A Dymond, 2014. " TUNING OPTIMIZATION ALGORITHMS UNDER MULTIPLE OBJECTIVE FUNCTION EVALUATION BUDGETS "

The performance of optimization algorithms is sensitive to both the optimization problem's numerical characteristics and termination criteria. Given these considerations two tuning algorithms named tMOPSO and MOTA are proposed to assist optimization practitioners to find algorithm settings which are approximate for the problem at hand. For a specified problem tMOPSO aims to determine multiple groups of control parameter values, each of which results in optimal performance at a different objective function evaluation budget. To achieve this, the control parameter tuning problem is formulated as a multi-objective optimization problem. Furthermore, tMOPSO uses a noise-handling strategy and control parameter value assessment procedure, which are specialized for tuning stochastic optimization algorithms. The principles upon which tMOPSO were designed are expanded into the context of many objective optimization, to create the MOTA tuning algorithm. MOTA tunes an optimization algorithm to multiple problems over a range of objective function evaluation budgets. To optimize the resulting many objective tuning problem, MOTA makes use of bi-objective decomposition. The last section of work entails an application of the tMOPSO and MOTA algorithms to benchmark optimization algorithms according to their tunability. Benchmarking via tunability is shown to be an effective approach for comparing optimization algorithms, where the various control parameter choices available to an optimization practitioner are included into the benchmarking process.

Supervisors: Prof. P.S. Heyns, Prof. S. Kock

 

 

JJA Eksteen, 2014. " ADVANCES IN ITERATIVE LEARING CONTROL WITH APPLICATION TO STRUCTURAL DYNAMICS RESPONSE RECONTRUCTION "

Iterative learning control (ILC) is a repetitive control scheme that uses a learning capability to improve the tracking accuracy of a desired test system output over repeated test trials. ILC is sometimes used in response reconstruction on complex engineering structures, such as ground vehicles, for purposes of fatigue/durability testing. The compensator that is employed in ILC in such cases is traditionally an approximate, linear inverse model of the closed-loop test system. This research presents advances in ILC, particularly with respect to its application in response reconstruction for fatigue testing purposes. The contribution of this research focuses on three aspects: the use of a nonlinear inverse model in ILC as ILC compensator instead of a linear inverse model; the use of multiple inverse models, each one defined over a different part of the test frequency band, instead of one model that covers the entire test frequency band; and on the development and use of a new type of ILC algorithm. The contributions are implemented and demonstrated on a quarter vehicle road simulator, with favourable results for the use of nonlinear inverse models and multiple inverse models, and with the new ILC algorithm giving comparable to slightly worse results than the conventional ILC algorithm. In order to invert the nonlinear inverse models this research also presents advances in the stable inversion method that is used to invert such models.

Supervisor: Prof. P.S. Heyns

 

S Aye, 2014. " ACUSTIC EMISSION-BASED DIAGNOSTICS AND PROGNOSTICS OF SLOE ROTATING BEARINGS USING BAYESIAN TECHNIQUES"

Diagnostics and prognostics in rotating machinery is a subject of much on-going research. There are three approaches to diagnostics and prognostics. These include experience-based approaches, data-driven techniques and model-based techniques. Bayesian data-driven techniques are gaining widespread application in diagnostics and prognostics of mechanical and allied systems including slow rotating bearings, as a result of their ability to handle the stochastic nature of the measured data well. The aim of the study is to detect incipient damage of slow rotating bearings and develop diagnostics which will be robust under changing operating conditions. Further it is required to explore and develop an optimal prognostic model for the prediction of remaining useful life (RUL) of slow rotating bearings.

This research develops a novel integrated nonlinear method for the effective feature extraction from acoustic emission (AE) signals and the construction of a degradation assessment index (DAI), which are subsequently used for the fault diagnostics of slow rotating bearings. A slow rotating bearing test rig was developed to measure AE data under variable operational conditions. The proposed novel DAI obtained by the integration of the PKPCA (polynomial kernel principal component analysis), a Gaussian mixture model (GMM) and an exponentially weighted moving average (EWMA) is shown to be effective and suitable for monitoring the degradation of slow rotating bearings and is robust under variable operating conditions. Furthermore, this study integrates the novel DAI into alternative Bayesian methods for the prediction of remaining useful life (RUL). The DAI is used as input in several Bayesian regression models such as the multi-layer perceptron (MLP), radial basis function (RBF), Bayesian linear regression (BLR), Gaussian mixture

regression (GMR) and the Gaussian process regression (GPR) for RUL prediction. The combination of the DAI with the GPR model, otherwise, known as the DAI-GPR gives the best prediction. The findings show that the GPR model is suitable and effective in the prediction of RUL of slow rotating bearings and robust to varying operating conditions. Further, the models are also robust when the training and tests sets are obtained from dependent and independent samples.

Finally, an optimal GPR for the prediction of RUL of slow rotating bearings based on a DAI is developed. The model performance is evaluated for cases where the training and test samples from cross validation approach are dependent as well as when they are independent. The optimal GPR is obtained from the integration or combination of existing simple mean and covariance functions in order to capture the observed trend of the bearing degradation as well as the irregularities in the data. The resulting integrated GPR model provides an excellent fit to the data and improvements over the simple GPR models that are based on simple mean and covariance functions. In addition, it achieves a near zero percentage error prediction of the RUL of slow rotating bearings when the training and test sets are from dependent samples but slightly different values when the estimation is based on independent samples. These findings are robust under varying operating conditions such as loading and speed. The proposed methodology can be applied to nonlinear and non-stationary machine response signals and is useful for preventive machine maintenance purposes.

Supervisor: Prof. P.S. Heyns

 

 

Olabode Olakoyejo, 2013. " GEOMETRIC OPTIMISATION OF CONJUGATE COOLING CHANNELS WITH DIFFERENT CROSS-SECTIONAL SHAPES"

In modern heat transfer, shape and geometric optimisation are new considerations in the evaluation of thermal performance. In this research, we employed constructal theory and design to present three-dimensional theoretical and numerical solutions of conjugate forced convection heat transfer in heat generating devices with cooling channels of different cross-sectional shapes.

In recent times, geometric configurations of cooling channel have been found to play an important role in thermal performance. Therefore, an efficient ways of optimally designing these cooling channels shapes is required.

Experimentation has been extensively used in the past to understand the behaviour of heat removals from devices. In this research, the shapes of the cooling channels and the configurations of heat-generating devices were analytically and numerically studied to minimise thermal resistance and thus illustrate cooling performance under various design conditions.

The cooling channels of five different cross-sectional shapes were studied: Circular, square, rectangular, isosceles right triangular and equilateral triangular. They were uniformly packed and arranged to form larger constructs.

The theoretical analysis is presented and developed using the intersection of asymptotes method. This proves the existence of an optimal geometry of parallel channels of different cross-sectional shapes that penetrate and cool a volume with uniformly distributed internal heat generation and heat flux, thus minimising the global thermal resistance.

A three-dimensional finite volume-based numerical model was used to analyse the heat transfer characteristics of the cross-sectional shapes of various cooling channels. The numerical computational fluid dynamics (CFD) package recently provided a more cost-effective and less time-consuming means of achieving the same objective. However, in order to achieve optimal design solutions using CFD, the thermal designers have to be well experienced and carry out a number of trial-and-error simulations. Unfortunately, this can not always guarantee an accurate optimal design solution. In this thesis a mathematical optimisation algorithm (a leapfrog optimisation program and DYNAMIC-Q algorithm) coupled with numerical CFD was employed and incorporated into the finite volume solver, –FLUENT, and grid (geometry and mesh) generation package, – GAMBIT to search and identify the optimal design variables at which the system would perform optimally for greater efficiency and better accuracy. The algorithm was also specifically designed to handle constraint problems where the objective and constraint functions were expensive to evaluate.

The automated process was applied to different design cases of cooling channels shapes. These cooling channels were embedded in a highly conductive solid and the peak temperature was minimised.

The trend and performance of all the cooling channel shapes cases studied were compared analytically and numerically. It was concluded that an optimal design can be achieved with a combination of CFD and mathematical optimisation.

 Furthermore, a geometric optimisation of cooling channels in the forced convection of a vascularised material (with a localised self-cooling property subjected to a heat flux) was also considered. A square configuration was studied with different porosities. Analytical and numerical solutions were provided. This gradient-based optimisation algorithm coupled with CFD was used to determine numerically the optimal geometry that gave the lowest thermal resistance. This optimiser adequately handled the numerical objective function obtained from numerical simulations of the fluid flow and heat transfer.

 The numerical results obtained were in good agreement with results obtained in the approximate solutions based on scale analyses at optimal geometry dimensions. The approximate dimensionless global thermal resistance predicted the trend obtained in the numerical results.  This shows that there were unique optimal design variables (geometries) for a given applied dimensionless pressure number for fixed porosity.

The results also showed that the material property had a significant influence on the performance of the cooling channel.

 Therefore, when designing the cooling structure of vascularised material, the internal and external geometries of the structure, material properties and pump power requirements would be very important parameters to be considered in achieving efficient and optimal designs for the best performance.

 Finally, this research investigated a three-dimensional geometric optimisation of conjugate cooling channels in forced convection with an internal heat generation within the solid for an array of cooling channels. Three different flow orientations based on constructal theory were studied numerically- firstly, an array of channels with parallel flow; secondly, an array of channels in which flow of every second row was in a counter direction and finally, an array of channels in which the flow direction in every channel was opposite to that of previous channel. The geometric configurations and flow orientations were optimised in such a way that the peak temperature was minimised subject to the constraint of fixed global volume of solid material. The optimisation algorithm coupled with CFD was also used to determine numerically the optimal geometry that gave the lowest thermal resistance. 

The use of the optimisation algorithm coupled with the computational fluid dynamics package; render the numerical results more robust with respect to the selection of optimal structure geometries, internal configurations of the flow channels and dimensionless pressure difference.

Keywords:     Geometric configurations, computational fluid dynamics, mathematical optimisation, thermal conductivity, constraints, laminar flow, forced convection, optimal geometry, peak temperature, constructal theory, thermal resistance, Dynamic-Q,  flow orientation

Supervisors:  Prof. T. Bello-Ochende/Prof. J.P. Meyer

 

Mr. SO OBAYOPO 2012 "Performance enhancement in proton exchange membrane fuel cell – numerical modelling and optimisation"

Sustainable growth and development in a society requires energy supply that is efficient, affordable, readily available and, in the long term, sustainable without causing negative societal impacts, such as environmental pollution and its attendant consequences. In this respect, proton exchange membrane (PEM) fuel cells offer a promising alternative to existing conventional fossil fuel sources for transport and stationary applications due to its high efficiency, low-temperature operation, high power density, fast start-up and its portability for mobile applications. However, to fully harness the potential of PEM fuel cells, there is a need for improvement in the operational performance, durability and reliability during usage. There is also a need to reduce the cost of production to achieve commercialisation and thus compete with existing energy sources. The present study has therefore focused on developing novel approaches aimed at improving output performance for this class of fuel cell.

In this study, an innovative combined numerical computation and optimisation techniques, which could serve as alternative to the laborious and time-consuming trial-and-error approach to fuel cell design, is presented. In this novel approach, the limitation to the optimal design of a fuel cell was overcome by the search algorithm (Dynamic-Q) which is robust at finding optimal design parameters. The methodology involves integrating the computational fluid dynamics equations with a gradient-based optimiser (Dynamic-Q) which uses the successive objective and constraint function approximations to obtain the optimum design parameters. Specifically, using this methodology, we optimised the PEM fuel cell internal structures, such as the gas channels, gas diffusion layer (GDL) - relative thickness and porosity - and reactant gas transport, with the aim of maximising the net power output. Thermal-cooling modelling technique was also conducted to maximise the system performance at elevated working temperatures.

The study started with a steady-state three-dimensional computational model to study the performance of a single channel proton exchange membrane fuel cell under varying operating conditions and combined effect of these operating conditions was also investigated. From the results, temperature, gas diffusion layer porosity, cathode gas mass flow rate and species flow orientation significantly affect the performance of the fuel cell. The effect of the operating and design parameters on PEM fuel cell performance is also more dominant at low operating cell voltages than at higher operating fuel cell voltages. In addition, this study establishes the need to match the PEM fuel cell parameters such as porosity, species reactant mass flow rates and fuel gas channels geometry in the system design for maximum power output.

This study also presents a novel design, using pin fins, to enhance the performance of the PEM fuel cell through optimised reactant gas transport at a reduced pumping power requirement for the reactant gases. The results obtained indicated that the flow Reynolds number had a significant effect on the flow field and the diffusion of the reactant gas through the GDL medium. In addition, an enhanced fuel cell performance was achieved using pin fins in a fuel cell gas channel, which ensured high performance and low fuel channel pressure drop of the fuel cell system. It should be noted that this study is the first attempt at enhancing the oxygen mass transfer through the PEM fuel cell GDL at reduced pressure drop, using pin fin.

Finally, the impact of cooling channel geometric configuration (in combination with stoichiometry ratio, relative humidity and coolant Reynolds number) on effective thermal heat transfer and performance in the fuel cell system was investigated. This is with a view to determine effective thermal management designs for this class of fuel cell. Numerical results shows that operating parameters such as stoichiometry ratio, relative humidity and cooling channel aspect ratio have significant effect on fuel cell performance, primarily by determining the level of membrane dehydration of the PEM fuel cell. The result showed the possibility of operating a PEM fuel cell beyond the critical temperature ( 80°C), using the combined optimised stoichiometry ratio, relative humidity and cooling channel geometry without the need for special temperature resistant materials for the PEM fuel cell which are very expensive.

The study methodology is the first of its kind in South Africa on PEM fuel cell system and deemed to pave way for enhanced sustainable energy development through fuel cell technology.

Supervisors:  Prof. T. Bello-Ochende/Prof. J.P. Meyer

 

 

Shafiqur Rehman 2012 "WIND POWER RESOURCE ASSESSMENT, DESIGN OF GRID-CONNECTED WIND FARM AND HYBRID POWER SYSTEM"

 

An exponentially growing global population, power demands, pollution levels and, on the other hand, rapid advances in means of communication have made the public aware of the complex energy situation. The Kingdom of Saudi Arabia has vast open land, an abundance of fossil fuel, a small population but has always been among the front-runners where the development and utilisation of clean sources of energy are concerned. Several studies on wind, solar and geothermal sources of energy have been conducted in Saudi Arabia. Solar photovoltaic (pv) has been used for a long time in many applications such as cathodic protection, communication towers and remotely located oil field installations. Recently, a 2MW grid-connected pv power plant has been put online and much larger solar desalination plants are in planning stage.

Wind resource assessment, hub height optimisation, grid-connected wind farm and hybrid power system design were conducted in this study using existing methods. Historical daily mean wind speed data measured at 8 to 12metres above ground level at national and international airports in the kingdom over a period of 37 years was used to obtain long-term annual and monthly mean wind speeds, annual mean wind speed trends, frequency distribution, Weibull parameters, wind speed maps, hub height optimisation and energy yield using an efficient modern wind turbine of 2.75MW rated power. A further detailed analysis (such as estimation of wind shear exponent, Weibull parameters at different heights, frequency distribution at different heights, energy yield and plant capacity factor and wind speed variation with height) was conducted using wind speed measurements made at 20, 30 and 40metres above ground level.

As a first attempt, an empirical correlation was developed for the estimation of near-optimal hub height (HH = 142.035 * (α) + 40.33) as a function of local wind shear exponent (α) with a correlation coefficient of 97%. This correlation was developed using the energy yield from a wind turbine of 1 000kW rated power and wind speed and local exponent for seven locations in Saudi Arabia. A wind-pv-diesel hybrid power system was designed and specifications were made for a remotely located village, which is being fed 100% by diesel power generating units. The proposed system, if developed, will offset around 35% of the diesel load and therefore will result in decreased air pollution by almost the same amount.     

The developed wind speed maps, the frequency distributions and estimated local wind shear exponents for seven locations and energy yield will be of great help in defining the further line of action and policy-building towards wind power development and utilisation in the kingdom. The study also recommends conducting a wind measurement campaign using tall towers with wind measurements at more than one height and estimating the local wind shear exponents and developing a wind atlas for the kingdom. The study further states that a grid-connected wind farm of moderate capacity of 40MW should be developed using turbines of varying rated powers. The wind speed data was also analysed using wavelet transform and Fast Fourier Transform (FFT) to understand the fluctuation in wind speed time series for some of the stations. It is also recommended that policy-makers should take firm decision on the development of hybrid power systems for remotely located populations which are not yet connected with the grid. There are two challenges which need research: one is the effect of dust on the moving and structural elements of the wind turbines and the second is the effect of high prevailing temperatures on the performance and efficiency of the same.

Supervisor: Dr. Md. Mahbub Alam

Co-supervisors: Prof. JP Meyer and Dr. Luai M. Al-Hadhrami

C Kat  2012 "Validated leaf spring suspension models "

As all simulation models in this study are required to be validated against experimental measurements a thorough experimental characterisation of the suspension system of interest, as well as two different leaf springs, are performed. In order to measure the forces between the suspension attachment points and the chassis two six component load cells were developed, calibrated, verified and validated.

This study will primarily focus on the modelling of multi-leaf and parabolic leaf springs. The study starts with a literature study into the various existing modelling techniques for leaf springs. A novel multi-leaf spring model, which is based on a macro modelling view point similar to that used for modelling material behaviour, is developed. One of the modelling techniques found in the literature i.e. neural networks is also used to model the leaf spring. The use of neural networks is applied and some of the challenges associated with the method is indicated. The accuracy and efficiency of the physics-based elasto-plastic leaf spring model and the non physics-based neural network model are compared. The accuracy is calculated using a new quantitative validation metric that is able to give an intuitive and reliable account of the accuracy between two signals. The quantitative validation metric is based on the well-known, and frequently used, relative error. The modified percentage relative error metric that is developed accounts for the known challenges associated with using the relative error on signals with a periodic nature around zero. The modified percentage relative error metric is compared to two other quantitative validation metrics that were identified from the literature study. It is concluded that the modified percentage relative error has certain limitations but that it is able to give an accurate and reliable account of the agreement/disagreement between two periodic signals around zero. The modified percentage relative error is used to obtain the accuracies of the two models and both give good results with the neural network being almost 3 times more computationally efficient. The elasto-plastic leaf spring model, for the multi-leaf spring, is then further extended to model the behaviour of a parabolic leaf spring. It was also combined with a method that is able to capture the effect of changes in the spring stiffness due to changes in the loaded length. Qualitative validation using experimental data shows that the elasto-plastic leaf spring model is able to predict the vertical behaviour of the parabolic leaf spring. Quantitative validation also shows that the method proposed for accounting for the change in stiffness due to changes in the loaded length is able to capture this characteristic of the physical leaf spring.

Following the systematic modelling approach the elasto-plastic multi-leaf spring model is incorporated into a model of a simplified version of the physical suspension system. The quantitative validation results from this model show that the model is able to accurately predict the forces that are transmitted from the suspension system to the chassis.

Supervisor: Prof PS Els

KeSheng Wang  2011 "Approaches to the improvement of order tracking techniques for vibration based diagnostics in rotating machines"

In order to perform effective and reliable simulations in the CAE process, accurate simulation models of the product and its associated systems, sub-systems and components are required. In the vehicle dynamics context simulation models of the tyres, suspension, springs, damper, etc, are needed. This study will look at creating validated models of leaf spring suspension systems used on commercial vehicles. The primary goal set for the models are to be able to predict the forces at the attachment points where the suspension system is attached to the vehicle chassis as the models are to be used in full vehicle durability simulations. The most important component in this suspension model is the leaf spring. Leaf springs have been used in vehicle suspensions for many years. Even though leaf springs are frequently used in practice they still hold great challenges in creating accurate mathematical models. It is needless to say that an accurate model of a leaf spring is needed if accurate full vehicle models are to be created.

Conventional rotating machine vibration monitoring techniques are based on the assumption that changes in the measured structural response are caused by deterioration in the condition of the rotating machine. However, due to variations of the rotational speed, the measured signal may be non-stationary and difficult to interpret. For this reason order tracking technique is often introduced. One of the main advantages of order tracking over traditional vibration monitoring lies in its ability to clearly identify non-stationary vibration data and to a large extent exclude the influences of varying rotational speed.

This work synthesizes different order tracking techniques. Through combination, exchange and reconciliation of ideas between these order tracking techniques, three improved order tracking techniques are developed for the purpose of enhancing order tracking analysis in condition monitoring. The techniques are Vold-Kalman filter and computed order tracking (VKC-OT), intrinsic mode function and Vold-Kalman filter order tracking (IVK-OT) and intrinsic cycle re-sampling (ICR). Indeed, these improved approaches contribute to current order tracking practice, by providing new order tracking methods with new capabilities for condition monitoring of systems which are intractable by traditional order tracking methods, or which enhances results obtained by these traditional methods.

Supervisor: Prof PS Heyns