skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: A Global Sensitivity Analysis Methodology for Multi-physics Applications

Abstract

Experiments are conducted to draw inferences about an entire ensemble based on a selected number of observations. This applies to both physical experiments as well as computer experiments, the latter of which are performed by running the simulation models at different input configurations and analyzing the output responses. Computer experiments are instrumental in enabling model analyses such as uncertainty quantification and sensitivity analysis. This report focuses on a global sensitivity analysis methodology that relies on a divide-and-conquer strategy and uses intelligent computer experiments. The objective is to assess qualitatively and/or quantitatively how the variabilities of simulation output responses can be accounted for by input variabilities. We address global sensitivity analysis in three aspects: methodology, sampling/analysis strategies, and an implementation framework. The methodology consists of three major steps: (1) construct credible input ranges; (2) perform a parameter screening study; and (3) perform a quantitative sensitivity analysis on a reduced set of parameters. Once identified, research effort should be directed to the most sensitive parameters to reduce their uncertainty bounds. This process is repeated with tightened uncertainty bounds for the sensitive parameters until the output uncertainties become acceptable. To accommodate the needs of multi-physics application, this methodology should be recursively applied tomore » individual physics modules. The methodology is also distinguished by an efficient technique for computing parameter interactions. Details for each step will be given using simple examples. Numerical results on large scale multi-physics applications will be available in another report. Computational techniques targeted for this methodology have been implemented in a software package called PSUADE.« less

Authors:
;
Publication Date:
Research Org.:
Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
Sponsoring Org.:
USDOE
OSTI Identifier:
1036842
Report Number(s):
UCRL-TR-227800
TRN: US201207%%23
DOE Contract Number:
W-7405-ENG-48
Resource Type:
Technical Report
Country of Publication:
United States
Language:
English
Subject:
71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSICS; 97 MATHEMATICAL METHODS AND COMPUTING; COMPUTERS; IMPLEMENTATION; PHYSICS; SENSITIVITY ANALYSIS; SIMULATION

Citation Formats

Tong, C H, and Graziani, F R. A Global Sensitivity Analysis Methodology for Multi-physics Applications. United States: N. p., 2007. Web. doi:10.2172/1036842.
Tong, C H, & Graziani, F R. A Global Sensitivity Analysis Methodology for Multi-physics Applications. United States. doi:10.2172/1036842.
Tong, C H, and Graziani, F R. Fri . "A Global Sensitivity Analysis Methodology for Multi-physics Applications". United States. doi:10.2172/1036842. https://www.osti.gov/servlets/purl/1036842.
@article{osti_1036842,
title = {A Global Sensitivity Analysis Methodology for Multi-physics Applications},
author = {Tong, C H and Graziani, F R},
abstractNote = {Experiments are conducted to draw inferences about an entire ensemble based on a selected number of observations. This applies to both physical experiments as well as computer experiments, the latter of which are performed by running the simulation models at different input configurations and analyzing the output responses. Computer experiments are instrumental in enabling model analyses such as uncertainty quantification and sensitivity analysis. This report focuses on a global sensitivity analysis methodology that relies on a divide-and-conquer strategy and uses intelligent computer experiments. The objective is to assess qualitatively and/or quantitatively how the variabilities of simulation output responses can be accounted for by input variabilities. We address global sensitivity analysis in three aspects: methodology, sampling/analysis strategies, and an implementation framework. The methodology consists of three major steps: (1) construct credible input ranges; (2) perform a parameter screening study; and (3) perform a quantitative sensitivity analysis on a reduced set of parameters. Once identified, research effort should be directed to the most sensitive parameters to reduce their uncertainty bounds. This process is repeated with tightened uncertainty bounds for the sensitive parameters until the output uncertainties become acceptable. To accommodate the needs of multi-physics application, this methodology should be recursively applied to individual physics modules. The methodology is also distinguished by an efficient technique for computing parameter interactions. Details for each step will be given using simple examples. Numerical results on large scale multi-physics applications will be available in another report. Computational techniques targeted for this methodology have been implemented in a software package called PSUADE.},
doi = {10.2172/1036842},
journal = {},
number = ,
volume = ,
place = {United States},
year = {Fri Feb 02 00:00:00 EST 2007},
month = {Fri Feb 02 00:00:00 EST 2007}
}

Technical Report:

Save / Share:
  • Reduction in safety margin can be expected as passive structures and components undergo degradation with time. Limitations in the traditional probabilistic risk assessment (PRA) methodology constrain its value as an effective tool to address the impact of aging effects on risk and for quantifying the impact of aging management strategies in maintaining safety margins. A methodology has been developed to address multiple aging mechanisms involving large numbers of components (with possibly statistically dependent failures) within the PRA framework in a computationally feasible manner when the sequencing of events is conditioned on the physical conditions predicted in a simulation environment, suchmore » as the New Generation System Code (NGSC) concept. Both epistemic and aleatory uncertainties can be accounted for within the same phenomenological framework and maintenance can be accounted for in a coherent fashion. The framework accommodates the prospective impacts of various intervention strategies such as testing, maintenance, and refurbishment. The methodology is illustrated with several examples.« less
  • As modeling efforts expand to a broader spectrum of areas the amount of computer time required to exercise the corresponding computer codes has become quite costly. This costly process can be directly tied to the complexity of the modeling, which makes the relationships among the input variables not mathematically tractable. In this setting it is desired to perform sensitivity studies of the input-output relationships. A variety of situations require that decisions and judgments be made in the face of uncertainty, such as lack of knowledge about probability distributions associated with input variables, different hypothesized future conditions, or different strategies associatedmore » with a decision making process. In this paper a generalization of Latin hypercube sampling is given that allows these areas to be investigated without making additional computer runs.« less
  • As part of the Site Specific Spectra Project, this report seeks to identify the sources of and minimize uncertainty in estimates of seismic hazards in the Eastern United States. Findings are being used by the Nuclear Regulatory Commission to develop a synthesis among various methods that can be used in evaluating seismic hazard at the various plants in the Eastern United States. In this volume, one of a five-volume series, we discuss the application of the probabilistic approach using expert opinion. The seismic hazard is developed at nine sites in the Central and Northeastern United States, and both individual experts'more » and synthesis results are obtained. We also discuss and evaluate the ground motion models used to develop the seismic hazard at the various sites, analyzing extensive sensitivity studies to determine the important parameters and the significance of uncertainty in them. Comparisons are made between probabilistic and real spectral for a number of Eastern earthquakes. The uncertainty in the real spectra is examined as a function of the key earthquake source parameters. In our opinion, the single most important conclusion of this study is that the use of expert opinion to supplement the sparse data available on Eastern United States earthquakes is a viable approach for determining estimted seismic hazard in this region of the country. 29 refs., 15 tabs.« less
  • The division uses the project appraisal methodology to make decisions about how to allocate federal funds to candidate R and D projects. The division is responsible for over 30 research and development (R and D) projects. Due to the large number of projects and the great variety in projects, funding decisions would be difficult without a methodology. The methodology scores R and D projects according to the following eight criteria: energy savings, technical risk, cost, commercial potential, uniqueness, resource availability, environmental impact, and legal, social, and institutional effects. Projects with high scores are deemed more favorable for funding than projectsmore » with low scores. Two issues have been raised concerning the methodology. The first issue relates to the question: how accurate must project data be in order to make a reasonably sound assessment of a project's merit relative to other candidate projects. An understanding of the sensitivity of scores to data inputs would enable the division to set priorities for refining and validating data inputs. The second issue pertains to the question: which elements of the methodology should be altered to reflect changes in the division's mission. The division's mission has been changing rapidly over the last two years as a result of reorganization and evolution of energy policy. Further changes are anticipated in the future. Thus, it was determined that a sensitivity analysis should also be conducted on the energy savings weighting factors.« less