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.
The Lockheed TOYSAT experiment is described. The experiment was designed to test hypothesis concerning the application of optimal control theory to flexible spacecraft. The theory is presented, and results described.
Silicon nitride is a high temperature material currently under consideration for heat engine and other applications. The objective is to improve the net shape fabrication technology of Si3N4 by injection molding. This is to be accomplished by optimizing the process through a series of statistically designed matrix experiments. To provide input to the matrix experiments, a wide range of alternate materials and processing parameters was investigated throughout the whole program. The improvement in the processing is to be demonstrated by a 20 percent increase in strength and a 100 percent increase in the Weibull modulus over that of the baseline material. A full characterization of the baseline process was completed. Material properties were found to be highly dependent on each step of the process. Several important parameters identified thus far are the starting raw materials, sinter/hot isostatic pressing cycle, powder bed, mixing methods, and sintering aid levels.
To show the feasibility of simultaneous optimization as design procedure, low order problems were used in conjunction with simple control formulations. The numerical results indicate that simultaneous optimization is not only feasible, but also advantageous. Such advantages come at the expense of introducing complexities beyond those encountered in structure optimization alone, or control optimization alone. Examples include: larger design parameter space, optimization may combine continuous and combinatoric variables, and the combined objective function may be nonconvex. Future extensions to include large order problems, more complex objective functions and constraints, and more sophisticated control formulations will require further research to ensure that the additional complexities do not outweigh the advantages of simultaneous optimization. Some areas requiring more efficient tools than currently available include: multiobjective criteria and nonconvex optimization. Efficient techniques to deal with optimization over combinatoric and continuous variables, and with truncation issues for structure and control parameters of both the model space as well as the design space need to be developed.
In this work, we describe methods for selecting the neural network architectures and input spaces to implement belief state inference on generic commercial transport aircraft. First, we highlight a case study on the planning, execution, and analysis of a set of experiments to determine the configurations of a conditional variational autoencoder (CVAE). We present a structured method that can be used in a number of aerospace applications, to optimize the structure and training parameters of the CVAE for belief state inference, using Design of Experiments (DOE) statistical methodologies. The motivation for this specific DOE was to identify the appropriate hyperparameters for measuring the CVAE reconstruction probability and latent space, such that the measurements can be used to infer qualitative state changes for the aircraft. We demonstrate that this process yields information about a trained neural network’s utility for this specific application, along with a quantifiable range of certainty. We execute 84 experiments using loss-of-control flight maneuver data from a NASA T-2 aircraft, demonstrating that this empirical process allows us to construct cheap and simple models with specific attributes amenable to belief state inference in aerospace applications. While theoretically, we could create a single CVAE with an input space the size of all measurable flight variables and environmental dynamics, it becomes intractable to use such a neural network in an in-situ intelligent multi-agent system. Using the recommendations from our case study, we introduce a technical approach for feasibly describing the belief space by (1) identifying significant statistical relationships among flight variables using rule induction, (2) using a set of rules that cover all features to define the input space of multiple CVAEs, and (3) forming a belief space based on the joint probability density of their collective latent spaces. This results in a series of relatively small matrix multiplications that can be performed in real time, as opposed to large matrix computations in a single CVAE. We demonstrate the application of this approach on the T-2 flight loss-of control experiments, using the architecture and hyperparameter recommendations from the case study. We compare the utilities of an individual CVAE trained on all flight variables and multiple CVAEs defined on subsets of flight variables for detecting qualitative changes in flight. We demonstrate that the use of multiple CVAEs with smaller input spaces permits the CVAE to capture more granular relationships in the latent space, permitting better state space characterization and loss-of-control detection.
An extensive numerical experiment, using the developed computer code, was conducted to design an optimized laser-sustained hydrogen plasma thruster. The plasma was sustained using a 30 kW CO2 laser beam operated at 10.6 micrometers focused inside the thruster. The adopted physical model considers two-dimensional compressible Navier-Stokes equations coupled with the laser power absorption process, geometric ray tracing for the laser beam, and the thermodynamically equilibrium (LTE) assumption for the plasma thermophysical and optical properties. A pressure based Navier-Stokes solver using body-fitted coordinate was used to calculate the laser-supported rocket flow which consists of both recirculating and transonic flow regions. The computer code was used to study the behavior of laser-sustained plasmas within a pipe over a wide range of forced convection and optical arrangements before it was applied to the thruster design, and these theoretical calculations agree well with existing experimental results. Several different throat size thrusters operated at 150 and 300 kPa chamber pressure were evaluated in the numerical experiment. It is found that the thruster performance (vacuum specific impulse) is highly dependent on the operating conditions, and that an adequately designed laser-supported thruster can have a specific impulse around 1500 sec. The heat loading on the wall of the calculated thrusters were also estimated, and it is comparable to heat loading on the conventional chemical rocket. It was also found that the specific impulse of the calculated thrusters can be reduced by 200 secs due to the finite chemical reaction rate.
This paper discusses an experimental investigation conducted to optimize a free-to-rotate wing for use on a small unmanned aircraft system (UAS). Although free-to-rotate wings have been used for decades on various small UAS and small manned aircraft, little is known about how to optimize these unusual wings for a specific application. The paper discusses some of the design rationale of the basic wing. In addition, three main parameters were selected for "optimization", wing camber, wing pivot location, and wing center of gravity (c.g.) location. A small apparatus was constructed to enable some simple experimental analysis of these parameters. A design-of-experiment series of tests were first conducted to discern which of the main optimization parameters were most likely to have the greatest impact on the outputs of interest, namely, some measure of "stability", some measure of the lift being generated at the neutral position, and how quickly the wing "recovers" from an upset. A second set of tests were conducted to develop a response-surface numerical representation of these outputs as functions of the three primary inputs. The response surface numerical representations are then used to develop an "optimum" within the trade space investigated. The results of the optimization are then tested experimentally to validate the predictions.
The Airborne Imaging Spectrometer (AIS) is a multispectral infrared imaging instrument utilizing a 32- x 32-element mercury-cadmium-telluride detector array to achieve high spatial resolution. A novel optical design provides high spectral resolution, and, in conjunction with the 32-element spatial resolution, makes optimal use of the capabilities of the area array. The experiment requirements and the details of instrument design, implementation, and calibration are presented. The superior resolution, accuracy, and data acquisition rate of this sensor/instrument concept for reconnaissance and geophysical survey is detailed and plans for upgrading/improving the instrument and future applications of this design concept are discussed.
Airframe noise is a significant component of environmental noise in the vicinity of airports. The noise associated with the leading-edge slat of typical transport aircraft is a prominent source of airframe noise. Previous work suggests that a slat-cove filler (SCF) may be an effective noise treatment. Hence, development and optimization of a practical slat-cove-filler structure is a priority. The objectives of this work are to optimize the design of a functioning SCF which incorporates superelastic shape memory alloy (SMA) materials as flexures that permit the deformations involved in the configuration change. The goal of the optimization is to minimize the actuation force needed to retract the slat-SCF assembly while satisfying constraints on the maximum SMA stress and on the SCF deflection under static aerodynamic pressure loads, while also satisfying the condition that the SCF self-deploy during slat extension. A finite element analysis model based on a physical bench-top model is created in Abaqus such that automated iterative analysis of the design could be performed. In order to achieve an optimized design, several design variables associated with the current SCF configuration are considered, such as the thicknesses of SMA flexures and the dimensions of various components, SMA and conventional. Designs of experiment (DOE) are performed to investigate structural response to an aerodynamic pressure load and to slat retraction and deployment. DOE results are then used to inform the optimization process, which determines a design minimizing actuator forces while satisfying the required constraints.
Reliable, secure access to energy is a major focus for national security efforts. One potential route to such energy is through fusion reactions in inertial confinement fusion (ICF) experiments. Such experiments are carried out at facilities such as the National Ignition Facility (NIF) in Livermore, California, where high powered lasers are used to compress a DT fuel-containing target to the necessary high temperature, high pressure conditions. These experiments are limited in number, which creates a heavy dependence on high fidelity predictive physics simulations and analysis performed “pre shot,” or before the experiment occurs. Many of these simulations in higher dimensions (2D and 3D) are computationally expensive, so finding optimal simulation-based designs presents its own challenges. In this work, we present our multi-fidelity Bayesian optimization with Gaussian processes (GPs) for ICF double shell targets, where a 1D surrogate model is used to help find a 2D surrogate model, enabling us to find optimal targets in the higher fidelity (2D), while saving computational cost.
The traditional approach to the design of engine inlets for commercial transport aircraft is a tedious process that ends with a less-than-optimum design. With the advent of high-speed computers and the availability of more accurate and reliable computational fluid dynamics (CFD) solvers, numerical optimization processes can effectively be used to design an aerodynamic inlet lip that enhances engine performance. The designers' experience at Boeing Corporation showed that for a peak Mach number on the inlet surface beyond some upper limit, the performance of the engine degrades excessively. Thus, our objective was to optimize efficiency (minimize the peak Mach number) at maximum cruise without compromising performance at other operating conditions. Using a CFD code NPARC, the NASA Lewis Research Center, in collaboration with Boeing, developed an integrated procedure at Lewis to find the optimum shape of a subsonic inlet lip and a numerical optimization code, ADS. We used a GRAPE-based three-dimensional grid generator to help automate the optimization procedure. The inlet lip shape at the crown and the keel was described as a superellipse, and the superellipse exponents and radii ratios were considered as design variables. Three operating conditions: cruise, takeoff, and rolling takeoff, were considered in this study. Three-dimensional Euler computations were carried out to obtain the flow field. At the initial design, the peak Mach numbers for maximum cruise, takeoff, and rolling takeoff conditions were 0.88, 1.772, and 1.61, respectively. The acceptable upper limits on the takeoff and rolling takeoff Mach numbers were 1.55 and 1.45. Since the initial design provided by Boeing was found to be optimum with respect to the maximum cruise condition, the sum of the peak Mach numbers at takeoff and rolling takeoff were minimized in the current study while the maximum cruise Mach number was constrained to be close to that at the existing design. With this objective, the optimum design satisfied the upper limits at takeoff and rolling takeoff while retaining the desirable cruise performance. Further studies are being conducted to include static and cross-wind operating conditions in the design optimization procedure. This work was carried out in collaboration with Dr. E.S. Reddy of NYMA, Inc.
Mixing of gaseous jets in a cross-flow has significant applications in engineering, one example of which is the dilution zone of a gas turbine combustor. Despite years of study, the design of jet injection in combustors is largely based on practical experience. A series of experiments was undertaken to delineate the optimal mixer orifice geometry. A cross-flow to core-flow momentum-flux ratio of 40 and a mass flow ratio of 2.5 were selected as representative of an advanced design. An experimental test matrix was designed around three variables: the number of orifices, the orifice aspect ratio (long-to-short dimension), and the orifice angle. A regression analysis was performed on the data to arrive at an interpolating equation that predicted the mixing performance of orifice geometry combinations within the range of the test matrix parameters. Results indicate that mixture uniformity is a non-linear function of the number of orifices, the orifice aspect ratio, and the orifice angle. Optimum mixing occurs when the asymptotic mean jet trajectories are in the range of 0.35 less than r/R less than 0.5 (where r = 0 is at the mixer wall) at z/R = 1.0. At the optimum number of orifices, the difference between shallow-angled slots with large aspect ratios and round holes is minimal and either approach will lead to good mixing performance. At the optimum number of orifices, it appears possible to have two local optimums where one corresponds to an aspect ratio of 1.0 and the other to a high aspect ratio.
The use of magnetohydrodynamic (MHD) power conversion systems, potentially coupled with a fission power source, is currently being investigated as a driver for an advanced propulsion system, such as a plasma thruster. The efficiency of a MHD generator is strongly dependent on the electrical conductivity of the fluid that passes through the generator; power density increases as fluid conductivity increases. Although traditional MHD flows depend on thermal ionization to enhance the electrical conductivity, ionization due to nuclear interactions may achieve a comparable or improved conductivity enhancement while avoiding many of the limitations inherent to thermal ionization. Calculations suggest that nuclear-enhanced electrical conductivity increases as the neutron flux increases; conductivity of pure He-3 greater than 10 mho/m may be achievable if exposed to a flux greater than 10(exp 12) neutrons/cm2/s.) However, this remains to be demonstrated experimentally. An experimental facility has been constructed at the Propulsion Research Center at the NASA Marshall Space Flight Center, using helium as the test fluid. High energy electrons will be used to simulate the effects of neutron-induced ionization of helium gas to produce a plasma. These experiments will be focused on diagnosis of the plasma in a virtually static system; results will be applied to future tests with a MHD system. Initial experiments will utilize a 50 keV electron gun that can operate at up to a current of 200 micro A. Spreading the electron beam over a four inch diameter window results in an electron flux of 1.5x 10(exp 13) e/sq cm/s. The equivalent neutron flux that would produce the same ionization fraction in helium is 1x10(exp 12) n/sq cm/s. Experiments will simulate the neutron generated plasma modeled by Bitteker, which takes into account the products of thermal neutron absorption in He-3, and includes various ion species in estimating the conductivity of the resulting plasma. Several different probes will be designed and implemented to verify the plasma kinetics model. System parameters and estimated operating ranges are summarized. The predicted ionization fraction, electron density, and conductivity levels are provided in for an equivalent neutron flux of 1x10(exp 12) n/cm2/s. Understanding the complex plasma kinetics throughout a MHD channel is necessary to design an optimal power conversion system for space propulsion applications. The proposed experiments seek to fully characterize the helium plasma and to determine the reliability of each measurement technique, such that they may be applied to more advanced MHD studies. The expected value of each plasma parameter determined from theoretical models will be verified experimentally by several independent techniques to determine the most reliable method of obtaining each parameter. The results of these experiments will be presented in the final paper.
We describe empirical methods for selecting a neural network architecture to implement belief state inference on generic commercial transport aircraft. We highlight a case study on the planning, execution, and analysis of a set of experiments to determine the configurations of a conditional variational autoencoder (CVAE). Our main contribution is the application of a structured method that can be used for machine learning in many aerospace applications. This method optimizes the structure and training parameters of a neural network for belief state inference, using Design of Experiments (DOE) statistical methodologies. The motivation for this specific DOE analysis was to identify the appropriate hyperparameters for measuring the CVAE reconstruction probability and latent space, such that the measurements can be used to infer qualitative state changes for the aircraft. We demonstrate that this process yields information about a trained neural network’s utility for this specific application, along with a quantifiable range of certainty. We execute 84 experiments using loss-of-control flight maneuver data from the NASA T-2 aircraft, demonstrating that this empirical process allows us to construct cheap and simple models with specific attributes amenable to belief state inference in aerospace applications.
We describe empirical methods for selecting a neural network architecture to implement belief state inference on generic commercial transport aircraft. We highlight a case study on the planning, execution, and analysis of a set of experiments to determine the configurations of a conditional variational autoencoder (CVAE). Our main contribution is the application of a structured method that can be used for machine learning in many aerospace applications. This method optimizes the structure and training parameters of a neural network for belief state inference, using Design of Experiments (DOE) statistical methodologies. The motivation for this specific DOE analysis was to identify the appropriate hyperparameters for measuring the CVAE reconstruction probability and latent space, such that the measurements can be used to infer qualitative state changes for the aircraft. We demonstrate that this process yields information about a trained neural network’s utility for this specific application, along with a quantifiable range of certainty. We execute 84 experiments using loss-of-control flight maneuver data from the NASA T 2 aircraft, demonstrating that this empirical process allows us to construct cheap and simple models with specific attributes amenable to belief state inference in aerospace applications.
Q-Law is a Lyapunov-based control law used to determine optimal controls for a low thrust trajectory. One major issue with its use is the difficult derivatives re-quired for calculating optimal controls at a given time. In this paper, an implementation of Q-Law with automatic differentiation via a Python package called JAX is applied. With automatic differentiation, the difficult derivatives for Q-Law’s optimal controls are calculated with ease, and derivatives of final states with respect to Q-Law’s weights are found enabling gradient-based optimization of Q-Law for the first time. Different search and optimization methods for finding optimal weights are then compared using the LEO to GEO problem, and it was found that gradient-free methods like design of experiments and genetic algorithm produced the best results, but they took the longest time to get a solution, while the gradient-based method found a locally optimal result in a much faster time. Overall, the run time for a single propagation is manageable and well-suited for a mission designer to use as an initial guess generator for trajectory optimization, or for simple orbit transfer analysis.
An instrument to perform in-reactor, real-time thermal property measurements of nuclear fuels and materials was recently designed, optimized, and tested at Massachusetts Institute of Technology Research Reactor. It was the second insertion experiments of a total of two rounds. Here we report the collection and analysis of the real-time captured thermal diffusivity data on the preloaded SiC. With the optimized design, the instrument temperature had been successfully raised as we expected. Only one probe (the top one) returned real-time thermal diffusivity in a time span of approximately two days. Measured thermal diffusivity of the preloaded SiC was significantly lower than the literature value at the same temperature, indicating a significant influence from the irradiation induced microstructure damage. Meanwhile, the real-time thermal diffusivity value continuously reduced with time, indicating the dynamic microstructure evolution. The possible reasons of the probes that did not return data were hypothesized. The further optimization of the instrument with the objective to improve the measurement accuracy and instrument survivability is also discussed.
Abstract—We describe empirical methods for selecting a neural network architecture to implement belief state inference on generic commercial transport aircraft. We highlight a case study on the planning, execution, and analysis of a set of experiments to determine the configurations of a conditional variational autoencoder (CVAE). Our main contribution is the application of a structured method that can be used for machine learning in many aerospace applications. This method optimizes the structure and training parameters of a neural network for belief state inference, using Design of Experiments (DOE) statistical methodologies. The motivation for this specific DOE analysis was to identify the appropriate hyperparameters for measuring the CVAE reconstruction probability and latent space, such that the measurements can be used to infer qualitative state changes for the aircraft. We demonstrate that this process yields information about a trained neural network’s utility for this specific application, along with a quantifiable range of certainty. We execute 84 experiments using loss-of-control flight maneuver data from the NASA T-2 aircraft, demonstrating that this empirical process allows us to construct cheap and simple models with specific attributes amenable to belief state inference in aerospace applications.
An attempt is made to summarize the experience of the design of antennas for various types of spacecraft: fully oriented, partially oriented, and with arbitrary orientation. Several examples of the construction of antenna systems and their characteristics are given. The optimal relationship between weight, dimension and power characteristics of on-board devices allows selection of the most acceptable versions. Questions are presented for consideration in design of optimal spacecraft antenna-feeder systems.