ارزیابی عملکرد الگوریتم‌های فراابتکاری در طراحی بهینه‌ی شمع‌های سازه‌ی نگهبان ترانشه‌های عمیق

در نوشتار حاضر به بررسی تاثیر الگوریتم‌های فراابتکاری در طراحی بهینه‌ی شمع‌های سازه‌ی نگهبان ایستگاه شماره‌ی 3 متروی تبریز پرداخته شده است. برای این منظور، ابتدا ترانشه‌ی موردنظر در نرم‌افزار اُپن‌سیس مدل‌سازی و فرآیند گام به گام خاک‌برداری آن، مطابق روند اجرایی، شبیه‌سازی شده است. چهار الگوریتم فراابتکاری متداول، یعنی الگوریتم‌های مبتنی بر جغرافیای زیستی، ژنتیک، ازدحام ذرات، و زنبورعسل برای طراحی بهینه استفاده شده‌اند، تا علاوه‌بر مقایسه‌ی عملکرد هر یک در حل مسئله‌ی مذکور، احتمال دستیابی به بهترین پاسخ افزایش یابد. نتایج به‌دست‌آمده حاکی از عملکرد بسیار خوب الگوریتم ژنتیک نسبت به سایر الگوریتم‌های استفاده‌شده در دستیابی به طرح بهینه بوده است. به‌منظور بررسی بهتر، توزیع تنش خاک در اطراف سازه‌ی نگهبان و همچنین تغییرشکل المان‌های شمع بررسی شده‌ و نتایج نشان داده‌اند که استفاده از مهار متقابل جهت ایجاد تعادل تغییرشکل‌های قسمت فوقانی و تحتانی شمع‌ها و کاهش عمق مدفون المان‌های شمع ضروری به‌نظر می‌رسد.

evaluation of the efficiency of metaheuristic algorithms in the optimal design of pile wall retaining systems

the effectiveness of the application of metaheuristic algorithms in the optimal design of retaining structures is investigated in this paper. for this purpose, an ongoing tabriz metro station project with a deep excavation pit is selected here as a case study. the retaining system of the project consists of secant pile walls supported by a layer of struts. the piles have a circular section consisting of reinforced concrete cores covered by steel sleeves, and the struts are made of steel rectangular hollow sections. a detailed finite element model is developed in the opensees platform, including all the construction processes, in order to perform static analyses. four different metaheuristic algorithms, namely genetic, particle swarm optimization, bee, and biogeography-based algorithms, are chosen for the optimization problem. the pile external diameter, the steel tube stiffness, the number of longitudinal bars inside the concrete core and their diameters, the center-to-center spaces of the pile elements, the dimensions of structs and their center-to-center spaces, the location of the structs in depth and the buried depth of pile elements are selected as optimization variables. the total cost of the retaining system is considered as an objective function that should be minimized in the design space of the variables. for optimization purposes, an integration of the opensees software with the matlab platform is done to join the modeling space with the mentioned optimization algorithms. the number of iterations for each run is assumed to be 400, which is also considered a termination criterion. the optimization process is performed 50 times, and the best response is reported here. the results demonstrate an excellent performance of the genetic algorithm in obtaining the optimum solution with respect to the other three considered algorithms. it exhibits a proper standard deviation and convergence rate in producing the optimum response. it is shown that the soil stress is increased in the depth where struts are installed, while they are reduced near the ground level, where the deflection of piles creates an active situation for the soil. this is true considering the results of all algorithms. proceeding with the excavation phase increases the soil stress as well as the pile deformation. it can also be obtained that providing a layer of strut seems necessary for reducing pile movements as well as their buried depth.