1 Introduction

Battery housings are a central element of modern electric vehicles. In pure electric vehicles, they are usually placed in the vehicle underbody. Although there is only a limited space available, high requirements are set to protect the sensitive battery cells inside. The battery housing must protect the cells from mechanical damage in all directions and meet high sealing requirements to shield the electrical system from moisture [1]. To be able to meet these requirements while at the same time offering a high level of packaging density and a low weight, current battery housings are often built in an aluminum assembly design. In this type of construction, the battery housing consists of a battery tray assembled from various aluminum parts and a flat or domed lid. The frame structure of the battery tray is mainly made of aluminum extrusions and for the bottom plate either flat extrusions or sheet metal is used. Flat extruded profiles provide additional protection against damage from below and offer the possibility of directly integrating the cooling required for temperature control of the cells [2]. However, the long joining seams between profiles pose challenges for conventional joining methods.

Conventional welding processes induce large amounts of heat that lead to large welding distortions [3]. Therefore currently, the friction stir welding (FSW) process is predominantly used to produce these leak-tight linear joints [4]. FSW is a slow and, consequently, expensive process [5, 6], which also requires very large machines for large components like battery trays. Furthermore, despite its advantages, distortion of the components can still occur due to the heat introduced [7].

Known mechanical joining processes such as clinching or self-pierce riveting are generally limited to localized, usually point-shaped joint formation. To enable the tightness of the joint zone over the entire profile composite, a linear joint is necessary.

While there are already a few longitudinal mechanical joining systems that are used in board walls of trucks, none of them currently fulfill the high sealing requirements that are necessary for battery housings [8]. These systems typically rely on force-fit and form-fit joints. To ensure tight joints, a material bond between the joining partners is desirable. Forming processes offer the possibility of producing load-bearing joints with low heat and distortion and, under certain conditions, create material bonds [9].

The first continuously linear joints produced by forming technology are known as roll joining, which allows the production of profiles from flat material [10]. However, continuous joining by means of rolls requires high support stiffnesses in the zone close to the joining process. The patent specification DE102016117524A1 addresses processes for producing materially bonded joints by means of adhesives [11]. In the context of battery tray production, adhesive joints in the floor assembly should be avoided, as they can be affected by high thermal stresses of subsequent welding processes used for attaching transverse or longitudinal load paths.

Based on the state of the art and the identified need for research, this research project favors the approach of implementing the linear joint by means of a single press stroke. The approach includes the objective of using numerical simulation and suitable design methods to develop an economical profile system with integrated functional section for longitudinal mechanical joining, considering increased tightness requirements. Figure 1 shows the model of a battery tray designed according to this idea.

Fig.1
figure 1

Model of a battery tray joined by longitudinal mechanical joining

Full size image

2 Linear Mechanical Joining Process for Extruded Aluminum Profiles

The development of new linear mechanical joints includes both the development of the joining geometries and the process developments for manufacturing the aluminum extrusion profiles, as well as for the joining process itself.

2.1 Development of New Geometries for Linear Mechanical Joining

In the development of joining geometries, firstly the boundary conditions regarding material and the base geometry of the extrusion profile are determined. The specified material is EN AW-6060 aluminum, as this is commonly used in structurally relevant automotive applications.

To avoid a subsequent artificial aging process of entire battery trays, the aim is to develop a joint that can be carried out with profiles that have already been aged to the T6 state (solution annealed + artificial aging). The basic profile geometry is based on a hollow profile plate already used in battery trays. The height of the profile is 12 mm, with the top and bottom walls each possessing a thickness of 3 mm. To be able to produce the profiles on the 10 MN direct bar extrusion press available at the Institute of Forming Technology and Lightweight Components, the multi-chamber profile plate, which typically has a width of up to 400 mm, is reduced to 100 mm width and one single chamber, as shown in Fig. 2. Scaling the profile width is not expected to impact the joint.

Fig. 2
figure 2

Basic geometry of the aluminum extrusion profiles and important dimensions

Full size image

Taking these boundary conditions into account, three joining geometries have been determined in a simulation-based development process, allowing the profiles to be joined within a single press stroke. The simulation-based geometry development was divided into two steps. In step one, the forming of the joints was iteratively improved by analyzing the areas where forming took place and controlling them by adjustments of the joint geometry. In step two, shear lap and 3-point bending tests were simulated for the joint models.

Figure 3 illustrates the joining geometries V1, V2 and V6, each corresponding to different joining concepts.

Fig. 3
figure 3

Profile geometries for different joining concepts

Full size image

The approach for V1 uses a geometry inspired by semi-tubular rivets. The joint is created by plastic expansion of the foot geometry. Profile concept V2 follows the approach of creating an undercut by upsetting the inner joining partner. The asymmetric geometry has the advantage that there are fewer fits, and the joining process is less sensitive to manufacturing deviations. V6 uses the toggle lever principle. Opposing forces prevent the joint from opening. Due to uneven support of the profiles before joining, additional effort for the joining tool may be required for series production.

2.2 Process Development for the Bar Extrusion Process

The process development for the extrusion process has been carried out using extrusion simulations. Commercial simulation programs have been developed for the numerical investigation of extrusion processes by means of Finite Element Method (FEM), which are used both academically and industrially [12]. Due to the continual evolution of software and the increasing application of the FEM in the simulation of extrusion, a general investigation method has been developed, which makes it possible to physically represent the extrusion process with complex boundary conditions [13].

Altair's software Inspire Extrude Metal 2022 was used to simulate the extrusion process for the die geometries. A rigid-viscoplastic material model of the aluminum alloy EN AW-6060 was used. The aluminum was meshed with tetrahedral elements, with a decreasing average element length from 10 mm in the billet to 1 mm in the weld chamber and 0.75 mm in the profile. Rigid tools were assumed. The die temperature was kept constant at 450 °C, as this is the maximum for the available extrusion press. A shear factor of m = 1 was assumed between billet and container. In the die, a coefficient of friction of µ = 0.3 was considered.

Since the three extrusion profiles differ only in the joining geometry, they have similar extrusion ratios, which are calculated according to

$$R=\frac{{A}_{\mathrm{Container}}}{{A}_{\mathrm{Profile}}}$$
(1)

where \(R\approx 26\). Based on the numerical analysis of the temperature distribution, a ram velocity of 3 mm/s and a billet ingot temperature of 510 °C have been determined. The extrusion simulations predicted uniform velocity fields at the profile exit for all three die geometries, which suggests a uniform and straight exit of the extrusion profiles. Accordingly, all three extrusion profiles have been successfully produced with the tools and the determined parameters.

Inspire Extrude Metal 2022 was also used to determine the length of the scrap in the area of the charge weld. The charge weld represents the welding of the previous aluminum billet to the next billet. The weld seam does not form evenly across the profile cross-section, but the material of the new billet replaces the material of the old billet at different positions of a complex profile at different rates [14]. In this area, the mechanical properties can be reduced. To avoid that this influence affects the results of this work, it has been defined that the cross-section of the profile must consist of 95% of new material. For the investigated geometries, a total of 1500 mm of the extruded profiles had to be scrapped after the start of each extrusion process.

2.3 Heat Treatment of the Extrusion Profiles

Heat treatment can be used to adjust the mechanical properties of the produced profiles. Extrusion with subsequent quenching by air, water, spray mist, etc. is comparable to a solution annealing process. The strength and elongation at break of the profiles can be adjusted by subsequent aging [15]. Figure 4 illustrates the possible heat treatment routes for heat treatable alloys such as EN AW- 6060.

Fig. 4
figure 4

Heat treatment of aluminum alloys, acc. [15]

Full size image

Due to the complex thermodynamic and microstructural mechanisms, it is not yet possible to predict profile properties after the extrusion process sufficiently by numerical simulation, as this requires accurate knowledge of the microstructure and the downstream aging process [16]. Therefore, the following artificial ageing strategy for aging the aluminum alloy EN AW-6060 to state T6 has been determined based on an experimental parameter study. To achieve the strength values specified by standards, the profiles were heated to 190 °C within 2 h and then held at this temperature for 7 h.

Two quenching strategies have been considered. One part of the extruded profiles was cooled on the cooling bed in stationary air after the extrusion process. The other part was quenched in a spray cooling system with water and compressed air. According to Milkereit et al., the aluminum alloy EN AW-6060 achieves the highest strengths (after artificial aging) with a quenching rate of 50 K/min [17]. This critical quenching rate has not been achieved by the profiles cooled on the cooling bed in stationary air but has been exceeded by quenching using spray cooling. Tensile tests show a 10% higher tensile strength for the profiles quenched in the spray cooling compared to the slowly cooled specimens. Therefore, the following load-bearing capacity tests were carried out with spray-cooled profiles (Rm ≈ 200 MPa).

3 Joint Formation

To evaluate the joining process, the occurring joining forces and micrographs of the joints were analyzed. The contact surfaces of the joining partners were measured using an optical 3D profiler to identify possible interactions between them.

3.1 Joining Forces

Joining tests have been carried out with 50 mm wide specimens on a ZwickRoell universal testing machine. For this purpose, the specimens were cut from the profiles after extrusion and artificial aging. The profiles have been joined between upsetting tools. Figure 5 shows the force displacement curves of the three investigated profile variants. The joining forces were normalized to 1 mm joining length. The experimental results were compared with results from simulations performed with the software Simufact Forming 2021. For this purpose, 2D planar simulation models with a plane strain state were set up, in which the profiles were joined between plane tools. The profiles were modeled using a quadtree mesh with a mesh size of 0.1 mm in the joining area. The surfaces were refined to a mesh size of 0.5 mm. The tools were modeled as rigid. Flow curves determined during the project were used for the material model. The friction conditions are discussed in Sect. 3.2. The results were also used for subsequent simulations with similar parameters to determine the shear lap and bending capacities in Sect. 4.

Fig. 5
figure 5

Comparison of joining simulation and experiments for the different profile variants

Full size image

The experimental force curve for variant V1 can be divided into two areas. Firstly, there is a linear increase in force, during which the foot geometry forms into the cavity. The total force results from bending the foot geometry, frictional forces, and a slight upsetting of the foot geometry.

Once the foot geometry is fully formed into the cavity, there is a faster increase in force. The higher forces are mainly due to the upsetting of significantly wider cross sections. To complete the joint, a total force per joining length of 4 kN/mm is required for variant V1.

In the simulation, a small plateau can be observed at about 0.3 mm displacement. In this area, bending of the base geometry occurs almost exclusively in the simulation. This behavior cannot be found in the experimental curves, most probably due to manufacturing tolerances and different friction conditions.

Since the joint in variant V2 is created by upsetting the inner joining partner, a typical force curve for upsetting of aluminum (EN AW-6060) can be observed. After a rapid increase in the joining force (elastic deformation), the force then continues to increase linearly with a lower gradient (plastic deformation). The necessary joining force of 1.4 kN/mm is significantly lower compared to V1.

For variant V6, characteristic of the knee lever principle, there is initially a long and flat increase in force. The dominant deformation is a bending of the fine joint geometry. Shortly before completing the joint, a force increase to 0.45 kN/mm can be observed when the two joining partners are tensed against each other. Variant V6 thus requires by far the lowest joining forces among the investigated joints.

Overall, all three variants can be successfully joined, and the necessary joining forces correspond to the prediction from the simulation. The simulation models can thus be used in the future for further optimization of existing joining geometries as well as for the development of new ones. Variant V6, requiring comparatively low forces, is particularly interesting if no large presses are available.

3.2 Micrographic Examination of the Joints

The micrographs shown in Fig. 6 have been created to evaluate the shape of the joints. The variants V2 and V6 show a good shaping of the joining geometry and a high agreement with the simulations shown in Fig. 3.

Fig. 6
figure 6

Micrographs of the completed joints

Full size image

Variant V1 shows a significantly lower degree of filling of the cavity compared to the simulation. This correlates with the observations on the joining forces. The behavior can probably be attributed to friction effects, but also to manufacturing deviations. As a result, the foot geometry starts to upset earlier, impeding the cavity's filling.

In close-up micrographs of V2, a rough surface can be observed on the flanks where the joining partners are in contact. To further investigate this contact area, 3D images of these contact surfaces have been taken using a Keyence VR-5200 3D profiler. For this purpose, the joint has been mechanically separated in a way that no damage to the contact surfaces is to be expected because of the separation process. Afterwards, a 3D image of the contact surface has been taken. Figure 7 shows a top view (a) and a cross-sectional view (b) made from this image.

Fig. 7
figure 7

3D image of the contact surface of the previously separated joint V2

Full size image

The top view of this 20 mm wide specimen shows an oval area with a different structure compared to the surrounding surface. A similar rough surface of the same shape and size can be observed on the other joining partner. This suggests that the surface changes during joining due to forming and contact with the joining partner. The cross-sectional view shows a significantly rougher surface in this area, indicating higher friction in the joint. The rough surfaces observed on the contact surfaces of both joining partners suggest that microform closures are present here, with the surfaces interlocking. This can have a positive effect on load-bearing capacities and leak tightness.

The top view also shows some edge effects, likely caused by material escaping from the joining zone during the joining process. This can result in leaks in the edge areas. To avoid reworking the corners, lateral clamping could be used during joining.

4 Joint Properties

In this section, the load-bearing capacity of the profile joints are compared with respect to the shear strengths and bending strength with a friction stir welded reference specimen. This is followed by investigations about fatigue strength and finally leak tests of the different profile variants are conducted.

4.1 Shear Lap Strength

To determine the shear strength of the joints, the specimens with a length of 20 mm (in extrusion direction) were taken from the 50 mm wide joined profile composites. This is necessary to eliminate the mentioned edge effects. The test was conducted on a quasi-static testing machine with a hydraulic clamping tool and at a speed of 10 mm/min. To prevent plastic deformation of the profiles by the clamping tool, the profiles were additionally fixed with steel plates (see Fig. 8). The measured forces were normalized to 1 mm joining length.

Fig. 8
figure 8

Experimental setup of the shear lap tests of the joints

Full size image

Figure 9 shows the force–displacement curves for different specimens. The results of shear test for single sided friction stir welded (FSW) specimen are provided as a reference in black. The V1 (blue) and V2 (green) profile variants have lower load-bearing capacities at 0.35 kN/mm and 0.29 kN/mm, respectively, compared to the FSW specimen. It can be observed that an increase in the load-bearing capacity can be achieved by adjusting the geometry and optimizing the undercut of the profile composites. The V6 variant (gray) at 0.47 kN/mm demonstrates a higher shear strength than the reference FSW specimen.

Fig. 9
figure 9

Force–displacement curves of the joints subjected to shear lap loading

Full size image

In addition, the micrographs of the fractured joints are presented in Fig. 10. It can be observed that due to the low undercuts and the associated low shear lap strength of the V1 and V2 variants, the specimens are deformed. This causes the profiles to simply slide apart. Due to the asymmetric geometry, V1 has a larger joining area and thus a larger contact surface, which leads to a higher shear tensile strength compared to the V2 variant.

Fig. 10
figure 10

Microscopic images of the fractured joints subjected to shear lap loading

Full size image

In Fig. 11 the results from shear lap tests of Profile V2 are compared with simulations using different friction models. The force–displacement curve for the simulation using coulomb friction µ = 0.2 underestimates the achievable shear forces of the joint by 15%.

Fig. 11
figure 11

a Comparison of numerical and experimental results for shear lap tests and b failure modes at s = 2 mm using different friction models

Full size image

Images of the failure behavior from the experiment and the simulation after a displacement of 2 mm each show that the joint partners slip apart in the simulation, while in the experiment a hooking without relative movement at the contact point is still recognizable. Even further increasing the friction factor does not prevent the joint from slipping.

The Tresca friction model is used to better represent the failure behavior in the simulation. Tresca friction model has been developed for high contact pressures. The shear stress τ is calculated as

$$\tau =m\bullet k$$
(2)

where m is the friction factor and k is the materials shear yield strength. The friction factor m can vary between 0 and 1, where m = 0 describes frictionless sliding, and m = 1 represents sticking friction [18]. For the simulation sticking friction means that as long as a normal force is present, the nodes stick together.

The simulation using the Tresca friction model with sticking friction fits very well with the experimental results, also for higher displacements. The failure mode after s = 2 mm is also identical. This observation supports the previously established thesis that microform closures are present in the joint.

4.2 Bending Strength

The bending strength is determined by means of the 3-point bending test (see Fig. 12). For this purpose, the specimen is positioned centrally on two cylindrical supports with a diameter of 30 mm and a distance of 80 mm. Analogous to the shear lap test, 20 mm sections are cut out of 50 mm profiles. The punch is positioned centrally between the contact surfaces and has a diameter of 20 mm. Figure 13 illustrates the force–displacement profiles of the specimens, with due consideration given to the normalization of measured forces to a 1 mm joining length. The reference specimen, which was joined by means of single-sided friction stir welding (FSW), has a bending strength of 85.2 N/mm joint length. However, considering the results of the variant V6, it can be observed that the bending strength of the specimen is 129.3 N/mm and is higher than the strength of the reference specimen. In contrast, the bending strengths of variants V1 and V2 are lower than strength of the reference specimen and are 52.25 N/mm and 72.15 N/mm, respectively.

Fig. 12
figure 12

Experimental setup of the 3-point bending test of the joints

Full size image
Fig. 13
figure 13

Force–displacement curves of the joints subjected to bending load

Full size image

The microscopic images of the fractured joints subjected to bending load is shown in Fig. 14. It can be seen that the joining area of variants V1 and V2 remains intact after the bending test. This preservation of the joint zone is advantageous, ensuring continued tightness after testing or in the event of a crash.

Fig. 14
figure 14

Microscopic images of the fractured joints subjected to bending load

Full size image

The force is applied in the edge areas of the joining zone with the lowest material thickness. As a result, only the aluminum base material is subjected to bending stress and thus only low bending strengths are achieved. Adjusting the base geometry of the joint would lead to an increase in the load-bearing capacity.

4.3 Fatigue Strength

Fatigue analysis was performed to describe the strength of the joints under cyclic loading. The experiments were conducted on a Rumul Mikrotron 20 kN servo-hydraulic machine. This is a dynamic testing machine operating at full resonance. For each force amplitude, three specimens were tested. The test was performed on profile sections with a length of 20 mm (in extrusion direction). For all tests, the test frequency f = 50 Hz and the load ratio R = 0.1 were kept constant. Number of cycles for low cycle tests was set between N = 1 × 10–4 to 1 × 10–5 and for high cycle tests in the range of N = 1 × 10–6 to 2 × 10–6. From the test results, Woehler curves were generated for both profile variants V1 and V6, and the respective k-factors (slope exponent of the Woehler curve) were determined. Due to the lower shear strength of variant V2, it was not considered in the fatigue strength analysis. Figure 15 shows the Woehler curves for the profile variants V1 and V6.

Fig. 15
figure 15

Woehler diagram of the profile variants V1 and V6 subjected to the cyclic loading

Full size image

Observably, V6 has a higher fatigue strength at the same number of cycles than variant V1. For variant V1, the fatigue strengths at low and high load cycles are 45.0 and 27.0 N/mm, respectively. However, for variant V6, the fatigue strengths at low and high load cycles are 103.5 and 50.4 N/mm, respectively. This represents an increase of 130.0% in the low-cycle fatigue strength and 86.6% increase in the high-cycle fatigue strength. Similar to the results of the bending test, the joining zone of variant V1 remains intact after the failure of the profile bond. In addition, the failure occurs at the edge of the joining zone in the area of the lowest material thickness, which leads to a reduction in the fatigue strength.

4.4 Leak Tightness Properties

Battery housings are subject to increased sealing requirements. Firstly, the penetration of harmful media from the environment into the high-voltage housing must be prevented. In fact, the ingress of water or humidity can lead to electrical short circuits or, in reaction with the battery electrolyte, to the formation of hydrofluoric acid. In addition, leakage of electrolyte into the environment should be prevented. There are various standards for testing the tightness of the battery system, in which different methods and procedures for leak testing are described. These can be divided into vacuum and overpressure methods. The test methods are also classified according to their sensitivity.

The XL3000flex from INFICON GmbH was used for helium sniffer leak detection. This is a high-precision sniffer leak detector. The measuring principle is based on a resistive sector field mass spectrometer, which can detect helium and hydrogen. In this way, even the smallest leaks can be reliably detected and localized. The test procedure is shown schematically in Fig. 16.

Fig. 16
figure 16

Test procedure for Helium sniffer leak detection

Full size image

The limit value for the leakage rate was set at 1 × 10 3 mbar·l/s, which corresponds to the limit of vapor tightness [10]. It was observed that the profile variants V1 and V6 profile composites show high leak tightness. All leakage rates were below 5 × 10–5 mbar·l/s. It should be mentioned that at leakage rates below 5 × 10–5 mbar·l/s, there is a large influence of ambient exposure of helium. The variant V2 exhibited leakage rates of over 2.04 × 10–2 mbar·l/s. By increasing the joining force from 1.6 kN/mm to 2.6 kN/mm and optimizing the punch geometry so that the joining force is applied directly to the joining area instead of being distributed over the entire profile, leakage rates of less than 1 × 10–3 mbar·l/s were achieved (Fig. 17).

Fig. 17
figure 17

Leakage rates as a function of joining force for variants V1, V2, and V6

Full size image

Furthermore, the influence of various profile properties on the sealing properties of the composite was also investigated.

Figure 18 presents the leakage rate for profile V6 as a function of different cooling strategies (in air or in water), surface conditions (with or without additional lubricants), joining speeds (low punch speed: 0.16 mm/s or high punch speed: 3.3 mm/s) and the mechanical (quasi-static or cyclic) as well as thermal preload. For the mechanical preload, forces were selected from the elastic range of the quasi-static shear lap test and from the fatigue range of the fatigue test. In addition, the influence of downstream over-welding on the tightness of the profile composite was investigated. For this purpose, weld lines were provided longitudinally and transversely to the linear joint. Furthermore, the tightness was measured on specimens in real component size. In this regard, profiles with a length of 600 mm were joined and then 20 mm specimens were taken from the center. The limit value for the leakage rate is marked with a red line. It can be observed that for all specimens, the leakage rate is below the limit value of 1 × 10–3 mbar·l/s. In addition, no influence on the leakage properties could be seen. The same tendency can be observed for variant V1. It can be concluded that both variants V1 and V6 ensure complete impermeability to water and water vapor.

Fig. 18
figure 18

Leakage rates of different profile properties for variant V6

Full size image

5 Strength Grading by Local Cooling

In pursuit of enhancing joint shaping and achieving superior undercuts during joining, local cooling strategies were explored to create local property profiles with variations across the cross-section of the extruded profiles. Tekkaya et al. presented a method involving the local quenching of extruded profiles, demonstrating the capability to establish strength differences of up to 100 MPa along the profile length for the EN AW-6082 aluminum alloy [19]. The approach is to transfer this process to a strength grading over the cross-section. The principle is shown in Fig. 19. The section of the profile with the outer joining geometry is quenched locally by water mist. According to Totten and MacKenzie, sufficiently rapid cooling leads to a fine grain and a supersaturated solid solution [20].

Fig. 19
figure 19

Strength grading by local Quenching, according to Ref. [20]

Full size image

After artificial aging, the fine grains and finely dispersed precipitations cause a higher strength. The part with the inner joining geometry is not actively cooled. Slow cooling generates a coarse grain and coarse precipitations that persist after artificial aging, leading to reduced strength.

Experimental investigations have shown that the proposed method for strength grading is not feasible for the profile dimensions and aluminum alloy investigated. The high thermal conductivity of aluminum leads to rapid cooling, even for the uncooled side, given the small profile dimensions and the investigated EN AW-6060 is much less quenching sensitive then EN AW-6082 [17]. Although not suitable for the studied alloy and dimensions, strength grading through local cooling might be conceivable for larger profiles and quench-sensitive alloys. Consequently, the influence of strength grading on the linear joining of aluminum extrusion profiles has been investigated by means of simulations.

The V2 profile variant has been optimized specifically for this approach. Since the male and female geometry of the joint are each located at only one end of the profile, a difference in the cooling rate can be achieved for sufficiently wide profiles. In case of the V1 and V6 variants, the geometries that should be kept soft to localize the forming are too close to those that should be as hard as possible.

In the simulation a flow curve was assigned to the profile with the outer joining geometry, which corresponds to the material properties of the present alloy EN AW-6060 after quenching by spray cooling and artificial aging to T6. For the profile with the inner joining geometry, the flow curve was gradually reduced to 50% by a scaling factor. In Fig. 20 (a), different geometrical properties are defined, which are used to describe the influence of strength grading on the joints shape. Hmax is the maximum undercut of the joint and is predicted to have an influence on the joint’s strength while the contact length lc and the undercut with contact Hc are predicted to have an influence on the leak tightness of the joint, because there is a permanent pressure between the surfaces. In order to determine the influence on the shear strength, shear lap simulations have also been carried out and the shear lap load capacity Fshear,max, that is the maximum shear lap strength, was determined.

Fig. 20
figure 20

a Definition of joint properties and b the effects of increasing strength grading

Full size image

Figure 20 (b) shows the effect of strength grading on the defined properties. With a 50% strength grading, the undercut with contact can be increased by 100%, and the contact length can increase by up to 300%. This is attributed to the increased deformation of the inner joining partner with increasing strength grading causing less deformation in the outer joining partner due to its higher strength. As a result, the outer joint geometry expands less and the contact in the upper area of the joint can be increased.

Since the maximum possible undercut Hmax has already been reached without strength grading, and the outer, stronger joining partner is not further deformed by reducing the flow stress of the inner joining partner, Hmax is not affected by the strength grading. Further simulations altering the outer geometry of the joint to allow larger undercuts showed no significant influence on Hmax. This can be explained by the upsetting behavior of the inner part, which is comparable to an upsetting test. The contact in the upper area of the joint prevents further formation of the undercut.

In contrast, the shear tensile load capacity is negatively affected by the strength grading. A linear correlation is evident here. By reducing the strength of the inner joining partner to 50%, the maximum shear tensile strength of the joint is reduced by 31%. This is because the failure of the joint in the shear tensile test is mainly caused by the bending up of the profile with the inner joint geometry. Reducing strength in this area directly impacts joint strength. As this area is very close to the joining area, it will not be possible to implement strength grading by local cooling without also reducing the strength here.

In summary, it can be said that the negative influence on the joint strength prevails. The better shape after joining indicates an increased leak tightness. However, since the required leak tightness can already be fulfilled without strength grading, strength grading by local cooling is not considered to be effective for this application.

6 Conclusions

In this study, three aluminum extrusion profile systems for leak tight linear mechanical joining were developed through numerical simulations, successfully produced, and effectively joined in a single press stroke. The main findings of this study include:

  • EN AW-6060 alloy extruded profiles demonstrated successful joining after artificial aging to the T6 condition.

  • Joining forces ranging from 0.45 kN/mm to 4 kN/mm (depending on the variant) were necessary for the joining process.

  • Shear tensile tests revealed that the load-bearing capacity of the joints, compared to a friction stir welded joint, falls within the range of 75% and 105% depending on the variant.

  • The 3-point bending load-bearing capacity of the joints ranges from 60 to 150% compared to a friction stir welded joint.

  • Fatigue strength for V1 is achieved at a force amplitude per joint length of 27.0 N/mm, and for V6, it's 50.4 N/mm, demonstrating good comparability in inclination exponent.

  • Profile systems V1 and V6 exhibit high leak tightness, with leak rates below 5 × 10–5 mbar·l/s.

These findings suggest that the production of leak-tight linear joints of aluminum extrusion profiles is not only feasible but also suitable for integration into battery trays. Additionally, the attempt to enhance joint properties through strength grading was found to be ineffective, as lower material strength in the failure area outweighs the positive effects of improved joint shaping.