Project # NCTRKE & NCTRMS
Research Team
- Michael Scott, Associate Professor, Oregon State (PI)
- Kevin Mackie, Assistant Professor, University of Central Florida (co-PI)
Research Abstract
PEER has invested many years of research in the development of validated/calibrated modeling technologies that enable the simulation of complex, three-dimensional, nonlinear, dynamic, coupled soil-foundation-structure-interaction problems. OpenSees, and the larger community that supports the framework, is now at a level of maturity that allows significant analytical research studies on performance-based earthquake assessment of large transportation network components, such as bridges. However, with this ability comes numerous key challenges. The first challenge is, given the proliferation of parameters necessary to describe nonlinear models, to efficiently assess the sensitivity of response to modeling and design parameters. The second challenge is to properly assess the reliability and resulting risk to the transportation network components by capturing all of the relevant uncertainties in the model. The third challenge is to extend the traditional notion of design optimization to complex systems and objectives, potentially defined in terms of performance-based limit states.
PEERʼs mission to support research in performance-based earthquake engineering (PBEE) requires the development, maintenance, and evolution of robust software modules within the OpenSees framework for computational simulation, reliability analysis, and optimization of structural systems. The high level software abstractions of the reliability and optimization components of OpenSees require significant refactoring, and in some cases complete redesign, in order to support advanced applications of these modules to current and future PEER projects. The objective of this proposal is to re-engineer a majority of the reliability and sensitivity modules in OpenSees in order to bring them to a level of modern software design that is comparable to that present in OpenSeesʼ core finite element analysis modules. This will enable the open-source paradigm for gradient-based applications. Previously, this paradigm has resulted in significant growth of the other OpenSees analysis modules and will meet the three aforementioned challenges in PBEE assessment of large transportation networks.
Research Outcomes
The basic functionality of the existing reliability and sensitivity modules in OpenSees will be addressed (see Phase I below) initially. This will enable users to begin using the syntax and a simple subset of tools while high level changes to the reliability and optimization modules are made. All sensitivity and reliability examples currently posted on the OpenSees webpage will be re-tooled to reflect these “Phase I” changes. These examples, as well as new examples, will be presented at OpenSees Days and posted on the OpenSees webpage. In addition, documentation of the examples will be posted online to accompany the input files.
The complete code base modified during Phase II will be checked in to the OpenSees source code repository. The modifications, class design, and suggested application programming interface for future development will be presented at the OpenSees Days and will be frequently discussed with the developers and researchers. Documentation of code base changes will be provided in a final report format. The ultimate deliverable will be the stable state of the reliability and sensitivity modules, which can then be more easily maintained along with the other objectoriented code in the OpenSees repository.
Research Impact
Thorough documentation and annotated examples of reliability and optimization analysis in OpenSees will educate practicing engineers on emerging performance-based analysis methods as wells as basic concepts of uncertainty in finite element analysis of structural and geotechnical components of transportation systems. In addition, with a functional open-source and object-oriented design, experts in specific subdisciplines within reliability, sensitivity, and optimization will be able to contribute to OpenSees without needing to re-implement any of the core technologies.