IV.

DISCUSS THE RELATIONSHIP OF GROUND MOTION, LOAD AND

STRUCTURAL MODELLING AND SEISMIC LOAD COMBINATION IN
RELATION TO THE DYNAMIC ANALYSIS OF STRUCTURE.

In implementing a performance-based design, it is required to select one or more

performance objectives associated with different severities of earthquake. To ensure that the

designed structure can meet these performance objectives, analytical techniques, which allow

one to predict degrees of damage in the various elements based on the inelastic demands

predicted for these elements, must be adopted. One method is the so-called pushover

approach which consists of performing a series of incremental nonlinear static analyses on

the structural model. It has been reported that if the response of the structure is dominated by

first mode, this approach can give a reasonable estimate of inelastic demand distributions and

damages in the structure; however, for long period structures, the estimate of damages made

using this approach is poor. The other method is the nonlinear dynamic time-history analysis

which is more complex to perform and time-consuming and is deemed to be able to give

accurate predictions if assumed constitutive relation for structural elements reflects their true

behavior.

In this study the seismic ground response is computed using the program which is a

two-dimensional nonlinear effective stress finite element program based on equations. In this

program the soil displacement and the pore pressure can be obtained directly from the

solutions of equations. In addition, viscous boundary accounting for two-phase nature of soil

is used to model the lateral infinite extent of soil stratum. The nonlinear dynamic analysis of

building in this study is performed using the program. This program is versatile and can
perform the static analysis, linear and nonlinear dynamic time-history analysis, modal

analysis and static push-over analysis.

It is known that there are infinite number of artificial earthquakes which can be

generated to envelop the same design spectrum. In order to investigate effect of these

earthquakes on the seismic responses of buildings, we generate ten artificial earthquakes

using the following number as the seed numbers for the random number generator

subroutine. It can be seen that the time to reach the maximum acceleration varies

significantly with the seed number, indicating that the seed number plays an important

role in generating the spectrum compatible artificial earthquakes.

In the dynamic analysis of a building located on top of an improved soil zone, if

the input ground motion is computed using 1D analysis, then the structural response will

be underestimated. Thus, for such a case, the surface ground motion must be obtained

using 2D analysis if the design ground motion is specified at the engineering bedrock.

Also, the results of currently adopted averaging process given in the code are affected by

how the samples are chosen and a clear guideline should be developed.
V. DISCUSS THE FRAMEWORK OF MATERIAL MODELLING OF
REINFORCED CONCRETE (WITHOUT TORSION).

A material model for the analysis of reinforced concrete surface structures is

developed. The constitutive model employs the smeared crack concept, i.e. only average

stresses are considered at an integration point. While uncracked concrete is modelled with a

plasticity approach, the paper focuses on the numerical treatment of cracked reinforced

concrete in a state of plane stress. Special attention is given to the modelling of the tension

stiffening effect, the re-orientation of the principal tensile strain direction and the

compressive strength of cracked concrete. 

When approached using nonlinear finite element techniques, structural analyses

generate, for real RC structures, large complex numerical problems. Damage is a major part

of concrete behavior, and the discretization technique is critical to limiting the size of the

problem. Based on previous work, the mu damage model has been designed to activate the

various damage effects correlated with monotonic and cyclic loading, including unilateral

effects. Assumptions are formulated to simplify constitutive relationships while still allowing

for a correct description of the main nonlinear effects.

A non-linear multi-fiber finite element model is developed for RC element under pure

torsion. In the elastic domain, the comparison of numerical results with the analytical

formulation is excellent. The warping phenomenon is fully represented at sectional level as

well as its influence on element deformations. Large rotation conditions have also been

investigated and included successfully in the proposed numerical model for RC members
subjected to torsion. Regarding the research significance, the modification proposed by the

authors for the tensile behavior of concrete and the parametric formulation developed by the

authors for effective wall thickness shows a reasonable and correct influence in predicting

the torsional response in the inelastic domain. Although the model is developed with a

displacement-based formulation, the sectional model proposed is general enough to apply in

any finite element model or formulation, such as force-based or flexibility-based.

Finally, with the proposed model, the torsional response can be obtained in a simple

manner with reasonable computational efficiency. Moreover, the same sectional

discretization and constitutive material approach has to be applied in a more general model,

which is able to consider the effect of combined loading shear-bending-torsion. For future

studies, this model may be further developed with a larger calibration study and other new

formulations for different reinforcement repartitions may be proposed.