Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “prompt engineering”

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.

At least 73 records · Page 4
All sourcesNASA NTRSDOE OSTI
NASA NTRS1993

An expert system shell for inferring vegetation characteristics: Interface for the addition of techniques (Task H)

All the NASA VEGetation Workbench (VEG) goals except the Learning System provide the scientist with several different techniques. When VEG is run, rules assist the scientist in selecting the best of the available techniques to apply to the sample of cover type data being studied. The techniques are stored in the VEG knowledge base. The design and implementation of an interface that allows the scientist to add new techniques to VEG without assistance from the developer were completed. A new interface that enables the scientist to add techniques to VEG without assistance from the developer was designed and implemented. This interface does not require the scientist to have a thorough knowledge of Knowledge Engineering Environment (KEE) by Intellicorp or a detailed knowledge of the structure of VEG. The interface prompts the scientist to enter the required information about the new technique. It prompts the scientist to enter the required Common Lisp functions for executing the technique and the left hand side of the rule that causes the technique to be selected. A template for each function and rule and detailed instructions about the arguments of the functions, the values they should return, and the format of the rule are displayed. Checks are made to ensure that the required data were entered, the functions compiled correctly, and the rule parsed correctly before the new technique is stored. The additional techniques are stored separately from the VEG knowledge base. When the VEG knowledge base is loaded, the additional techniques are not normally loaded. The interface allows the scientist the option of adding all the previously defined new techniques before running VEG. When the techniques are added, the required units to store the additional techniques are created automatically in the correct places in the VEG knowledge base. The methods file containing the functions required by the additional techniques is loaded. New rule units are created to store the new rules. The interface that allow the scientist to select which techniques to use is updated automatically to include the new techniques. Task H was completed. The interface that allows the scientist to add techniques to VEG was implemented and comprehensively tested. The Common Lisp code for the Add Techniques system is listed in Appendix A.

Harrison, P. Ann↗
NASA NTRSDate not supplied

A Comparison of Model Predictive Control Architectures for Application to Electrified Aircraft Propulsion Systems

As electrified aircraft propulsion (EAP) systems continue to mature, more sophisticated hardware and software are being developed to balance operations among electric machines and gas-turbine engines. In hybrid-electric propulsion systems, the increased complexity resulting from integrating turbine-engine shafts with electric machines necessitates control methodologies to account for various physical domains. Ideal controllers for hybrid-electric engines manage systems, subsystems, and their interactions in a coordinated fashion, able to account for safety and performance goals while being computationally efficient. In a previous work, linear model predictive control (MPC) schemes were implemented in centralized and distributed frameworks on a nonlinear turbofan engine model as a proof of concept. However, these schemes were not evaluated for computational complexity, prompting further study. The research presented here develops hierarchical MPC schemes to reduce the computational burden of the previous MPC schemes. A two-tier framework is implemented, where a slower sampling MPC controls electric machines and determines fan-speed tracking goals for a faster sampling controller, which is either a MPC or a proportional-integral (PI) controller. The proposed designs are compared to the centralized MPC investigated previously, and performance is measured via fan speed tracking error, energy storage state-of-charge, and computation time. Results reveal that the hierarchical MPC scheme employing a lower-level PI controller improves computation time while maintaining comparable tracking and state-of-charge regulation to the centralized scheme.

model predictive control↗
NASA NTRSDate not supplied

A Comparison of Model Predictive Control Architectures for Application to Electrified Aircraft Propulsion Systems

As electrified aircraft propulsion (EAP) systems continue to mature, more sophisticated hardware and software are being developed to balance operations among electric machines and gas-turbine engines. In hybrid-electric propulsion systems, the increased complexity resulting from integrating turbine-engine shafts with electric machines necessitates control methodologies to account for various physical domains. Ideal controllers for hybrid-electric engines manage systems, subsystems, and their interactions in a coordinated fashion, able to account for safety and performance goals while being computationally efficient. In a previous work, linear model predictive control (MPC) schemes were implemented in centralized and distributed frameworks on a nonlinear turbofan engine model as a proof of concept. However, these schemes were not evaluated for computational complexity, prompting further study. The research presented here develops hierarchical MPC schemes to reduce the computational burden of the previous MPC schemes. A two-tier framework is implemented, where a slower sampling MPC controls electric machines and determines fan-speed tracking goals for a faster sampling controller, which is either a MPC or a proportional-integral (PI) controller. The proposed designs are compared to the centralized MPC investigated previously, and performance is measured via fan speed tracking error, energy storage state-of-charge, and computation time. Results reveal that the hierarchical MPC scheme employing a lower-level PI controller improves computation time while maintaining comparable tracking and state-of-charge regulation to the centralized scheme.

model predictive control↗
NASA NTRS1996

Fullerene Propellant Research for Electric Propulsion

The large mass, low first ionization potential and large electron impact ionization cross-section make Buckminsterfullerene (carbon 60) potentially attractive as an ion engine propellant. It has the potential for significant increases in engine efficiency over that obtained with xenon at specific impulses less than 3 000 s. One problem encountered in fullerene ion engines has been dissociation of the propellant. Previously this was attributed to thermal decomposition due to operation of the ion engine at temperatures greater than 1073 K. However, during tests conducted at temperatures lower than 1073 K fullerene fragmentation was still observed. This prompted an investigation to determine if the dissociation was still due to thermal effects or if it was due to collisional processes in the discharge chamber.

fullerenes↗
DOE OSTI2026

Leveraging large language models to address data scarcity in machine learning for graphene synthesis

Machine learning in experimental materials science faces significant challenges due to the scarcity of data, which are costly and time-consuming to generate, particularly when relying on in-house experiments. Literature data mining offers a potential solution but introduces issues like mixed data quality, inconsistent formats, and non-uniform reporting of synthesis parameters, resulting in partially missing and heterogeneous features across the dataset. Here, we propose data imputation and feature engineering methods that employ pre-trained large language models (LLMs) to enhance machine learning performance on scarce, heterogeneous datasets, demonstrated on graphene CVD synthesis data and the ML-HydPARK hydrogen storage dataset. GPT models perform data imputation via tailored prompting and semantic normalization of inconsistently reported features through embeddings, for example, to harmonize the complex nomenclature of CVD substrates. Beyond yielding more diverse and richer feature representations than traditional methods such as K-nearest neighbors (KNN) and Multivariate Imputation by Chained Equations (MICE), LLM-based data imputation is evaluated against dataset characteristics and prompting strategies. We vary the level of autonomy granted to the LLM, from generic prompting that leverages pre-trained knowledge for autonomous data generation to data-informed prompting that constrains outputs using target-specific information, and demonstrate which level of autonomy yields superior imputation performance across datasets and feature types. The proposed data engineering methods markedly improve downstream performance; for example, in graphene layer number classification using a support vector machine (SVM), binary accuracy increases from 39% to 65% and ternary accuracy from 52% to 72%. Fine-tuning experiments on both datasets show that combining our proposed LLM-based data imputation and feature encoding methods with numerical machine learning predictors outperforms standalone fine-tuned LLM predictors in data-scarce settings. The proposed strategies emphasize data enhancement techniques rather than refining learning architectures or regularizing loss functions, offering a broadly applicable framework for improving machine learning performance on scarce, inhomogeneous datasets.

Chemical vapor deposition↗
DOE OSTI2026

Mechanism, Thermochemistry, and Kinetics for the CH + N 2 Reaction Leading to Prompt NO Formation in Combustion

Here, the reaction of CH + N 2 forming H + NCN is a remarkable example of activation of the nitrogen triple bond and is an important source of prompt NO in combustion. The reaction pathway is complex and proceeds through two competing mechanisms: a cyclic addition channel initiated by c-HC(NN) and a chain-addition channel initiated by HCNN, both of which eventually form HNCN prior to dissociation to H + NCN. This work reinvestigates this reaction with composite coupled cluster protocols, including a novel spin-flip equation of motion coupled cluster scheme, combined with pragmatic two-dimensional master equation simulations of the resulting rate coefficients. These improved calculations predict the CH + N 2 rate coefficient between the two more recent previous theoretical results and reduce the uncertainties of the best theoretical models of this reaction to less than a factor of 1.3. Additionally, we provide a closer theoretical investigation of the simultaneous dependence of the CH + N 2 rate coefficient on pressure and temperature, and affirm that collisionally stabilized HNCN, another potential source of prompt NO, emerges as an appreciable product of this reaction under conditions relevant to automotive internal combustion engines and aircraft gas turbine engines.

Nguyen, Thanh Lam [Univ. of Florida, Gainesville, ↗
NASA NTRS1996

Iridium Aluminide Coats For Protection Against Ox idation

Iridium aluminide coats investigated for use in protecting some metallic substrates against oxidation at high temperatures. Investigation prompted by need for cost-effective anti-oxidation coats for walls of combustion chambers in rocket engines. Also useful in special terrestrial applications like laboratory combustion chambers and some chemical-processing chambers.

Kaplan, Richard B.↗
NASA NTRS2002

Investigation of Friction Stir Welding and Laser Engineered Net Shaping of Metal Matrix Composite Materials

The improvement in weld quality by the friction stir welding (FSW) process invented by TWI of Cambridge, England, patented in 1991, has prompted investigation of this process for advanced structural materials including Al metal matrix composite (Al-MMC) materials. Such materials can have high specific stiffness and other potential beneficial properties for the extreme environments in space. Developments of discontinuous reinforced Al-MMCs have found potential space applications and the future for such applications is quite promising. The space industry has recognized advantages of the FSW process over conventional welding processes such as the absence of a melt zone, reduced distortion, elimination of the need for shielding gases, and ease of automation. The process has been well proven for aluminum alloys, and work is being carried out for ferrous materials, magnesium alloys and copper alloys. Development work in the FSW welding process for joining of Al-MMCs is relatively recent and some of this and related work can be found in referenced research publications. NASA engineers have undertaken to spear head this research development work for FSW process investigation of Al-MMCs. Some of the reported related work has pointed out the difficulty in fusion welding of particulate reinforced MMCs where liquid Al will react with SiC to precipitate aluminum carbide (Al4C3). Advantages of no such reaction and no need for joint preparation for the FSW process is anticipated in the welding of Al-MMCs. The FSW process has been best described as a combination of extrusion and forging of metals. This is carried out as the pin tool rotates and is slowly plunged into the bond line of the joint as the pin tool's shoulder is in intimate contact with the work piece. The material is friction-stirred into a quality weld.

Ravinder M. Diwan↗
DOE OSTI2024

Radiation characterization summary of the NETL beam port 1/5 free-field environment at the 128-inch core centerline adjacent location

The characterization of the neutron, prompt gamma-ray, and delayed gamma-ray radiation fields in the University of Texas at Austin Nuclear Engineering Teaching Laboratory (NETL) TRIGA reactor for the beam port (BP) 1/5 free-field environment at the 128-inch location adjacent to the core centerline has been accomplished. NETL is being explored as an auxiliary neutron test facility for the Sandia National Laboratories radiation effects sciences research and development campaigns. The NETL reactor is a TRIGA Mark-II pulse and steady-state, above-ground pool-type reactor. NETL is intended as a university research reactor typically used to perform irradiation experiments for students and customers, radioisotope production, as well as a training reactor. Initial criticality of the NETL TRIGA reactor was achieved on March 12, 1992, making it one of the newest test reactor facilities in the US. The neutron energy spectra, uncertainties, and covariance matrices are presented as well as a neutron fluence map of the experiment area of the cavity. For an unmoderated condition, the neutron fluence at the center of BP 1/5, at the adjacent core axial centerline, is about 8.2×10 12 n/cm 2 per MJ of reactor energy. About 67% of the neutron fluence is below 1 keV and 22% above 100 keV. The 1-MeV Damage-Equivalent Silicon (DES) fluence is roughly 1.6×10 12 n/cm 2 per MJ of reactor energy.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
NASA NTRS1992

An interactive preprocessor for the NASA engine performance program

The Simplified NEPP Automated Preprocessor (SNAP), which is written to aid in the preparation of input data files for the NASA Engine Performance Program (NEPP), is described. Specifically, SNAP is a software package on the Virtual Machine operating system that prompts the NEPP user for input information via a series of menus. The data collected from these menus are assimilated into an input file suitable for the running of NEPP. SNAP acts as a user-friendly preprocessing interface for NEPP. This serves as an introduction to the SNAP software, a user's manual, a description of the program logic, and a maintenance manual for future modifications to the software.

Berton, Jeffrey J.↗
NASA NTRS2015

Inheriting Curiosity: Leveraging MBSE to Build Mars2020

The success of the Jet Propulsion Laboratory's (JPL) Martian mission Mars Science Laboratory (MSL) prompted NASA to challenge JPL to build a second rover, Mars2020. Mars2020 has chosen to infuse Model Based Systems Engineering (MBSE) in pursuit of aiding the design of the Flight System. This paper will derive the motivation for MBSE infusion and will explain the current state of the Mars2020 Flight System Model. Successes in MBSE adoption will be discussed, as will limitations to the methodology.

Mars↗
DOE OSTI2026

Bayesian Optimization of Catalysis with In-Context Learning

Large language models (LLMs) can perform accurate classification with zero or few examples through in-context learning (ICL), allowing the model to observe query-relevant examples at inference time and eliminating the need for additional weight updates to generalize beyond its original training data. We extend this capability to regression with uncertainty estimation using frozen LLMs (e.g., GPT-4o, Gemini), enabling Bayesian optimization (BO) in natural language without explicit model training or feature engineering. We apply this to materials discovery by representing materials as synthesis and testing procedures for use in natural language prompts. This Bayesian, design-first approach prioritizes optimization toward target material properties before detailed characterization, in contrast to conventional experimental workflows that often emphasize characterization of suboptimal materials. On benchmarks like aqueous solubility and oxidative coupling of methane (OCM), BO-ICL matches or outperforms Gaussian processes. In live experiments on the reverse water–gas shift (RWGS) reaction, BO-ICL identifies multimetallic catalysts that approach equilibrium CO yield within 6 and 10 iterations from a pool of 3,700 and 360,000 candidates, respectively. Our method redefines materials representation and accelerates discovery, with broad applications across catalysis, materials science, and AI.

Calibration↗
NASA NTRS1977

Interactive LORAN-C to geographic and geographic-to-LORAN-C computation

The LORAN program is stored in CMS disk files for use by Avionics Engineering Center terminal users. A CMS EXEC file named LORAN controls program operation. The user types LORAN and the program then prompts for data input and produces output on the terminal. The FORTRAN program refers to a disk file of LORAN master data giving station locations, coding delays, repetition rate and station pair identification letters. For Geographic-to-LORAN conversion, no iterative computations are required; the program is a straightforward coordinate conversion based upon the techniques described by the Navy. For LORAN-to-Geographic conversion, the original Navy program required a dead-reckoned position, near the actual unknown fix, to begin computations. No iteration was performed to obtain the LORAN fix, but internal program errors occurred at execution time if the dead-reckoned fix were displaced from the actual fix by more than a few minutes of latitude or longitude. In order to enhance usefulness of the program for the terminal user, an iterative routine was added which allows a single dead-reckoned position to be entered from the master data file for each LORAN chain. The results compare exactly with the LORAN-C navigation chart, and provide adequate benchmark data for general aviation flight planning and data analysis.

Piecuch, L. M.↗
NASA NTRS1989

The role of the host in a cooperating mainframe and workstation environment, volumes 1 and 2

In recent years, advancements made in computer systems have prompted a move from centralized computing based on timesharing a large mainframe computer to distributed computing based on a connected set of engineering workstations. A major factor in this advancement is the increased performance and lower cost of engineering workstations. The shift to distributed computing from centralized computing has led to challenges associated with the residency of application programs within the system. In a combined system of multiple engineering workstations attached to a mainframe host, the question arises as to how does a system designer assign applications between the larger mainframe host and the smaller, yet powerful, workstation. The concepts related to real time data processing are analyzed and systems are displayed which use a host mainframe and a number of engineering workstations interconnected by a local area network. In most cases, distributed systems can be classified as having a single function or multiple functions and as executing programs in real time or nonreal time. In a system of multiple computers, the degree of autonomy of the computers is important; a system with one master control computer generally differs in reliability, performance, and complexity from a system in which all computers share the control. This research is concerned with generating general criteria principles for software residency decisions (host or workstation) for a diverse yet coupled group of users (the clustered workstations) which may need the use of a shared resource (the mainframe) to perform their functions.

Kusmanoff, Antone↗
NASA NTRS1993

Stresses In A Cylinder Welded With Mismatch

NASA technical memorandum presents parametric study of additional stresses introduced into cylindrical pressure vessel that is round except for radial mismatch in longitudinal butt weld in sidewall. Prompted by concern over fatigue caused by cyclic stresses in longitudinally welded cylindrical titanium inlet of high-pressure turbopump in main engine of Space Shuttle. Results applicable to any round cylindrical pressure vessel characterized by parameters, and conform to simplifying assumptions, used in analysis.

Min, J. B.↗
NASA NTRS1992

Attitude and articulation control of the CRAF/Cassini spacecraft in the presence of structural flexibilities and propellant slosh

Preliminary attitude and articulation controller designs for the CRAF/Cassini spacecraft have prompted a consideration of potential interactions with structural dynamics and propellant motion. The two algorithms significantly affected are attitude control during main engine burns, and articulation control of the high-precision scan platform (HPSP). For main engine burns, the primary concerns are the sloshing of the bipropellant, which constitutes 70 percent of initial spacecraft mass, and the vibration of the three large appendages, especially the HPSP, on which the gyros are mounted. For the HPSP pointing algorithm, the main concern is the fairly large offset of the HPSP center-of-mass from the gimbals, which is motivated by other system and subsystem considerations. This offset results in significant coupling between HPSP articulation and HPSP boom structural dynamics. This paper surveys the operational scenarios of these algorithms, and the impacts of nonrigid dynamics on controller design and performance.

Enright, Paul J.↗
NASA NTRS1990

An efficient design sensitivity analysis of eigenvectors

Subspace iteration has been a major advance in solving large eigen problems when only a subset of eigen-pairs is required. The essence of this method is a transformation from displacement coordinates of an n-th order eigensystem to generalized coordinates of a smaller q-th order. The eigenvalue problem is then solved in the reduced space. The method was first developed by Clint and Jennings for real symmetric systems and was then called 'simultaneous iteration'. The success of the method prompted further research along this line and there have been many improved algorithms developed. This approach has been widely used by structural engineers for extracting the most useful natural frequencies and mode shapes of large-scale dynamic systems. This paper exploits into a new direction which is in the form of iterative process for simultaneously calculating eigenvector derivatives of many eigenvectors with respect to multi-variables. The method fully uses all the available information from preceding eigenvalue solution and, thus, effectively economizes computational efforts. It iterates through two equations derived from the first variation of the two fundamental equations used in subspace iteration method. There is no expensive large matrix decomposition required and the process converges to acceptable solution in a finite number of iterations. Therefore, the procedure increases its efficiency superiority over the others as the system size or the number of interested eigenvectors become larger and larger.

Ting, T.↗