Design, Characterisation and Numerical Investigations of Additively Manufactured H10 Hybrid-Forging Dies with Conformal Cooling Channels

Abstract

Internal die cooling during forging can reduce thermal loads, counteracting surface softening, plastic deformation and abrasive die wear. Additive manufacturing has great potential for producing complex geometries of the internal cooling channels. In this study, hybrid forging dies were developed combining conventional manufacturing processes and laser powder bed fusion (L-PBF) achieving conformal cooling channels. A characterisation of the used hot-work tool steel’s AISI H10 powder material was carried out in order to determine suitable parameters for L-PBF processing and heat treatment parameters. Additionally, the mechanical properties of L-PBF-processed AISI H10 specimens were investigated. Furthermore, the influence of different internal cooling channels regarding a possible structural weakening of the die were analysed by means of a finite element method (FEM) applied to a hot-forging process. The numerical results indicated that the developed forging dies withstood the mechanical loads during a forging process. However, during the investigation a large dependency between the resulting stresses and the chosen parameters were observed. By choosing the best combination of parameters, a reduction of the equivalent stress by 1000 MPa can be achieved. Finally, a prototype of the hybrid-forging dies featuring the most promising cooling channel geometry was manufactured.

1. Introduction

Hot-forging dies are subjected to high mechanical, thermal and tribological loads resulting from the high forces and the contact of the die with the heated semi-finished products. The occurring loads soften the near surface areas of the die, promoting abrasive wear and plastic deformation. Currently, cooling lubricants are applied by spray units to reduce the thermal load on the die. However, spray cooling is limited to be used in between forming operations only [1]. Here, additional internal die cooling, by means of conformal internal cooling channels, can reduce the thermal load throughout the forming process. Additionally, these channels can heat up the dies during machine downtime, reducing ramp-up time and the associated costs. By combining internal and external cooling, it is possible to increase the heat flow, which can reduce die wear and increase die lifetime [2]. However, the application of such channels leads to new challenges in terms of manufacturing techniques. The laser powder bed fusion (L-PBF) process allows the production of complex internal channel geometries and was used in this study [3]. Another challenge is the stability of the forging dies, which is reduced by internal cooling channels. The finite element method (FEM) approach is suitable for the design of such channels.

The aim of this work was to present the development of a hybrid-forging die with internal cooling channels and the difficulties in the design, with the aim of achieving a possible design for later hot-forging tests. Preliminary work has successfully shown that dies fabricated by L-PBF (without inner cavities) can withstand the thermomechanical loads of hot forging [4]. In this study, AISI H10 hot-work tool steel was used to build up complex geometries for hybrid-forging dies with internal channels. The term hybrid-forging dies is used in this context to refer to a combination of conventionally manufactured and additively manufactured components of the die. Hybrid die concepts provide a great potential for reducing costs, as wear critical parts can be designed differently. Additive manufacturing processes especially result in higher costs than conventional machining due to the longer processing time, higher material and machine costs. Additional costs must be compared to the increased die lifetime to ensure an economical application [5]. For this purpose, the used powder and the L-PBF-fabricated samples were characterised with regard to defects and mechanical material properties. In order to determine stable cooling channel geometries, numerical investigations were carried out in this work with regard to die loading and the design of the cooling channels. Finally, a prototype featuring the most suitable channel geometry was manufactured.

2. State of the Art

Die forging is known for its wide range of applications, for example in automotive engineering to produce crankshafts. Compared to cold forging, a greater degree of deformation is achievable for the production of more complex components, which result from the reduced yield strength of heated semi-finished products. An important factor influencing the economic efficiency of forging processes is die lifetime. Maximum thermal loads occur in the final stages of the forming process, but cooling lubricants can only be applied before the forming process [6]. Figure 1a shows a calculated temperature distribution at the surface for a forging die which was subjected to concentrated thermal loads at the convex mandrel radius. In this case, the semi-finished product (Ø 30 mm, height 40 mm, AISI 1045) was heated up to 1200 °C, the die temperature was adjusted to 150 °C, the cycle time of the hole process was 8 s and the temperature distribution was simulated at the end of the forming process [7]. In addition to the simulations, Behrens et al. [8] also conducted experimental tests on this process. With their research, it could be proven that abrasive wear and plastic deformation were present in the mandrel area after 1000 strokes (Figure 1b) [8].

Different approaches have been explored to reduce abrasive wear and plastic deformation. For example, previous investigations of hybrid-forging dies were often carried out with wear-resistant die inserts. Kwon et al. [9] developed ceramic zirconia inserts that were thermally shrunk into forging dies and could withstand high thermal and mechanical loads. No cooling channels were present in these inserts and yet increased wear resistance could be achieved as a result. However, ceramic inserts feature a low ductility and are prone to material breakouts [9]. Another approach deals with the application of cooling channels. By using internal cooling, the thermal softening of the near surface layers of forging dies can be reduced. This cooling concept is state of the art for injection moulding but is challenging in die forging due to high mechanical die loads [10]. Accordingly, there are currently only initial approaches to the use of internal cooling channels in hot-forging dies, such as in the work of Behrens et al. [11]. In their work, two different approaches to the use of the channels were designed and numerically investigated. According to the numerical investigations, both variants were suitable for use in the forging process. The research in the comprehensive review paper by Feng et al. [12] makes it clear that almost no research has been done on internal cooling channels in the field of hot forging. Only the work of Behrens et al. [11] is cited for this application area. For the field of injection moulding, different approaches are extensively presented [12]. In the context of the work of Kwon et al. [9], a suitable geometry with parameter combination for the cooling channel within the process of injection moulding was calculated by means of a mathematical optimisation. The focus was on optimised heat dissipation. Another application of internal channels is presented in the work of Delikatas et al. [13]. Here, the die for cold extrusion was produced with the help of additive manufacturing. Cooling channels as well as channels and reservoirs of lubricant were introduced inside the die. These dies were tested numerically and experimentally for their usability. Overall, the cooling and lubricant channels had a positive influence on the process.

L-PBF is a layer-by-layer additive manufacturing (AM) technology. The L-PBF process has been established rapidly in recent years to produce metal parts directly according to CAD models. The principle of L-PBF is based on the local melting of fine metallic powder layers by a high-power laser beam. The interaction of the laser with the metal powders and the solidified layers regarding the material quality is mainly influenced by the laser power, laser scan speed, hatch distance and layer thickness [14]. By building up the model in layers, it is possible to implement more complex geometries than with standard processes. Examples for the use in the production of dies for forming processes are mentioned in [15]. The authors additively manufactured dies by different techniques (layer-lamination, powder bed and powder-nozzle-based).

The growing demand of L-PBF parts has increased the interest of processing application-oriented materials. The die- and the mould-producing industry requires adapted steel alloys with a high wear resistance. Krell et al. [16] focused on the development of processing carbon-alloyed martensitic die steel powders for hot-working applications. One promising aspect is the processing of this approach to achieve the needed mechanical and thermal resistance to endure high mechanical and thermal loads [17]. The main challenge of processing carbon-alloyed die steels is crack avoidance and the successful processing of the L-PBF-fabricated inhomogeneous segregated microstructure caused by process-related high cooling rates. High residual stresses because of a fine martensitic microstructure and a high amount of retained austenite lead to a high chance of crack formation during the L-PBF process. Substrate preheating is used to achieve a homogeneously distributed microstructure in combination with isothermal phase transformation effects and thus reducing crack formation during the L-PBF process [18].

In order to calculate the high thermomechanical loads of a hot-forming die, it has become established to carry out a finite element (FE) simulation for the design of the dies in advance. In some areas, plastic deformations can occur, which can lead to material fatigue and finally to die failure due to cracks. For this reason, the FE method is increasingly used to determine the local stress values for the design of dies [19]. Optimised numerical die design can take into account the process variables that occur in the manufacturing process [20] and enable a more efficient design of the dies and step sequences. As a result, an optimisation of the die configuration is already possible in the design phase of the forging process through stress-adapted dimensioning with little experimental effort [21]. For this purpose, it is important to determine the occurring temperatures and stresses as well as the effects of design variations [22].

3. Materials and Methods

3.1. Characterisation of the L-PBF Process

The successful L-PBF fabrication of high dense parts with a low number of defects depends on the process parameters and the properties of the steel powder. In this study, a gas-atomised hot-work tool steel AISI H10 powder was used to fabricate hybrid-forging dies with inner cooling channel structures. The hot-work tool steel AISI H10 powder was produced by Atomising Systems Ltd. UK, Shellfield, UK in a powder fraction of 10 µm to 45 µm [23]. This provides a homogenous powder bed and melting behaviour of the powder. Scanning electron microscope (SEM) investigations have indicated a high sphericity of powder particles. Spherical powders feature advantageous flow behaviour and are optimally suited for L-PBF processing. The morphology of the particles is shown in Figure 2.

The chemical composition of the used powder is listed in

Table 1. Chemical composition of hot-work tool steel AISI H10 powder according to the manufacturer’s specifications [23] and verification of the data in our own laboratory tests with combustion analysis and X-ray fluorescence analysis.

Elements (mass-%)	C	Si	Mn	Cr	Mo	V
	0.28	0.20	0.40	3.20	2.70	0.40

according to the manufacturer’s specifications.

In this research a parameter study was performed to achieve crack-free and high-density L-PBF-fabricated samples made of hot-work tool steel AISI H10. In total, 33 parameter variations were tested on the L-PBF system (Concept Laser M2). In this case, cube-shaped specimens with a dimension of 10 × 10 × 10 mm³ were chosen for the parameter study. Three cubes were fabricated for each parameter setting. The laser power was varied between 145 and 300 W and scanning speeds were adjusted between 500 and 1300 mm/s. The powder layer thickness of 30 µm, focus diameter of 150 µm, hatch distance of 105 µm and oxygen content < 0.05% were constant. In addition, the substrate preheating temperature was varied between 200 and 250 °C.

All L-PBF-fabricated cubes were austenitised at 1030 °C for 25 min in an air atmosphere (chamber furnace) and quenched in oil at room temperature. Tempering was performed in three stages. First, tempering was carried out at a temperature of 475 °C for two hours. The second and third tempering stages were performed in a range of 475 to 625 °C for two hours to determine the hardness-tempering behaviour of the L-PBF-fabricated AISI H10 hot-work tool steel. Hardening and tempering temperatures were based on conventional cast AISI H10 hot-work tool steel [24]. The results of the optimal heat treatment route are presented in Section 4.1. Additional cuboid-shaped specimens (90 × 11 × 11 mm³) were built up by L-PBF. These cuboid specimens were processed for the determination of the tensile strengths according to ISO 6892-1:2019. A total of six specimens were examined at room temperature with the same heat treatment route. The samples were printed so that the direction of tension was the same as the subsequent direction of loading in the forming process. Throughout the entire test, the tensile speed was kept constant at 1 mm/s and the test was stopped when the sample failed.

The material density of the L-PBF-fabricated cubes was determined according the Archimedes’s method using deionised water at room temperature of 21 °C (Kern & Sohn balance ALS 220-4). The hardness of L-PBF-fabricated cubes was measured with Rockwell C in as-built and heat-treated (quenched and tempered) conditions (GNEHM, Rocky 2000). L-PBF defects, porosities and microstructure analyses were performed by using optical microscopy (Olympus BX 51, Olympus Stream image analysis software). L-PBF-fabricated cubes were ground and polished for metallographic investigations. Optical microscopic sample preparation was carried out by grinding with 220, 500, 1200 SiC paper and polishing with 3 and 1 µm suspensions. The microstructure was analysed in etched condition (3% nital etchant). In addition, the morphology of the AISI H10 powder particles were investigated by using a secondary electron microscope (SEM) (Tescan, Mira 3). Furthermore, an identification and a quantification of martensite and retained austenite was carried out by using X-ray diffraction (XRD, Bruker, D8 Advance, Mo Kα1 radiation).

3.2. Design of L-PBF Hybrid-Forging Dies

The cost efficiency via increased die life by internal cooling must be evaluated considering the additional costs associated with using L-PBF parts, in order to be industrially utilised. Compared to conventional production, L-PBF is slower and more expensive. Thus, only parts of the die with cooling channels were produced using L-PBF. The other parts such as the base of the mandrel were conventionally machined. Based on these reasons, a hybrid die design was applied, in which only components with conformal cooling channels were produced by L-PBF technique. The remaining subcomponents were machined as shown in Figure 3. This enabled a reduction in manufacturing time and costs compared to manufacturing purely by the L-PBF process.

In order to achieve a design suitable for production, the complete manufacturing process chain of the dies must be taken into account. L-PBF processes require a flat surface with free accessibility from all sides, which is why the mandrel was designed as a separate component from the surrounding die engraving. The production of the hybrid die was planned in such way that in a first step, a base of the upper die and the base of the mandrel were produced by conventional machining processes (Figure 3). Subsequently, the mandrel was built up by L-PBF directly onto the mandrel base, which features inner cooling channels. Components manufactured by L-PBF have an increased roughness, and therefore both components were oversized for further finishing by machining. The postprocessing of the heat-treated material’s bonded die components was performed in the third step. Afterwards, the die engraved component was heat-treated as well (step 4). Finally, all components with the narrow tolerances H7/g6 were assembled manually by hand (step 5) and finished off in step 6. The assembled components can be installed in the forming machines via a clamping ring on the outer radius. Subsequent surface modifications, i.e., diffusion treatments and coatings, can further increase the wear resistance.

Different geometries of the conformal cooling channels were designed using CAD and numerically investigated. The focus was on the question of whether the design can withstand the high loads during die forging and provide a high heat dissipation. In this process, the inner cooling channels were designed to match the mandrel contour. Thereby, the shape (F) of the inner cooling channels was changed, i.e., circular and elliptical variants were investigated. Furthermore, the channel diameters (Ø) were varied, with larger channel diameters featuring an increased heat dissipation and reduced load capacity of the dies. Moreover, the surface distances (S) and the rotational axis distances (R), which also have an influence on the cooling and the load capacity, were varied as well. The influence of each parameter was determined by applying two variations each. A total of eight different variants of the circular channel shapes and four for the elliptical shape (in total twelve variants) were worked out.

3.3. Numerical Simulation

The aim of the numerical finite element method (FEM) in this work was to evaluate the twelve channel variants regarding mechanical stability during the forging process. In this research, the numerical simulation was based on an uncoupled die stress analysis. At first, the forming process using numerically rigid dies was calculated. Then, the stress analysis on the then elastic deformable die was conducted in an additional simulation step. In this decoupled forming process analysis, the loads on the rigid dies are imported as reaction forces into the subsequent die analysis and the stresses in the dies are determined. In addition to this approach, there is also the coupled approach [25]. In the coupled simulation, the die is modelled as an elastic deformable body. Since in this work the focus was on identifying the influence of the individual design parameters on the occurring stresses in the die, the approach of an uncoupled simulation can be chosen. Based on the simulations, a suitable die variant can be recommended for the subsequent experimental tests.

All simulations of the considered hot-forging process were carried out with Simufact Forming (16.0 SP1, Simufact Engineering GmbH) software. The choice of the press parameters provides an abstraction for the press that is later considered for the experimental tests. A constant speed of 250 mm/s and a stroke of 35.15 mm (100% stroke) were defined, which were chosen at the beginning because of previous tests of this process with a conventional die. The friction between the die and semi-finished product was modelled using a combined friction model with a friction coefficient µ of 0.15 and a friction factor m of 0.30 [26]. An initial temperature of 1250 °C was assigned to the semi-finished product (dimensions: 30 mm diameter and 40 mm height). This temperature was slightly reduced by a cooling time of five seconds, which depicts the transfer from the heating unit to the forming die. The thermomechanical properties of the used semi-finished product AISI 4140 (42CrMo4) were taken from a previous work described in [19] and modelled using the analytical approach by Hensel and Spittel. The preheated dies were set to a starting temperature of 180 °C.

The process simulation model included the lower die, the semi-finished product as well as the hybrid upper die (Figure 4). The upper die itself was divided into three individual parts, the die engraving, the base of the upper die and the mandrel with integrated cooling channels. In the numerical study, only the mandrel geometry was changed according to the parameters in Figure 3. According to the procedure for the decoupled die stress analysis, all parts of the die were modelled as rigid bodies in the forging process. For an efficient design process, a planar symmetry was used and only a 3D half model was employed. For the subsequent die load analysis, the components were meshed with tetrahedral elements. Thereby, the mandrel had smaller elements of the mesh especially in the lower area of the cooling channels (minimum element edge length of 0.1 mm). The choice of mesh size was based on the geometric specifications. This means that the mesh parameters in this research were chosen so that the radii were always mapped by several elements. The main area for this research was in the mandrel area, where the transition from vertical to horizontal cooling channels is located. In addition, the smaller horizontal cooling channels with a dimension of 1.5 mm to 2.0 mm should be mapped by more than ten elements. Due to these geometric conditions, different element edge lengths between 0.5 mm and 0.05 mm were investigated. Overall, a minimum element edge length of 0.1 mm turned out to be the best option in terms of the accuracy of the results and the computing time. Since it can be assumed that the other two components did not have such critical and geometrically complex areas, a slightly larger element edge length with a minimum of 0.2 mm was selected here. Overall, the simulation was carried out by the implicit MSC Marc solver.

Within the following die stress analysis, the calculated loads of the hot-forging process were mapped on the upper die and the corresponding elastic material properties were assigned. For the mandrel, the properties of L-PBF-processed hot-work tool steel AISI H10 were used. Within the die stress analysis, the dies were considered to be only elastically deformable. The resulting stresses, in particular the maximum principal stress and von Mises’s stress, were compared with the material specific properties. By matching the equivalent stress with the yield strength according to the hypothesis of von Mises and the maximum principal stress with the tensile strength according to the principal normal stress hypothesis of Lamé and Rankin, it is possible to estimate how likely failure due to cracks or plastic deformation is, within the first cycle, as a result of die overload [27]. The corresponding material parameters yield strength and tensile strength were determined for the L-PBF-processed hot-work tool steel AISI H10 within the tensile tests (described in Section 3.1). In addition, the material data were supplemented with literature values [28] and data from JMatPro. Data for the conventionally produced AISI H10 steel were determined using JMatPro. Since the density analysis and the micrograph analysis (comparison Section 4.1) showed that a material with a high density of about 99.95% can be produced, a homogenous material without defects was assumed. For the initial evaluation of the die stability with regard to the inserted cooling channels, an isotropic material behaviour was also assumed. The tensile tests carried out to determine the critical material properties showed the same printing direction to the main loading direction as in the die considered later on.

3.4. Comparison and Modification of the Variants

The simulation results were examined in relation to the equivalent stress and the maximum principal stress in order to find the ideal channel geometry for the trade-off between die weakening and maximum heat dissipation. With the help of Minitab, a correlation analysis was carried out and a superimposed contour plot was created. In this way, we determined which parameters were recommended to stay below the tolerable equivalent stress and maximum principal stress. Based on these results, a final mandrel geometry with the determined ideal parameters was designed and numerically investigated. The calculated result was compared with the Minitab evaluation to validate the superimposed contour diagram.

4. Results and Discussion

4.1. L-PBF Process

As part of the parameter study, 33 different combinations of L-PBF process parameters were investigated. The quality of the parameter combination was determined via defect analysis and hardness. In this study, the laser power was adjusted in a range of 145 to 300 W. It was found that a laser power of 225 W was promising. If the laser power was adjusted too low, the thermal input was not high enough and local areas were not completely melted. This led to an increased formation of fusion defects. In case of higher laser powers, the defect density and the porosity were increased as a result of keyhole formation. This correlation was additionally influenced by the scanning speed, because the laser interacted with steel powder for a shorter or longer duration. For a laser power of 225 W, a scanning speed of 700 mm/s was determined as suitable, which can be classified in the lower examination interval (500–1300 mm/s). The substrate preheating temperature showed an influence on the temperature gradient. Consequently, an improvement in cracking behaviour was generally observed at a substrate preheating temperature of 250 °C, compared to 200 °C. The process parameters hatch distance of 0.105 mm, layer thickness of 0.03 mm, focus diameter of 0.150 mm and shielding gas N₂ were not varied, but these parameters were found to be suitable in preliminary test series [29].

Table 2. Used L-PBF parameter setting for hot-work tool steel AISI H10 with a focus diameter of 0.150 mm (temp. = temperature; Vol. = Volume).

Subject	Laser Power (W)	Scan Speed (mm/s)	Hatch (mm)	Layer Thickness (mm)	Substrate Preheat Temp. (°C)	Vol. Energy Density (J/mm3)	Shielding Gas (-)
Inspected range	145.0–300.0	500.0–1300.0	0.105	0.030	200.0–250.0	-	N2
Optimal combination	225.0	700.0	0.105	0.030	250.0	102.0	N2

summarises the most promising L-PBF parameter setting for processing the investigated hot-work tool steel AISI H10 on Concept Laser M2, and all subsequent results refer to this parameter combination in this study.

As mentioned above, cube- and cuboid-shaped specimens were produced. Based on these specimens, a tempering chart was determined, which is shown in Figure 5. A detailed description of the heat treatment is given in Section 3.1. In a temperature range of 475–575 °C, high hardness values of approximately 51–53 HRC were achieved, which are comparable to conventionally processed and heat-treated hot-work tool steel AISI H10. As a result of the tempering study, the secondary hardness maximum deviated from conventional steel, resulting in a hardness of 51–52 HRC being achieved at approximately 550 °C. Accordingly, all following components and specimens produced by L-PBF processes were heat-treated at this temperature.

XRD investigations were carried out on the L-PBF-fabricated specimens in the as-built condition and after heat treatment to characterise the phase distribution. Figure 6a shows the results, with the black curve describing the as-built condition and the red curve corresponds to after the heat treatment. Both curves show characteristic increases in intensity at certain 2-theta angles, describing the present phases. In the as-built condition, the γ(200), γ(220) and γ(311) structures are detectable. In the case of high thermal and mechanical loads of the forging dies, these unstable austenitic phases can be transformed into martensite. The resulting increase in volume promotes the formation and the propagation of cracks. After heat treatment, minor increases in intensity are detectable. Accordingly, minor amounts of retained austenite are present and the cracking behaviour is improved. Figure 6b shows the microstructure in the as-built condition (top) and after heat treatment (bottom). Compared to the as-built condition, a fine-grained microstructure can be observed after heat treatment. By means of hardness tests, an as-built hardness of 44.3 ± 0.6 HRC was determined and for the fine-grained structure after heat treatment, a hardness of approximately 51.3 ± 0.3 HRC. Furthermore, a cross-sectional metallographic image of the cube-shaped sample is shown in Figure 6c. It can be observed that there is no cracking near the surface. In addition, a high relative density of 99.95% was determined. Specimens in the as-built condition show an Archimedean density of 7.850 ± 0.003 g/cm³. Accordingly, it was proven that the L-PBF process parameters (Table 2) were suitable for crack-free processing of the hot-work tool steel AISI H10 powder to high-density specimens.

Finally, the mechanical properties of the L-PBF-processed hot-work tool steel AISI H10 were determined. All specimens were fabricated using the L-PBF process parameters described previously (Table 2) and heat-treated as described in this Section 4.1. The uniaxial tensile tests were carried out at room temperature. The effective direction of the tractive force ran in the same direction as the later load in the forming process. A tensile strength of 1778 ± 38.3 MPa, yield strength of 1392 ± 42.9 MPa, elongation of 9.0 ± 1.0% and reduction of area of 18.9% were determined as part of the investigations. The results of the tensile and the hardness tests are summarised in

Table 3. Determined material data of L-PBF samples at room temperature by tensile tests.

Hardness (HRC)	Tensile Strength Rm (MPa)	Yield Strength Rp0.2 (MPa)	Elongation A5 (%)	Reduction of Area z (%)
51.3 ± 0.3	1778.0 ± 38.3	1392.0 ± 42.9	9.0 ± 1.0	14.7 ± 3.1

. Overall, the results are in a comparable but slightly lower range of values compared to conventionally produced samples from this material. These parameters were used in the following process simulation.

4.2. Design of L-PBF Hybrid-Forging Dies

In Section 3.2, the design of hybrid-forging dies was explained. The turned and milled die engraving was identical for all twelve variations of the hybrid-forging dies. In the case of the base of the upper die, which was also produced through conventional machining processes, only the rotational axis distance of the media feed was adjusted. For the L-PBF-fabricated mandrel, different channel geometries were developed. The defined geometries of the conformal cooling channels are presented in Figure 7.

Eight designs (Variant 01 to Variant 08) featured circular channel geometries, the other four an elliptical one (Variant 09 to Variant 12). The selected channel diameters were 1.50 mm and 2.00 mm. Furthermore, different channel distances to the surface were designed (3.25 mm and 5.00 mm). The rotational axis distances were defined at 4.00 mm and 5.75 mm.

The transition radii to the medium feed were particularly challenging because they usually tended to increase the stresses and therefore the risk of failure. Thus, there were no abrupt geometrical changes. The cross-section from media feed to mandrel was reduced, to achieve an increased media flow in the channels.

4.3. Numerical Simulation

For a reliable evaluation and selection of the variants, the allowable stress must first be identified from the determined material data. For the maximum principal stress, a value of approx. 1400 MPa at 200 °C is acceptable. This corresponds to the approximate tensile strength at 200 °C based on the material data at room temperature (Table 3) and the comparison with conventionally produced samples. The determined material data were extrapolated to 200 °C. Then, the value determined in this way was compared with material data from conventionally manufactured samples. The temperature of 200 °C was chosen because previous numerical and experimental temperature tests on conventional dies have shown that it is optimal [7], since the highest stresses are not located in the surface area but in the area of the cooling channels for this kind of construction. The assumption is based on Figure 1 [7] and the starting temperature of the upper die. For all considered channel variants, the maximum principal stress was within the tolerable range at a stroke of 100%. Thus, there was no risk of die failure due to cracking (Figure 8).

To estimate the risk of plastic deformation of the dies, the von Mises equivalent stress is compared to the yield strength. To avoid plastic deformation the von Mises equivalent stress should not exceed 1100 MPa, which corresponds to the extrapolated yield strength at 200 °C. It was found that in relation to the equivalent stress, none of the selected parameter combinations could withstand the occurring load. Thus, the targeted press stroke was reduced, which resulted in an increased flash thickness and decreased maximum forming load. Besides a stroke of 100.0% (35.15 mm), two reduced stroke set points of 97.5% and 95.0% were analysed. Despite the reduction of the stroke (97.5% and 95.0%), a complete mould filling of the upper die could be achieved with each variant.

In the following, Variant 03 (F = circular, Ø = 2.00 mm, S = 3.25 mm and R = 4.00 mm) and Variant 10 (F = elliptic, Ø = 2.00/1.50 mm, S = 5.00 mm and R = 5.75 mm) of the twelve variants are compared and discussed in detail. Variant 03 had a smaller surface distance and a smaller rotation axis distance compared to Variant 10. In addition, they differed in the shape of the channel. For Variant 03, an equivalent stress of 2180 MPa was reached in the inner circuit of the cooling channel at a stroke of 100.0% (Figure 9a). A reduction of the stroke to 97.5% led to a reduction of the occurring equivalent stress (Figure 9b). However, the equivalent stress still exceeded the threshold of 1100 MPa. A further reduction of the stroke to 95.0% led to an equivalent stress that was below 1100 MPa (Figure 9b). Overall, plastic deformation can only be avoided through a significant reduction of the stroke. For all illustrations, the maximum value of the legend has been set to the described maximum sustainable equivalent stresses.

The best parameter combination of the twelve models examined was found in Variant 10 (F = elliptic, Ø = 2.00/1.50 mm, S = 5.00 mm and R = 5.75 mm), shown in Figure 10. However, even this variant showed an equivalent stress of 1500 MPa at 100.0% stroke. The highest value of the equivalent stress was close to the level where the cooling channels change from the horizontal direction to the vertical direction and there was a reduction of the channel diameter (Figure 10a). By reducing the stroke to approx. 97.5%, the equivalent stress could be reduced to such an extent that the maximum equivalent stresses were below 1100 MPa (Figure 10b). Thus, the calculated equivalent stress was below the given yield strength at 200 °C so that the risk of plastic deformation was reduced.

When comparing the two variants presented here, in addition to the occurring stresses such as the equivalent stress or the maximum principal stress, the possible cooling effect due to the position of the cooling channels can also be considered. It can be assumed that Variant 03 has a higher cooling capacity due to the small surface distance. Variant 10, on the other hand, has a higher rotational axis distance, which is why the cooling channels are closer to the slope in the mandrel area. In addition, the oval cooling geometry has a larger surface area, which enables a higher heat exchange. Accordingly, Variant 10 is also to be preferred in terms of the expected cooling effect, although experimental tests on the cooling effect still have to be carried out.

4.4. Comparison and Modification of the Variants

With the results of all simulations at 97.5% stroke and the specified result ranges of the equivalent stress and the principal stress from Section 4.3, a superimposed contour diagram can be created using the statistical software package Minitab (Figure 11). The tolerable range is given for the equivalent stress and the maximum principal stress. In each case, the dashed line indicates the highest set value and the solid line the lowest set value. Based on the tensile strength of the material from the L-PBF process and information from the literature, the tolerable principal stress was determined. The positive maximum principal stress can be neglected when choosing the parameter combination, as the previous investigations showed that the occurring equivalent stresses were more critical. The limits for the equivalent stress were set close to the yield strength for approx. 200 °C (1100 MPa), so that a tolerable range could be recognised on the basis of the calculated results. These created lines represent limit functions based on the calculated stresses (equivalent and principal stress) from the simulation. Accordingly, the calculated values of the different variants represent the supporting points of the curves. In both diagrams, the support points used and thus also the variants examined are marked by crosses in circles. Some of the lines are outside the definition range of the axes due to the shift and are therefore not recognisable. The definition ranges of the axes were chosen in such a way that they still show values that can be converted in reality.

By specifying the tolerable mechanical stresses, a range was obtained, in which the variants with tolerable stresses were located. This area is marked yellow in the diagrams. The tolerable range of stresses and the axes show which parameter combination would have to be selected in order to achieve the tolerable stresses (comparison of importance of variables, Figure 3). The axes indicate the variables that were varied in this study. For better comprehensibility, the numerical results of the circular (Figure 11a) and elliptical (Figure 11b) channel shapes are considered individually. In the circular design, the influence of the surface distance cannot be mapped due to too many variables. Therefore, the parameter setting with the lower stresses in the simulation was chosen, which is why a distance of 5.00 mm was specified.

At closer inspection of the circular cooling channels, it is noticeable that no variant fits in the target range. When comparing the elliptical channel geometries, one variant (V. 10 = Variant 10), which has also already been presented (Figure 10), lies in the area marked in yellow (in the light blue area). The arrangement of the cross shows that it is not close to the established limits for the equivalent stress and the principal stress. This makes it clear that the presented variant has lower equivalent stresses than the maximum permissible. Accordingly, this variant has a parameter combination that is desirable for achieving values below the maximum tolerable stresses.

In the comparison of the two diagrams, it is noticeable that there are differences in the tolerable range. On the one hand, the yellow area is larger for the elliptical channel geometries than for the circular shapes. On the other hand, a higher rotation axis distance is desirable for the elliptical variants. The circular variants should have a rotation axis distance between 3.90 mm and 4.60 mm depending on the channel diameter. For the elliptical variants, a rotation axis distance between 4.90 mm and 5.75 mm can be selected depending on the surface distance. Accordingly, it is possible to get closer to the surface, i.e., further away from the centre of the die, with the elliptical variants. The variable of the surface distance should also be close to 5.00 mm with the elliptical shape as is the case with the circular channel shape. All in all, it can be stated that the elliptical shape is more suitable with regard to the stresses that occur. A plausible reason for this is the transition area to the media feed, which differs from circular channel geometries and where the maximum stresses occur. Due to the channel routing, there is a homogeneous course of the surface without hard changes, whereby stress peaks occur. Accordingly, there is a more even distribution of forces in the mandrel, which leads to a reduction in stress peaks.

Based on the Minitab evaluations, an additional circular model (Variant 13) was designed and numerically investigated. This variant had the following parameter combinations: diameter (Ø) of 1.35 mm, rotational axis distance (R) of 4.00 mm and surface distance (S) of 5.00 mm. In the evaluation, the focus was placed on the stress load at 97.5% stroke in relation to the equivalent stress according to von Mises (Figure 12). A maximum value of 1050 MPa was calculated in the numerical simulation. Thus, the stress was below the maximum target stress of 1100 MPa and was within the yellow range in comparison with Figure 11. Accordingly, it can be stated that the established Minitab representation can be used for the selection of suitable parameter combinations. However, Variant 13 should be further tested with regard to the effect on the permeability of the cooling medium due to the small channel diameter before possible experimental forging tests are carried out.

Compared to Variant 10, the newly designed Variant 13 had a smaller channel diameter in addition to the different channel shape. Due to the smaller channel diameter, it could be assumed that the cooling capacity was smaller due to a smaller surface area and accordingly also less cooling medium. In addition, the pressure and the flow characteristics inside the channel were altered due to the change in cross-section from the inflow to the mandrel area. This can lead to further problems that were considered in this work. These include, for example, the reduction of heat removal due to an increased flow velocity or the blockage of the channels due to increased turbulent flow. Furthermore, Variant 13 had a smaller rotation axis distance (Variant 13 with 4.00 mm and Variant 10 with 5.75 mm). This means that the cooling channel was located further inwards in the mandrel and the cooling effect was probably further reduced. Overall, the focus for the experimental forging tests were therefore placed on Variant 10. This variant represented a good compromise between sufficient stability and probably adequate cooling effect with a not-too-high pressure increase.

4.5. Prototype Processing

A prototype of Variant 10 (F = elliptic, Ø = 2.00/1.50 mm, S = 5.00 mm and R = 5.75 mm) successfully passed through the manufacturing sequence previously shown in Figure 3. The L-PBF mandrel in the as-built condition is shown in Figure 13a. Macroscopically, an increased surface roughness can be observed. Due to the different surface conditions, a distinction is noticeable in the joining zone between the mandrel base and mandrel. This issue is considered in the manufacturing sequence by oversizing the built-up structure. Figure 13b shows the die components after heat treatment and postprocessing. A joining zone between both components is hardly recognisable and indicates a sufficient material bond. The manually assembled hybrid-forging die is shown in Figure 13c. This prototype will be used in future forging tests and analysed in terms of wear. In addition, these tests serve to validate the simulation results and to evaluate the potential of the cooling channels.

5. Conclusions

In this study, hybrid-forging dies featuring conformal cooling channels were designed. Suitable parameters for the L-PBF process were determined, and different design possibilities were numerically analysed and validated by prototype processing.

The process parameter study for L-PBF-fabricated AISI H10 hot-work tool steel showed a high relative density of 99.95% and a low number of defects, such as pores. Substrate preheating to 250 °C was necessary to produce crack-free hot-work tool steel AISI H10 samples. An optimised material-specific heat treatment strategy for L-PBF-fabricated samples shows promising mechanical properties. In the heat-treated condition, a hardness of 51.3 HRC, a tensile strength of 1778.0 MPa, a yield strength of 1392.0 MPa and an elongation of 9.0% could be achieved at room temperature.

The hybrid-forging dies were designed in such a way that the use of the L-PBF process was limited to the mandrel, which is subject to high thermal loads, thus achieving economical use compared to conventional hot-forging dies. The mandrel base and the die engraving component could be conventionally manufactured. All die components could be assembled manually.

Numerical investigations showed a significant influence of the shape parameters such as the channel diameter and surface distance in the resulting stresses within the die. In addition, a dependency between the individual parameters could be demonstrated by the Minitab evaluation. However, all channel geometries investigated led to high equivalent von Mises stresses and indicated a risk of plastic deformation at a stroke of 100.0%. Therefore, stroke reduction was taken into account. The targeted 100.0% stroke was chosen based on previous reference forging tests with the same geometry as a full die. Another conclusion from the results is that in future applications of cooling channels, attention must be paid to the stresses within the die. Finally, an elliptical channel shape offered the best results. At a stroke of 97.5%, no critical stresses regarding crack formation and plastic deformation occurred and complete form filling was achieved. To determine a possible parameter combination for the circular channel shapes, the analysis with the statistical software package Minitab was used. Using this software package, it was possible to design another parameter combination that achieved equivalent stresses below the maximum tolerable equivalent stress at a 97.5% stroke.

Based on the results, a prototype of the most promising elliptic variant was manufactured using the determined L-PBF parameters and developed manufacturing concept. The hybrid L-PBF forging dies will be subsequently tested in experimental forging tests regarding wear reduction by means of internal cooling channels and the achievable cooling effect. Furthermore, the experimental forging tests will be used for validation of the numerical simulation for the respective variants.