Faculty of Engineering, Built Environment and Information Technology
School of Engineering
Department of Mechanical and Aeronautical Engineering
Selected Highlights from Research Findings
Varying speed and load conditions on real machinery such as draglines used in open cast mining operations, present considerable challenges for structural health monitoring. In his master’s research, Berndt Eggers investigated the application of computed order tracking with subsequent rotation domain averaging and statistical analysis, to health monitoring on the drag gearbox of such a dragline. Computed order tracking is a fault detection method developed to deal with the varying speed conditions and has been proven effective in laboratory conditions. Its application to real situations, however, requires some adjustment to deal with issues such as inadequate speed data and the fact that the drag gear rotates in two directions. This provides a unique opportunity to observe the performance of the order tracking method in a bi-directional rotating environment, allowing an investigation of the relationships between the results obtained from each operating direction. A monitoring station was set up aboard the dragline and data was captured for a period spanning approximately one year. The data captured consisted of accelerometer and proximity sensor data. The results indicate that computed order tracking can be successfully employed in real environments. Furthermore it is shown that the rotation direction that opens the gear-tooth crack gives a better indication of incipient failure. It is therefore important not to disregard either direction when monitoring rotating machinery of this kind.
Contact person: Prof PS Heyns.
The main aim of this study was to develop a numerical model whereby the rotational motion of a tippler structure can be simulated accurately by using linear static finite element models solved for set intervals. The complex rotational working of a tippler structure complicates the analytical evaluation of the structure. A further complication is the ever-changing boundary conditions while the structure rotates, together with the weight reduction of the coal in the wagons when the wagons are offloaded. Both these factors need to be taken into account when determining the stress levels in the structure while operational. To verify the accuracy of a finite element simulation of a tipping cycle, strain gauge measurements obtained from the actual tippler structure was compared with stress results obtained from linear static finite element analyses of the structure, simulating different tip positions at set time intervals. The results obtained from the comparison indicated an accurate simulation of the tipping cycle by means of the finite element simulation.
Contact person: Prof NDL Burger.
Particle swarm optimization (PSO) algorithms are inspired by studies of the social and cognitive behaviour of groups for example bird swarms and fish schools. A particle (individual) is “flown” through a multidimensional search space in an attempt to locate a good solution. In an optimization framework, a good solution implies a low function value. Individuals recall their own best solution to date, and also have access to the best solution of the group to date. PSO algorithms have been studied extensively since its invention in 1995. The researchers conducted two studies to gain additional insight into PSO algorithms. In their first investigation they considered the effect of diversity in PSO algorithms. Diversity is the ability of an algorithm to perform space-filling searches. Specifically, two implementations of the PSO were studied. While the behavior of the respective implementations is markedly different, they only differ in the formulation of the velocity updating rule. In fact, the differences are merely due to subtle differences in the introduction of randomness into the algorithm. They showed that for the first implementation (linear PSO), the particle trajectories collapse to line searches in the multidimensional search space. The second implementation (general PSO) does not suffer this drawback. Instead, diverse stochastic search trajectories are retained. They then showed that some popular heuristics like maximum velocity limit, position restriction, craziness and high initial velocities also introduce diversity into the algorithm. In the second study, the frame (in)variance of PSO algorithms were studied. Frame invariance requires specific relationships between the measurements of the same event by two different observers. A simple example is that the length of a particular object is independent of the observer that performs the measurement. Although frame invariance is a law of nature, optimization algorithms do not necessarily obey this law. Violation of frame invariance results in algorithm performance that depends on the choice of reference frame (observer). The frame (in)variance of the same two implementations were investigated. The researchers verified that the linear PSO is frame invariant as opposed to the classical PSO which is frame variant. Since the classical PSO is diverse but frame variant, the researchers investigated the possibility to introduce diversity in an invariant algorithm. The proposed formulation indeed generated space filling trajectories and was frame invariant in a stochastic sense. They obtained promising numerical results which verified that PSO algorithms can be formulated independent of a reference frame without sacrificing diversity.
Contact person: Prof S Kok.
This study is designed to increase fundamental understanding of transition single phase flows inside commercial enhanced and smooth tubes in heat exchangers. The heat exchangers are found in chillers, steam boilers, steam and nuclear power plants, chemical processes, air-conditioning systems and so on. Traditionally smooth tubes have been used in heat exchangers but recently enhanced tubes have been introduced that increase the heat transfer surface area and the flow path which in many cases increases the turbulence. High heat transfer can be maintained at high flow rates (Reynolds numbers) but it also increases the pump power needed. The optimum flow rate from an energy point of view with enhanced tubes is in many cases in the transitional flow regime between laminar and turbulent flow. The transitional regime is usually avoided by engineers as no data exist on the heat transfer and pressure drop characteristics and type of inlet of both smooth and enhanced tubes. Textbooks proffer a transitional Reynolds number of around 2 100 in smooth circular tubes. In this work it has been shown that the onset of turbulent flow can be stalled to a Reynolds number of 7 500 and beyond. The transitional Reynolds number is shown to be dependent on the type of tube surface and diameter, on the heat transferred, and on the inlet configuration of the heat exchanger. The results shown below are adiabatic friction factors for a smooth tube, low-finned tube with 18º helix angle, and for a low-finned tube with 27° helix angle, for four different types of inlets. The four inlets are a fully (hydrodynamically) developed inlet, a re-entrant inlet, a square-edge inlet, and a bell mouth inlet. The results show that the transition from laminar to transition flow is highly influenced by the inlet geometry. Transition can be delayed to up to Reynolds numbers in the order of 7 500. In practice an understanding of these phenomena can be used to design large industrial systems with significantly improved energy efficiency and hence cost effectiveness.
Contact person: Prof JP Meyer.
Vapour and liquid phases flow in horizontal tubes according to the prevailing flow regime. The existence of a particular flow regime depends on a variety of parameters, including the thermophysical properties of the fluid; the tube diameter, orientation and geometry; force field; mass flux; heat flux; and vapour quality. The researchers have shown that flow regimes significantly influence the heat and momentum transfer processes during two-phase flow, inferring that any accurate prediction of heat transfer and pressure drop during flow condensation should be based on the analysis of the prevailing flow regime. In addition, identification and modelling of the flow regime can enhance the overall performance and safety in two-phase flow systems. Although a reliable, objective and quantitative instrumentation-based indicator of flow regime is desirable, definitions of flow regimes are usually based on visual (photographs or graphical illustrations) and linguistic descriptions, with a corresponding element of subjectivity. The complex nature of two-phase flow, which is characterized by turbulence, deformable phase interface, phase interaction, phase slip, and compressibility of the gas phase, makes it difficult to obtain reliable flow models. The main drawback of previous techniques for flow regime identification is their lack of generality as their ability to detect transitions depends on the specific heat exchanger and fluid used. The researchers performed flow regime recognition based on spectral analysis of pressure signal fluctuations of the condensing gas, combined with high-speed videography and photography of the prevailing flow regime. They subsequently produced numerous semi-empirical correlations that among other accurately predict flow regime transitions during refrigerant condensation in circular smooth-, helical micro-fin-, and herringbone micro-fin tubes, all of which are extensively used in the heat transfer industry. The developed correlations that are plotted on a so-called flow regime map make it possible for design engineers to produce accurate designs of refrigeration condensers.
Contact person: Prof L Liebenberg.