The coupling of artificial intelligence and materials characterizations has been a center piece of almost all materials discovery efforts since 1990. Furthermore, with the constant development in probabilistic machine learning tools (i.e. machine learning models that can efficiently propagate uncertainty from inputs to outputs), the synergistic interplay between materials expert knowledge and machine learning model accuracy has been a major driver for state-of-the-art materials research. However, there still exist major challenges facing the machine learning community today when it comes to materials discovery. Be it properly representing uncertainties in the data, or leveraging materials expert knowledge in the learning phase, or actively learning from the usually-sparse, and in many cases non-comprehensive, data sets available to the machine learning expert. These challenges need to be overcome to accelerate research endeavors in the domain of material discovery. To this end, General Electric Research (GER) has been developing and applying highly robust, industry-grade tools that can do tasks like uncertainty quantification (UQ), probabilistic calibration, meta-modelling, active learning, and many more. Application and demonstration of two such tools namely Bayesian Hybrid Modeling (GEBHM) and Intelligent Design and Analysis of Computer Experiments (IDACE) are presented in this paper in the context of materials modeling and discovery.

Thermochemical processing of sustainable biomassBiomass paired with carbon captureCarbon capture and storage has the potential to provide 1/7th of the emission mitigations necessary for the world to meet net-zero targets by 2050 while also providing carbon-negative fuel, electricity, and economic development. This work has developed coating solutions to mitigate the hot corrosion that occurs when biomassBiomass is processed in boilers, gasifiers, and other thermal conversion equipment, in addition to coating solutions to mitigate solid particle erosion that occurs when steam turbines are used for aggressive load following. Analysis of 66 hot corrosion coatings and 75 solid particle erosion coatings reveals unique mechanisms that enable significantly improved performance relative to conventional coatings used today without increasing material costCost. Implications of this work to fuel flexibility, process efficiencyEfficiency, and lessons learned utilizing ICME will also be discussed. This material is based upon work supported by the Department of EnergyEnergy under Award Number DE-FE0031911.

This paper focuses on the development of process parameters of additive machines to manufacture high-quality parts using intelligent experimental design techniques. Direct metal laser melting modality of additive manufacturing is of focus. Intelligent design of experiments protocol is used in conjunction with highly accurate Gaussian Process based surrogate models of the microstructural and mechanical quantities of interest as functions of the process parameters. Results from the modeling approach and the experimental design methodology are presented, and the sharp savings using the probabilistic approaches on the cycle time is highlighted.

At GE Research, we are combining “physics” with artificial intelligence and machine learning to advance manufacturing design, processing, and inspection, turning innovative technologies into real products and solutions across our industrial portfolio. This article provides a snapshot of how this physical plus digital transformation is evolving at GE.

A novel framework including experimental and model-based techniques saves time and enables the introduction of new alloys for additive manufacturing. This article describes the first phase of the probabilistic machine learning framework that was successfully demonstrated to rapidly define optimum parameter sets for commercial high-temperature nickel superalloys, as well as to guide alloy design and selection for compatibility with laser powder bed fusion additive manufacturing.

Component design, maturation and material development for sCO2 turbomachines in the 10MW to 500 MW range have been identified to be in the critical path for the successful implementation sCO2 power plants. However, the performance and life limiting mechanisms of these components in high pressure, high temperature sCO2 environment are not well understood. These mechanisms are governed by multi-scale coupled physics interactions combined with strong perturbations in performance variables around the critical point. In this paper, some of the work done by GE, under U.S. DOE PREDICTS program, in developing performance and life prediction models for Hybrid Gas Bearing (HGB) and Dry Gas Seal (DGS) is presented. From a system perspective, HGB can provide substantial benefits (~3 to 5%) to the modular CSP operation, while use of DGS limits the leakage losses (~2-3% improvement in cycle efficiency for 10 MW scale) as well as the windage losses in the generator (increase in generator efficiency by ~8%). Multi-scale coupled physics models to predict dynamic performance of HGB and DGS are developed. The models try to capture sCO2 specific phenomena like sonic transitions, possibility of phase change, flow induced and rotordynamic instabilities and large perturbations in apparent heat transfer coefficients due to Leidenfrost effect. The output of performance model is fed into 3D fracture mechanics based life prediction framework. Test campaigns to characterize corrosion of Nickel base super alloys in sCO2 environment are conducted and chemical kinetics models are built. LCF behavior of Ni base super alloys in high pressure, high temperature sCO2 is also being investigated using a novel experimental setup. Bayesian hybrid probabilistic models are developed to quantify uncertainty in multi-physics models and to validate models with statistical confidence. This strongly coupled performance and life prediction framework is a valuable tool to design a wide variety of sCO2 turbomachinery components and heat exchangers, analyze their performance in supercritical and trans-critical mission cycles and predict their life for long term durability of sCO2 turbomachines.

U.S. Department of Energy (DOE) has recently sponsored research programs to develop megawatt scale supercritical CO2 (sCO2) turbine for use in concentrated solar power (CSP) and fossil based applications. To achieve the CSP goal of power at $0.06/kW-hr LCOE and energy conversion efficiency > 50%, the sCO2 turbine relies critically on extremely low leakage film riding seals like dry gas seal (DGS). Although DGS technology has been used in other applications before. making it successful for stringent conditions of an sCO2 turbo-expander is challenging. This paper presents results from a multi-scale coupled physics model that predicts the performance of DGS under a typical sCO2 turbine mission cycle and addresses some of the risks specific to operation in sCO2. Real gas equations of state are incorporated in the models to capture large discontinuities in fluid properties close to the critical point. A novel experimental setup is developed to observe and characterize transition of CO2 through liquid-vapor and supercritical phases. Coupled fluid-structure-thermal interaction model investigates the effect of aerodynamic and thermal perturbations on the structural and rotordynamic instabilities. Dynamic instabilities arising from sonic transition in thin sCO2 film of DGS pose additional challenges while the large surface roughness changes due to sCO2 corrosion warrant further design considerations. Effectiveness of features like spiral grooves in converting fluid momentum into pressure rise in the thin film and also in achieving local flow reversals is investigated. Effect of various design features on the optimal performance is quantified and insights for a successful DGS operation in a sCO2 turbomachine are provided.

Cyclic oxidation behavior of β NiAlCr coatings was investigated at 1150°C with emphasis on effective Pd+Hf and Pt+Hf co-doping composition and the concomitant phase transformations and interdiffusion of Pd and Pt into the substrate. Coatings with compositions in the range of Ni–(33–39)Al–5Cr–(2–5)Pd/Pt were deposited on René N5 single-crystal substrates and subjected to long-term cyclic oxidation. The Pt+Hf- and Pd+Hf-modified overlay coatings exhibited comparable oxidation kinetics and matched the range of oxidation behavior exhibited by the standard Pt-aluminide coating. In particular, additions of about (2–5)Pd+(0.15–0.3)Hf and (2–5)Pt+(0.15–0.3)Hf among the studied composition range exhibited long lifetimes of >2000 oxidation cycles. A positive synergistic effect between Pd–Hf and Pt–Hf was clearly observed when comparing the lifetimes of β NiAlCr-Pd and β NiAlCr-Pt quaternary coatings with the quinary coatings. Transformations from β→3R+7R-martensites at low temperature and β→γ′→γ+γ′ at high temperature occurred in both Pd- and Pt-modified coatings due to Al depletion to substrate and surface oxide. A greater depletion of Pd to the substrate compared to Pt was observed at 1150°C compared to 1100°C; the interdiffusion coefficients were consistent with those measured in NiAl–NiAl+Pt/Pd diffusion couples. These results are very encouraging in supporting the exploration of Pd-modified coating supplanting the currently used, expensive Pt-based coatings.

A combinatorial approach was used to investigate the effects of higher-order additions to NiAl on the oxidation, phase stability and interdiffusion behavior of the coatings. Overlay coatings with compositions (in at.%) in the range Ni–38Al–5Cr–(0.1–0.5)Hf–(2–8)Pd and Ni–36Al–5Cr–(0.1–0.5)Hf–(2–8)Pt were deposited on René N5 single crystal substrates via ion plasma deposition (IPD). Cyclic oxidation experiments on ten different Pd+Hf- or Pt+Hf-modified NiAlCr coatings at 1100°C revealed similar weight gain and oxide scale thickness. While the thickness of alumina scales formed on Pd- and Pt-modified coatings were comparable, a higher density of oxide spalls were observed in the former. High-temperature transformations from β→γ′→γ+γ′ occurred during extended exposure in both Pd- and Pt-modified coatings owing to Al depletion to substrate and surface oxide. On cooling, Al-depleted β-NiAl transformed to two martensite phases, 3R and fine-twinned 7R. Rumpling or surface roughening occurred for some coating compositions. Microprobe measurements showed that the difference in penetration depths of Pd into the substrate was not as large as expected from the higher interdiffusion coefficient of Pd in β-NiAl and γ-Ni.

Alloys in the Mo-rich corner of the Mo-Ti-Zr-C system have found broad applications in non-oxidizing environments requiring structural integrity well beyond 1273 K (1000 °C). Alloys such as TZM (Mo-0.5Ti-0.08Zr-0.03C by weight %) and TZC (Mo-1.2Ti-0.3Zr-0.1C by weight) owe much of their high temperature strength and microstructural stability to MC and M2C carbide phases. In turn, the stability of the respective carbides and the subsequent mechanical behavior of the alloys are strongly dependent on the alloying additions and thermal history. A CALPHAD-based thermodynamic modeling approach is employed to develop a quaternary thermodynamic database for the Mo-Ti-Zr-C system. The thermodynamic database thus developed is validated with diffusion multiple experiments and the validated database is exercised to elucidate the effects of alloying and thermal history on the phase equilibrium in Mo-rich alloys.

Planar solid oxide fuel cells with yttria-stabilized zirconia electrolytes typically operate at temperatures in the range of 700–850 °C. The maximum temperature is limited by the use of ferritic stainless steel interconnects which offer significant advantages such as low cost and high thermal and electronic conductivity over traditional ceramic interconnects. However, these alloys rely on the formation of a protective chromia scale for oxidation resistance that results in an increase in ohmic resistance and can volatilize leading to a loss of cathode catalytic activity. To better understand the oxidation behavior of chromia forming ferritic stainless steels, thermodynamic modeling was performed in conjunction with experimental oxidation testing and empirical kinetic evaluation. The phase stability and oxidation behavior of the Fe–Cr–O ternary system were assessed using thermodynamic calculations. Calculated ternary phase diagrams were validated against the experimental oxidation data of Fe–20Cr and Fe–18Cr ferritic stainless steels, GE-13L and AL-441HP, respectively. Results indicate that the use of accelerated testing, such as exposing the system to higher temperatures, can lead to changes in phase equilibria and the oxidation kinetics of the alloys. Through combined thermodynamic assessment and controlled oxidation experiments, the oxidation behavior of high-chromium ferritic stainless steels is presented and discussed.

Diffusional analyses were performed to understand the oxidation at 1300 °C of a multiphase Mo-13.2Si-13.2B (at.%) alloy. During oxidation, a protective glass scale formed with an intermediate layer of (Mo+glass) between the base alloy and external glass scale. Compositional profiles across the (Mo+glass) layer and the external glass scale were determined, and interdiffusion fluxes and effective interdiffusion coefficients for the various components were determined by using “MultiDiFlux” software. The motion of the (alloy/Mo+glass) and (Mo+glass/glass) interphase boundaries after passivation was examined. Additionally, vapor-solid diffusion experiments at 1300 °C were carried out with single-phase Mo3Si and T2 specimens in addition to a multiphase Mo-10Si-10B (at.%) alloy. These specimens were exposed to vacuum to induce silicon loss resulting in the formation of a Mo layer. An average effective interdiffusion coefficient of Si in Mo at 1300 °C was estimated from the Mo3Si-vapor couple to be in the order of 8×10−17 m2/s.