Additive Manufacturing of a High Temperature, Ni-Based Superalloy Compact Heat Exchanger: A Study on the Role of Select Key Printing Parameters

Additive manufacturing (AM) is an advanced fabrication technique in which parts are fabricated layer by layer from a preprogrammed digital model. In the last decade, AM has experienced tremendous growth, as it allows the fabrication of complex objects which are too challenging to be fabricated using conventional fabrication methods. In addition, when properly utilized, AM also generates less waste material than subtractive processes. Also, AM allows the fabrication of multiple parts into a single component, which significantly simplifies the assembly process and reduces the lead time [1].

One of the areas that can benefit from AM is the fabrication of high-performance heat exchangers (HXs) made out of high-temperature superalloys such as Ni-based alloys. HXs have a wide array of applications, such as aerospace; power production; computer cooling; heating, ventilation, and air conditioning; and ground and sea-based transportation. Currently, there is high demand for compact, high-temperature HXs capable of delivering both high volumetric (kW/m³) and mass-based (kW/kg) heat removal density [2]. Such HXs are especially needed for high-efficiency power generation cycles, in which the cycle efficiency is proportional to the maximum operating temperature at the turbine inlet, and for aerospace applications to cool down hot bleed air for the environmental control system of the cabin. However, high-temperature resistance and compact design requirements limit the number of suitable materials and fabrication technologies. As shown by Zhang et al. [2] material selection and fabrication procedures are the main challenges for the development of high-temperature heat exchangers.

Several studies in the literature have successfully used AM to fabricate complex HXs. Tsopanos et al. [3] fabricated a microcross flow HX with rectangular channels in both cold and hot sides with 316 L stainless steel, using selective laser melting (SLM). Their HX achieved an overall heat transfer coefficient of 2.22 kW/m²K and a volumetric heat transfer density of 3.14 W/m³K. Arie et al. [4] reported the fabrication of manifold-microchannel HXs (MMHXs) out of stainless steel, titanium, and aluminum alloys using selective laser sintering (SLS), with an overall size of 25 × 44 × 62 mm³ and fin thickness of 0.23 mm. Based on the successful results of that work, a bigger unit (32 × 150 × 150 mm³) was printed out of a titanium alloy with the same fin thickness [5]. They subsequently reported enhanced heat transfer coefficient of up to 190% for the same pressure drop when compared to the state-of-the-art HXs [6]. Gerstler and Erno [7] successfully fabricated multifurcating HXs using SLM with aluminum, titanium alloy, cobalt chrome, and Inconel 718. The HXs met the expected pressure drop and heat transfer requirements, with a 66% reduction in mass and 50% reduction in volume. Da Silva et al. [8] used the SLM process to fabricate a cross-flow HX with circular channels of 316 L stainless steel for use as a replacement for printed circuit heat exchangers. The HX was tested with air and water as working fluids resulting in a heat removal density smaller than 1 kW/kg. A detailed literature survey on the use of AM to fabricate HXs can be found in the review papers [9–11].

This study concentrates on metallic HXs fabricated using AM, with a specific focus on high-temperature materials such as Ni-based superalloys. Superalloys are typically challenging to process using conventional fabrication methods. The high strength, toughness, and heat resistance of superalloys render them difficult to machine (shape and cut), form, and deform [12]. Moreover, joining superalloys using brazing and welding is also challenging [13,14]. But these challenges can largely be avoided with AM due to its additive nature [15]. Zhang et al. [16–18] successfully fabricated an MMHX out of Inconel 718 using SLS. The HX had an overall size of 66 × 27 × 74 mm³ with a fin thickness of 0.18 mm on both sides (air-to-air application). The HX was tested at over 600 °C, and the resulting heat removal density was in the order of 10 kW/kg, which is a 25% improvement over conventional plate-fin HXs.

However, the HX fabricated by Zhang et al. [18] had rather low effectiveness (50–60%) and coefficient of performance, COP equal to 62.5. Moreover, due to the printing orientation of the unit, one side of the HX had straight fins, while the other had inclined fins, which yields a lower performance than straight fins in MMHXs. This is a problem because both higher effectiveness and COP are often required for aerospace and power production applications. The objective of this work is, therefore, to use the direct metal laser sintering (DMLS) technique to fabricate high-temperature MMHXs for aerospace applications, with the objective of achieving higher effectiveness and COP values for the HX without penalizing the mass-based (kW/kg) heat removal density. To achieve this goal, the MMHXs used for this work were fabricated for the first time with straight fins and straight manifolds on both sides, which allowed higher mass-based (kW/kg) heat removal density of the HX for the same values of effectiveness and COP. This enhancement was achieved by introducing a new printing orientation.

To achieve high power density for the HX, one of the most important parameters in an MMHX is a small enough fin thickness, as previously shown by Arie et al. [19]. Additionally, the minimum base thickness, which separates the hot and cold streams, plays an important role. The base thickness needs to withstand the designed pressure; however, the thinner the base, the more reduction in mass and higher power density for the heat exchanger. Therefore, base thickness becomes a crucial optimization problem. To minimize fin thickness, multiple MMHX coupons were fabricated with different fin thicknesses to determine the smallest fins that could be successfully fabricated. A similar approach was applied for the base thickness with a series of pressure test coupons. The size of the sample demonstrator HXs was increased from that of Zhang et al.'s MMHX [16–18], using a larger build printer, but the fin thickness was kept the same. The use of larger printers usually results in a loss of resolution compared to smaller printers due to the laser travel distance. To assure good resolution in the larger printer, we optimized the printing parameters, rather than using the default ones. This allowed us to print a 76 $×$ 76 $×$ 76 mm³ and a 94 × 87.6 × 94.4 mm³ MMHX with straight fins and tapered manifolds.

Manifold-microchannel HXs are a favorable alternative to state-of-the-art HX technologies because they deliver high heat transfer performance [20–22] (since flow can be typically in the entry region [23–25]) while minimizing the pressure drop. MMHXs utilize a system of manifolds to convert an array of long microchannels into a system of short parallel microchannels. As shown in Fig. 1, the manifold structure and its flow distribution channels are placed on top of the microchannels and oriented perpendicularly to them. The flow first enters through the manifold channels, then it is distributed into the parallel microchannels on the heat transfer surface, where it travels a short length in each microchannel before it is guided out. As discussed by Cetegen [26], the main advantage of the manifold-microchannel concept is the reduced pressure drop due to the simultaneous reduction in both flow length and flowrate. Although multiple works in the literature have reported the superior heat transfer enhancement and reduced pressure drops associated with MMHXs compared to conventional HXs [16–18,20], their fabrication from high-temperature superalloy materials has presented challenges that limit their widespread use in HX design options.

The conventional fabrication process of an MMHX involves multiple steps, since the microchannel surfaces and manifolds have to be fabricated separately. When the MMHX has a cross-flow configuration as shown in Fig. 1, either photochemical etching or laser micromachining is needed to fabricate the double-sided cross-flow microchannel heat transfer surface. Photochemical etching is a fabrication process that utilizes a photoresist and etchants to remove a certain area of a metal plate. Due to its ability to etch various metals, the process is usually used to fabricate microchannel surfaces for printed circuit heat exchangers [2,27–29]. The etching process starts with a photoresist layer deposited on the metal surface. Then, the metal material with the photoresist layer is exposed to UV light through a phototool. After the photoresist layer is developed, the exposed metal is dissolved via an etching process to form semicircular channels with typical channel widths of 0.5–2 mm [30]. There are two disadvantages to this process. First, compared to rectangular channels, semicircular channels are less compact and heavier due to their thicker fins, thus adversely affecting the HX heat removal density. Second, significant postprocessing would be needed for the etched double-side microchannel structure to meet the required flatness for assembly. For these reasons, this technology may not be appropriate for MMHXs.

Laser micromachining is another approach for fabricating a double-sided cross-flow microchannel surface. By applying a high-power laser beam, the energy of the laser is converted to heat, which vaporizes or melts the metal to form the microchannel structure [31]. However, laser micromachining is a very time-consuming and expensive process, as it may take up to two days to fabricate one 389 × 132 × 1.3 mm³ microchannel surface [32], rendering it unfeasible for a full-scale HX.

On the other hand, computer numerical control (CNC) machining, sheet metal forming, and injection molding can be used to fabricate the manifold. Considering the fabrication time and cost, CNC machining is a good approach for prototyping, while sheet metal forming and injection molding can be used for mass production. Since an MMHX generally consists of many layers of manifolds and microchannel surfaces as shown in Fig. 1, brazing, welding, or diffusion bonding are needed to assemble the parts. However, due to the reduced size of the channels, improper brazing or welding between the manifold and microchannels can cause clogging in the channel side, thus reducing heat transfer performance and significantly increasing pressure drop. If diffusion bonding is used, considerable effort is required to postprocess the microchannel surfaces and manifolds to meet the parallel and flatness requirements for the bonding surfaces.

To address these challenges posed by conventional fabrication methods, in this study the DMLS technique was used to fabricate the MMHXs of this study. Using the AM approach, an MMHX can be 3D printed as a single unit without requiring any assembly process, which can significantly reduce the fabrication lead time and cost. An DMLS machine usually consists of two platforms, a powder dispenser platform, which houses the metal powder, and a build platform on which the MMHX structure is built.

Fabrication of an MMHX using AM on the orientation shown in Fig. 1 will cause an overhang, as shown in Fig. 2. Overhang occurs when the next layer in the printing direction is larger than the previous layer. The overhang is tolerable in structures with an inclination up to an angle of 45 deg with respect to the printing direction, while for angles greater than 45 deg a supporting structure is needed to avoid part failure [33]. Since it is difficult to remove the support structure from the interior structure of an MMHX due to its structural complexity, all of the structures need to be fabricated with an inclination smaller than 45 deg. Zhang et al. [16–18] have successfully fabricated an MMHX in the orientation shown in Fig. 3(a) using inclined fins and manifolds on one side of the HX. Both inclined fins and inclined manifold walls have a self-supporting angle of 45 deg, avoiding the problem of the overhang. However, despite the printability of this HX, this configuration affects the flow in the HX. In the flow scheme of Fig. 1, the fluid enters and leaves the microchannels through 90 deg angle turns, while with the design in Fig. 3(a) the fluid enters at an angle of 45 deg and leaves at an angle of 135 deg, which is fluid dynamically less efficient than 90 deg. A study of the effect on the HX performance by Arie [34] found that the inclined-fin HXs had a similar heat duty as the straight channels but higher pressure drops.

This work demonstrates that, instead of modifying the structures within the HX, the entire unit can be oriented at a 45 deg angle with respect to the build plate (plane x–y). Figure 3(b) shows the two sides of the MMHX structure. In this orientation, the unsupported overhang inside of the HX can be avoided by adding supporting structures outside as shown in Fig. 3(b), which eliminates the need for inclined fins and manifolds. The first advantage of printing in this orientation is that it allows for the fabrication of straight fins and manifold channels, thus improving performance. As mentioned previously, when the fluid flows in the microchannel following the ideal MMHX configuration of 90 deg turns (as shown in the top stack of Fig. 1), the pressure drop in the microchannels is smaller than that for the inclined fins. Moreover, the 45 deg inclined printing orientation allows higher fin density to be printed compared to the printing orientation and inclined fin design adopted in the solution of Fig. 3(a). To better explain this statement, Fig. 4 provides a visualization of the respective printing process for the configurations shown in Figs. 3(a) and 3(b). The figure shows the sequences of layers used to create the desired designs: the hatched area represents the portion of the layer in which the powder has been sintered, while the nonsintered one is left blank.

Thus, for the same fin thickness (⁠$tfin)$⁠, the space occupied at the base with straight fins obtained by changing the inclination of the print to a 45 deg orientation (⁠$tfin−base,straight$⁠) is 30% less than the vertical print with the inclined fins (⁠$tfin−base,inclined$⁠). This demonstrates the possibility of printing higher fin density surfaces with the orientation shown in Fig. 3(b). This analysis is valid for both sides of the MMHX, thus allowing the HX to have the same fin thickness (⁠$tfin)$ on both sides.

To fabricate the MMHX using DMLS, one of the most important factors is the minimum fin thickness that can be fabricated through the 3D printing process, since a higher fin density can increase the mass-based heat removal density (kW/kg) of the HX. The 3D-printed minimum feature size varies from machine to machine, depending on the size of both the laser spot and the metal powder diameter. Currently, the minimum feature size reported appears to vary from 0.15 mm to 0.38 mm [35]. This study is focused on optimizing the printing orientation, laser power, and material consideration, while the other parameters were kept constant. Other conditions such as speed and scanning strategy were setup using the default setting of the 3D printer. As the idea is, by choosing the optimized printing orientation and laser power for selected material, most metal 3D printers can achieve similar fabrication results through the default setting. Additional printing conditions, such as AM laser speed, and AM laser jump speed, among others, are shown as follows in Table 1.

Arie et al. [19] demonstrated that the performance of an MMHX design with a printable minimum fin thickness of 0.15 mm can surpass that of most conventional designs: for example, compared to a wavy fin design, they achieved up to nearly 60% increase in gravimetric heat transfer density (kW/kg K). In order to evaluate the capability of AM to fabricate MMHXs with small fin thickness, multiple coupons with different fin thicknesses and channel sizes were fabricated using different 3D printer machines. These coupons were fabricated in the same orientation as the orientation of Fig. 2(b), to be representative of the final HX. The coupon consisted of fins with three different nominal design sizes: 0.100 mm, 0.125 mm, and 0.150 mm. The fin spacing was set at 0.260 mm for all three cases. Multiple coupons were fabricated using a ProX DMP 200 (SC) [36], a ProX DMP 300 [37], and an EOSINT M 290 3D printer [38]. Table 2 lists the material used for each printer, the maximum printer build size, and the layer thickness. The layer thickness is a measure of the layer height of each successive addition of material in the 3D printing process in which the layers are stacked. Layer height is essentially the vertical resolution in the z-axis (as shown in Fig. 4). Figure 5 shows the measurements of the fabricated fin thicknesses compared to the design fin thickness used to create the computer-aided design (CAD) file for the different 3D printers.

Figure 5 shows that the ProX 200 machine had the closest agreement between the design fin thickness and actual fin thickness for the maraging steel. This result can be explained by the information reported in Table 2. First, the printer has a smaller build volume, meaning the distance between the laser and the build plate is smaller than that of the other printers. This allows a higher resolution in the printing feature. Second, the layer thickness is smaller than the conventional 40 μm. Third, the ProX 200 was the only in-house printer, enabling the parameters to be customized to enhance the feature resolution. Despite these promising results, maraging steel is not suitable for high-temperature applications which require superalloy materials. Maraging steel has a maximum service temperature of 400 °C, while Inconel 625 and 718 can withstand up to 900 °C [39]. On the other hand, given the need for a bigger printer to accommodate bigger HXs (since the build orientation negatively affects the build size requirement), other machines and different materials were tested. The ProX 300 machine was investigated in two different configurations. The ProX 300 (Inc625) showed a small deviation from the fin design values for 0.100 and 0.150 mm, but a significant deviation for 0.125 mm. This lack of consistency in the printer could become an issue when printing bigger parts. On the other hand, the ProX 300 (Inc718) showed a more consistent trend, but the deviation from the desired values must be carefully considered during the design stage: since the printed part will be bigger than the CAD file, the features must be designed to be smaller to mitigate the deviation. A similar trend was also shown by the EOS M290 (40 μm).

As shown in Fig. 5, which compares EOS M290 (40 μm) with EOS M290 (20 μm), the fin thickness can be significantly reduced by reducing the layer thickness. Changing the layer thickness of Inconel 625 from 40 to 20 μm allows features to be fabricated with a much smaller deviation. Both powder sizes show consistency in this trend. The fins shown in Fig. 6 were 3D printed by the EOS M290 (20 μm) and analyzed using a Keyence 3D digital microscope.

Based on the measurements reported in Fig. 5, the deviation of the actual fin thickness from the design value can be quantified as 13% for the design values of 0.150 mm, 15% for 0.125 mm, and 35% for 0.100 mm. The drawbacks of reducing the layer thickness are longer printing time and higher cost, mainly due to the longer printing time. Smaller powder size should therefore be selected only in those cases where fin thickness reduction and consistency are needed.

Printing parameters can be adjusted by machine operators and can significantly affect the quality of the printed parts [40]. Laser power intensity strongly affects the fabrication quality of the parts printed using DMLS. We investigated the effect of laser power on the fins' thickness by printing multiple coupons using maraging steel and the ProX 200, with three different fin sizes and two different laser power settings. Table 3 shows the design fin thicknesses and fabricated fin thicknesses fabricated using the 300 and 225 W laser power settings on the ProX 200 machine.

From Table 3, it can be concluded that smaller fin size can be fabricated using a lower laser power intensity. This is because the higher the laser power, the larger the melting pool, and the more difficult it is to fabricate fine features. However, reducing laser power can increase the porosity of the part and lead to leakage when tested under high design pressures. While porosity may not be a problem for the fins, it can negatively affect the thickness of the separator plates.

To evaluate the surface finish (channel roughness) of the selected printers, the pressure test coupons were analyzed using the Keyence 3D digital microscope available at the University of Maryland Additive Manufacturing Center.

Figure 7 shows the surface finishing for two coupons printed with ProX 300 (Inc718) and EOS M290 (20 μm). As one can note, from Fig. 1(b) (left), the surface is smooth, but irregular closer to the edges. On the other hand, the EOS M290 (20 μm) coupon, is more uniform but localized spots with variable heights are visible throughout the surface, thus increasing the roughness. The surface finish was evaluated in terms of the root-mean-square height (⁠$Sq$⁠), arithmetic mean peak curve (⁠$Spc$⁠), and developed interfacial area ratio (⁠$Sdr$⁠). The results of the roughness measurement are shown as follows and are in concordance with other reported values found in the literature [11,23] (Table 4).

Wall roughness in additive manufactured parts can play an important factor in the heat exchanger performance, i.e., pressure drop and heat transfer. However, in the case of manifold microchannels, the flow is mainly laminar and therefore the effect of roughness on pressure drop will be small [23]. In addition, depending on the level of roughness an early laminar to turbulent can be observed [11].

In addition to the minimum printable fin thickness, another important design limitation is the minimum base or separator plate thickness (the distance between the microchannel base which separates the hot and cold flow streams as shown in Fig. 1) that is required to withstand the HX design operating pressures on the hot and cold sides. Ideally, the smaller the base thickness, the higher the HX performance, both in terms of thermal performance and mass reduction, and the higher the heat removal density (kW/kg). Reducing the base thickness reduces the thermal resistance due to the conductive wall resistance, while also reducing the total mass of the HX. However, the minimum base thickness requires specific study considering the porosity of thin-wall structures resulting from the sintering process of SLS. As reported by Zhang et al. [2], an earlier version MMHX fabricated for aerospace applications using SLS and Inconel 718 had a microchannel base thickness of 500 μm, and was leak-free under a system pressure of 450 kPa. To further investigate the minimum base thickness effect, several coupon designs were printed. As shown in Fig. 8, each coupon consisted of a chamber whose testing walls were parallel to the x–z plane.

The printers selected for this study were those that gave the most consistent results during the fin thickness study: the ProX 300 (Inc718) and EOS M290 (20 μm). Each machine printed three coupons with tested wall thicknesses of, respectively, 300, 350, and 400 μm, as shown in Fig. 9.

For the range of applications considered in this study, high-temperature HXs generally operate under 340 kPa pressure [41]. For this reason, the six coupons were pressurized under the design condition, using nitrogen as a test medium. The test results are shown in Table 5.

As reported in Table 5, EOS M290 (20 μm) proved capable of printing three pressure-tight coupons, demonstrating dense structures for the three selected base thicknesses. ProX 300 (Inc718) failed in the smallest base thickness selected. As expected, moving from 40- to 20-μm layer thickness resulted in more leak-proof coupons. The layer thickness required to obtain leak-proof HXs varies based on the pressure selected. For this study, base thickness of at least 350–400 μm is needed if the HX is printed using ProX 300 (Inc718).

A 76 × 76 × 76 mm³ MMHX was fabricated to confirm that the results of the coupon study could be scaled up. Based on the knowledge gained from the fin thickness study, the fins were designed at 0.1 mm so that a fin thickness of 0.11 mm could be fabricated. As no specific test was performed on the pressure containment of maraging steel, a conservative base thickness of 0.500 mm was selected. The support structure was the same as the one designed in Fig. 2(b).

Figure 10(a) shows that besides the previously mentioned 45 deg angle in the y–z plane (identified as α in Fig. 10(a)), another 10 deg angle was given to the unit in the x–y plane (identified as β in Fig. 10(a)). The reason for this was to prevent the recoater from hitting the entire size of the HX perpendicularly to the recoater direction, as shown in Fig. 10(b). Introducing this angle β, the subsequent recoater fronts in the recoater moving direction impact progressively larger portions of the unit, thus reducing the stress of the impact and increasing the chance of a successful print.

Figure 11 shows the MMHX fabricated out of maraging steel using the ProX 200 at the University of Maryland's College Park AM Center. The HX fabrication can be considered an important success for two reasons. First, the 76 × 76 × 76 mm³ HX was leak-proof, and the measured fin thickness was 0.110 ± 0.01 mm, which is consistent with the results of the fin thickness study of the ProX 200. Second, the build orientation and the printer parameters for the coupons were validated also for the large scale, thus demonstrating the scale-up procedure. However, as the HX is only 76 × 76 × 76 mm³ and the material is maraging steel, other printers may have to be used to print a bigger build volume out of higher temperature materials such as Ni-based superalloys.

Based on the successful scaled-up procedure demonstrated with the ProX 200, the ProX 300 (Inc718) was selected to fabricate the 94 × 87.6 × 94.4 mm³ MMHX with high-temperature material (Inconel). Despite having lower printing resolution compared to the EOS M290 (20 μm), the ProX 300 (Inc718) reduces fabrication time and cost, which are crucial for making the MMHXs competitive in the aerospace market. The printing time for this MMHX using the ProX 300 (Inc718) was four days, 50% faster than the EOS M290 (20 μm). The same percentage was also reflected in the quoted cost of the HX.

Based on the results of the fin thickness study, in order to fabricate this MMHX with a minimum fin thickness, the designed fin thickness was set at 0.100 mm. The base thickness was set to 0.500 mm to ensure a leak-proof unit. Two smaller sectional sample coupons were also created for measuring internal structure dimensions, considering it would be very difficult to directly measure the features' dimensions inside the unit. Among the possible postprocessing steps, the unit underwent powder removal, stress relief, and removal from the base plate. Further high-temperature treatment or high-pressure treatment could be applied in case the unit leaked to reduce the porosity. The unit was cleaned multiple times using an ultrasonic cleaner to ensure complete removal of the remaining nonsintered powder. After cleaning, the actual mass was compared to the theoretical mass, resulting in a satisfactory difference between the two of 3%. Figure 12 shows the additively manufactured Inconel 718 unit. The actual fin thickness of the 94 × 87.6 × 94.4 mm³ MMHX fabricated unit using the ProX 300 (Inc718) was determined based on the measurements of the small section coupons. As expected from the fin study, the fin thickness of the sectional coupons was confirmed to be 0.220 mm. The HX core was leak-tested on both sides under pressure of 340 kPa. The unit was pressurized with nitrogen while being submerged in water. No leaks were observed, confirming that the separator plates met the requirements. Therefore, no additional postprocessing (e.g., hot isostatic pressing) was needed.

Based on this successful outcome, a comprehensive thermal and hydraulic characterization was made, where the heat removal density of the HX was determined to be 6.4 kW/kg, with an effectiveness of 90% and COP of 150, with a 600 °C temperature difference between the hot and cold streams at their respective inlets [42]. In order to evaluate the MMHX performance, we compared it with state-of-the-art HXs built with AM. Table 6 shows a comparison among studies of AM heat exchangers based on mass-based power density (kW/kg), volume-based power density (W/cm³), COP, and heat exchanger effectiveness, ε.

As can be seen in Table 6, our present MMHX suffered a reduction of about 36% in mass-based power density over that of Zhang et al. [18]. However, we achieved a substantial improvement in COP over Zhang et al. [16]–[18] MMHX, which achieved an effectiveness of 55% and COP of 62.5. The fin density of our 94 × 87.6 × 94.4 mm³ MMHX is 63.5 fins per inch (FPI), while Zhang et al.'s MMHX [18] had fin densities of 50.3 FPI and 31.5 FPI for the straight and inclined sides, respectively. As shown by Arie et al. [19] and Battaglia [42], the MMHX showed better performance when compared with state-of-the-art plate and fin heat exchangers in terms of both mass-based power density and COP, showing the excellent capabilities of the MMHX.

As presented by Moon et al. [44] a higher mass-based power density equal to 15.7 kW/kg was obtained, which represents a four times enhancement in heat transfer performance and 11 times pressure drop penalty when compared to their tube-in-tube HX baseline. When compared to our MMHX this work, Moon et al. [44] obtained a 60% in mass-based power density enhancement. Note that in the Moon et al. [44] HX the thermal conductivity of the AlSi10 Mg was about eight times higher than our MMHX made of Inconel 718. In addition, our MMHX was tested with nitrogen and air, which heat thermal capacity was about four times lower than water and WEG mixture, thus strongly affecting our HX heat transfer rate.

In order to assess the fabrication quality, CT scanning was performed on the HX shown in Fig. 13. The scans were performed twice. The first scan was a full-size scan of the heat exchanger as shown in Fig. 13(a). This CT scan had a resolution of 70–80 μm. An analysis of the full-scale result showed no cracking between the hot and cold streams. However, leftover powder was observed to be blocking some of the channels as shown in Fig. 13(b). Despite using the ultrasonic cleaner, not all of the powder could be removed, as some of the powder might have been sintered during the stress relief process.

A more detailed scan was performed on a portion of the HX (38 $×$ $38×38$ mm³) to analyze the quality of the fins. The detailed scan had a resolution of 35 μm. From the scan results, the thickness of the fins was measured as 0.220 mm, confirming once more the expected value. However, a small portion of the cold side's fins was found to show traces of some defects, as shown in Fig. 13(c). This could be because of an interruption in printing due to the need to refill the powder in the middle of the print. Since the defects did not cause any leaking between the hot and cold streams, they were not considered detrimental to the extent that would make the unit untestable.

This paper focused on the effect of selected printing parameters on additive manufacturing of high temperature materials for high performance and compact manifold microchannel heat exchangers using the selected laser sintering technique. While MMHXs provide optimum distribution of the flow among the heat transfer surface microchannels, as well as keeping the flow within the hydrodynamically developing flow regime, thus offering superior heat transfer performance and heat exchanger heat removal density, they present fabrication and joining/assembly challenges for build volume and performance metrics of practical significance. To improve the MMHX's performance, the additive printing orientation was modified to increase the fin density. By rotating the entire HX by a 45 deg angle with respect to the build plate, it was possible to successfully print for the first time both sides of the cross-flow MMHX with straight fins, thus increasing the fin density by 30%. As for the feature resolution, to improve the heat removal density of the HX, the fins had to be fabricated as thinly as possible, and the base thickness needed to be minimized while meeting the structural integrity requirements.

A study was performed to determine the smallest fin thickness that could be fabricated using the selected laser sintering technique and the minimum base thickness that could ensure a structurally robust and leak-proof HX. Multiple coupons with different fin and base thicknesses were printed using multiple printers with different printer parameters. In addition, the fin structure distortion was considered. Different test coupons were fabricated to examine the fin quality before printing the full-size HX, and it was confirmed that with optimized printer orientation and laser power, no structure distortion was observed from the test coupons. For the full-size HX, CT scan results show that there is fin distortion during the larger size print, but the distortion is a very small portion (less than ∼2%) which does not significantly impact the HX performance. The results showed that fin thicknesses as small as 0.110 mm could be fabricated using the ProX 200 machine. The ProX 300 and EOS M290 machines were less accurate in printing small thickness fins but can accommodate fabrication of bigger built-volume HXs. Reducing the printing layer thickness and laser power intensity can improve the fabrication quality. For 340 kPa, a base layer thickness of 300 μm proved to be leak-proof and robust. Based on the knowledge acquired from the coupon study, multiple MMHXs were fabricated up to a size of 94 × 87.6 × 94.4 mm³. The leak tightness of the MMHXs was demonstrated through a leakage test under the desired design pressure. As to the fabrication, the 94 × 87.6 × 94.4 mm³ manifold microchannel represented the largest of such HXs printed out of Ni-based superalloy materials, using additive manufacturing and with a fin thickness of 0.220. As to its performance, the HX achieved an expected heat removal density of 6.4 kW/kg and volumetric power density of 9.3 kW/liter, with an effectiveness of 90% and COP of 150, and a 600 °C temperature difference between the hot and cold streams at the inlets.

The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof or those of Boeing's. The authors wish in particular to thank ARPA-E Program Directors, Dr. David Tew, and Dr. Adison Stark, and other ARPA-E staff including Mr. Geoffrey Short, and Mr. Joel Fetter, for their technical insight and advice over the course of this project.

• U.S. Department of Energy, ARPA-E division, under subrecipient (Award No. DE-AR0000692; Funder ID: 10.13039/100009224).

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.

$s$ =

spot size (m)

$t$ =

thickness (m)

Greek Symbols

$α$ =

angle of inclination of the unit during the print (deg)

$β$ =

angle of inclination of the unit during the print (deg)

Subscripts

fin =

fin

fin-base =

measure of the fin taken at the base

inclined =

relative to the print with 45 deg inclined orientation

laser =

relative to the 3D printer laser

vertical =

relative to the print with vertical orientation

Abbreviations

AM =

additive manufacturing

CNC =

computer numerical control

COP =

coefficient of performance

FPI =

fin per inch

HX =

heat exchanger

INC625 =

Inconel 625

INC718 =

Inconel 718

MMHX =

manifold microchannel heat exchanger

SLM =

selective laser melting

DMLS =

direct metal laser sintering