), one specimen was tested under the same load conditions as specimens (a)–(c). The load vs. cross head displacement curve is shown in Fig. 4. Notice that the maximum critical load for configuration (d) is close to those previously measured and shown in Table 3. This is explained by the fact that the strength of the weaker core is a limiting factor in the beam design, and the strength of RohacellÒ 51WF is close to the strength of the DivinycellÒ H60 (cf. Table 2). Damage initiation and development of failure occurred according to the predicted scenario, as shown in Figs. 5 and 6. High speed video recordings were used to identify the location of failure initiation and its progression in the sandwich specimens. In all cases, failure initiated as a shear crack in the centre of the weak core, the crack tip kinked towards the faces and continued as a delamination along face–core interfaces. Notice that a full delamination of the face occurs for the cases of conventional edge stiffeners (Fig. 5), while the peel stopper in Fig. 6 clearly confines the crack inside the weak core, and, moreover, lets the sandwich face still be attached to the sandwich beam edge. The sandwich beams with embedded peel stoppers were subsequently loaded in the three-point bending fixture in order to inspect the damage zone in the vicinity of the peel stoppers as seen in Fig. 7. All beams of configurations (a) and (b) ended up with completely delaminated face sheets, while peel stoppers in configuration (c) effectively stopped delamination. A failure scenario similar to the observations for configuration (c) was observed for configuration (d), as illustrated by the high-speed recording shown in Fig. 6. Again, the peel stopper confirmed its efficiency in rerouting
3383
Fig. 7. Post-mortem inspection of the cracked part of test specimen c2 (cf. Fig. 6).
the crack propagation, and by stopping the crack propagation beyond the bounds of the area marked by the peel stoppers (cf. Fig. 8).
Fig. 6. Failure of test specimen c2 and d (cf. Fig. 3 and Table 3).
3384
J. Jakobsen et al. / Composites Science and Technology 67 (2007) 3378–3385
three-point bending. To validate the functionality of the peel stopper concept, three test configurations were examined: one without any peel stopping reinforcement, another one with a conventional butt-reinforcement and the third configuration with the proposed peel stoppers. Different material constituents were used in the test specimens with embedded peel stoppers. Only the configurations with the embedded peel stoppers were able to arrest the development of debonding/delamination beyond the bounds of the peel stoppers, and thereby prevent total delamination of the sandwich faces. Acknowledgements
Fig. 8. Post-mortem inspection of the cracked part of test specimen d (cf. Fig. 6).
A post-mortem inspection of the damaged beams has demonstrated that in all cases the beams containing the peel stoppers retained at least 10% of their initial bending stiffness, while the conventional beams retained none. The use of peel stoppers or other crack stopping devices are usually expected to lead to a higher ultimate load carrying capability (strength) of sandwich structures. However, the ‘‘one-dimensional’’ nature of the investigated sandwich beams did not allow to fully verify the ability of peel stoppers to increase the post-damage capacity of these structural elements. Moreover, in the present investigation it is important to realise that for ‘‘one-dimensional’’ sandwich beams/panels it is unrealistic to expect a significant increase of the strength capacity of a partly damaged (delaminated) sandwich, as there is very limited (none) capacity for redistribution of loads. Thus, only experiments with and modelling of ‘‘two-dimensional’’ structural sandwich elements, i.e., sandwich panels, plates or shells, with peel stoppers can be expected to show an increased strength. This will be further examined and clarified in experiments with ‘‘two-dimensional’’ sandwich structures like panels and plates. 5. Conclusions A new peel stopper design, which effectively prevents face sheet peeling and confines debonding/delamination to the restricted areas in sandwich components, is introduced. The proposed peel stopper design has been tested on sandwich beam configurations loaded quasi-statically in
The work presented was supported by the Danish Research Council for Technology and Production Sciences; Grant N 26-04-0160, ‘‘Structural Grading – a novel concept for design of sandwich sub-structures’’, and the Innovation Consortium ‘‘Integrated Design and Processing of Lightweight Composite and Sandwich Structures’’ (abbreviated ‘‘KOMPOSAND’’) funded by the Danish Ministry of Science, Technology and Development. The support received is gratefully acknowledged. The authors also acknowledge DIAB group (Sweden) and Degussa Ro¨hm GmbH (Germany) for supplying the sandwich core materials used in this investigation. References [1] Zenkert D. An introduction to sandwich constructions. London: EMAS; 1995. [2] Zenkert D. The handbook of sandwich construction. London: EMAS; 1997. [3] Allen HG. Analysis and design of structural sandwich plates. Franklin Book Co; 1969. [4] Gay D, Hoa SV, Tsai SN. Composite materials: design and application. New York: CRC; 2003. [5] Burman M, Zenkert D. Fatigue of foam core sandwich beams-1: undamaged specimens. Int J Fatigue 1997;19(7):551–61. [6] Burman M, Zenkert D. Fatigue of foam core sandwich beams-2: effect of initial damage. Int J Fatigue 1997;19(7):563–78. [7] Zenkert D, Schubert O, Burman M. Fracture initiation in foam–core sandwich structures due to singular stresses at corners of flawed butt joints. Mech Compos Mater Struct 1997;4(1):1–21. [8] Shipsha A, Burman M, Zenkert D. Interfacial fatigue crack growth in foam core sandwich structures. Fatigue Fract Eng Mater Struct 1999;22(2):123–31. [9] Vadakke V, Carlsson L. Experimental investigation of compression failure mechanisms of composite faced foam core sandwich specimens. J Sandwich Struct Mater 2004;6(4):327–42. [10] Carlsson L, Matteson R, Aviles F, Loup D. Crack path in foam cored DCB sandwich fracture specimens. Compos Sci Technol 2005;65(15– 16, special issue):2612–21. [11] Avile´s F, Carlsson L. Experimental study of debonded sandwich panels under compressive loading. J Sandwich Struct Mater 2006;8(1):7–31. [12] Thomsen OT, Bozhevolnaya E, Lyckegaard A. Structurally graded core junctions in sandwich elements. Compos Part A 2005;36(10, special issue):1397–411. [13] Bozhevolnaya E, Thomsen OT. Structurally graded core junctions in sandwich beams: fatigue loading conditions. Compos Struct 2005;70(1):12–23.
J. Jakobsen et al. / Composites Science and Technology 67 (2007) 3378–3385 [14] Bozhevolnaya E, Thomsen OT. Structurally graded core junctions in sandwich beams: quasi-static loading conditions. Compos Struct 2005;70(1):1–11. [15] Bozhevolnaya E, Lyckegaard A. Local effects at core junctions of sandwich structures under different types of loads. Compos Struct 2006;73(1):24–32. [16] Lyckegaard A, Bozhevolnaya E, Thomsen OT. Parametric study of structurally graded core junctions. J Sandwich Struct Mater 2006;8(5):423–35. [17] Jakobsen J, Johannes M, Bozhevolnaya E. Failure prediction of inplane loaded sandwich structures. In: NAFEMs conference: prediction and modelling of failure using FEA, Denmark: Copenhagen/ Roskilde; 2006. [18] Grenestedt J. Development of a new peel-stopper for sandwich structures. Compos Sci Tech 2001;61(11):1555–9. [19] Wonderly C, Grenestedt J. Dynamic performance of a peel stopper for composite sandwich ship structures. J Compos Mater 2004;38(10):805–31.
3385
[20] Kim JH, Lee YS, Park BJ, Kim DH. Evaluation of durability and strength of stitched foam-cored sandwich structures. Compos Struct 1999;47(1–4):543–50. [21] Lascoup B, Aboura Z, Khellil K, Benzeggagh M. On the mechanical effect of stitch addition in sandwich panel. Compos Sci Tech 2006;66(10):1385–98. [22] Potluri P, Kusak E, Reddy T. Novel stitch-bonded sandwich composite structures. Compos Struct 2003;59(2):251–9. [23] Gurit Product Manual: