Johann Sienz

Fully stressed topological design of structures using an evolutionary procedure

An automated fully stressed design approach based on the Xie and Steven algorithm is presented. With this algorithm a fully stressed design is obtained by a gradual removal of low stressed material. By applying this evolutionary procedure a layout or topology of a structure can be found from an initial block of material. A fully integrated, interactive program is presented which incorporates automatic mesh generation, finite element analysis and the fully stressed design algorithm. The feasibility of the approach is demonstrated using several examples.

Formulation of the Optimal Latin Hypercube Design of Experiments Using a Permutation Genetic Algorithm

The choice of location of the evaluation points is important in response surface generation, especially when the evaluations are expensive. Space-filling designs can be used to specify the points so that as much of the design space is sampled as possible with the minimum number of response evaluations. One popular technique is the optimal Latin hypercube design of experiments. However, its generation is non-trivial, time consuming and is – but for the simplest problems – infeasible to carry out by enumeration. Therefore, solving this problem requires an optimization technique to search the design space. As the problem is discrete, it is ideally suited to the use of discrete optimization techniques such as genetic algorithms. This paper describes a method for generating optimal latin hypercubes using a permutation genetic algorithm and compares it with a standard binary genetic algorithm. The objective of the optimization is based on minimizing a function that is analogous the potential energy of the system of material points. The developed method offers considerable improvements over previous solutions; it generates better solutions and the computational effort in reaching those solutions is significantly reduced.

Formulation of the Audze--Eglais uniform Latin hypercube design of experiments

This paper describes a method for formulating the Audze-Eglais Uniform Latin Hypercube design of experiments (DoE). The formulation of the Audze-Eglais DoE has not been reported in any previous research.The principle of the Audze-Eglals DoE is to distribute experiment points as uniformly as possible within the design variable domain. This is achieved by minimizing the potential energy of the points of a DoE. The DoE for N variables and P experiments is independent of the application under consideration, so once the design is formulated for P points and N design variables, it is stored in a matrix and need not be formulated again.The generation of the Audze-Eglais DoE is time consuming and requires optimization to solve the minimization problem. Therefore, the major aim of the paper is to identify a design variable encoding method for use in optimization. The two methods for encoding the DoE for use in the optimizer are presented. The method adopted in this study uses the co-ordinates of the plan points as design variables. The results for the potential energy are compared to published Audze-Eglais Uniform Latin Hypercube DoE and to random sampling Latin Hypercube DoE. The results indicate that the method works well and improvement over previous results has been achieved.

Advanced solution methods in topology optimization and shape sensitivity analysis

Investigates the efficiency of hybrid solution methods when incorporated into large‐scale topology and shape optimization problems and to demonstrate their influence on the overall performance of the optimization algorithms. Implements three innovative solution methods based on the preconditioned conjugate gradient (PCG) and Lanczos algorithms. The first method is a PCG algorithm with a preconditioner resulted from a complete or an incomplete Cholesky factorization, the second is a PCG algorithm in which a truncated Neumann series expansion is used as preconditioner, and the third is a preconditioned Lanczos algorithm properly modified to treat multiple right‐hand sides. The numerical tests presented demonstrate the computational advantages of the proposed methods which become more pronounced in large‐scale and/or computationally intensive optimization problems.

Computer Aided Optimisation of Profile Extrusion Dies

Abstract An objective function is proposed for use in optimisation of profile die design, and defined in terms of mass flow rates to topological partitions of the profile cross section. In a series of uPVC extrusion trials its dependence on land channel height is investigated for two profiles to assess its suitability for gradient based optimisation of flow distribution. Preliminary finite element flow simulations are carried out, and comparisons of computed and experimental objective function values show that the mathematical modelling should be extended to include extensional effects and draw down at the die exit. Examination of the experimental profiles indicates the limitations of die optimisation in determining the detailed geometry of the finished profile cross sections.

Computational modelling of 3D objects by using fitting techniques and subsequent mesh generation

Abstract The aim of this paper is to present a method to generate computational, geometric models of 3D objects defined in the form of a ‘cloud of points’ (CoP). These objects can then be discretized for subsequent finite element analyses. Various procedures required to achieve this are briefly described, including the mathematical representation of an integrated image acquisition/modelling system, which allows for the automatic creation of accurate geometric models of either scanned images of objects or of CoP generally, as well as the mesh generation algorithm employed. The application of this combined method is then demonstrated with respect to modelling human organs and agricultural objects. The main emphasis of the paper is on the wide applicability of the developed techniques.

Structural optimization strategies for simple and integrally stiffened plates and shells

Purpose – Shells are widely used structural systems in engineering practice. These structures have been used in the civil, automobile and aerospace industries. Many shells are designed using the finite element analysis through the conventional and costly trial and error scheme. As a more efficient alternative, optimization procedures can be used to design economic and safe structures.Design/methodology/approach – This paper presents developments, integration and applications of reliable and efficient computational tools for the structural optimization of variable thickness plates and free‐form shells. Topology, sizing and shape optimization procedures are considered here. They are applied first as isolated subjects. Then these tools are combined to form a robust and reliable fully integrated design optimization tool to obtain optimum designs. The unique feature is the application of a flexible integrally stiffened plate and shell formulation to the design of stiffened plates and shells.Findings – This wor...

Nonlocal buckling behavior of bonded double-nanoplate-systems

Buckling behavior of a bonded, uni-axially compressed double-nanoplate-system is investigated in this work. Both the synchronous and asynchronous-type buckling is considered in detail. The two nanoplates are assumed elastically bonded by a polymer resin. The nano-scale effects of nanoplates are dealt with in the analysis by using nonlocal elasticity theory. The theory is utilized for deriving the expressions for a buckling load of a double-nanoplate-system. A simple analytical method is introduced for determining the buckling load of a nonlocal double-nanoplate-system. Explicit closed-form expressions for the buckling load are derived for the case when all four ends are simply supported. Single-layered graphene-sheets are considered for the study. The study highlights that the nonlocal effects considerably influence the buckling behavior of the double-graphene-sheet-system. Unlike the buckling behavior of a single graphene sheet, the double-graphene-sheet-system undergoes both synchronous as well as async...

Formulation of the Audze-Eglais uniform Latin hypercube design of experiments for constrained design spaces

Optimal Latin Hypercubes (OLH) created in a constrained design space might produce Design of Experiments (DoE) containing infeasible points if the underlying formulation disregards the constraints. Simply omitting these infeasible points leads to a DoE with fewer experiments than desired and to a set of points that is not optimally distributed. By using the same number of points a better mapping of the feasible space can be achieved. This paper describes the development of a procedure that creates OLHs for constrained design spaces. An existing formulation is extended to meet this requirement. Here, the OLH is found by minimizing the Audze-Eglais potential energy of the points using a permutation genetic algorithm. Examples validate the procedure and demonstrate its capabilities in finding space-filling Latin Hypercubes in arbitrarily shaped design spaces.

Water assisted injection moulding : development of insights and predictive capabilities through experiments on instrumented process in parallel with computer simulations

Abstract An idealised model of core-out in water assisted injection moulding (WAIM) is set up to isolate the effect of cooling by the water on the deposited layer thickness. Based on simulations, this is investigated for a specific case as a function of Pearson number and power law index. It is found that cooling significantly reduces the layer thickness to the extent that a change in the flow regime ahead of the bubble, from bypass to recirculating flow, is possible. For shear thinning melts with high temperature coefficient of viscosity, the simulations show very low layer thickness, which may indicate unfavourable conditions for WAIM. Although in the real moulding situation, other effects will be superimposed on those found here, the results provide new insights into the fundamentals of WAIM. Investigation of other effects characterised by Fourier and Reynolds numbers will be reported subsequently. Some early process measurement results from an experimental WAIM mould are presented. Reductions in residual wall thickness are observed as the water injection set pressure is increased and the duration of water bubble penetration through the melt is determined experimentally. The formation of voids within the residual wall is noted and observed to reduce in severity with increasing water injection pressure. The presence of such voids can be detected by the signature from an infrared temperatures sensor.