AFP Track Planning and Calculation Approach

AFP Tool Path Planning and Calculation Approach for Variable Stiffness Composite Manufacturing Variable stiffness composites (VSC) exhibit advanced mechanical properties compared to conventional carbon fiber reinforced plastics (CFRP). During the design phase, finite element analysis (FEA) is used to optimize fiber orientations that must be converted into feasible tool paths for automated fiber placement (AFP) operations; manufacturing defects such as wrinkles, gaps and overlaps must be addressed without exception. This article discusses a novel path planning algorithm with variable orientation in AFP. In this approach, fiber orientations are clustered and for each cluster a reference curve is fitted as a B-Spline taking into account manufacturing constraints. Case studies are presented to validate the discussed tool path planning approach. [caption id="attachment_157289" align="aligncenter"] Figure 1: AFP lay-up illustration4 / Illustration of AFP lay-up4[/caption] Weight reduction is indispensable for sustainability in high-value-added manufacturing sectors. Since its first commercial use in the 1960s¹, CFRP has offered high strength-to-weight ratios, making these materials excellent candidates for applications in aerospace, automotive and marine industries. In a life cycle analysis study of CFRP in the automotive sector conducted by Duflou², it was emphasized that these materials provide lower energy consumption compared to metal materials during intensive use. In CFRP production, composite layers for large parts such as aircraft wings are traditionally laid up manually, requiring significant time, cost and skill. To address such issues, AFP technology with the tow concept was introduced in the 1970s and deployed in commercial aviation³. A gantry-type CNC machine or robotic system is equipped with a fiber placement head specifically designed to lay down strips forming a series of bands and assembled to create a layer (see Figure 1⁴). [caption id="attachment_157290" align="aligncenter"] Figure 2: Composite material usage rate in aerospace. (https://www.machining4.eu/Technolotgy)/[/caption] In a recent study⁵, stress-oriented tool path planning in 3D printing was shown to lead to significant improvements in mechanical properties, and a similar approach is applicable in AFP, where the orientation and coverage of the part surface depend on the placement strategy. This relates to flexibility in design and manufacturing. Typical strategies are: (i) constant angle, (ii) geodesic and (iii) variable orientation. Constant angle orientation involves establishing a reference curve at a fixed angle to the desired direction on the part surface. Geodesic orientation is based on the shortest path between two points on the part surface and follows zero curvature. Variable angle orientation has the advantage that in the production of variable stiffness composites, the layup can be aligned with the load direction. [caption id="attachment_157292" align="aligncenter"] Figure 3: Conventional AFP lay-up approach⁴[/caption] In one of the early works in this field, Shirinzadeh et al.⁶ developed a feasible approach for generating constant angle paths. Here, a reference curve is created by projecting a fixed axis and then this curve is spread over a step distance depending on the local surface curvature and fiber strip width. Recently, VSC design has received attention with emphasis on optimizing placement orientation for weight reduction, with fiber directions being evaluated as a new design variable.⁷ However, variable orientation increases design complexity and the likelihood of manufacturing defects. Therefore, the conformity of the tool path to optimized placement orientations and manufacturability are critical for achieving desired mechanical performance.⁸ [caption id="attachment_157293" align="aligncenter"] Figure 4: Comparison of conventional and proposed methods. (a) constant angle lay-up (b) flow-directed lay-up (c) clustering-based lay-up⁴.*M: Manufacturability, D: Design suitability⁴[/caption] This article describes a novel method integrated with FEA for tool path generation with variable orientation in AFP. In this method, manufacturing constraints are applied during the path planning phase to keep defects under control. To realize VSC, the anisotropic mechanical properties of CFRP materials are utilized, where composite fibers are aligned in the load direction. This differs from conventional AFP tool path generation using fixed orientation placement, where the optimum obtained from FEA involves varying fiber orientations. Therefore, when variable orientation is ignored, part performance is not at its best level (see Figure 4). To optimize the mechanical structure of the part, it is possible to define objective functions such as strain energy, buckling or natural frequency. In this article, optimization was performed only as a case study according to the first natural frequency.

Tool Path Planning for AFP

In FEA, fiber orientations can be calculated according to several different patterns: (i) constant angle, (ii) flow line, (iii) fractional variable angle; each provides a balance between design optimization and manufacturability. For example, constant angle optimization (see Figure 4a⁴) has the highest manufacturability at the cost of reducing strength-to-weight ratio. Conversely, flow line fiber alignment supports design optimization but is difficult to manufacture (see Figure 4b⁴). Therefore, fiber orientations obtained from FEA must effectively address tool path planning to prevent manufacturing defects that have negative effects on mechanical performance. In this work, clustering-based tool path planning was used as shown in Figure 4c⁴, and its application is shown step by step in Figure 5.

Manufacturing Defects in AFP

In AFP, gap and overlap area percentage are used to evaluate part performance compared to design. The literature shows that gap values between 2% and 12% proportionally reduce buckling and in-plane stiffness, and the acceptable limit depends on the application. Lay-up defects typically arise from flaws such as surface coverage amount, lay-up head position, strip deviation, missing strips, lay-up speed and pressing force. In Figure 6a and Figure 6b, fiber courses of a layer laid according to parallel and shifted offset methods are shown. In these images, wrinkles, missing strips, overlaps and undulations are observed.

Experimental Demonstration

The proposed approach was demonstrated in the lay-up of the first layer of a multi-layer, simply supported plate (1000 mm x 500 mm) on a robotic AFP system located within Sabancı University and supplied by Coriolis© (see Figure 8). Fiber strips consist of thermoset CFRP material with 3000 filaments and weighing 150 g/m2, manufactured by KordSA©. Experimental validation was performed on the AFP system shown in Figure 8.

Tool Path Calculation

As an example case, the fiber vector distribution determined with a lay-up radius constraint of 0.5 m and aimed at increasing the first natural frequency to 16 Hz is shown in Figure 7a. Reference curves for each cluster and the generated lay-up paths are shown in Figures 7c and 7d, respectively. Finally, the calculated lay-up paths were converted to IGES file format for kinematic simulation in CAD-Fiber©, a commercial software. Validation and AFP Production Kinematic absence of gaps and overlaps in lay-up paths is the most important and first step for proper lay-up execution. Therefore, the initial validation was performed kinematically by comparing the actual product with CAD-Fiber©, an AFP software used commercially and in industry. As a result of this comparison, the maximum gap was measured as 0.29 mm as shown in Figures 9a and 9b, which is less than the ±0.50 mm specified in the technical requirements of the robotic AFP system. However, defects resulting from process parameters, robot positioning and speed cannot be defined in the software. Therefore, actual lay-up trials were also conducted. Close-up views of critical areas are shown in Figure 9. Gap area percentage measurements were performed using image processing, and the entire layer is shown in Figure 9e. The AFP system feeds composite strips 80 mm beyond the spools. Therefore, when the strips are cut, the last 80 mm section cannot be controlled. As a result, the laid strips deviate from the desired path and create gaps or overlaps of up to 2 mm (see Figure 9c). To resolve this issue, the dimensions of the desired geometry are extended by 80 mm as shown in Figure 9e. Here, the actual part geometry is the area within the white band boundaries. In measurements after lay-up, a maximum gap of 1.00 mm was observed, which corresponds to approximately 16% of the 6.35 mm tow width. In the region where gaps were most frequently observed, the average gap area percentage was calculated as 5%. Based on the obtained measurement results, it can be concluded that the variable lay-up direction path calculation approach can produce acceptable layers. However, no wrinkles or undulations were observed in the visual inspection of the layer. [caption id="attachment_157299" align="aligncenter"] Figure 9: Lay-up path validation, (a) validation with CADFiber software, (b) gap measurement, (c) close-up of critical areas, (d) areal gap percentage measurement by image processing, (e) full view of the laid layer⁴.[/caption]

Conclusion

This article describes an approach to tool path planning and calculation for applying AFP with variable lay-up orientation in the production of variable stiffness composite parts. Tool path calculation is performed using fiber orientations obtained from FEA, at which stage possible manufacturing defects such as gaps, overlaps and wrinkles have not yet been considered. Fiber orientations are clustered through normalized section segmentation⁹. In each cluster, a reference curve is fitted according to fiber orientations. Subsequently, strip paths are generated by parallel spreading of the reference curve subject to minimum placement length and maximum placement curvature. Thus, manufacturability and optimized fiber courses are calculated with an innovative approach in AFP application. Experimental demonstration and validation were performed in an example AFP lay-up process. In kinematic validation of the generated lay-up paths on the commercial AFP software CAD-Fiber©, maximum gaps of 0.29 mm were observed. However, where necessary, this gap value can be improved by using a finer mesh structure in FEA analysis. Subsequently, the approach proposed for the first time in the literature for variable orientation was validated in a real AFP placement process. Maximum gap was measured around 1.00 mm with a strip width of 6.35 mm, which corresponds to 16% of the strip width. Gap area percentage was measured at less than 5%.

Acknowledgments

This work was conducted as part of project number 218M715 under the TÜBİTAK 1003–Priority Areas Research Projects Program. Furthermore, the authors of the study thank KordSA for their support in the supply of composite fiber materials.   References: 1Dorey G (1987) Carbon fibres and their applications. Journal of Physics D: Applied Physics, 20/3:245. 2Duflou JR, De Moor J, Verpoest I, Dewulf W (2009) Environmental impact analysis of composite use in car manufacturing. CIRP Annals. 3Anderson RL, Grant CG (1991) Advanced fiber placement of composite fuselage structures. In NASA. Langley Research Center, First NASA Advanced Composites Technology Conference, Part 2. 4 Tunc, L. T., & Sheikhi, M. (2023). AFP tool path planning for manufacture of variable stiffness composites. CIRP Annals.5Li Y, Xu K, Liu X, Yang M, Gao J, Maropoulos P (2021) Stress-oriented 3D printing path optimization based on image processing algorithms for reinforced loadbearing parts. CIRP Annals, 70(1), 195-198. 6Shirinzadeh B, Cassidy G, Oetomo D, Alici G, Ang Jr MH (2007) Trajectory generation for open-contoured structures in robotic fibre placement, Robotics and Computer-Integrated Manufacturing, 23/4:380-394. 7Rasool M, Singha MK (2019) Stability of variable stiffness composite laminates under compressive and shearing follower forces, Composite Structures, 225, 111003. 8Brooks TR, Martins JR (2018) On manufacturing constraints for tow-steered composite design optimization, Composite structures, 204:548-559. 9Nik MA, Fayazbakhsh K, Pasini D, Lessard L (2014). Optimization of variable stiffness composites with embedded defects induced by automated fiber placement. Composite Structures, 107:160-166. 10Shi J, Malik J (2000). Normalized cuts and image segmentation. IEEE Trans. on Pattern Analysis and Mach. Intelligence,22/8:888–905.   Assoc. Prof. L. Taner Tunç Faculty of Engineering and Natural Sciences Manufacturing Research Laboratory Sabancı University