Deep Drawing Behaviour of Steel–Glass Fibre-Reinforced and Non-Reinforced Polyamide–Steel Sandwich Materials

: Thermoplastic-based ﬁbre metal laminates (FMLs) have gained increasing interest in the automotive industry due to their forming potential—especially at higher temperatures—into complex components compared to thermoset-based ones. However, several challenges arise while processing thermoplastic-based FMLs. One the one hand, forming at room temperature (RT) leads to early failure modes, e.g., fracture and delamination. On the other hand, warm forming can extend their forming limits, although further defects arise, such as severe thickness irregularities and wrinkling problems. Therefore, this study focuses on developing different approaches for deep drawing conditions to deliver a promising, feasible, and cost-effective method for deep-drawn FML parts. We also describe the defects experimentally and numerically via the ﬁnite element method (FEM). The FMLs based on steel/glass ﬁbre-reinforced polyamide 6 (GF-PA6/steel) are studied under different deep drawing conditions (temperatures, punch, and die dimensions). In addition, mono-materials and sandwich materials without ﬁbre reinforcement are investigated as benchmarks. The results showed that the best deep drawing condition was at a temperature of 200 ◦ C and a die/punch radius ratio of 0.67, with a gap/thickness ratio of ≤ 2.0. The FEM simulation via Abaqus 6.14 was able to successfully replicate the anisotropic properties and wrinkling of the GF-PA6 core in an FML, resembling the experimental results.


Sandwich Materials
In the 1980s, an FML based on resin matrix composites was developed, which was successfully used in the wings of the large passenger aircraft A380 due to the combined high strength of metal layers with the lightweight characteristics and high fatigue resistance of composites [1][2][3]. Their research and development have expanded in both the automotive industry and the aerospace sectors. Some of the traditional FMLs are GLARE ® (Al/glass-fibre-resin/Al), ARALL ® (Al/aramid-fibre-resin/Al), and CARALL ® (Al/carbon-fibre-resin/Al) [4], which mainly consist of aluminium alloy sheets and different fibre-reinforced resins. They are produced mostly by curing of the fibre-reinforced epoxy resin into panel structures. In industrial applications, their singular structures and time-consuming production processes limit their mass production and application in the automotive industry. In contrast to FML, another type of sandwich material containing thermoplastic polymer core without fibre reinforcement, i.e., metal/polymer/metal (MPM), was developed in the last two decades. These are lightweight and offer significant forming potential, as well as outstanding thermal and acoustic isolation properties. However, the reduced absolute strength and stiffness values-especially with the higher volume fractions of the core-are the main drawbacks. Alucobond ® , Dibond ® , Hylite ® , Litecor ® , and other examples of this category are mainly used in automotive structures such as the roof and door structures of the Volkswagen Lupo and Audi A2 [5]. In order to enhance the stiffness of these sandwiches while keeping their outstanding lightweight potential formability, thermoplastic-based FMLs have been developed [6,7]. For this purpose, different cover sheet materials have been applied, e.g., steel [8], which is gaining more and more attention due to its good formability and cost advantages. In contrast, thermoplastic polymer matrices such as PP (polypropene), PA6 (polyamide 6), PEEK (polyetheretherketone), and PU (polyurethane) have been considered, such as glass fibre-reinforced polymers (GFRPs) and carbon fibre-reinforced polymers (CFRPs). The Leika ® project, for example, developed a forming technology of thermoplastic-based FML containing steel/GF-PA/steel, which have weights reduced by around 25% compared to the use of monolithic steel sheets [9]. Other sheet-like thermoplastic-based FML products have been developed, such as CAPAAL ® (Al/carbon-fibre-PA6/Al) and CAPET ® (titanium/carbon-fibre-PEEK/titanium) [10].

Deep Drawing Behaviour of Sandwich Materials
In order to ensure the industrial applicability of sandwich materials, reliable and efficient sheet-forming techniques (herein, deep drawing) need to be developed and characterized. Deep drawing of monolithic sheet materials is well-established and widely used in the automotive and aerospace industries. In this case, an optimum holding force is applied in the flange region for the purpose of reaching the maximum drawing ratio. The stress-strain evolution is concentrated on the edge of the punch and the die, while the flange area is subject to a circumferential stress compression, which can lead to wrinkling or tearing in the case of unsuitable drawing parameters. Avoiding these defects can be achieved by adjusting the friction conditions, the drawing ratio, the die radius, the gap size, the process temperature, and the blank holding force [11]. In the case of deep drawing of sandwich materials, the situations are more complex. For MPM sandwich materials, studies have shown that deep drawing limits the decrease progressively with increasing core layer thickness [12][13][14]. This is due to the high tensile stresses exerted by the soft polymer core on the outer metal cover near the edge of the punch. This high tensile stress leads to an increase in the void volume fraction within the metal, which, in turn, eventually leads to earlier large crack formation and a reduction in the limit drawing ratio [13]. However, this is only a reduction relative to the monolithic sheet metal, and the formability of the MPM is still satisfactory at room temperature (RT) [14].
It is well-known that at RT, the mobility of the fibre reinforcement in the FML core layer is severely restricted in the matrix polymer, and that the low plastic deformation of the core layer can lead directly to a sharp increase in tensile stress on the outer steel sheet and premature cracking. Therefore, deep drawing using thermoforming (i.e., forming under controlled heating and cooling conditions) is considered an innovative approach for processing thermoplastic-based FML. Previous research has been limited to deep drawing of thermoset-based FML under semi-cured or non-cured conditions of the core layer in order to enable the fibre mobility, as reheating of the cured resin for forming is not possible. For instance, using a combination of metal deep-drawing and core curing during forming, the semi-cured FML exhibits better formability, where both the cover and core are simultaneously stamped and then joined [15]. However, due to the viscous flow nature of the resin, a stable flow condition during the forming process was not possible, resulting in irregular thickness distribution of the formed part. In addition, the support of the core layer for the metal cover during the deep-drawing process is also low due to the resin being in a liquid state, which can lead to wrinkling of the metal cover under circumferential compressive stresses. Therefore, a method of applying holding force in a stepwise manner to avoid such defects is introduced. Using semi-cured GLARE ® panels in this method ensured a better holding force effect in the initial step to reduce wrinkling, because in the earlier stage, the circumferential compressive stresses were mainly distributed in the flange area, which is prone to wrinkling. In the later stages, however, the radial tensile stresses dominated mainly in the wall area, which was susceptible to tearing. Thus, the pressure needs to be varied and adjusted during the deep-drawing process rather than remaining constant. For a semi-cured GLARE ® sheet with a blank diameter of 140 mm and a punch diameter of 75 mm, it can be deep-drawn to 30 mm without wrinkling or fracture by varying the holding force from 110 kN to 40 kN in two steps [16]. For the non-cured thermoplastic resin-based FML, a thermoformed laminate structure can be produced by combining polymer injection moulding, where the metal cover and fibres are formed first, then thermoplastic polymer is injected from the edges between the formed metal laminates at a high temperature and pressure, and, finally, the thermoplastic resin is cooled and polymerised to produce a 3D sandwich structure through a combination of deep drawing and thermoplastic resin transfer moulding (T-RTM) [17]. However, moulding techniques using semi-cured or non-cured resin-based FML require high equipment accuracy and control. Moreover, semi-cured resins still require 60-90 min at RT to fully cure at the end of the process. Thus, the efficiency of this approach does not meet the conditions for mass production.
Therefore, a new forming technology was introduced based on traditional sheet stamping technology combined with temperature-controlled forming, called thermoforming. For example, the deep drawing behaviour of flat thermoplastic-based FML semi-products based on steel/GF-PA6)/steel was studied [18], where the sample size was Ø140 mm and the deep drawing ratio was 1.9. The study showed that defect-free samples could only be obtained at very high holding forces of 200 kN to 300 kN for drawing depths of less than 17 mm. Due to the molten state of the core matrix and the reduced resistance to circumferential pressure between the laminates and the inner side of the thermoplasticbased FML, severe wrinkles were formed [18]. Further research was published using Al/unidirectional-GF-PA6/Al laminates while forming at a tool temperature of 270 • C (greater than the melting point at 220 • C of PA6) so that PA6 was squeezed out during deep drawing and the PA6 turned brown due to the thermal degradation at this temperature [19]. In addition, a removable base was added to the bottom of the die in order to ensure the bond between the layers at the bottom of the cup, thus ensuring the bond strength of the sample. However, wrinkles in the inner aluminium layer into the soft matrix of GF-PA6 could not be avoided. Wrinkling of the inner Al sheet on the sidewalls led to the accumulation and uneven thickness distribution of the GF-PA6 core. A number of similar studies were performed [20][21][22][23][24][25][26][27] wherein the formability of sandwich materials at RT and elevated temperatures through a combination of conventional sheet stamping and thermoforming was the focus. The influence of the core layer on the formability of thermoplastic-based FMLs, mainly strain evolution, and the resulting defects were analysed, highlighting the approaches intended to achieve improved deep-drawing results.

Aim and Structure of the Paper
The deep drawing technique is of practical importance for the application of thermoplasticbased FMLs in the automotive industry in the foreseeable future. The current research explores the deep drawing behaviours of thermoplastic-based FMLs consisting of steel covers and GF-PA6 (organosheets) under controlled heating and cooling conditions. This process faces the challenges of generating various defects, such as wrinkling and the complex causative factors leading to them, when compared to monolithic metal sheets. Therefore, the aim of this paper is to investigate approaches to improve these defects by setting different process parameters, such as temperature, holding force, cover/core thickness ratio, punch radius, and gap size. Furthermore, the mono-materials and MPM sandwich panels are investigated as benchmarks to characterise the influence of the core layer organosheet on forming thermoplastic-based FMLs.
For this purpose, a systematic investigation method based on ascending scales was used. The paper is structured as follows: Firstly, some basic results on the tensile properties of thin steel sheets, such as anisotropy, are discussed. Secondly, some of the preparatory work is presented, including sandwich panel production, tool design, and the determination of lubrication conditions. Thirdly, an experimental study discussing the deep-drawing behaviours of each mono-material, MPM panel, and thermoplastic-based FML panel, including the force and strain evolutions, is carried out sequentially. Finally, the failure modes, including cover sheet wrinkling, thickness irregularities, delamination, and fibre fracture, are described and verified using the finite element model regarding the formation of the core layer.

Materials and Sandwich Production
In this study, two thicknesses (0.3 mm and 0.4 mm) of electrolytic-galvanised steel sheets were used as cover sheets for the production of sandwich panels, namely, TS290 and TS275, which were supplied by Thyssenkrupp Steel Europe (Duisburg, Germany). These unalloyed low carbon steel grades are typical of those used in the food packaging industry and, more recently, also in the automotive industry. For the non-reinforced sandwich panels, PA6 was used as the core material (supported by Infiana Germany GmbH & Co. KG, Forchheim, Germany), the thickness of which could be varied through stacking and thermal fusing. The organosheet (GF-PA6 from Lanxess Deutschland GmbH, Köln, Germany) was used for the preparation of a semi-finished thermoplastic-based FML, Tepex (abbreviated as RG), containing a fibre volume fraction of 47 vol-% (the fibres were equally distributed in the weft and wrap directions, i.e., 50 vol-% fibres in each weft and warp direction and weaving style of twill 2/2). This material was supplied by Lanxess. A summary of the materials used is given in Table 1. The mechanical properties of the mono-materials were firstly determined at a strain rate of 0.008 s −1 (gauge length of 80 mm) via the video extensometer, as this affects their later forming behaviour. For instance, the anisotropy of the steel sheet influences the strain evolution pattern of the sandwich panel and failure location under deep drawing conditions, e.g., the earing effect. In order to predict and explain this forming behaviour, it is necessary to determine the mechanical anisotropy of the monolithic steel sheets. For this purpose, tensile tests were carried out on TS275 and TS290 at RT in three directions (0 • , 45 • and 90 • ) to the rolling direction (RD), and their results are shown in Figure 1a. It can be observed that the strength levels of TS275 were similar in all three directions, indicating a higher degree of isotropic behaviour. In contrast, the strength levels of TS290 showed clear differences in the three directions, i.e., the strengths measured in the 0 • and 90 • directions were slightly higher than that in the 45 • direction. Moreover, the mechanical properties along RD of the monolithic steel sheets decreased with the increasing temperature (see Figure 1b), as expected. elastic modulus was 18 GPa in the warp (0°) and weft (90°) directions, respectively. Moreover, the tensile behaviour of the RG core layer in the 45° direction showed higher elongation at failure as well as reduced strength, mainly due to the fact that the fibres were more susceptible to shear deformation in the 45° direction. This phenomenon has been described previously [28]. The soft core PA6 has a 200 % elongation at failure and a tensile strength of 73 MPa, showing rather good formability. This study focuses on the deep drawing of sandwich panels, which were developed arbitrarily for a systematic understanding of the different material combinations and process parameters on their deep drawability. Therefore, firstly, sandwich panels were produced. In this case, the mono-materials utilized (cover, core, and adhesive agent) were prepared as shown in Figure 2. The core material was simply cleaned with acetone and dried at 80 °C for at least 12 h, while the steel sheets were cleaned, ground (grid size of 60: abbreviated as G_60) and tempered (1 min at 440 °C: abbreviated as HT_440 °C/1 min); then, the adhesion promoter SI-Coating was spread onto them and activated for 3 min at 250 °C. This preparation scenario was recommended based on a previous study [29]. Subsequently, the sandwich panels were produced in this study by hot-pressing. It is important to note that for the production of FML, the RG organosheets were stacked with their 0°/90° directions in the RD of the steel sheets. They were placed between the two hot press plates equipped with heating cartridges and water cooling. The temperature and pressure were monitored and adjusted continuously, keeping the hot-pressing conditions at 245 °C, 3 bar, and 5 min. After consolidation by cooling under pressure for 5 min, the sandwich panels were demoulded at approx. 80 °C. In contrast to the production of FML, the production of MPM (without fibre reinforcement) required the insertion of additional spacers between the press plates to avoid excessive squeezing and spillage of PA6. The thickness of the spacers was generally 0.1 to 0.2 mm thicker than the total initial thickness of the MPM due to the thermal expansion and swelling of the PA6. Under these conditions, sufficient bonding forces were achieved, i.e., lap shear strength values of 20 to 25 MPa for the MPM samples [29]. The entire production process is shown schematically in Figure 2. The produced 200 × 200 mm² square semi-finished products were finally punched into round pieces 180 mm in diameter in the punching machine for further sheet testing. In addition, the hot press had a working area of 400 × 350 °C, in which a uniform As for the core material, more attention was paid to the organosheet RG, which showed a linear elastic behaviour. The ultra-tensile strength was about 340 MPa, and the elastic modulus was 18 GPa in the warp (0 • ) and weft (90 • ) directions, respectively. Moreover, the tensile behaviour of the RG core layer in the 45 • direction showed higher elongation at failure as well as reduced strength, mainly due to the fact that the fibres were more susceptible to shear deformation in the 45 • direction. This phenomenon has been described previously [28]. The soft core PA6 has a 200% elongation at failure and a tensile strength of 73 MPa, showing rather good formability.
This study focuses on the deep drawing of sandwich panels, which were developed arbitrarily for a systematic understanding of the different material combinations and process parameters on their deep drawability. Therefore, firstly, sandwich panels were produced. In this case, the mono-materials utilized (cover, core, and adhesive agent) were prepared as shown in Figure 2. The core material was simply cleaned with acetone and dried at 80 • C for at least 12 h, while the steel sheets were cleaned, ground (grid size of 60: abbreviated as G_60) and tempered (1 min at 440 • C: abbreviated as HT_440 • C/1 min); then, the adhesion promoter SI-Coating was spread onto them and activated for 3 min at 250 • C. This preparation scenario was recommended based on a previous study [29]. Subsequently, the sandwich panels were produced in this study by hot-pressing. It is important to note that for the production of FML, the RG organosheets were stacked with their 0 • /90 • directions in the RD of the steel sheets. They were placed between the two hot press plates equipped with heating cartridges and water cooling. The temperature and pressure were monitored and adjusted continuously, keeping the hot-pressing conditions at 245 • C, 3 bar, and 5 min. After consolidation by cooling under pressure for 5 min, the sandwich panels were demoulded at approx. 80 • C. In contrast to the production of FML, the production of MPM (without fibre reinforcement) required the insertion of additional spacers between the press plates to avoid excessive squeezing and spillage of PA6. The thickness of the spacers was generally 0.1 to 0.2 mm thicker than the total initial thickness of the MPM due to the thermal expansion and swelling of the PA6. Under these conditions, sufficient bonding forces were achieved, i.e., lap shear strength values of 20 to 25 MPa for the MPM samples [29]. The entire production process is shown schematically in Figure 2. The produced 200 × 200 mm 2 square semi-finished products were finally punched into round pieces 180 mm in diameter in the punching machine for further sheet testing. In addition, the hot press had a working area of 400 × 350 • C, in which a uniform temperature field was determined to minimise the influence of the temperature gradient on the mechanical properties of PA6. More results regarding the cooling rate and its influence on the recrystallization of PA6 were published in the previous study [30]. Fla sandwich panels were prepared for the following deep drawing test, for which the influence of the different coefficients of thermal expansion of the mono-materials on the shrinkage was ignored. The relevant results for sandwich panel production via the hot press and the analysis on its structural properties via a top-hat profile were published in the previous study [28]. Appl. Sci. 2023, 13, x FOR PEER REVIEW 6 of 28 temperature field was determined to minimise the influence of the temperature gradient on the mechanical properties of PA6. More results regarding the cooling rate and its influence on the recrystallization of PA6 were published in the previous study [30]. Fla sandwich panels were prepared for the following deep drawing test, for which the influence of the different coefficients of thermal expansion of the mono-materials on the shrinkage was ignored. The relevant results for sandwich panel production via the hot press and the analysis on its structural properties via a top-hat profile were published in the previous study [28].

Figure 2.
Surface treatments of cover and core layers and the production scheme of the sandwich panels by means of hot pressing.

Tool Design and Lubrication
For this research, the test conditions for sheet deep drawing are summarised in Table  2. The diameter of the blank and punch, the punch radius, and the test speed were kept constant, while the die radius, holding force, and forming temperature were varied. Two dies with radii of ten and fifteen millimetres and separate inner diameters were prepared (see Figure 3). In order to obtain the predesigned temperature, the test machine was modified; the blank holder was equipped with heating cartridges to provide the required forming temperature. In addition, the samples, punch, and die were preheated in an external furnace to 20 °C above the predesigned temperature to compensate for the temperature loss occurring during their transfer. It should be mentioned that these three tool components were separated from the machine by means of insulating ceramics to avoid excessive heating of the machine and to maintain the temperature in the sample. The forming temperatures were adjusted at 235 ± 5 °C and 200 ± 5 °C; herewith, (quasi) isothermal warm forming could be assumed and considered. For further results, the real-time temperature progress is depicted in [31].

Tool Design and Lubrication
For this research, the test conditions for sheet deep drawing are summarised in Table 2. The diameter of the blank and punch, the punch radius, and the test speed were kept constant, while the die radius, holding force, and forming temperature were varied. Two dies with radii of ten and fifteen millimetres and separate inner diameters were prepared (see Figure 3). In order to obtain the predesigned temperature, the test machine was modified; the blank holder was equipped with heating cartridges to provide the required forming temperature. In addition, the samples, punch, and die were preheated in an external furnace to 20 • C above the predesigned temperature to compensate for the temperature loss occurring during their transfer. It should be mentioned that these three tool components were separated from the machine by means of insulating ceramics to avoid excessive heating of the machine and to maintain the temperature in the sample. The forming temperatures were adjusted at 235 ± 5 • C and 200 ± 5 • C; herewith, (quasi) isothermal warm forming could be assumed and considered. For further results, the real-time temperature progress is depicted in [31]. In addition, early defects such as tears and wrinkles in the deep-drawing process can be minimised or even avoided if suitable lubrication conditions are present. Therefore, different lubrication conditions were tested for cold and warm deep drawing: 1.
In Figure 4a, using a pure thermoplastic film (TP film: 0.05 mm), the drawing behaviour of a monolithic TS290 was significantly limited, with a maximum drawing depth of 15 mm at RT.

2.
With the addition of grease in the setting shown in Figure 4b, the drawing depth was slightly improved after reaching 25 mm. Moreover, waviness on the steel sheet was observed following the configuration in Figure 4c.
1. In Figure 4a, using a pure thermoplastic film (TP film: 0.05 mm), the drawing behaviour of a monolithic TS290 was significantly limited, with a maximum drawing depth of 15 mm at RT. 2. With the addition of grease in the setting shown in Figure 4b, the drawing depth was slightly improved after reaching 25 mm. Moreover, waviness on the steel sheet was observed following the configuration in Figure 4c.

3.
A further lubricant configuration was tested, as shown in Figure 4d, and led to a significant improvement, as the drawing depth reached 50 mm. This was the lubrication setting used for further cold deep-drawing experiments. 1. In Figure 4a, using a pure thermoplastic film (TP film: 0.05 mm), the drawing behaviour of a monolithic TS290 was significantly limited, with a maximum drawing depth of 15 mm at RT. 2. With the addition of grease in the setting shown in Figure 4b, the drawing depth was slightly improved after reaching 25 mm. Moreover, waviness on the steel sheet was observed following the configuration in Figure 4c.

3.
A further lubricant configuration was tested, as shown in Figure 4d, and led to a significant improvement, as the drawing depth reached 50 mm. This was the lubrication setting used for further cold deep-drawing experiments.  Figure 5 illustrates the force-displacement curve of the deep-drawn TS290 and FML for lubrication conditions shown in Figure 4a-f. In the absence of grease, the slope of the force-displacement curve was highest for the lubrication setting shown in Figure 4a. Despite the addition of grease in Figure 4b, the drawing depth was limited to 23 mm and the slope of the force-displacement curve was only slightly reduced compared to the setting shown in Figure 4a. Despite the high drawing depth and lower forming force of the setting in Figure 4c, waviness was observed in the sidewalls. The best lubrication condition was found for the setting given in Figure 4d, where TP was used at both of the upper and lower sides of the sample and a further grease film was added at the bottom side of the TP. The force plateau was due to the reduced area of the sheets between the blank holders with increasing drawing depth [32]. Moreover, the force evolution for the lubrication settings shown in Figure 4e,f is demonstrated in the diagram as well. The decreased force evolution of the monolithic steel sheet at 235 • C compared to that at RT was due to the reduced mechanical properties of the steel sheets. More details about the force evolution of the FML are described in the following chapters. As a criterion for successful drawing, a ≥40 mm depth of drawing with no cracks, wrinkles, or wavy defects was defined. Figure 5 illustrates the force-displacement curve of the deep-drawn TS290 and FML for lubrication conditions shown in Figure 4a-f. In the absence of grease, the slope of the force-displacement curve was highest for the lubrication setting shown in Figure 4a. Despite the addition of grease in Figure 4b, the drawing depth was limited to 23 mm and the slope of the force-displacement curve was only slightly reduced compared to the setting shown in Figure 4a. Despite the high drawing depth and lower forming force of the setting in Figure 4c, waviness was observed in the sidewalls. The best lubrication condition was found for the setting given in Figure 4d, where TP was used at both of the upper and lower sides of the sample and a further grease film was added at the bottom side of the TP. The force plateau was due to the reduced area of the sheets between the blank holders with increasing drawing depth [32]. Moreover, the force evolution for the lubrication settings shown in Figure 4e,f is demonstrated in the diagram as well. The decreased force evolution of the monolithic steel sheet at 235 °C compared to that at RT was due to the reduced mechanical properties of the steel sheets. More details about the force evolution of the FML are described in the following chapters. As a criterion for successful drawing, a ≥40 mm depth of drawing with no cracks, wrinkles, or wavy defects was defined.
For the strain analysis, a photogrammetry method was applied utilizing a dot pattern (1 mm diameter with 2 mm centre-to-centre distance) on the steel surface using the electrolytic marking system (EU-Classic, Östling Marking Systems GmbH, Solingen, Germany), applying 8 V and a frequency of 3.5 Hz for a precise etched dot pattern.

Experimental Test Plan
The focus of this research is the deep-drawing behaviour of FML. Prior to the experiment, it was necessary to carry out deep-drawing investigations on mono-materials, as well as sandwich panels, without fibre reinforcement and to use them as benchmarks for the FML. A brief abbreviation of the sandwich materials is necessary in order to clearly identify each material combination in the diagrams, and Table 3 shows the naming rules for the different sandwich material combinations. For instance, MPM 02 -PA0.5 refers to non-reinforced MPM based on TS290/PA6/TS290 with a steel thickness of 0.3 mm and a core layer thickness of 0.5 mm. The effect of the process parameters on the deep-drawing behaviour of the sheet materials was investigated. These included forming temperature, For the strain analysis, a photogrammetry method was applied utilizing a dot pattern (1 mm diameter with 2 mm centre-to-centre distance) on the steel surface using the electrolytic marking system (EU-Classic, Östling Marking Systems GmbH, Solingen, Germany), applying 8 V and a frequency of 3.5 Hz for a precise etched dot pattern.

Experimental Test Plan
The focus of this research is the deep-drawing behaviour of FML. Prior to the experiment, it was necessary to carry out deep-drawing investigations on mono-materials, as well as sandwich panels, without fibre reinforcement and to use them as benchmarks for the FML. A brief abbreviation of the sandwich materials is necessary in order to clearly identify each material combination in the diagrams, and Table 3 shows the naming rules for the different sandwich material combinations. For instance, MPM 02 -PA0.5 refers to non-reinforced MPM based on TS290/PA6/TS290 with a steel thickness of 0.3 mm and a core layer thickness of 0.5 mm. The effect of the process parameters on the deep-drawing behaviour of the sheet materials was investigated. These included forming temperature, core thickness, die radius, and blank holding force. The experimental plan is shown in Table 4.  x: refers to the chosen material combination. The explanation for "01" and "02" is found in footer of Table 3.

Finite Element Modelling
To comprehend the deformation behaviour of organosheet RG and FML under deepdrawing conditions, finite element analysis via Abaqus/Explicit was introduced. The plastic response of steel-grade TS275 was modelled by assuming isotropic hardening, but in this case, the damage criterion to simulate the fracture behaviour was not given. The plastic properties of TS275 were extrapolated according to Swift law. Figure 6 shows the relationship between flow stress and plastic strain for TS275 at RT and 235 • C, as well as the other properties, namely, the density (ρ), E-modulus (E), Poisson' ratio (v), and yield strength (YS). The corresponding data were obtained from the experimental stress-strain curves in Figure 1b. Moreover, the influence of the strain rate on the mechanical properties of the steel sheet could be ignored. The difference between the tensile strengths from 0.04 s −1 to 0.004 s −1 was minor for this kind of mild steel (TS275), a result which was published in the previous study [33]. The test speed for the deep-drawing (0.5 mm/s) and tensile tests (0.64 mm/s) were at the approximate magnitude, same for the characterization of organosheet and PA6. In this study, the simulation of deep-drawn FML was performed at 200 • C, where the difference in mechanical properties between 200 • C and 235 • C was neglected. More results on the mechanical characterization of this organosheet and PA6 regarding the temperature were published in [34], in which the same organosheet was considered. Thus, RG was modelled as an anisotropic elastic material with nine engineering parameters, E 1, E 2 , and E 3 (E-moduli in their respective directions); v 12 , v 23 , and v 13 (Poisson ratios in their respective directions); and G 12 , G 23 , and G 13 (Shear-moduli in their respective directions), whose values at RT and 200 • C were obtained from [33,34] (see Table 5). The density of RG was 1.8 g/cm 3 . Table 5. Mechanical properties of RG at RT and 200 • C [33,34]. the interface or purely in the first or the second shear direction, respectively [35]. For the damage evolution criterion, , , and represent the work done by the tractions and their conjugate relative displacements in the normal, first, and second shear directions, respectively, and is the total energy release rate. Moreover, and 0 specify the effective displacement at complete failure and that relative to the effective displacement at the initiation of damage, respectively; is the energy dissipated due to failure; and is a material parameter [35].   Furthermore, the interfacial contact behaviour between the layers of FML was modelled using three separate approaches. First, it was modelled using cohesive elements (COH3D8) based on nominal stresses and energies, which were defined in terms of traction separation, i.e., mixed-mode delamination (see Figure 7). The maximum normal stress damage initiation criterion (Maxs) and the Benzeggagh-Kenane (BK) damage evolution criterion were used, respectively. For the damage initiation criterion in this figure, t n , t s , and t t represent the normal and the two shear tractions; and t 0 n , t 0 s , and t 0 t represent the peak values of the nominal stress when the deformation was either purely normal to the interface or purely in the first or the second shear direction, respectively [35]. For the damage evolution criterion, G n , G s , and G t represent the work done by the tractions and their conjugate relative displacements in the normal, first, and second shear directions, respectively, and G T is the total energy release rate. Moreover, δ f m and δ 0 m specify the effective displacement at complete failure and that relative to the effective displacement at the initiation of damage, respectively; G c is the energy dissipated due to failure; and η is a material parameter [35]. The adhesion strength between the cover and core layers decreased sharply wit increasing temperature, as the higher temperatures increased the overall ductility o adhesive, making crack propagation much smoother [36]. The properties of the coh elements in this study were defined based on the results presented in previous st [37,38], in which the peak traction stresses in the normal and shear modes were red by 83 % each when the temperature increased from RT to 110 °C for an epoxy-base hesive. The aim was to replicate and validate the model, where after the deep-dra The adhesion strength between the cover and core layers decreased sharply with the increasing temperature, as the higher temperatures increased the overall ductility of the adhesive, making crack propagation much smoother [36]. The properties of the cohesive elements in this study were defined based on the results presented in previous studies [37,38], in which the peak traction stresses in the normal and shear modes were reduced by 83% each when the temperature increased from RT to 110 • C for an epoxy-based adhesive. The aim was to replicate and validate the model, where after the deep-drawing experiments, the adhesion was partially deteriorated. In a perfect case, the parameters for such cohesive zone modelling must be determined for the current material combinations; this point is currently under investigation in the framework of the follow-up DFG project, number 330043166. In this study, the deep-drawing behaviour of FML was simulated below the melting point of the PA6 matrix in the organosheet RG, i.e., 200 • C. The material properties for the cohesive elements in the model are summarized in Table 6. The size of the cohesive element was set to 1 mm and the thickness to 15 µm. As the adhesion promotor was based on PA6, the density of the cohesive layer was set to be equal to PA6, i.e., 1.08 g/mm 3 . In addition, the interaction between the FML cover and core layer could also be assumed to follow a tangential friction behaviour, taking into account the possible relative sliding between the cover and core layers at 200 • C, which was similar to the relative sliding that occurred in thermoplastic-based FML composed of Al/GF-PA6/Al when the melting point of the core layer PA6 was exceeded, i.e., the friction coefficient between Al and PA6 varied between 0.1 and 0.9 at temperatures above 220 • C [34]. This is influenced by temperature, pressure, and sliding velocity. In this study, Columbian friction coefficients µ of 0.5 and 0.9 were considered to define the frictional behaviour in the model. Finally, a tie constraint was applied between the cover and core layers, assuming no relative sliding between the cover and core layers at 200 • C. The simulation results of different interaction conditions of the layers are compared with the experimental results at the end of this paper. The different conditions for defining the contact behaviour between the cover and core layers are summarized in Table 7. Similarly, the damage fracture behaviour of the RG core was not considered in the current case.

Tie Constraint Coulomb Friction Model Cohesive Element Model
Set 1 x Set 2 x : µ = 0.5 Set 3 x: µ = 0.9 Set 4 x * x: performed simulations with different contact settings, *: performed with the parameters in Table 6.
The dimension of the round sample in the simulation model was 180 mm in diameter. The upper surface of the die in the model was fixed, and a holding force was applied to the lower surface of the holder. The sample was compressed by the holder under this load, and once the punch began to move, the sample was able to slide inwards between the die and the holder (see Figure 8). The element type of the sample was chosen as Hex by default, and the meshing was performed using the sweep technique, which includes 8-code solid element C3D8 with an element size of 2 mm, for which the mesh analysis was converged [31] The punch was idealised as an analytical shell and the die was a discrete rigid body (R3D4) with a mesh size of 5 mm; the die edge area was refined to a 1 mm element size to avoid unsmooth contact between the tool and the sample. Interactions between the sample and the model were defined as general contact interactions, set with a Coulomb friction coefficient of 0.05. and the meshing was performed using the sweep technique, which includes 8-code solid element C3D8 with an element size of 2 mm, for which the mesh analysis was converged [31] The punch was idealised as an analytical shell and the die was a discrete rigid body (R3D4) with a mesh size of 5 mm; the die edge area was refined to a 1 mm element size to avoid unsmooth contact between the tool and the sample. Interactions between the sample and the model were defined as general contact interactions, set with a Coulomb friction coefficient of 0.05.

Deep Drawing of the Mono-Materials
According to the experimental plan, the deep-drawing behaviour of the monolithic steel sheets was first analysed in order to further determine the influence of the core layer PA6 and RG on the forming behaviour of the steel sheets in the sandwich panels. Figure  9a shows the force evolution of TS290 for deep-drawing conditions with different forming temperatures and holding forces. It can be seen that the drawing force decreased as the temperature increased and the holding force decreased, as could be expected. In this respect, the drawing behaviour of TS275 was similar to that of TS290.

Deep Drawing of the Mono-Materials
According to the experimental plan, the deep-drawing behaviour of the monolithic steel sheets was first analysed in order to further determine the influence of the core layer PA6 and RG on the forming behaviour of the steel sheets in the sandwich panels. Figure 9a shows the force evolution of TS290 for deep-drawing conditions with different forming temperatures and holding forces. It can be seen that the drawing force decreased as the temperature increased and the holding force decreased, as could be expected. In this respect, the drawing behaviour of TS275 was similar to that of TS290. In Figure 9b, the strain evolution on both sides of TS290 at RT with major ( 1 ) and minor ( 2 ) strains converging to 1 = − 2 was similar to the pure shear deformation behaviour in the Nakajima test [39]. By means of the FLC curve for TS290 at RT (marked with a yellow curve), it can be judged that the strain distribution on the steel sheet remained within the safety zone up to a drawing depth of 50 mm. This highlights the good and deep drawability of the steel sheet TS290, and that the strain evolution on both of its sides was almost identical. The FLC curves in Figure 9b, as well as the ones in the following figures, were all determined at a test speed of 1.5 mm/s in the previous study [31].
In addition, the anisotropy of TS275 and TS290, obtained earlier by tensile tests, was reflected by their strain evolution under deep-drawing conditions. Figure 10 shows the strain evolution along three cross-sections at 0°, 45°, and 90° through the sample's centre point. It can be seen that the strain distribution for TS275 was almost identical in the different directions (Figure 10a), as is similar to the previous tensile tests. In contrast, the strain distribution in the flange region for TS290 was significantly higher in the 45° direction (see Figure 10b). This was due to its anisotropy in the 45° direction causing more In Figure 9b, the strain evolution on both sides of TS290 at RT with major (ε 1 ) and minor (ε 2 ) strains converging to ε 1 = −ε 2 was similar to the pure shear deformation behaviour in the Nakajima test [39]. By means of the FLC curve for TS290 at RT (marked with a yellow curve), it can be judged that the strain distribution on the steel sheet remained within the safety zone up to a drawing depth of 50 mm. This highlights the good and deep drawability of the steel sheet TS290, and that the strain evolution on both of its sides was almost identical. The FLC curves in Figure 9b, as well as the ones in the following figures, were all determined at a test speed of 1.5 mm/s in the previous study [31].
In addition, the anisotropy of TS275 and TS290, obtained earlier by tensile tests, was reflected by their strain evolution under deep-drawing conditions. Figure 10 shows the strain evolution along three cross-sections at 0 • , 45 • , and 90 • through the sample's centre point. It can be seen that the strain distribution for TS275 was almost identical in the different directions (Figure 10a), as is similar to the previous tensile tests. In contrast, the strain distribution in the flange region for TS290 was significantly higher in the 45 • direction (see Figure 10b). This was due to its anisotropy in the 45 • direction causing more strain evolution and plastic deformation. with a yellow curve), it can be judged that the strain distribution on the steel sheet remained within the safety zone up to a drawing depth of 50 mm. This highlights the good and deep drawability of the steel sheet TS290, and that the strain evolution on both of its sides was almost identical. The FLC curves in Figure 9b, as well as the ones in the following figures, were all determined at a test speed of 1.5 mm/s in the previous study [31].
In addition, the anisotropy of TS275 and TS290, obtained earlier by tensile tests, was reflected by their strain evolution under deep-drawing conditions. Figure 10 shows the strain evolution along three cross-sections at 0°, 45°, and 90° through the sample's centre point. It can be seen that the strain distribution for TS275 was almost identical in the different directions (Figure 10a), as is similar to the previous tensile tests. In contrast, the strain distribution in the flange region for TS290 was significantly higher in the 45° direction (see Figure 10b). This was due to its anisotropy in the 45° direction causing more strain evolution and plastic deformation.  Figure 11a shows that as the organosheet RG thickness increased, the drawing force was roughly doubled. However, at a drawing depth of about 23 mm, a plateau in the force-displacement curve occurred for both thicknesses of RG, which was due to fibre  Figure 11a shows that as the organosheet RG thickness increased, the drawing force was roughly doubled. However, at a drawing depth of about 23 mm, a plateau in the force-displacement curve occurred for both thicknesses of RG, which was due to fibre fracture in the organosheet core, marked by yellow arrows in Figure 11b for the RG1.0 sample. By means of simulation, it can be observed that the RG1.0 was deep-drawn at F h = 45 kN and r d = 15 mm, and when the stroke reached 25 mm, RG1.0 in the flange and die edge area showed significant stress concentration. By increasing the holding force (100 kN) and reduction in the die radius (10 mm), significant wrinkling occurred due to the large circumferential compressive stress and the restricted plastic deformability. As no fracture criterion was defined for the RG core layer, only areas of stress concentration and wrinkling could be replicated in the simulation image, but fibre fracture could not, as shown in Figure 11c. However experimentally, fracture occurred as the fibres in the flange region were compressed and wrinkled, along with the compression failure of fibres and matrices such as fibre micro-buckling, matrix shear failure, or fibre failure, leading to a decrease in the punch force [40]. large circumferential compressive stress and the restricted plastic deformability. As no fracture criterion was defined for the RG core layer, only areas of stress concentration and wrinkling could be replicated in the simulation image, but fibre fracture could not, as shown in Figure 11c. However experimentally, fracture occurred as the fibres in the flange region were compressed and wrinkled, along with the compression failure of fibres and matrices such as fibre micro-buckling, matrix shear failure, or fibre failure, leading to a decrease in the punch force [40].

Deep Drawing of the Non-Reinforced MPM
According to the experimental plan, the effect of the core layer PA6 on the deep drawing of the sandwich panel was investigated subsequently. If only one variable, i.e., the core layer thickness, was varied, it could be observed that the drawing force increased with increasing core thickness. In addition, the MPM demonstrated good drawability in the range of 6 kN to 100 kN, with no remarkable defects such as wrinkling or cracking, indicating the large flexibility of the MPM's working window under deep-drawing conditions, as shown in Figure 12a. However, due to the high tensile stresses exerted by the soft polymer core on the outer steel sheet, the MPM showed differences in the strain evolution

Deep Drawing of the Non-Reinforced MPM
According to the experimental plan, the effect of the core layer PA6 on the deep drawing of the sandwich panel was investigated subsequently. If only one variable, i.e., the core layer thickness, was varied, it could be observed that the drawing force increased with in-creasing core thickness. In addition, the MPM demonstrated good drawability in the range of 6 kN to 100 kN, with no remarkable defects such as wrinkling or cracking, indicating the large flexibility of the MPM's working window under deep-drawing conditions, as shown in Figure 12a. However, due to the high tensile stresses exerted by the soft polymer core on the outer steel sheet, the MPM showed differences in the strain evolution of the inner and outer steel sheets compared to that of the monolithic steel sheets. As shown in Figure 12b, the outer steel sheet of MPM 02 -PA0.5 gained further extension in the ε 2 =0 and ε 2 = 1 2 ε 1 directions, which corresponded to the plane strain and biaxial tension states in the Nakajima test, respectively [39]. In the ε 1 = −ε 2 direction, on the other hand, the strain evolution was similar to that of the monolithic steel sheet. This phenomenon became more pronounced as the thickness of the core layer increased, as can be seen in Figure 12c. It can be observed that the strain of the outer cover steel sheet in the ε 2 = 1 2 ε 1 direction for MPM 02 -PA1.0 exceeded its corresponding FLC curve, indicating a much higher probability of cracking. The MPM 02 -PA0.5, on the other hand, showed a strain distribution that was still within the safety zone compared to the corresponding FLC curve. Therefore, the formability of the MPM decreases with increasing core thickness.

Deep Drawing of the Non-Reinforced MPM
According to the experimental plan, the effect of the core layer PA6 on the deep drawing of the sandwich panel was investigated subsequently. If only one variable, i.e., the core layer thickness, was varied, it could be observed that the drawing force increased with increasing core thickness. In addition, the MPM demonstrated good drawability in the range of 6 kN to 100 kN, with no remarkable defects such as wrinkling or cracking, indicating the large flexibility of the MPM's working window under deep-drawing conditions, as shown in Figure 12a. However, due to the high tensile stresses exerted by the soft polymer core on the outer steel sheet, the MPM showed differences in the strain evolution of the inner and outer steel sheets compared to that of the monolithic steel sheets. As shown in Figure 12b, the outer steel sheet of MPM 02 -PA0.5 gained further extension in the 2 = 0 and 2 = 1 2 1 directions, which corresponded to the plane strain and biaxial tension states in the Nakajima test, respectively [39]. In the 1 = − 2 direction, on the other hand, the strain evolution was similar to that of the monolithic steel sheet. This phenomenon became more pronounced as the thickness of the core layer increased, as can be seen in Figure 12c. It can be observed that the strain of the outer cover steel sheet in the 2 = 1 2 1 direction for MPM 02 -PA1.0 exceeded its corresponding FLC curve, indicating a much higher probability of cracking. The MPM 02 -PA0.5, on the other hand, showed a strain distribution that was still within the safety zone compared to the corresponding FLC curve. Therefore, the formability of the MPM decreases with increasing core thickness. The strain distribution of a cross-sectional line through the centre pointof the drawn cups enabled further analysis of the effect of the core layer PA6 on the strain evolution of the sandwich panel. Firstly, the major strain values were compared for the drawing depths of 20 and 40 mm. It can be seen that at a drawing depth of 20 mm, the strain evolution first appeared in the area of the punch edge (in the range of approx. 50 to 150 mm along the section line). Subsequently, as the drawing depth increased to 40 mm, the strain evolution in the die edge region developed rapidly (see Figure 13a). In contrast, by comparing the strain evolution of TS290 at both drawing depths (25 mm and 50 mm), it can be seen that The strain distribution of a cross-sectional line through the centre pointof the drawn cups enabled further analysis of the effect of the core layer PA6 on the strain evolution of the sandwich panel. Firstly, the major strain values were compared for the drawing depths of 20 and 40 mm. It can be seen that at a drawing depth of 20 mm, the strain evolution first appeared in the area of the punch edge (in the range of approx. 50 to 150 mm along the section line). Subsequently, as the drawing depth increased to 40 mm, the strain evolution in the die edge region developed rapidly (see Figure 13a). In contrast, by comparing the strain evolution of TS290 at both drawing depths (25 mm and 50 mm), it can be seen that the maximum major strain values always appeared in the die edge region and doubled as the stroke increased (see Figure 13b). Due to the high tensile stresses exerted by the soft polymer core, the strain evolution of the MPM increased in the punch-edge region and decreased in the die-edge region [13]. Combining Figure 13a,c, it can be concluded that at a 20 mm stroke, the main strain evolution was concentrated at the punch edge, while the corresponding strain distribution was in the ε 2 = 1 2 ε 1 direction. It was the acceleration of the strain evolution around the die radius as the stroke was increased that led to the rapid expansion of the strain distribution along the ε 1 = −ε 2 direction in Figure 13c. It can be concluded that the effect of the core layer PA6 on the outer steel sheet of the sandwich is mainly concentrated on the punch edge, i.e., the biaxial tension along the ε 2 = 1 2 ε 1 direction. Its influence is secondary to the pure shear deformation and strain evolution of the cover steel sheet at the die edge.
tion of the inner and outer steel sheets of MPM 02 -PA0.5; and (c) strain evolution of the outer steel sheet of MPM 02 with different core thicknesses and their corresponding FLC curves.
The strain distribution of a cross-sectional line through the centre pointof the drawn cups enabled further analysis of the effect of the core layer PA6 on the strain evolution of the sandwich panel. Firstly, the major strain values were compared for the drawing depths of 20 and 40 mm. It can be seen that at a drawing depth of 20 mm, the strain evolution first appeared in the area of the punch edge (in the range of approx. 50 to 150 mm along the section line). Subsequently, as the drawing depth increased to 40 mm, the strain evolution in the die edge region developed rapidly (see Figure 13a). In contrast, by comparing the strain evolution of TS290 at both drawing depths (25 mm and 50 mm), it can be seen that the maximum major strain values always appeared in the die edge region and doubled as the stroke increased (see Figure 13b). Due to the high tensile stresses exerted by the soft polymer core, the strain evolution of the MPM increased in the punch-edge region and decreased in the die-edge region [13]. Combining Figure 13a,c, it can be concluded that at a 20 mm stroke, the main strain evolution was concentrated at the punch edge, while the corresponding strain distribution was in the 2 = 1 2 1 direction. It was the acceleration of the strain evolution around the die radius as the stroke was increased that led to the rapid expansion of the strain distribution along the 1 = − 2 direction in Figure 13c. It can be concluded that the effect of the core layer PA6 on the outer steel sheet of the sandwich is mainly concentrated on the punch edge, i.e., the biaxial tension along the 2 = 1 2 1 direction. Its influence is secondary to the pure shear deformation and strain evolution of the cover steel sheet at the die edge.

Deep Drawing of the FML
At RT, the drawing force of the FML under deep-drawing conditions decreased with the increasing die edge radius, and the slope of the force-displacement curve increased significantly with increasing core thickness. Basically, the higher slope refers to higher strengthening potential, as shown earlier through the uniaxial tensile test in Figure 1. However, the maximum drawing depth and the corresponding maximum drawing force decreased, as expected [41], as shown in Figure 14a. With a thicker core layer, the tensile stress applied to the outer steel sheet at the punch edge was greater, as was the cracking probability, as shown in the sample image. Adjusting the process parameters and the

Deep Drawing of the FML
At RT, the drawing force of the FML under deep-drawing conditions decreased with the increasing die edge radius, and the slope of the force-displacement curve increased significantly with increasing core thickness. Basically, the higher slope refers to higher strengthening potential, as shown earlier through the uniaxial tensile test in Figure 1.
However, the maximum drawing depth and the corresponding maximum drawing force decreased, as expected [41], as shown in Figure 14a. With a thicker core layer, the tensile stress applied to the outer steel sheet at the punch edge was greater, as was the cracking probability, as shown in the sample image. Adjusting the process parameters and the thickness ratio did not lead to a change in the maximum stroke of the FML, which was limited to less than 25 mm. The detailed strain evolution in each direction was analysed with the help of centreline cross-sections. Figure 15 shows the strain evolution of the FML for different material combinations. It can be observed that the final strain evolution in the flange area showed anisotropy for both isotropic (TS275) and anisotropic (TS290) cover sheets, with slightly higher strain values in the 45° direction. For the warm deep drawing of FML, at temperatures above the melting point of PA6 (220 °C), severe wrinkling on the inner cover sheet and an appearance of waviness on the outer side took place, as shown in Figure 16a. The reason for this was that the molten matrix in the core of the FML did not provide the cover with the ability to resist the circumferential compressive stress of the blank, as well as due to a larger gap size (3.3 mm) by = 15 mm. A large gap-to-sheet thickness ratio often leads to waviness during the deep drawing of the sheet [43]. Thus, in subsequent tests, improvements in the deepdrawability of the FML can be achieved by increasing the holding force (100 kN); thickening the covers; reducing the die radius and gap size. This can be substantially improved Furthermore, the strain evolution of the FML differed from that of the monolithic steel sheet and the MPM in that the anisotropy of the core RG led to directional strain evolution along the warp and weft at the punch edge, as shown in Figure 14b. Because the fibres were less mobile in the warp and weft directions, the tensile stresses applied to the outer steel sheet in the 0 • and 90 • directions were greater in comparison to the 45 • direction and, thus, more likely to cause cracks. The strain at the punch edge shifted to the ε 2 = 0 direction after a transient evolution along the ε 2 = 0.5ε 1 direction due to the stress concentration and necking of the cover steel sheet along the warp and weft directions of RG. Another difference is that the strain evolution in the flange area was also directional, with the strain evolution along the 45 • direction being higher than that in the 0 • and 90 • directions, due to the fact that the core layer was prone to shear deformation in the 45 • direction [28,42]. Moreover, the strain evolution of the outer steel sheet of FML already exceeded its corresponding FLC level (solid red line in Figure 14b) at RT, indicating the cracking of the steel.
The detailed strain evolution in each direction was analysed with the help of centreline cross-sections. Figure 15 shows the strain evolution of the FML for different material combinations. It can be observed that the final strain evolution in the flange area showed anisotropy for both isotropic (TS275) and anisotropic (TS290) cover sheets, with slightly higher strain values in the 45 • direction.
For the warm deep drawing of FML, at temperatures above the melting point of PA6 (220 • C), severe wrinkling on the inner cover sheet and an appearance of waviness on the outer side took place, as shown in Figure 16a. The reason for this was that the molten matrix in the core of the FML did not provide the cover with the ability to resist the circumferential compressive stress of the blank, as well as due to a larger gap size (3.3 mm) by r d = 15 mm. A large gap-to-sheet thickness ratio often leads to waviness during the deep drawing of the sheet [43]. Thus, in subsequent tests, improvements in the deep-drawability of the FML can be achieved by increasing the holding force (100 kN); thickening the covers; reducing the die radius and gap size. This can be substantially improved by reducing the forming temperature to 200 • C. Figure 16b shows the results in terms of drawing force and depth.
The increase in temperature allowed the FML to reach a stroke of 40 mm, and the punch was experimentally stopped at this displacement. The reduction in drawing force at 200 • C or 235 • C compared to RT was due to the softening of the core material and the consequent increase in fibre mobility in the PA6 matrix. Furthermore, the drawing force increased with the increasing core thickness at 200 • C and remained almost constant with the increasing core thickness at 235 • C. The drawing force at 200 • C (red lines) decreased more rapidly from the drawing depth of 25 mm compared to that at 235 • C (green and black lines). This was due to fibre fracture in the flange area; this phenomenon is discussed later. It should be mentioned that a sliding of the interlayer could occur at 235 • C, which is above the melting point of PA6. This is due to the thermoplastic nature of PA6 as well as the adhesion promoter based on PA6, which is degraded at high forming temperatures and forms adhesion strength again after cooling and recrystallization. The detailed strain evolution in each direction was analysed with the help of centreline cross-sections. Figure 15 shows the strain evolution of the FML for different material combinations. It can be observed that the final strain evolution in the flange area showed anisotropy for both isotropic (TS275) and anisotropic (TS290) cover sheets, with slightly higher strain values in the 45° direction. For the warm deep drawing of FML, at temperatures above the melting point of PA6 (220 °C), severe wrinkling on the inner cover sheet and an appearance of waviness on the outer side took place, as shown in Figure 16a. The reason for this was that the molten matrix in the core of the FML did not provide the cover with the ability to resist the circumferential compressive stress of the blank, as well as due to a larger gap size (3.3 mm) by = 15 mm. A large gap-to-sheet thickness ratio often leads to waviness during the deep drawing of the sheet [43]. Thus, in subsequent tests, improvements in the deepdrawability of the FML can be achieved by increasing the holding force (100 kN); thickening the covers; reducing the die radius and gap size. This can be substantially improved by reducing the forming temperature to 200 °C. Figure 16b shows the results in terms of drawing force and depth. The increase in temperature allowed the FML to reach a stroke of 40 mm, and the punch was experimentally stopped at this displacement. The reduction in drawing force at 200 °C or 235 °C compared to RT was due to the softening of the core material and the consequent increase in fibre mobility in the PA6 matrix. Furthermore, the drawing force increased with the increasing core thickness at 200 °C and remained almost constant with the increasing core thickness at 235 °C. The drawing force at 200 °C (red lines) decreased more rapidly from the drawing depth of 25 mm compared to that at 235 °C (green and black lines). This was due to fibre fracture in the flange area; this phenomenon is discussed later. It should be mentioned that a sliding of the interlayer could occur at 235 °C, which is above the melting point of PA6. This is due to the thermoplastic nature of PA6 as well as the adhesion promoter based on PA6, which is degraded at high forming temperatures and forms adhesion strength again after cooling and recrystallization.
As observed from the strain evolution analysis, the differences between warm deep drawing and that at RT were significant. The strain in the outer steel sheet developed in the 1 = − 2 direction; however, that in the inner steel sheet developed in the 1 = −2 2 direction, showing a uniaxial tension state as in Nakajima's test [39] (Figure 17a,b).
The higher strain evolution in the inner steel sheet was the opposite of that for MPM under deep-drawing conditions. The reason is that under warm deep drawing conditions, the inner steel sheet not only showed wrinkling, but also experienced higher tensile stresses caused by the fibre-reinforced core at the die radius edge (Figure 17c). In addition, the strain evolution on the outer steel sheet decreased slightly with the increasing temperature and decreasing core layer thickness at 200 °C (Figure 17d,e). Since the organosheet did not melt completely at 200 °C, it still stimulated tensile stresses on the steel sheets, leading to an increase in the strain evolution. This difference gradually disappeared as the temperature rose to 235 °C (Figure 17f). Due to the rapid mobility of the fibrous structure after melting of the PA matrix, their effect on the strain distribution in the outer steel sheet As observed from the strain evolution analysis, the differences between warm deep drawing and that at RT were significant. The strain in the outer steel sheet developed in the ε 1 = −ε 2 direction; however, that in the inner steel sheet developed in the ε 1 = −2ε 2 direction, showing a uniaxial tension state as in Nakajima's test [39] (Figure 17a,b). The higher strain evolution in the inner steel sheet was the opposite of that for MPM under deep-drawing conditions. The reason is that under warm deep drawing conditions, the inner steel sheet not only showed wrinkling, but also experienced higher tensile stresses caused by the fibre-reinforced core at the die radius edge (Figure 17c). In addition, the strain evolution on the outer steel sheet decreased slightly with the increasing temperature and decreasing core layer thickness at 200 • C (Figure 17d,e). Since the organosheet did not melt completely at 200 • C, it still stimulated tensile stresses on the steel sheets, leading to an increase in the strain evolution. This difference gradually disappeared as the temperature rose to 235 • C (Figure 17f). Due to the rapid mobility of the fibrous structure after melting of the PA matrix, their effect on the strain distribution in the outer steel sheet could be neglected. The corresponding FLC levels shown in Figure 17b,d were taken from the previous study, indicating that the typical strain evolution of FML at elevated temperatures was still in a safe forming zone. In addition, the strain evolution along the cross-sectional lines showed the differences in the various material combinations and forming temperatures. In contrast to the strain evolution at RT, the strain evolution at 235 °C was significantly extended, as shown in Figure 18a. Although the flange area of FML showed anisotropy at RT, FML 01 did not exhibit different strain values in either direction at 235 °C, i.e., reduced anisotropy and In addition, the strain evolution along the cross-sectional lines showed the differences in the various material combinations and forming temperatures. In contrast to the strain evolution at RT, the strain evolution at 235 • C was significantly extended, as shown in Figure 18a. Although the flange area of FML showed anisotropy at RT, FML 01 did not exhibit different strain values in either direction at 235 • C, i.e., reduced anisotropy and homogeneous strain distribution. In contrast, FML 02 still showed higher strain values at the die radius edge in the 45 • direction. This was mainly because TS290 itself is anisotropic, whereas TS275 is almost isotropic (see the tensile test results in Figure 1a). If the temperature was reduced to 200 • C, the strain evolution along the cross-sectional line differed slightly from that at 235 • C, showing a slightly greater strain evolution in the 45 • direction than in the 0 • and 90 • directions (see Figure 18b). It can, therefore, be concluded that at RT, RG plays a critical role in the formability and strain evolution on the cover of the FML. As the temperature increases, its influence diminishes. Until the temperature is above the PA6 melting point, the anisotropy of the organosheet has a negligible effect on the strain evolution on the steel sheet, which dominates at this point. homogeneous strain distribution. In contrast, FML 02 still showed higher strain values at the die radius edge in the 45° direction. This was mainly because TS290 itself is anisotropic, whereas TS275 is almost isotropic (see the tensile test results in Figure 1a). If the temperature was reduced to 200 °C, the strain evolution along the cross-sectional line differed slightly from that at 235 °C, showing a slightly greater strain evolution in the 45° direction than in the 0° and 90° directions (see Figure 18b). It can, therefore, be concluded that at RT, RG plays a critical role in the formability and strain evolution on the cover of the FML. As the temperature increases, its influence diminishes. Until the temperature is above the PA6 melting point, the anisotropy of the organosheet has a negligible effect on the strain evolution on the steel sheet, which dominates at this point.

Validation of the Simulation Results and Failure Modes
After analysing the force and strain evolution of the different materials under deepdrawing conditions, this section focuses on the defects that appear in the FML obtained numerically to be validated with the experiments. After raising the blank holder force to 100 kN and lowering the die radius to 10 mm, the FML 02 -RG0.5-235 °C (composed of the thin 0.3-mm TS290 steel) still showed severe internal wrinkling and waviness in the outer steel sheet (Figure 19a). However, these conditions showed significant improvements for FML 01 -RG0.5 using the thicker TS275 (0.4 mm) (Figure 19b). This means that with an increase in the cover/core thickness ratio together with the lower gap size between the punch and inner diameter of the die, the waviness of the outer steel sheet and the wrinkles of the inner steel sheet could be reduced, as the thicker cover sheets were able to withstand the circumferential compressive stresses due to their higher geometric stiffness. As the temperature was reduced below the melting point of PA6, i.e., at 200 °C, the wrinkling of the inner steel sheet was further reduced without a significant reduction in formability (see Figure 19c). Moreover, by increasing the core layer thickness at 200 °C, the wrinkling of the inner steel sheet was still present and no significant improvement was indicated (Figure 19d).

Validation of the Simulation Results and Failure Modes
After analysing the force and strain evolution of the different materials under deepdrawing conditions, this section focuses on the defects that appear in the FML obtained numerically to be validated with the experiments. After raising the blank holder force to 100 kN and lowering the die radius to 10 mm, the FML 02 -RG0.5-235 • C (composed of the thin 0.3-mm TS290 steel) still showed severe internal wrinkling and waviness in the outer steel sheet (Figure 19a). However, these conditions showed significant improvements for FML 01 -RG0.5 using the thicker TS275 (0.4 mm) (Figure 19b). This means that with an increase in the cover/core thickness ratio together with the lower gap size between the punch and inner diameter of the die, the waviness of the outer steel sheet and the wrinkles of the inner steel sheet could be reduced, as the thicker cover sheets were able to withstand the circumferential compressive stresses due to their higher geometric stiffness. As the temperature was reduced below the melting point of PA6, i.e., at 200 • C, the wrinkling of the inner steel sheet was further reduced without a significant reduction in formability (see Figure 19c). Moreover, by increasing the core layer thickness at 200 • C, the wrinkling of the inner steel sheet was still present and no significant improvement was indicated (Figure 19d). Appl. Sci. 2023, 13,   The wrinkling behaviour of the FML's inner cover sheet at 200 °C can be attributed to the failure of the core material, as illustrated in Figure 20, where fracture of the fibres at 200 °C at a drawing depth of 25 mm can be observed. Once fibre fracture occurred, the backing support for the inner steel plate was reduced in the fractured area and the inner steel sheet began to wrinkle. As can be seen in the middle image of Figure 20, fractures similar to those seen in monolithic organosheets when deep-drawn at RT appeared in the organosheet in the flange and die edge regions due to high circumferential compressive stresses. Considering the finite element simulation results, it can also be seen that the FML 01 -RG0.5 showed no coincident fibre wrinkles in the flange area in the warp or weft directions with the experimental results in the case of the contact condition of Set 1 in Table 7 (see Figure 21). However, by considering the tangential frictional contact behaviour (Sets 2 and 3) and the cohesive contact behaviour between the FML layers at 200 °C, all simulation results resulted in wrinkling in the warp and weft directions of the organosheet RG in the flange and die edge areas. Moreover, the degree of the wrinkling of contact Sets 3 and 4 The wrinkling behaviour of the FML's inner cover sheet at 200 • C can be attributed to the failure of the core material, as illustrated in Figure 20, where fracture of the fibres at 200 • C at a drawing depth of 25 mm can be observed. Once fibre fracture occurred, the backing support for the inner steel plate was reduced in the fractured area and the inner steel sheet began to wrinkle. As can be seen in the middle image of Figure 20, fractures similar to those seen in monolithic organosheets when deep-drawn at RT appeared in the organosheet in the flange and die edge regions due to high circumferential compressive stresses. The wrinkling behaviour of the FML's inner cover sheet at 200 °C can be a to the failure of the core material, as illustrated in Figure 20, where fracture of t at 200 °C at a drawing depth of 25 mm can be observed. Once fibre fracture occu backing support for the inner steel plate was reduced in the fractured area and t steel sheet began to wrinkle. As can be seen in the middle image of Figure 20, f similar to those seen in monolithic organosheets when deep-drawn at RT appear organosheet in the flange and die edge regions due to high circumferential com stresses.  Considering the finite element simulation results, it can also be seen that the FML 01 -RG0.5 showed no coincident fibre wrinkles in the flange area in the warp or weft directions with the experimental results in the case of the contact condition of Set 1 in Table 7 (see Figure 21). However, by considering the tangential frictional contact behaviour (Sets 2 and 3) and the cohesive contact behaviour between the FML layers at 200 • C, all simulation results resulted in wrinkling in the warp and weft directions of the organosheet RG in the flange and die edge areas. Moreover, the degree of the wrinkling of contact Sets 3 and 4 was more severe than that of Set 2 due to the higher stress values indicated in the simulation images.
By comparing the contact of Sets 3 and 4, a significant difference in the stress evolution in the punch edge area was observed, where the stress distribution was more homogeneous in the circular direction, but that of Set 3 showed obvious localization in the warp and weft directions. As the real contact between layers of the semi-finished FML panel could not be simplified as frictional sliding from the beginning of the warm deep-drawing process, the results of Set 4 through mixed-mode delamination of cohesive elements were found to be more reliable. Appl. Sci. 2023, 13, x FOR PEER REVIEW 23 of 28 was more severe than that of Set 2 due to the higher stress values indicated in the simulation images. By comparing the contact of Sets 3 and 4, a significant difference in the stress evolution in the punch edge area was observed, where the stress distribution was more homogeneous in the circular direction, but that of Set 3 showed obvious localization in the warp and weft directions. As the real contact between layers of the semi-finished FML panel could not be simplified as frictional sliding from the beginning of the warm deepdrawing process, the results of Set 4 through mixed-mode delamination of cohesive elements were found to be more reliable.  Table 7 at strokes of 25 mm and 40 mm.
As can be seen in Figure 22a, the cohesive element deletion first occurred at the side wall area at a drawing depth of 10 mm. By further increasing the drawing depth up to 25 mm, the deletion of cohesive elements propagated accordingly to reach the flange area. The laminate lost support from the tool first in the side wall area when the laminate was deep-drawn into the die. With further punch displacement, the flange area of the laminate flowed inwards. Plastic deformation and relative sliding could occur when the adhesion strength between layers was weakened due to the high temperature [37]. As no fibre fracture criterion was defined in the material modelling, the simulation results showed deviation from the experimental ones. Thus, wrinkling of organosheet RG, but not of the fibre fracture, in simulated conditions could be successfully validated by the experiments. By analysing the stress and strain evolution of the inner and outer steel sheets at a punch displacement of 40 mm, it can be seen that the stress evolution of inner and outer steel sheets was homogeneous, which is a different result from the anisotropic organosheet core RG (Figure 22b). Moreover, the strain evolution of RG was localized in the 0°, 45°, and 90° directions in the flange area (see Figure 22c). As was different from the experimental results in terms of the strain evolution of the cover sheets, the simulation showed the anisotropic strain evolution of the inner and outer steel sheets, which was due to the anisotropic behaviour of the core RG. In contrast, the experimental results showed more isotropic strain evolution in Figure 22c, which might have been due to the fibre fracture in the warp and weft directions. In addition, the flow behaviour of the cover sheets in the 0° (warp) and 90° (weft) directions at the flange and die edge area changed. Thus, the influence of anisotropic RG was minimized and the nature of TS275 itself dominated at 200 °C. To interpret the different strain evolution results between the experiments and the simulation, cross-sectional strain analysis was carried out as shown in in Figure 23a. As can be seen, the simulated strain evolution of the outer steel sheet in FML 01 -RG0.5 at 0° and 90° was significantly lower than that oneat 45° and at 200 °C, which was obviously different from the experimental one in Figure 18b. However, the simulation result of the monolithic steel sheet TS275 matched with the experimental one very well. In addition, the force  Table 7 at strokes of 25 mm and 40 mm.
As can be seen in Figure 22a, the cohesive element deletion first occurred at the side wall area at a drawing depth of 10 mm. By further increasing the drawing depth up to 25 mm, the deletion of cohesive elements propagated accordingly to reach the flange area. The laminate lost support from the tool first in the side wall area when the laminate was deep-drawn into the die. With further punch displacement, the flange area of the laminate flowed inwards. Plastic deformation and relative sliding could occur when the adhesion strength between layers was weakened due to the high temperature [37]. As no fibre fracture criterion was defined in the material modelling, the simulation results showed deviation from the experimental ones. Thus, wrinkling of organosheet RG, but not of the fibre fracture, in simulated conditions could be successfully validated by the experiments. By analysing the stress and strain evolution of the inner and outer steel sheets at a punch displacement of 40 mm, it can be seen that the stress evolution of inner and outer steel sheets was homogeneous, which is a different result from the anisotropic organosheet core RG (Figure 22b). Moreover, the strain evolution of RG was localized in the 0 • , 45 • , and 90 • directions in the flange area (see Figure 22c). As was different from the experimental results in terms of the strain evolution of the cover sheets, the simulation showed the anisotropic strain evolution of the inner and outer steel sheets, which was due to the anisotropic behaviour of the core RG. In contrast, the experimental results showed more isotropic strain evolution in Figure 22c, which might have been due to the fibre fracture in the warp and weft directions. In addition, the flow behaviour of the cover sheets in the 0 • (warp) and 90 • (weft) directions at the flange and die edge area changed. Thus, the influence of anisotropic RG was minimized and the nature of TS275 itself dominated at 200 • C. To interpret the different strain evolution results between the experiments and the simulation, cross-sectional strain analysis was carried out as shown in in Figure 23a. As can be seen, the simulated strain evolution of the outer steel sheet in FML 01 -RG0.5 at 0 • and 90 • was significantly lower than that oneat 45 • and at 200 • C, which was obviously different from the experimental one in Figure 18b. However, the simulation result of the monolithic steel sheet TS275 matched with the experimental one very well. In addition, the force evolution indicated that the simulation result of monolithic RG at RT and FML 01 -RG0.5 at 200 • C deviated significantly from the experimental results at drawing depths of 10 mm and 15 mm, respectively (Figure 23b). The reason may be the fact that the fibre degradation and damage criterion were not implemented in the simulation model. However, the simulation results coincided with the experimental result of monolithic TS275 very well. Based on the current simulation techniques, further modelling and simulation studies in terms of damage initiation and evolution of the organosheets are necessary, and will be performed in the extension of the current project. Moreover, the influence of temperatures and strain rate on the material properties could also be taken into account and embedded into the simulation model if necessary. Appl. Sci. 2023, 13, x FOR PEER REVIEW evolution indicated that the simulation result of monolithic RG at RT and FML 01 -RG 200 °C deviated significantly from the experimental results at drawing depths of 1 and 15 mm, respectively (Figure 23b). The reason may be the fact that the fibre deg tion and damage criterion were not implemented in the simulation model. Howeve simulation results coincided with the experimental result of monolithic TS275 very Based on the current simulation techniques, further modelling and simulation stud terms of damage initiation and evolution of the organosheets are necessary, and w performed in the extension of the current project. Moreover, the influence of temper and strain rate on the material properties could also be taken into account and embe into the simulation model if necessary. Finally, the thickness distribution and the macro-defects of the deep-drawn F different forming temperatures were qualitatively investigated. The thickness distrib was homogeneous at RT in the FML 01 -RG0.5 ( Figure 24a) and remained uniform forming at 200 °C, along with substantially improved formability, as shown in Figur However, at 235 °C, a significant thickness irregularity was observed, together with a ning of the core material in the punch edge and flange regions and a thickening in th area (Figure 24c). This was due to the flow of the softened and fluid polymer matrix caused problems such as wrinkling, waviness, and unavoidable thickness irregula Based on this, it can be stated that forming at a temperature of 200 °C (< melting po the polymer matrix) is more appropriate for ensuring good formability and improvi quality of the deep-drawn parts. With warm deep drawing, delamination can also after the cooling process if there is no counterpressure to the punch in the sample b region, as in Figure 24d. This is due to the lack of pressure at the die bottom as well different coefficients of thermal expansion of the core and cover layers, which can l a reduction in adhesion and separation while sawing the drawn cups.  Finally, the thickness distribution and the macro-defects of the deep-drawn FML at different forming temperatures were qualitatively investigated. The thickness distribution was homogeneous at RT in the FML 01 -RG0.5 ( Figure 24a) and remained uniform while forming at 200 • C, along with substantially improved formability, as shown in Figure 24b. However, at 235 • C, a significant thickness irregularity was observed, together with a thinning of the core material in the punch edge and flange regions and a thickening in the wall area (Figure 24c). This was due to the flow of the softened and fluid polymer matrix. This caused problems such as wrinkling, waviness, and unavoidable thickness irregularities. Based on this, it can be stated that forming at a temperature of 200 • C (< melting point of the polymer matrix) is more appropriate for ensuring good formability and improving the quality of the deep-drawn parts. With warm deep drawing, delamination can also occur after the cooling process if there is no counterpressure to the punch in the sample bottom region, as in Figure 24d. This is due to the lack of pressure at the die bottom as well as the different coefficients of thermal expansion of the core and cover layers, which can lead to a reduction in adhesion and separation while sawing the drawn cups.

Conclusions and Outlook
In this paper, the influences of the material properties (cover sheets: TS275-0.4 mm and TS290-0.3 mm, and core layers: PA6, RG), thicknesses of the core materials (0.5 and 1.0 mm), tool geometries (die radius and gap size), and forming temperatures (RT, 200 °C and 235 °C) on the deep-drawing behaviour of several thermoplastic-based FMLs are characterized, using mono-materials and MPM sandwich panels as benchmarks. Based on the obtained results, the following conclusions can be made:

•
The strain evolution of a steel sheet is influenced by its anisotropy. The effect of the core layer PA6 on the MPM strain evolution is mainly due to the tensile stress it exerts on the outer steel sheet at the punch edge, resulting in an increase in strain evolution along the 2 = 1 2 1 direction and leading to earlier failure with the increase in the core thickness;

Conclusions and Outlook
In this paper, the influences of the material properties (cover sheets: TS275-0.4 mm and TS290-0.3 mm, and core layers: PA6, RG), thicknesses of the core materials (0.5 and 1.0 mm), tool geometries (die radius and gap size), and forming temperatures (RT, 200 • C and 235 • C) on the deep-drawing behaviour of several thermoplastic-based FMLs are characterized, using mono-materials and MPM sandwich panels as benchmarks. Based on the obtained results, the following conclusions can be made:

•
The strain evolution of a steel sheet is influenced by its anisotropy. The effect of the core layer PA6 on the MPM strain evolution is mainly due to the tensile stress it exerts on the outer steel sheet at the punch edge, resulting in an increase in strain evolution along the ε 2 = 1 2 ε 1 direction and leading to earlier failure with the increase in the core thickness; • At RT, the deep-drawability of RG is limited by failure in the flange and die edge regions, attributable to in-plane tangential deformation and fibre compression fracture due to the acting circumferential compressive stresses; • At RT, the tensile stresses exerted by RG on the outer cover of FML exhibit directional and localised fracture occurrence at the punch edge due to the lack of fibre mobility and anisotropy; • At RT, the influence of the core layer RG on the strain distribution of the FML in the flange region dominates. This influence decreases with the increasing forming temperature until it is above the PA6 melting point, where the anisotropy of the cover steel sheet itself dominates; • At 235 • C, wrinkling of the inner cover of FML and the waviness of the outer cover can be reduced by increasing the holding force and the cover/core thickness ratio; reducing the die radius and gap size; as well as lowering the forming temperature (here, 200 • C) to just below the PA6 melting point; • The minor wrinkling of the inner cover of FML at 200 • C results from in-plane shear and fibre fractures of the RG in the warp and weft directions due to their limited fibre mobility compared to 235 • C; • Warm deep drawings at 200 • C and 235 • C can both achieve good improvement of the drawability of the thermoplastic-based FML with an increase in depth by at least 200%; however, the thickness distribution uniformity of FML at 200 • C is better than that at 235 • C.
Moreover, for deep drawing of thermoplastic-based FMLs, varying the process parameters of thermoforming has great potential to improve the product quality of FMLs. Further research regarding the cover/core thickness ratio and the selection of a suitable forming temperature are both possible future research directions. Of greater interest is the selection of fibre-reinforced polymers, especially their weaving style and fibre volume content, both of which have an influence on the mechanical properties and the mobility of the fibres in the molten state of the matrix. Furthermore, for the purpose of forming thermoplastic-based FMLs, reasonable damage models for the FML layers and a cohesive model for interlaminar interaction need to be developed to describe their fracture and delamination behaviours at different temperatures, which will play a decisive role in further numerical research; predicting their forming behaviours; and, thus, improving the thermoformability and final product quality of thermoplastic-based FMLs.