Understanding the Impact of Cooling Systems on Bulk Additive Friction Stir Deposition of Aluminum

Abstract

This work presents an investigation of two cooling systems used in additive friction stir depositions (AFSD) and the related effect on process temperature, feed material, and life of processing equipment. A new AFSD cooling system, Mazak MegaStir Liquid-cooled Toolholder (LCTH), was integrated onto MELD Manufacturing AFSD machines using both continuous and discrete material feeding systems. The LCTH is compared to the original MELD cooling system to understand how process temperatures are controlled by the cooling systems, how feed material interacts with the process under different cooling conditions, and how well the cooling systems protect the equipment. It is shown that spindle revolution rate impacts the temperature experienced by the bearings more greatly than the utilization of cooling. The attempt of three large volume depositions demonstrates the necessity of a cooling system utilization for AFSD. Additionally, four configurations of tool and cooling system are analyzed to understand parameters that affect depositions temperature and the thermal effect on feed material. Configurations with short working face to cooling distances show ~40 K cooler deposition temperatures and configurations with the new LCTH deposited ~9 K cooler than those deposited with the original MELD cooling system. Microhardness analysis on feed material revealed the varying efficacy of each cooling configuration’s ability to enable bulk deposition. The influence of feed system on process temperatures is also presented with a comparison of discrete and continuous feed system depositions.

1. Introduction

Metal-based additive manufacturing (MAM) is of interest to many industries ranging from biomedical to aerospace for both repair and new build of many materials [1]. MAM is divided into two major categories: solidification-based and solid-state-based processes [2,3]. Solidification-based processes comprise three main types: powder bed fusion, binder jetting, and directed energy deposition [4]. With both advantages and disadvantages, each of these MAM categories has a specific use case that it is well suited for. Solidification-based MAM tends to have defect types including: lack of fusion, voids, porosity, cracks, delamination, and residual stresses [5,6]. These defects are inherent to the materials and the processes utilized.

Additive friction stir deposition (AFSD) is a new large area, solid-state MAM process. AFSD is capable of many additive techniques: deposition, joining, repair, and cladding [7,8]. The solid-state nature of the process prevents many of the solidification-based defects previously discussed [2]; however, delamination and residual stress may still be concerns [9,10,11,12]. AFSD has the potential to be a replacement technology for forging components [13,14,15]. AFSD is capable of depositing in an open environment, allowing for large depositions with less size constraints than powder bed fusion or binder jetting [16,17]. The use of solid bar feed stock also makes AFSD more environmentally conscious than powder or wire-based feed common in powder bed fusion, binder jetting, and directed energy deposition [8]. AFSD has a deposition rate that is approximately ten times faster than solidification processes [18]. The rapid, solid-state, and open environment capabilities of this process are some of the many advantages that the AFSD process has over traditional methods.

AFSD is a relatively new MAM process, but it is being widely studied. The process has been evaluated for many material systems: Al [19,20,21,22], Mg [23,24], Fe [14,25], Ni [26,27], Ti [28], Cu [29]. Generally, solid feed material of ~9.5 mm² or ~13 mm² cross section is used, but powder and chips are also possible formats for feed material [17]. AFSD process deposits through a thermomechanical mechanism utilizing frictional heating and rotational deformation rather than a solidification mechanism [30]. The process is capable of being utilized by large-scale machines for components or deposition, currently the largest machine has a 6 × 9 × 4 m print volume possible [31,32]. The use of solid bar feed material and thermomechanical processing enables successful depositions across a wide variety of material systems and component sizes.

Work has been done by various groups to understand how various parameters of AFSD impact the deposition’s thermal history and microstructure. In a work by Qiao et al. [33] the impact of spindle rotation rate on thermal history was explored. Higher rotation rates produced depositions with larger grain size and reduced precipitation strengthening. Hoarston et al. [34] varied spindle rotation, traverse, and feed rates that resulted in varying surface quality and varying heat input to the deposition. Increased heat input to these depositions negatively impacted depositions surface quality. Additionally, Qiao et al. [35] analyzed depositions performed at ambient temperature and ones cooled during deposition by a water bath. Depositions performed in the water had more favorable properties and were processed at a lower temperature. These three studies all manipulate the thermal condition of AFSDed builds through various mechanisms, i.e., spindle rotation rate, heat input, and external cooling on the deposition.

This work will explore the impact of varying the cooling system and distance between working face and cooling on various configurations; these changes will impact the thermal gradients in AFSD tools. Additionally, deposition size will be explored in this work. One of the next critical steps for industrial implementation of AFSD is transitioning the scale of AFSD from academic- or research-size machines to industrial. Despite numerous material systems being explored in the literature, many studies continue to use very small deposition sizes that exhibit vastly different thermal conditions than what has been observed at the industrial scale. The transition from academic size-machines and depositions will require an understanding of the role of cooling in the AFSD process. This understanding will need to include the thermal impact on feed material, equipment life, and the process during AFSD, in order to facilitate a transition from small research depositions to large, full-scale components for industry application. This work presents data that expands the understanding of the impact that cooling configurations and deposition methods (i.e., discrete or continuous) have on these areas of focus. This work can inform the transition from academic or research-scale depositions to industrial scale.

2. Materials and Methods

2.1. Machinery

For this work a MELD L3 and a CD-14, from MELD Manufacturing Corporation (Christiansburg, VA, USA), were used for AFSD builds. Both systems use a hollow rotating 38 mm diameter tool through which solid, square feed material is fed, see Figure 1a A and B. The feed material is deformed against the previous layer or substrate (see Figure 1a A and C or Figure 1a A and D), which generates heat and joins the two together through diffusion bonding that has been accelerated with increases in temperature and pressure. The L3 in this study has a discrete feeding system so that feed material is printed one piece at a time with a manual reload and interruption (see Figure 1b); however, the CD-14 has a continuous feeding system that loads new material, so the printing is continuous and without interruption, see Figure 1c.

Both the L3 and CD-14 have a manufacturer cooling system that consists of an upper and lower seal with coolant flowing directly in contact with the tool; this system is referred to as MELD throughout. A new AFSD cooling system, Mazak MegaStir Liquid-cooled Toolholder, that will be referred to as LCTH throughout, was integrated with the deposition machines. The LCTH is an integrated toolholder and cooling system, while MELD is a two-component system. With the new LCTH system, the tool is indirectly cooled by the tool holder through metal-on-metal contact between tool and holder.

The MELD and LCTH systems have several differences: flow style, cooling style, cooling area, and coolant flow rate. The MELD system uses a vacuum style pump that pulls the coolant at ~8 L per minute through the cooling loop to directly cool ~2000 mm² of tool area before returning to a reservoir. The reservoir is chilled by an external chiller. The LCTH system uses a pressure style pump to push the coolant through the 15,000 mm² area of the cooling loops within the tool holder. This coolant flows through the toolholder directly from the external chiller at ~12 L per minute. The tool is held in the LCTH with ~8000 mm² of chilled tool surface area contact. See

Table 1. Cooling system comparison.

Systems	Flow Style	Cooling Style	Cooling Area (mm2)	Flow Rate (L/min)
MELD	Vacuum (pull)	Direct	~2000	~8.0
LCTH	Pressure (push)	Indirect	~8000 *	~12.0
			~15,000 **

* LCTH to Tool. ** Coolant to LCTH.

for a summary of these cooling systems.

Process data was collected using the included data acquisition system (DAQ) on the L3 and CD-14 machines. Both DAQs were configured to utilize a K-Type thermocouple probe in the deposition tool with the tip exposed to the deposited aluminum and recorded by the DAQ wirelessly. The L3 DAQ collected temperature (°C) from two locations: at the tool face and at the bearing low in the spindle. The L3 machine utilized in the study could record at a maximum of one hertz. The CD-14 DAQ collected tool temperature (°C) at 20 hertz but bearing temperature was not available.

2.2. Experimentation and Evaluation

2.2.1. Spindle Bearing Temperature Baseline

To understand the impact of spindle rotation rate on spindle bearing temperatures, a simple experimental design was conducted to test the full range of rates used during depositions. Two tool cooling states were used to impact the temperature of the spindle bearing: cooled and uncooled spindle rotation. During the uncooled tests, no tool cooling was used while rotating the machine spindle at a constant rotation rate for 90 min. This was done for five rates: 100, 200, 300, 400, and 500 rpm. During the cooled tests, these tests were repeated for each rate; however, a 90 min period of pre-cooling was implemented before engaging the spindle, and coolant flow continued for the duration of the test. Starting conditions were preserved by allowing 24 h between tests to allow the system and equipment to completely restabilize thermally.

2.2.2. Cooling System Comparison

To compare the cooling systems’ effect on machine critical temperatures, specifically the temperature of the spindle bearing, a build was produced using three experimental conditions: no cooling, MELD cooling system, and LCTH cooling system. A path geometry was selected that maximized localized heat input. The planned path will result in a deposition that is ~150 × 150 mm² and ~200 mm tall using 100 layers of ~2 mm thickness. For each condition this path plan was attempted. These depositions were made on the MELD L3 in a discrete manner, where after each layer was deposited, the process was halted to allow for reloading of feed material and then resumed in an automated fashion. Each condition was tested layer by layer until the deposition failed to continue deposition or the deposition stack became too unstable to continue with the next layer, i.e., complete separation of the deposition stack from the substrate.

This experiment utilized tools with a square ~10 × 10 mm² through hole with no face features. Each tool was made of H13 and had internal features following the internal geometric consideration previously described in work done by Brigham Young University (BYU) [36]. Each tool had a 5-degree draft angle with a draft depth of 20 mm and a 45-degree chamfer angle with a chamfer depth of ~0.5 mm from the tool face. Draft and chamfer may be identified in Figure 2 where these features are pointed out. The featureless tool deposited material with a constant 350 rpm spindle speed and material was fed at 380 mm per minute while the tool translated at 650 mm per minute. All depositions were made without graphite contamination. Commercially available AA6061-T6511 aluminum extrusion was used to make ~525 × 9.5 × 9.5 mm³ long pieces of feed material. Substrates were produced from commercially available ~9.5 mm thick AA6061-T651 plates of ~300 × 300 mm². Substrates were cleaned with isopropyl alcohol to remove oils and sanded to remove thick oxide layer with 80 grit sandpaper. Both feed material and substrates were of nominal chemical composition, as shown in

Table 2. Chemical composition of AA6061 material for feed material and substrates. Adapted from Ref. [37].

Component, Wt. %	Al	Cr	Cu	Fe	Mg	Mn	Si	Ti	Zn
AA 6061-T6	Bal.	0.04–0.35	0.15–0.4	Max 0.7	0.8–1.2	Max 0.15	0.4–0.8	Max 0.15	Max 0.25

. Commercially available AA6061, of associated heat treatments, was selected due to the prevalence of this alloy being used in AFSD research and availability for procurement by the researchers.

2.2.3. Cooling Configuration Comparison

Additional three-dimensional depositions were produced to compare tool lengths and cooling systems in four configurations. This experiment utilized H13 tool bodies with a ~13 mm² through hole with internal features following the internal geometric consideration previously described [36]. Each tool has two sets of positive protruding teardrops on the face of the tool, alternating 90 degrees apart from each other with two protrusions per set. Each set of teardrops is arranged on two diameters of 25.4 mm and 31.8 mm; the protrusions have a depth of ~2.3 mm and ~1.5 mm, for the two diameters, respectively. These design features and details can be seen in Figure 2. These tool features were added to the tool to increase mixing between layers and between deposition and substrate, as shown by work done previously at BYU [38]. All depositions were made without the use of graphite lubricant on the feed material. Feed material was made of commercially available AA6061-T6511 aluminum extrusion cut to ~525 × 12.7 × 12.7 mm³ long and made of nominal chemical composition, as shown in Table 2. Tool face thermal data was collected with a K-type thermocouple with the probe tip exposed, for direct contact between deposition and thermocouple, approximately located, as shown in Figure 2.

These depositions are ~100 mm in diameter and ~160 mm in height in a helical path with individual layers being ~2 mm in thickness. These depositions were made on a MELD L3 in a discrete manner as described above; however, four layers were able to be deposited from each piece of feed material before needing to reload material. The helical path and multiple layers per piece of feed material were used to as closely approximate the condition and path needed for a stable continuous deposition. Depositions were performed on a substrate of an aluminum AA6061-T6 plate with dimensions of ~200 × ~200 × 9.5 mm³. The substrate was stacked on top of another identical plate used as a backing or anvil plate. Each configuration was tested using the same process parameters. Depositions were all done using a 210 rpm spindle rotation rate, ~340 mm/min tool traverse rate, and ~190 mm/min material feed rate. The 2 mm layer thickness and deposition parameters were utilized due to internal deposition development performed at BYU. This development optimized for repeatability of deposition for all comparisons in this work.

Depositions were made using four configurations of two cooling systems and two distances between the tool face and lowest portion of the cooling system, referred to as DC. Two extreme DCs were explored to maximize differences in cooling profile for each cooling system. Each cooling system can manipulate DC through different mechanisms. The MELD system can change the position of the cooling system along the tool length to modify DC. The LCTH system must have tools of differing length to change the DC. The extremes of the MELD system were limited by the physical constraints of the cooling system along the length of the tool. The LCTH system extremes were limited at the maximum by the tool length provided by Mazak-MegaStir and at the minimum by the length necessary to collect thermal data with a thermocouple.

The first two configurations used the MELD cooling system where coolant directly cools the tool and can be placed anywhere along the length of the tool, see purple representations of MELD system in Figure 3a,b. The MELD system has two extreme DCs possible: ~12 mm and ~36 mm, for ease of reference these will be referred to as MELD-S and MELD-L, respectively. The second two configurations use the LCTH with indirect cooling applied to the tool. The cooling system is fixed in position, so two different tool lengths were utilized to adjust the DC: ~12 mm and ~47 mm; for ease of reference these will be referred to as LCTH-S and LCTH-L, respectively. Figure 3 shows these four configurations with representation of the two cooling systems.

Table 3. Summary of experiment utilized in study identifying tool and cooling systems used for each test that deposited material.

Test	Cooling	Tool	Experiment	Feed System
1 *	N/A	Long—(Figure 3a)	Cooling System	Discrete
2	MELD	Long—(Figure 3a)	Cooling System	Discrete
3	LCTH	Long—(Figure 3c)	Cooling System	Discrete
4 †	MELD	Long—(Figure 3a)	Cooling Configuration	Discrete
5 †	MELD	Short—(Figure 3b)	Cooling Configuration	Discrete
6 †	LCTH	Long—(Figure 3c)	Cooling Configuration	Discrete
7 †	LCTH	Short—(Figure 3d)	Cooling Configuration	Discrete
8 **	MELD	Short—(Figure 3c)	Discrete vs. Continuous	Continuous
9	LCTH	Short—(Figure 3d)	Discrete vs. Continuous	Continuous

* configuration MELD-L was used, but the water jacket was removed to utilize no cooling. ^† Feed material from these depositions were used for microhardness testing for Thermal Effect experiment. ** Configuration MELD-S was used, but the tool body was of Cu–Be not H13 tool steel.

describes the details of the deposition: cooling system, tool, and feed system used in each experiment of this study.

2.2.4. Thermal Effects on Feed Material

Microhardness line traces were performed on the used feed material from the cooling configuration builds. After each stop, the remaining feed material is ejected from the tool and collected. The line traces were done on four ejected feed rods for each cooling configuration test: the first, second, 10th, and 21st bars. The deformed end of the feed material was milled to return it back to prismatic dimension before sanding and polishing to a mirror finish using: 240, 400, 600, 800, 1200, and 1200 fine grit sandpapers then 1 micron alumina suspension polish on a LECO PX300 Manual Polisher (St Joseph, MI, USA). Microhardness line traces were done on a Clark Instrument CM-402-AT (Novi, MI, USA) with 500 g mass at 10 s dwell with 0.5 mm spacing along a 50 mm path resulting in 100 Vickers hardness data points per trace. The line traces start with 1st indent being 0.25 mm from the deformed end of the feed material and 100th indent is towards the undeformed end of the feed material. For each configuration, the hardness data collected is smoothed using a seven-point window moving average for each trace, then averaged for each bar for one resulting profile per configuration.

2.2.5. Continuous vs. Discrete

The same ~100 mm diameter helical depositions were also completed on a CD-14 AFSD machine as they were made on the discrete L3 AFSD machine. This CD-14 machine has continuous deposition capability and as such does not have reload breaks. Two of the cooling configurations were duplicated on this machine with continuous deposition: both the short DC conditions from the LCTH and MELD systems. These configurations were repeated based on initial deposition results. The LCTH tool was an identical H13 tool used previously. The tool for MELD system on the CD-14 uses Cu–Be alloy as body material with H13 as face material rather than using H13 for the entire tool. This Cu–Be tool is the manufacturer provided solution for graphite or contaminant free deposition of aluminum alloys. The same tool face features were present as on the tooling used for the discrete configuration testing. The Cu–Be tool, however, did not have any internal geometric changes (i.e., draft or chamfer) but was a straight square through-hole tools. Table 3 describes the details of the deposition: cooling system, tool, and feed system used in each experiment of this study.

3. Results and Discussion

3.1. Baseline Bearing Temperatures

As the spindle operates, heat is generated through friction within the bearing assembly. This frictional heat could increase the temperature of the bearings if not managed. To understand the cooling systems’ ability to protect the critical spindle and spindle bearings, a baseline must be established for bearing temperatures with only spindle rotation and no external thermal loads. Regardless of the cooling system utilized, an increase in spindle rotation rate increases the bearing temperatures, as can be seen in Figure 4, by comparing each color of curves. Each spindle rotation rate increase of 100 rpm from 100 to 500 rpm increased the bearing temperature by approximately 15, 28, 9, and 23 °C, respectively. This shows that for only a 90 min spindle rotation condition, increasing rotation rate by 100 rpm increases the spindle bearing temperature by ~20 °C. This follows the expectation that additional frictional heating would occur with an increase in rotational velocity within the bearings.

With the impact of spindle rotation rate established, cooling systems can now be considered for the no load spindle bearing temperature. For each rotation rate tested between 100 and 500 rpm, the cooling systems decreased the bearing temperature by approximately 4, 10, 0, 15, and 13 °C, respectively, compared to the no cooling condition. For each rotation rate, utilization of cooling decreased the spindle bearing temperature by ~9 °C on average. The utilization of cooling decreases the temperature of spindle bearings for a given rotation rate when compared to no cooling. This can be seen in Figure 4 by comparing curves with filled and hollow markers of the same color. The impact of cooling versus no cooling is less impactful than the rotation rate used. This trend is logical due to relatively low contact area between the spindle and the cooling system.

Spindle bearing temperature is an important factor for the life of AFSD equipment. If spindle life is decreased due to high bearing wear, related to high bearing temperatures, this can be detrimental to the life of the component. The MELD user manual states that, “The maximum operating temperature for the bearings is [99 °C]” [39]. With no thermal load, spindle rotation rates of 500 rpm exceed the spindle bearing limit, this can be seen by both types of inverted triangle in Figure 4. It also seems likely that 400 rpm with no cooling will also surpass the bearing temperature limit if the spindle was run for a longer duration. This suggests that spindle rotation rate selection should be considered when developing parameters to enable bulk deposition of aluminum using MELD AFSD equipment. This would typically not be a problem because deposition spindle rotation rates are significantly lower for aluminum depositions but should be considered for equipment protection.

3.2. Cooling System Comparison

To test the cooling systems, three large depositions were attempted. For the control, a no cooling deposition was attempted and had very limited ability to deposit. During this deposition, two complete layers were deposited. These layers were of normal quality and process forces; this can be seen in Figure 5a. Shortly after starting the third layer, the print failed to deposit material, due to swaging or material deformation inside the tool. Swaging of feed material reduces the force being applied to the deposition because of frictional drag inside the tool. The no cooling attempt was used as a control to confirm the necessity of cooling for aluminum deposition; it seems likely that cooling would be a necessity for any AFSD at meaningful scale.

In contrast to the control, both cooling systems performed very well. These depositions were completed with normal forces and quality. Utilizing the MELD cooling system, the full planned build height was completed. However, while using LCTH cooling system, the deposition was unable to complete the full height, 100 layers, due to first layer adhesion issues. These depositions were done with flat or featureless tooling, which has poor interlayer and especially layer-to-substrate interaction. Additionally, any increase in cooling could cause the material to deform less readily with the substrate material. Future work using knub tools has eliminated this concern [38]. When cooling was utilized, either system produced meaningful depositions of large scale.

An analysis of the first 50 layers of each build, after approximately four hours of deposition, shows both cooling systems appear to maintain the spindle bearing temperatures equally effectively. Despite the difference in the cooling system, as detailed in Table 1, the spindle bearings’ temperatures were relatively equal for both systems at the same stage in the builds. By layer 50 both spindle bearing sets had heated to ~70 °C regardless of system, see Figure 6. This figure shows average bearing temperatures for each layer of each deposition. Despite differing slightly in temperature leading up to the end of the test, the final bearing temperatures of these two builds are the same in the end. This suggests that each cooling system protects the spindle bearings from elevated temperatures to the same degree.

There appears to be low thermal contribution from the process to the spindle bearings, especially when cooling is utilized. When compared to the baseline spindle temperature profiles from the no thermal load study previously discussed, continuous spindle rotation related friction would result in bearing temperature of ~80 °C, see Figure 6. Each deposition build, by layer 50, had only generated enough heat for the spindle to be at ~70 °C, as shown in Figure 6, which is expected because these depositions do not have continuous spindle activity. These depositions each have about one minute between layers with the spindle stopped. This is due to the discrete nature and needing to reload the feed material after each layer.

3.3. Cooling Configuration

The helical depositions exploring the impacts of various configurations of DC and cooling system found that each configuration seemed to have similar deposition quality. Each deposition has generally clean edges and similar surface texture both on the top surface and the side of the depositions. All helical depositions were completed in each configuration without any deviation from planned height; this can be seen in Figure 7. This demonstrates that each configuration is capable of being used for deposition of AA6061 and can produce similar surface quality.

Despite each configuration being capable of depositing AA6061 there are differences in thermal impact to the process and feed material. As shown in Figure 8, each configuration has a unique tool temperature during deposition. The MELD-S appears to have most consistent deposition temperature of ~450 °C for the entire deposition. The MELD-L had the highest process temperatures during the second half of the deposition, 2500 s to end. LCTH-S had the lowest process temperatures for the second half of the deposition. The LCTH-L has very similar temperature profile to the MELD-L, except it remained slightly cooler in the last half of the deposition. There is a notable difference in the temperature achieved for short and long DC, being approximately 40 K hotter for long DCs. This indicates that the DC is an important process parameter that needs to be controlled due to its relationship with the achievable process temperatures. Additionally, the DC plays a role in enabling the process by preventing feed material from deforming and swaging into the tool, which can interrupt and prevent the deposition process. Therefore, DC is a parameter that requires judicious manipulation to control process temperature while enabling processing.

Although the effective length of the tool or the DC has the largest impact on process temperature, the cooling system is also an unneglectable factor to the thermal system. Utilizing the LCTH compared to MELD for long DC and short DC tools, the deposition temperature was reduced by ~2.5 K and ~15.0 K, respectively. The length of the DC magnifies the impact of the cooling system because the LCTH is more effective, as shown by the steeper gradient, with the short DC tools than with the long DC tools. LCTH has an increased cooling ability compared to the MELD cooling system, but both are viable cooling system for deposition of aluminum AA6061. Also, the MELD-S appears to have the most consistent deposition temperature of ~450 °C for the entire deposition. Consistency is highly desirable for the AFSD process, which the MELD-S was able to achieve successfully in the discrete system. It is clear that the cooling system and the length of DC both impact the process temperature.

3.4. Feed Material Evaluation

The tool DC impacts the ability of the cooling system to prevent feed material strength degradation. AA6061 is a precipitate strengthened aluminum alloy, meaning that with increased thermal input, the material becomes weaker due to precipitate coarsening. This coarsening decreases the hardness of the material. If temperatures are additionally elevated, then the strengthening precipitates can be put back into solution and precipitate during natural aging process. This phenomenon can be seen in the ‘check mark’ shape of each of the hardness plots in Figure 9, where the material to the left of the check mark (at the minima found at ~2–4 mm) has increased strength likely due to natural aging while material to the right has likely experienced over aging. The thermal impact on the feed material has degraded the strength. With room temperature natural aging, i.e., storage, the material to the right was not able to have the strength recovered but material to the left did. The material to the left of the check mark was able to regain strength after being left at room temperature, likely due to being at a higher temperature during deposition putting strengthening mechanism into solution that were able to be precipitated at room temperature.

The extent of this over aging negatively impacts the strength of material entering process or the feed material hardness. This degradation can cause premature deformation, a deformation inside the tool, that can prevent deposition. In Figure 9 shorter tools had decreased thermal effect on the feed material, i.e., minimized heat affected zone (HAZ) depth. These long tools have HAZ that are 10 mm deeper than those seen in the short tools, ~28 mm rather than ~18 mm for long HAZ and short HAZ. It is also important to note that the MELD-L tool has ~10 mm shorter DC than the LCTH-L because of the limitations of the tooling, see Figure 3. The MELD-L has a DC that is shorter than that of the LCTH-L; despite this it thermally softened the feed material equally to that of the longer LCTH-L. This equivalent thermal softening but short length is evidence that the MELD cooling system has a reduced cooling impact or ability.

Each of the tools explored in the study have ~20 mm of draft depth, which appears to affect the cooling systems’ efficacy. Short DC tools were able to maintain feed material strengths within in the draft where contact between tool and cooling would not be possible geometrically. In configurations with a short DC the strength was unaffected ~2 mm into the draft from the feed direction. In contrast, configurations with long DC could not maintain these strengths even inside the undrafted feed section of the tool. Material degradation occurred ~10 mm before the draft, up the tool in undrafted sections, from the feed direction in long DC tools. These differences show that short DC tools have improved control of the process when compared to long DC tools.

Additional to the materials’ precipitate strength, the yield strength is negatively correlated with temperature. In the temperature range of 25 to 500 °C seen during AFSD processing, the yield strength drops from ~310 MPa to ~10 MPa for AA6061, as seen in Figure 10. This significant decrease in feed material strength can be cause for concern for material swaging inside the tool, which prevents deposition of material. If feed material temperature gets high enough that the material yields in the undrafted region of the tool, the process could be halted due to excessive force. The inclusion of the draft in these tools prevents the feed material from swaging enough to inhibit deposition for all tool configurations tested within this study. For cooling configurations where the HAZ extends above the drafted region of the tool, swaging may be a concern if using different materials or parameters.

3.5. Continuous vs. Discrete

Continuous AFSD does not have any feed-material-reload-related additional cooling effects. Each continuous configuration deposited completely and of equal quality to each other and the related discrete deposition. These continuous AFSDs, despite having the same number of layers in the build, took less time to deposit than the discrete counterpart, as shown in Figure 11. Without the interlayer cooling effect, the continuous MELD-S, designated as “MELD-S_Cont”, cooling configuration deposition experiences a slow temperature climb from 490 to 510 °C during the continuous helical deposition, as shown in Figure 11. In contrast, the continuous LCTH-S, designated as “LCTH-S_Cont”, has a very stable temperature of 480–500 °C, this can also be seen in Figure 11. The increasing process temperature over time is very different than what is observed in the discrete system where peak temperature of each reload is constant or potentially slightly decreasing over time. This is likely due to thermal saturation of the boundary condition overtime.

Both continuous configurations were able to deposit to completion. Regardless of the cooling system used, all depositions were completed as planned. This completion of these test parts indicates that each cooling system is viable for depositions of AA6061 at scale. There is an apparent temperature difference between the two continuous depositions, ~20 K. Additionally, there is ~50 K increase in deposition temperature when completed through a continuous deposition method rather than a discrete method, these trends can be seen in Figure 11. Large AFSD deposited using a discrete method obviously do not have the same thermal history (cycles and trends) or processing temperature despite equivalent processing parameters. These differences should be considered when comparing discrete depositions to continuous depositions.

The impact of the tool cooling system becomes more apparent when the interlayer cooling time is not convoluting the observation. Regardless of deposition style (continuous or discrete), the LCTH is more effective at cooling the deposition process as shown in Figure 11. It is also important to draw attention to the difference in tool material for the MELD-S continuous tool, it is a Cu–Be alloy for the tool material which has much higher thermal conductivity than that of tool steel. As shown in Figure 11, despite this change in thermal conductivity the LCTH with low conductivity tools was able to cool more significantly than the MELD cooling system with high thermal conductivity tooling. The improved efficacy allows for more stable process temperatures and potentially greater control of deposition temperature if paired with a temperature control method.

4. Conclusions

In this study, a new AFSD cooling system is discussed and compared with the original manufacturer provided cooling system. This work shows the impact of cooling on deposition temperatures, feed material and essential machine components. From the data presented and analyzed the following conclusions are drawn:

•

Cooling is essential to the AFSD process for aluminum deposition. Without sufficient cooling, swaging will occur which can prevent the deposition of material. Deposition surface quality does not seem to be impacted by cooling configuration as long as a cooling system is used.

•

Rotational rates above 400 rpm will negatively impact spindle bearing life. Without thermal load from deposition, spindle bearing temperatures exceed the operating limit of 99 °C.

•

Either cooling system enables higher spindle rotation rates because both cooling systems lower spindle bearing temperatures when rotation rate is held constant. Spindle rotation rates must carefully be considered when developing parameters to enable prolonged deposition.

•

Both the distance to cooling and the cooling system influence process temperature during deposition. Cooling system provides ~15 K difference in temperature and distance provides ~40 K difference in temperature.

•

Short tools used in either cooling system (MELD or LCTH) demonstrate smaller HAZ in the feed material. During larger and continuous depositions this can prevent swaging of material in the tool.

•

All configurations of tool length and cooling system were found to successfully print AA6061 using the AFSD process during bulk deposition. Thus, both cooling systems are sufficient for deposition of AA6061 aluminum alloys (~500 °C processing). From an industrial perspective, aiming for large-scale continuous production, the most robust (i.e., thermal stability and minimization of thermally degraded feed material) configuration is LCTH cooling system with a short DC.

This work has shown the importance of cooling on both the process, feed material, and equipment used for AFSD. Though in this work all depositions are large relative to those done in the literature, these are still small relative to depositions needed by industry. While AFSD builds size will continue to increase during the transition to industrial scale, the necessity of cooling will become more prevalent as time at elevated temperature will be increased. The prolonged time at elevated temperature will test the reliability of various cooling systems. Additionally increasing the build size changes the boundary condition of the deposition and may impact the efficacy of different configurations. Also, work done in higher strength aluminum alloys to deposit, namely AA7xxx, cooling may become more important for large-scale success. Cooling will become especially important when working with high melting temperature materials (titanium, steel, and stainless–steel alloys). The depositions temperatures are and will be notable higher temperature. This increased temperature is taxing on the deposition equipment if cooling systems are insufficient. The ability to cool while depositing such high temperature materials will be important for both tool life and equipment life. The robust cooling system provided by the LCTH will allow for this cooling at high temperatures.