This study proposed a discrete structural optimization method for a framed automotive body. Up to four types of discrete design variables are considered simultaneously, that is, the sizing, cross-sectional shape, topology, and material variables. Firstly, to solve the nonconvex and nonlinear optimization problem, the original non-dominated sorting genetic algorithm, the third version (NSGA-III), is adapted. An improved extreme points identification scheme and a new mutation operator are proposed to stabilize the normalization of the population and accommodate the manufacturing constraints, respectively. Two test problems demonstrate that the modified NSGA-III can handle continuous and discontinuous multiple objective optimization. Subsequently, the classical 10-bar truss is used to illustrate the proposed method. A weight reduction of 4.5 kg is achieved as compared to previous optimal designs in the literature. Finally, a framed automotive body is optimized for maximizing the first order natural frequency and minimizing the total mass, the maximum stresses and the maximum displacements in different load cases and the manufacturing cost. The results obtained by different optimization procedures are presented and discussed. The results demonstrate the feasibility and effectiveness of the proposed method. A weight reduction of 17.59% is achieved while other structural performances satisfy the design requirements.
5 December 2019 in Journal of Computational and Nonlinear Dynamics
In mechanisms, the guide and roller form joints with clearance and the roller may make contact with the convex surface or concave surface of the guide. Correctly modeling this type of joint is critical in a mechanism's dynamic performance analysis. This paper proposes a contact kinematic model and a hybrid contact force model for the planar joint and investigates both the convex and concave contact cases. The contact kinematic model is derived from the relative motion between the guide and roller and the hybrid contact force model consists of an elastic force and a damping force. The sliding door mechanism is used as an illustrative example to demonstrate the proposed models. Experimental data are collected for a vehicle's sliding door mechanism. Results show that the proposed models are effective and the simulation results match the general trend of experimental data based on RMS errors.
This study proposes a multi-objective optimization (MOO) strategy with an improved constraint-handling technique to improve the crashworthiness of an excavator rollover protective structure (ROPS). First, the experimental test under the ISO 12117 criteria is conducted and the developed numerical model is verified. Then, the amounts of energy absorption and the cross-sectional forces of components in the ROPS are analyzed. The main energy absorbing and load carrying components are identified. Finally, the thicknesses of the identified components are considered as the design variables. A multi-objective crashworthiness optimization process aims at improving the safety distance and reducing the total mass is designed by the finite element analysis-based surrogate model technique and a modified MOO algorithm. The proposed algorithm modifies the objective function values of an individual with its constraint violations and the true objective function values, of which adaptive penalty weights fed back from the constraint violations are used to keep the balance. Compared with the existing methods, it is found that the optimal solutions obtained by the proposed algorithm show superiority on convergence rate and diversity of distribution. The optimal results show that the safety distance is 27.42% higher while the total mass is 7.06% lower than those of the baseline design when it meets the requirements of ISO 12117. This study provides an alternative crashworthiness design route for the ROPS of the construction machines.
1 January 2020 in IOP Conference Series: Materials Science and Engineering
Most of the structural topology optimization problems involved in practical engineering applications are problems with many constraints. Thus, this paper presented an improved bi-directional evolutionary structural optimization (BESO) method for problems with many constraints. Slack variables were introduced to transform the inequality constraints into equality constraints. Then the Lagrange multiplier method converted the multi-constrained optimization problem to an unconstrained optimization problem and an advanced multiplier calculation was proposed. Elemental sensitivity numbers were derived according to the obtained multipliers. The design variables were updated by the BESO method. One numerical example that aimed to minimizing the structural mean compliance and involved volume, fundamental frequency and displacement constraints was used to validate the proposed method. The new method extends the BESO method to topology optimization that has a number of constraints, such as volume, displacement and frequency constraints.