Multi-objective Bayesian optimization of complex, concentrated alloys
This is a presentation on using multi-objective bayesian optimization and machine methods to guide the optimization of metal alloy compositions.
SEARCH · Search NASA
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
This is a presentation on using multi-objective bayesian optimization and machine methods to guide the optimization of metal alloy compositions.
Several experimental trials were conducted to determine optimal sintering parameters and compositions for purchased boron carbide powder using spark plasma sintering/direct current sintering. Sputtering targets used to create ablators for the National Ignition Facility have been unable to survive post processing procedures suggesting that in-house fabrication may yield better results due to extended control.
Sugarcane (Saccharum spp. hybrids) has great potential as a major feedstock for biofuel production worldwide. It is considered among the best options for producing biofuels today due to an exceptional biomass production capacity, high carbohydrate (sugar + fiber) content, and a favorable energy input/output ratio. To maximize the conversion of sugarcane biomass into biofuels, it is imperative to generate improved sugarcane varieties with better biomass degradability. However, unlike many diploid plants, where genetic tools are well developed, biotechnological improvement is hindered in sugarcane by our current limited understanding of the large and complex genome. Therefore, understanding the genetics of the key biofuel traits in sugarcane and optimization of sugarcane biomass composition will advance efficient conversion of sugarcane biomass into fermentable sugars for biofuel production. The large existing phenotypic variation in Saccharum germplasm and the availability of the current genomics technologies will allow biofuel traits to be characterized, the genetic basis of critical differences in biomass composition to be determined, and targets for improvement of sugarcane for biofuels to be established. Emerging options for genetic improvement of sugarcane for the use as a bioenergy crop are reviewed. This will better define the targets for potential genetic manipulation of sugarcane biomass composition for biofuels.
In order to increase the efficiency of light-duty gasoline engines, the Co-Optimization of Fuels and Engines (Co-Optima) initiative from the U.S. Department of Energy is investigating multi-mode combustion strategies. Multi-mode combustion can be describe as using conventional spark-ignited combustion at high loads, and at the part-load operating conditions, various advanced compression ignition (ACI) strategies are being investigated to increase efficiency. Of particular interest to the Co-Optima initiative is the extent to which optimal fuel properties and compositions can enable higher efficiency ACI combustion over larger portions of the operating map. Extending the speed-load range of these ACI modes can enable greater part-load efficiency improvements for multi-mode combustion strategies. In this manuscript, we investigate fuel effects for six different fuels, including four with a research octane number (RON) of 98 and differing fuel chemistries, iso-octane, and a market representative E10 fuel, on the load limits for two different ACI strategies: spark-assisted compression ignition (SACI) and partial fuel stratification-gasoline compression ignition (PFS-GCI) operation. Experimental results show that limits to intake boosting limit high load operation for most fuels, but high smoke emissions for high particulate matter index (PMI) fuels under SACI conditions could also be a limitation. Contrastingly, low load is limited by combustion efficiency, but these effects have more pronounced variation with fuel chemistry for PFS-GCI than with SACI. Additional, distinct effects affecting autoignition timing and peak heat release at higher speeds were identified for fuels having different low temperature heat release (LTHR) propensities for both ACI modes.
Supported catalysts are widely used in industry and can be optimized by tuning the composition and interface of the metal nanoparticles and oxide supports. Rational design of metal-metal oxide interfaces in nanostructured catalysts is critical to achieve better reaction activities and selectivities. We introduce here a new class of nanocrystal tandem catalysts that have multiple metal-metal oxide interfaces for the catalysis of sequential reactions. We utilized a nanocrystal bilayer structure formed by assembling platinum and cerium oxide nanocube monolayers of less than 10 nm on a silica substrate. The two distinct metal-metal oxide interfaces, CeO2-Pt and Pt-SiO2, can be used to catalyse two distinct sequential reactions. The CeO2-Pt interface catalysed methanol decomposition to produce CO and H2, which were subsequently used for ethylene hydroformylation catalysed by the nearby Pt-SiO2 interface. Consequently, propanal was produced selectively from methanol and ethylene on the nanocrystal bilayer tandem catalyst. This new concept of nanocrystal tandem catalysis represents a powerful approach towards designing high-performance, multifunctional nanostructured catalysts
Disclosed herein are novel families of silicon carbidonitride phosphor compositions. In certain embodiments, optimal ranges of carbon content have been identified which provide excellent luminescence and thermal stability characteristics.
Synthesis gas (syngas) can be used to synthesize a variety of fuels and chemicals. Domestic transportation and military operational interests have driven continued focus on domestic syngas-based fuels production. Liquid transportation fuels may be made from syngas via four basic processes: 1) higher alcohols, 2) Fischer-Tropsch (FT), 3) methanol-to-gasoline (MTG), and 4) methanol-to-olefins (MTO) and olefins-to-gasoline/distillate (MOGD). Compared to FT and higher alcohols, MTG and MTO-MOGD have received less attention in recent years. Due to the high capital cost of these synthetic fuel plants, the production cost of the finished fuel cannot compete with petroleum-derived fuel. Pacific Northwest National Laboratory has recently evaluated one way to potentially reduce capital cost and overall production cost for MTG by combining the methanol and MTG syntheses in a single reactor. The concept consists of mixing the conventional MTG catalyst (i.e. HZSM-5) with an alcohol synthesis catalyst. It was found that a methanol synthesis catalyst, stable at high temperature (i.e. Pd/ZnO/Al2O3) [1], when mixed with ZSM-5, was active for syngas conversion. Relatively high syngas conversion can be achieved as the equilibrium-driven conversion limitations for methanol and dimethyl ether are removed as they are intermediates to the final hydrocarbon product. However, selectivity control was difficult to achieve as formation of undesirable durene and light hydrocarbons was problematic [2]. The objective of the present study was thus to evaluate other potential composite catalyst systems and optimize the reactions conditions for the conversion of syngas to hydrocarbon fuels, through the use of mixed alcohol intermediates. Mixed alcohols are of interest as they have recently been reported to produce higher yields of gasoline compared to methanol [3]. 1. Lebarbier, V.M., Dagle, R.A., Kovarik, L., Lizarazo-Adarme, J.A., King, D.L., Palo, D.R., Catalyst Science & Technology, 2012, 2, 2116-2127. 2. Zhu, Y., Jones, S.B., Biddy, M.J., Dagle, R.A., Palo, D.P., Bioresource Technology, 2012, 117, 341-351. 3. Gujar, A.C., Guda, V.K., Nolan, M., Yan W., Toghiani, H., White, M.G., Applied Catalysis A: General, 2009, 363, 115-121.
Ferritic-martensitic (FM) steel Grade 92, with or without thermomechanical treatment (TMT), and austenitic stainless steels HT-UPS (high-temperature ultrafine precipitate strengthening) and NF709 were selected as potential candidate structural materials in the U.S. Sodium-cooled Fast Reactor (SFR) program. The objective is to develop advanced steels with improved properties as compared with reference materials such as Grade 91 and Type 316H steels that are currently in nuclear design codes. Composition modification and/or processing optimization (e.g., TMT and cold-work) were performed to improve properties such as resistance to thermal aging, creep, creep-fatigue, fracture, and sodium corrosion. Testings to characterize these properties for the advanced steels were conducted by the Idaho National Laboratory, the Argonne National Laboratory and the Oak Ridge National Laboratory under the U.S. SFR program. This paper focuses on the resistance to thermal aging and creep of the advanced steels. The advanced steels exhibited up to two orders of magnitude increase in creep life compared to the reference materials. Preliminary results on the weldment performance of the advanced steels are also presented. The superior performance of the advanced steels would improve reactor design flexibility, safety margins and economics.
We present an approach for maximizing a global utility function by learning how to allocate resources in an unsupervised way. We expect interactions between allocation targets to be important and therefore propose to learn the reward structure for near-optimal allocation policies with a GNN. By relaxing the resource constraint, we can employ gradient-based optimization in contrast to more standard evolutionary algorithms. Our algorithm is motivated by a problem in modern astronomy, where one needs to select-based on limited initial information-among $10^9$ galaxies those whose detailed measurement will lead to optimal inference of the composition of the universe. Our technique presents a way of flexibly learning an allocation strategy by only requiring forward simulators for the physics of interest and the measurement process. We anticipate that our technique will also find applications in a range of resource allocation problems.
The UK's vitrified high-level-waste (HLW) is destined to be disposed of in a deep geological disposal facility (GDF). Contact with groundwater will be the process by which radionuclides are released during 100,000+ years of disposal. Understanding the mechanisms and kinetics of glass dissolution allow models and predictions to be made on the long-term fate of the disposed HLW. Current standard laboratory scale experimental methodologies may be unrepresentative of the geological disposal environment. Aim: Provide insight into the dissolution behaviour of alkali-alumino-borosilicate UK vitrified waste in geological disposal environments. A 'cyclic vapor-static' dissolution methodology is proposed, capturing expected temporal behaviour. This will elucidate dissolution kinetics and mechanisms, optimize the base glass composition, waste loading content and generate significant, pragmatic disposal performance assessment data. Variants of three glasses to be investigated: waste-loaded MW25 and ZnCa MW28, and the ZnCa MW base. Preliminary Results: Suggests drying process decreases the chemical durability of the glass. Addition of ZnCa to UK HLW improves chemical durability. Increased waste-loading improves chemical durability. Banded structure in altered glass possibly attributed to dissolution/reprecipitation of secondary phases). Conclusion - Further Work: Cyclic vapor-static dissolution methodology may elucidate dissolution behaviour expected within a UK GDF. Potential to provide beneficial, pragmatic data for the performance assessment to implement geological disposal. Fundamental mechanisms to be understood.
There is a need in many fields, such as nuclear medicine, non-proliferation, energy exploration, national security, homeland security, nuclear energy, etc, for miniature, thermal neutron detectors. Until recently, thermal neutron detection has required physically large devices to provide sufficient neutron interaction and transduction signal. Miniaturization would allow broader use in the fields just mentioned and open up other applications potentially. Recent research shows promise in creating smaller neutron detectors through the combination of high-neutron-cross-section converter materials and solid-state devices. Yet, till recently it is difficult to measure low neutron fluxes by solidstate means given the need for optimized converter materials (purity, chemical composition and thickness) and a lack of designs capable of efficient transduction of the neutron conversion products (x-rays, electrons, gamma rays). Gadolinium-based semiconductor heterojunctions have detected electrons produced by Gd-neutron reactions but only at high neutron fluxes. One of the main limitations to this type of approach is the use of thin converter layers and the inability to utilize all the conversion products. In this LDRD we have optimized the converter material thickness and chemical composition to improve capture of conversion electrons and have detected thermal neutrons with high fidelity at low flux. We are also examining different semiconductor materials and converter materials to attempt to capture a greater percentage of the conversion electrons, both low and higher energy varieties. We have studied detector size and bias scaling, and cross-sensitivity to xrays and shown that we can detect low fluxes of thermal neutrons in less than 30 minutes with high selectivity by our approach. We are currently studying improvements in performance with direct placement of the Gd converter on the detector. The advancement of sensitive, miniature neutron detectors will have benefits in energy production, nonproliferation and medicine.
Human carbonic anhydrase II (hCAII) is a metalloenzyme essential to critical physiological processes in the body. hCA inhibitors are used clinically for the treatment of indications ranging from glaucoma to epilepsy. Targeted protein degraders have emerged as a promising means of inducing the degradation of disease-implicated proteins by using the endogenous quality control mechanisms of a cell. Here, a series of heterobifunctional degrader candidates targeting hCAII were developed from a simple aryl sulfonamide fragment. Degrader candidates were functionalized to produce either cereblon E3 ubiquitin ligase (CRBN) recruiting proteolysis targeting chimeras (PROTACs) or adamantyl-based hydrophobic tags (HyTs). Screens in HEK293 cells identified two PROTAC small-molecule degraders of hCA. Optimization of linker length and composition yielded a degrader with sub-nanomolar potency and sustained depletion of hCAII over prolonged treatments. Mechanistic studies suggest that this optimized degrader depletes hCAII through the same mechanism as previously reported CRBN-recruiting heterobifunctional degraders.
The composition of organic aerosol under different ambient conditions as well as their phase state have been a subject of intense study in recent years. One way to study particle properties is to measure the particle size shrinkage in a diluted environment at isothermal conditions. From these measurements it is possible to separate the fraction of low-volatility compounds from high-volatility compounds. In this work, we analyse and evaluate a method for obtaining particle composition and viscosity from measurements using process models coupled with input optimization algorithms. Two optimization methods, the Monte Carlo genetic algorithm and Bayesian inference, are used together with process models describing the dynamics of particle evaporation. The process model optimization scheme in inferring particle composition in a volatility-basis-set sense and composition-dependent particle viscosity is tested with artificially generated data sets and real experimental data. Optimizing model input so that the output matches these data yields a good match for the estimated quantities. Both optimization methods give equally good results when they are used to estimate particle composition to artificially test data. Finally, the timescale of the experiments and the initial particle size are found to be important in defining the range of values that can be identified for the properties from the optimization.
The methodology described in this article enables a type of holistic fleet optimization that simultaneously considers the composition and activity of a fleet through time as well as the design of individual systems within the fleet. Often, real-world system design optimization and fleet-level acquisition optimization are treated separately due to the prohibitive scale and complexity of each problem. Importantly, this means that fleet-level schedules are typically limited to the inclusion of predefined system configurations and are blind to a rich spectrum of system design alternatives. Similarly, system design optimization often considers a system in isolation from the fleet and is blind to numerous, complex portfolio-level considerations. In reality, these two problems are highly interconnected. To properly address this system-fleet design interdependence, we present a general method for efficiently incorporating multi-objective system design trade-off information into a mixed-integer linear programming (MILP) fleet-level optimization. This work is motivated by the authors' experience with large-scale DOD acquisition portfolios. However, the methodology is general to any application where the fleet-level problem is a MILP and there exists at least one system having a design trade space in which two or more design objectives are parameters in the fleet-level MILP.
This project was designed to develop acid-stable PGM-free transition metal oxide oxygen reduction reaction (ORR) electrocatalysts to meet or exceed the performance and durability of the DOE 2020 Technical Targets for platinum-group metal (PGM) free electrocatalysts from first-principles to incorporation into membrane-electrode assemblies (MEAs) for polymer electrolyte membrane (PEM) fuel cells. The planned project was to accomplish this goal with a multi-step approach: 1) materials modeling and experimental screening to identify acid-stable oxides with high ORR activity, 2) optimization of catalyst particle size and catalyst/carbon/ionomer catalyst layer composition, 3) fabrication of MEAs for performance and durability testing. During the course of this project: 1) acid-stability descriptors were developed for manganese oxides; 2) a family of acid-stable multicomponent oxides based on antimony were developed; 3) a flexible synthesis for the formation of nanocrystalline nonstoichiometric oxides was developed; 4) limited ORR activity but modest and improving oxygen evolution reaction (OER) activity was measured for multicomponent antimony oxides. The project was programmed into five technical tasks: The development of acid-stable ORR electrocatalytic oxides through 1) identification and optimization of acid-stable oxide compositions, and 2) electrochemical characterization; 3) optimization of catalyst layer composition for MEAs using identified ORR electrocatalysts; 4) MEA fabrication and performance testing to result in performance of 44 mA-cm-2 at 0.9 V vs. RHE; and 5) accelerated-stress testing of optimized MEAs. Because no acid-stable oxide was identified with the requisite activity for ORR (4.4 µA-cm-2oxide intrinsic activity at 0.9 V vs. RHE) within the time and budget allotted in Tasks 1 and 2, the project was halted at the end of phase 1, with no activity in Tasks 3-5. The research output, while not succeeding in developing ORR electrocatalysts, advanced the development of acid-stable oxide materials, showing potential for further improvement as OER electrocatalysts.
The Rotating Impeller In-Situ Gas–Liquid Reaction Method is employed for the production of Al-TiC composites. The method relies on injecting a carbon-bearing gas by means of a rotating impeller into a specially formulated molten aluminum-titanium alloy. Under the optimal conditions of temperature and composition, the gas reacts preferentially with titanium to form titanium carbide particles. The design of the apparatus, the process operation window, and the routes for forming titanium carbide particles with different sizes are elucidated.
The Savannah River National Laboratory (SRNL) evaluated a large number of Sludge Batch 6 (SB6) composition projections to support frit optimization for SB6 vitrification at the Defense Waste Processing Facility (DWPF). The evaluations discussed in this report occurred over a period of about 4 months, and included about 40 composition projections, developed by both Savannah River Remediation (SRR) and SRNL. Paper study assessments were used to evaluate the sludge composition projections with arrays of potential frit compositions using the predictive models in the DWPF Product Composition Control System (PCCS). Both nominal sludge compositions and sludge compositions with anticipated compositional variation were considered. The model predictions were used to identify candidate frit compositions for each SB6 projection and to provide some guidance to SRR on washing and blending strategies for SB6 preparation. This report presents a chronological review of this process and summarizes the findings at each stage. Following initial feedback from this work, the number of washes in Tank 51 was reduced to increase the projected sodium concentration in SB6. Analyses of predicted frit performance before and after a potential decant of Tank 40 showed that the post-decant SB6 composition would be difficult to process with any frit composition and that this scenario should be avoided. Based on the most recent SB6 projections (February 2010 SB6 composition projections developed at SRNL using the measured SB6 qualification sample composition and the revised Tank Farm washing plan), Frit 418 appears to be viable for SB6 processing at a target waste loading of 36%. A Nominal Stage PCCS Measurement Acceptability Region (MAR) assessment gave projected operating windows of 25-41% waste loading, limited by predictions of nepheline crystallization. The projected operating window is reduced to 25-38% waste loading when anticipated compositional variation is considered, again limited by predictions of nepheline crystallization. Blend points between 62 and 40 inches of Sludge Batch 5 (SB5) heel in Tank 40 had no practical impact on the projected performance of Frit 418. This assessment is made from a paper study approach only and assumes that no decant of Tank 40 will occur during SB6 processing. A decant of Tank 40 would reduce the Na2O concentration in Tank 40 to a point where it would be very difficult to target a waste loading of 36% for SB6 with Frit 418. The performance of Frit 418 with SB6 is limited by predictions of nepheline crystallization, which is a durability limiting constraint. Alternatives to Frit 418 are available that can provide equivalent projected operating windows and are limited by process related constraints (i.e., liquidus temperature predictions) rather than durability limiting constraints. A separate memorandum has recently been issued that discusses the use of Frit 418 for SB6 vitrification.1 potential differences in melt rate among these alternative frits will be discussed in a forthcoming technical report.
Abstract The compositional and structural variety inherent to oxide perovskites spawn wide-ranging applications. In perovskites, the band gap E g , a key material parameter for these applications, can be optimally controlled by varying the composition. Here, we implement a hierarchical screening process in which two cross-validated and predictive machine learning models for band gap classification and regression, trained using exhaustive datasets that span 68 elements of the periodic table, are applied sequentially. The classification model separates wide band gap materials, with E g ≥ 0.5 eV, from materials which have zero or relatively small band gaps, namely E g < 0.5 eV, and the second regression model quantitatively predicts the gap value of the wide band gap compounds. The study down-selects 13,589 cubic oxide perovskite compositions that are predicted to be experimentally formable, thermodynamically stable, and have a wide band gap. Of these, a subset of 310 compounds, which are predicted to be stable and formable with a confidence greater than 90%, are identified for further investigation. Our models are methodically analyzed via performance metrics and inter-dependence of model features to gain physical insight into the band gap prediction problem. Design maps to identify the variation of band gap with substitution of different elements are also presented.