Mechanical structures of large-scale antennas are sheet metals connected by thousands of rivets. The antenna dimensional error after riveting often violates the limit allowed. The prediction of the global dimensional error induced by many rivet connections requires a rapid and accurate assembly deformation calculation method. Main process parameters of these local rivet connections are the local connection dimension, material property, local clamp position, rivet upsetting direction, and the hammer time-to-displacement impact, except for the riveting sequence. We neglect the process parameter uncertainties and consider that the main riveting parameters equate to a dynamic finite-element (FE) model of single rivet connection. The dynamic FE analysis result yields an inherent strain database for the riveted local parts. Then, we propose an iterative loop of static FE analyses for the global structure taking the inherent strain database and possible former static FE analysis result as the boundary conditions. The loop forms a local-to-global framework. Two examples are involved through the framework representation and realistic application. Framework advantages include: (1) a good balance between the cost and precision of dimensional error calculation; (2) the sequence simulation of all the riveting operations; and (3) supporting the further assembly process optimization to reduce the global dimensional error of the assembly with thousands of rivets.
25 March 2017 in Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture
Due to the complexities of calculations based on different representations of constituent parts in an assembly, the combined improvement of efficiency and precision for the riveting has seldom been studied by the sequence planning or dimensional engineering. This article develops an assembly sequence optimization method that simultaneously minimizes the path length and overall dimensional error for the solid riveting. The sequence is denoted by the order of classifications for the nodes around the rivet holes in the finite element model for the assembly. Ant colony algorithm is applied on the optimization by integrating two calculations, that is, the length calculation of riveting path and the calculation of the dimensional error related to rivet upsetting directions and assembly sequence. Results for three assemblies are presented which show a good degree of optimization in performance for different scales of assemblies with different numbers of rivets that share the same solid riveting process.
7 June 2015 in Journal of Intelligent Manufacturing
This work targets the development of the multidisciplinary design for centrifugal fans, based on the calculation of the fan performance following changes in the fan structure and manufacture. A parametric method that starts from the blade number and manufacturing parameters is first developed to control the shape of fan assembly and its fluid domain. The parametric code, and the aerodynamic and structure analysing codes are then integrated, from the solution of which algorithms are developed to calculate the fan indexes including the flow rate, pressure rise, fan efficiency, fan strength, and the motor power. Validation of the proposed method is performed by the comparison with an experimental test. An orthogonal sampling opens out relevant changes of the fan working performance with respect to the blade number and manufacturing parameters. Results suggest ranges for the blade number and manufacturing parameters that have a sound performance and structure safety when converting the motor power to airflow.