Additive manufacturing process design for complex aircraft components

Certain aspects of the metal additive manufacturing (AM) processes depend to some extent or other on the use of a heat source. This can be in the form of an electron beam, a laser or an arc or it can even depend on the process of raw material supply. There are two approaches to AM, specifically the powder bed fusion (PBF) and direct energy deposition (DED) processes, that can be used for manufacturing metal parts with no requirements for a secondary casting or infiltration process. Both processes can be used for manufacturing metal components with near-full density [1]. PBF generates a three-dimensional part by means of a layer-by-layer laser fusion of metallic powders [2] and can be split into two sub-processes, where these are defined principally by the source of fusion, by the use of either an electron beam or a laser [3]. Electron beam powder bed fusion (EBPBF) uses a high energy electron beam to melt sequential layers of a metal powder to lead to the manufacture of solid components of the required density. This category of part has to be fabricated in a vacuum due to the physical requirements of the electron beam. AM systems of the sort just described usually require a larger feedstock powder size, and this results in higher surface roughness. The AM build time is, however, generally lower because of the use of high power and rapid scanning for the electron beam [4, 5]. Another AM process is termed laser powder bed fusion (LPBF) and this is increasingly common in industry, thanks to its capability for high resolution and high-quality part production. LPBF employs at least one high power laser to melt and solidify the layers of powdered metal. This procedure is based on a powder feedstock delivered to and then coated on a platform, at which point the powder is melted and solidified in a precise manner under the control of a computer-aided design (CAD) model. Key parameters, such as laser power and layer thickness in particular, are user controllable for a particular material and application.

DED technologies operate in a different manner to PBF, requiring local deposition of the material stock through the use of a powder or wire-feed directly into the melt pool, noting that this can be initiated by a high energy in the form of a controlled electric arc, a suitable laser or a high energy electron beam. DED can be used to manufacture very large components because it does not need a powder bed. It should be clarified that DED generally produces components at lower resolution compared to those obtainable from powder bed–based technologies [4]. AM in general is now popular within the aerospace industry for the production of critically important metallic parts [6] due to the inherently reduced lead time and costs, and also the fact that complex lightweight geometries of high performance can be reliably produced through these manufacturing technologies. AM offers an inherently high-speed capability for the manufacture of replacement parts, and can be significantly faster than other manufacturing processes, with excellent resources available for localised heat treatment, surface hardening and alloying, the precision removal of material, repairs and high-performance prototyping [7]. In addition to this, AM provides the designer with options for integrated lightweight design, and a consequential reduction in the number of system components [8]. Processes of metal AM are inherently different to the more conventional subtractive machining, using feedstocks of some form to build the component up in a layer-by-layer manner, and employ powders or wires melted or fused by heat sources. These materials are then solidified to generate the required geometry on the basis of a heat source path that can be defined within a digital process [9]. AM capabilities can be optimised for specific applications and such processes are now standardised within ISO/ASTM 52,900 [10]. We note that PBF is the principal technique for the production of metallic parts [11]. Current research shows that the AM part orientation has a big influence on the properties of the final part and the amount of the support structure needed. Furthermore, the selected part orientation affects the support contact area, surface roughness and build time, as well as the cost of the manufactured parts [12]. Support structures can be perceived as a waste in the AM process since they consume powder and energy [13], and the build time is increased as well as the cost of the part due to the additional material and the removal process [14]. The support structures should be analysed by a draft angle and potentially modified to minimise the amount of material and manufacturing costs involved [15]. The main purpose of creating support structures is to assist printability in the manufacturing process. A 3D model with different geometries with holes or edge features, for example, will need a support structure for a successful fabrication since the support structure is necessary to keep the printed parts in the desired position.

In the aerospace industry, the most appropriate process for fabricating aluminium parts is the selective laser melting (SLM), a layer-by-layer process, which will be involved at this process design stage. The fabrication of smart structures using the AM technique can allow the flexibility of embedding sensors within a structure without compromising the structure and/or its functionality. Conversely, using traditional manufacturing by placing sensors increases the possibility of disturbing the normal operation of systems due to the sensor placement design [16]. There are two parameters that have most attention in current studies: laser power and scan/laser speed. In [17], a matrix of power and speed settings was created in order to understand the relationship between power and speed for metal powder bed fusion processes. They varied power from 50 to 200 W and speed from 100 to 250 mm/s. For stainless steel, the optimum weld bead was found between 150 and 200 W and 175 and 225 mm/s. Wang et al. investigated the effect of power and speed machine in SLM of aluminium and its alloys. The laser scan speed was varied from 1455 to 182 mm/s, whilst the laser power was reduced from an optimum value of 320 to 80 W in 6 steps of 40 W each, to maintain a constant energy density of 55 J/mm³. They showed that strength and ductility of the samples decrease gradually when both laser power and laser scan speed are decreased [18].

LPBF technique leads to residual stresses that are tensile near the top and compressive near the bottom of the part. If the part is removed without caution, it will bend excessively. To avoid such displacements, a stress-relieving heat treatment may be applied. The scope of the heat treatment is to relax stresses and to reduce them, whilst fixing the part displacements. Modelling of the heat treatment has been investigated. In the current research, a beam is manufactured using PBF at five different temperatures and for six different dwell times. It is demonstrated that the displacement clearly decreases when the temperature is increased, and the ramp-up stage plays an important role in the final deflection. However, during long heat treatments at temperatures above 750 °C, the deflection increases [19]. In addition, there are studies on heat treatments that consist of more than one stage. For example, heat treatment experiments in two stages at different temperatures were carried out to verify the mechanical properties of AA 7050 [20]. For aluminium alloy 7050, other researchers investigated different durations of heat treatment to obtain a refined and homogeneous microstructure: 2 h at 420 °C and then 8 h at 470 °C before being quenched in water; and heat-treated 30 min at 470 °C, quenched in water and artificially aged at 121 °C for 16 h [21]. Moreover, the heat treatment modifies the microstructure in order to adjust the mechanical properties of a material to match its function in an application. Since SLM produces microstructures different from those of conventional manufacturing, the response to a particular heat treatment will not be similar to that of a conventionally manufactured part [22]. One parameter required for the application of a heat treatment in Simufact Additive™ is the surface heat transfer coefficient, which is the relation between the heat flux and the temperature difference. The heat transfer coefficient depends on material properties, cooling medium properties and the initial temperature and cooling time, amongst other factors [23]. For the work discussed in this paper, natural convection will be considered, with a heat transfer coefficient of h = 5 W/(m²·K) as constant value [24].

For critical applications that require high elongation, the HIP treatment is a reliable approach for improving ductility, and it allows the optimisation of SLM parameters [25]. HIP is used to eliminate pores in the material and to release residual stresses [26], as well as to densify materials due to the achieved degree of homogenisation of the alloy composition, although there is no general scheme for the implementation of a HIP treatment since it depends highly on the application [27]. For powder metallurgy parts, combining high temperature and high pressure has shown advantages such as consolidation of material properties by creep and diffusion, healing of internal voids and strength improvement. The HIP cycle depends on the parameter temperature and pressure parameters as well as the dwell time. Both ramp-up and ramp-down slopes, also known as heating and cooling rates respectively, are determined by the maximum temperature and the dwell time of the treatment. The pressure is gradually increased, and when the temperature reaches its maximum, the pressure is then set to the maximum pressure [28‐31]. For AA7050 powder, an HIP treatment was proposed between 250 and 260 °C at a pressure of 92 MPa [32], noting that the most common applied pressure in HIP treatment for aluminium alloy is 100 MPa [33, 34]. In general, there are no specific requirements for temperature, pressure, dwell time and heating and cooling rates, since these parameters depend on the application. In the literature, a wide range of temperatures (up to 2000 °C) are being investigated [33]. HIP units at maximum of 1 GPa have been already manufactured [35, 36], whilst heating and cooling rates up to 25 °C/min [37] and 3000 °C/min have been used respectively [28, 38]. The current research focuses on the ribs and spars as the most relevant components in the aircraft wing box. There are already investigations on ribs using AM techniques. For example, a rib is proposed using a specific technique known as wire arc additive manufacturing (WAAM), which manufactures components layer-by-layer using welding technology controlled by a CNC machine. This technique combines an electric arc as the heat source and a wire as the feedstock. The advantages of this method are in cost savings when manufacturing high value parts, and in shortened lead times and reduced waste material when compared to conventional manufacturing [39]. However, a rib with IoT capabilities that can be used in an Industry 4.0 platform for data collection and processing has not been considered. There is therefore a need to explore various ways of integrating sensory capability into manufacturing design processes to improve reliability of aircraft components. Embedding sensors within a desired location leads to the ability of an end-user to monitor parameters such as high temperature and pressure in specific critical regions.

The Simufact Additive™ software is based on the finite element method (FEM) and has been used in this work to examine the manufacturing process for wing ribs to assess the displacements and the stresses. In this article, the SLM technique was used within PBF in AM. Details are presented in the “Results and discussion” section.

This work is structured into the following sections: “Introduction”, “Materials and methods”, “Results and discussion”, “Conclusions and further work”, and References.

The simulation of the AM process for the wing rib was carried out in Simufact Additive™ 2021. The AM technique used to generate samples in this paper includes PBF (SLM process) since it is the most appropriate process for the fabrication of aluminium parts in the aerospace industry.

Simufact Additive™ is a piece of commercial software developed especially for PBF AM simulations. It allows the analysis of four techniques: SLM, electron beam melting (EBM), laser beam melting (LBM) and direct metal laser sintering (DMLS). It has the advantage of providing the necessary functionality to simulate the entire AM process: printing, support cutting, heat treatment, hot isostatic pressing and the evaluation of residual stresses and distortions. The voxel mesh created by Simufact Additive™ is only offered as hexahedral element and once this mesh has been created, a volumetric structure can be built, providing a 3D network. Finally, a surface mesh is generated, noting that both the voxel and surface meshes complement each other to guarantee that the whole part is covered.

Before simulating the manufacturing process for complex aircraft components using the wing rib as an example, a generic manufacturing process framework for these groups of components is designed that represents the stages the part goes through. The flowchart for this process is shown in Fig. 1.

The outcomes of the above research show that the maximum displacement and maximum equivalent stress after a conventional heat treatment process still remain high for the wing rib. Therefore, a HIP process was applied to reduce the displacements and relieve stresses, and a reduction in maximum displacement of 86.5% was achieved. For the cover’s manufacturing process, the HIP treatment did not show any significant influence on the displacement and the stress: though for the HIP process there is a reduction of 8% in the maximum displacement. In this case, the cost-effective manufacturing process for the covers remains the conventional heat treatment. For both parts, sets of simulations were carried out by varying the heating and cooling rates, dwell time, maximum temperature of the heat treatment and maximum pressure of the HIP.

Figures 16 and 17 show the designed heat treatment processes for the wing rib and the covers.

In order to reduce costs, a support optimisation was investigated in Simufact Additive™ for the covers and the wing rib. After the support optimisation, the mass of the main wing rib support was significantly reduced by 91.24%: from 10.22 to 0.895 kg. The other supports for the sensory locations on the wing rib reduced their mass significantly by around 52% and 80%. For the covers, the mass was reduced by 40.89%. Table 8 shows an overview of the support properties before and after the support optimisation.

Considering the support optimisation, the previous designed HIP process (Fig. 16) and the conventional heat treatment (Fig. 17) were applied to the wing rib and covers, respectively. The simulations led to the following results (Table 9).

For the sensory cover, the maximum displacement was reduced by 20.20% and the maximum equivalent stress was increased by 7.58% as a result of the support optimisation. In the wing rib case, the maximum displacement decreased by 4.66% whilst the maximum equivalent stress increased by 14%.

After the support optimisation, the maximum total displacement for the wing rib is located on the leading edge, as shown in Fig. 18, whilst the maximum equivalent stress is on the trailing edge, as clarified in Fig. 19.

The simulations of the covers result in a maximum displacement and equivalent stress located on the outer surface of the cover (Figs. 20 and 21).

The convergence of the simulation was checked considering the point for which the displacement is maximum (Fig. 22). The displacement reaches a constant value after around 18,000 s of CPU time. The value of the maximum equivalent stress is also reached after this time. The convergence analysis for the sensory cover shows that the displacement reaches a constant value after around 7000 s of CPU time, whilst the value of the maximum equivalent stress was reached after around 10,000 s of CPU time (Fig. 23).

The manufacturing time was estimated by Simufact Additive™ in 31 h for the wing rib and 3 h for the sensory cover. These times were confirmed by a company that provided a quotation for printing small-scaled parts in aluminium. This company is a 3D-printing supplier based in Spain and provided 23.5 manufacturing hour for the wing rib and 30 min for the sensory cover. Due to the reduced dimensions, a HIP for the wing rib, which lasts 4.5 h in Simufact Additive™, was not necessary. In addition, the heat treatment of 2.5 h for the sensory cover was not considered.

The scope of this work was to redesign the manufacturing process for wing ribs by considering embedded sensory capabilities and AM technologies. In order to design an optimal and cost-effective manufacturing process, a heat treatment procedure was investigated and proposed by varying the cooling and heating rates, the dwell time, the maximum temperature of the heat treatment and the maximum pressure of the hot isostatic pressing process. For the wing rib, the conventional heat treatment did not allow a significant reduction of displacement and stress, whilst the HIP process achieved a significant reduction of maximum displacement. For the covers, a conventional heat treatment process was selected since the HIP did not bring much improvement in displacement and stress. In addition, the support structure for both the wing rib and the covers was optimised. The results showed a significant mass and volume reduction: the mass and the volume of the main rib support were reduced by 91.24%, whilst the support of the sensory cover achieved a reduction by 52.71% in mass and 40.89% in volume. Besides, the maximum displacement of both sensory cover and rib was reduced after support optimisation: 20.20% and 4.66% respectively.

Further work will deal with the implementation of the IoT capability into a wing rib structure manufactured using rapid prototyping for academic purposes. Recommendations can be further investigated by combining more than one thermal cycle considering different heating and cooling rates, dwell times, heat treatment temperatures and pressure variations. In addition, the thermal profile (treatment cycles) could be designed considering first the HIP process instead of a conventional heat treatment.