Font Size: a A A

Seismic Reliability Analysis Of The Free Spanning Pipeline Based On Neural Network

Posted on:2019-09-06Degree:MasterType:Thesis
Country:ChinaCandidate:Y ZengFull Text:PDF
GTID:2381330623962562Subject:Water conservancy project
Abstract/Summary:PDF Full Text Request
In recent years,the construction process of submarine oil and gas pipeline is speeding up.In the course of laying and running of sea pipe,the uneven seabed,the scour of current and wave to submarine pipeline and seabed may cause the overhanging of submarine pipeline.Under the action of earthquake load,the free spanning pipelines will produce larger stress and strain value,which may cause serious economic loss and environmental pollution.Seismic problem is an inevitable factor in the process of submarine pipeline design.Therefore,it is urgent to explore the factors that affect the stress and strain of the spanning pipe under earthquake action and to study the practical engineering methods for reliability evaluation.Firstly,ANSYS was used as the platform for seismic time-history analysis of the spanning pipe,the steel pipe with Ramberg-Osgood as the constitutive model was established.Considering the constraint effect of soil on the pipe,and the dynamic water force acting on the suspended span was simplified as the additional water mass added to the model.The influencing factors of the suspended sea-crossing pipe under earthquake action are studied.The calculation results show that the stress and strain of the pipe is affected by the soil property.When the soil in the buried section is clay,the stress and strain value of the pipe is the maximum.The stress-strain value of pipeline is proportional to span length,seismic acceleration and buried depth of pipeline.Then by using the artificial neural network in MATLAB,the sample data of the suspended sea-crossing pipe under seismic action is trained.The stress neural network and strain neural network of free spanning pipeline are obtained,and the error meets the requirements.The ultimate state function of the pipeline is obtained through the stress neural network and the strain neural network.Based on the stress and strain failure criteria,the reliability calculation methods of Monte Carlo method,JC method and central point method are compared.Finally,the reliability relationship and variation rule of pipeline parameters,span length,seismic acceleration and buried depth are obtained by Monte Carlo method.
Keywords/Search Tags:Free Spanning Pipeline, Earthquake, Neural Network, Limit State Function, Reliability, Monte Carlo Method
PDF Full Text Request
Related items