Mechanical Constructions and Mechanization

The complex problem of truss structural optimization, based on the discrete design variables assumption, can be approached through optimizing aspects of sizing, shape, and topology or their combinations. This paper aims to show the differences in results depending on which aspect, or combination of aspects of a standard 10 bar truss problem is optimized. In addition to standard constraints for stress, cross section area, and displacement, this paper includes the dynamic constraint for buckling of compressed truss elements. The addition of buckling testing ensures that the optimal solutions are practically applicable. An original optimization approach using genetic algorithm is verified through comparison with literature, and used for all the optimization combinations in this research. The resulting optimized model masses for sizing, shape, and topology or their combinations are compared. A discussion is given to explain the results and to suggest which combination would be best in a generalized example.

In this paper parametric modeling of sizing optimization truss models is developed. Sizing optimization of trusses views element cross-sections as variables with the goal of minimizing overall mass while maintaining equivalent stresses within acceptable ranges, as well as limiting displacement. In order to conduct such a process, models needs to be created with parameters, and outputs which can be used to create an objective function. Furthermore the models in each iteration of the optimization are subjected to finite element analyses to determine stress. Parametric models of standard 10 bar, 17 bar, and 25 bar trusses are created to facilitate optimization. The heuristic optimization method used is genetic algorithm. Optimization results obtained from these models are compared to those from literature and the initial model.

Using continuous variables in truss structural optimization results in solutions which have a large number of different cross section sizes whose specific dimensions would in practice be difficult or expensive to create. This approach also creates optimal models which if varied, even slightly, result in structures which do not meet constraint criteria. This research proposes the discretization of cross section sizes to standard sizes of stock produced for the particular cross section and material, and a 1mm precision for node location when using shape optimization. Additionally, Euler buckling constraints are added to all models in order to achieve optimal solutions which can find use in practical application. Several standard test models of trusses from literature, which use continuous variables, are compared to the discrete variable models under the same conditions. Models are optimized for minimal weight using sizing, shape, topology, and combinations of these approaches.

The use of continuous variables for cross-sectional dimensions in truss structural optimization gives solutions with a large number of different cross sections with specific dimensions which in practice would be expensive, or impossible to create. Even slight variations from optimal sizes can result in unstable structures which do not meet constraint criteria. This paper shows the influence of the use of discrete cross section sizes in optimization and compares results to continuous variable counterparts. In order to achieve the most practically applicable design solutions, Euler buckling dynamic constraints are added to all models. A typical space truss model from literature, which use continuous variables, is compared to the discrete variable models under the same conditions. The example model is optimized for minimal weight using sizing and all possible combinations of shape and topology optimizations with sizing.

In this paper the effects of adding buckling constraints to truss sizing optimization for minimizing mass are investigated. Introduction of buckling testing increases the complexity of the optimization process as Euler buckling criteria changes with each iteration of the optimization process due to the changes in element cross section dimensions. The resulting models which consider this criteria are practically applicable. For the purposes of showing the effects of dynamic constraints for buckling, optimal parametric standard test models of 10 bar, 17 bar, and 25 bar trusses from the literature are tested for buckling and compared to the models with the added constraint. Models which do not consider buckling criteria have a considerable number of elements which do not meet buckling criteria. The masses of these models are substantially smaller than their counterparts which consider buckling.

The use of continuous variables for cross-sectional dimensions in truss structural optimization gives solutions with a large number of different cross sections with specific dimensions which in practice would be expensive, or impossible to create. Even slight variations from optimal sizes can result in unstable structures which do not meet constraint criteria. This paper shows the influence of the use of discrete cross section sizes in optimization and compares results to continuous variable counterparts. In order to achieve the most practically applicable design solutions, Euler buckling dynamic constraints are added to all models. A typical space truss model from literature, which use continuous variables, is compared to the discrete variable models under the same conditions. The example model is optimized for minimal weight using sizing and all possible combinations of shape and topology optimizations with sizing.

This paper looks at sizing optimization results, and attempts to show the practical implications of using a novel constraint. Most truss structural optimization problems, which consider sizing in order to minimize weight, do not consider the number of different cross-sections that the optimal solution can have. It was observed that all, or almost all, cross-sections were different when conducting the sizing optimization. In practice, truss structures have a small, manageable number of different cross-sections. The constraint of the number of different cross-sections, proposed here, drastically increases the complexity of solving the problem. In this paper, the number of different cross-sections is limited, and optimization is done for four different sizing optimization problems. This is done for every number of different cross-section profiles which is smaller than the number of cross-sections in the optimal solution, and for a few numbers greater than that number. All examples are optimized using dynamic constraints for Euler buckling and discrete sets of cross-section variables. Results are compared to the optimal solution without a constrained number of different cross-sections and to an optimal model with just a single cross-section for all elements. The results show a small difference between optimal solutions and the optimal solutions with a limited number of different profiles which are more readily applicable in practice.

This paper looks at sizing optimization results, and attempts to show the practical implications of using a novel constraint. Most truss structural optimization problems, which consider sizing in order to minimize weight, do not consider the number of different crosssections that the optimal solution can have. It was observed that all, or almost all, crosssections were different when conducting the sizing optimization. In practice, truss structures have a small, manageable number of different cross-sections. The constraint of the number of different cross-sections, proposed here, drastically increases the complexity of solving the problem. In this paper, the number of different cross-sections is limited, and optimization is done for four different sizing optimization problems. This is done for every number of different cross-section profiles which is smaller than the number of cross-sections in the optimal solution, and for a few numbers greater than that number. All examples are optimized using dynamic constraints for Euler buckling and discrete sets of cross-section variables. Results are compared to the optimal solution without a constrained number of different crosssections and to an optimal model with just a single cross-section for all elements. The results show a small difference between optimal solutions and the optimal solutions with a limited number of different profiles which are more readily applicable in practice.