Faculty of Engineering, Built Environment and Information Technology
School of Engineering
Department of Mechanical and Aeronautical Engineering
Selected Highlights from Research Findings
Continuous casting. The project has as its aim the development of optimal tundish and mould designs. The tundish and the furniture in it form a crucial part of the continuous casting process of steel in that it acts as a reservoir, grade separator and as an inclusion particle removal device. The design optimization process uses Computational Fluid Dynamics (CFD) combined with mathematical optimization to produce tundish designs that result in cleaner steel and reduced losses due to mixed-grade length. Industrial partners that have contributed to the research effort through THRIP projects, include steel makers Columbus Stainless and Iscor, and refractory suppliers Vesuvius and LTM Technologies. The research on tundishes has recently been extended to studying the mould of the continuous caster, in particular, in the design optimization of submerged entry nozzles.
Design optimization of automotive fuel tank for sloshing and impact. Researchers are collaborating with US researchers (Livermore Software Technology Corp.) in the multi-disciplinary design optimization of automotive fuel tanks for sloshing and impact. Computational Fluid Dynamics (CFD) software (FLUENT) is linked with non-linear dynamic finite element software (LS-DYNA) to combine the sloshing and impact event simulations using mathematical optimisation techniques (LS-OPT software). A physical demonstration prototype tank has been built and is used for validation of the technology. Interest has been shown from the civil structural engineering community to utilize this research in the design of liquid-transporting containers.
Reliable and robust design. The research has as its main aim the optimal design of engineering systems in the presence of random variations. More specifically, this entails including reliability (e.g. failure) criteria in the design optimization process. Currently, design optimization is mostly performed using deterministic simulation models of engineering processes. The study concentrates on treating manufacturing-related design variables, e.g. structural member thickness and material properties, as statistical field quantities to assess the influence of their variation on the performance of the manufacturing process or system design. An example of where the research is to be applied is in the automotive manufacturing and design sector. Structural finiteelement analyses are used when car parts are desgned. Nominal thicknesses and material properties are specified and e.g. crash performance characteristics are determined and optimized using mathematical optimization techniques. What would happen if during the manufacturing process of some of these parts , the thickness varies throughout the part, and material properties have a certain nominal value but deviate with an associated statistical distribution? The reliability-based design optimization process being developed will ensure a cost-effective way to include uncertainties like these into the design process, thereby resulting in a reliable as well as robust product with optimal performance.
Contact person: Prof KJ Craig.
Chisel wear. In the Dynamic System Group a significant breakthrough was made, in developing a methodology to indirectly determine the wear and tear on chisels on lathes, by measuring the vibration on the clamp which keeps the chisel in place. The process entails the calculation of the characteristics of the signal and the resultant analysis by means of a system of neural networks. It was tested experimentally at the Kolbenco manufacturing plant in Alrode.
Existing techniques for the monitoring of gear damage pre-suppose a constant workload. These techniques are therefore inadequate for a variety of practical problems which generally occur in industry i.e. the identification of damage occuring on the gearboxes used by large drag lines. A new technique was developed with which to monitor the damage occuring in gears which are subjected to a varying workload. This technique has already been tested extensively in the laboratory and will shortly be tested on an actual drag line at Syferfontein mine.
Contact person: Prof PS Heyns.
Design of heat sinks. Over the past few years our research group has developed simplified simulation software that can be used to optimise the thermal design of mainly heat sinks. The method used is one where the flow through the heat sink is solved using a simplified network analysis to calculate the velocities at different positions in a heat sink. It uses these velocities to calculate the heat transfer rates at the heat sink boundaries. A finite volume solver is then used to calculate the temperature distribution in the heat sink. Although not solving the flow field, the model still requires an accurate formulation of the flow distribution around the heat sink to correctly predict the heat transfer rate to the environment.
During 2002 two important advances were made towards applying this approach to more complex geometries. The first was the development of a more advanced network solver that, when combined with the existing correlations for inlet and exit losses around objects, will enable this group to model flow around fully 3-D geometries. With this addition, previous limits regarding geometry limitations have been removed completely.
The second very important technical achievement was the integration of a 1-D solver for the heat transfer to the cooling fluid between the finned sections. This means that although the fluid flow is solved separately using a network solver, the heat transfer between the fluid and the solid is now solved simultaneously. This is very important when the temperature of the air between the fins increases substantially.
Contact person: Prof JA Visser.
|
