Bio-based hybrid cabin door of ultralight helicopter with variable-axis fiber design

This paper presents the development of an integrated technology concept for the production of semi-structural fiber composite aerospace components with a bio-based hybrid laminate structure. In this context, the high lightweight potential of a variable-axial hybrid fiber-reinforced laminates compared to a multiaxial laminate design is of particular importance. As a demonstrator, a double-shell cabin door based on the cabin door of the ultra-light helicopter CoAx 2D of EDM Aerotec GmbH was redesigned in hybrid bio-based mixed composite construction using selected finite-element method (FEM) simulation and optimization tools, manufactured and characterized. The obtained results illustrate that compared to the reference carbon fiber-reinforced polymer (CFRP) component, the developed bio-based hybrid composite with local tailored fiber placement (TFP) reinforcement has a 30% biomass content, exhibits comparable mechanical properties and significantly contributes to increase energy and resource efficiency significantly.


Introduction
One of the common solutions for achieving a high lightweight degree in aircraft and helicopter construction is the substitution of light metal aerospace materials, such as aluminium or titanium, by fiber-reinforced plastic composites (FRP) made of glass, carbon or aramid fibers. The production of FRP structures requires a high input of resources; in addition, the recycling of old components is a major challenge. This has a negative impact on recycling costs and the overall carbon footprint of the aircraft structure [1]. To increase the eco-efficiency of flying, bio-based fiber composites can be used to minimize the final product's environmental impact. So far, highly stressed aerospace components have not yet been implemented from natural fiber-based lightweight materials due to the fact that its mechanical properties are significantly lower than those of CFRP. This means that pure material substitution is rarely a practical option.
A promising solution to meet the aerospace requirements is the hybridisation of natural fiber-based FRP, 1 3 which implies a tailored integration of carbon fibers in the direction of the application critical stresses present in the component. For such hybridization, variable-axis fiber deposition methods (e.g., TFP) or smart hybrid multiaxial laminate design can be used [2,3]. The use of TFP reinforcements requires the development of a holistic technological approach for the numerical design as well as FEM-based simulation of hybrid composites using modern topology and material optimization techniques, with sufficient consideration of manufacturing aspects [4,5].
This work provides an insight into the performed design of a cabin door of an ultra-light helicopter (ULH) in hybrid design with a partial TFP reinforcement as well as into the development of an integrated technology concept. It is based on an interactive solution approach for the simultaneous development of optimum material composites, manufacturing processes and integrative construction methods. Although there have been similar works in the past [6], the focus of the research there was on geometry optimization and the presentation of a general viability under integration of partial UD-CFRP patches into the flax fiber composite. The patch geometry was defined empirically based on the main stress vectors. The work [6] can be considered as a preliminary stage, but the main difference lies in the unified analytical solution approach, where the design is primarily based on the FEM results. Such a solution approach has barely been pursued so far for sustainable, bio-based aerospace structures.

Materials
For the work carried out on the design of the hybrid helicopter door, commercially available materials were selected and processed into thermoset fiber composite panels and test specimens. To determine the required characteristic values, mechanical tests were carried out in compliance with DIN EN ISO 527-5, EN ISO 14129 and DIN EN ISO 14126. Other important characteristic values were partly taken from data sheets and literature or assumed on the basis of empirical values.

Polymer matrices
To increase the biogenic content of the hybrid composite, a 19 wt% bio-based thermosetting epoxy resin SuperSap INR/INS (EntropyResins) was used as matrix material in the work. This EP system has a viscosity B of 25 cPs at 25 °C as well as a pot life of 180 min and can be processed by VARI or RTM.

Carbon fibers
The CF semi-finished products used were a UD scrim of the type ZOLTEK PX 35 with a basis weight of 150 g/ m 2 and a 2/2 twill fabric of the type Style 462 Aero with a basis weight of 250 g/m 2 , which are approved for aviation applications. In the autoclave supported vacuum assisted resin injection process (aVARI), a fiber mass fraction of 55-60 wt% can be assumed, which corresponds to a fiber volume fraction of approx. 45 vol.%.

Flax fibers
Two unidirectional flax fiber fabrics (FF-UD) FlaxTape 110 and UD FAW 150 (quasi-UD fabrics) and a 2/2 twill fabric BD FAW 150 from EcoTechnilin were used as biogenic reinforcement material for the hybrid helicopter door. The numbers in the material designation represent the grammage in g/m 2 .
Compared with CF and GF semi-finished products, the FF semi-finished products have some quality deficits with regard to fiber orientation and homogeneity of fiber distribution. This inevitably leads to a higher statistical characteristic value scatter in the test laminates. Furthermore, the fiber volume fractions can be achieved are significantly below those of CF or GF semi-finished products.

Tailored fiber placement
Washable paper or thin polyester fleece is often used as embroidery base for TFP production. Textile semi-finished products can also be embroidered directly, thus simplifying the subsequent handling and precise positioning of the local TFP reinforcements. An embroidery machine from TAJIMA was used as the TFP system, with which the CF rovings can be processed up to 24 K. However, very stiff HM-CF or UHM-CF types of are not suitable for processreliable depositing of embroidery patterns with small radii. Therefore, 3 K, 6 K and 12 K HT-CF rovings were used. A 10 tex polyester thread was used as the embroidery thread. The stitch width (R value)-a measure of the bundling or spreading of the roving-and the roving spacing value l s are determinant parameters in TFP processing [5] and were extensively investigated here. To determine the optimum embroidery parameters, a quasi-isotropic polyester nonwoven (50 g/m 2 ) was first used as the embroidery base. Later, the determined embroidery parameters with necessary corrections were transferred to the bio-based semi-finished products BD FAW 150 and UD FAW 150.
The TFP embroidery of uni/bi-directional FF semi-finished products showed significant differences compared to polyester nonwovens. Due to the stretching of the bio-based embroidery base in the stent frame, the R values determined with polyester nonwovens had to be increased by 0.2 mm and l s had to be decreased by 0.2 mm. The flax fiber fabric BD FAW 150 could be embroidered well both parallel to the fiber and at dent angles. By reducing the embroidery spacing, it was possible to almost completely eliminate fiber endulation on BD FAW 150, especially with 12 CF rovings. In contrast, considerable processing problems were observed with the flax fabric UD FAW 150. As can be seen in Fig. 1, numerous voids occurred both in the deposited CF reinforcement and in the embroidery base due to the embroidery thread tension. Therefore, BD FAW 150 was favored for later investigations. Parameters with necessary corrections were transferred to the bio-based semi-finished products BD FAW 150 and UD FAW 150.

Specimen and component manufacturing
The autoclave assisted vacuum infusion process (aVARI) was used for specimen and component production. Using an autoclave (min. 5 bar overpressure), the laminate can be compressed more, which leads to higher fiber volume fractions and better mechanical properties even for natural fiber composites. A further advantage is the use of a bio-based thermosetting EP system. In aVARI, as in VARI processes, dry, ready-cut semi-finished reinforcements are placed in a laminating mould according to the laminate build-up plan, sealed airtight with a vacuum film and vacuum impregnated with the matrix resin at room temperature. In the autoclave, the excess resin is then pressed out by the overpressure and the curing cycle is completed. In this way, fiber contents of up to 50% by volume, which are otherwise normal for prepregs, can be achieved in natural fiber laminates. Compared to prepreg processing, however, the manual production effort is very high.

Characteristic values for the FEM simulation
To determine the mechanical properties, UD composites and flat TFP semi-finished products were manufactured and consolidated in the VARI process with SuperSap INR/INS to form a symmetrical two-ply UD laminate. Standardized tensile specimens were taken from the laminates by waterjet cutting according to DIN ISO 527-5 and subjected to tensile testing. Table 1 illustrates the mechanical parameters used for FE-based design, calculation and optimisation. Tensilebased characteristic values were determined experimentally. The pressure-based characteristic values were taken from technical data sheets and/or literature [7][8][9][10] or assumed on the basis of empirical values. Fiber volume content was determined gravimetrically. Until all necessary characteristic values are verified by an experimental investigation, a deviation between the simulated and experimental results must be accept.
Specifically, the TFP-FFRP semi-finished product with 12 K CF roving (last column in Table 1) achieved parallel to the fiber a Young's modulus of 47.5 GPa, a tensile strength at break of 738.0 MPa and an elongation at break of 1.5%. Transverse to the fiber is a Young's modulus of 8.7 GPa, tensile strength of 103.8 MPa, and elongation at break of 1.9%. In the parallel fiber direction, these values are well below a CF-UD laminate and slightly above a CF-BD laminate currently used for cabin door production.
In the case of consolidated TFP composite laminates, dry spots were observed between individual CF fiber bundles and, as a result, rough and visually unappealing surfaces. Furthermore, the embroidery thread was clearly visible. In the component design, the TFP reinforcement was not placed in the outer laminate layer or on the visible side of the component to improve the appearance of the component.

FEM simulation and optimization
The use of TFP reinforcements requires a detailed understanding of the existing stress state (principal stress distribution) under different load scenarios to design and dimension the component in a stress/production-oriented manner. To get one step closer to this goal, a CAD model of the left cabin door of the UHL CoAx2D was created and the deformation was simulated in Ansys Workbench 19.2 with an The deformations resulting from the FEM simulation are shown in Fig. 3. This shows a maximum deformation of the bottom corner of 21.537 mm.
In actual operation, the cabin door is, therefore, lifted outward by approx. 21.5 mm due to the prevailing vacuum. This gap may no longer be bridged by the rubber seal, which means that the tightness of the booth can no longer be guaranteed. For this reason, a smaller permissible deformation of approx. 20 mm is defined as the target value for the design. The laminate weight and the calculated deformation or the resulting efforts are used as reference variables in the further course of the component optimisation.

Definition of the TFP geometry
From the literature, some academic and commercial approaches for the derivation of the 2D or 3D TFP geometry are known. Software tools such as Abaqus Add-on CAIO (Computer-Aided Internal Optimization) [4], TACO (Tailored Composite Design Code) [11] or the EDOstructure (AOPS) [5] are particularly worth mentioning in this context. On the basis of various computational algorithms, corresponding principal stress lines are visualised automatically using principal stress integration and trajectories for the TFP geometry are derived in a further step. In this work, a new type of routine was developed for the design of the TFP geometry using the two available FE tools for the calculation and optimisation of fiber composites-Altair HyperWorks 2019 and Ansys Workbench 19.2 (Fig. 4). In addition, the CAD programs Solidworks and SpaceClaim were used to edit the geometry, and the Ansys Direct Optimisation plugin was used for parametric FRP optimization.
The design routine essentially consists of 2 calculation steps. In the first step, Altair Hyperworks uses what is known as Topography Optimization (also referred to as Free Sizing Optimization) to determine the particularly stressed ing Optimization is commonly used to generate a thickness distribution for each defined fiber orientation and can be used to interpret the cutting geometry of the semi-finished products [12]. In this case, an analogous procedure is followed, using optimization results to define the TFP geometry and TFP direction.
At the beginning of the optimization, the FEM model for the outer and inner shells has a four-layer laminate structure [0°; 45°; − 45°; 90°] sym . The laminates are arranged symmetrically to each other. The base material is defined as FlaxPly 110 with a single layer thickness of 0.125 mm. This material has the highest mechanical properties compared to results of other investigated flax fiber semi-finished products [13]. The stack thickness is initially 1 mm (8 plies) per fiber direction. Topography optimization reduces the layer thicknesses until the specified convergence criterion is reached. In Fig. 5, the deformation at the beginning and at the end of the topography optimization is shown. Convergence after 56 iteration steps is reached at max. deformation of 19.76 mm. Figure 6 illustrates the discrete thickness distribution of the stack depending on the fiber direction. The areas marked in light blue are locations that can only be modelled with a flax UD layer. The red areas indicate heavily stressed structural segments that need to be reinforced with an 8-layer UD laminate. In this case, the use of TFP reinforcement is appropriate. Based on the FE results and the lack of possibility to change the profile geometry of the circumferential door profile with the smallest cross section for the purpose of increasing the area moment of inertia, this area (Fig. 6, [90°] right) is modelled with TFP. Due to a rather diffuse principal stress distribution, the insertion of the CF fabric is preferable to the TFP reinforcement in the hinge area. Green-red area at 0° directions is implemented with 4 layers of CF-UD fabric.
The base laminate for the outer shell was a 4-ply [0°; 45°; − 45°; 90°] laminate developed from FlaxTape 110. Due to limited draping properties of FlaxTape 110, a 2-ply [0°; 45°] laminate of flax fabric FlaxPly BD FAW 150 is used for the inside shell. The hinge areas of then cabin door are made of CF fabric with a [0°; 45°] laminate. To improve the accuracy of the simulation results, the overlapping areas (20 mm) are also designed. The window was defined as a 4 mm thick shell of isotropic transparent PMMA (Plexiglas). In the design no core material was defined. In SolidWorks, the geometry for the TFP reinforcement is defined using the results shown in Fig. 6 for 90° direction and manually converted into an embroidery pattern. The laminate thickness distribution of the bio-based cabin door is shown in Fig. 7.

Optimization of the laminate
The Ansys plug-in "Direct Optimization" offers several methods for single and multi-objective optimization for FRP problems: Screening (based on sampling and sorting), NSGA-II-based generic algorithm, NSGA-II-based adaptive multi-objective method (restricted to continuous input parameters), nonlinear programming using quadratic Lagrangian (SQP, suitable for local refinement of single objectives with continuous parameters), mixed integer sequential quadratic programming (MISQP), and gradientbased single objective method for continuous parameters [14]. Multiple mutually conflicting objective functions can also be considered to find reasonable weighting factors for subsequent single-objective optimization and to develop a suitable design.
The fiber material, the fiber angle, the number of layers and the layer sequence can be defined as input variables for the optimization. For the objective functions, a choice can be made between component weight and deformation, local or global stresses and strains, or local or global failure. Due to the relatively simple laminate design with minimum laminate thickness, the input variables were limited to fiber direction and layer stacking order and, for the objective function, to part global deformation. MISQP was selected as the optimization method for its ability to solve mixed-integer nonlinear problems with a modified gradient-based single objective method for continuous parameters. Assuming that integer variables have a continuous influence on the model, sequences of quadratic approximations are applied. These support the computation of a single output parameter objective with multiple constraints [15].

Results of the laminate design
During the design and optimization process, a preferred solution was developed with a basic laminate structure consisting of a FF UD (FlaxTape 110) or FF fabric (Flax-Ply BD 150) and partial reinforcements of CF TFP patches and CF scrim (Zoltek TMP PX35)/fabric (BD Style 462) in heavily loaded areas. In Fig. 8, the deformations of the CFRP and bio-based car doors are compared. With a deformation of 17.415 mm, the optimized door is stiffer by a factor of 1.24 (Fig. 9). The deformation pattern is also more homogeneous.
The calculated biomass fraction of the developed cabin door is only about 29 wt% due to the use of a 19 wt% biobased resin component of the EP system, synthetic rigid foam inserts and structural adhesives. With a fully bio-based EP system and bio-based adhesives and sealants, a biomass content of more than 70 wt% could be achieved.
The failure behaviour was analysed with the failure criterion according to PUCK 2D (Fig. 9) [16]. The puck constants were taken from CFRP, since no characteristic values could be determined for flax fiber composites yet. Table 2 illustrates the parameters used in the FE simulation for Puck's fracture criterion. It has to be noted that the results are only of limited significance due to the assumed pressure-based characteristic values and correction factors. For example, lower puck constants and smaller compression-induced interfiber break spaces have been determined in [8] for comparable unidirectional flax fiber laminates. Nevertheless, the failure pattern, especially in the area of the narrowest circumferential cross section (enhanced by TFP patches), could be improved significantly. No convergence could be achieved for failure as an objective function for optimization. This was due to excessive local stresses in the area of the bearing.
In addition to mechanical stress due to (fluid) mechanical loads, the ULH is subject to high vibration stress. The improvement of the damping properties is among the main reasons for the use of flax fibers [13]. To illustrate this material advantage, an additional FE-based modal analysis was performed in this work. According to the specifications, resonances with 1 to 8 times the rotor and motor frequencies have to be eliminated for the cabin door (Table 3). A tolerance of ± 3% applies in this case. In addition to avoiding resonances, low eigenfrequencies are also to be aimed at for acoustic reasons.
In the FE modal analysis, the basic frequencies of the first 20 modes were checked for resonance with the (multiple) rotor and motor frequencies. As soon as the basic frequency corresponded to ± 3% of the excitation frequencies, resonance was assumed. For the CFRP cabin door, the eigenfrequencies occur at 5 modes and for the bio-based variant only at 3 modes (Table 4).
Particularly in the higher modes, the damping properties of the flax fibers have a clear influence on the vibration behavior (see Fig. 10). Here, the basic frequencies are significantly below those of the reference cabin door made of CFRP. At this point, however, it has to be noted that the "soft" PMMA window contributes significantly to the vibration behavior and influences it more strongly than the fiber composite structure.

Validation of the FEM results
For the production of the bio-based cabin door, laminating molds were developed for the VARI or prepreg process and made from an Ebalta EP 978 foam material suitable for autoclave production. To reduce thermal stress during production, the molds are hollow on the inside. Due to the high infiltration resistance of flax fiber semi-finished products and the resulting slow flow front progress, a cascadetype infiltration system was made. The use of a Compolex 150 RF3 flow aid additionally accelerated the infiltration process. The cabin door was produced using the SuperSap INR/INS resin system in the aVARI process. The two halfshells were joined directly in the laminating mold using an adhesive Scotch Weld DP 490 2 K and specially made rigid foam inserts. The PMMA window pane was then inserted using a UV-resistant polyurethane hybrid sealant 0.225 Reduction factor s (0 < s < 1) 0.5 Reduction factor M (0 < M < 1) 0.5 Interface weakening factor 0.8   Sikaflex 521. To prevent local failure at the hinge attachment a PVC foam core (80 g/mm 2 ) were bonded in the places of the hinges, which could minimally influence the measurement results. Figure 11 shows the completed biobased cabin door. For demonstration purposes, the surface was partially painted.

Mechanical testing of the cabin door
Based on the simulation results and design features, a realistic test setup (Fig. 12) was developed and simulated using FEM analysis. In real operation, the cabin door is supported in the helicopter cabin at three points, with all translational and rotational degrees of freedom in the xand y-directions being locked at the hinge. When closed, the cabin door is locked at positions 3 and 4 in Fig. 12 by a manually operated locking mechanism (two locking bolts). Only the degree of translation along the bolt axis and the degree of rotational freedom are retained at the locking positions. During the test, the lower corner of the cabin door (Position 1) was deflected via a bonded force application element and a hinge joint by means of a tensile/compression testing machine by 20 mm or 30 mm [corresponding to a safety factor (SF) of 1.5] in the z-direction. The resulting force was recorded with a load cell. The test was repeated 10 times in each case, an average value was calculated from the measurement results and compared with the calculated resultant force. During the test, the deformations at the stress-critical point with the smallest profile cross section (reinforced by TFP patches) were recorded three-dimensionally using an ARAMIS 3D optical deformation measurement system (Fig. 13). The measurement results have been used to verify the FE results.

Results of the structural test
The force-displacement curve of the bio-based cabin door shows a reversible linear behaviour and a good agreement with the FE results (Table 5). With required deflection of the bottom corner of 20 mm, a mean force of 108.62 N results. Calculated value is 119 N, which is 10% higher than measured. Thus, at 20 mm deflection, the hybrid car door is only 10% more compliant than calculated, indicating good accuracy of the simulation.
Despite the less stiff behavior of the bio-based cabin door, an excellent qualitative and, above all, quantitative course of the displacements in the critical area (smallest profile cross section) can be observed (Fig. 14). The deviations from simulated values are between 7% and 12%. It can be concluded that the deformation of the cabin door is numerically well-represented and that the requirements have been met.

Summary and outlook
The design of the bio-based cabin door of the UHL CoAx2D was performed using a newly developed routine for deriving the TFP geometry with programs Hyperworks and Ansys

Fig. 13
Detection of surface displacements with an optical 3D measuring system ARAMIS Workbench using a combination of free sizing, bundle sizing and shuffle optimisation. The primary optimisation objective variable was the deformation of the unattached lower car door corner of 20 mm, which is defined in the course of the aerodynamic load specified in the specifications and a wind gust of 10 m/s acting laterally outward. The FE results were validated by mechanical testing of the fabricated biobased cabin door. For this purpose, a realistic test setup was developed and simulated based on the attachment and locking of the cabin door in the helicopter cabin. During component testing, the displacements of the component surface at selected points were recorded using an optical 3D measuring system (Aramis) and compared with the FE results.
As a result, a compromise solution with a basic laminate structure consisting of a flax fiber scrim/fabric and partial reinforcements of CF TFP patches and CF scrim/fabric in heavily stressed areas was worked out after comprehensive material and technological consideration. With an actual weight of 985 g (without glued-in PMMA windows) and a biomass content of around 29 wt%, the hybrid bio-based cabin door is 255 g heavier than the reference variant made of CFRP. This is still within the permissible total weight defined in the specifications. In terms of ecological assessment, the use of bio-based raw materials can generate considerable added value, especially in the raw material and component manufacturing phase, but the associated increase in weight can lead to a higher CO 2 emission value in the long service life of the components. One conceivable approach to weight reduction is to use a lighter pre-impregnated flax fiber scrim (UD prepreg) instead of the "dry" semi-finished products. The high substitution potential of flax fibers and potentially achievable CO 2 savings compared to CFRP make the development of higher performance, fully bio-based semi-finished products a sensible and promising option.
Despite the added weight, the developed bio-based cabin door shows a better deformation and, according to FEM calculations, a damping behaviour at critical resonance frequencies in the range of 400-850 Hz that is about 30% more favourable. Furthermore, the measurement results confirm that topography optimization is well-suited for the design of TFP reinforcements with simple variable fiber design. In contrary, the downstream optimizations of the fiber direction, the discrete layer thicknesses as well as the stacking order of the base laminate structure did not improve the deformation behavior or the minimisation of the component weight any further. Considering the reasonable computational effort, this design methodology for developing of TFP reinforcements will be further pursued and optimised. Regarding the nonlinear, irreversible material behavior of UD flat fiber laminates under tensile/compressive loading as well as the fracture behavior of hybrid CF-FF laminates, numerous further essential questions remain to be addressed.  Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.