Graduation - 2024 - March (Open Access)

Permanent URI for this collection

https://scholar.sun.ac.za/handle/10019.1/125747

Browse

Now showing 1 - 7 of 7

A decision support tool for implementing machine learning in SME manufacturing companies.
(2024-03) van Coller, Christiaan; Louw, Louis; Palm, Daniel; Stellenbosch University. Faculty of Engineering. Dept. of Industrial Engineering. Engineering Management (MEM).
ENGLISH ABSTRACT: The manufacturing industry is an integral part of a country’s economy. Having a strong manufacturing industry, and especially strong small and medium-sized enterprises (SMEs) in manufacturing, brings a significant benefit to countries and their people. A tool that can be used for improving manufacturing companies is machine learning. Machine learning can be used to significantly improve efficiency, traceability, reduce costs, as well as many other benefits. There are, however, many limitations, especially for SMEs, that can withhold them from unlocking the full potential of machine learning. This thesis aims to provide SMEs with the necessary tools to implement machine learning into their manufacturing operations. This includes the necessary requirements, benefits, and drawbacks to implementing machine learning. This aim is achieved through a literature review identifying the needs, requirements, and benefits of implementing machine learning in manufacturing, as well as the use of a decision support tool for implementing machine learning into the manufacturing operations of SMEs. This thesis focuses specifically on the manufacturing industry; therefore, all context falls within the manufacturing industry. Although the tool might have limited applications outside of manufacturing, it is designed for the manufacturing industry. The tool generates a project plan, guidelines for data preparation, the most applicable algorithm for the scenario as well as template code, evaluation guidelines, and deployment guidelines. These are generated using information provided by the user of the tool that is processed by OpenAI’s GPT-4 large language model. These outputs are validated using dataset tests, interviews, and a case study. The validation process showed that the tool is accurate and reliable, relevant in its recommendations, user friendly in terms of its user interface, and that it provides accurate and practical code for the appropriate algorithms. The thesis as a whole highlights the unique challenges and limitations faced by SME manufacturing companies, after which it addresses the challenges associated with machine learning. For further research, improvements can be made to the decision support tool to make it a direct channel for machine learning implementation. Having a functionality that allows users to upload their data will significantly improve the guidance of the tool as well as the algorithm that will be suggested.
Development and heat treatment of functionally graded 316L stainless steel-tungsten carbide coatings manufactured using laser cladding
(Stellenbosch : Stellenbosch University, 2024-02) Tendere, Takudzwa Lewis; Sacks, Natasha; Stellenbosch University. Faculty of Engineering. Dept. of Industrial Engineering.
ENGLISH ABSTRACT: This thesis details the development and analysis of a functionally graded tungsten carbide (WC)-316L stainless steel (SS316L) coating fabricated through laser cladding (LC) technology, followed by stress relief heat treatment. The primary goal was to construct a three-layer functionally graded coating and assess its mechanical and microstructural properties pre- and post-heat treatment. A comprehensive literature review identified a notable gap concerning functionally graded SS316L and WC coatings, with existing one-layer coatings often prone to delamination and cracking. This research gap underlined the necessity for developing functionally graded coatings. The study utilized a full factorial Design of Experiments (DoE) framework to determine the optimal laser cladding processing parameters. This approach involved adjusting laser power and scanning speed to achieve the desired coating properties of 1 mm thickness per layer with minimal porosity. The optimal parameters, identified via regression analysis and analysis of variance, facilitated the deposition of a three-layer coating, each layer distinct in its SS316L and WC composition, thereby establishing a functionally graded structure. Prior to and following heat treatment, the coatings underwent extensive microstructural characterization through XRD analysis, optical microscopy, SEM/EDS, and Vickers hardness testing. Simulations with JmatPro software for phase diagrams ensured the selected heat treatment temperature effectively relieved residual stresses without inducing secondary phases or altering the core microstructure. Post-treatment comparative analysis using SEM, XRD, and hardness tests indicated no substantial microstructural changes. However, the hardness from the substrate and to the coating layer had a noticeable decrease of 5 %, 8 %, 10 % and 15 % respectively. The results of the study have significant implications for surface engineering applications requiring both corrosion and wear resistance. The research contributes to a better understanding of functionally graded coatings, particularly in the context of WC/SS316L materials. While it addresses specific challenges in surface engineering, it also suggests potential directions for further studies in the application of functionally graded materials in various engineering environments.
Development of a test setup to determine frictional losses on bicycle drivetrains.
(Stellenbosch : Stellenbosch University, 2024-02) Loedolff, C; Venter, G; Stellenbosch University. Faculty of Engineering. Dept. of Mechanical and Mechatronic Engineering.
ENGLISH ABSTRACT: Cyclists compete purely under their own power, however the power output to the wheels is influenced by their equipment. Bicycle drivetrains are thus engineered to be as light and efficient as possible, but this may impact longevity without proper care and lubrication. Chain lubricants are developed for specific conditions, with characteristics resulting in better efficiency for shorter durations, or better resistance to contamination. To accurately determine the power losses in a given drivetrain for any given load condition, a test bench is required to simulate the real-world conditions these drivetrains are subjected to. This work presents the development of a test bench that is capable of performing three discreet tests: the frictional losses due the static radial loading on ball bearings used in the bottom brackets of bicycles, the isolated chain losses due to chain tension, and the complete system losses under full load conditions. All these tests require the measurement of speed and torque data to determine the power losses. In addition, the full load test requires a braking mechanism to supply the load. The speed and torque measurements were incorporated into a single, bespoke transducer. Strain gauge based torque transducers are the norm for industrial equipment, however a unique approach was pursued here. The transducer was constructed from a thin aluminium tube that deflects angularly under load. The ends of the tube were fitted with slotted disks which run through the line-of-sight path of a set of interrupter optical sensors. The sensor output signals were used to determine the angular velocity of each disk, as well as the relative angular deflection between the two disks. The angular deflection of the tube was used together with its known stiffness to determine the torque measured by the transducer. The sensor was calibrated and rated to measure torque above 0.2 Nm. Alongside the torque transducer, a custom eddy current brake was developed using permanent magnets with a spinning aluminium disk. This braking mechanism was chosen as it provides smooth loading and can be designed to accurately modulate the braking power. Control was successfully achieved by changing the relative position of the magnets to adjust the strength of the magnetic flux through which the disk is moving. The exact braking power of the eddy current brake is adjusted by using feedback control from the developed torque sensors. Filtering and smoothing of the measured speed and torque data was performed to ensure more stable control. Full system testing shows the torque induced by radially loaded bearings are lower than expected, while the developed sensor accurately measured the higher loading associated with full load conditions. Additionally, the eddy current brake is verified to supply the required braking power of 250Wduring full load testing. Drivetrain losses are seen to depend on the applied lubricant, with the highest efficiency measured to be 97.5 %.
Exploring narratives as a tool to manage psychosocial distress in children hospitalised for tuberculosis in the Western Cape, South Africa
(Stellenbosch : Stellenbosch University, 2024-03) October, Caitlin Danielle; Hoddinott, Graeme; Coetzee, Bronwyne; Stellenbosch University. Faculty of Arts and Social Sciences. Department of Psychology.
ENGLISH ABSTRACT: Background: Children who are hospitalised for long periods are at risk of developing emotional, behavioural and social difficulties. Hospitalisation for any reason is known to be a stressful experience for children. Some interventions aimed at alleviating the psychosocial distress associated with long-term hospitalisation among children have been successful, although none have focused specifically on children hospitalised for tuberculosis care in South Africa. “Courageous Curly” is a set of narrative tools (a story with activities for use by hospital staff to facilitate the child’s understanding of their hospitalisation) adapted from successful interventions and further developed in Cape Town. Aims: I aimed to explore children’s experiences of a pilot implementation of Courageous Curly to inform further refinement. My specific objectives were: 1) To describe how children talk about their experiences of being in hospital for TB, 2) To observe and document how children interact with the teacher and each other during the presentation of Courageous Curly, and 3) To describe how children talk about their experiences of being in hospital for TB treatment after they have participated in Courageous Curly. Methods: The study followed an exploratory qualitative design situated within the narrative paradigm. To achieve objective 1, I analysed case descriptions of interviews by doing a secondary analysis with children hospitalised with MDR-TB at Brooklyn Chest Hospital to identify the negative experiences children had during hospitalisation. To achieve objective 2, I conducted observations of the Courageous Curly story being told to children at the school situated at BCH over a period of 13 days. To achieve objective 3, I interviewed children (n=6) about their experience of hearing the Courageous Curly story. I analysed the data of all three objectives through thematic analysis. Findings: Children had four types of negative experiences due to hospitalisation, namely: negative emotions, behavioural responses, a lack of knowledge of being admitted to BCH, and a lack of understanding of other individuals at BCH. Children were able to participate in Courageous Curly and interacted well during implementation. Children who had experienced Courageous Curly said that they enjoyed it and that it made them feel better about being in hospital. Conclusion: Similar to children hospitalised for other reasons, children who are hospitalised for TB care experience psychosocial distress due to long-term hospitalisation. Participants in this study enjoyed participating in the narrative intervention, Courageous Curly, and they used it to narrate their own experiences while participating in the activities as part of the story. Such interventions are crucial for mitigating the potential negative effects on psychological, social, emotional development and well-being in children who are hospitalised for prolonged periods. The initial evidence of the pilot project with Courageous Curly is that it is possible to implement it with good participation from children. Children enjoyed participating in Courageous Curly, and it positively influenced their experience of hospitalisation. Further research is needed to refine and evaluate its effectiveness in multiple contexts.
Improvement of the acoustic properties of tonewoods
(Stellenbosch : Stellenbosch University, 2024-02) Nefdt, Keenan Colin; Meincken, Martina; Stellenbosch University. Faculty of AgriSciences. Dept. of Forest and Wood Science.
ENGLISH ABSTRACT: This work investigated different wood modification techniques to improve the acoustical properties of tonewoods, in particular the sound radiation ratio (R). The treatments used were heat-, fungal - and UV exposure, as well as immersion into NaOH and Na2SO3 solution and a combination of the most successful treatments. All initial experiments were performed on pine, before replicating the best performing treatment on spruce due to cost factors. The main objective was to reduce the hemicellulose content without severely degrading cellulose, which results in a reduction of density while maintaining, or even improving the elasticity (MOEL). The results indicated that it is possible to increase R and MOEL significantly with heat-, fungal – and sodium treatment. Overall, the combined heat-fungal and heat-sodium treatments performed the best and managed to increase R by up to 20 %. Sodium treatment led to the best increases in R, but compromised the wood structure and the treatment protocol needs to be developed further. Consequently, this study found the most successful wood treatment to be exposure to white rot combined with heat treatment.
Integral abutment bridges – continuum modelling of soil-structure interaction using finite element analysis with interface elements.
(Stellenbosch : Stellenbosch University, 2024-02) Dreyer, Tashreeq; Van Zijl, Gideon P. A. G.; Fataar, Humaira; Stellenbosch University. Faculty of Engineering. Dept. of Civil Engineering.
ENGLISH ABSTRACT: This thesis presents a computational study on the soil-structure interaction (SSI) behaviour stemming from cyclic thermal-induced movements of integral abutment bridges (IABs) using finite-element engineering software DIANA-FEA. An existing IAB, the Van Zylspruit Bridge in South Africa, is used as a template for the modelling presented in this study, with instrumented earth pressure and abutment displacement readings serving as validation of the models. The abutment-backfill section of the bridge is considered for the SSI modelling using continuum element methodology. The hardening soil model (HSM or HS model) is a numerical soil model that approximates real soil behaviour, including densification due to plastic deformation, stress-dependent stiffness, plastic straining (or yielding) in primary compression and deviatoric loading, isotropic hardening, and dilation. Soil failure is governed by a modified Mohr-Coulomb criterion in this model. The HS model is used to model the backfill, whereas the concrete and in-situ soil materials are assigned linear elastic models for simplicity. In continuum mechanics, ratcheting is the incremental build-up in pressure due to imposed cyclic loading. Ratcheting is prevalent in the backfills of IABs due to imposed cyclic thermal-induced displacements in the deck which are transferred to the abutments. Models presented in this thesis include interface modelling to allow separation between the integral abutment and retained granular (silica sand) fill to capture ratcheting and settlement behaviour in the backfill. The effective bridge temperature (EBT) with daily and seasonal cycles was measured on site from January 2016 to August 2017. This EBT “double cycle” is used to model imposed deformations on the abutment in order to capture ratcheting behaviour in the backfill. The Van Zylspruit bridge has a buried approach slab with one end fixed to the abutment (as a cantilever member). To evaluate the effect that approach slabs have on SSI behaviour, two models were created: one with an approach slab (Slab model) and the other without (No-Slab model). In the Slab model, interface elements were defined between the edges of the approach slab and the backfill to capture SSI behaviour stemming from the slab’s inclusion. The presence of the slab typically reduced settlements, but the development of a gap beneath the slab that grew with seasonal cycles prompts engineers to exercise caution with cantilevered slab designs. Results from both models showed reasonable agreement with measured data in terms of lateral earth pressure and abutment displacement. Two additional clay materials (one stiff and one soft) are used as alternative fill materials (with HSM parameters) to evaluate their effectiveness as backfill materials for IABs, with and without the approach slab present. The stiff clay was found to behave similarly to granular soil both in terms of stress ratcheting and backfill settlement. The soft clay was found to behave unfavourably, especially with the slab present, causing excessive backfill settlements over three times that of the stiff clay. A parameter study on the Mohr-Coulomb parameters (cohesion, friction angle, and dilatancy angle) of the HS model is conducted on the Slab model with granular fill. The values are modulated in reasonable ranges to provide settlement data on granular soils that might be considered as fill material in South Africa. It was observed that high dilatant granular materials with low friction cause excessive heave at the far end of backfill, in excess of 5mm after 5 cycles of ±5.5 mm abutment displacement amplitude. Low-friction and low-dilatant materials typically behaved unfavourably, with settlements across the backfill reaching 4 mm with the same aforementioned displacement criteria. The findings in this study may be beneficial to the current discourse surrounding IABs in the bridge engineering industry as well as the enhancement of modelling IABs using continuum mechanics methodology.
A model for accurate error propagation in a convergent stereovision system.
(Stellenbosch : Stellenbosch University, 2024-02) Ezebili, Ifeanyi Francis; Schreve, K.; Stellenbosch University. Faculty of Engineering. Dept. of Mechanical and Mechatronic Engineering.
ENGLISH ABSTRACT: Stereovision is a camera-based imaging technique that facilitates the reconstruction of the 3-space coordinates and depth of a scene point using the images acquired from two cameras. Generally, stereovision finds application in autonomous vehicle navigation, mobile robots, parts inspection for quality assurance, and tracking and identification of objects. Like the output of a typical metrological system, the 3D coordinates measured with a stereovision system have associated measurement uncertainties. Such uncertainties practically emerge from the errors which are concomitant with each of the steps involved in stereo camera measurement. In this dissertation an analytic epipole-featured model is developed and proposed for structure computation and 3-space depth measurement in convergent stereo camera imaging. The proposed reconstruction model is predicated on the image sensor parameters of both cameras, left and right, together with two extrinsic parameters, namely the baseline distance and the stereo projection angle of the scene point. The intrinsic parameters are normalized with respect to the focal lengths of the cameras. The proposed model is characterized by less computational complexity and short execution time and can be employed in active vision-based metrology in which the imaging stereo cameras are rotated about their vertical axes relative to each other. The terms virtual depth and depth factor or depth coefficient are subsequently introduced and described. Both quantities together define the depth of a world point relative to the coordinate frame of the reference camera. From the developed reconstruction model, an equivalence relation between coplanar parallel and convergent stereo camera imaging systems is established. The relation states that for double-view geometry in computer vision, disparity in a row-aligned, coplanar-parallel stereo camera configuration is equivalent to the baseline-to-depth-factor ratio in a convergent stereo camera configuration. This baseline-to-depth-factor ratio in convergent stereo camera imaging is termed convergent stereo disparity and can be identified and equated with the image rectification process in a practical conventional coplanar-parallel stereo camera setup. Incorporating the epipoles in the developed reconstruction model facilitates the establishment of the stereo equivalence relation and the definition of convergent stereo disparity in stereovision. Furthermore, generalized mathematical analyses are done to model and study the variation of depth sensitivity and relative depth uncertainty with respect to convergent stereovision system parameters for 3-space points using the developed reconstruction model. It is observed that different values of left and right focal lengths are required to achieve high sensitivity coefficient, a condition that is not conformable with the conventional practice of having the same left and right focal lengths in stereo camera imaging. Regarding the variation of the stereo projection and stereo convergence angles, there are trade-offs between depth sensitivity coefficient and relative depth uncertainty. It is found that a stereo convergence angle of 90° yields the best relative depth uncertainty value at which the focal length-normalized epipole-to-principal point distances on both image planes are reciprocals. The analytic derivations and graphical characteristics of the depth sensitivity coefficient would give a stereovision system designer some information regarding the margin by which the estimated depth of a scene point changes for a drift in the value of any stereo camera parameter, and also some idea in respect of the potential trade-offs involved in the choice of certain parameters. The performance of the developed reconstruction model is studied, and its accuracy tested by comparing the 3-space coordinates it predicts to those obtained by Gold Standard triangulation algorithm and to the ground truth results. In terms of execution speed the proposed reconstruction model exhibited a computation time of 0.6 ms compared to 6.2 ms and 9.9 ms recorded for the Direct Linear Transformation (DLT) and Gold Standard triangulation algorithms respectively. The measurement errors determined by theoretical methods based on the law of error propagation incorporating the analytic reconstruction model (with and without full input covariance matrices) are compared with those obtained by the experimental approach. Strong correlations are found to exist between the two sets of values obtained, indicating the validity of the error model. The study of measurement error using the reconstruction model shows that accounting for the covariances of all the stereo camera parameters in vision-based metrology predicts smaller errors compared to when the covariances of the parameters are ignored. It is also found that it makes no significant difference if full or diagonal input covariance matrices are used in the theoretical computation of error compared to the experimental approach to determining the error. The error model derived in this work and predicated on the developed epipole-dependent reconstruction model would be useful in the design of high-precision stereovision systems.

Browse

Browsing Graduation - 2024 - March (Open Access) by Title

Results Per Page

Sort Options