M. L. Combrinck,2016 "BOUNDARY LAYER RESPONSE TO ARBITRARY ACCELERATING FLOW"
This thesis was aimed developing a fundamental understanding of the boundary layer response to arbitrary motion. In this context arbitrary motion was defined as the unsteady translation and rotation of an object.
Research objectives were developed from the gaps in knowledge as defined during the literature survey. The objectives were divided into three main activities; mathematical formulations for non-inertial bulk flow and boundary layer equations, implementation of said formulations in a numerical solver and simulations for various applications in arbitrary motion.
Mathematical formulations were developed for the bulk flow and boundary layer equations in arbitrary motion. It was shown that the conservation of momentum and energy equation remains invariant in the non-inertial forms. The conservations of momentum equation can at most have six fictitious terms for unsteady arbitrary motion. The origins of the terms were found to be from transformation of the material derivative to the non-inertial frame. All fictitious terms were found to be present in the boundary layer equations; none could be eliminated during an order of magnitude analysis.
The vector forms of the non-inertial equations were implemented in a novel OpenFOAM solver. The non-inertial solver requires prescribed motion input and operate on a stationary mesh. Validation of the solver was done using analytical solutions of a steady, laminar flat plate and rotating disk respectively.
Numerical simulations were done for laminar flow on a translating plate, rotating disk and rotating cone in axial flow. A test matrix was executed to investigated various cases of acceleration and deceleration over a range of 70 g to 700 000g. The boundary layer profiles, boundary layer parameters and skin friction coefficients were reported.
Three types of boundary layer responses to arbitrary motion were defined. Response Type I is viscous dominant and mimics the steady state velocity profile. In Response Type II certain regions of the boundary layer are dominated by viscosity and others by momentum. Response Type III is dominated by momentum. In acceleration the near-wall velocity gradient increases with increasing acceleration. In deceleration separation occurs at a result of momentum changes in the flow.
The mechanism that causes these responses has been identified using the developed boundary layer equations. In acceleration the relative frame fictitious terms become a momentum source which results in an increase in velocity gradient at the wall.
In deceleration the relative frame fictitious terms become a momentum sink that induced an adverse pressure gradient and subsequently laminar separation.
Keywords:Non-Inertial Reference Frames, Fictitious Forces, Boundary Layer Equations, OpenFOAM, Laminar Flat Plate, Laminar Rotating Disk, Rotating Cone in Axial Flow.
Supervisor: Prof. L.N. Dala
Chabala Lloyd Ngo, 2016 "NATURAL CONVECTION AND RADIATION HEAT LOSS IN SOLAR CAVITY RECEIVERS – NUMERICAL MODELLING, PERFORMANCE ENHANCEMENT AND OPTIMISATION"
Energy access is increasingly seen as a vital catalyst to wider social and economic development, which enables education, health and sustainable agriculture, and creates jobs. Therefore, sustainable growth and development in society needs energy supply that is readily available, affordable, renewable and efficient without causing many negative societal impacts, such as environmental pollution and its consequences. In this regard, concentrating solar power technology has great potential to be used for energy production and it is a promising alternative to conventional fossil fuel-based energy technologies, such as coal power plants, due to the abundance of solar energy as an energy resource, as well as its minimal impact on the environment. The parabolic dish receiver assembly is one such promising concentrating solar power technology. It usually consists of a reflector in the form of a dish with a downward-facing receiver at the focus of the dish. A cavity receiver is used to maximise the absorption of the concentrated flux. However, the receiver is subjected to environmental variations, as well as changes in receiver inclination angle, which lead to heat losses that affect the overall receiver’s performance.
The need for the commercialisation of economically viable parabolic dish systems necessitates further in-depth investigation into cavity receiver designs. As the cavity receiver plays a critical role in transferring solar heat to the engine, any heat loss from the cavity receiver can significantly reduce the efficiency and, consequently, the system’s cost effectiveness. It is therefore essential to assess and effectively minimise heat loss in the cavity receiver to improve the thermal performance of the system, which can contribute to the commercialisation of this type of technology.
This research focused on the modified cavity receivers that are employed in medium- and high-temperature solar dish systems with operating temperatures of up to 1 200 K. Firstly, a three-dimensional numerical investigation was conducted on a modified cavity receiver to quantify the natural convection heat loss, and to determine the effects of the operating temperature, receiver inclination angle and aperture size on heat loss. Furthermore, visualisation results, such as temperature contours, were presented to gain insight into the effects of natural convection. The Boussinesq and non-Boussinesq fluid models were used in the numerical investigation and a comparison was made between them.
Secondly, a novel approach of suppressing natural convection heat loss in a cavity receiver was investigated. The proposed model has not been observed in literature. A cavity receiver with plate fins attached to the inner aperture surface was investigated as a possible low-cost means of suppressing natural convection heat loss in a cavity receiver. Employing air as the working fluid, laminar natural convection heat transfer from the cavity receiver with plate fins attached to the inner aperture surface was investigated for a range of Rayleigh numbers, inclination angles, and fin heights and thicknesses. Furthermore, visualisation results, such as fluid flow and temperature contours, were presented to gain insight into the suppression of natural convection. In addition, a numerical optimisation tool was used to select the best plate fin geometric configuration that improves cavity receiver performance at minimum natural convection heat loss.
Finally, a numerical study and optimisation of the combined laminar natural convection and surface radiation heat transfer in the cavity receiver with plate fins were conducted, and a three-dimensional simulation model was developed to estimate and optimise the convective and radiative heat loss. The influence of operating temperature, emissivity of the surface, orientation and the geometric parameters on total heat loss (convection and radiation) from the receiver were investigated. The results in steady state were obtained for a Rayleigh number range of 105 to 107. The overall thermal efficiency of the receiver was also analysed at different operating temperatures.
From this research, it can be concluded that there is a significant deviation between the Boussinesq and non-Boussinesq models of up to 20% at high temperatures. Therefore, natural convection at high temperature differences can accurately be predicted using the non-Boussinesq model. It was also observed that a significant reduction in natural convection heat loss (up to 20%) from the cavity receiver can be achieved through plate fins, which act as heat suppressors. The results obtained provide a novel approach for improving the design of cavity receivers for optimal performance.
When natural convection was studied together with radiation, the overall cavity efficiency marginally increased by approximately 2% with the insertion of fin plates in the cavity receiver, although the convective heat loss was suppressed by about 20%. This is due to the fact that radiation heat loss dominates at high operating temperatures compared to convective heat loss.
Keywords: parabolic dish; cavity receiver; natural convection; radiation; plate fin; Rayleigh number.
Supervisor: Prof Tunde Bello-Ochende
Co-Supervisor: Prof J.P. Meyer
O.O. Noah,2016 "EXPERIMENTAL, THEORETICAL AND NUMERICAL INVESTIGATION OF NATURAL CONVECTION HEAT TRANSFER FROM HEATED MICROSPHERES IN A SLENDER CYLINDRICAL GEOMETRY"
The ability of coated particles of enriched uranium dioxide (UO2) fuel to withstand high temperatures and contain the fission products in the case of a loss of cooling event is a vital passive safety measure over traditional nuclear fuel requiring active safety systems to provide cooling. As a possible solution towards enhancing the safety of light-water reactors (LWRs), it is envisaged that the fuel in the form of loose-coated particles in a helium atmosphere can be introduced inside Silicon-Carbide nuclear reactor fuel cladding tubes of the fuel elements. The coated particles in this investigation were treated as a bed from where heat was transferred to the cladding tube by means of helium gas and the gas movement was by natural convection. Hence it was proposed that light-water reactors (LWR) could be made safer by redesigning the fuel in the fuel assembly (see Fig. 1.3b).
As a first step towards the implementation of this proposal, a proper understanding of the mechanisms of heat transfer, fluid flow and pressure drop through a packed bed of spheres during natural convection was of utmost importance. Such an understanding was achieved through a review of existing literature on porous media. However, most heat transfer correlations and models in heated packed beds are for forced convectional conditions and as such characterise porous media as a function of Reynolds number only rather than expressing media heat transfer performance as a function of thermal properties of the bed in combination with the various components of the overall heat transfer. The media heat transfer performance considered as a function of thermal properties of the bed in the proposed design is found to be a more appropriate approach than the media as a function of Reynolds number.
The quest to examine the particle-to-fluid heat transfer characteristics expected in the proposed new fuel design led to implementing this research work in three phases, namely experimental, theoretical and numerical simulation. An experimental investigation of fluid-to-particle natural convection heat transfer characteristics in packed beds heated from below was carried out. Captured data readings from the experiment were analysed and heat transfer characteristics in the medium evaluated by applying the first principle heat transfer concept. A basic unit cell (BUC) model was developed for the theoretical analysis and applied to determine the heat transfer coefficient, h, of the medium. The model adopted a concept in which a single unit of the packed bed was analysed and taken as representative of the entire bed; it related the convective heat transfer effect of the flowing fluid with the conduction and radiative effect at the finite contact spot between adjacent unit cell particles. As a result, the model could account for the thermophysical properties of sphere particles and the heated gas, the interstitial gas effect, gas temperature, contact interface between particles, particle size and particle temperature distribution in the investigated medium. Although the heat transfer phenomenon experienced in the experimental set-up was a reverse case of the proposed fuel design, the study with the achievement in the validation with the Gunn correlation aided in developing the appropriate theoretical relations required for evaluating the heat transfer characteristics in the proposed nuclear fuel design.
A slender geometrical model mimicking the proposed nuclear fuel in the cladding was numerically simulated to investigate the heat transfer characteristics and flow distribution under the natural convective conditions anticipated in beds of randomly packed spheres (coated fuel particles) using a commercial code. Random packing of the particles was achieved by discrete element method (DEM) simulation with the aid of Star CCM+ while particle-to-particle and particle-to-wall contacts were achieved through the combined use of the commercial code and a SolidWorks CAD package. Surface-to-surface radiative heat transfer was modelled in the simulation reflecting real-life application. The numerical results obtained allowed for the determination of parameters such as particle-to-fluid heat transfer coefficient, Nusselt number, Grashof number and Rayleigh number. These parameters were of prime importance when analysing the heat transfer performance of a fixed bed reactor.
A comparison of three approaches indicated that the application of the CFD combined with the BUC model gave a better expression of the heat transfer phenomenon in the medium mimicking the heat transfer in the new fuel design
Keywords: nuclear fuel; light-water reactor; porous medium; numerical simulation; natural convection heat transfer
Supervisor: Prof J. Slabber
Co-Supervisor: Prof J.P. Meyer
B.H.Freyer, 2016. " Active tool vibration control and tool condition monitoring using a selfsensing actuator"
The studies consist of two simulations of active tool vibration control and tool condition monitoring
respectively and a hardware-in-the-loop laboratory demonstration of active tool vibration control
typical to turning.
Besides reducing the restricting effects of tool vibrations on productivity, work-piece surface finish
and tool life, it is desirable to handle lack of space at the tool tip and the cost of control systems in
turning processes in an effective way. These two aspects are here considered by means of the concept
of a self-sensing actuator (SSA) in the simulation of tool vibration control. In the simulation an IIRfilter
represents the structure of the passive tool holder. A known pre-filtering technique was applied
to the error in a feedback filtered-x LMS algorithm to maintain the stability of the control system. The
self-sensing path is modelled and illustrated. The IIR-filters and their inverses were used for
modelling this path, with equations resulting from the nodal displacements associated with nodes that
have forces acting on them. For the cantilever type structure a considerable reduction of 93% of the
displacement r.m.s. values of the tool tip, was obtained when using this control system.
Signal processing using orthogonal cutting force components for tool condition monitoring (TCM)
has established itself in literature. Single axis strain sensors however limit TCM to linear combination
of cutting force components. This situation may arise when a single axis piezoelectric actuator is
simultaneously used as an actuator and a sensor, e.g. its vibration control feedback signal exploited
for monitoring purposes. Processing of a linear combination of cutting force components to the
reference case of processing orthogonal components is compared. The same time-delay neural
network structure has been applied in each case. Reconstruction of the dynamic force acting at the
tool tip in a turning process is described. By simulation this dynamic force signal was applied to a
model of the tool holder equipped with a SSA. Using a wavelet packet analysis, wear-sensitive
features were extracted. The probability of a difference less than 5 percentage points between the
flank wear estimation errors of abovementioned two processing strategies is at least 95 %.
This study proves the basic concept of adaptive feedback active vibration control in combination with
a self-sensing actuator to control tool vibrations. The structure involved is representative of a tool post
clamped tool holder. The advantages that adaptive control hold when applied to non-stationary
vibrations motivate this investigation. Secondly the dual functionality of a piezoelectric element is
utilized for system simplification. Actuator linearization measures are considered and a model for the
system’s forward path identified. The tool vibrations signal for this work is of 100 Hz bandwidth
around the representative tool holder bending mode. A downscaled force based on real cutting force
characteristics was artificially applied to the representative tool holder. Limited form locking contact
with the tool holder restricted the actuator’s reaction to compressive forces only. Results of up to 70%
attenuation of vibration induced strain on the SSA were achieved. This method clearly shows concept
viability.
Supervisor: Prof. N.J. Theron
Co-supervisor: Prof. P.S. Heyns
Erfan Asaadi,2016. "Improved inverse methods in the characterisation of mechanical behaviour of materials"
Supervisor: Prof P.S Heyns
G.J.Jansen van Rensburg, 2016. "Development and Implementation of State Variable Based User Materials in Computational Plasticity"
The Finite Element Method is a powerful tool that can be used to test, improve or better understand an industrially relevant problem. There are numerous Finite Element Analysis (FEA) software packages that operate either in the commercial, open source or research space. Different application specific codes also have specialised model formulations. Most software packages have a comprehensive list of material models already implemented. If a different material model is required, some form of user material can often be implemented and linked to the software package. In some cases the effective implementation and testing of a user implemented material requires knowledge on the effect and handling of strain formulations, element technologies and the desired material behaviour. With sophisticated material models available in the research space, this thesis focuses on the identification and implementation of existing computational plasticity models for use within FEA.
The effect of different strain formulation choices is first illustrated and discussed using different sample problems. Three different FEA software packages are also compared before discussion and implementation of a general numerical framework for corotated hypo-elastoplasticity in isotropic and combined hardening. The numerical framework allows expansion to include different, more sophisticated hardening behaviour by simply altering the scalar equation used to update the von Mises yield surface.
The Mechanical Threshold Stress (MTS) material model is implemented within the hypo-elastoplastic numerical framework. Material parameter identification is investigated using linear regression on data followed by numerical optimisation. The MTS model is a rate and temperature dependent state variable based material model. The model is tuned to fit imperfect cemented carbide data in compression, where material test frame compliance or some eccentricity caused inhomogeneous deformation through the test section of the specimen. The characterised model is then used on a sample problem to investigate the plastic deformation in the cemented carbide anvils during the High Pressure, High Temperature (HPHT) synthesis of diamond.
Further extensions, built on the dislocation density based modelling theory of the MTS model, are investigated by selecting an alternate form of the state dependent variable. A dislocation density ratio is used instead of the original stress like variable in the MTS model. The evolution of this internal state variable is altered, along with additional state dependent variables, to include additional deformation and thermal mechanisms. The model extensions in the case of rate and temperature dependent cyclic deformation as well as multiple waves of recrystallisation are discussed and implemented. The recrystallisation and through thickness microstructural variation of a High Strength, Low Alloy (HSLA) steel are finally investigated during the process of industrial hot rolling or roughing simulations.
Supervisors: Prof. S. Kok
Co-supervisor: Dr D.N. Wilke
Saheed Adewale Adio, 2016. "Experimental investigation and mathematical modelling of thermophysical properties of ethylene glycol and glycerol-based nanofluids"
Nanofluids are a new class of heat transfer fluids that aim to improve the poor thermal efficiency of conventional heat transfer fluids. The dispersion of nanoparticles into traditional heat transfer fluids, such as water, ethylene glycol, glycerol, engine oil and gear oil, improves the thermal conductivity of base fluids, which has attracted researchers to apply nanofluids in engineering systems. Nanofluids show higher thermal and electrical conductivity. However, in terms of heat transfer performance, viscosity is also important. The viscosity of nanofluids increases due to an increase in the nanoparticle volume fraction, which needs attention and proper experimental investigation to improve the efficiency of nanofluids in heat transfer applications. Consequently, investigation into the effective viscosity of nanofluids is as important as the thermal conductivity.
On the other hand, how nanofluids are prepared can have an effect on the resultant performance. Using an ultrasonication mixer for the dispersion of nanoparticles in the base fluid is one of the most effective and popular methods of preparing nanofluids, especially from the two-step method. Almost all the experimental studies available on nanofluids chose an arbitrary time for the preparation of nanofluids. Choosing an arbitrary time for ultrasonication or any other physical preparation mechanism may be counterproductive. Therefore, in this research, nanofluids are prepared through an optimised two-step method that is assisted with ultrasonic vibration. The resulting homogenised nanofluids are further investigated for the influence of temperature, particle size, volume fraction, base fluid type and particle type on the evolution of the viscosity, pH and electrical conductivity. The temperature range investigated in this thesis is 20 to 70 oC; the nanoparticle volume fraction is up to 5%; the base fluids are ethylene glycol (EG) and glycerol, while the nanoparticle types are MgO, Al2O3 and SiO2 in different sizes.
Viscosity is a very important parameter, especially in systems that involve fluid flow (forced or natural convection) and for numerical analysis. However, most generic models in the literature underpredicted the viscosity evolution of nanofluids. Therefore, it is essential that very accurate models need to be developed for the prediction of the viscosity of nanofluids. To this end, this research also models the viscosity of the different nanofluids using dimensional analysis and regression analysis based on the experimental input-output data. Furthermore, artificial intelligence methods, such as the group method of data handling-neural network (GMDH-NN), genetic algorithm-polynomial neural network (GA-PNN) and fuzzy C-mean clustering-based adaptive neuro-fuzzy inference system (FCM-ANFIS) methods, are used to model the relationships between the experimental input parameters and the viscosity of the nanofluids.
Generally, the viscosity of the nanofluids reduced exponentially with temperature increase and the trends are similar to those displayed by the respective base fluids. However, the viscosity of the nanofluids is higher depending on the concentration of the nanoparticles contained in the nanofluids. Suspending nanoparticles in the base fluid increased the viscosity of the resulting nanofluid, and a further increase in the volume fraction of the nanoparticles increased the effective viscosity of the nanofluids. The viscosity trend of the nanofluids of Al2O3-glycerol is non-linear to volume fraction increase while MgO-EG nanofluids displayed a linear dependence. Regarding the influence of particle size, smaller particles produced a higher energy dissipation rate due to the higher number density, increased Brownian velocity and particle-particle interactions. Therefore, the viscosity was higher in nanofluid samples prepared from smaller nanoparticles. When the same nanoparticle samples were dispersed in different base fluids, it was found that the relative viscosity is different in the different nanofluids, which suggests that base fluid properties are indispensable when discussing the viscosity of nanofluids.
The pH and electrical conductivity of the base fluids did not change much with an increase in temperature, and their values were smaller than unity. The suspension of nanoparticles saw an increase in the values of both the pH and electrical conductivity. As the volume fraction of suspended particles increased, the value of the electrical conductivity and pH also increased commensurately, until counterion condensation effects set in. Increasing the temperature of the nanofluids led to an increase in the electrical conductivity, while the pH generally reduced with an increase in temperature. Although smaller nanoparticles showed slightly higher electrical conductivity values, pH values were convincingly higher across all the volume fractions. The ionisation process is different for the different base fluids, therefore, the pH and electrical conductivity were different for the same nanoparticles suspended in different base fluids.
The viscosity correlations developed in this thesis through dimensional analysis with regression all gave good agreements with the experimental data. When the correlations were compared with some of the prominent, well-cited models in the open literature, they performed better in producing experimental results. Furthermore, the use of GMDH-NN, GA-PNN and FCM-ANFIS for modelling and predicting the effective viscosity for the nanofluids as a function of particle diameter, temperature and volume fraction are presented. The results of the GMDH-NN, GA-PNN and FCM-ANFIS models all showed very good agreement when compared with the experimental data. Therefore, these models come in handy when using these nanofluids for computational fluid dynamics or any other design analyses.
Keywords: Nanofluid, effective viscosity, electrical conductivity, pH, MgO, ethylene glycol, Al2O3, glycerol, SiO2, temperature, volume fraction, nanoparticle size, relative viscosity, ultrasonication, energy density, empirical models, dimensional analysis modelling, GMDH-NN, GA-PNN, FCM-ANFIS.
Supervisors : Dr Mohsen Sharifpur and Prof Josua P Meyer
Copyright © University of Pretoria 2024. All rights reserved.
Virtual Campus
Get Social With Us
Download the UP Mobile App