Faculty of Engineering and the Built Environment (ETDs)
Permanent URI for this communityhttps://hdl.handle.net/10539/37934
For queries relating to content and technical issues, please contact IR specialists via this email address : openscholarship.library@wits.ac.za, Tel: 011 717 4652 or 011 717 1954
Browse
56 results
Search Results
Item Improving grade estimation using machine learning: a comparative study of ordinary kriging against machine learning algorithms(University of the Witwatersrand, Johannesburg, 2024) Akpabio, AniekanThis study investigated the efficiency of machine learning (ML) methods in the accurate prediction of ore grades, placing them in direct comparison with the established Ordinary Kriging (OK) methodology, a mainstay in geostatistical analysis. Utilising a dataset from a complex platinum group elements (PGE) deposit, the research assessed a suite of ML algorithms—namely, Random Forest (RF), Decision Trees (DT), Support Vector Regression (SVR), and particularly 𝑘- Nearest Neighbours (𝑘NN). The latter is highlighted for its adeptness in assimilating spatial data correlations intrinsically, echoing the insights from Nwalia's analytical explorations. The research engages with detailed swath plot analyses, comparative metric evaluations, and a nuanced understanding of spatial continuity, to illustrate the distinct advantages and operational competencies of the models. 𝑘NN, with its reliance on local data proximities and non-parametric nature, alongside RF, with its ensemble-based approach, emerged as capable in point estimate predictions. These models adeptly delineated local grade variations, demonstrating a high degree of reliability to the observed data and outperforming the OK model in both precision and accuracy. Further, the study examined block estimate predictions, a cornerstone in practical mining and resource estimation, where both 𝑘NN and RF demonstrated a commendable ability to generalise predictions over larger spatial extents. This translates into significant potential for enhancing mineral resource estimation processes, tailoring them to the granular specifics of a given ore body, and refining block model accuracy to inform more strategic mining operations. While the results endorse the ML methodologies as robust alternatives to traditional geostatistical techniques, the research also highlights the nuanced nature of these predictions. Factors such as the ore body's heterogeneity, the appropriateness of the variogram model, and the interplay between prediction scale and algorithmic performance are examined, offering a critical lens through which the suitability of each method is assessed. iv The research suggests that while some models like LR and SVR are bounded by linear assumptions and hyperparameter sensitivities, non-linear models such as DT and RF can innately navigate the complex, multifaceted layers of geological data. The comprehensive evaluation extends to propose a novel set of performance metrics designed to capture the intricacies of grade prediction, thereby aligning closely with the operational demands and decision-making processes in the mining industry.Item The improvement of the on-time delivery for “company x” e-commerce orders during the golden quarter(University of the Witwatersrand, Johannesburg, 2024) Sukazi, Thobile Nomalungelo; Sunjka, BernadetteE-commerce has revolutionized global business and consumer interactions, offering convenience and accessibility across various domains like Business-to-Business (B2B), Business-to-Consumer (B2C), and Consumer-to-Consumer (C2C). The COVID-19 pandemic has accelerated digital transformation, with South Africa's e-commerce market showing robust growth projections, fuelled by factors such as improved internet penetration and shifting consumer behaviours. The Omni-channel strategy has become standard, with leading players leveraging digital capabilities to maintain market share. Notably, the "Golden Quarter" of retail, encompassing events like Black Friday and Singles Day, presents a pivotal opportunity for retailers to boost profits through strategic promotional efforts. As the market matures, focus shifts to optimizing the final mile of delivery, aiming to improve efficiency and reduce costs. This project seeks to explore tailored strategies for final mile optimization in Company X, aligning with the broader goal of enhancing efficiency and customer experiences in South Africa's growing e-commerce sector. Despite being the second-largest wholesale food distributor in South Africa, Company X experienced significant on-time delivery performance declines, particularly in its discount retailer brand, Banner 3. The analysis identified logistical bottlenecks in the final mile as the primary contributor to these challenges, resulting in an average delay of 4.3 days in the order fulfilment process. Additionally, the study highlighted the importance of addressing these challenges to maintain customer satisfaction, loyalty, and competitiveness in the rapidly evolving South African e-commerce landscape. This study employs a comprehensive framework and systematic approach to investigate the research questions and objectives. A qualitative research design involves one-on-one interviews conducted digitally via Microsoft Teams. Ethics clearance (MIAEC 099/23) was obtained, ensuring transparency and participant understanding. The sampling strategy prioritizes quality over quantity, with six diverse participants selected to provide rich qualitative data. Data analysis follows Braun and Clarke's thematic analysis approach, incorporating triangulation methods and emphasizing thorough documentation to ensure validity and reliability. This research has thoroughly investigated Company X's final mile delivery challenges during the Golden Quarter, providing comprehensive insights and recommendations for enhancement. Key findings underscore the significance of accurate forecasts, planning collaboration, proximity to customers, fleet and technology utilization, customer service levels, and delivery types in optimizing delivery performance. Recommendations encompass advanced forecasting models, collaborative planning efforts, tailored customer promises, technological enhancements, and automation to address identified challenges and capitalize on opportunities for improvement. The proposed strategies offer a strategic roadmap for Company X to enhance efficiency, customer satisfaction, and competitiveness in the e-commerce landscape, aligning with the study's objectives and concluding the project successfully. The tailored recommendations contribute valuable strategies for improving efficiency, customer satisfaction, and competitiveness. Future research could focus on evaluating the implementation of these strategies and exploring emerging technologies to further optimize the delivery process and adapt to evolving market dynamics.Item The Effects of Rectilinear Acceleration and Deceleration on Shock Formation near a Stationary Boundary(University of the Witwatersrand, Johannesburg, 2024) Morrow, Sean RobertInspired by the world land speed record vehicles the Thrust and Bloodhound supersonic cars (SSC), the focus of this dissertation is to investigate how rapid acceleration affects the formation of shock waves coming off an object travelling in ground effect. Due to the proximity of the ground, these shock waves are not able to freely propagate under the object and must interact with, and reflect off, the ground. Steady state and transient models of aerofoils, accelerating from Mach 0.05 to Mach 2.00 at a test run acceleration of 3 g and an extreme acceleration of 176 g are developed and compared to reveal that the transient shock wave development trails that of the constant velocity aerofoil. The main reason for this difference is that the transient flow is unable to fully develop and reach a state of equilibrium. The extreme acceleration allowed even less time for the flow to develop, and the difference in the shock location continuously increased throughout the acceleration. The same difference in shock location was evident when these models were decelerated back down to Mach 0.05. However, the extreme deceleration and increasing difference in shock location drastically changed the transonic and subsonic flow field, especially as flow features and shock waves from the higher velocity flow overtook the model. In each acceleration and deceleration case, the transient flow history effects subsided and the aerodynamic performance from the transient analysis converged with the aerodynamic performance from the steady state analysis. Under acceleration the transient performance converged at a higher steady state Mach number, while under deceleration the transient performance converged at a lower steady state Mach number. As the magnitude of the acceleration and deceleration increased the Mach number at which the results converged shifted to higher and lower Mach numbers respectively. Models with different orientations and ground clearances were also compared against each other and a case at free flight to determine the impact ground effect has on the formations and locations of the shock waves. Increasing ground effect was shown to promote the formation of shock waves under the inverted aerofoil and in general delay the propagation of the bow shock between the model and the ground. Once the bow shock propagations passed underneath the models, the resulting flow field converged with free flight conditions and ground effect no longer had an impact on the supersonic aerodynamic performance of these models. Under some conditions, the combination of ground effect and the transient effects of acceleration or deceleration can cause dangerous lift and pitch conditions.Item Understanding the challenges of implementing an effective Requirements Analysis process within an engineering R&D environment(University of the Witwatersrand, Johannesburg, 2024) Lydall, Peter Wykeham; Law, CraigRequirements Analysis is widely regarded within the Systems Engineering community as an activity that has a significant impact on project outcomes. However, it is an activity that is often overlooked or poorly executed. This report details the application of Yin (2003)’s Case Study Method to a single case, involving an engineering research and development group at a South African science council. The case study attempted to gain insights into the perceptions and attitudes of engineers and managers towards Requirements Analysis, that might explain why it is performed inconsistently or less effectively than it could be. Key findings include: that there is a poor understanding of what Requirements Analysis is; the importance of assigning a Requirements Analyst, in a dedicated role, with the appropriate level of engineering experience and Systems Engineering training, and a desire to perform the activity; the necessity of having a cost effective and tailored process which evolves over time.Item Metallic Equivalent of Aircraft Landing Gear Using Composite Materials(University of the Witwatersrand, Johannesburg, 2024) Kotze, Marius Hugo; Boer, MichaelThere are two types of Light Sport Aircraft landing gear configuration. The taildragger and tricycle arrangement where the difference is specified by the position of the main landing gear. Shipment delay of the current Aluminium 7075 T6 landing gear has caused further delays in the manufacturing of the BushCat Light Sport Aircraft. Thus, a composite alternative was required which could be manufactured locally. The objective was to determine which locally available material was best suited as an alternative to the current Aluminium 7075 T6 design. This included estimation of the correct design loads acting on the BushCat aircraft main landing gear and to specify a composite alternative that could withstand these calculated design application loads. The loads that were used would be obtained from the ASTM F2245-14 regulations and EASA CS-23 amendments. The loads were validated by means of Finite Element Analysis and analytical calculations. Drop tests were also conducted by the company and image processing was used to compare the calculated deformations to the FEA results. This was used to validate the load and constraint applications in Ansys 2023 R2 software. The composite materials used for analysis were unidirectional epoxy e-glass wet layup and prepregs fibres. A coupon study was conducted on Aluminium 7075 T6 alloy and [0/90/90/0], [0/45/45/0], [0/90/45/0] layered unidirectional epoxy e-glass wet layup and prepreg coupons loaded under tension, compression, bending and torsion. The FEA results were validated using analytical calculations obtained from the Classical Lamination Theory. It was concluded that the unidirectional epoxy e-glass prepreg coupons were best suited as an alternative as better results in withstanding the applied load applications were obtained. The prepreg fibres also contained a lower void content in comparison to the wet layup fibres, thus increasing the fatigue life of the composite laminate as well as reducing the moisture absorption. The final composite landing gear was analysed using the Puck-failure criterion and it was found that after analysis and modifications were conducted, the newly designed composite landing gear could withstand the applied loads during limit load and ultimate load conditions without any fibre or inter-fibre failure in the strut of the landing gear. It was found that, failure had occurred in one of the fibre plies near the bolted regions of the axle section during ultimate (emergency) landing conditions and was thus concluded that the composite landing gear should still be inspected when attempting emergency landing at higher load conditions at an aircraft maximum take-off weight of 600 kg. The final composite landing gear design after modifications was 4.613 kg heavier than the Aluminium 7075 T6 landing gear. With regards to manufacturing the final composite landing gear a vacuum bagging process should be followed where the final vacuum bagging assembly containing the composite layup of the landing gear should be placed inside an oven or autoclave to start the curing process. Once the composite landing gear is cured, it could be machined into its final shape were non-destructive techniques such as ultrasound of thermography should be used to inspect the final composite landing gear for any air of volatile compounds withing the laminate. Static and dynamic destructive testing should also be used to validate if the final composite landing gear can withstand all landing conditions aircraft maximum weight without any fibre failure or delamination occurring.Item Effects of Ni-Mo binder and laser surface engineering of NbC based cutting inserts during face-milling of automotive grey cast iron(University of the Witwatersrand, Johannesburg, 2024) Rabothata, Mahlatse Solomon; Genga, Rodney; Polese, ClaudiaThe main aim of this study was to design, develop and produce NbC-Ni cermet based cutting inserts as potential substitutes for conventional WC-Co based inserts for the face-milling machining of automotive grey cast iron (a-GCI), an alloy that plays a critical role in the automotive manufacturing industry. For this purpose, rapid pulsed electric current sintering (PECS), additions of Mo as a partial binder replacement and TiC and TiC 7 N3 as secondary hard phases, and femtosecond laser surface modification (LSM) technique were used in an effort to enhance the NbC-Ni based cutting inserts’ machining capabilities during face-milling of a- GCI. All the sintered samples achieved relative densities of 97% and above, irrespective of the sintering process. Adding Mo, TiC and TiC 7 N3 to the NbC-12Ni (wt%) composition refined the NbC grain size in PECS samples, enhancing hardness and wear resistance. Mechanical impact shock and wear resistance of inserts were further improved via femtosecond LSM, creating pyramid (P) LSM and shark skin (S) LSM based micro-patterns on the surface of the cutting inserts. Face milling machining tests of a-GCI were performed at 200-400 m/min cutting speed (𝑉𝑉𝑐𝑐) and 0.25-1.0 mm depth of cut (ap ). The inserts’ cutting-edge wear and failure were evaluated after every pass using optical microscopy and analysed via high angular annular dark field (HAADF)-scanning electron transmission microscopy (STEM). Machining performance was assessed by technique for order of preference by similarity to ideal solution (TOPSIS) based model using insert tool life (𝐼𝐼𝑙𝑙), specific cutting energy (𝑈𝑈𝑐𝑐) and maximum resultant cutting forces (Fmax ) as criteria and including surface roughness (Ra) during finishing operations. The pyramid LSM PECS NbC-10TiC-12N[Ni/Mo] (wt%) (R2MS-P1) insert was the top performer during semi-finishing, with 20 min 𝐼𝐼𝑙𝑙 , 22 J/mm 3 𝑈𝑈𝑐𝑐 and 1087 N Fmax , obtaining an overall preference score (𝑂𝑂𝑖𝑖) of 0.953. The best inserts during finishing 2 were the blank (B) (i.e. unmodified cutting edge) PECS NbC-10TiC 7 N3 -12Ni (wt.%) (TCN1S-B3) and LPS WC-10TiC-10[Co/Mo] (wt.%) (T1ML-B3) inserts with both inserts obtaining 𝑂𝑂𝑖𝑖s of 0.826, respectively. In general, additions of Mo, TiC, TiC7 N3 , PECS and LSM improved hardness and abrasion wear resistance, resulting in enhanced performance of NbC-Ni based cutting inserts during machining.Item Fatigue Crack Propagation in AlSi10Mg Additive Manufactured Aeronautical Parts Processed by Laser Shock Peening(University of the Witwatersrand, Johannesburg, 2024) Chinyama, Joel; Polese, ClaudiaAdditive manufacturing (AM) offers advantages for complex aeronautical parts, but inherent defects can reduce fatigue life. Post-processing techniques such as laser shock peening (LSP) can be used to introduce beneficial compressive residual stress that hinders crack propagation. This study investigates LSP as a method to improve fatigue performance in additively manufactured AlSi10Mg aeronautical parts. It examined how varying LSP treatment laser power intensity (1.5 – 4.5 GW/cm2) affects LSP's effectiveness and identified optimal LSP residual stress profiles for peak fatigue performance. The residual stress profiles that were used in this research were adopted from previous experimental work done on AM-manufactured AlSi10Mg alloys within the Wits AM/LSP group. Previous experimental work results on wrought AA2024-T351 (untreated and LSP-treated cases) indicated that LSP-treated samples have a fatigue life of at least four times longer as compared to as-built samples. AFGROW models with similar geometry, material properties and load conditions were used to predict the fatigue life of as-built and LSP-treated cases. An improvement in fatigue life of at least 3.8 times was observed, which was within an acceptable deviation from the experimental results. These results were used to validate AFGROW models for exploring different specimens. Fracture mechanics models (AFGROW) were used to compare the fatigue life of as-built and LSP-treated AlSi10Mg samples with different LSP power intensity parameters. The results showed that LSP treatment can significantly extend fatigue life, with the optimal laser power intensity found to be 3.0 GW/cm2. This improvement is attributed to the introduction of compressive residual stresses by LSP, which suppress crack initiation and propagation. The effectiveness of LSP was further explored in the context of the Cessna 172/175 horizontal stabilizer, a part that could benefit from AM for weight reduction and structural integrity. AFGROW models were developed to predict the fatigue life of the centre lightening hole in the forward spar, a critical location for crack initiation. The models incorporated a beta correction factor to account for the specific crack geometry and stress distribution. The beta correction factor was determined by comparing the stress intensity factors from the Finite Element Analysis (FEA) and AFGROW models. The results again demonstrated a significant increase in fatigue life (of at least six times) for LSP-treated parts compared to as-built parts. AFGROW models with a beta correction factor proved valuable for predicting fatigue life in components with complex geometries. This study confirms that LSP is an effective post-processing technique for mitigating fatigue crack propagation in AlSi10Mg AM aeronautical parts.Item Prediction of Water Hyacinth Coverage on Hartbeespoort Dam(University of the Witwatersrand, Johannesburg, 2024) de Gouveia, Claudia D. Camacho; Bührmann, Doctor JokeWater hyacinth is an invasive weed contributing to Hartbeespoort Dam’s poor water quality. Although biological control is the most effective and sustainable method of controlling water hyacinth, the dam has unfavourable conditions for agents that the weed thrives in. Literature uses mathematical models and remote sensing to theorise growth rates or estimate coverage. However, prediction could prove beneficial as planning biological control is essential to its success. Hence, a model to predict water hyacinth coverage was developed. This research simplified the complex relationships involved in water hyacinth growth to focus on the most influential factors: temperature and nutrients. Missing data were imputed using multiple k-nearest neighbours. Nutrient datasets had limited data, thus five scenarios were developed to extrapolate datasets, using Monte Carlo simulation and seasonal patterns. The features were used to build ensemble, decision tree, artificial neural network and support vector machine models. Ensemble using the bagging method was the best model resulting in a root mean square error of 4.01 for water hyacinth coverage predictions from 1 June 2018 to 1 May 2019.Item e-Commerce adoption in South African SMMEs(University of the Witwatersrand, Johannesburg, 2024) Mathe, Barney; Sunjka, BernadetteIn this research study e-Commerce adoption in Small, Medium and Micro enterprises (SMMEs) in South Africa was studied. The purpose of this study was to determine the factors that constitute for a successful implementation of the e-Commerce strategy and compare this to the adoption of e- Commerce in SMMEs. The study evaluated frameworks/standards for successful e-Commerce implementation and with this evaluation a consolidated conceptual framework for the successful implementation of e-Commerce was developed. Through samples of both SMME and successful/most visited platforms, functionality and performance assessments of the platforms were conducted using the Jaccard similarity method, and time and method study. T-tests and ANOVA were conducted for performance assessments, to assess if there was a significant difference in performance between SMME and successful e-Commerce sites. Through these assessments it was found that there is a clear gap in adoption from both a functionality and performance perspective, 6 out of 7 of the successful sites had advanced to the last stage of adoption while only one SMME platform reached the last stage which is an external integration stage of the developed framework.Item Inventory Management using Artificial Intelligence(University of the Witwatersrand, Johannesburg, 2024) Garg, Arnav; Smith, Bevan; Rich, WilliamPoor inventory management negatively affects a company’s profits. Too little stock limits potential sales and customer satisfaction while too much stock increases storage costs and potential damage. Company X distributes butter in a multi-echelon supply chain consisting of multiple entities such as a manufacturing plant, distribution centre and retailers before reaching the customer. For every 1% of the demand that is not met, revenue in excess of R18 million is lost per year. This study aims to use machine learning methods (supervised and reinforcement learning) for optimal decision making that maximizes profits. Supervised learning methods (random forest, recurrent neural network and support vector machines) were used to forecast the demand based on historical data. Thereafter, deep reinforcement learning (DRL) was used to train an agent to decide when and how much to order over a period of a year. Various algorithms (PPO and DDPG) and unique reward functions were tested and the performance was compared to a benchmark heuristic that stocks inventory based on a sum of the forecasted demand. The random forest algorithm performed the best at predicting the forecasted demand. The DRL model using a continuous action and state space together with the DDPG algorithm and a reward function based on a combination of the current profit, order fulfilment rate, units available and units unsatisfied performed the best. The DDPG algorithm outperformed the PPO algorithm with the DDPG model being able to provide a 21% increase in net profit over the benchmark heuristic when the production and warehouse facilities of the supply chain were merged. The DRL models were not able to provide a higher order fulfilment rate compared to the benchmark heuristic but they were able to provide better asset utilization by sending full trucks and minimizing the inventory held to maximize the profitability in the supply chain. The results suggest that RL has potential of better handling the stochastic constraints (demand and lead times) in real supply chains to automate the ordering process. It was found that increasing the order fulfilment rate does not necessarily lead to higher profits and the reward function has a significant effect on the net profit which can be further optimized in the future.