Automated Sensitivity Analysis for FORTRAN Models

by J. E. Horwedel, Martin Marietta Energy Systems, Inc, Oak Ridge, United States,
S. F. Railsback, Martin Marietta Energy Systems, Inc, Oak Ridge, United States,
H. I. Jager, Martin Marietta Energy Systems, Inc, Oak Ridge, United States,



Document Type: Proceeding Paper

Part of: Water Resources Planning and Management and Urban Water Resources

Abstract: For complex computer models used in engineering design and assessment, a cost-efficient procedure for identifying the parameters that are important to a given model prediction is a necessity. The GRESS code was developed to meet this need. GRESS employs a precompiler to add derivative-taking capabilities to FORTRAN computer programs. Early implementations of GRESS were used to calculate sensitivities of model results to user-selected input parameters, but the present version of GRESS includes the ability to calculate and report the sensitivities of model results to all input data. This valuable extension of the GRESS technology makes it possible to rapidly perform a comprehensive sensitivity analysis of FORTRAN models. This paper describes GRESS and its application to a dissolved oxygen (DO) simulation model.

Subject Headings: Sensitivity analysis | Computer models | Computer software | Simulation models | Dissolved oxygen | Data processing | Model analysis

Services: Buy this book/Buy this article

 

Return to search