A team of researchers at the U.S. Army Research Laboratory has detailed industry progress toward developing an optimal sand-phobic thermal barrier coating for gas turbine engines in military and commercial fixed- and rotary-wing aircraft.
With some commercial and military aircraft often operating over sandy regions, existing engine inlet particle separators are not entirely efficient in filtering fine sand particles below 75 microns, affecting the durability of turbine blade thermal barrier coatings and overall performance of the aircraft engine.
The lab aims to make further improvements on the coatings' tolerance to these and similar foreign particulates that may be contained in the intake air. Such coatings are used in the high-temperature sections of the engine.
R&WI is republishing the research laboratory's full paper discussing modeling tests and current findings from these efforts below. The paper was published as part of the American Society of Mechanical Engineers Turbo Expo 2017 proceedings. The event was held in Charlotte, North Carolina, June 26 to 30, 2017.
Over the years, through consistent efforts in advanced materials development, scientists and engineers have significantly improved the efficiency and power densities of gas turbine engines by increasing the turbine inlet temperature using high-temperature capable blade materials and coatings. Traditionally, foreign object damage (FOD) has been the primary concern in aviation and tank automotive gas turbine engines. Current state-of-the-art inlet particle separators equipped in gas turbine engines can remove most of the larger sized particles (greater than ~75 µm) from the inlet air. This, in combination with high-quality, erosion-resistant coatings on the compressor sections, has significantly reduced the risk from FOD in the cold sections of the engine. However, fine micron-sized particles (consisting of ash, soot, dust, and/or sand) can still pass through the engine and create problems in the hot-sections, where the operating temperatures of modern gas turbine engines typically surpass the melting point of many of these contaminants. The resulting impact, adhesion, melt infiltration and/or glassy solidification of these small particles can significantly damage the hot-section components within the gas turbine engine.
High pressure gas generator turbine blades typically experience inlet temperatures of 1300-1600ºC, requiring TBCs to be used as thermal insulators on the turbine blades to reduce blade cooling needs. Current state-of-the-art gas turbine blade technology has four layers in the high temperature turbine blade coating system that consists of different materials with specific properties and functions. These layers are (i) the Ni-based superalloy substrate, (ii) bond coat, (iii) thermally grown oxide (TGO) (Alumina, Al2O3), and (iv) the ceramic top-coat (YSZ, Yttria stabilized Zirconia). These discrete multilayer materials have shown vulnerability to intake flows that contain sand, dust, and/or fly ash, which adhere and react at high temperatures and cause premature component failure through both impingement and combined mechanical-thermal-chemical attack. Currently there are no valid fundamental physics based models to describe the complex phenomena of molten particulate impingement, adhesion and melt infiltration into the ceramic TBCs, followed by subsequent phase change to glassification and underlying chemical oxidation leading to failure/delamination of the coating layers. A collaborative research effort at the U.S. Army Research Laboratory (ARL), with the U.S. Army Aviation and Missile Research, Development, and Engineering Center (AMRDEC), the U.S. Navy Naval Air Systems Command (NAVAIR) and the National Aeronautics and Space Admininstration (NASA), has been initiated to conduct research to understand the complex fine sand particulate interaction mechanisms with thermal barrier coatings and bulk material substrates.
Field-returned engine hardware from Army rotorcraft from Southwest Asia (SWA) shows a significant number of occurrences of sand-induced damage. Army rotorcraft engines are being pulled out of service at a substantial knockdown on the design life. Commercial jet engines flying over dust polluted urban areas in some parts of Far East Asia and South Asia also find sub-micron particles in a molten/semi-molten state after passing through the combustor. Novel TBCs which are also resistant to sand-induced surface degradation can greatly reduce vulnerability in hostile particle-laden environments and increase fuel efficiency of gas turbine engines for both the military rotorcraft and civil aircraft fleet. This paper outlines the continued research efforts at ARL to develop novel TBCs that are sand-phobic with excellent resistance to impact wear and CMAS adherence, together with excellent thermal strain tolerance and very low thermal conductivity properties for improved performance of future gas turbine engines.
Technical challenges involve modeling complex thermo-mechanical interactions of mechanics of particle impact and adherence. The particles can be fully or partially molten or remain in their solid state after flowing through the combustor depending upon their size, morphology, geological location and composition (sand melting point varies between ~ 1050 – 1250 °C). Since the particle impact on turbine blades is much more complex with multi-physics phenomena including thermo-mechanical interactions and chemical reactions, a step-by-step canonical approach is being taken at the U.S. Army Research Laboratory to study the blade surface-particle interaction. Intrinsic damage evolutions are caused by thermo-mechanical fatigue characteristics of the coating and substrate materials and their interface bonding strengths. Extrinsic damage evolutions are influenced by particle impact causing wear and erosion of TBC. Further, the molten sand deposition, infiltration into the micro-gaps of TBC and the solidification causes the TBC to lose its thermal strain tolerance leading to typical horizontal cracks and subsequent delamination of TBC.
As a first step, a spherical particle (quartz sand composition) impacting a flat target made of typical turbomachinery material was developed using a nonlinear dynamic impact analysis finite-element code, LS-DYNA. Researchers in the past have developed semi-empirical erosion models by studying the rebound characteristics of sand particles impacting turbomachinery materials. In this research effort, a micro-structure finite-element model similar to the micro-structure of TBC on a substrate has been developed to study the particle impact physics including particle kinematics, propensity of particulate adhesion to the surface and buildup in addition to local damage on target material surfaces. The meso-scale finite-element model is being extended to further develop simulation of liquid sand particle (molten sand) impacting a thermal barrier coated turbine blade surface as the next step. Further, the model was modified to simulate multiple sharp-edged fine sand particles.
Non-linear dynamic impact analysis software, LS-DYNA was used to simulate multiple sand particle impacts on a micro-structure model consisting of typical interface layers. In this simulation, the irregular, sharp-edged sand particles were assigned silica quartz material properties, and were assumed rigid through the use of Mat_Rigid material model available in LS-DYNA. Material data were obtained from published references and used for the four layers of blade materials. Elastic-plastic material model, called Mat_Plastic available in LS-DYNA solver was used for all four layers. Surface_to_Surface contact algorithm available in the software was utilized for each particle-surface interaction contacts with columnar TBC that has inter-columnar gaps of 2 µm. Tied_Contact algorithm (available in LS-DYNA) was used for the interfaces between layers. The particle velocities and plastic strain build up on the TBC were tracked through the simulation runs. For each particle, initial velocity of 200 m/s (Mach number ~0.6 at ambient temperature) with an impact angle of 45 degrees was used to run the impact simulation. This simulation method will be further developed to analyze molten sand particle impact interactions with target material properties at gas turbine relevant temperatures and a range of particle velocities and impact angles.
The numerical solution provides stresses, strains, deformations, etc., including velocity history during particle-surface impact interactions. The explicit computations were terminated after the particles reached almost steady rebound velocities. These results show that a there is significant plastic strain build up accumulation on the TBC surface during multiple particle impact. The plastic strain build up is due to the transfer of kinetic energy from multiple particles being imparted to the TBC surface. Due to multiple particle impact, considerable amount of particle kinetic energy is absorbed by the TBC since the rebound velocity for each particle is reduced to about 25% of initial velocity. Although the comparison with spherical particles is not shown in this paper, we have also observed that the impact energy absorption resulting from the irregular shaped particulates is higher leading to a deeper plasticity zone.
In order to verify whether a fine sand particle can melt during its flight through the combustor, a multi-physics finite-element simulation using Multiphysics v5.0 (COMSOL, Inc.) was carried out. Multiphysics is a finite-element based simulation program that solves the coupled partial differential equations (of different physio-chemical phenomena) simultaneously. The first order model helps us to understand the amount of thermal exposure within the combustor leading to a molten or semi-molten condition of the sand particulate. The COMSOL simulation is a 2D axisymmetric model, with the diameter of the spherical particle varying between 100 nm – 500 µm. The initial particle core temperature is assumed to be 425 °C (typical air flow temperature at the exit of final stage of compressor in a rotorcraft gas turbine engine) and the outer wall temperature of the particle is ramped to 1400 °C. Particles smaller than 100 µm will fully melt within 10 ms of exposure. For a particle in the turbulent flow within the combustor, this is sufficient exposure time to fully melt.
The test sand properties values are based on published data for the test sand, AFRL 02. AFRL 02 test sand is a commercially-available mineral-based synthetic sand (34 wt.% quartz, 30 wt.% gypsum, 17 wt.% aplite, 14 wt.% dolomite, and 5 wt.% salt). This material was developed to mimic the effects seen on fielded engine hardware in a laboratory test environment. The resulting synthetic sand has a non-spherical morphology, with a wide particle distribution (0.5 – 150 µm; mean particle diameter ~ 15 µm), and the melting temperature (1140 °C) falls well within the range of natural sands (1050 – 1250 °C).
High-speed imagery of sand impacting on the surface of a TBC-coated Inconel 718 one inch disk confirms the molten nature of the sand contaminants in the combusted gas airflow. The image sequence shows the flight path of a sand particle, its impact splat on the sample surface, and subsequent plume of material resulting from the impact. This confirms the first-order melting model results.
In addition, particle velocity measurements were carried out using a Dantec two-component laser Doppler velocimetry system during thermal cycling testing of coated samples with sand ingestion using the hot particulate burner rig facility at ARL. Molten sand particle velocity measurements were obtained at the exit of the jet burner in the center of the flame, and at a point close to the surface of a coated Inconel circular disk sample in the flame. From the data of the free stream impacting velocities of molten sand particles and the rebound velocities of molten liquid sand particles close to the sample surface, it is evident that most of the impacting molten sand particles lose their kinetic energy resulting in deposition onto the TBC surface.
In order to determine the vulnerable impact zones inside a typical turbine nozzle passage, computational fluid dynamics simulations were performed and then particle trajectory simulations were carried out. From these simulations, vulnerable zones of the blades where sand accumulation could occur, were identified. The particle trajectory and wall interactions were simulated using a fixed stator/rotor cascade geometry. The flow inlet conditions studied were as follows, inflow velocity of 82.3 m/s, temperature of 1400 °C, with pressure at 294 psi, and the geometry was positioned sufficiently downstream and away from boundaries as to avoid outflow numerical disturbances. The concept model is based on earlier works using Eulerian- Lagrangian particle injection methodology and it is coupled with the surrounding carrier gas-phase turbulent environments. The flow turbulence is modeled via a Reynolds Average Navier Stokes (RANS) approach and adaptive mesh refinement (AMR) that resolves the flow field in local intensity regions accurately and efficiently. At the blade surface, a law of the wall model is utilized at a location of y+ = 50 to avoid the excessive cost of resolving the viscous region. The surrogate model assumes the sand particle is in the molten state, as it leaves the combustion chamber, and prescribes molten silicon dioxide (SiO2) physical transport properties.
Particles that have sufficient kinetic energy to deposit on the blade surface forming a film within the regions of viscous flow near the wall. The simulations show a higher rate of deposition on the stator row blades in comparison to the rotor blades. In general, the particle deposition is higher on the leading edge and the pressure surface.
ARL spray technologies were used to tailor dissimilar interfaces for densification of rigid particles during spray deposition processes (SPD, SPPS, HVAF, HVOF and APS) to generate the graded ceramic-metal and ceramic-composite material to be characterized. Simultaneous deposition of ceramic and metallic materials can be achieved through the use of two or more powder feeders. Post-processing of functionally-graded material (FGM) deposits via sintering and HIPing can be used to remove any porosity, increase the density and harden the outer ceramics for abrasion resistance. Coating constituent materials were selected based on evaluations with respect to thermal conductivity, phase stability, adhesion, strain tolerance, impact/erosion resistance, and CMAS-interaction. These coatings were deposited onto Inconel 718 one inch circular disks using Air Plasma Spray (APS), Solution Precursor Plasma Spray (SPPS), and Electron-Beam Physical Vapor Deposition (EB-PVD). The coatings were then evaluated using a high temperature button cell rig followed by a hot burner rig facility at ARL. Typically, the coated samples were tested for three alternating thermal cycles of 3 minutes each (hot/cold) with high temperature maintained at 1300 °C (above sand melting point of 1140 °C) and at 1g/min sand ingestion rate (AFRL 02 sand). AFRL 02 and AFRL 03 represent the state-of-the-art universal “average sand” developed by AFRL for conducting standardized sand ingestion tests for validation of gas turbine components’ resistance against CMAS attack. AFRL 02 is generally comprised of smaller grains and used for bench-level and component-level testing. AFRL 03 exhibits larger grain distribution and is used for engine-level testing. Based on material morphology data after thermal cycling tests with sand particle ingestion, the performances of the coatings were determined and ranked. In addition to coating types, superfinishing of the coated surfaces were considered in the selection process.
From the candidate coatings, the top 11 promising coatings were selected and applied to 11 turbine nozzle doublets of a typical high pressure turbine nozzle ring. This turbine nozzle ring consisted of 12 nozzle doublets, with the remaining nozzle doublet used as the baseline (Reference nozzle doublet) with the coating as manufactured by the OEM. ARL-NASA and two coating vendors supplied the selected 11 different coatings/surface treatments for the eleven turbine nozzle doublets, as listed in Table-1. The turbine nozzle ring with the eleven candidate coatings was then subjected to full engine test with mission relevant sand ingestion using the AFRL03 sand.
Post-test examination of nozzle vanes included optical microscopy, confocal microscopy, scanning acoustic microscopy, CT-scan, and SEM to quantify the damage caused by CMAS on the various nozzle vanes with different types of coatings. The post-test image of the reference nozzle vane doublet with the OEM supplied coating is shown. This picture shows some micro sand particles entrained in the CMAS build-up on the vane surface.
Scanning acoustic microscopy (SAM) analysis is a technique that utilizes ultrasonic waves to non-destructively characterize sub-surface defects present within a material specimen or component. For the current study, the SAM was operated in reflection mode to generate a C-Scan (horizontal section), in which the elastic sound waves propagated onto the sample by the transducer are reflected back when there is a change in impedance, which is indicative of a change in acoustic properties due to defects or material property changes. The presence of CMAS was expected to generate an echo or reflection due to the differences in composition, as well as the relatively porous structure of the solidified CMAS.
Analysis of the post-test vanes was performed using a Sonoscan Gen6 SAM, operated with a 15 MHz transducer and deionized water as the transducing medium. Subsurface imaging of the nozzle vanes was conducted by sequentially gating the reflected echo (ultrasound signal reflected back by sample) using a time interval of 1 µs. Sub-surface images of an air-cleaned nozzle vanes (little to no CMAS present) and the ARL-NASA-05 specimen are shown in Fig. 22. In the SAM analysis presented in Fig. 22(a) and (b), the sample was oriented so that the leading edges of the vanes were facing the transducer, that is, the orientation of the images is the same as that observed in the optical images in Fig. 17 and 21. Only one sub-surface slice is presented here for brevity. It can be seen that in addition to CMAS adhered on the outer walls, the sub-surface air channel is also significantly clogged by sand. A similar analysis was conducted with the pressure side of the nozzle vanes oriented facing the transducer, as shown in Fig. 21 (c) and (d). Again, it can be seen that significant amounts of sand are seen throughout the image, corresponding to sand being present on both the surface and subsurface (i.e., cooling channels) of the airfoil. The presence of CMAS on the exterior and interior of nozzle vanes can have significant adverse effects on the aerodynamic and thermal performance of the component. The presence of sand within internal cooling channels indicates that the bleed air used was substantially contaminated by the sand laden flow.
This paper was written and published by Muthuvel Murugan, Anindya Ghoshal, Michael Walock, Andy Nieto, Luis Bravo, Blake Barnett, Marc Pepi and Jeffrey Swab of the U.S. Army Research Laboratory; Robert Tyler Pegg and Chris Rowe of NAVAIR; Dongming Zhu of NASA Glenn Research Center; and Kevin Kerner of the U. S. Army Aviation and Missile Research Development and Engineering Center Aviation Development Directorate.