This contribution addresses the complete process chain of an annular aerospike breadboard engine fabricated by laser powder bed fusion using the nickel-based superalloy Inconel® 718. In order to qualify the material and process for this high-temperature application, an extensive material characterization campaign including density and roughness measurements, as well as tensile tests at room temperature, 700, and 900 °C, was conducted. In addition, various geometric features such as triangles, ellipses, and circular shapes were generated to determine the maximum unsupported overhang angle and geometrical accuracy. The results were taken into account in the design maturation of the manifold and the cooling channels of the aerospike breadboard engine. Postprocessing included heat treatment to increase mechanical properties, milling, turning, and eroding of interfaces to fulfill the geometrical tolerances, thermal barrier coating of thermally stressed surfaces for better protection of thermal loads, and laser welding of spike and shroud for the final assembly as well as quality assurance. This contribution goes beyond small density cubes and tensile samples and offers details on the iterations necessary for the successful printing of large complex shaped functional parts. The scientific question is how to verify the additive manufacturing process through tensile testing, simulation, and design iterations for complex geometries and reduce the number of failed prints.

Aerospike engines are a promising type of rocket engines that have the potential to improve the efficiency of rocket propulsion systems. Unlike traditional bell-shaped engines, aerospikes have a cone-shaped external contour that provides a more efficient combustion of the propellant under various environmental conditions.1–3 The design and manufacturing of aerospike engines presented a challenging task due to their complex internal and external geometries. However, the recent advancements in additive manufacturing (AM) have made it possible to create these complex shapes with greater accuracy and precision.4 Among the various AM techniques, powder bed fusion (PBF-LB/M) has become a crucial technology for manufacturing aerospikes, due to its ability to produce parts with intricate geometries and internal structures made of high-temperature and corrosion-resistant materials like nickel-based superalloys in near-net shape. PBF-LB/M is a layer-based AM process that uses a laser as an energy source to melt powder feedstock to generate complex three-dimensional components.5,6 Inconel® 718 is known for high strength and corrosion resistance at elevated temperatures up to 650 °C as well as high wear resistance.7 The combination of high strength and ductility results from solution annealing to dissolve alloying elements into the matrix and double aging to form γ′ Ni3(Al,Ti) and γ″ (Ni3Nb) phases. If the solution annealing is not performed properly, undesirable brittle phases such as NbC, δ-Ni3Nb, and Laves phase form.8 Conventional machining of Inconel® 718 proves difficult due to the low thermal conductivity and, therefore, high surface temperatures, leading to high tool wear.9 The weldability, on the other hand, is very good, which is why the alloy is now commonly used in AM.10 Studies on processing Inconel® 718 with PBF-LB/M have focused on density,11,12 microstructure,13,14 mechanical properties,14–16 or geometry features.12 Herzog et al.12 have studied design features such as walls, overhangs, bore holes, and teardrop shapes, with their minimal feature sizes and effects on accuracy and roughness. Gas tightness and heat dissipation in complex shapes were not analyzed. In 1995, NASA was funding a reusable launch vehicle program, where Lockheed Martin won with a design using a linear aerospike engine developed by Rocketdyne,1,17–19 but the program was discontinued in 2001 after delays and technical difficulties on other parts of the project. Besnard et al.20 reported on a student activity and the design of an annular graphite aerospike conventionally machined, which exploded 200 ms after ignition on the static test stand. Two years later, two test flights could be conducted, but again, parts of the engine melted away.21 In recent years, the interest in aerospikes has increased with several companies (e.g., Pangea Aerospace,22–24 RocketStar, Stoke Space, and Polaris) and research organizations25 working toward realizing their full potential using AM. Pangea Aerospace has been the first company to successfully test an AM aerospike engine, but so far, no aerospike has made it into space. Despite the promising results, there is still much work to be done in the area of aerospike manufacturing since the high operating temperatures in the spike need to be dissipated through either the high thermal conductivity of the chosen material or the complex cooling design. In order to print a functional part, the AM process needs to be verified regarding density, roughness, mechanical properties, and geometrical limitations first. Then, the design follows function taking into account the geometrical limitations of AM. All functional parts then need postprocessing to finish interfaces for assembly.

This contribution shows the results of the European Space Agency’s (ESA) ASPIRER (AeroSPIke Rocket Engine Realisation) project program, where Fraunhofer IWS was responsible for the manufacture of the spike and shroud, as well as the catalyst housing designed by the Institute of Aerospace Engineering (ILR) (see Fig. 1 and explained in detail by Dorau et al.26). It describes the whole process chain from AM process verification to design studies to final part printing, postprocessing, and final assembly.

FIG. 1

ASPIRER aerospike engine exploded view. Adapted with permission from Dorau et al., “Numerical analysis of an additively manufactured 6kN hydrogen peroxide/kerosene aerospike breadboard engine,” in 8th Edition of the Space Propulsion Conference, Estoril, Portugal, 9–13 May 2022 (Association Aéronautique et Astronautique de France, Paris, 2022) (Ref. 27).

FIG. 1

ASPIRER aerospike engine exploded view. Adapted with permission from Dorau et al., “Numerical analysis of an additively manufactured 6kN hydrogen peroxide/kerosene aerospike breadboard engine,” in 8th Edition of the Space Propulsion Conference, Estoril, Portugal, 9–13 May 2022 (Association Aéronautique et Astronautique de France, Paris, 2022) (Ref. 27).

Close modal

While the process chain for the conventionally manufactured catalyst housing only includes mechanical processing, quality assurance, and aging, the more complex AM process chain (see in Fig. 2) consists of production and quality assurance, test specimen manufacturing, and testing as well as joining via laser welding and the deposition of a thermal barrier coating (TBC) by means of thermal spraying.

FIG. 2.

Process chain for the spike and the shroud.

FIG. 2.

Process chain for the spike and the shroud.

Close modal

All components were manufactured using a Renishaw AM250 system. The powder used is the nickel-based superalloy Inconel® 718 (NiCr19Fe18Nb5Mo3) supplied by EOS GmbH. The particle size distribution determined by a particle analyzer CAMSIZER® X2 was d10: 19.5 μm, d50: 33.2 μm, d90: 53.8 μm, with a mean sphericity value of 0.883. A scanning electron microscope (SEM) image on a JEOL JSM 6610LV confirms the presence of mostly spherical particles with some small satellites (see Fig. 3).

FIG. 3.

SEM image of the Inconel® 718 powder.

FIG. 3.

SEM image of the Inconel® 718 powder.

Close modal

Two parameter sets previously developed at Fraunhofer IWS with the layer thickness (LT) of 30 and 60 μm were used for material characterization. The laser power was 150 W for both layer thicknesses, the scanning speed was 600 mm/s for LT = 30 μm and 875 mm/s for LT = 60 μm, and the hatch distance was 100 and 90 μm. For up-skin and down-skin surfaces, the following adapted parameters were used:

  • Up-skin: LT = 30 μm not activated, LT = 60 μm one remelt with same parameters; and

  • Down-skin: LT = 30 μm one remelt with 100 W laser power and 110 μm hatch distance, LT = 60 μm three layers with 50 W, 80 μm hatch distance, and 1000 mm/s scanning velocity.

For density measurements, three cubes of 10 × 15 × 10 mm3 were built for each parameter set and microsectioned in the XY (in layer) and XZ (along build direction) planes. Micrographs were then taken with an Olympus GX51 light microscope, and the porosity was determined via image analysis. For the tactile measurement, specimens with different overhang angles defined by the angle between the substrate and the surface of 15°, 30°, 45°, 60°, 75°, and 90° were built and measured at the up-skin and down-skin surfaces with Accretech Surfcom Touch 50. For the tensile tests, cylinders were printed and machined according to DIN EN ISO 50125 B5 × 25. Varying parameters were LT = 30 and 60 μm, the manufacturing orientation 0° (horizontal), 45° (diagonal), and 90° (vertical) as well as the conditions as built (AB) and solution annealed and aged (SAA). Each condition was tested with at least three specimens. At room temperature (RT), the Inspekt Table 50 kN according to DIN EN ISO 6892-1 and, for 700 and 900 °C, the universal testing machine ZwickRoell Z1476 with an air oven from MAYTEC in accordance with DIN EN ISO 6892-2 were used. Simulations were performed using Amphyon 2020 by Oqton with a calibration linked to the used machine and powder. Solution annealing (SA) to dissolve the alloying elements such as Al, Nb, and Ti was performed using a Nabertherm N 41/H chamber oven in accordance with AMS5663™28 at 980 °C for 2 h and air-cooled. Subsequent aging (SAA) to precipitate second phases γ′ and γ″ was performed at 720 °C for 8 h, followed by 620 °C for 8 h using a Xerion X.VAC-VF Bottom oven. The final engine components were solution annealed and aged (SAA) in X.VAC, as they did not fit the N 41/H chamber oven. 3D scanning was carried out using GOM ATOS Core 135 SN for preliminary parts and GOM ATOS Core 300 SN for the full size components larger than 130 mm. Atmospheric plasma spraying was conducted using a GTV F6 plasma gun using an Ar/H2 plasma gas mixture and commercially available Oerlikon Metco AMDRY NiCoCrAlY spray powder to deposit the metallic bondcoats. Suspension plasma spraying was performed using an Oerlikon Metco Simplex Pro plasma gun with an Ar/H2 plasma gas mixture in order to coat the ceramic yttria-stabilized zirconia (YSZ) topcoats.29 For laser beam joining, a 50 μm fiber 1070 μm wavelength alta prime™ 4000 nLIGHT laser with a spot size of 150 μm, a power in the range of 0.4–2.2 kW, and a K-Lab Scout300 scanner with a scanning speed of 83.3 m/s were used.

1. Density

The parameter set with LT = 30 μm reached a mean relative density of 99.97 ± 0.01% utilizing a volume energy density (VED) of 83.3 J/mm3. For the 60 μm layer thickness parameter set, the relative density using a VED of 42.3 J/mm3 was slightly lower with 99.82 ± 0.04%. Micrographs of the corresponding specimens are shown in Fig. 4.

FIG. 4.

Exemplary XZ plane density cube micrographs with LT = 30 μm (a) and LT = 60 μm (b).

FIG. 4.

Exemplary XZ plane density cube micrographs with LT = 30 μm (a) and LT = 60 μm (b).

Close modal

2. Roughness

A comparison of up- and down-skin surface roughness average (Ra) for both layer thicknesses can be seen in Fig. 5. Down-skin surfaces could only be built without support starting from 45° overhang angle. Besides 15° and 30° overhang angles at the up-skin surface, Ra is lower for all surfaces of LT = 30 μm specimens, which is expected due to a lower staircase effect. In general, Ra is lower for larger overhang angles, as expected from the literature.30,31 Overall, mean Ra is around 40% smaller for LT = 30 μm vs LT = 60 μm. The roughness depends not only on the main process parameters but also on special up-skin or down-skin parameters like remelting or adapted laser power or scanning velocity for several layers as described in Sec. II. The high roughness values Ra for overhang angles 15° and 30° for LT = 30 μm can be attributed to the lack of an up-skin parameter set activated. The high roughness values Ra for overhang down-skin surfaces for LT = 60 μm can be explained with the low laser power for the down-skin parameter set, leading to insufficient melting and, therefore, higher roughness.

FIG. 5.

Surface roughness on up- and down-skin surfaces of different overhang angles.

FIG. 5.

Surface roughness on up- and down-skin surfaces of different overhang angles.

Close modal

3. Tensile testing

The results for the 0.2% offset yield strength (YS) and ultimate tensile strength (UTS), as well as the fracture strain (A) from tensile tests at RT, are shown in Table I and Fig. 6. SAA increased the YS as well as the UTS independent of the orientation or layer thickness of the samples as expected due to solution annealing and precipitation hardening.32 For all samples, the 45° orientation is superior in YS, which was also seen by Ref. 15 probably due to a favorable stress state since maximum shear stress is at 45° to the loading direction, which translates to perpendicular to the build direction. The same accounts for UTS at the 30 μm layer thickness samples, while all the 60 μm samples show the highest UTS for 0° orientation. All SAA specimens exceed the requirements of AMS5663™ for YS = 1034 MPa. The UTS requirements of 1276 MPa are fulfilled by most of the SAA specimens, except for the LT = 60 μm and 90° oriented parts.

FIG. 6.

Tensile stress-strain plot of the 90° oriented SAA specimens.

FIG. 6.

Tensile stress-strain plot of the 90° oriented SAA specimens.

Close modal
TABLE I.

Tensile test results at RT with LT.

LT (μm)ConditionOrientation (deg)YS (MPa)UTS (MPa)A (%)
30 AB 777 ± 12 1059 ± 7 28 ± 3 
45 789 ± 11 1067 ± 24 22 ± 7 
90 656 ± 11 966 ± 7 29 ± 3 
SAA 1304 1453 17 
45 1331 ± 36 1467 ± 40 12 ± 7 
90 1211 ± 19 1368 ± 14 11 ± 3 
60 AB 704 ± 22 976 ± 54 16 ± 4 
45 709 ± 17 922 ± 62 9 ± 4 
90 582 ± 30 789 ± 71 6 ± 3 
SAA 1168 ± 26 1370 ± 37 6 ± 2 
45 1168 ± 24 1300 ± 60 3 ± 2 
90 1085 ± 68 1202 ± 10 2 ± 1 
LT (μm)ConditionOrientation (deg)YS (MPa)UTS (MPa)A (%)
30 AB 777 ± 12 1059 ± 7 28 ± 3 
45 789 ± 11 1067 ± 24 22 ± 7 
90 656 ± 11 966 ± 7 29 ± 3 
SAA 1304 1453 17 
45 1331 ± 36 1467 ± 40 12 ± 7 
90 1211 ± 19 1368 ± 14 11 ± 3 
60 AB 704 ± 22 976 ± 54 16 ± 4 
45 709 ± 17 922 ± 62 9 ± 4 
90 582 ± 30 789 ± 71 6 ± 3 
SAA 1168 ± 26 1370 ± 37 6 ± 2 
45 1168 ± 24 1300 ± 60 3 ± 2 
90 1085 ± 68 1202 ± 10 2 ± 1 

SAA leads to an increase in the strength and a decrease in the ductility as expected. Elongation A of LT = 60 μm samples is high for 0°, moderate for 45°, and low for 90° oriented samples. This was not observed for LT = 30 μm samples. Depending on which specimen orientation is considered from the AMS5663™, the 30 μm layer thickness SAA specimens pass or closely fail the minimal elongation, ranging between 12% and 6%, while the 60 μm layer thickness SAA specimens are less ductile and fail the standard. Considering the high exceedance of the YS and UTS and the close failure of A from the AMS5663™ values, the heat treatment should be altered to perform less hardening, with slightly lower strength but more elongation. A reason for the overhardening we report might be the used oven, which is not capable of reaching cooling rates similar to air cooling, which would be necessary for complete dissolution.13 The high-temperature test results at 700 and 900 °C can be seen in Table II and Fig. 6.

TABLE II.

Tensile test results at high temperature with different LTs under SAA condition and 90° orientation.

LT (μm)Temp. (°C)YS (MPa)UTS (MPa)A (%)
30 700 875 ± 35 973 ± 20 6.9 ± 3.9 
30 900 190 ± 8 198 ± 4 36.2 ± 12 
60 700 801 815 ± 69 0.3 ± 0.3 
60 900 178 ± 7 198 ± 9 10.0 ± 6 
LT (μm)Temp. (°C)YS (MPa)UTS (MPa)A (%)
30 700 875 ± 35 973 ± 20 6.9 ± 3.9 
30 900 190 ± 8 198 ± 4 36.2 ± 12 
60 700 801 815 ± 69 0.3 ± 0.3 
60 900 178 ± 7 198 ± 9 10.0 ± 6 

The LT = 30 μm specimens exhibit 72% of the YS and UTS strengths at 700 °C compared to RT. The value for YS required in AMS5663™ at 649 °C is still achieved, whereas UTS misses the required value by 27 MPa. The strengths then drop at 900 °C to 15% of the RT values, with YS and UTS approaching each other. For the elongation in 4D (initial gauge length for the determination of strain, four times the diameter of the round specimen), AMS5663™ requires no reduction at 649 °C compared to room temperature. In contrast, for the 700 °C 30 μm samples, we report a significant reduction of 38% for A relative to RT. At 900 °C, A then increases more than threefold. The 60 μm layer thickness specimens break in the 700 °C tensile tests in two out of three cases before reaching an elastic yield point. Thereby, YS is determined by only one specimen. UTS has a high standard deviation due to early fractures. For the specimens that broke prior to the yield point, the average UTS is 776 MPa, whereas for the third specimen that reached the yield point, the UTS is 891 MPa. The values for the strains are very low, corresponding to the high brittleness. None of the 60 μm layer thickness samples meet the standard requirements. As a result of the tensile tests, the 30 μm layer thickness samples consistently prove to be superior over the 60 μm samples. The static and dynamic strength is influenced by the relative density and the surface roughness, with higher relative density and smoother surfaces, resulting in increased strength.33 As the postprocessing of all surfaces of the component to reduce the surface roughness is precluded and due to the superior mechanical properties, surface roughness, and density of the 30 μm parameter set, this set is used for the AM of all following engine components.

Several design features of the spike and shroud were manufactured in order to alter the component’s design and to prove the manufacturability of those features. An initial series of two build jobs was manufactured to determine the maximum radius and ideal shape of channels and manifolds for cooling water and propellant.

The first build contained circular, elliptical (B/H = 2, D = 1/3 H), drop-, diamond (D = H)-, and triangle-shaped (D = S/3−2) cross sections running horizontally in 50 mm long massive cuboids parallel to the substrate surface (see Fig. 7). The dimensions of all cross sections are designed to match the hydraulic diameter D of 8, 9, 10, and 12 mm. All specimens were successfully manufactured. Only the parts with the biggest radii and circular form show significant shape deviation in the overhang area (see Fig. 7) as was also observed in Ref. 12. The second build, therefore, only contained parts with corresponding circle radii of 5 and 6 mm. Furthermore, all the parts of the second build are of tubular shape with a wall thickness of 1 mm as intended for the final components (see Fig. 7). The circular shapes that showed large deviations in the first build were additionally printed supported using thin walls, as well as cylindrical support structures. All parts could be manufactured entirely. The results show again that the circular channels have major deviations in the overhang area. In particular, the circular channel with 6 mm radius is close to an insufficient and incomplete structure. The thin-walled support inside the circular channel could not be manufactured, as it was too thin. Using cylindrical supports, instead, seems promising for circular channels as the deviations are decreased. The elliptical-standing and drop-shaped components could be manufactured in very good quality in all build jobs, whereby the elliptical shape is superior to the teardrop shape in that the rough down-skin surface accounts for a smaller proportion of the channel surface. The elliptical shape is preferred for channels and manifolds where the higher vertical space requirement can be admitted. The results of the initial critical design study are being incorporated into the development of the engine design (see Fig. 8): the implementation of the water inlet and outlet through the spokes of the spike [Fig. 8(b), red and blue] as well as the definition of the maximum radii at 4.5 mm in the core holes for the connections [Fig. 8(a), red], which are later machined to threads.

FIG. 7.

Critical design study to determine the maximum channel dimensions.

FIG. 7.

Critical design study to determine the maximum channel dimensions.

Close modal
FIG. 8.

Design adaptions following the results of the critical design feature studies.

FIG. 8.

Design adaptions following the results of the critical design feature studies.

Close modal

In a second series of critical design feature builds, a close to final design was studied with regard to critical overhangs, minimum wall thicknesses, rapid cross-sectional area changes, and the kerosene injector holes, which was already presented in Ref. 26.

The shroud water inlet manifold [Fig. 9(e)] showed multiple regions that could lead to manufacturing defects, such as thin to broken walls of smaller than 1 mm thickness and deflections that lead to wiper and part damage, as well as areas affecting the pressure to drop during throughflow. Therefore, it was completely redesigned. On the spike dome [Fig. 9(b)], minor adjustments to the overhang angle and channel size were made in order to increase the manufacturability and flow velocity based on simulations.27 

FIG. 9.

Critical design features of the first iteration finished engine design. Adapted with permission from Dorau et al., “Development of an additively manufactured hydrogen peroxide/kerosene 6kN aerospike breadboard engine,” in 72nd International Astronautical Congress (IAC), Dubai, United Arab Emirates, 25–29 October 2021 (International Astronautical Federation, Paris, 2021) (Ref. 26).

FIG. 9.

Critical design features of the first iteration finished engine design. Adapted with permission from Dorau et al., “Development of an additively manufactured hydrogen peroxide/kerosene 6kN aerospike breadboard engine,” in 72nd International Astronautical Congress (IAC), Dubai, United Arab Emirates, 25–29 October 2021 (International Astronautical Federation, Paris, 2021) (Ref. 26).

Close modal

Selbmann et al.34 have already shown that injector holes cannot be manufactured by PBF-LB/M, as they deviate by around 12% from the desired diameter and cross-sectional area. Observing the deflections of the spike segment with the joining interface (see Fig. 10), machining offsets at the spike flange are increased to 1.5 mm.

FIG. 10.

Deviation between the designed and manufactured spike spoke and flange segment.

FIG. 10.

Deviation between the designed and manufactured spike spoke and flange segment.

Close modal

The spike and shroud cannot be manufactured monolithically, as accessibility for the TBC coating on the spike and the eroding of the injector holes in the shroud must be guaranteed. An open tip has been incorporated into the spike design to optimize the removal of powder from its internal cooling channels. Laser beam welding is utilized to close the tip by joining a lid. After the postprocessing of the final spike and shroud, both are joined together with circumferential laser beam welding. Preliminary tests were conducted to find suitable process parameters for joining, to validate the process toward tightness, as well as to estimate deformations. A laser power of 1100 W led to a collapsing seam [see Fig. 11(a)] and the undesired penetration of the underlying material. With a laser power of 1000 W, a uniform quality seam can be produced along the 1.3 mm deep interface between the spike and lid [see Fig. 11(b)]. As expected from the micrographs, the seam was tight to fluids with high wetting ability as ethanol and acetone at atmospheric pressure. Machining only the lid to get a tight fit is sufficient for joining. For the second study, the circumferential geometry between the spike and shroud interface is rolled off into the plane to simplify the investigations.

FIG. 11.

Micrographs of the welding seam (1) between the spike (2) and lid (3) at laser powers of 1100 (a) and 1000 W (b).

FIG. 11.

Micrographs of the welding seam (1) between the spike (2) and lid (3) at laser powers of 1100 (a) and 1000 W (b).

Close modal

Welding the 3 mm deep butt joint between the spike and shroud also shows a high sensitivity of the weld to the used laser power. Using 1900 W [see Fig. 12(a)] does not result in a complete through-weld and complete penetration. With just 100 W more laser power, a quality, uniform, flank-parallel seam can be produced [see Fig. 12(b)].

FIG. 12.

Micrographs of the welding seam (1) between the shroud (2) and spike (3) at laser powers of 1900 (a) and 2000 W (b).

FIG. 12.

Micrographs of the welding seam (1) between the shroud (2) and spike (3) at laser powers of 1900 (a) and 2000 W (b).

Close modal

To protect the spike from the high thermal loads, a TBC was sprayed onto its shaft. In order to identify expedient surface and spraying conditions that withstand the highest temperatures, a comprehensive preliminary study was performed and published.29 It was found that the columnar-like YSZ coatings are superior compared to the water based vertically cracked YSZ coating. To validate the spraying on the final geometry, two samples of the final spike containing only the complex shaft geometry have been manufactured. The lid was welded onto the tip and finally thermally sprayed.35 Although the coating thicknesses varied along the shaft, the TBC coating system adhered well in the thermally high loaded regions.

From the manufacturing simulation of the spike, expected deformations of ∼1.5 mm result at the flange (see Fig. 13). In order to counteract these defects and to realize successful component manufacturing, the ILR reduced the flange area and we added a bevel compensating for the shrinkage that allowed later layers to bond.

FIG. 13.

Simulated (a) and observed displacement and delamination (b) on the spike flange.

FIG. 13.

Simulated (a) and observed displacement and delamination (b) on the spike flange.

Close modal

Due to the design iteration, the spike could be manufactured completely after 8 days and 19 h, having a 10% reduced mass of 6.5 kg compared to the design of the first iteration. The contour of the outer catalyst housing was consequently adapted to the spike contour (see Fig. 1), so that uneven thermal stresses between the two components are reduced during the firing tests.

Manufacturing of the shroud required several trials. The simulation showed large deformations at connectors for water and sensors (see Fig. 14).

FIG. 14.

Simulation of the displacement for the first built shroud iteration highlighting potentially defective regions below the connectors (yellow).

FIG. 14.

Simulation of the displacement for the first built shroud iteration highlighting potentially defective regions below the connectors (yellow).

Close modal

The initial design [Fig. 15(a)] manufacturing had to be stopped in the area of the largest cross section of the component above the water outlets due to joining defects [see Fig. 15(b)]. Based on this first iteration, the braces below the connections were adjusted in such a way that the exposed cross section and, thus, the amount of heat introduced could be reduced for as long as possible [see Fig. 15(c)] (1). Furthermore, channels were introduced in the area of the largest cross-sectional areas between 170 and 190 mm build-up height to reduce the cross-sectional area (2), along with changes to the number and geometry of external ribs below the kerosene manifold to allow for better heat dissipation (3). Despite the adjustments, joining errors occurred again in the area of the connectors [see Fig. 15(d)].

FIG. 15.

Initial design (a) and corresponding defects (b), design changes second iteration (c), and corresponding defects (d).

FIG. 15.

Initial design (a) and corresponding defects (b), design changes second iteration (c), and corresponding defects (d).

Close modal

All connectors were then made with a combination of a dense material to maximize heat dissipation just below the component surface and cylindrical support to connect the component and dense support for good support removability. This reduced the maximum cross-sectional areas at 170 and 180 mm build height. The shroud was then successfully manufactured after a process time of 6 days and 13 h [see Fig. 16(c)].

FIG. 16.

Built spike (a) compared to machining model to check for postprocessing clearance (positive, red) (b), built shroud (c), compared to CAD model to check for manufacturing deviations (d).

FIG. 16.

Built spike (a) compared to machining model to check for postprocessing clearance (positive, red) (b), built shroud (c), compared to CAD model to check for manufacturing deviations (d).

Close modal

The manufactured spike was 3D scanned and compared with the CAD model in the finished machined state to ensure that all surfaces to be machined had sufficient oversize (positive values) [Fig. 16(b)]. It was found that the pitch radius of the holes in the flange (1) was reduced by ∼0.6 mm more than assumed. As a result, the pitch diameter of the outer catalyst housing to be manufactured was adjusted to the pitch diameter of the spike. At the outermost diameter of the flange, the reserves of the planar machining allowance are reduced toward 0 mm (2). The allowance should be further increased in future build jobs. While the interface between the spike and shroud has more than 1 mm machining allowance on the inside due to the overall component shrinkage (3) (deviations above 1 mm are shown in gray), the radius on the outside is reduced by about 0.6 mm (4). An adjustment of the manufacturing data on the interface can only be made after checking the manufactured shroud. The shaft of the spike has shrunk by about 0.4 mm (5), which is well compensated by the TBC to be sprayed onto it.13,21

A parallelism error of around 0.9 mm occurred between the axis from the spike flange and an axis fitted to the shaft of the spike, as well as a concentricity error between the spike interface and a cylinder fitted to its shaft of 0.4 mm. The spike shaft bends against the direction of the air flow over the powder bed. The shroud shows a general shrinkage of around 0.5 mm [see ±0.5 mm in the histogram of Fig. 16(d)]. On its interface (1), the radius on the outside is reduced by about 0.6 mm, which fits the interface of the spike very well.

During the mechanical postprocessing of the spike and shroud, the highest priority was given to a parallel and concentric alignment of the axis of the spike shaft to the axis of the shroud inner wall. Subsequently, the narrowest cross section for the hot gas flow was set by reworking the flat surfaces at the interface between the two components and their shrinkage was compensated. The intersections between the spike and shroud were turned. The underside of the flange, where the seal to the outer catalyst housing is located, was manufactured by electrical discharge machining turning (EDM) to achieve the lowest possible surface roughness. All screw connections were milled, and the injector holes and pressure and temperature measuring points were made by die sinking EDM. Following the mechanical rework, both components were 3D scanned again to check for dimensional accuracy. The scanned models were then virtually assembled via their interface in order to measure the narrowest hot gas flow cross section between the spike and shroud. A script was used to rotate the spike relative to the shroud in the fit of the interface to determine an optimized alignment. The aim of the optimization was to minimize the standard deviation of all minimum distances between the spike and the shroud along the circumference, which should ensure a particularly uniform gap area along the circumference. The value of the mean minimum gap distance at the narrowest flow cross section is 7.13 ±0.025 mm, which is about 6.5% deviation from the ideal gap diameter of 6.7 mm. It should be noted that this gap is still filled by the 0.6 mm thick TBC layer, which ultimately makes the gap about 0.25 mm smaller than intended. This will likely influence the engine performance and was already studied by Dorau et al.18 

The TBC consisting of a bond and top coat was sprayed onto the spike shaft. A cover was used to protect the machined base of the shroud during spraying.35 

For the joining of the spike and the shroud, both components were tightened together with ropes running inside the shroud (see Fig. 17). This fixation allows the entire joining process to be carried out in the same position PA (flat position) as in the preliminary tests.

FIG. 17.

Joining of spike and shroud utilizing laser beam welding.

FIG. 17.

Joining of spike and shroud utilizing laser beam welding.

Close modal
  • An aerospike rocket engine made of Inconel® 718 was successfully manufactured using PBF-LB/M in combination with machining, heat treatment, welding, and TBC.

  • In-depth material property analysis such as tensile testing at room temperature, and elevated temperatures at 700 and 900 °C for horizontal and vertical build directions with two layer thicknesses of 30 and 60 μm, proved that the AMS5663™ aerospace material specification is met.

  • AM process simulation qualitatively indicates the critical areas of possible defects and serves as a baseline for design iterations reducing overheating and large deformations by reducing cross-sectional area gradients.

  • Scalability from test parts to complex large parts remains challenging in AM as test cuboids have a constant, small scanning area and, therefore, a constant energy input, while the aerospike engine geometry with changing scanning area per layer has a variable energy input, leading to unforeseen defects such as overheating and warping.

  • The developed process chain, which included design iterations, process simulation, material property studies, and critical design studies, can be transferred to other Inconel® 718 applications with high thermal loads and corrosion requirements, e.g., heat exchangers, engines, or turbines.

  • Companion samples from the final parts should undergo the same material analysis to extend the preverification study.

  • Further studies on corrosion behavior and thermal cycling would expand the data available under operation conditions of the engine.

We would like to thank ESA for funding the ASPIRER project through a General Support Technology Program (GSTP) with Contract No. 4000130551/20/NL/MG. Furthermore, we thank ESA Technical Officer Simon Hyde and acknowledge ArianeGroup SAS for the consulting input regarding AM process stability and thank all our students and technical staff, who contributed substantially to the realization of the project’s multidisciplinary goals.

The authors have no conflicts to declare.

Alex Selbmann: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Validation (equal); Visualization (equal); Writing – original draft (equal). Samira Gruber: Conceptualization (equal); Data curation (equal); Investigation (equal); Methodology (equal); Project administration (equal); Writing – original draft (equal); Writing – review & editing (equal). Martin Propst: Conceptualization (supporting); Investigation (supporting); Project administration (lead). Tim Dorau: Investigation (supporting); Project administration (lead). Robert Drexler: Data curation (supporting); Writing – review & editing (supporting). Filofteia-Laura Toma: Data curation (supporting); Writing – review & editing (supporting). Michael Mueller: Funding acquisition (lead). Lukas Stepien: Supervision (lead). Elena Lopez: Supervision (supporting); Writing – review & editing (supporting). Christian Bach: Funding acquisition (lead); Project administration (lead); Writing – review & editing (supporting). Frank Brueckner: Supervision (supporting); Writing – review & editing (supporting). Christoph Leyens: Funding acquisition (supporting); Supervision (supporting); Writing – review & editing (supporting).

1.
Rocketdyne
, see https://apps.dtic.mil/sti/citations/AD0384856 for “Final report, advanced aerodynamic spike configurations: AD0384856, AFRPL-TR-67-246-Vol II” (retrieved 22 September 2023).
2.
S.
Corda
,
A. N.
Bradford
,
T. R.
Moes
,
T. H.
Cox
,
R. S.
Monaghan
,
L. S.
Voelkner
,
G. C.
Corpening
, and
R. R.
Larson
, “Flight testing the linear aerospike SR-71 experiment (LASRE),” NASA/TM-1998-206567 (NASA Center for Aerospace Information, Hanover, MD, 1998) (retrieved 22 September 2023).
3.
J.
Ruf
and
P.
McConaughey
, “
The plume physics behind aerospike nozzle altitude compensation and slipstream effect
,” in
33rd Joint Propulsion Conference and Exhibit
,
Reston, VA,
6–9 July 1997 (
American Institute of Aeronautics and Astronautics,
Reston, VA, 1997), p.
07061997
.
4.
S. A.
Whitmore
and
I. W.
Armstrong
, “
Development and testing of conventional and additively manufactured aerospike nozzles for small satellite propulsion
,” in
AIAA Propulsion and Energy 2019 Forum, Indianapolis, IN, 19–22 August 2019
(
American Institute of Aeronautics and Astronautics,
Reston, VA, 2019), p.
08192019
.
5.
VDI 3405:2014-12, “Additive manufacturing processes, rapid manufacturing—Basics, definitions, processes,” Verein deutscher Ingenieure e.V., Dusseldorf, Germany, 2014.
6.
ISO/ASTM 52900:2021, “Additive manufacturing—General principles—Fundamentals and vocabulary,” Beuth Verlag GmbH, Berlin; Vol. 01.040.25.
7.
M. J.
Donachie
and
S. J.
Donachie
,
Superalloys
(
ASM International
, Materials Park, OH,
2002
).
8.
W. M.
Tucho
,
P.
Cuvillier
,
A.
Sjolyst-Kverneland
, and
V.
Hansen
, “
Microstructure and hardness studies of Inconel 718 manufactured by selective laser melting before and after solution heat treatment
,”
Mater. Sci. Eng. A
689
,
220
232
(
2017
).
9.
Q.
Yin
,
Z.
Liu
,
B.
Wang
,
Q.
Song
, and
Y.
Cai
, “
Recent progress of machinability and surface integrity for mechanical machining Inconel 718: A review
,”
Int. J. Adv. Manuf. Technol.
109
,
215
245
(
2020
).
10.
A.
Marques
,
Â.
Cunha
,
M. R.
Silva
,
M. I.
Osendi
,
F. S.
Silva
,
Ó.
Carvalho
, and
F.
Bartolomeu
, “
Inconel 718 produced by laser powder bed fusion: An overview of the influence of processing parameters on microstructural and mechanical properties
,”
Int. J. Adv. Manuf. Technol.
121
,
5651
5675
(
2022
).
11.
P.
Kumar
,
J.
Farah
,
J.
Akram
,
C.
Teng
,
J.
Ginn
, and
M.
Misra
, “
Influence of laser processing parameters on porosity in Inconel 718 during additive manufacturing
,”
Int. J. Adv. Manuf. Technol.
103
,
1497
1507
(
2019
).
12.
D.
Herzog
,
K.
Asami
,
C.
Scholl
,
C.
Ohle
,
C.
Emmelmann
,
A.
Sharma
,
N.
Markovic
, and
A.
Harris
, “
Design guidelines for laser powder bed fusion in Inconel 718
,”
J. Laser Appl.
34
, 012015 (
2022
).
13.
K. N.
Amato
,
S. M.
Gaytan
,
L. E.
Murr
,
E.
Martinez
,
P. W.
Shindo
,
J.
Hernandez
,
S.
Collins
, and
F.
Medina
, “
Microstructures and mechanical behavior of Inconel 718 fabricated by selective laser melting
,”
Acta Mater.
60
,
2229
2239
(
2012
).
14.
K.
Moussaoui
,
W.
Rubio
,
M.
Mousseigne
,
T.
Sultan
, and
F.
Rezai
, “
Effects of selective laser melting additive manufacturing parameters of Inconel 718 on porosity, microstructure and mechanical properties
,”
Mater. Sci. Eng. A
735
,
182
190
(
2018
).
15.
F.
Caiazzo
,
V.
Alfieri
,
G.
Corrado
, and
P.
Argenio
, “
Mechanical properties of Inconel 718 in additive manufacturing via selective laser melting: An investigation on possible anisotropy of tensile strength
,” in
3rd International Forum on Research and Technologies for Society and Industry (RTSI), Modena, Italy, 11–13 September 2017
(
IEEE,
Piscataway, NJ, 2017), pp.
1
4
, 92017.
16.
A.
Thomas
,
M.
El-Wahabi
,
J. M.
Cabrera
, and
J. M.
Prado
, “
High temperature deformation of Inconel 718
,”
J. Mater. Process. Technol.
177
,
469
472
(
2006
).
17.
F. B.
Lary
, “Advanced cryogenic rocket engine program, aerospike nozzle concept. Volume 2: AD0387190,” Report No. AFRPL-TR-67-280 (Air Force Rocket Propulsion Laboratory, Edwards AFB, CA, 1968) (retrieved 22 September 2023).
18.
M.
Rowbotham
, “
XRS-2200 linear aerospike engine—Use of Pro/ENGINEER for determining mass properties
,”
35th Joint Propulsion Conference and Exhibit, Los Angeles, CA, 20–24 June 1999
(
American Institute of Aeronautics and Astronautics
, Reston, VA,
1999
).
19.
M. G.
D'Agostino
,
Y. C.
Lee
, and
T. S.
Wang
, “
X-33 XRS-2200 linear aerospike engine sea level plume radiation
,” in
26th JANNAF Exhaust Plume Technology Subcommittee Meeting, San Antonio, TX, 5–9 November 2001
(NASA, Washington, DC,
2001
).
20.
E.
Besnard
,
J.
Garvey
,
T.
Holleman
,
T.
Mueller
,
H. H.
Chen
, and
H.-P.
Chen
, “
Student development and test of a gimbaled annular aerospike engine
,” in
38th Joint Propulsion Conference & Exhibit, Indianapolis, IN, 7–10 July 2002
(American Institute of Aeronautics and Astronautics, Reston, VA,
2002
).
21.
E.
Besnard
and
J.
Garvey
, “
Development and flight-testing of liquid propellant aerospike engines
,” in
40th AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, Fort Lauderdale, FL, 11–14 July 2004
(
American Institute of Aeronautics and Astronautics
, Reston, VA,
2004
).
22.
Z.
Sapi
,
N.
Palumbo
, and
F.
Rossi
, “
Structural development and manufacturing of the DemoP1 demonstrator
,” in
Aerospace Europe Conference 2021, Warsaw, Poland, 23–26 November 2021
(Council of European Aerospace Societies, Brussels, Belgium,
2021
).
23.
F.
Rossi
,
G.
Esnault
,
Z.
Sápi
,
N.
Palumbo
,
A.
Argemi
, and
R.
Bergström
, “
Research activities in the development of DemoP1: A LOX/LNG aero-spike engine demonstrator
,” in
7th Edition of the Space Propulsion Conference, virtual event, 17–19 March 2021
(Association A豯nautique et Astronautique de France, Paris, France,
2021
).
24.
F.
Rossi
,
Z.
Sápi
,
N.
Palumbo
,
A.
Demediuk
, and
M.
Ampudia
, “
Manufacturing and hot-fire test campaign of the DemoP1 aerospike engine demonstrator
,” in
8th Edition of the Space Propulsion Conference, Estoril, Portugal, 9–13 May 2022
(Association Aéronautique et Astronautique de France, Paris, France,
2022
).
25.
M.
Buchholz
et al, “
Developing a roadmap for the post-processing of additively manufactured aerospike engines
,” in
71st International Astronautical Congress (IAC)—The CyberSpace Edition, virtual event, 12–14 October 2020
(International Astronautical Federation, Paris, France,
2020
).
26.
T.
Dorau
et al, “
Development of an additively manufactured hydrogen peroxide/kerosene 6kN aerospike breadboard engine
,” in
72nd International Astronautical Congress (IAC)
, Dubai, United Arab Emirates, 25–29 October 2021 (International Astronautical Federation, Paris,
2021
).
27.
T.
Dorau
et al, “
Numerical analysis of an additively manufactured 6kN hydrogen peroxide/kerosene aerospike breadboard engine
,” in
8th Edition of the Space Propulsion Conference, Estoril, Portugal, 9–13 May 2022
(Association Aéronautique et Astronautique de France, Paris, France,
2022
).
28.
SAE International
, “AMS5663P nickel alloy, corrosion- and heat-resistant, bars, forgings, rings, and stock for forgings and rings 52.5Ni-19Cr-3.0Mo-5.1Cb (Nb)-0.90Ti-0.50Al-18Fe consumable electrode or vacuum induction melted 1775 °F (968 °C) solution heat treated, precipitation-hardenable” (
2022
).
29.
F.-L.
Toma
,
S.
Gruber
,
A.
Selbmann
,
S.
Tkachenko
,
O.
Kunze
,
K.
Zabransky
,
C.
Leyens
,
L.
Čelko
, and
M.
Tajmar
, “
Performance of suspension sprayed YSZ thermal barrier coatings on Inconel 718 substrates produced by laser powder bed fusion
,” in
Thermal Spray 2022: Proceedings From the International Thermal Spray Conference, Vienna, Austria, 4–6 May 2022
(
DVS Media GmbH
, Düsseldorf, Germany,
2022
), pp.
645
653
.
30.
G.
Strano
,
L.
Hao
,
R. M.
Everson
, and
K. E.
Evans
, “
Surface roughness analysis, modelling and prediction in selective laser melting
,”
J. Mater. Process Technol.
213
,
589
597
(
2013
).
31.
E. E.
Covarrubias
and
M.
Eshraghi
, “
Effect of build angle on surface properties of nickel superalloys processed by selective laser melting
,”
JOM
70
,
336
342
(
2018
).
32.
R.
Zhao
,
Z.
Zhao
,
P.
Bai
,
W.
Du
,
L.
Zhang
, and
H.
Qu
, “
Effect of heat treatment on the microstructure and properties of Inconel 718 alloy fabricated by selective laser melting
,”
J. Mater. Eng. Perform.
31
,
353
364
(
2022
).
33.
N.
Bass
,
S.
Jalui
, and
G.
Manogharan
, “
Net-shape tensile specimens as representatives of material properties of metal additive manufacturing: evaluation and correction factor
,” in ASME 2022 17th International Manufacturing Science and Engineering Conference: Volume 2, West Lafayette, IN, 27 June–1 July 2022 (
American Society of Mechanical Engineers
, New York, 2022), p.
06272022
.
34.
A.
Selbmann
,
S.
Gruber
,
L.
Stepien
,
E.
López
,
A.
Marquardt
,
F.
Brückner
, and
C.
Leyens
, “
Mechanical and geometrical characterization of additively manufactured Inconel® 718 porous structures for transpiration cooling in space applications
,”
SPIE Proc.
11992
,
16–
22 (
2022
).
35.
F.-L.
Toma
,
S.
Gruber
,
A.
Selbmann
,
C.
Leyens
,
M.
Propst
,
T.
Dorau
, and
C.
Bach
, “
Suspension sprayed YSZ thermal barrier coatings on Inconel 718 aerospike manufactured by laser powder bed fusion
,” in
Thermal Spray 2023: Proceedings from the International Thermal Spray Conference (ITSC 2023), Québec City, QC, Canada, 22–25 May 2023
(
ASM International
, Materials Park, OH,
2023
), pp.
675
682
.