OpenSees software architecture for the analysis of structures in fire

摘要

Computational modeling of structures subjected to extreme static and dynamic loads (such as snow, wind, impact, and earthquake) using finite-element software are part of mainstream structural engineering curricula in universities (at least at graduate level), and many experts can be found in industry who routinely undertake such analyses. However, only a handful or institutions around the world teach structural response to fire (at any level) and only a few of the top consulting engineers in the world truly specialize in this niche area. Among the reasons for this are the lack of cheap and easily accessible software to carry out such analyses and the highly tedious nature of modeling the full (often coupled) sequence of a realistic fire scenario, heat transfer to structure and structural response (currently impossible using a single software). The authors in this paper describe how finite-element software can be extended to include the modeling of structures under fire load. The added advantage of extending existing finite-element codes, as opposed to creating fire-specific applications, is due to ability to perform multihazard type analysis, e.g., fire following earthquake. Due to its open source nature and object-oriented design, the OpenSees software framework is used for this purpose. In this work, the OpenSees framework, which was initially designed for the earthquake analysis of structures, is extended by the addition of new concrete classes for thermal loads, temperature distributions across element cross sections, and material laws based on Eurocodes. Through class and sequence diagrams, this paper shows the interaction of these classes with the existing classes in the OpenSees framework. The performance of this development is tested using benchmark solutions of a single beam with finite stiffness boundary conditions and a steel frame test. The results from OpenSees agree well with analytical solutions for the benchmark problem chosen and provide reasonable agreement with the test. The experience with OpenSees so far suggests that it has excellent potential to be the basis of a unified software framework for enabling computational modeling of realistic fires, and further work is continuing towards the achievement of this goal. The extensions made to OpenSees described in this work, in keeping with the open source ideals of the framework, have been included in the current OpenSees code and are available for researchers and practicing engineers to test, develop, and use for their own purposes.