Publications

We describe an algorithm for the static load balancing of scientific computations that generalizes and improves upon spectral bisection. Through a novel use of multiple eigenvectors, our new spectral algorithm can divide a computation into 4 or 8 pieces at once. This leads to balanced partitions that have lower communication overhead and are less expensive to compute than those of spectral bisection. In addition, our approach automatically works to minimize message contention on a hypercube or mesh architecture.

This paper describes a collaborative effort between Sandia National Laboratories and the Rocketdyne Division of Rockwell International Corporation to develop an automated braze paste dispensing system for rocket engine nozzle manufacturing. The motivation for automating this manufacturing process is to reduce the amount of labor and excess material required. A critical requirement for this system is the automatic location of key nozzle features using non-contact sensors. Sandia has demonstrated that the low-cost Multi-Axis Seam Tracking (MAST) capacitive sensor can be used to accurately locate the nozzle surface and tube gaps.

This paper examines some of the metrics that are commonly used to design correlation filter's for optical pattern recognition, including: the Fisher ratio, the signal-to-noise ratio, the equal correlation peak (ECP) constraint, and normalized correlation. Attention is given to the underlying assumptions that are required to move from Bayesian decision theory to a particular metric or design principle. Since a Bayes classifier is statistically optimum, this provides a means for assessing the merit of a particular approach. Although the authors only examine a few metrics in this paper, the approach is general and should be useful for assessing the merit and applicability of any of the numerous filter designs that have been proposed in the optical pattern recognition community.

This report discusses electronic isolators which are used to maintain electrical separation between safety and non-safety systems in nuclear power plants. The concern is that these devices may fail allowing unwanted signals or energy to act upon safety systems, or preventing desired signals from performing their intended function. While operational history shows many isolation device problems requiring adjustments and maintenance, we could not find incidents where there was a safety implication. Even hypothesizing multiple simultaneous failures did not lead to significant contributions to core damage frequency. Although the analyses performed in this study were not extensive or detailed, there seems to be no evidence to suspect that isolation device failure is an issue which should be studied further.

This report documents the fiscal year 1992 activities of the, Utility Battery Storage Systems Program (UBS) of the US Department of Energy (DOE), Office of Energy Management (OEM). The UBS program is conducted by Sandia National Laboratories (SNL). UBS is responsible for the engineering development of integrated battery systems for use in utility-energy-storage (UES) and other stationary applications. Development is accomplished primarily through cost-shared contracts with industrial organizations. An important part of the development process is the identification, analysis, and characterization of attractive UES applications. UBS is organized into five projects: Utility Battery Systems Analyses; Battery Systems Engineering; Zinc/Bromine; Sodium/Sulfur; Supplemental Evaluations and Field Tests. The results of the Utility Systems Analyses are used to identify several utility-based applications for which battery storage can effectively solve existing problems. The results will also specify the engineering requirements for widespread applications and motivate and define needed field evaluations of full-size battery systems.

The sixth experiment of the Integral Effects Test (IET-6) series was conducted to investigate the effects of high pressure melt ejection on direct containment heating. Scale models of the Zion reactor pressure vessel (RPV), cavity, instrument tunnel, and subcompartment structures were constructed in the Surtsey Test Facility at Sandia National Laboratories. The RPV was modeled with a melt generator that consisted of a steel pressure barrier, a cast MgO crucible, and a thin steel inner liner. The melt generator/crucible had a hemispherical bottom head containing a graphite limitor plate with a 4-cm exit hole to simulate the ablated hole in the RPV bottom head that would be formed by ejection of an instrument guide tube in a severe nuclear power plant accident. The cavity contained 3.48 kg of water, which corresponds to condensate levels in the Zion plant, and the containment basement floor was dry. A 43-kg initial charge of iron oxide/aluminum/chromium thermite was used to simulate corium debris on the bottom head of the RPV. Molten thermite was ejected by steam at an initial pressure of 6.3 MPa into the reactor cavity. The Surtsey vessel atmosphere contained pre-existing hydrogen to represent partial oxidation of the zirconium in the Zion core. The initial composition of the vessel atmosphere was 87.1 mol.% N[sub 2], 9.79 mol.% O[sub 2], and 2.59 mol.% H[sub 2], and the initial absolute pressure was 198 kPa. A partial hydrogen burn occurred in the Surtsey vessel. The peak vessel pressure increase was 279 kPa in IET-6, compared to 246 kPa in the IET-3 test. The total debris mass ejected into the Surtsey vessel in IET-6 was 42.5 kg. The gas grab sample analysis indicated that there were 180 g[center dot] moles of pre-existing hydrogen, and that 308[center dot]moles of hydrogen were produced by steam/metal reactions. About 335 g[center dot]moles of hydrogen burned, and 153 g[center dot]moles remained unreacted.

The Sandia National Laboratories (SNL) Engineering Analysis Code Access System (SEACAS) is a collection of structural and thermal codes and utilities used by analysts at SNL. The system includes pre- and post-processing codes, analysis codes, database translation codes, support libraries, UNIX{trademark} shell scripts, and an installation system. SEACAS is used at SNL on a daily basis as a production, research, and development system for the engineering analysts and code developers. Over the past year, approximately 180 days of Cray Y-MP{trademark} CPU time have been used at SNL by SEACAS codes. The job mix includes jobs using only a few seconds of CPU time, up to jobs using two and one-half days of CPU time. SEACAS is running on several different systems at SNL including Cray Unicos, Hewlett Packard HP-UX{trademark}, Digital Equipment Ultrix{trademark}, and Sun SunOS{trademark}. This document is a short description of the codes the SEACAS system.

This bulletin from Sandia Laboratories presents current research on testing technology. Fiber optics systems at the Nevada Test Site is replacing coaxial cables. The hypervelocity launcher is being used to test orbital debris impacts with space station shielding. A digital recorder makes testing of high-speed water entries possible. Automobile engine design is aided by an instrumented head gasket that detects the combustion zone. And composite-to-metal strength and fatigue tests provide new data on joint failures in wind turbine joint tests.

The CONTAIN computer code is a best-estimate, integrated analysis tool for predicting the physical, chemical, and radiological conditions inside a nuclear reactor containment building following the release of core material from the primary system. CONTAIN is supported primarily by the U. S. Nuclear Regulatory Commission (USNRC), and the official code versions produced with this support are intended primarily for the analysis of light water reactors (LWR). The present manual describes CONTAIN LMR/1B-Mod. 1, a code version designed for the analysis of reactors with liquid metal coolant. It is a variant of the official CONTAIN 1.11 LWR code version. Some of the features of CONTAIN-LMR for treating the behavior of liquid metal coolant are in fact present in the LWR code versions but are discussed here rather than in the User`s Manual for the LWR versions. These features include models for sodium pool and spray fires. In addition to these models, new or substantially improved models have been installed in CONTAIN-LMR. The latter include models for treating two condensables (sodium and water) simultaneously, sodium atmosphere and pool chemistry, sodium condensation on aerosols, heat transfer from core-debris beds and to sodium pools, and sodium-concrete interactions. A detailed description of each of the above models is given, along with the code input requirements.

Inelastic material constitutive relations for elastoplasticity coupled with continuum damage mechanics are investigated. For elastoplasticity, continuum damage mechanics, and the coupled formulations, rigorous thermodynamic frameworks are derived. The elastoplasticity framework is shown to be sufficiently general to encompass J{sub 2} plasticity theories including general isotropic and kinematic hardening relations. The concepts of an intermediate undamaged configuration and a fictitious deformation gradient are used to develop a damage representation theory. An empirically-based, damage evolution theory is proposed to overcome some observed deficiencies. Damage deactivation, which is the negation of the effects of damage under certain loading conditions, is investigated. An improved deactivation algorithm is developed for both damaged elasticity and coupled elastoplasticity formulations. The applicability of coupled formulations is validated by comparing theoretical predictions to experimental data for a spectrum of materials and loads paths. The pressure-dependent brittle-to-ductile transitional behavior of concrete is replicated. The deactivation algorithm is validated using tensile and compression data for concrete. For a ductile material, the behavior of an aluminum alloy is simulated including the temperature-dependent ductile-to-brittle behavior features. The direct application of a coupled model to fatigue is introduced. In addition, the deactivation algorithm in conjunction with an assumed initial damage and strain is introduced as a novel method of simulating the densification phenomenon in cellular solids.

Target recognition requires the ability to distinguish targets from non-targets, a capability called one-class generalization. Many neural network pattern classifiers fail as one-class classifiers because they use open decision boundaries. To function as one-class classifier, a neural network must have three types of generalization: within-class, between-class, and out-of-class. We discuss these three types of generalization and identify neural network architectures that meet these requirements. We have applied our one-class classifier ideas to the problem of automatic target recognition in synthetic aperture radar. We have compared three neural network algorithms: Carpenter and Grossberg`s algorithmic version of the Adaptive Resonance Theory (ART-2A), Kohonen`s Learning Vector Quantization (LVQ), and Reilly and Cooper`s Restricted Coulomb Energy network (RCE). The ART 2-A neural network gives the best results, with 100% within-class, between-class, and out-of-class generalization. Experiments show that the network`s performance is sensitive to vigilance and number of training set presentations.

This report discusses the seventh experiment of the Integral Effects Test (IET-7) series. The experiment was conducted to investigate the effects of preexisting hydrogen in the Surtsey vessel on direct containment heating. Scale models of the Zion reactor pressure vessel (RPV), cavity, instrument tunnel, and subcompartment structures were constructed in the Surtsey Test Facility at Sandia National Laboratories. The RPV was modeled with a melt generator that consisted of a steel pressure barrier, a cast MgO crucible, and a thin steel inner liner. The melt generator/crucible had a hemispherical bottom head containing a graphite limitor plate with a 4-cm exit hole to simulate the ablated hole in the RPV bottom head that would be formed by ejection of an instrument guide tube in a severe nuclear power plant accident. The cavity contained 3.48 kg of water, and the containment basement floor inside the cranewall contained 71 kg of water, which corresponds to scaled condensate levels in the Zion plant. A 43-kg initial charge of iron oxide/aluminum/chromium thermite was used to simulate corium debris on the bottom head of the RPV. Molten thermite was ejected by steam at an initial pressure of 5.9 MPa into the reactor cavity.

Charpy V-notch specimens (ASTM Type A) and 5.74-mm diameter tension test specimens of the Shippingport Reactor Neutron Shield Tank (NST) (outer wall material) were irradiated together with Charpy V-notch specimens of the Oak Ridge National Laboratory (ORNI), High,, Flux Isotope Reactor (HFIR) vessel (shell material), to 5.07 {times} 10{sup 17} n/cm{sup 2}, E > 1 MeV. The irradiation was performed in the Ford Nuclear Reactor (FNR), a test reactor, at a controlled temperature of 54{degrees}C (130{degrees}F) selected to approximate the prior service temperatures of the cited reactor structures. Radiation-induced elevations in the Charpy 41-J transition temperature and the ambient temperature yield strength were small and independent of specimen test orientation (ASTM LT vs. TL). The observations are consistent with prior findings for the two materials (A 212-B plate) and other like materials irradiated at low temperature (< 200{degrees}C) to low fluence. The high radiation embrittlement sensitivity observed in HFIR vessel surveillance program tests was not found in the present accelerated irradiation test. Response to 288{degrees}C-168 h postirradiation annealing was explored for the NST material. Notch ductility recovery was found independent of specimen test orientation but dependent on the temperature within the transition region at which the specimens were tested.

A method is presented for determining the nonlinear stability of undamped flexible structures spinning about a principal axis of inertia. Equations of motion are developed for structures that are free of applied forces and moments. The development makes use of a floating reference frame which follows the overall rigid body motion. Within this frame, elastic deformations are assumed to be given functions of n generalized coordinates. A transformation of variables is devised which shows the equivalence of the equations of motion to a Hamiltonian system with n + 1 degrees of freedom. Using this equivalence, stability criteria are developed based upon the normal form of the Hamiltonian. It is shown that a motion which is spin stable in the linear approximation may be unstable when nonlinear terms are included. A stability analysis of a simple flexible structure is provided to demonstrate the application of the stability criteria. Results from numerical integration of the equations of motion are shown to be consistent with the predictions of the stability analysis. A new method for modeling the dynamics of rotating flexible structures is developed and investigated. The method is similar to conventional assumed displacement (modal) approaches with the addition that quadratic terms are retained in the kinematics of deformation. Retention of these terms is shown to account for the geometric stiffening effects which occur in rotating structures. Computational techniques are developed for the practical implementation of the method. The techniques make use of finite element analysis results, and thus are applicable to a wide variety of structures. Motion studies of specific problems are provided to demonstrate the validity of the method. Excellent agreement is found both with simulations presented in the literature for different approaches and with results from a commercial finite element analysis code. The computational advantages of the method are demonstrated.

We describe an algorithm for the static load balancing of scientific computations that generalizes and improves upon spectral bisection. Through a novel use of multiple eigenvectors, our new spectral algorithm can divide a computation into 4 or 8 pieces at once. These multidimensional spectral partitioning algorithms generate balanced partitions that have lower communication overhead and are less expensive to compute than those produced by spectral bisection. In addition, they automatically work to minimize message contention on a hypercube or mesh architecture. These spectral partitions are further improved by a multidimensional generalization of the Kernighan-Lin graph partitioning algorithm. Results on several computational grids are given and compared with other popular methods.

A robotic rover vehicle designed for use in the exploration of the Lunar surface is described. The Robotic All-Terrain Lunar Exploration Rover (R-A.T.L.E.R-) is a four wheeled all-wheel-drive dual-body vehicle. A uniquely simple method of chassis articulation is employed which allows all four wheels to remain in contact with the ground, even while climbing over step-like obstacles as large as [approximately]1.3 wheel diameters. Skid steering and modular construction are used to produce a simple, rugged, highly agile mobility chassis with fewer parts required compared to other designs being considered for planetary exploration missions. The design configuration, mobility parameters, and performance of several existing R.A.T.L.E.R prototypes are discussed, with emphasis on an analysis of the configuration parameters which directly affect the designs mobility performance.

This report describes preliminary experiments to investigate the feasibility of using electron beam (e-beam) radiolysis to destroy the organic compounds in simulated Hanford tank waste. For these experiments a simulated Hanford Tank 101-SY waste mixture was radiolyzed in a {sup 60}Co facility to simulate radiolysis in the waste tank. This slurry was then exposed without dilution to dose levels up to 1600 Mrad at instantaneous dose rates of 2.5 {times} 10{sup 8} and 2. 7 {times} 10{sup 11} rad/s. The inferred dose to destroy all the organic material in the simulated waste, assuming destruction is linear with dose, is 1000 Mrads for the higher dose rate. The cost for organic destruction of Hanford waste at a treatment rate of 20 gpm is roughly estimated to be $10. 60 per gallon. Such a system would treat all the waste in a 1 million gallon Hanford tank in about 40 days. Estimates of capital costs are given in the body of this report. While ferrocyanide destruction was not experimentally investigated in this work, previous experiments by others suggest that ferrocyanide would also be destroyed in such a system.

This report describes the Training and Qualification Program at the Saturn Facility. The main energy source at Saturn is the Saturn accelerator which is used to test military hardware for vulnerability to X-rays, as well as to perform various types of plasma radiation source experiments. The facility is operated and maintained by a staff of twenty scientists and technicians. This program is designed to ensure these personnel are adequately trained and qualified to perform their jobs in a safe and efficient manner. Copies of actual documents used in the program are included as appendices. This program meets all the requirements for training and qualification in the DOE Orders on Conduct of Operations and Quality Assurance, and may be useful to other organizations desiring to come into compliance with these orders.

Experiments were run to determine if oxidized Kovar could be chemically cleaned so that copper would wet the Kovar in a wet hydrogen atmosphere at 1100{degrees}C. We found that a multi-stepped acid etch process cleaned the Kovar so that copper would wet it. We also found that the degree of copper cracking after melting and cool-down correlated well with the degree of wetting.

The relatively thin web of salt that separates Bayou Choctaw Caverns 15 and 17 was evaluated using the finite-element method. The stability calculations provided insight as to whether or not any operationrestrictions or recommendations are necessary. Because of the uncertainty in the exact dimensions of the salt web, various web thicknesses were examined under different operating scenarios that included individual cavern workovers and drawdowns. Cavern workovers were defined by a sudden drop in the oil side pressure at the wellhead to atmospheric. Workovers represent periods of low cavern pressure. Cavern drawdowns were simulated by enlargening the cavern diameters, thus decreasing the thickness of the web. The calculations predict that Cavern 15 dominates the behavior of the web because of its larger diameter. Thus, giventhe choice of caverns, Cavern 17 should be used for oil withdrawal in order to minimize the adverse impacts on web resulting from pressure drops or cavern enlargement. From a stability point of view, maintaining normal pressures in Cavern 15 was found to be more important than operating the caverns as a gallery where both caverns are maintained at the same pressure. However, during a workover, it may be prudent to operate the caverns under similar pressures to avoid the possibility of a sudden pressure surge at the wellhead should the web fail.

A feasibility study for developing an improved tool and improved models for performing event assessments is described. The study indicates that the IRRAS code should become the base tool for performing event assessments, but that modifications would be needed to make it more suitable for routine use. Alternative system modeling approaches are explored and an approach is recommended that is based on improved train-level models. These models are demonstrated for Grand Gulf and Sequoyah. The insights that can be gained from importance measures are also demonstrated. The feasibility of using Individual Plant Examination (IPE) submittals as the basis for train-level models for precursor studies was also examined. The level of reported detail was found to vary widely, but in general, the submittals did not provide sufficient information to fully define the model. The feasibility of developing an industry risk profile from precursor results and of trending precursor results for individual plants were considered. The data sparsity would need to be considered when using the results from these types of evaluations, and because of the extremely sparse data for individual plants we found that trending evaluations for groups of plants would be more meaningful than trending evaluations for individual plants.

The discrete Fourier transform (DFT) is used frequently used in the computation of the signal-to-noise ratio (SNR) and harmonic distortion. To estimate the SNR or harmonic distortion, a sine wave is applied to the digitizing system under test. When the data record contains an integer number of cycles of the sine wave, energy from the sine wave and its harmonics does not leak into adjacent DFT frequency bins. Each harmonic occupies one an only one DFT frequency bin. To find the root-mean-square (RMS) value of a harmonic from its DFT, one computes the magnitude of the DFT value at the single frequency of the harmonic. When the DFTs of the fundamental and its harmonics are single lines, the SNR and harmonic distortion are easy to compute. When the data record contains a non-integer number of cycles of the sine wave, energy leaks from the sine wave and its harmonics to adjacent frequencies. The literature contains several approaches to problem of determining which DFT components correspond to a sine wave harmonic. This paper describes how to estimate the RMS value of a sine wave from its DFT with special attention to the selection of the DFT window. The set of DFT frequencies which comprise a harmonic depends on the DFT window, the length of the DFT, and the number of bits of the digitizer. Criteria are developed for choosing the DFT frequencies that correspond to a sine wave. These criteria lead to better choices of DFT windows for SNR and harmonic distortion calculations.

Sandia has developed an advanced operational control system approach, caged Graphical Programming, to design and operate robotic waste cleanup and other hazardous duty robotic systems. The Graphical Programming approach produces robot systems that are faster to develop and use, safer in operation, and cheaper overall than altemative teleoperation or autonomous robot control systems. The Graphical Programming approach uses 3-D visualization and simulation software with intuitive operator interfaces for the programming and control of complex robotic systems. Graphical Programming Supervisor software modules allow an operator to command and simulate complex tasks in a graphic preview mode and, when acceptable, command the actual robots and monitor their motions with the graphic system. Graphical Progranuning Supervisors maintain registration with the real world and allow the robot to perform tasks that cannot be accurately represented with models alone by using a combination of model and sensor-based control. This paper describes the Graphical Programming approach, several example control systems that use Graphical Programming, and key features necessary for implementing successful Graphical Programming systems.

Hazardous operations which in the past have been completed by technicians are under increased scrutiny due to high costs and low productivity associated with providing protective clothing and environments. As a result, remote systems are needed to accomplish many hazardous materials handling tasks such as the clean up of waste sites in which the exposure of personnel to radiation, chemical, explosive, and other hazardous constituents is unacceptable. Traditional remote manual operations have proven to have very low productivity when compared with unencumbered humans. Computer models augmented by sensing and structured, modular computing environments are proving to be effective in automating many unstructured hazardous tasks.

Testing and analysis of shock wave characteristics such as produced by detonators and ground shock propagation frequently require a method of measuring velocity and displacement of the surface of interest. One method of measurement is doppler interferometry. The VISAR (Velocity Interferometer System for Any Reflector) uses doppler interferometry and has pined wide acceptance as the preferred tool for shock measurement. An important asset of VISAR is that it measures velocity and displacement non intrusively. The conventional VISAR is not well suited for portability because of its sensitive components, large power and cooling requirements, and hazardous laser beam. A new VISAR using the latest technology in solid state lasers and detectors has been developed and tested. To further enhance this system's versatility, the unit is fiber optic coupled which allows remote testing, permitting the VISAR to be placed over a kilometer away from the target being measured. Because the laser light is contained in the fiber optic, operation of the system around personnel is far less hazardous. A software package for data reduction has also been developed for use with a personal computer. These new advances have produced a very versatile system with full portability which can be totally powered by batteries or a small generator. This paper describes the solid state VISAR and its peripheral components, fiber optic coupling methods and the fiber optic coupled sensors used for sending and receiving laser radiation.