RBF-BASED MESHLESS METHOD FOR LARGE DEFLECTION OF

PLATES ON NONLINEAR FOUNDATIONS

Author :
Advisor : Husain Jubran Al-Gahtani
Date : 2015-01-05
Access : restricted

Abstract
Abstract : Modeling the interaction between the soil and the structural foundation is a
major concern in many engineering applications. The soil foundations very often
represent a complex medium and therefore it is extremely difficult to obtain
analytical-based solutions for this interaction. The problem becomes more difficult if
the structure is undergoing large deformation. Numerical methods are powerful
alternative tools to solve such complicated problems. There are two main categories
of these methods: element-based methods such as the finite element (FEM) and the
boundary element (BEM) and element free-based or mesh-less methods like radial
basis functions (RBF). The FEM and BEM have been studied extensively so far,
whereas the RBF still needs to be investigated for the use in various engineering
application areas. Free-element or free-mesh methods are advantageous for
problems that require frequent re-meshing such as those arising in non-linear
analysis. In this work, the use of multi-quadratic radial basis function (MQ-RBF) is
proposed to solve the problem of large deflection of thin elastic plates resting on
nonlinear foundations. The study covers several variables, including the plate shape,
the loading type, the boundary conditions, the foundation type and the foundation
parameter. Two types of foundation models (discrete and continuous) are adopted.
The discrete foundation model consists of Winkler, non-linear Winkler, Pasternak
and non-linear Winkler-Pasternak model whereas the Boussinesq model is used to
represent interaction of the plate with the continuous foundation. WALFRAM

MATHEMATICA software was utilized for the development of the computer codes.
The accuracy and efficiency of the method is demonstrated by comparing the
obtained solution with readily available and corresponding numerical solutions. From
the comparison, the obtained results from MQ-RBF method are close to the ones
obtained from FEM. As a main conclusion, the proposed meshless-based method
(MQ-RBF) can be utilized as a good alternative numerical method for the element
based numerical methods such as FEM and BEM in the solution of the large
deflection of thin elastic plates on nonlinear foundations. The developed computer
codes have been utilized to generate correlations between discrete and continuous
foundation parameters. The obtained correlations of the foundation parameters have
been verified through several examples which revealed that the Pasternak and nonlinear Winkler-Pasternak discrete models are the most in agreement with the
continuous foundation model.

ANALYSIS AND DESIGN OF HIGH STRENGTH CONCRETE

COLUMNS SUBJECTED TO COMPRESSION AND BIAXIAL
BENDING
An effort has been made in this work to develop interaction curves and contours
to be used for the analysis and design of HSRC columns. Based on the results of
this work, the following conclusions can be drawn:
1. The use of equivalent rectangular stress block for concrete under compression is
equivalent to the exact stress-strain relationships.
2. Concrete stress-strain relationship can be idealized by the equivalent rectangular
stress block for both NSC and HSC. A new rectangular stress block has been
proposed for HSC, because NSC stress block (assumed in ACI curves)
overestimates the column capacity for the case of HSC.
3. Mathematica is capable of developing short and efficient codes for generating
interaction diagrams for any circular and rectangular NSRC or HSRC columns.
4. Mathematica-generated interaction curves for NSRC columns have been
successfully validated using ACI curves, PCA software and the available data from
the literature.
5. Mathematica-generated interaction curves for HSRC columns have been
successfully validated using PCA software and the available data from the literature.
6. Both ACI and Mathematica-developed interaction curves are more conservative
than PCA software for NSRC columns.
7. The deviations between ACI/Mathematica-developed and PCA interaction curves
occur in the compression-controlled part of the curves.
8. Oztekins model is the best among all investigated modes for representing the
tressstrain relationship of HSC. The stress block parameters, computed based on
this model, have yielded results that are closer to the literature experimental data.
9. Bresler load contour method is proposed for biaxial bending columns, and it is
found conservative for such cases based on the experimental validation.

7.2 Recommendations
Finally, as is the case with any other computational-based research, there is a
scope for further enhancement of the study. Below are some of
recommendations that can be made:

the

1. The developed interaction curves are based on concrete stress-strain

relationships for unconfined concrete (neglecting the effect of the column tie bars).
Although this yields conservative design column capacity, it does not accurately
resemble the real case and therefore, more realistic results may be obtained if the
concrete confinement is accounted for.
2. The current literature survey shows that there is a lack of experimental data for
HSRC columns with higher steel ratios ( > 1.3%). More comprehensive literature
search should be performed to get the ones corresponding to medium to high values
of . Otherwise; new experiments have to be conducted to generate the missing
data.

EXPERIMENTAL INVESTIGATION AND NUMERICAL

MODELLING OF TWO DIMENSIONAL CHLORIDE IN
CONCRETE
Conclusions
An effort was made in this work to study the dependence of chloride diffusion on time,
chloride binding, and multi-directional ingress of chloride, which is normally not addressed by
the models based on Ficks 2nd Law of diffusion. Results have shown the significant effects of
these factors on chloride diffusion coefficient and therefore on time of initiation of reinforcement
corrosion.
Based on the outcomes of this study following conclusions can be drawn:
1. Chloride profiles plotted using the experimental data show that the total chloride
concentration is always greater than free chloride concentration, confirming the effect of
chloride binding in concrete. This binding effect was found to be more prominent in silica
fume concrete than OPC concrete due to the denser microstructure of the silica fume
concrete.
2. Chloride diffusion coefficient was found to decrease significantly with time due to the
densification of concrete microstructure. A sharp decrease in the chloride diffusion
coefficient was observed during the earlier exposure durations due to the faster rate of
hydration reactions.
3. Besides calculating chloride diffusion coefficient, the experimental data was utilized to
determine corrosion initiation time using a set of five alternate models. Comparison of
the results obtained using different models enabled to examine the effects of various
factors considered in the study.

4. The values of time to initiation of reinforcement corrosion calculated using threshold

chloride concentration set by ACI either in terms of free chloride or total chloride came
out to be almost similar. This confirms the adequacy of the both limits set by the ACI.
5. A chloride diffusion model (3rd Model) was proposed along with the approach for solving
it numerically incorporating the effects of time and chloride binding. The accuracy of this
model in calculating the chloride diffusion coefficient and corrosion initiation time was
compared with that of the other alternate models. It was observed that the proposed
model captured the interaction effect of chloride diffusion due to multi-directional
exposure besides capturing the effects of time and chloride binding. Therefore, the
proposed model can be used in service life prediction against chloride-induced corrosion
with more degree of accuracy.