Mini-dissertations (Industrial and Systems Engineering)

Permanent URI for this collectionhttp://hdl.handle.net/2263/10571

This collection contains research documents that presents the author's research and findings and is submitted in support of candidature for a final year undergraduate programme, an Honours degree, or as part of a Masters degree that involves course-work at the University of Pretoria, within the department indicated.

From the Letter of Consent:
"As per the policy of the University of Pretoria The Department of Industrial and Systems Engineering is authorised to grant permission for the archiving of the final year project mini-dissertations proceedings."

Browse

Now showing 1 - 20 of 438

Sugo Bryanpark
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Van Den Berg, Emile Garrett
Work scheduling in the restaurant industry is an often neglected task due to the abnormal complexities of having to allocate multi-skilled staff over multiple overlapping shifts to meet dynamic demand. Client satisfaction is a significant aspect of the industry, in which the time to produce is almost as important as the quality of the food. Thus it is vital for a restaurant to be able to anticipate and meet demand. It is found that the use of quantitative (formula driven) methods does not ensure higher accuracy in both the forecasting and work-scheduling, but rather when used in combination with qualitative (experience input) is the accuracy majorly increased. The forecast model operates on data gathered from the point of sale system, and is then reworked into 100 di↵erent time intervals over the course of each day in a week. The output is then validated and adjusted before it is imported into the work-schedule function. The work-schedule function is able to output the ideal work schedule per staff member while also illustrating the shortages and surplus staff per time interval. This then allows the manager to ensure that service delivery is met as efficiently as possible.
Moving forward
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Taute, Jana
This document is the final project report for a final year project in the Department of Industrial and Systems Engineering, module code BPJ 420. Company ABC is a company that started out as a family-owned business in 1993. Despite an annual growth of 21% for the last three years, a recent survey showed deterioration in customer service. As customer satisfaction is vital to sustainable growth of the company, the underlining problems to poor customer service are identified and quantified in the project motivation and problem analysis. Business process management (BPM) is identified as an appropriate theoretical area in industrial engineering to deal with the process inefficiencies causing the decline in customer service. Information system design (ISD), operations research (OR) and the quick changeover philosophy single minute exchange of die (SMED) are identified to semi-automate and optimise some of the processes within the company to enhance customer service. A literature study is done on the above mentioned theoretical contexts and their corresponding project approaches. Problem analysis is done using the drivers and triggers for a BPM related project identified in the literature study, as well as the PIECES analysis used to identify information system (IS) related projects. The requirements elicitation is done, where the appropriate strategy with its process architecture is developed with initial BPMN process models and the basic enterprise construction. Corporate strategy is developed using strategic analysis tools, like PESTEL analysis, Porter’s five forces and value disciplines. The balanced scorecard approach is used to identify appropriate measures for the chosen strategy. The requirements serve as a foundation for the solution selection where a cost-benefit analysis indicates that implementing the new business processes, especially those involving the IS and SMED, could potentially increase the overall quote acceptance rate by 20% and reduce costs by R 411 550.00 annually. BPMN and SMED are used to construct the new process models in the solution construction. These process models are developed with corresponding training manuals for the expanded IS, role descriptions with specific KPIs and customer communication scripts. The implementation approach is discussed with some findings as conclusion. Since the evaluation results were obtained only a month after implementation, some of the expected long-term benefits have not been fully realised. The company did however experience an immediate 28.9% growth in revenue from the corresponding month of the previous year with further benefits discussed in the conclusion and recommendations.
Airport route development decision support tool for Cape Town Air Access
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Robertson, Kyle M.
The connectedness and accessibility of Cape Town International Airport (CTIA) is critical for the economic growth and development of the Western Cape. Cape Town Air Access (CTAA) is responsible for bolstering the long-term growth and sustainability of Cape Town's air access by making decisions on which routes to develop and which existing routes to expand. These decisions are multi-faceted and have wide-ranging impacts on the Air Transport Network (ATN). This study aims to assist the decision-makers at CTAA by creating a model with a network view of their route development decisions. Complex Network Theory (CNT) is used to create the model, which has been applied in many elds to study both the topology and dynamic nature of complex networks. In this study, it is used to model CTIA's ight network and in turn measure its accessibility under various scenarios. Three scenarios are tested: Increasing the capacity on key ights, decreasing the capacity on key ights and introducing new ights to the network. The key ights are identi ed by calculating the vulnerability of airports in the network, which measures the impact that removing an airport has on the accessibility of CTIA. The new ights added to the network are the busiest international airports in each of the 5 strategic countries identi ed by CTAA (Angola, Canada, Ethiopia, Kenya and the United States of America (USA)). Two heuristics are employed in the model to account for the number of seats stolen as a result of an increase in capacity on a direct ight. The Proportionality heuristic `steals' seats from connecting ights with the same proportion as passengers currently using each route to travel to CTIA. The Switching behaviour heuristic uses ight-cost and travel-time data between regions to predict the number of passengers to switch to the direct ight as opposed to using the connecting ight. Three metrics (betweenness centrality, weighted clustering coe cient and e ciency) are used to calculate the accessibility of CTIA when new ights are added to the network, or the capacity on existing ights is increased or decreased. These metrics translate the word de nition of accessibility ( exibility, time-e ciency and cost-e ciency of travel) into a measurable quantity. The research question is: \Can a CNT approach assist decision makers in evaluating the impact of route changes on the Western Cape's air access?". Positive feedback from the company was received regarding the project's usefulness. The project manager described it as \Novel and worthwhile for what we are doing at CTAA", and \The methodology and the way it is structured makes a lot of sense". He also gave constructive criticism regarding the results of the analysis, which need to be more suitable for audiences such as politicians and the CTAA steering committee. Based on the results of the simulations, Schiphol Airport in Amsterdam and Franz Josef Strauss Airport in Munich should be expanded. This challenges the current thinking at CTAA which focuses on developing and expanding smaller routes. In terms of developing new direct ights, the results suggest that Miami International, Edmonton International and John F. Kennedy International Airport (JFK) should be developed.
Forecasting transportation service demand for fleet optimization: the case of a third party logistics provider
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Van Aarde, Margo
Outsourcing is becoming increasingly popular in today’s age where customers have higher expectations from suppliers than ever before. Businesses are collaborating to aim to meet customer demand and stay competitive in the market. Some businesses turn to outsourcing. This industry is rapidly growing and faces a unique set of challenges. Customers value flexibility and punctuality in this service provided. In order to do this, logistics providers have to account for the stochastic nature of customer demand and thereby incur additional costs. The focus of the project is on a transportation based logistics provider that provides transportation-, cross docking- and distribution services to a collection of retail clients. The company is a leveraged logistics provider, which means that it uses its own vehicles as well as vehicles hired from external companies to provide the service. This project aims to investigate ways to reduce the operating costs of a division in a logistics provider company specifically by suggesting the optimal fixed fleet size the company should maintain. This was done by forecasting the anticipated number of vehicles that will be required to meet customers’ service demand for a period of twelve months and finding the optimal fleet size of the 3PL’s fixed fleet. The primary challenge for the 3PL is to maintain their flexibility in the service they provide to clients while introducing stability in their own operations. The report gives a detailed plan that was followed to achieve this goal and outlines the tools and techniques that was applied to obtain the results. Furthermore, a literature study specifically aimed at forecasting and optimization models was done to broaden knowledge and understanding of these subjects. The results suggests that the 3PL should maintain a fixed fleet size of 129 vehicles in order to meet the following twelve months of anticipated customer service demand. This was calculated as the optimal fixed fleet size that will ensure their customer service level is not affected and lead to a potential saving of almost R 40 000. This analysis will enable the 3PL to make informed decisions regarding the available options they have to reduce their operating expenses.
The Optimization of the Woodlam Furniture Factory with Specific Reference to It’s Electricity Consumption
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Wagener, R.
Woodlam is the leading plywood furniture manufacturer in South Africa, experiencing steady growth in the last few years. Due to the increasing cost of electricity in South Africa it has become important for companies within the manufacturing industry to find methods to optimize their electricity consumption. This enables these manufacturers to reduce operating costs. Globalization and market pressures also make it necessary to find a means to reduce the cost of production in order to increase profit and remain competitive both locally and internationally. Addressing this problem correctly results in a more profitable business model while allowing production capacity to meet the customer demand. This report includes a literature review to understand key concepts, electricity data collection and analysis, solution research and design of the approach to solving the problem as well as the evaluation and validation of the proposed improvements. This report outlines the importance of finding an effective solution to rising electricity manufacturing costs and depicts the methodology and steps necessary to obtain the desired optimal solution. An indepth study was conducted to provide insight into the manufacturing process. Electricity tariffs and solution approaches were also examined from which an operations research mixed integer linear programming model (MILP) was selected. This model was then formulated to achieve the desired financial benefits that can be achieved through the intelligent scheduling of operations. The monetary and environmental results of the (MILP) model depict that by scheduling the production of the machinery has the ability to assist the production manager to reduce the electricity cost by avoiding peak electricity tariff time periods. The validation process to determine the success of the project is also outlined.
Resource scheduling algorithm for maintenance planning
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Young, Kirsten
“Company XYZ” is a company which outsources maintenance to various enterprises all over South Africa. Technicians are hired to travel to their customers which are geographically far from one another to perform maintenance on electrical devices such as servers, computers and air conditioners. An employee’s workday consists of both their travel time and working time and so routing must be carefully considered in order to reduce travel costs. Company XYZ’s employees find that their workloads are unbalanced i.e. some days they will work much longer hours than others. This has led to Company XYZ requiring a way to efficiently schedule their employees so that customers demand can be met, while keeping costs low, resource utilization high and workloads balanced. Fourier-E attempted solving Company XYZ’s problem by creating a linear programming resource allocation model. The model worked but there is still much room for improvement. All the data was therefore already available in a device database which could be used in the development of a new solution. After performing a literature study it was found that the problem at hand has many similar aspects to that of a Multiple Travelling Salesman Problem and so the many methods of solving this kind of problem were researched. The genetic algorithm was selected as the most suitable algorithm for solving the problem because of its short running time and the student’s ability to code it. Specific selection, crossover and mutation techniques were used to evolve the initial population of solutions. With every new generation, a better schedule was found. The best solution of the final generation was selected as the schedule to analyse. The genetic algorithm exhibited many advantages over using the existing linear programming method. The chosen schedule significantly reduced overtime, reduced travel distances and balanced resource workloads. It is up to the company to decide whether they should implement it or not. Company XYZ should validate the final schedule by using a testing team to ensure that the assumptions on which the model was based are acceptable.
Minimise loom changeover time at MONN
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Nouwens, Marcus
MONN is a manufacturer of premium quality and custom woven Axminster carpeting. The process of making a custom woven Axminster carpet involves many processes but the weaving process o ered room for improvement in the throughput and changeover time of the 5 looms, which weave the carpets. It was identi ed that the looms created a bottleneck in the factory making any improvement in loom changeover time valuable. The processes and methods in place when setting up a loom for a new job were uner- gonomic and slow. Using multiple literature sources that looked into facilities planning and the ergonomics of preparing the looms, new methods improving ergonomics and speed were devel- oped. Two new scenarios were developed which could lead to possible improvements over the current system. In order to assign a monetary value to the di erent scenarios, AnyLogic was used as a simulation tool to model the di erent scenarios. AnyLogic is a simulation program that by utilising the concept of discrete and agent based modelling provides a platform to simulate a process with accuracy. The current processes and systems MONN uses was simulated providing a foundation for the new scenarios for and benchmark to which they can be compared to. To ensure that the base model was realistic; validation and calibration experiments were conducted. Once the based model was completed, the new scenarios could be developed using simulation. Two scenarios were developed where changes in the material handling systems and improve- ments in general ergonomics were made. An additional scenario was added where sta were relocated to new positions to further improve performance. The results of the scenarios' simulation models were documented, and the monetary values of the simulations were assessed to determine the feasibility. It was found that all proposed sce- narios were an improvement over the current system and would that they would pay themselves in 5{8 years contributing R500 000 { R1 350 000 a year to pay themselves o , depending on the scenario.
A facility solution required to incorporate cold sterilisation in the supply chain of citrus exported to Europe
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Le Roux, Hester
This document describes a Final Year Project (BPJ 420) required by the University of Pretoria in partial fulfilment of Systems and Industrial Engineering. This report describes the solution to an identified problem encountered by the South African Citrus Industry with the assistance of Citrus Growers Association (CGA). This project entailed a thorough study of the citrus supply chain, facility and capacity planning of cold storages used for citrus exports through the Durban Harbour. The South African Citrus Industry is currently facing a phytosanitary risk of False Codling Moth (FCM) that might prevent the industry gaining further global market access. FCM is only detected in Africa and stricter bio-security protocol measures imposed by the European Union (EU) can prevent further exports to their markets. South Africa exports almost fifty million citrus cartons to the EU, which approximate 40% of all citrus fruit per year at an estimated amount of R15.7 billion in 2015/2016 based on exporting prices (DAF, 2016). Current prevention measures are in place that includes: a System Approach (Moore and Hattingh, 2013) and Nuclear-based Sterile Insect Technique (SIT) (IAEA, 2017), although research has shown that cold sterilisation is the best FCM prevention measure. The aim is to identify facility solution required that would ensure that South Africa meets the regulations imposed by the European Union (EU) to prevent FCM from spreading to their countries. Northern South African farms, packhouses, cold storages in Durban and Transnet Durban Port Terminal were visited to obtain information. The data acquired is used to adhere to cold sterilisation protocols when identifying current cold room facilities. The cold storages are divided into two types either a pre-cooling or forced air-cooling (cold sterilisation treatment) chamber. The capacity is determined with three conditions: Prototype model 1: Current citrus production 2017 Prototype model 2: Forecasted demand for 2020 Prototype model 3: Forecasted demand for 2020 including all fruit volumes exported to the EU Industrial Engineering (IE) techniques are used to approach, analyse and develop a possible solution. A Process Flow Diagram is drawn to illustrate key role players in the citrus supply chain. A Method Study portrays the detail of the citrus processes compiled in the Literature Review. The focus of this project would entail a thorough Facility Planning analysis of the cold storage operations in Durban. This study determined the current 38 300 pre-cooling and 13 350 forced-air cooling capacity. The CGA provided 2020 forecasts of citrus export through Durban Harbour. The forecasts are based on tree census data, bud-wood sales, land availability, climatic conditions and historic production since 2012. It is calculated for Prototype Model 2 that in 2020 the demand will be 45 543 precooling and 15 404 forced-air cooling pallet slots. The additional capacity required is 1085 pre-cooling and 1416 forced air-cooling facilities. This is a moderate growth expected annually. Prototype Model 3 is calculated if the European Union imposes stricter protective measures with 45% extra volumes to undergo cold sterilisation treatment in 2020. It is calculated that the demand to cold sterilise the extra volumes will be 43 602 forced aircooling pallet slots and pre-cooling facilities will be in excess of 21 710 pallet slots. However these facilities can be converted into forced-air cooling facilities to accommodate space restrictions in the CBD. The required 29 614 forced air-cooling pallet slots can either be transformed in current cold storages or new cold storages should be built in Elangeni, Kwa-Zulu Natal an Industrial Zoned area. The cost involved to facilitate this forced-air cooling chambers to either convert current facilities or develop new would amount R531 129 915.00. A Trade-off Analysis is used to compare the current and additional cost per 15 kg carton to undergo cold sterilisation treatment. An increase per carton of R18.28 would accumulate to a total of R519 116 500.00 additional annually for all citrus fruit exported to the EU. In conclusion of this Final Year Project is the result subjective to condition. The current forced air-cooling facility in Durban will not accommodate cold sterilisation treatment for extra volumes in 2020 if the European Union enforces stricter protocols. The increased overheads involved is not a viable solution and would be in the best interest of the South African Citrus Industry to prove different measures with research done to prevent False Codling Moth without imposing cold sterilisation treatment.
Using a discrete-event simulation model to identify bottlenecks and determine the efficiency in a maintenance workshop
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Venter, Francois Marthinus Fredrick
Developing a discrete event simulation model based on key aspects in the maintenance workshop in order to identify bottlenecks and determine the efficiency of maintenance practices.
Developing an industry solution to service bottom-end markets in South Africa
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Prinsloo, Riandre
Logistics and distribution services have been around since the early 1950's. During this time it was only thought of as military services, but since then it has developed into a system that can give any company a competitive edge, if the network is designed e ciently. Due to the increasing competition in the logistics and distribution industry, it has become increasingly important for companies to gain a competitive advantage over their competitors. This entails exploring untapped markets, such as the BOP (bottom of the pyramid) market segment. The aim of this project is to determine whether there is a market opportunity in South Africa to service this segment. The market evaluation can be broadly classi ed into the market size, structure, and preferences of the consumers constituting this market segment. The problem investigation phase of this project evaluated the market opportunity, which concluded that the BOP market consists of four billion people, which constitutes 72% of the global population. In South Africa the BOP market consists of approximately 30 million people, which is 47% of the population. This validates that there is a great market opportunity for food producers and distributors to enter the market. Servicing and distributing to the BOP market comes with added complexity, thus traditional supply chain thinking will not su ce. The complexity in this market are due to a few contributing factors, such as the cash poor consumers constituting the BOP market, the lack of infrastructure, longer time frames and limited product awareness. Numerous companies have tried to improve their pro ts and overcome the challenges in distributing to the BOP market by bridging the infrastructure gap through selecting the appropriate technologies and controlling costs through di erential or layered distribution networks. The BOP market has a relatively high success rate in terms of market penetration (5% to 10%) but operation in this market proves to be di cult. A literature review based on companies who have success- fully entered the BOP market segment proved that it is possible to operate successfully in this market segment. The key success factors of these distribution companies were noted and the distribution models implemented were further researched. The literature case studies presented in this report validate that traditional distribution channels will not be su cient to service the BOP market in African countries. Al- ternative distribution channels need to be developed which are innovative and based on inclusive business models. An evaluation framework was developed based on key criteria in order to evaluate the several distribution models. The distribution models were further evaluated based on technical and nancial aspects. Based on the results of the evaluation framework, technical and nancial evaluation, the applicable distribution method was modelled in order to make an e ective recommendation. The dataset was reduced in order to create a proof of concept for the model, as the probability that this can be replicated on a larger set of data is very likely. The ideal location for the hub was determined by using the centre of gravity method with the respective inputs as discussed in the literature review. The dataset was further reduced to form clusters in order to conduct spoke location analysis. The ideal spoke locations were determined by using the centre of gravity method, as for the hub location, thus creating prede ned locations for the spokes to be situated at. The optimal number of spoke locations was determined by creating an optimization model, through the use of the Operations Research approach, to minimize the number of spoke locations, which was formulated using Lingo. Through the use of the optimization model, an entire distribution network was developed based on the test case data. The developed model was validated by conducting a sensitivity analysis, which proved that the model behaves as expected.
The development of a decision support model for the picking, staging, combining and loading activities in the grocery distribution warehouse of a large fast moving consumer goods retailer
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Van Niekerk, W.W.
The Pick 'n Pay grocery warehouse situated at the Longmeadow distribution centre (DC) has 10 790 stock keeping units that are received from 123 vendors and are distributed to 421 stores located in several provinces as well as three neighbouring countries. The stock keeping units which are referred to as delivery units (DUs) are picked into handling units (HUs) (rolltainers used to transport DUs) and staged in staging lanes. Loading sta load the staged HUs onto transport units (TUs). The TUs then deliver the stock to various Pick 'n Pay stores. This process is referred to as the outbound process. The two concerns currently experienced by the outbound process is an excessive num- ber of dropped HUs (HUs that are not loaded onto their planned TUs) and excessive TU loading times. It was determined that a decision support model was required to provide insight on how to reduce dropped HUs and TU loading times. A Literature review was performed to investigate agent based simulation (ABS), system dynamics (SD) and discrete event simulation (DES) as possible simulation techniques. Tako and Robinson (2012) deter- mined that DES was most frequently used within the supply chain environment to assist in making tactical and operational decisions. It was determined that a DES model would be used in this project. AnyLogic was selected as the simulation software as this software provides the functionality required for a DES model. The simulation model was developed by simulating the outbound process's picking/staging, combining and loading events. Data captured between the 1st of May 2017 and the 30th of June 2017 was used to develop the simulation model. Key measures were identi ed against which the simulation model was validated. These measures included the number of DUs per HU after loading, loading time per TU, number of HUs staged, combined, dropped and loaded per shift and the number of TUs loaded per shift. After compar- ing the data generated for each measure from 100 simulation runs to the observed data with distribution plots and 99% con dence intervals it was concluded that the developed simulation model was a valid representation of the outbound process. It was determined that high levels of congestion in the staging lanes contribute to the excessive number of dropped HUs and TU loading times. Two scenarios were identi ed which could reduce staging lane congestion namely the increase in the number of stores with night-time receiving and the distribution of weekly volumes. The two identi ed scenarios were evaluated with the developed simulation model using data captured between the 1st of August 2017 and the 31st of August 2017. The e ectiveness of the models were determined by evaluating the number of HUs dropped per shift, loading time per TU,the number of HUs per TU, the number of HUs in the staging lanes and the ow of HUs into and out of the staging lanes. It was concluded that both scenarios could reduce the number of HUs dropped per shift. The scenarios did not indicate a signi cant reduction in TU loading times but did produce an increase in the HUs per TU and loading rate per HU. The scenario proposing an increase in the number of stores with night-time receiving was selected as the most suitable solution as this solution was the most e ective in reducing the number of dropped HUs and increasing the loading rate per HU.
Greedy Algorithm and Model for Analysis of a Bus Fleet's Operations
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) McGladdery, Duncan
Bus Company (BUSCO) approached Fourier-E, an industrial engineering company, to develop and optimise the Duty Master for their Sowetan operations. The Duty Master is a scheduling blue print that dictates the operations and management of a company's bus eet. Through the manipulation of positioning routes and depot allocations, a Duty Master should minimise distances travelled and eet size. In doing so, operating costs are reduced. Fourier-E had already developed their own algorithm for BUSCO's Sowetan Duty Master, which had resulted in substantial savings in the number of busses required to complete all revenue routes and total distances travelled. However, the run-time of their model of approximately 20 minutes to generate a solution is a serious limitation and prohibitive for client use. The primary aim of this project was to develop an online user-friendly model to en- able BUSCO to rapidly solve their eet scheduling requirements. Speci cally, the new algorithm had to be capable of generating a solution for the Duty Master in less than ten seconds. To accommodate the reduction in run-time, Fourier-E speci ed that the devel- oped algorithm must generate a solution which is at least 80% as good as their existing one in terms of busses saved and positioning kilometres driven. The problem of developing the Duty Master can be modelled as a Mulit-Depot Vehi- cle Routing Problem with Pickup and Delivery, Time Windows and Intermediate Facili- ties (MDVRPPDTWIF). To solve the problem, a greedy-heuristic was developed, called \Greedy-Bin". Computational tests showed that the algorithm exceeds all Fourier-E's speci cations and performs particularly well in run speed, whch at 1.23 seconds, is 976 times faster than the existing approach. The Greedy Bin algorithm performs at 88% of Fourier-E's with regards to busses saved and 96.7% for kilometres saved. The Greedy-Bin algorithm met all validation criteria. In order for clients to access the algorithm, an online user interface was developed which enables rapid evaluation of various operatonal scenarios. To achieve this, ve variables that can be manipulated by the client were added to the model, namely: bus speed, revenue routes, loading bu er, distance bu er and day and night depot capacities. Model outputs are: the number of busses needed to complete all revenue routes, positioning distance, depot allocations and eet utilisation throughout the day. Manipulation of the Duty Master model also demonstrates its capacity to save on oper- ating expenses by generating alternative solutions for depot allocations. Additionally, the nancial implications of changing bus speeds and manipiulating loading and positioning bu ers are demonstrated. Finally, in an industry where tenders for new revenue routes are highly competitive, the model can be used to assess the potential nancial implications of adding new routes and in doing so, inform the decision of whether or not to tender for those routes.
Facilities planning : case study at Enterprise, a division of Tiger Brands
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Keeve, A.F.S.
A facility planning project was conducted at Enterprise, a meat processing facility in Gauteng. The project involved the facility layout and design of the hamper room. The hamper room is defined as an area where four different hampers are manually packed with a range of 19 different products. (Each hamper consists of a mixture of some of the products.) The day after the hampers have been packed, they are sold at Enterprise’s factory shops. These shops are located in Germiston and Polokwane. The hampers are in high demand and thus each hamper not packed is considered as a loss of sales. An in-depth as-is analysis of the hamper room facility was performed. A literature study was conducted to understand the problems. This document contains recommendations to rectify issues of labour utilization, wastages, high inventory levels and inadequate facility layout. The degree to which the output of the hamper room does not conform to the demand was calculated and converted into cost of lost sales. The problem investigation and the conceptual design revealed alternative layouts that will allow the hamper room to fulfil the required demand while addressing scheduling and safety hazard issues. Employee ownership, order replenishment schedules and alternative layouts were investigated and recommendations were made. The aim of the recommendations was to improve overall efficiency. The final recommended solution was then verified and validated. The most significant improvement from the current facility is alternative layout 3. Implementation of this layout will ensure that the unfilled demand of 13 pallets per day is fulfilled. This fulfilment will decrease the cost of lost sales that amounted to R175 500-00 per day.
Investigating Sales and Production Volumes of Metals using a Bayesian Change Point Detection Approach
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Ingham, Michael
This paper identified change points in the production volumes and sales of various metals in South Africa from the period 2003-2016. The socioeconomic environment in which the mining sector operates is complicated, with countless factors influencing the production volumes and sales of metals. This complexity makes it difficult for mining stakeholders to accurately forecast the production and sales for specific metals. Possible causative events or factors were linked to identified change points in order to create a deeper understanding of the environment in which mining operates. This deeper understanding provides mining stakeholders with information on the events or factors that have the greatest impact on the performance of the various metals. This information allows mining stakeholders to focus forecasting efforts on the identified factors. Combining the focused forecasts with the impact that similar events had in the past helps the mining stakeholders to alter production levels or schedule investments before forecasted events take place, minimising the potential negative impact of said event. In this study, the monthly production volumes and sales of Gold, Platinum Group Metals (PGMs), Iron Ore and Manganese were analysed. The data spanned from 2003- 2016 and was supplied by StatsSA. The data was analysed using the Bayesian Change Point Analysis (BCP) [Barry and Hartigan, 1993], the Dynamic Programming Algorithm (DP)[Bai and Perron, 2003] and the Non- Parametric Multiple Change-Point Analysis (MCP) [Matteson and James, 2014]. The results reveal that production drops were caused predominantly by mining strikes and increases ii in production costs, while the sales were influenced by changes in the exchange rates and rand value of the commodity. Future studies will use sales volumes instead of Actual Rand values in order to identify changes that can be attributed to shifts in the demand of each metal.
Single machine scheduling for powder coating aluminium profiles
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Kruger, Karien
Scheduling is one of the most important issues in the planning and operation of manufacturing systems (Hoitomt, Luh, & Pattipati, 1993). It is becoming crucial for businesses to be able to schedule effectively. This is due to the effective management of time and money which is now more important to maintain a successful business. This document describes the proposed final year project that will be performed at Wispeco Pty (LTD). Wispeco is a company extruding aluminium profiles. The project was executed at the powder coating department of the Alberton plant. The project aims to find an effective scheduling system, considering customer service and the cost of the different sequences that can be scheduled for the aluminium profiles that must be powder coated. The document contains the preliminary concepts which will be used in the final deliverable. This document defines the problems at the powder coating department and motivates the need for this project. A literature study and a basic solution are also enclosed with an analysis done on these different solutions. This project will cause the scheduling of the aluminium profiles which must be powder coated to be more cost-effective and customer-orientated. Providing Wispeco with this scheduling program will be a significant contribution to the powder coating department.
Improving probabilistic record linkage with a single-layer neural network
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Hamersma, Kris A.
Data analysis requires data to be of a high quality. Unfortunately this is not always the case, especially when data is extracted from di erent data sources. In the case where there is no unique identi er to match data records from multiple data sources alternative methods need to be developed to match the records. Record linkage attempts to do this primarily with deterministic and probabilistic approaches. Deterministic models depend on certain corresponding elds from each record pair to be identical matches to match the record pair together. Probabilistic methods use a set of equations called the Fellegi- Sunter formulae to calculate decision-making weights, which is used to score a record pair on how well they match. If the matching score is above a certain threshold, the record pair is considered to be a match. This project investigates whether the development of a learning algorithm that re nes the weights will improve the probabilistic model's matching accuracy. The dataset that was used to train and test the record linkage models was a set of 92650 record pairs, some of which were matches and some of which were non-matches. It was found that a learning algorithm did improve the matching accuracy of the probabilistic model, although it is likely that the increase in the number of input features will improve the matching performance even more.
An investigation into the improvement of the inventory management process, in the automotive OEM company's supply chain
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Greyling, M.T.
The automotive industry tends to have high volumes of accumulation of inventory. Likely reasons for this is that the automotive industry aim to have high customer service levels, which implies that they need to stock service-, crash- and maintenance parts which are prone to become Slow Moving and Obsolete (SLOB) inventory (inventory that did not sell in the last four years). Inventory can be divided into three types, namely: fast moving-, medium movingand slow-moving inventory. Inventory management is vital to minimise the accumulation of all three types of inventory in an automotive supply chain. Previous research has mostly focussed on fast moving inventory, as its revenue generating ability is higher than SLOB inventory, however the pro t impact of resource wastage on SLOB inventory is likely much higher than assumed in the industry and better management of this type of inventory can change bottom line returns to investors. Therefore, the project investigates the improvement of inventory management on all three types of inventory identi ed, however the project makes speci c reference to SLOB inventory as it is particularly relevant in the automotive industry. To obtain the necessary data for the analysis both qualitative data collected from interviews and quantitative data were obtained from the automotive Original Equipment Manufacturer (OEM) company's Rosslyn warehouse's Warehouse Management System (WMS). To answer the research question, numerous inventory management techniques were researched. Two particular inventory management techniques stood out for this speci c problem. Business Process Modelling Notation (BPMN) charts are used to map current inventory management process and a linear programming model is developed to optimise the automotive OEM company's inventory levels across the supply chain. The current inventory management process was mapped to identify "problem areas" within the process that could be solved by developing unique solutions for each "problem area". In the preliminary mapping it seems that more regular scrapping processes are clearly needed to minimise storage costs and improve the supply chain's efficiency. Furthermore, minor changes such as doing activities simultaneously instead of sequentially, can improve the inventory management process's efficiency, by drastically reducing idle time. The aim of the linear programming model is to minimise transportation- and storage costs across the automotive OEM company's supply chain. Therefore the model used various input data (as discussed in the quantitative data analysis chapter) to generate estimated optimal quantity of parts that the warehouse should order. If the automotive acOEM company implemented the suggested warehouse ordering quantities derived from the Linear Programming (LP) model, the automotive OEM company's supply chain can save an estimated R16 472 807 within two years, which is about a third of their current expenditure on medium- and fast moving parts in the supply chain. Furthermore the SLOB inventory, obtained from the sample data provided by the automotive OEM company's WMS costs the automotive supply chain about R 28 000 000 to store these parts for two years. If these parts can be reduced by implementing regular scrapping processes the inventory management process will operate more effectively and save vast amounts of money.
Food Security in South Africa : improving the availability, affordability, and quality of fresh produce in informal communities
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) De Beer, Darren
This project addresses the growing concern of food security in informal communities in South Africa by carrying out a case study on the fresh produce supply chain in Soshanguve South. The focus of the project is to determine the best way to efficiently and effectively supply Modular Innobox’s Street Fresh with fresh produce. The solution that has been recommended is called the Street Fresh Delivery Service (SFDS). The service takes orders from the Street Fresh and other street vendors in Soshanguve South, purchases from the Tshwane Fresh Produce Market in bulk on their behalf, and then delivers the orders. The service saves the Street Fresh and street vendors time and money, and is key to improving the availability, affordability, and quality of fresh produce in the community. The service also gives the Street Fresh a competitive edge in the community as a greater variety of produce can be stocked and customer demands can be met. The implementation plan of the SFDS is outlined in this report. Start-up capital of R65 000 is needed to successfully implement and maintain the service, and a net profit before taxes of R138 700 can be realised after the first 12 months of operation. The SFDS can save a street vendor R50 each day, increase the vendors profits by 35%, and eliminate the need for the vendor to travel to the market every morning. The SFDS can improve the degree of food security in Soshanguve South, and can be implemented in low-income informal communities in Gauteng to improve their fresh produce supply chains. It is recommended that further research is carried out in the use of cashless systems for the SFDS, establishing a centralised marketplace in the community, establishing direct links with local farmers, and finally the option of re-selling food waste from supermarkets or farms.
Reducing the amount of defects at the quality control station by using statistical process control
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Van Loggerenberg, Nichole
The aim of this document is to provide the reader with information regarding the quality problems the company Parsec (Pty) Ltd experiences in their manufacturing facility, and the solution to solving this problem. Research has been done in order to understand what companies in the same industry with similar problems have done to address the problem. The document first provides the background of the company and its related industry. The problem investigation follows, with detailed information regarding the production process of producing electronic printed circuit boards. After this, a thorough literature study is presented, with technical aspects of printed circuit boards, as well as an investigation into the case studies regarding companies with similar problems. Relevant data was then captured and analysed, after which a demonstration of the solution to reducing defects on printed circuit boards, follows. With the limited data captured and analysed in this project, it can be concluded that Parsec has to focus energy in investigating the reason for the significant amount of missing components, damaged components, and dry joints on their PCBs. Further, the solution will aid production managers in monitoring future defects on the PCBs. Once a defect to be reduced has been identified, it would be recommended for Parsec’s technicians to use SPC as well to monitor and control the characteristics of that specific defect.
Production scheduling of Aluminium Anodising plant
(University of Pretoria. Faculty of Engineering, Built Environment and Information Technology. Dept. of Industrial and Systems Engineering, 2017) Behr, Carina
Wispeco Aluminium decided to reduce a double line aluminium anodising plant to a single line. This change forces the anodising plant to run at maximum capacity. To be able to run at maximum capacity, it is suggested that a production schedule be put in place to optimise the movement of the cranes and minimise the makespan of a day's work. Production scheduling literature, including the popular / / notation was in- vestigated to assist with identifying the characteristics of the relevant problem. An important characteristic that was identi ed is that the anodising problem is a ow shop problem. Once these characteristics were identi ed, peer-reviewed articles were investi- gated to identify similar problems. An article was identi ed which addresses the multiple cranes that cannot cross over one another, the ow shop characteristic and the parallel machines. This article was used to start formulating an algorithmic model as solution. This algorithm was coded in Python. To ensure that the results generated by the algo- rithm are relevant and correct, the results were compared to actual time-studies. The model tested the in uence of three heuristic scheduling approaches on the makespan of the products. The heuristics tested were FIFO, Priority orders and Shortest Processing Times. The results compared over three production days indicated that the schedule does not have a large impact on the makespan of a day's work. This project suggests that the scheduling on the anodising plant is continued as it currently operates. The current method might not be the optimal solution for any given day, but produces an adequate makespan, which will not disrupt the normal processes that workers are used to.

Browse

Recent Submissions