Andreas Schellenberg

A SOFTWARE FRAMEWORK FOR HYBRID SIMULATION OF LARGE STRUCTURAL SYSTEMS

Hybrid simulation is a versatile, powerful and economically viable experimental method. In particular, through the use of numerical partitioning concepts, it should be possible to test portions of a large structure in one of more laboratories, and to simulate the remainder of the structure/foundation system numerically. Considerable efforts have been expended to develop such capabilities, but their adoption has been hampered by the absence of a common framework for conducting hybrid simulation across various laboratories and computing environments. Therefore, based on object-oriented software design methodologies, a set of interrelated software classes is described herein, which in their unity form a network-enabled framework for integrating experimental testing with standard structural analysis frameworks. In this paper, the open-source Framework for Experimental Setup and Control (OpenFresco) is used in combination with the object-oriented finite element software framework OpenSees (Open System for Earthquake Engineering Simulation) and the graphical user interface OpenSees Navigator. This enables an unlimited number of users to connect multiple laboratories throughout the world, to define specialized experimental setups, to implement advanced integration operators and to utilize high performance control methodologies, in a highly modular and scalable fashion. By utilizing both the object-oriented experimental and finite element software frameworks, response features that are frequently encountered in large structural systems (e.g., geometric nonlinearities, three-dimensional effects, multiple support excitation and soil-structure interaction) can be investigated by incorporating them into the analytical model. One of the most advantageous applications to incorporate geometric nonlinearities into the analytical model is in large displacement tests where a structure is tested until collapse. Contrary to shaking table testing, no large physical masses have to be present in the experimental part of a hybrid simulation. Another example presented illustrates the ability to account for multiple support excitations in long span structures. By using a standardized framework for communication, simulation and control, such tests can be easily set up in various laboratories to test large structures not possible by other means.