In Situ Investigation of the Frictional Behaviour in Friction-Spinning

Abstract

Friction-spinning is an incremental thermomechanical forming process that has huge potential due to its simple yet effective mechanism of utilising friction between a rotating workpiece and a forming tool to increase the workpiece’s temperature, which reduces the required forces and increases formability during the forming process. Despite the simplicity of the process’s setup, the thermomechanical loads and high relative velocities involved, especially in the contact zone, make the application of classical methods for characterising friction inaccurate. It is therefore essential to find a way to describe the frictional behaviour under real process conditions to be able to gain a holistic understanding of the process and the effect of the adjustable parameters on the outcome, especially the temperature. To achieve this goal, an experimental setup that considers the actual process boundary conditions in forming tubes made of EN AW-6060 was used to measure in situ normal and frictional forces, in addition to process temperatures, under varying rotational speed and feed rate values.

1. Introduction

Friction-spinning is an incremental thermomechanical metal-forming process that, similarly to the process of friction welding, utilises the temperature generated between a rotating workpiece and a forming tool to elevate the process temperature and thereby reduce the required forming forces. There are two adjustable parameters in this process, the rotating speed, ω, of the workpiece and the feed rate, f, of the tool, in addition to the contact area and the forming path. These parameters directly affect the normal forces, the relative velocities and process temperatures. In friction-spinning, the relative velocities involved in manufacturing tubular workpieces made of aluminium alloys typically range between v = 0.8 and 7.5 m/s, depending on the rotational speed and the radial position at which the forming tool is located with respect to the centre of rotation of the workpiece. The forming temperatures for aluminium alloys can reach up to 550 °C. Since friction is the primary mechanism used to generate heat, it is important to understand and characterise this phenomenon while considering all relevant process boundary conditions.

To date, there has been no dedicated research analysing the frictional behaviour in friction-spinning, which highlights the need to examine similar thermomechanical manufacturing processes with comparable boundary conditions. Becker and Hora et al. introduced three friction models for extrusion of EN AW 6082 using a Tribo-Torsion-Test setup, which enables the measurement of normal and frictional stresses under elevated temperatures using an inductive heat source. These models included a Bingham/Coulomb temperature-dependent approach, a shear friction approach in which the friction factor is a function of pressure and relative velocity and a model that directly describes the friction stress as a function of temperature, pressure and relative velocity. Becker used a Hockett–Sherby-based method and a modified Zener–Hollomon equation for determining the flow stress values [1,2,3].

Buchner investigated the frictional behaviour in closed-die warm forging of AA 2618 using ring-on-disc and pin-on-disc tests, performing experiments at temperatures up to 400 °C. To model friction, Buchner employed a physical approach that incorporates the evolution of the real surface area. He determined the tool–workpiece interface conditions by subjecting a sandblasted aluminium sheet to compression with a flat steel punch and inspected the surface before and after compression. He analytically calculated the surface growth based on this investigation [4,5]. However, Buchner’s analysis did not consider the influence of relative velocity.

Behrens et al. developed a friction model for hot bulk-forming processes, considering several process variables, including temperature, lubrication and forming velocity among others [6]. Shi et al. designed a new rig for investigating the frictional behaviour of sheet metal forming under warm conditions and used the Coulomb model to characterise friction at temperatures up to 350 °C employing a furnace [7].

In many mentioned publications in which the friction phenomenon was examined under elevated temperatures, an external heat source was used to achieve the required temperature [3,4,7,8]. In friction-spinning, however, heat generation occurs locally at the contact surface and is inhomogeneously distributed across the material, which experiences high relative velocities along with high loads and plastic deformation. Therefore, typical friction tests such as the pin-on-disc test, strip drawing test and spiral sliding test—as used by Matsumoto—are not expected to yield accurate results for friction-spinning due to different boundary conditions, namely in heat distribution, relative velocity and plastic deformation [7,9].

The Coulomb friction law is often used due to its simplicity and, in many cases, serves to estimate friction forces based on normal forces. However, the coefficient of friction depends on both temperature and sliding velocity [10]. It has also been observed that in extrusion processes, the Coulomb friction law tends to overestimate the frictional stress, sometimes exceeding the flow stress of the deformed material [1,11]. Despite this, most publications on modelling processes involving linear and orbital friction—as well as friction stir welding, which shares key features with friction-spinning, particularly high relative velocities and intentionally induced high temperatures—use the Coulomb-based coefficient of friction, µ, for modelling [12,13,14,15,16,17,18,19,20].

In an attempt to determine the friction coefficient using the Coulomb friction law considering temperature and relative velocity, Lossen used the strip drawing test on workpieces preheated in an industrial oven under temperatures of up to 500 °C, at relative speeds of 10 and 50 mm/s [21]. However, these relative velocities and the temperature distribution in the workpieces do not correspond to the conditions found in the friction-spinning process and plastic deformation was not considered. This work is considered a continuation of Lossen’s [21] analysis of friction in friction-spinning. The novelty of this investigation lies in the fact that the friction behaviour is examined at real process speeds and in situ under actual process conditions. In friction-spinning of aluminium, material adhesion to the tool is so far inevitable, leading to reduced material efficiency and deviations in workpiece geometry. Consequently, the study also investigates a novel tool coating aimed at minimising aluminium adhesion and enhancing thermal protection of the tool, whereby the influence on the friction behaviour is observed. The results will be used to improve understanding of the process input and output parameters, enabling the development of accurate numerical FEM models for the process.

2. Materials and Methods

The experiments were conducted with tubes made of EN AW-6060, with an outer diameter of 40 mm and a wall thickness of 5 mm, for rotational speeds of up to ω = 900 rpm (relative velocity v = 1.6 m/s). For higher relative velocities of up to v = 9.3 m/s, tubes of the same material with a diameter of 70 mm and a wall thickness of 5 mm were used (rotational speed of up to ω = 2730 rpm). The tool material was made of high-speed steel 1.3343.

The experiments were conducted on the same two machines that were also used to perform the process of friction-spinning, enabling the application of identical process conditions with minimal effort. The machine used in experiments with relative velocities of up to 1.6 m/s is a modified PLB 400 spinning machine from Leifeld (Ahlen, Germany). It has a headstock with a 11 kW motor and a cross slide that moves in X and Y directions, driven by a hydraulic system. The control and monitoring system consists of an NI realtime system cRIO-9082 from National Instruments Germany GmbH (München, Germany). Continuous path control was achieved using the NI software LabVIEW Version 2020.

For positioning accuracy, Temposonics R-Series sensors from MTS Sensor Technologies GmbH & Co. KG (Lundenscheid, Germany) were used. Normal force measurements were conducted using analogue hydraulic pressure transmitters S10 from WIKA (Klingenberg, Germany). The combination of the tool and tool fixture normally used in friction-spinning was replaced by a setup with a tool plate made from high-speed steel 1.3343, identical to the tool material. For the measurement of the frictional forces, F_Frictional, at the contact zone (see Figure 1), a methodology similar to that used by Buchner [4] was implemented:

$$F_{\mathrm{F}\mathrm{r}\mathrm{i}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{a}\mathrm{l}}={\displaystyle \frac{M}{R_2}}={\displaystyle \frac{F_{\mathrm{s} }·R_1}{R_2}}$$

(1)

Here, F_Frictional is the frictional force at R₂ = 17.5 mm, and F_s is the force measured by a KD80s-20 kN sensor from ME-Meßsysteme GmbH (Henningsdorf, Germany), with a nominal load of 20 kN (see Figure 1). Due to machine-related limitations, experiments with higher relative velocities of up to v = 9.3 m/s and rotational speeds of up to ω = 2730 rpm were conducted on an additional CNC-controlled friction-spinning test rig, developed in-house at the institute, while retaining the same tool arrangement. The test rig features a spindle driven by two 50 kW motors and is equipped with two cross slides, each capable of applying an axial force of up to 50 kN per axis. All components are controlled via a Siemens Sinumerik S7-d840 system (Nürnberg, Germany). To measure forces and moments, a six-axis force/torque sensor from ME-Meßsysteme GmbH (Henningsdorf, Germany) was installed between the upsetting plate and the cross slide. This sensor offers a measurement range of 50 kN and 5 kNm and is classified with an accuracy class of 0.5.

The experiments were performed by moving the tool plate, mounted on a cross-slide, vertically at a defined feed rate, f, and a specified rotational speed, ω, towards a rotating EN AW-6060 tube clamped in a lathe chuck. This led to the upsetting of the tube, similar to a compression test, except that the reduction in length occurred locally near the contact area due to friction-induced heat generation, rather than globally over the entire workpiece as in a conventional upsetting test. In this work, this experiment will be referred to as the friction-upsetting test.

The local compression in the workpiece results from localised heat generation, which reduces the flow stress at the contact region. The infeed of all experiments was set to 15 mm. Preliminary tests showed that a friction-upsetting test using a flat tool plate produces a growing ring-shaped contact area, which causes large differences in the relative velocities between the inner and outer parts of the contact area, depending on the radial position. This makes the analysis of relative velocity in friction-spinning inaccurate.

To limit the difference in the relative velocities between the outer and the inner edges of the ring-shaped contact area, the tool plate was designed such that the contact area becomes constant shortly after the tool plate and the rotating workpiece meet. This was achieved by constructing a ring-shaped surface elevated from the rest of the plate, as shown in Figure 1. This ring-shaped contact area was also implemented for friction analysis by Horwatitsch [22]. The width of this ring is 9 mm, which exceeds the workpiece’s wall thickness of 5 mm, to compensate for possible imperfections of the tube geometry or positioning inaccuracies, as well as for the high vibration occurring especially in the initial stages of contact. This means that only limited growth in the contact area occurs at the start of the experiments. Tests showed that the contact surface becomes constant after approximately 1.5 mm of feed. The surface roughness of the contact area had an average Ra = 0.3 µm and a standard deviation of 0.2.

For temperature measurements, three slots were machined into the tool plate for positioning type-K thermocouples, placed at a distance of 1 mm from the ring-shaped contact surface at R₂ = 17.5 mm or R₂ = 32.5 mm. The investigated process parameters are listed in

Table 1. Parameters.

Tube Diameter (mm)	Rotational Speed ω (rpm) Workpiece	Av. Relative Velocity v (m/s) Workpiece	Feed Rate f (mm/s) Tool Plate
40	600	1.1	0.25/2.00
40	700	1.3	0.5
40	800	1.5	1.00
40	900	1.6	0.25/2.00
70	345	1.2	0.25/2.00
70	515	1.8	0.25/2.00
70	820	2.8	0.25/2.00
70	1190	4.0	0.25/2.00
70	1365	4.6	0.25/2.00
70	1640	5.6	0.25/2.00
70	1910	6.5	0.25/2.00
70	2185	7.4	0.25/2.00
70	2455	8.4	0.25/2.00
70	2730	9.3	0.25/2.00

.

Here, the average relative velocity, v, was calculated using the ω at the middle of the contact area, where the radius is r = R₂ = 17.5 mm or r = R₂ = 32.5 mm. For statistical validity, three experiments were conducted for each required value in the tests with the tube diameter of D40 mm. The experiments with a tube diameter of D70 mm were carried out to map and identify trends and were therefore performed only once.

v = 2π rω

(2)

3. Results

The experimental measurements show a typical progression of the normal and frictional forces, with a sharp rise occurring at the initial stages of contact, followed by a decrease in force values accompanied by an increase in temperature, until the force profiles reach an approximate steady state, as shown for two different parameter sets in Figure 2.

At the early stages of the process, adhesive wear on the tool plate was observed, such that contact occurred between the workpiece material and a layer adhered to the tool, comparable to the built-up edge formed in turning [23].

The profile of the frictional forces agrees with the findings of Vairis [12] on the process of linear friction welding, the demonstration of friction force by Pietrzak et al. [19] for friction spin welding and the description of Raab [15] for friction welding. Raab describes the beginning of the friction welding process as starting with low friction forces that continually increase due to the growth in the real contact area, until adhesive wear occurs. This wear leads to a further increase in the real contact area, coupled with a rise in friction power, which results in high temperatures at the contact surface. This explanation was also shared by Vairis [12,15]. In addition to the increase in real area caused by the flattening of asperities, the contact area growth at the beginning of the experiments—before the contact surface is equal to the ring-shaped geometry on the tool plate (see Figure 1)—is possibly also responsible for the initial force increase.

The normal force profile in the steady state shows a gradual increase, which is more pronounced at high feed rates (see Figure 2). This behaviour is assumed to result from the dominant effect of work hardening caused by the forming process, which surpasses the thermal softening induced by the elevated temperatures at the contact zone. At higher feed rates, and thus at a shorter process duration, there is less time for the heat generated at the contact surface to be transferred to the inner parts of the material. This leads to higher flow stress and to an increase in the required forming force. A further explanation for the increase in normal force in the steady state is the slight increase in wall thickness of the workpiece near the deformation zone, which requires a corresponding rise in normal force for further deformation.

The average values of the normal and the frictional force measurements were calculated for the investigated process parameters, considering the average normal force in the steady state. A positive correlation between the forces (normal and frictional) and the feed rate, and a negative correlation with the relative velocity, was observed. The maximum average forces in the steady state occurred at the lowest relative velocity and highest feed rate (v = 1.1 m/s and f = 2 mm/s) with F_Normal = 11.1 kN and F_Frictional = 4.8 kN, whereas the lowest values appeared at v = 1.6/1.5 m/s and f = 0.25 mm/s with approximately F_Normal = 3.7 kN and F_Frictional = 1 kN.

To analyse the frictional behaviour in friction-spinning and its relationship with the process parameters, the normal and frictional stresses were calculated by dividing the average forces for the tested parameters in the steady-state region by the contact area, as shown in Figure 1.

$${\sigma }_{\mathrm{N}}={\displaystyle \frac{F_{\mathrm{Normal} }}{A_{\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{a}\mathrm{c}\mathrm{t}}}}$$

(3)

$${\tau }_{\mathrm{F}}={\displaystyle \frac{F_{\mathrm{Friction} }}{A_{\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{a}\mathrm{c}\mathrm{t}}}}$$

(4)

It was assumed that, at this point, the real contact area A_Real was 100%. The results are shown in Figure 3.

Figure 3 shows a clear influence of the relative velocity and feed rate on the frictional stress. A positive correlation with the feed rate and a negative correlation with the relative velocity can be observed. The behaviour of the normal stresses is similar to that of the frictional stresses, except that the normal stress values exhibit a greater increase with rising feed rates. This is a result of the increase in normal forces in the steady state, as explained earlier.

For the analysis of the temperature measurements generated at the contact area T_Contact, the maximum measured temperatures at 1 mm distance from the contact surface T_1mm and the measured frictional forces were used, applying Equations (3) and (4) according to Fourier’s law.

$$T_{\mathrm{Contact}}=T_{1\mathrm{mm}}+{\displaystyle \frac{{l\times \dot{q}}_{\mathrm{Friction}}}{\lambda \times A_{\mathrm{contact}}}}$$

(5)

$${\dot{q}}_{\mathit{Friction}} =0.5 F_{\mathrm{Friction}} \times v$$

(6)

Here λ = 27.4 W/mK is the thermal conductivity of the tool material 1.3343, ${\dot{q}}_{\mathit{Friction}}$ is the heat flux generated through friction, l = 1 mm is the distance between the measured temperature and the contact surface, and 0.5 is the assumed fraction of the total heat flux that flows into the tool plate. For simplicity, the thermal energy resulting from forming was not considered.

The results in Figure 4 illustrate the temperature measurements at 1 mm distance from the contact surface (solid lines) and the calculated temperatures at the contact surface (dashed lines) under varying feed rate and relative velocities. An obvious relationship between the calculated contact surface temperatures and the process parameters is not readily apparent from Figure 4. In general, the relative velocity leads to increased heat generation (see Equation (6)). However, Figure 3 shows that high relative velocities correlate with corresponding lower frictional stresses and frictional forces. Therefore, the relative velocity and the corresponding frictional forces have opposing effects on the generated heat flux and corresponding temperatures, i.e., an increase in the relative velocity leads to a reduction in the frictional forces and vice versa (see Equation (6)).

Furthermore, the measured temperature at 1 mm from the contact surface shows a negative correlation with the feed rate. This can be explained by the shorter process duration associated with higher feed rates, which means that the heat generated at the contact region has less time to be transferred to the inner parts of the material via conduction. Consequently, it is expected that even higher temperatures would occur at high feed rates if the process duration were longer.

In a further series of experiments aimed at gaining a better understanding of the effect of the feed rate on process temperature, the feed rate was increased during the process while the relative velocity was kept constant. Figure 5 shows a representative example in which the feed rate was initially set to 0.25 mm/s and then increased to 1 mm/s at a constant relative velocity of 1.3 m/s. The measurements reveal that the increase in feed rate led to a corresponding increase in the frictional force and, after a slight delay, in temperature. This result confirms that an increase in feed rate leads to a corresponding increase in temperature at a constant relative velocity.

In this case, the short duration associated with the high feed rate was overcome because, in the first step at f = 0.25 mm/s, there was sufficient time to reach the steady-state phase and allow enough heat to be transferred to the measuring position. The subsequent increase in feed rate could then be directly observed in the temperature measurements.

To analyse the friction behaviour at relative velocities typical for friction spinning—up to 9 m/s—experiments were carried out using an additional test rig capable of spindle speeds of up to 3000 rpm. To reach relative velocities of 9 m/s, not only the rotational speed but also the diameter of the tube specimen had to be increased from 40 mm to 70 mm.

To evaluate the frictional contact as a function of relative velocity, the ratio of shear stress to normal stress was investigated for different relative velocities and upsetting feed rates. Since the same contact area was used for both stresses, the directly measured normal force F_N and the frictional force F_Frictional, derived from the torque, can be used to determine the friction coefficient/value according to Coulomb’s model.

Figure 6 shows an exemplary force–time diagram for a relative velocity of 1.65 m/s and a feed rate of 2 mm/s. Additionally, the resulting friction value µ over the upsetting displacement is shown. The force peaks at the beginning of the friction-upsetting path lead to a short-term maximum in the friction value, which then decreases and approaches a steady-state value.

The initial peak in the friction coefficient has two primary causes: first, the temperature-induced drop in the flow stress of the workpiece had not yet occurred at the beginning of the process (see above); second, adhesion of the aluminium to the upsetting plate typically occurs at the onset of deformation. This leads to a momentary sticking contact and, accordingly, to a brief increase in the friction force. As the friction-upsetting process progresses, the forces stabilise and a steady-state process condition is reached, with the friction coefficient µ approaching a constant value. In the example shown in Figure 6, the friction coefficient µ converges to approximately 0.5.

In the following, average values of the friction value µ from the steady-state region are reported. The maximum “static/adhesion” friction values µ_max are generally about twice as high as those observed in the steady-state range. As shown in Figure 7, both the steady-state and maximum friction values decrease with increasing relative velocity and feed rate.

The rise in relative velocity and feed rate generates more heat in the contact zone between the tool and the workpiece. This increased local heat input leads to a reduction in the flow stress of the material in the contact zone. Consequently, the shear stress decreases more rapidly with increasing relative velocity than the normal stress. Additionally, the increase in normal stress at higher feed rates is disproportionately larger than the increase in shear stress, resulting in lower overall friction values.

Figure 8 shows that the observed trend persists across the entire investigated range of relative velocities. To confirm the independence of the friction value from the contact area, two relative velocities values are provided in the overlap region of the two test series with different tube diameters. These indicate that the differences in the friction values lie within the range of measurement variability, and no clear correlation with the contact area can be identified.

The comparison between higher feed rates (2 mm/s) and lower feed rates (0.25 mm/s), as illustrated in Figure 9, confirms that the general trend of decreasing friction values with increasing relative velocity also holds under these conditions. Furthermore, the increase in friction resulting from reduced feed rates is clearly observable across the entire range of relative velocities investigated. Variations at individual measurement points can be attributed, in part, to the fact that single measurements were used in this evaluation to identify general trends.

The trend of decreasing friction force with increasing relative velocity becomes even clearer when only the frictional forces are considered (see Figure 10). The unsteady course of the normal forces shows that decreasing friction force with increasing relative speed has no direct correlation to the normal force.

In addition to the previously discussed mechanisms contributing to reduced friction coefficients at increasing relative velocities, the presence of adiabatic shear bands is also assumed. Localised temperature rise in the contact zone, combined with severe plastic deformation and high shear stresses, leads to significant grain refinement and dynamic recrystallization within the microstructure. Due to the highly localised thermal input, plastic strain is strongly confined to narrow regions, giving rise to the observed phenomenon.

Microstructural analyses have confirmed the presence of shear bands in the contact zone at higher relative velocities. For example, Figure 11 presents the microstructure of two tubes from friction-upsetting experiments at a high relative velocity of 2.8 m/s and 7.4 m/s, compared with the initial microstructure of the specimen in the tangential direction. Grain refinement is clearly visible in the contact area and even increases with increasing relative velocity. In regions distant from the contact zone, even grain growth is observable, attributed to the elevated temperature and low levels of deformation. Furthermore, this grain growth increases with rising relative velocity due to enhanced heat input resulting from a longer friction path over the same time interval.

In order to reduce adhesion and tool wear, various coating combinations were applied to the forming tool and subsequently investigated. Previous studies have shown that a combination of thermal barrier coatings (TBC) and physical vapour deposition (PVD) coatings can reduce tool wear and improve workpiece accuracy during the friction-spinning process [21].

Also, in this study the aim of using coated tools was to further reduce the thermal load on the tools and additionally to prevent aluminium adhesion to the coated tools. For the tool coating investigated in this study, upsetting plates and round-bar forming tools were equipped with the multilayer coating system shown in Figure 12.

A 100 µm thick bond coat of NiCrAlY (Amperit 413.001) was first applied to the tool steel substrate (1.3343). Subsequently, a 150 µm thick thermal barrier coating (7YSZ, Amperit 827.007) was deposited on top. This layer was then overlaid with a 150 µm thick cold gas-sprayed titanium layer (Metco 4010C) serving as metallization. As the final top layer, a 2 µm TiB₂ coating was deposited via a physical vapour deposition (PVD) process using a hybrid configuration (dual-target): one TiB₂ dcMS cathode at 3 kW and one TiB₂ HiPIMS cathode at 3 kW (HiPIMS frequency: 1000 Hz, pulse duration: 200 µs).

Investigations using upsetting plates with this coating reveal that the temperature of the uncoated plate rises significantly faster at the beginning of the friction upsetting than the coated plate. In the steady-state region of the upsetting path, the temperature rises equally in coated and uncoated plates. The temperature of the coated tools is, independent of the relative speed, approximately 150 °C lower than that of the uncoated ones, as shown in Figure 13. This behaviour effectively protects the tool from excessive thermal loads.

The friction value of the coated upsetting plates is significantly lower at initial contact compared to that of the uncoated plate. However, after a short upsetting stroke, the friction values converge. This is attributed to aluminium adhesion on the upsetting plate, resulting in the formation of aluminium–aluminium contact between the upsetting specimen and the plate. From this point onward in the friction-upsetting process, both the friction and normal forces exhibit a highly similar progression between the comparative specimens, as illustrated in Figure 14.

Measurements of the adhered aluminium mass, shown in Figure 15, revealed that the use of coated upsetting plates as well as coated round-bar forming tools for flange forming resulted in a reduction in adhesion by at least 85% and up to 97%.

4. Discussion

The investigations presented in this study on the frictional behaviour during friction-spinning demonstrate that the coefficient of friction µ (friction value) generally decreases with increasing feed rate. However, when considering only the frictional forces (Figure 10), it becomes apparent that these increase with higher feed rates at constant relative velocity. The results show that the normal forces increase disproportionately with increasing feed rate, suggesting that application of the Coulomb model to friction-spinning yields an inaccurate representation of the process. To describe friction using the shear friction model—which would be more appropriate for friction-spinning—the temperature-dependent shear yield stresses of the materials are required. The observation that the normal force does not decrease continuously at high and increasing relative velocities, despite a continuous reduction in friction force, indicates a more pronounced decrease in shear strength than in flow stress with rising temperature, induced by the increasing relative velocity. With increasing relative velocity, on the other hand, the friction value very clearly continuously declines.

A clear correlation between the coefficients of friction and the normal forces determined in the friction-upsetting tests cannot be established. Although there is a tendency for the normal forces (or normal stresses) to decrease with increasing relative velocity due to a longer frictional path and the resulting higher thermal input, this trend is not consistently observed across all conditions at higher relative velocities (>4 m/s).

The temperature investigation under varying process parameters shows a rather complex relationship with the feed rate and relative velocity and can be summarised in two points:

• The heat flux resulting from friction is a product of the relative velocity and frictional force, which are interrelated parameters and therefore need to be considered collectively.
• A high feed rate results in elevated temperatures at the contact surface due to increased frictional forces. However, the amount of heat transmitted to the inner regions of the workpiece is reduced, as the shorter process duration associated with higher feed rates limits heat conduction. This is important to consider because the amount of heat that reaches the inner parts of the material determines the corresponding reduction in flow stress and thus the required forming forces.

The multilayer coating investigated in the friction-upsetting test and flange forming process demonstrates that the initial peak in the friction value at the beginning of the friction-upsetting process can be significantly reduced. However, due to aluminium adhesion—even on the PVD coating—the friction values after a short upsetting time are comparable to those of uncoated tools. Owing to the substantially reduced aluminium adhesion, a marked increase in material efficiency and an anticipated improvement in geometrical component repeatability can be expected. Both these aspects, together with extended tool life resulting from the lower thermal load though the thermal barrier effect of the coating, require validation through further experimental investigations.

The results of these investigations aim to enable a more accurate representation of the friction-spinning process within a thermomechanical FEM model. By incorporating detailed knowledge of process-dependent friction values, the thermo-mechanical-influenced forming behaviour should be able to be modelled with greater precision.