Selecting the appropriate earthquake record to perform dynamic analysis of the structure using genetic algorithm

Dissertation for Master's Degree

Field: Civil Engineering

Structural Orientation

Abstract:

In this paper, a new approach for optimal selection of acceleration maps and their scaling to perform dynamic time history analysis, to achieve the average response spectrum that has a suitable match and a small distance from the target spectrum and represents the expected earthquake of the building, is presented by binary genetic algorithm and natural numbers. Due to the difference in the nature of accelerometers and scale coefficients, the genetic algorithm presented in this article is a hybrid (with two chromosomes). The presented algorithm is able to create a new generation of people from among the infinite set of earth movement records, during a process in which natural selection, mating, and mutation take place, and continue this process until a person with desirable characteristics is obtained. One of the most important factors in the accuracy and efficiency of these programs is the correct estimation of their parameters. If these parameters are calculated correctly, the difference between the average response spectrum and the design spectrum is greatly reduced. Due to the relatively large number of these parameters, the application of trial and error based methods largely depends on the user's skill, the presented hybrid genetic algorithm program can solve this defect. This program has two genetics that are executed simultaneously and results close to the optimal solution. This program itself is able to provide the user with the range of optimal coefficients and values ??of covariance and mutation of each chromosome.

Key words: time history analysis, response spectrum, design spectrum, genetic algorithm, hybrid algorithm

Chapter 1

(Overview and background of the research)

1-1 Introduction

Plateau of Iran earthquake history It has a long rise, and the study of ancient history proves the occurrence of earthquakes in three thousand years before Christ. In a historical survey, Ambarsez has extracted and analyzed the history of nearly six thousand earthquakes that have occurred in this land since two thousand years ago from historical sources. These results show that the active areas in different periods more or less coincide with each other.

Given the fact that Iran is located on several earthquake faults and the unstable buildings that we witness every year being destroyed due to earthquakes, we must look for ways to solve this problem. Earth movement during an earthquake can cause severe damage to buildings and equipment inside them. When the acceleration, speed, and displacement of the ground are applied to the structure, in most cases, the strengthening of these movements causes the creation of forces and displacements in the structure. Many factors affect the ground movement and their strengthening. In order to investigate the behavior of a structure and its safe and economic design, it is necessary to take into account the effect of these factors[5].

Evaluating and recognizing earthquakes that may occur in the future is one of the important issues of earthquake and structural engineering, which requires knowing and predicting a possible earthquake and its characteristics in the region, as well as knowing the behavior of the structure under this earthquake. In the dynamic analysis methods, the lateral force of the earthquake is obtained by using the dynamic reflection that the structure shows as a result of the earth movement caused by the earthquake. These methods include "spectral analysis" method and "time history analysis" method. The ground motion, which is used in dynamic analysis, must have at least the design earthquake conditions. The effects of the earth's movement are determined in one of two ways: "acceleration reflection spectrum" or "acceleration time history" [3]. For the acceleration reflection spectrum, you can use the spectrum of the standard design or the spectrum of the special design of the building according to the regulations.

Generally, structures enter the nonlinear range when they are subjected to strong earthquakes, for this reason, the nonlinear analysis of the structure's time history is important. Nonlinear time history analysis[1] is more common in seismic analysis and structural design. Regulations related to seismic isolation structures include provisions governing nonlinear time history analyses. It has been about two decades that in Europe and America, the regulations governing the analysis of time history have been described. Although the seismic risk in a place (site) for design purposes is presented by the design spectrum[2], almost all design codes require a more accurate method for scaling[3] and selecting the earthquake time history according to the design spectrum.

Several methods for time history scaling have been proposed. These methods include: frequency domain methods [4] and time domain methods [5], which in frequency domain methods, the frequency value is manipulated to match the ground motion record. In the time domain method, the amplitude value of the ground motion record is scaled. Regardless of these methods, in almost all existing theories, earthquake selection and scaling processes are separate and distinct according to the design spectrum [30].

The selection of ground motion in dynamic analysis is very important because the motions have a significant impact on the analysis result as well as the design output. Therefore, it is very important to obtain a set of ground motions with accurate estimation of the seismic response of the structure based on the local seismic hazard where the structure is located. Recently, access to online digital data has increased, as well as access to real earthquake accelerometers. Although there are many differences depending on the conditions of the accelerometer recording station, the magnitude of the source earthquake, the recording location of the ground motion record, the type of fault, the type of soil, the duration of the movements, and the spectral characteristics.

The main goal of this research is to select a suitable combination of earthquake records in a specific location (site) that matches the spectrum of the plan and has the least difference with it. The specifications necessary to scale the earthquake record are numerical variables that are applied by the user in a certain range. Therefore, the phase and shape of the response spectrum [6] of the earthquake remain intact. Unlike conventional methods for scaling, in which a set of earthquake records is predetermined and then the scaling is adapted to the design spectrum, the presented method is able to search from a set including thousands of earthquake records and introduce a subset of records that match the target spectrum [7], which is done by genetic algorithm [8]. Genetic algorithm is the best evolutionary method and does not need gradient information. The implementation of genetic algorithms starts with the creation of an initial population[9] of chromosomes[10]. Then these initial structures are evaluated and they are given the opportunity to reproduce according to their merit. Usually, the degree of desirability of the solutions is determined according to the current population. The structure of the genetic algorithm is such that it must have at least one member in its initial population. This member is responsible for the production of a new population and its development to meet the final condition.

In the genetic algorithm, the first stage of evolution is the production of individuals. In these algorithms, after generating the initial population, it is time to select two parents[11] and combine[12] them in the form of one or two children[13] and finally mutation[14] of the children. New children replace one of the weaker individuals in the population.

Genetic algorithm is an iterative procedure involving a population of fixed size. Each individual of this population is presented according to a limited string of symbols, which are referred to as the genome. Each of these genomes encodes a possible solution in the problem space. The problem space is interpreted as the search space, which includes all possible solutions to the problem. Generally, genetic algorithms are used for problems whose search space is very large and the usual search methods are not applicable for them.

According to the main goal of this research, which is to obtain the set of ground motions by the genetic algorithm, which is in accordance with the spectrum of the design of the 2800 regulation of Iran[3], the genetic algorithm can, from among the community of real earth maps for a specific area with a specific soil type, select and scale to obtain a suitable combination of ground motions, in accordance with the spectrum of the introduced plan. in the 2800 regulations of Iran.

The research process is that, first, based on the characteristics of different earthquakes that have occurred in the world based on the type of soil, and the distance of the station from the source of the earthquake, a database [15] is selected, then these data will be expanded for different types of soil, and then the genetic algorithm will select the response spectrum of the combination of these records by comparing the spectrum of the regulation, and if the convergence criterion [16] is satisfied, that set [17] It is selected as an individual for the earthquake record. The results show that the genetic algorithm produces accurate results in selecting and scaling the main earthquake sets according to the design spectrum.