Publication: Multiobjective optimization of axially loaded non- prismatic columns
Abstract
Son otuz yıldır yapısal dizayn optimizasyonu en etkin araştırma alanlarından biridir. Yapısal dizayn optimizasyonu yardımıyla yapının ağırlığının yada kullanılan malzemenin azaltılmasının yanında dayanım, burkulma ve yapıyla ilgili diğer özellikler verimli bir şekilde geliştirilebilir. Burkulma problemlerine genel olarak yapı mühendisliğinde, özellikle de sıkıştırma üniteleri, kabuk yapılar veya ince cidarlı basınç kaplarında karşılaşılır. Burkulmayla ilgili dizayn yapının statik dayanımına göre daha zor olduğundan, dizayn optimizasyon metodları özellikle bu tür problemler için faydalıdır. Burkulmayı göz önüne alan yapısal optimizasyon problemleri ilk araştırmalardan oluğundan bu problemlerle ilgili çok çalışma vardır. Bu tezde, burkulma yükü altındaki prizmatik olmayan kolonlar çeşitli kesit şekilleri için incelenmiştir. Bu problemde her farklı durum için iki kriter göz önüne alınmıştır: Ağırlığı minimum ve Yükü maksimum yapmak. Ağırlıklı fonksiyon metodu kullanılarak, Pareto setler bulunmuştur. Pareto setler arasından parametrelerdeki değişimden en az etkilenen nokta optimum nokta olarak seçilmiştir. Bu metodla bulunan sonuçlar daha dayanıklı ve güvenilir dizaynlar oluşturmaktadır. Sonuçlar hassasiyet ve dayanıklılık açısından Global Criterion metodunun sonuçları ile karşılaştırılmıştır.
MULTIOBJECTIVE OPTIMIZATION OF AXIALLY LOADED NON-PRISMATIC COLUMNS Structural design optimization has been an active research field during the past three decades. By means of the design optimization methods, not only the weight and materials of structures can be reduced, but also the strength, buckling stability, and other performances of structures can be improved efficiently. Buckling problems extensively exist in structural engineering, in particular, in the design of structures such as compression members, shells and thin-walled pressurized containers. Since the design of buckling performance is more difficult than static strength of structures, the design optimization methods are particularly useful for this kind of design problem. Since the early research of structural design optimization with buckling constraint there has much work on these problems. In this thesis, non-prismatic columns under buckling loads are investigated for various cross-sectional shapes. For each case, two objectives are taken into account for this multicriteria problem: Minimizing Weight and Maximizing Load. By using weighted function method, the optimum Pareto sets is found. The optimum point is chosen by finding the least sensitive point to variable changes among the obtained Pareto sets. The results obtained from this methodology yield highly insensitive, robust and reliable designs. Results are compared with Global Criterion Method from point of sensitivity and robustness.
MULTIOBJECTIVE OPTIMIZATION OF AXIALLY LOADED NON-PRISMATIC COLUMNS Structural design optimization has been an active research field during the past three decades. By means of the design optimization methods, not only the weight and materials of structures can be reduced, but also the strength, buckling stability, and other performances of structures can be improved efficiently. Buckling problems extensively exist in structural engineering, in particular, in the design of structures such as compression members, shells and thin-walled pressurized containers. Since the design of buckling performance is more difficult than static strength of structures, the design optimization methods are particularly useful for this kind of design problem. Since the early research of structural design optimization with buckling constraint there has much work on these problems. In this thesis, non-prismatic columns under buckling loads are investigated for various cross-sectional shapes. For each case, two objectives are taken into account for this multicriteria problem: Minimizing Weight and Maximizing Load. By using weighted function method, the optimum Pareto sets is found. The optimum point is chosen by finding the least sensitive point to variable changes among the obtained Pareto sets. The results obtained from this methodology yield highly insensitive, robust and reliable designs. Results are compared with Global Criterion Method from point of sensitivity and robustness.