Open Access Open Access  Restricted Access Subscription Access

Optimal Tuning of Thermomechanical Hybrid Simulation Parameters



According to current engineering practice, the fire performance qualification of structural elements relies on single component tests, where specimens are exposed to standard fire curves. As a result, the interactions between elements within the overall structural assembly are lost. This situation has motivated researchers to extend the hybrid simulation method, which has been deeply investigated in the seismic domain, to structures-in-fire testing. By linking numerical and physical substructures, hybrid simulation offers a flexible, cost-effective approach. Purely mechanical hybrid simulation tests are often performed slower than real-time (extended time scale), dictated by actuator capacity limitations, quality of displacement tracking, and synchronization among computational and experimental drivers. This practice is generally not acceptable when a rate-dependent response of the specimen is expected, as is often the case in fire (e.g. creep strain occurs when steel is subjected to high temperatures). In this scenario, the proper selection of the testing time scale must balance the laboratory testing capacity requirements (operating too close to real-time compromises the test stability if the equipment is not suited for this) and the expected rate-dependency of the structural response (operating at an extended time scale could reduce the solution accuracy if the specimen undergoes excessive creep). From this perspective, it is crucial to accurately tune the testing time scale to minimize experimental approximation. Thus the optimal tuning of the time integration scale in thermomechanical hybrid simulation (TMHS) is presented herein. Furthermore, updates to the TMHS experimental element, newly implemented in the OpenFresco hybrid simulation middleware, provide the ability to fully simulate a TMHS test prior to experimentally substructuring the physical specimen in the laboratory

Full Text: