24.12.2014 Views

Earthquake Engineering Research - HKU Libraries - The University ...

Earthquake Engineering Research - HKU Libraries - The University ...

Earthquake Engineering Research - HKU Libraries - The University ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

30;<br />

excitation. Also, the software was built following the object-oriented (OO) programming<br />

paradigm by using Or computer language.<br />

To trace and simulate the overall behavior of the structures subjected to earthquake<br />

excitation, various types of non-linear finite element analysis for reinforced concrete<br />

frame structures have been pursued and investigated over the last thirty years. Most of<br />

them can be divided into three categories: local models (microscopic or macroscopic<br />

finite element models), global models (member models), and semi-local models (fiber<br />

element models) (Miramontes et al. 1996, Meyer et al. 1991).<br />

Semi-local models stand between global models and local models (Miramontes et al.<br />

1996). A simplified kinematic hypothesis was adopted and equations of equilibrium are<br />

solved only for nodal points, like global models. By employing a cyclic constitutive<br />

relationship of the constituent materials, stress and strain were calculated at local level<br />

and consequently no hysteretic model at section or member level was needed, like local<br />

models. Fiber models exploited in this research fall into the category of semi-local<br />

models and have been investigated by many researchers (Taucer et al. 1991, Rubianobenavides<br />

1998). Despite many merits, fiber models had been considered to be too<br />

expensive for practical structural analysis. As personal computers have been getting<br />

cheaper and more computationally efficient, however, fiber models may become a<br />

rational and feasible tool for practical structural analysis. In this research, force-based<br />

finite fiber elements for RC frame structures and its applications were investigated.<br />

In addition to the precise nonlinear FE analysis of the overall structural responses,<br />

larger-scaled quick estimation of critical structural responses are convenient tools for preearthquake<br />

damage evaluation and emergent post-earthquake repair. In this research, the<br />

application of artificial neural networks (ANN") to estimate the critical structural<br />

responses was explored. <strong>The</strong> application of multiplayer perceptron (MLP) networks was<br />

studied and an augmented form of MLP was proposed to improve the modeling accuracy<br />

of classical MLP.<br />

CYCLIC STRESS-STRAIN RELATIONSHIP OF REINFORCED CONCRETE<br />

MATERIALS<br />

Reinforced concrete is actually a composite material, in which its constituent<br />

materials, reinforcing steel and concrete, are subjected to quite a different loading<br />

condition than that of the tests for the individual materials. Hence, another approach to<br />

the investigation of RC materials is to derive the average (smeared) behavior of steel and<br />

concrete of reinforced concrete directly from experiments on larger reinforced concrete<br />

specimens, such as the tests on panel element conducted at the <strong>University</strong> of Houston<br />

(Hsu 1993, Mansour et al. 2001). Such constitutive relations can be defined as reinforced<br />

concrete material models (Meyer et al. 1991).<br />

This study was aimed at analyzing RC/PC frame structures subjected to cyclic static<br />

loading or dynamic earthquake excitation. With this goal, five material models (Jeng<br />

2002) were employed for the finite element analysis to attain sufficient accuracy and to<br />

maintain computational efficiency as well. <strong>The</strong> first four models are shown in Figs. 1 to 4.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!