Topology Optimization of Square Concrete-Filled Steel Tube Truss Structures Based on Quasi-Full Stress Genetic Algorithm
Overview of the Literature
This paper by Xie Jun, Guo Fei, Zhu Shouqin, Liang Jinxiao, Yan Jie and Dong Jie, published in the Journal of University of Jinan (Natural Science Edition) in 2023, addresses the discrete variable topology optimization problem for square concrete-filled steel tube (CFST) truss structures. The authors propose a topology optimization method based on a quasi-full stress genetic algorithm. The method uses a genetic algorithm to randomly generate initial topological configurations, then applies a heuristic algorithm to check and correct the configurations. The optimization variables include the square steel tube section type, the topology variables, and the concrete strength grade. The objective is to minimize the total structural cost. The quasi-full stress genetic algorithm generates part of the initial population from the optimal solution of the quasi-full stress algorithm, and applies the quasi-full stress algorithm as an operator after the genetic operations. The authors also improve the penalty function of the traditional genetic algorithm to improve its efficiency. The method is demonstrated on a 12-bar truss structure, comparing the optimized and non-optimized models and the topology optimization results.
Core Technical Points and Interpretation
Topology optimization is a powerful tool for finding the optimal layout of a structure, but it is computationally intensive and often difficult to apply to discrete variables such as section sizes and member presence. The paper's contribution is the development of a hybrid algorithm that combines the global search capability of a genetic algorithm with the local search capability of a quasi-full stress algorithm. The quasi-full stress algorithm is a heuristic method that iteratively adjusts the member sizes to achieve a uniform stress distribution, which is the optimal condition for a truss structure. By seeding the genetic algorithm with the quasi-full stress solution and applying the quasi-full stress algorithm as an operator, the hybrid method converges faster and finds better solutions than either method alone.
The optimization variables in this problem are particularly interesting. The square steel tube section type is a discrete variable that represents the standard section sizes available from the manufacturer. The topology variables are binary variables that indicate whether a member is present or absent. The concrete strength grade is also a discrete variable, representing the available grades of concrete. The objective function is the total structural cost, which includes the cost of the steel tube, the cost of the concrete, and the cost of fabrication. The constraint is that the structure must satisfy the strength and stiffness requirements under the applied loads.
The improvement to the penalty function is a practical contribution. In a genetic algorithm, the penalty function is used to handle constraints by adding a penalty to the objective function when the constraints are violated. The traditional penalty function is often too harsh, which can drive the population away from the feasible region. The paper proposes an improved penalty function that adapts the penalty factor based on the degree of constraint violation, which allows the algorithm to explore the infeasible region more effectively and find better solutions.
Integration with Engineering Practice
The topology optimization of CFST truss structures has direct applications in the design of bridges, roofs, and other large-span structures. From a steel pipe perspective, the optimization results determine the section sizes and the layout of the members, which directly affect the cost and the manufacturing process. The paper's finding that the optimized structure has fewer members and that each member is fully stressed is consistent with the principles of efficient structural design. For a pipe manufacturer, this means that the optimized structure will use a smaller number of larger pipes, rather than a larger number of smaller pipes, which reduces the number of connections and simplifies the fabrication.
The optimization of the concrete strength grade is also relevant to the pipe manufacturer. A higher concrete strength grade allows a smaller steel tube section, which reduces the steel weight but increases the concrete cost. The optimization balances these costs to find the most economical solution. In practice, the concrete strength grade is often limited by the availability of materials and the construction method. For example, a high-strength concrete may require a special mix design that is difficult to pump through a small steel tube, which would limit the practical application of the optimization results.
The paper's use of a 12-bar truss as a demonstration case is a simple example, but the method can be scaled to larger structures. However, the computational cost of the genetic algorithm increases significantly with the number of members and the number of variables. For a large truss structure with hundreds of members, the optimization may take a long time, which limits its practical application. The paper does not address this scalability issue, but it is a critical consideration for real-world applications.
Key Questions and Reflections
One of the challenges in the topology optimization of CFST truss structures is the definition of the cost function. The paper uses the total structural cost as the objective, but the cost of a CFST member is not simply the sum of the steel and concrete costs. The fabrication cost, including the welding of the connections and the handling of the members, can be a significant portion of the total cost. The paper does not include these costs in the objective function, which may lead to an optimized structure that is not truly the most economical when fabrication costs are considered.
Another practical issue is the manufacturability of the optimized topology. The genetic algorithm may produce a topology that includes members with very small sections or unusual angles, which are difficult to fabricate and connect. The paper does not address this issue, but it is a critical consideration for the practical application of the optimization method. In my experience, the most economical structure is not always the one with the minimum weight or the minimum material cost; it is the one that can be fabricated and erected most efficiently.
Study Insights and Implications
This paper presents a novel and effective method for the topology optimization of CFST truss structures, and it demonstrates the benefits of combining a global search algorithm with a local search heuristic. The improved penalty function is a practical contribution that enhances the efficiency of the genetic algorithm. For engineers involved in the design of CFST truss structures, this paper provides a useful tool for finding the optimal layout and section sizes, which can lead to significant cost savings. For pipe manufacturers, the paper highlights the importance of considering the cost of the steel tube as a key parameter in the design optimization, and it reinforces the need to provide a wide range of standard section sizes to allow the optimizer to find the most economical solution. The paper also highlights the need for further research on the inclusion of fabrication costs and manufacturability constraints in the optimization, which are essential for the practical application of the method to real-world structures.
Zhuojin Pipe Fitting Co., Ltd