Experimental verification of PEKK stiffened panel under compression

The paper presents the experimental strain and stress analysis of a thermoplastic composite stiffened panel subjected to compression load. The panel has five stringers with a non-symmetric design. Two panels are investigated. The experimental analysis deals with the pristine panel without any flaws and the panel with an artificial crack at the middle stringer interface with the skin. Panels were made from a thermoplastic polyetherketoneketone (PEKK) carbon composite. The buckling and mechanical analysis is based on digital image correlation, strain gauge measurements, and linear variable differential (LVDT) transducers. The crack subcritical extension was defined and analysed. The results show that crack propagation starts coupled with sudden buckling mode transition. The crack growth behaviour is influenced by the buckling shape, which consists of two main modes (with three and four half-waves) in the longitudinal direction in each skin bay.


Introduction
High-performance thermoplastic materials have been used in aeronautical composite structures since the eighties of the last century [1].The main advantages of thermoplastic materials lay in the high toughness, chemical resistance, and expanded manufacturing possibilities such as thermoplastic welding, hot press forming, and co-consolidation processes [2,3].On the other hand, the high price of these materials strongly limited their utilization in general.Thermoplastics enable the application of thermoforming technology and decrease the number of parts to assemble and out of autoclave techniques.All these techniques can significantly reduce manufacturing costs.
In recent years several projects on thermoplastic aerostructures have been successfully executed.An overview of such projects realized in the EU is mentioned in [4].Similarly, in the Czech Republic, several national programs focused on thermoplastics and on primary aircraft structures were carried out.VZLU (Czech Aerospace Research Centre) also participated in several EU projects focused on this topic.Results of such projects were published in the past [5 -10].The main goals of this research were the application of the thermoforming technique (but not only it), the influence of impact damage on strength behaviour, tailored blank utilization, etc.
The presented paper considers the analysis of a thermoplastic composite stiffened asymmetric panel manufactured using the assembly of several simply flat parts with emphasis on buckling behavior, damage propagation, and final collapse.The influence of the barely visible impact damage effect was investigated on similar three-stringer panels in the past [11].In this paper, the five stringers thermoplastic panels with an angled cap on one side and disbond are discussed.Two panels were investigated, the first one was reference (pristine) without any flaw.The second one includes a disbond / crack at the middle stringer interface.Panels were designed and manufactured by GKN Fokker.The experimental results allow a better understanding of thermoplastic structures and the effect of disbond / crack propagation.

Panel geometry and material
The five-stringer panel is shown in figure 1.The panel consists of a composite skin and 5 stringers.Panel dimensions after curing were 545 mm long, and 690 mm in width.The panels were manufactured and coated with white primer by Fokker Aerostructures.Both panels were additionally potted into metallic blocks and machined at VZLU.The final width of both panels was 650 mm, and the length was 539 mm including potting plates.The skin / middle stringer connection of one panel was cracked.The artificial crack was created by a 40 mm Teflon insert at first.After that, the artificial crack was extended up to 72 mm using a special fixture.The laminated parts consist of three different layups and are joined by a short carbon fiber PEKK filler.The layup of the laminated parts is described in table 1.The thickness of one ply is 0.14 mm.The laminated parts were manufactured by an automated fiber placement process, the filler material was an extrusion product, and the parts were joined by an autoclave process.

Test set-up
The MTS four-column load frame with a capacity of 1 MN equipped with a crosshead-mounted actuator was used for the compression test.No special buckling device for panel stabilization was used.Additionally, two MICRO EPSILON LVDT wire sensors and ARAMIS optical digital image correlation system were used for panel displacement and strain-stress measurements.FASTCAM SA-Z high-speed camera and standard video cameras were applied for recording crack and failure development.The compression five-stringer panel test arrangement is illustrated in figure 3.

Results and discussion
Figure 5 represents the displacement vs force dependence measured by using various methods for reference and pre-cracked panels.Although one of the panels was pre-cracked before the compression test, the residual compressive strength of both panels is very similar (456 kN vs 477 kN).The residual -0,10 -0,05 0,00 0,05 0,10 strength of the pre-cracked panel was only about 4% lower compared to the reference panel without any flaws.Displacement data confirmed the expected fact, that displacement values measured using an LVDT sensor located on the actuator of the loading facility (MTS) give incorrect information.Only the values measured directly on the panel area are true and correct for numerical simulation analyses and comparison.Differences between DIC and wire sensor data can be explained by initial clearances and slightly different bases of measurements.Slope modification of curves in figure 5 corresponds to sudden buckling mode changing (mode I to mode II) during loading (dashed lines represent pre/cracked panel data).The skin load transfer in mode II is significantly lower compared with mode I, and consequently the panel stiffness is different.
The strain development indicates very similar buckling behaviour for both panels as can be seen in figure 6 and figure 7. Two main buckling modes were observed.The first buckling mode is represented by the three buckles, and the second one by the four buckles.The buckling mode development of both tested panels is shown in figure 8 and figure 9.    Initiation of buckles started approximately at a similar load force for both panels (160 kN).Consequently, due to the crack presence, the deformation in the Z-axis has been developing to higher values in the pre-cracked panel.This occurred because the lower stiffness of the panel surrounding the middle stringer can be defined.This strain deviation from expected values for the reference panel indicates the possibility of using this behaviour for structure health monitoring (SHM).Additionally, the first buckling mode was kept to the higher load levels compared to the flawless panel.The sudden transition from the first to the second buckling mode of the pre-cracked panel was accompanied by sudden crack propagation (from 72 mm up to 162 mm on the left side of the middle stringer).The crack propagation curve of the pre-cracked panel is shown in figure 10.The sudden transition to the second skin buckling mode of the reference flawless panel was not accompanied by any crack or delamination occurrence.Deformation in the Z-axis of both panels before total collapse was very similar.The total collapse of the panels due to disbond between the stinger web and stringer cap was observed for both panels.A yellow ellipse in figure 11 marks an overview of collapse initiation.

Conclusion
The mechanical behaviour of the pristine (reference) and flaw-containing five-stringer thermoplastic composite panels are discussed.The effect of the artificial crack on the buckling development, damage propagation, and final collapse was experimentally investigated.Observed differences between damage, strains, and z-axis deformation development can be used for future structure health monitoring of a panel structure.Two buckling modes of the same type were observed in both panels.Skin buckles were initiated for flawless and pre-cracked panels approximately at the same load level.The first buckling mode of the pre-cracked panel developed to significantly high load levels and strains compared to the reference flawless panel.Transition to the second buckling mode of the pre-cracked panel was accompanied by sudden crack propagation contrary to the reference panel in which no crack or delamination was observed in this stage of the test.The strength of both panels was very similar; the difference was only 4%.The total collapse of both panels was initiated by the debonding of stringer parts.
The failure was initiated primarily in the stringer/skin contact and secondarily in the stringer/stringer cap contact.The weak point was the loss of the stringer stability.Although the panel with introduced artificial debonding showed a transition to the second buckling mode of the skin compared with the reference panel earlier and the transition was accompanied by an increase in the debonding area, the loss of stringer stability determined the strength of the panel.Similar conclusions were obtained by numerical analysis in [12].The influence of the filler material on the failure process cannot be directly concluded from the test.The main aspect clear after the rupture was that the filler bonding was not high.The panel sub-parts, i.e., the skin, the stringer web, and the cap, were separated from the filler according to the co-consolidation process during manufacturing.Increasing the bonding could stabilize the stringer against the loss of stability, but the final rupture load should not be expected much higher.

Figure 1 .
Figure 1.Five-stringer panel with rulers around debonded middle stinger Both panels were made of PolyEtherKetoneKetone carbon (C/PEKK) unidirectional plies.The cross-section of the stringer region consists of three laminated parts and the filler as shown in figure 2.The laminated parts consist of three different layups and are joined by a short carbon fiber PEKK filler.The layup of the laminated parts is described in table1.The thickness of one ply is 0.14 mm.The laminated parts were manufactured by an automated fiber placement process, the filler material was an extrusion product, and the parts were joined by an autoclave process.

Figure 2 .
Figure 2. Metallographic picture of individual panel parts after failure

Figure 3 .
Figure 3. Panel test arrangement Panels were instrumented with 22 axial KYOWA KFRP strain gauges (SG) on both sides of the panel -11 pieces on a flat skin surface and 11 pieces on the panel side with stringers.Back-to-back positioning was kept.The positioning of SG is depicted in Fig. 1 (skin/skin and stringer cap/skin).Fig. 4 illustrates differences between the mean values of skin/skin and stringer cap/skin positioned strain gauges and the mean strain value at a load of 50 kN.Although the panel structure is asymmetric the load distribution is sufficiently uniform.

Figure 4 .
Figure 4. Differences in mean values of skin/skin and stringer cap/skin positioned strain gauges and the mean panel strain value at a compression load of 50 kN

Figure 5 .
Figure 5. Displacement vs. load dependence measured of the reference and the pre-cracked panels

Figure 6 .
Figure 6.Strain development during compressive loading of the pre-cracked panel

Figure 7 .Figure 8 .Figure 9 .
Figure 7. Strain development during compressive loading of the reference panel

Figure 10 .
Figure 10.Subcritical crack propagation curve in the middle stringer area

Figure 11 .
Figure 11.Area of initiation of loss of panel stability and strength Crack length -left side Crack length -right side Crack length -mean value 7th International Conference of Engineering Against Failure Journal of Physics: Conference Series 2692 (2024) 012019