MODARES TECHNICAL AND ENGINEERING
Year:2004 | Volume:- | Issue:14|Papers Count:9

Prediction of soil engineering behavior, based on previous experiments and observations, plays an important role in geotechnical engineering. So, to cover the subject, many researchers have focused on the mathematical and statistical models during the last decades. In this paper, multi-layer perceptron (MLP), a well-known method in Artificial Neural Networks, is used for modeling stress- strain behavior of silty sands, slop stability analysis of river bank and design of concrete liner of water tunnels. In addition, MLP design and factor affecting its performance have been discussed briefly. Since all different kinds of problems have been covered (Modeling, Analysis and Design), authors see immense potential for the application of well-trained neural networks in geotechnical engineering.

In this research, two types of steel frames with saddle-like connection were tested. They were: a) unbraced frames with modified Khorjeeni connection and, b) braced frames with conventional Khorjeeni connection. Four half scale models of the above mentioned frames were tested in laboratory under semi-dynamic cycles of loading. From these experiments the ductility factor,µ, for unbraced frames were found to be at least equal to 2.4 and for braced frames to be equal to 2.0.

An analytical method has been derived for the fluid surface sloshing phenomenon of cylindrical water tanks. The contained liquid is incompressible and invoiced with small amplitude motion. Although sloshing modes of any desired order could be included with the method, here a simplified three degrees of freedom model is applied. It is based on rigid wall structure assumption and thus the hydrodynamic pressure is found. The model is based on the lumped mass concept. Model analysis method is shear force response history and then the maximum base shears for each. The results are compared with the two-dof Housner model and with the Iranian Earthquake Code (Standard 2800) model. It is shown that the Iranian Code model results are in conservative under certain circumstances or base excitations. Besides, the,2nd slouching mode has to be taken into account as well if the horizontal load resisting structure deems to be slender. Thus a new proposal with elevated water tanks in the new Iranian Earthquake Code is presented.

Soil contamination with petroleum hydrocarbons (TPH) such as aromatic hydrocarbons is one of the major problem that environment is facing today. In this research, petroleum contaminated soils around the Tehran Refinary have been examined to study the anthracene biodegradation rate using different microorganisms.Inoculum of microorganisms were adapted in mineral mediums with four types of carbon sources (Anthracene, Naphtalene, Toluene or mixture of all) in separate reactors. Grown microorganisms in reactors were named mixA, mixN, and mixANT based on carbon sources. Biodegradation of anthracene was investigated by using above mentioned group of microorganisms. Cells of microorganisms were inoculated in liquid mineral medium at 0.65 and 70 mg/1 of anthracene for the investigation of biodegradation potential. MixN showed 100%, and 98% respectively removal for 0.65 and 70 mg/1. Then mixN was selected for the investigation of biodegradation rate. Specific biodegradation rate of different concentration of anthracene by mix N was determined in supersaturated mediums of anthracene. Community of inoculated microorganisms in all concentrations of this stage was 9x107 N/100ml and specifice biodegradation rate of 1,5,10,50 and 100mg/1 of anthracene were 0.0514, o.o686, 0.0563, 0.1238 and 0.0384 day-1 respectively. The highest specific biodegradation rate was achieved at a level of 50mg/1.

Depth of water at the upstream of pier will increase when a bridge is constructed. The difference between water depth before and after the bridge construction is usually termed as afflux.   The purpose of this paper is to apply and evaluate the abilities of Artificial Neural Networks (ANN) to predict this phenomena. Mapping between input data (discharge, normal depth, length to width ratio of bridge pier, contraction ratio and angle of pier axis) and output data (afflux) has been provided by a Radial Basis Function (RBF) ANN based on imperical data produced by experimental measurements. The results show that a RBF artificial neural network including two hidden layers can predict intelligently the afflux and its performance is much better than other conventional approaches.

A separated plastic strength surface is used for second order plastic analysis of frames under week-axis bending effect of columns. This surface considers the interaction of axial force and bending in reduction of member stiffness. Also, two functions for reducing the stiffness are suggested. In order to show the ability of these functions, a variety of frames with weak-axis bending of columns are analyzed. These have been done by authors computer program. The results demonstrated that this plastic strength surface gives little error for week-axis bending effect.

Structural joints in precast concrete structures have a determinant role in strength, stability and reliability of such structures. Therefore it is necessary to investigate their behavior under various loading conditions. In this paper, numerous models of structural joints with different shapes and number of shear keys are selected and analyzed using finite element method and the results are compared to some experimental works and results obtained from codes of practices. Also, the effects of depth and number of shear keys on the reduction of shear stress in this type of joints are studied.