Design and Manufacture of SiC/SiC Nozzle Guide Vanes With Environmental Barrier Coatings for High-Pressure Turbines

Efficient and environmentally friendly flying is a key development goal in order to fulfill the worldwide climate targets, for example, the European Green Deal [1]. For aviation, the Green Deal requires a 75% reduction in CO₂ emissions and a 90% reduction in NO_x emissions by 2050 compared to 2020 [2]. In addition to considering alternative fuels and propulsion systems, the efficiency of conventional jet engines must be improved [3,4]. The use of new materials and technologies plays an important role in increasing efficiency [5].

Ceramics are a promising class of materials for significant improvements in the performance of aircraft gas turbines and stationary gas turbines. As early as the 1980s and 1990s, turbine components made of monolithic ceramics (silicon nitride, silicon carbide) were considered and their potential was investigated [6–8]. Due to the lack of damage tolerance of monolithic ceramics, fiber-reinforced ceramics (CMC) are mainly considered for potential use in high-pressure turbines (HPT) today. Due to the long service life required in oxidizing atmospheres, development is concentrating on silicon carbide fiber-reinforced silicon carbide ceramics (SiC/SiC). Protected from steam by an environmental barrier coating (EBC), they exhibit excellent creep resistance at high temperatures in oxidizing atmospheres [9,10]. Compared to the conventionally used, single-crystal nickel-based alloys, SiC/SiC materials offer a 60% lower density. With a similar required wall thickness as a metallic counterpart, which leads to a considerable reduction in weight. Application temperatures of up to 1315 °C result in a temperature increase of over 200 K compared to nickel-based alloys. SiC/SiC components thus require less cooling, which leads to a reduction of the required cooling air or an increase in process temperature [10,11].

The potential of CMC to improve engine efficiency has been extensively analyzed in the literature [12–14]. In particular, the use of cooled CMC stators in high-pressure turbines has been investigated, which enables the modification of important cycle process parameters such as the turbine inlet temperature (T40) or the rotor inlet temperature (T41). The thermal efficiency can be substantially improved by increasing the rotor inlet temperature T41. If metallic stators are replaced by CMC stators, the original turbine inlet temperature can be maintained or even increased, resulting in a higher rotor inlet temperature T41 due to the reduced cooling air requirement of CMCs. As the rotor inlet temperature increases, so does the cooling air requirement and therefore the chargeable coolant mass flow, which counteracts the potential benefits and must be considered. Another approach is, to keep the rotor inlet temperature T41 constant by reducing the turbine inlet temperature T40, which is possible due to the reduced cooling air requirement of CMC stators [12–14]. In terms of climate impact, the second approach is particularly interesting, as it has the potential to reduce NO_x emissions, which are significantly influenced by the flame temperature and the residence time in the combustion chamber [15]. In addition, a reduction in the required cooling air mass flow generally leads to a reduction in the mixing losses between hot gas and cooling air [16,17]. As long as the use of CMC stators is not associated with significantly worse profile and trailing edge losses, it is possible to increase the high-pressure turbine efficiency.

Initial potential studies have shown that the use of CMC stators in the first and second stages of a HPT can reduce the cooling air requirement by 5.5% at a constant rotor inlet temperature T41, and the thermal efficiency can also be improved by up to 1.5% [12]. Wehrel et al. [14] analyzed the replacement of metallic stators by CMC in the first stage HPT on engine performance level. Due to the lower cooling air demand, this led to a reduction of the turbine inlet temperature T40 by 50 K and lowered NO_x emissions by more than 5%, while maintaining a constant rotor inlet temperature T41. In terms of potential, General Electric (GE) estimates that the use of CMC rotors in a GE90-class engine can reduce the total weight by 455 kg or ≈6% of the engine's total weight [18] (dry weight ≈7892.5 kg [19]). In addition to the component weight, the weight of disks can also be reduced [18].

Current state-of-the-art SiC/SiC materials for jet engines are based on the melt infiltration (MI) process. GE's HiPerComp is manufactured via prepreg MI [20–22] and NASA and Safran SiC/SiC via slurry cast MI [23–29]. For the protection from steam-enhanced corrosion, EBC are required for SiC/SiC materials. Commonly used coatings are rare earth mono- or disilicates. Ytterbium disilicate (Yb₂Si₂O₇) and yttrium disilicate (Y₂Si₂O₇, Y-DS) are favorable due to the similar coefficients of thermal expansion compared to SiC/SiC. EBC coatings are mainly applied via thermal spraying or physical vapor deposition (PVD) processes [10,30]. If the coatings are sufficiently thick, they can also provide thermal insulation, since the thermal conductivity of the ceramic coatings is low. Recent research has focused on these so-called thermal and environmental barrier coatings [26,31,32]. The temperature capability of EBCs is between 1482 °C and 1650 °C and is to be increased in the future. The bond coat based on (doped) silicon can withstand 1400 °C–1482 °C [26,32].

After the introduction of SiC/SiC shrouds in the LEAP engine in 2016, the inner liner of the combustion chamber and the stators of the first and second stages of the high-pressure turbine will also be made of SiC/SiC for the new GE9X engine [22,33]. CMC components for stationary gas turbines and jet engines have been manufactured and tested by several companies and research institutes, an overview is given in Ref. [34]. SiC/SiC high-pressure turbine vanes have been researched and tested for the hot gas section of jet engines. The vanes in the first stage of the high-pressure turbine are the hottest components in the engine (≈1700 °C) [35]. The first SiC/SiC vanes were developed within the NASA Ultra-Efficient Engine Technology program [36–40]. The vanes were EBC coated and cooled via trailing edge cooling holes. Film cooling on the suction or pressure side of the surfaces was not used. Burner rig tests were performed in a gas turbine environment. Finite element analysis (FEA) was carried out to predict the thermal and mechanical loads. The work was continued in the Environmentally Responsible Aviation project [12,13]. GE has recently published research on cooled SiC/SiC vanes [41] concentrating on stationary gas turbines. The focus lies on a shell and spar concept, where the CMC shell takes the aerothermal load and the metal spar supports the shell. Film cooling was not integrated in these vanes and is in general not widely researched and published. Only few publications examine experimentally the effects of film cooling holes in flat CMC samples [42–44]. However, to achieve higher temperatures for future engines, Boyle et al. [12] and Delvaux et al. [41] advocate additional film cooling of HPT vanes in combination with impingement and trailing edge cooling.

Silicon carbide fiber-reinforced silicon carbide CMC HPT nozzle guide vanes (NGV) were successfully designed and manufactured within the German Aerospace Center (DLR) project 3DCeraTurb. These will further be investigated experimentally to evaluate performance, damage, and lifetime for a future application in an aircraft engine. A melt infiltration-based process called liquid silicon infiltration (LSI) [45,46] was used to manufacture the vanes. The focus of this work was to adapt the manufacturing process from plate dimensions to more complex three-dimensional (3D) components. A ceramic-based design served as the basis for the development and manufacture of a new NGV geometry, considering material- and manufacturing-specific constraints. For steam protection, an EBC coating system was applied on the vanes by means of PVD processing. For the determination of thermal and mechanical properties of the material, SiC/SiC plates were characterized. The results of three material iterations are shown. To evaluate the thermal and thermo-mechanical loads in the component, computational fluid dynamics (CFD) simulations were coupled with a FEA. A stress analysis was performed and will be published separately.

Experiments at technology readiness level (TRL) 4 are planned in the Wind Tunnel for Straight Cascades Göttingen (EGG) to investigate the vane performance. The degradation mechanisms and life assessment will afterwards be evaluated in a high temperature furnace under corrosive atmosphere.

First, the manufacturing methods of SiC/SiC plate material is described. Second, the design process of the NGV including vane geometry and cooling is explained, followed by the manufacturing approach of the vanes and the characterization of plates and vanes.

The LSI process was used to manufacture SiC/SiC plates and vanes. For plate manufacturing eight layers of woven Hi-Nicalon S fabric (eight harness satin) were stacked and a batch fiber coating of BN, SiC, and pyC was applied via chemical vapor infiltration (CVI) at Archer Technicoat Ltd (United Kingdom). Up to five preforms could be coated in one batch. Then, the coated fiber preforms were infiltrated with a phenolic resin (Hexion GmbH, Germany) using resin transfer molding (RTM). The infiltration was performed at 80 °C inside a steel mold, followed by an in-mold curing at 150 °C at a pressure >16 bar. During pyrolysis of the SiC fiber-reinforced polymer material at 1450 °C in inert atmosphere, the polymer converted to a porous carbon matrix. Subsequent melt infiltration, also called siliconization, was performed in standing configuration at 1400 °C under vacuum. A silicon-boron alloy with eight atomic percent boron was used and lead to the formation of a SiSiC matrix. Residual carbon and excess silicon remained inside the matrix. The dimensions of all manufactured SiC/SiC plates were 250 × 250 × 3.3 mm³.

Three different material iterations were developed. For DS I and DS III material, one plate was manufactured each, for DS II in total three plates were manufactured. The fiber preforms for DS II and DS III material were coated in the same CVI batch. Compared to DS I the arrangement in the coating chamber was changed and the BN coating times were doubled. For DS III, a new phenolic resin, internally called MF88G, was used. Compared to the standard resin MF88, the solvent furfuryl alcohol has been replaced by ethylene glycol for the new resin type MF88G. Ethylene glycol acted as foaming agent in order to ensure a porous carbon structure, which was fully convertible to silicon containing SiC (SiSiC) during siliconization [47–49]. When MF88G was used, all manufacturing parameters remained the same. However, an additional tempering step at 240 °C was needed after curing to fully harden the resin.

The NGV was designed using a comprehensive multidisciplinary design process for cooled axial turbines [50,51]. Starting with the data from a zero-dimensional engine thermodynamic performance model, a one-dimensional meanline method was used to create the general turbine layout, defining the number of stages, the location of the blade rows and the annulus lines that determine the flow path boundaries. The initial turbine vane and blade geometries were created using a parametric 3D airfoil geometry generation tool. Special attention was paid to the NGV profile and the cooling configuration, as they had to be redesigned compared to the engine concept developed within the DLR project PERFECT [52], to account for the manufacturing limits bound to CMCs. The external flow of the first stage HPT NGV was investigated using the two-dimensional (2D) flow solver Mises [53]. A semi-empirical software tool developed for the design of cooling systems for turbine vanes and blades was then used for the thermal analysis [51]. It combines analytical and empirical methods to calculate the temperature distribution along an airfoil profile. It is based on the one-dimensional form of Fourier's law of heat conduction, which is applied to the discretized pressure and suction side contours of a 2D airfoil profile. External and internal convection as well as conduction through the pressure and suction side walls are modeled as thermal resistances in series as outlined in Fig. 1 Note that the bond coat was modeled in the calculation, but it is not shown due to the negligible influence and for better readability.

After the finalization of the external geometry, three CAD models were designed for the manufacturing process using the CAD/CAM software NX10.0 (Siemens Digital Industries Software, Germany). One for the final CMC vane, one for the final CMC vane including EBC coating systems, and one for the raw component of the CMC vane, including an allowance for grinding. For the manufacturing of the vanes, the liquid silicon infiltration process described in Sec. 2.1 was applied. The same resin as in the plate material DS III was used as carbon source. The process parameters were identically with the DS III material.

The process began with the draping and CVI fiber coating of woven fabrics (Archer Technicoat Ltd, Wycombe, UK). During this stage, shaping occurred and the final component contour was defined. For applying the CVI fiber coatings, a graphite mold was designed. Mold design process was supported by the consultation of Archer Technicoat Ltd. The coated fiber preform was then infiltrated with a phenolic resin, which acted as a carbon source. For resin infiltration and curing, RTM processing and steel molds were used. A steel insert that can take two vanes was designed and was installed in existing RTM infiltration molds. For high-temperature pyrolysis, the graphite molds from CVI processing were recycled due to cost reasons. Siliconization of the vanes was performed in standing configuration, where no molds were needed. Thereby, a silicon-boron alloy was melted and infiltrated into the vanes via capillary forces at 1400 °C under vacuum. A SiSiC matrix formed due to the reaction of the melt with the porous carbon matrix. In total six vanes were manufactured.

On three vanes, final machining was performed after siliconization. First, the vane surface was ground to achieve the required external geometry of the vanes (Protobau GmbH, Germany). Then, cylindrical cooling holes were introduced to the trailing edge via Laser MicroJet® technology, where a water jet guided laser was used for the drilling of holes (Synova S.A., Switzerland).

In the final step, an EBC was applied for the protection from oxygen and steam. Therefore, an EBC system, consisting of a silicon bond coat (Si-BC), one yttrium disilicate (Y-DS) layer and one yttrium monosilicate (Y-MS) layer, was applied using PVD processing. An 80 μm thick Si-BC was applied via electron beam physical vapor deposition (EB-PVD) technique by using a coating plant from vonArdenne. The yttrium silicate layers, each 10 μm thick, were deposited by a reactive sputtering process in an industrial size coating facility (IMPAX 1000 HT, Systec SVS Vacuum Coatings, Germany). A two-source experimental setup was used consisting of a metallic Si and Y target. The components were rotated between the targets. The oxygen inlet for reactive process had a flowrate of 12.5 sccm. After the coating process, a thermal crystallization process with temperatures up to 1250 °C was necessary. Coating thicknesses of the applied EBC coatings are much smaller (10 μm each, in total 20 μm) compared to industry standard (>150 μm), which is mostly applied via thermal spraying techniques. The PVD process used in this work currently limits the thickness for each layer to ≈10–15 μm, which will be increased in the future.

The fiber volume content of the manufactured SiC/SiC plates was calculated by the initial weight of the SiC fibers and the total volume of the finished composite in as manufactured state. Archimedes method was used for density and porosity measurements of the plates and vanes in as manufactured state according to standard DIN EN 993-1. The coefficient of thermal expansion (CTE) was measured according to DIN EN 1159-1 via high-temperature dilatometry (Netzsch GmbH, Germany). Only in-plane samples, in on-axis (0/90 deg) orientation were prepared. For each material, three ground samples with dimensions 3 × 5 × 25 mm³ were measured in inert atmosphere (argon) from 50 °C up to 1350 °C with a heating rate of 5 K/min. Laser flash analysis method [54] as per DIN EN 821-2 was used to determine through-thickness thermal conductivity (λ) of the SiC/SiC material without EBC coating between 25 °C and 1000 °C. The thermal diffusivity was measured (LFA 457, Netzsch GmbH, Germany) and the heat capacity was then calculated with the Netzsch software by using a pyroceram 9606 sample as reference material. Together with the density, λ was calculated. The sample dimensions were 12.6 mm in diameter with a height of 2.5 mm. Two samples were measured for each material with three measurements per temperature step.

The mechanical properties at room temperature were measured under uniaxial quasi-static tensile (DIN EN 658-1), compression (DIN EN 658-2) and Iosipescu-shear loading DIN EN 12289). The experiments were performed up to failure on a universal testing machine Zwick 1494 (Zwick/Roell, Germany) at a controlled cross head speed of 1 mm/min with strain gauges for strain measurement. Proportional limit stress (PLS) was evaluated using the 0.005% offset method as per ASTM C1275 [9]. The coupons for mechanical testing were cut from the plates manufactured for material DS I, DS II, and DS III in as manufactured state. The thickness of the tensile and Iosipescu specimens was 3.3 mm. The compression test specimens were ground to 2.5 mm due to geometrical restrictions of the testing device. For DS I, material specimens were cut out in 0/90 deg and ±45 deg orientation. For DS II and DS III, only 0/90 deg orientation was tested. For all specimens, the direction of testing was the x-direction. A graphical overview of the used sample geometries, loading directions and test standards, is given in Refs. [55,56], whereby only in-plane properties were measured in the present study. Four to 13 samples were tested for each material.

Polished samples of the plate and vane material were analyzed by scanning electron microscopy (SEM, Gemini Ultra-Plus, Carl Zeiss NTS GmbH, Germany) and optical microscopy (digital microscope VHX-5000, Keyence Deutschland GmbH, Germany). The AsB detector (angle selective backscatter) was used for the scanning electron microscope (SEM) microstructure images. The phase composition was analyzed using the open source software ImageJ [57]. At least ten different positions of each material were analyzed. The thickness of SiC and BN fiber coatings was measured on SEM images at two positions of the plate with a magnification of 5000× and 10,000×, respectively. For SiC coating measurements, ten different fibers were analyzed, for BN coatings at least five fibers with 10 measuring points were considered.

Surface roughness of DS I SiC/SiC plates in as manufactured state and after grinding was measured using the mobile surface measuring instrument Handysurf E-35A (Carl Zeiss Industrielle Messtechnik GmbH, Germany). Three measurements in three different positions on the plate were performed for each state. Surface roughness of the vanes after final machining was measured with a profilometer T1000 from Hommel in different positions and directions according to DIN EN ISO 4288:1998 and DIN EN ISO 3274:1998.

In order to assess the uniformity of the manufactured components, the vanes were examined via computed tomography (μCT). The scans were performed with a high-resolution μCT-System (phoenix v|tome|x L240/450, Baker Hughes Digital Solutions GmbH, Germany) consisting of a microfocus X-ray tube with 240 kV maximum accelerating voltage and a 14-bit flat panel detector (active area 4048 × 4048 pixels at 1000 μm per pixel). The μCT scans were performed with the X-ray parameters 180–200 kV/160–220 μA at an exposure time of 334 ms. A voxel size of 90 μm was achieved for the overview scans of the whole vanes and 25 μm in detailed images of the cooling holes in the trailing edge. The μCT data were analyzed and visualized with the commercial software package VGStudioMax 3.4 (Volume Graphics, Germany). Overview scans of the vanes were conducted after siliconization to check the geometry and make sure that enough allowance is present to fulfill the required external geometry during the grinding step. To rule out damage caused by machining, the vanes were scanned after grinding and after laser drilling of the trailing edge cooling holes.

The external geometry of the vanes after grinding was additionally measured with an optical GOM Atos5 3D scanning device (Carl Zeiss GOM Metrology GmbH, Germany). It uses two stereo cameras to picture a projected pattern on the surface of the vanes and calculates the position of the reflected points in comparison to the sensor position. In order to determine the position of the sensor, reference points were applied to the entire vane surface before the measurement. Lenses with medium focal lengths were used to achieve highest possible resolution while still being able to measure the entire component. With the help of the GOM software 2021, a comparison of the nominal and actual geometry was carried out and the deviations were displayed graphically.

Thermal and mechanical properties of DLR SiC/SiC plate material are presented, followed by the results of the vane and cooling design as well as the manufacturing of SiC/SiC vanes.

In the last three years, three SiC/SiC iterations were manufactured at DLR using the LSI process. The analyzed properties were summarized in Tables 1 and 2. All plates had a fiber volume content of 31% and an open porosity below 1.8%. Through-thickness thermal conductivity of all manufactured SiC/SiC iterations is comparable, as well as the in-plane thermal expansion. Thermal conductivity at 1000 °C for DS III material is 14.7 W/(mK), thermal expansion at operating temperature is 5.0 × 10⁻⁶/K. Nickel-based alloys show higher values for both properties at elevated temperatures. Mechanical properties were analyzed in-plane, either in fiber direction 0/90 deg or off-axis in ±45 deg.

The first iteration DS I was manufactured at the very beginning of the project and was analyzed in detail to generate a set of basic mechanical properties for structural calculations performed in the project. DS I shows similar values for tensile strength σ_x and elastic modulus E_x in both directions 0/90 deg and ±45 deg. This indicates an orientation-independent and matrix-dominated behavior typical for weak interface composites [56,58]. In the meantime, the fiber coating thickness of BN and SiC was increased which lead to the material called DS II. Unfortunately, fracture strain ε_x in the tensile tests of DS II was not increased with thicker BN fiber coating thickness. However, fracture surfaces showed slightly improved pull-out behavior of DS II compared to DS I. In both iterations DS I and II, the remaining carbon content was rather high (9.9 and 11.7%). For the manufacturing of DS III, a new phenolic resin composition was used in order to improve carbon conversion. The remaining carbon content was successfully eliminated to 0% in plate material. Also, the free silicon was reduced. This resulted in a ≈30% increased proportion of SiC in the matrix. Consequently, the density of DS III is the highest (2.75 g/cm³) even though the porosity is slightly higher than with DS I and DS II material. Furthermore, improved matrix stiffness and strength followed so that the elastic modulus E_x and proportional limit stress increased strongly. Fracture strain ε_x of DS III samples was much lower compared to DS II because the samples failed shortly after reaching the PLS. The fiber coating (BN and SiC) of DS II and DS III was performed in the same CVI batch. Still, the coating thickness and damage tolerance are lower in DS III. This was caused by an error in the production of the tensile specimens (water jet cutting) so that the specimens were not taken from a comparable point on the panel as with DS II. Future material characterizations will repeat these analyses and will additionally include the determination of interlaminar properties. To improve high-temperature performance, the further reduction of free silicon is a main part of ongoing research. Either a second infiltration and pyrolysis step before siliconization, or an adaption in resin composition is considered. First trials showed that values below 8% are possible. The SiC/SiC vanes in Sec. 3.3 were manufactured via DS III manufacturing route.

The temperature limit for MI SiC/SiC materials was tested in literature and was found to be 1315 °C [9,13,21,23,26,59,60]. NASA [23,40] also used comparable manufacturing processes for vane manufacturing and assumed they could withstand maximum temperatures of 1315 °C within the SiC/SiC substrate. Compared to DLR's LSI process, comparable amounts of excess silicon occurred, which is the temperature limiting factor. Therefore, we assume that DLR SiC/SiC material can also withstand temperatures up to 1315 °C. Due to cost restrictions, high-temperature and creep testing could not be performed yet, but are hoped to be considered in the future.

The ultrahigh bypass ratio geared turbofan (UHBR GTF) engine concept from the DLR project PERFECT [52] was used as a basis for this study, as detailed datasets and geometries of the turbine were accessible and published. However, the turbine inlet temperature T40 of 1828 K is lower than the targeted temperatures for future engines. Therefore, it is important to mention that the aim of the 3DCeraTurb project is to raise the test readiness level of CMCs from TRL 3 to TRL 4, i.e., to develop it further from a flat sample to a component and to realize this on a first stage high-pressure turbine stator. A new engine development around the CMC NGV was not considered yet. A summary of the resulting thermal design boundary conditions for the cooling system at EoF conditions is shown in Table 3.

The NGV geometry had to be redesigned due to various reasons, including manufacturability and the need to further investigate the aerodynamic and cooling performance in a linear cascade test rig [61]. Consequently, a ceramic-compatible profile design [50,61] was created by adjusting the leading edge radius to be greater than twice the wall thickness of the vane [20]. Additionally, the profile was optimized to minimize aerodynamic losses. Furthermore, the 2D midsection of the NGV was considered representative for the whole airfoil and expanded radially, removing any twist. The resulting profile and the EoF Mach number distribution at 50% span of the optimized vane profile can be seen in Fig. 2. Note that the semicircular trailing edge geometry is removed for the 2D simulation. However, the additional losses associated with the blunt trailing edge are modeled [53].

The cooling system was designed based on the optimized external NGV geometry. In the absence of detailed combustion chamber data, neither a pattern factor nor heat radiation was considered, as these would introduce significant uncertainty. Due to the manufacturing constraints associated with fiber ceramics, the NGV was designed as a shell, resulting in a single internal chamber. Film cooling was not considered in this initial approach, as this would damage the fibers and could lead to a decrease in mechanical properties which cannot currently be estimated. However, film cooling has proven to be a necessity as soon as higher turbine inlet temperatures or even a realistic pattern factor are assumed. Consequently, future work will consider both factors to design a more realistic cooling system for a first stage HPT NGV. Based on the current boundary conditions, only internal cooling in the form of impingement cooling was used. A metallic impingement baffle was integrated into the vane, redirecting the coolant in concentrated jets onto the inner side of the ceramic shell. The impingement hole diameters and spacings were parameterized for pressure and suction side individually to balance the temperature gradients along the profile. Cylindrical cooling holes were placed in the trailing edge to exhaust the coolant. The diameter and the spacings of these holes were parameterized to control the coolant mass flowrate.

In order to carry out a realistic potential study, the design was based on material data from literature (Table 4), which cannot be realized at DLR yet, particularly with regard to EBC coatings. Consequently, the properties actually manufactured differed from those in Table 4, which were used for the design purpose. For example, the characteristic values of the coating differ from those used for production. Therefore, the resulting design and the temperatures only serve to estimate the potential of such a ceramic NGV configuration. The thermal design boundary conditions shown in Table 3 are based on the EoF operating point.

The generic cooling configuration was integrated into a multi-objective optimization process using the DLR's in-house optimizer [64]. An evolutionary algorithm was deployed to iteratively generate novel configurations within a defined parameter space. During this iterative process, eight cooling geometry parameters, covering the internal impingement cooling configuration and the trailing edge cooling holes, underwent systematic variation. The primary objectives of the optimization process were to minimize the demand of cooling air and to reduce the standard deviation of material temperature along the surface, thereby reducing thermal stresses. The maximum tolerable temperatures of the materials served as a crucial constraint. The cooling air requirements of the final ceramic NGV concept were reduced to a fifth of that of a metallic reference design [14]. An overview of the 3D cooling configuration can be seen in Fig. 3(a). The flow path of the cooling air, which enters the NGV inside the impingement baffle on both sides, is displayed. The coolant is deflected through the holes in the baffle and impinges on the inner wall of the ceramic shell. Finally, the coolant is injected into the main flow through the holes in the trailing edge. The design temperature distribution along the external SiC/SiC surface on a simplified geometry is shown in Fig. 3(b) based on literature material properties from Table 4.

After completion of the external geometry and the cooling system, the design of the CMC component followed. The final external geometry includes the actually manufactured Si-BC (80 μm) and EBC (20 μm) coatings. The thicknesses differ from the literature target values used for the cooling design. This is because PVD processing was used for the application of the coatings in this work instead of thermal spraying techniques. For the design of the CMC component, the thickness of the coating system (100 μm) was subtracted from the external geometry. The CMC wall thickness was set to 1.9 mm. The vane span of 135 mm was determined by geometric specifications from the wind tunnel. This procedure resulted in the CAD model of the final CMC vane shown in Fig. 4(a). The vane needs a grinding step in the end, therefore, an allowance of 5 mm had to be considered. That is why an additional CAD model for the raw component was created by adding a 5 mm offset onto the external surface of the vane. For the inner surface, no grinding was foreseen, therefore, no offset was needed there. This model was used to design the CMC manufacturing tools. A third CAD model was created, representing the final vane consisting of CMC structure and EBC coating system. For easier mounting of the vane in the wind tunnel, the EBC coating system was not applied on the edges of the vane (5 mm on each side) as can be seen in Fig. 4(b). CFD and finite element calculations will base on this model. The inner and outer radii at the leading edge are indicated in the vane profile in Fig. 5(a). The design recommendations mentioned in Corman et al. [20] were considered. The smallest inner radius should at least equal two times the thickness of the part. In this work, this is the thickness of the unground component (2.4 mm). Accordingly, the radius must be ≥ 4.8 mm. Cooling holes with a diameter of 0.8 mm were implemented in the trailing edge design (see Fig. 5(b)). The hole to hole distance (pitch) was 2.32 mm and the holes were inclined by 1.2 deg towards the pressure side. The holes were exclusively implemented in the middle part of the vane, because only this area will be analyzed during wind tunnel testing.

In the beginning of the manufacturing process, a graphite mold is needed for applying the CVI-based fiber coating system due to the high temperatures in the coating chamber. The shaping of the vanes is realized in this step. Therefore, the thermal expansion of graphite was considered in the design process. For better machinability and improved temperature and geometric stability, graphite with high density was chosen. The mold consisted of three outer parts and an inner core. The correct position of the core against the outer mold was realized with the help of spacers. The mold was held together using graphite and C/C screws. The removal of the core after the coating was ensured by a suitable coating of the cores. The manufacturing of the graphite mold was carried out externally.

For resin infiltration and curing via RTM processing steel molds were used. A steel insert that can take two vanes was designed and is installed in existing RTM infiltration molds. Two vanes fit into the insert and can be infiltrated in the mold simultaneously (see Fig. 6). Comparable to the graphite mold, correctly measured dowel pins fix the inner core in the desired position. The manufacturing of the steel insert was carried out externally. For pyrolysis, the graphite molds from CVI processing were recycled due to cost reasons. The temperatures used are comparable to CVI processing, so the thermal expansion of the mold remains the same. For better accuracy, separate molds only for pyrolysis are recommended.

In total, six vanes were manufactured successfully as described in Sec. 2.2. Final machining and EBC coating were performed on three of the vanes.

Six dry fabric layers were draped into the graphite mold as shown in Fig. 7(a), leading to a fiber volume content of 29% For this first trial of producing three-dimensional components, an easy approach for the vane manufacturing was chosen, especially regarding the trailing edge. In future work, it is planned to use optimized preforming techniques, which include fibers in z-direction in the trailing edge to improve interlaminar stress resistance. Options like sewing or needling will be considered. Verrilli and Brewer et al. [36,39], e.g., used a special “Y-cloth” weaving technique.

After CVI coating of a BN/SiC/pyC fiber coating system, the final contour was defined and rigid SiC preforms (Fig. 7(b)) were available for phenolic resin infiltration via RTM (Fig. 8). After tempering inside the steel mold at 240 °C, pyrolysis was performed inside the graphite molds also used for CVI processing. The resulting porous carbon matrix was infiltrated with a silicon–boron melt, leading to the conversion of the carbon into a SiSiC matrix. The mean value of the resulting open porosity of all vanes was 4.6±1.6%, density was 2.52±0.06 g/cm³. Porosity is higher compared to plate material. Some pores are visible in the microscope image of the leading edge cross section in Fig. 9(a). Very few pores are visible in the trailing edge cross section in Fig. 9(b). Vane microstructure is comparable to this of plate material. Matrix consists of SiSiC areas surrounded by excess silicon melt (Fig. 9(b)). Due to the usage of a silicon–boron melt for infiltration, dark gray SiB₃ and black B₄C phases are formed in the matrix area, especially in excess melt regions (Si). In contrast to the plates of DS III, some remaining carbon (black) was still found in the components in some places, mostly in the leading edge region, but not in the shown trailing edge area. An updated RTM mold design with improved pressure distribution is believed to solve this issue. As can be seen in Fig. 9(a), the leading edge was correctly formed and the fabric layers were not crushed or displaced. The target radius of 4.8 mm was achieved.

After surface grinding, a smooth and reflective surface was partially produced (Fig. 10(a)). CT measurements were performed before and after grinding of the vanes (Fig. 11). Pores were visible, but grinding-induced material damage was not detected. Only slight deviations from nominal geometry (green) were found, which were quantified later. Some defects were visible to the eye and were confirmed by CT as well as 3D scanning (GOM Atos5) of the surface and comparison with the nominal geometry (Fig. 12).

On the surface of the vanes, regularly aligned grooves can be detected, which result from the diamond tooling geometry in the grinding process. These could be reduced by increasing the number of grinding cycles. The processing costs would increase, but this should be accepted. Very rough areas occur where the material was not fully dense and the porosity was opened during grinding. The roughness caused by pores cannot be removed by more complex machining. Therefore, the siliconization must be improved and the SiC/SiC material must be as dense as possible. Surface roughness was measured on as manufactured and ground DS I SiC/SiC plate material and compared to the ground surface of the SiC/SiC vanes in Table 5. SiC/SiC plate material was fully dense with an open porosity of 1.28%, which resulted in the average roughness R_a of 1.4 μm, which is suited for later EBC coating with PVD processes. The difference between the tallest peak and the deepest valley in the surface R_z was 9.3 μm. Surface roughness measurements on the vanes were performed in different positions and directions. In positions in between the grooves and where the SiC/SiC material of the vane was fully dense, R_a values up to 0.5 μm were measured. But, there is much greater average roughness of up to 3.5 μm in positions with pores present in the material. This weakens the adhesion of PVD EBC coatings and might lead to defects in the coatings. One focus of future development is to realize full densification of the SiC/SiC material not only in plate material but also in the vane geometry.

Grinding further resulted in local deviation of geometric and positional tolerances of up to + 0.28 mm, which have to be considered when positioning the vanes inside the wind tunnel.

Laser-drilled cooling holes were introduced for cooling the trailing edge. CT analyses confirmed that the holes were perfectly cylindrical with no conical shape. The diameter of 0.8 mm was correct and no material damage occurred due to laser-drilling (Fig. 13(b)). The hole to hole distance (pitch) of 2.32 mm was met. After laser drilling, an EBC coating system was applied successfully as can be seen in Fig. 10(b). The coating system consisted of a Si-BC (EB-PVD), one Y-DS layer and one Y-MS layer (PVD, reactive sputtering processes). Detailed analyses of the EBC coatings will be published elsewhere.

First results of coupled CFD/FEA simulations indicate that severe thermal stresses occur in the vane. The applied temperatures in this work (1828 K) seem to represent an upper limit for CMC vanes without film cooling. The reason is a huge temperature gradient between inner and outer surface of the CMC shell. The integration of film cooling holes will be necessary, especially if a realistic combustor outlet pattern is considered, or if turbine inlet temperatures are further increased. The fact that targeted gas temperatures exceed the expected temperature capability of EBC systems supports this conclusion [12]. In future projects, film cooling of CMC NGVs should be considered and examined. In particular, the effect on mechanical properties is not studied in detail yet. The successful implementation of laser drilled cooling holes in the trailing edge was shown in this work and is an important first step toward film cooled CMC vanes. Cooling holes for film cooling seem to be feasible. But this will also require a more complex cooling design and will make the manufacturing of the vanes more difficult.

An NGV profile was optimized for CMC manufacturing techniques in the 3DCeraTurb project based on DLR's UHBR GTF engine from the DLR project PERFECT. The NGV geometry was redesigned due to CMC specific constraints and the boundaries of the linear cascade test rig for planned wind tunnel testing. The leading-edge radius was adjusted to a ceramic-compatible design, while simultaneously optimizing the profile to minimize aerodynamic losses.

An initial cooling system relying on impingement and trailing edge cooling was designed using a low-fidelity turbine vane and blade cooling design tool. Results show that the required amount of cooling air could be significantly reduced by the introduction of a SiC/SiC NGV to the existing engine concept. It was also observed, that further increases in turbine inlet temperature, such as consideration of a pattern factor, require film cooling.

Three iterations of SiC/SiC plate material were manufactured using a resin based reactive melt infiltration process (LSI). The material was tested mechanically and thermally to generate a dataset for future structural calculations. In DS III material, the remaining carbon content was eliminated due to the use of a new phenolic resin composition. Simultaneously, the SiC proportion in the matrix increased, which led to a higher elastic modulus and proportional limit stress.

German Aerospace Center's LSI process for SiC/SiC plate manufacturing was adapted to three-dimensional components for the first time. SiC/SiC NGVs were successfully manufactured for a TRL 4 wind tunnel test to evaluate the vane performance, followed by degradation mechanisms and life assessment in a corrosive atmosphere high temperature furnace. In total, six SiC/SiC vanes were fabricated and analyzed. The leading edge radius of 4.8 mm was met correctly without any fabric layers being displaced. Surface grinding was performed on three vanes and resulted in surface roughness between 0.5 and 3.5 μm. The fewer pores there were in the material, the smoother the surface was. The implementation of laser drilled, cylindrical cooling holes with a diameter of 0.8 mm to the trailing edge worked well. Material damage caused by laser drilling was not observed. Nevertheless, the applied temperatures in this work seem to represent an upper limit for CMC vanes without film cooling. The successful implementation of laser drilled cooling holes in the trailing edge was shown in this work. Therefore, film cooling of CMC vanes seems to be feasible in future projects to increase hot gas temperatures. Nevertheless, the complexity of the cooling design process and the vane manufacturing is greatly increased by the aspect of film cooling.

This work was performed within the DLR project 3DCeraTurb and was supported by the German Federal Ministry for economic Affairs and Climate Action.

Thank you to Clemens Grunwitz for the profile design and Patrick Wehrel for the helpful discussions about engines. The authors thank Raouf Jemmali for performing CT analyses and Daniel Fricke for GOM measurements of the SiC/SiC vanes. Matthias Scheiffele is gratefully acknowledged for RTM processing, Daniel Cepli for mechanical testing and Stefan Frick for CTE measurements.

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

Acronyms

BN =

boron nitride

CAD =

computer-aided design

C/C =

carbon fiber-reinforced carbon

CFD =

computational fluid dynamics

CMC =

ceramic matrix composites

CTE =

coefficient of thermal expansion

CVI =

chemical vapor infiltration

EBC =

environmental barrier coating

EB-PVD =

electron beam physical vapor deposition

EoF =

end of field

FEA =

finite element analysis

GE =

general electric

GTF =

geared turbofan

HPT =

high-pressure turbine

LSI =

liquid silicon infiltration

MI =

melt infiltration

NGV =

nozzle guide vane

PLS =

proportional limit stress

PVD =

physical vapor deposition

pyC =

pyrolytic carbon

RT =

room temperature

RTM =

resin transfer molding

SEM =

scanning electron microscope

Si-BC =

silicon bond coat

SiC =

silicon carbide

SiC/SiC =

silicon carbide fiber-reinforced silicon carbide

TRL =

technology readiness level

UHBR =

ultrahigh bypass ratio

Y-DS =

yttrium disilicate

Y-MS =

yttrium monosilicate

μCT =

micro computed tomography

3PB =

three-point bending

Symbols

A =

surface area

Ma =

Mach number

R =

thermal resistance

T40 =

turbine inlet temperature

T41 =

rotor inlet temperature

δ =

layer thickness

λ =

thermal conductivity