Doctoral Degrees (Industrial Engineering)

Permanent URI for this collection

https://scholar.sun.ac.za/handle/10019.1/775

Browse

Browsing Doctoral Degrees (Industrial Engineering) by browse.metadata.advisor "Bekker, James"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    An agent-based model of Eldana Saccharina Walker
    (Stellenbosch : Stellenbosch University, 2016-12) Van Vuuren, Brian John; Potgieter, Linke; Bekker, James; Stellenbosch University. Faculty of Engineering. Dept. of Industrial Engineering.
    ENGLISH ABSTRACT: An agent-based simulation model is formulated in this dissertation in order to simulate the population dynamics of Eldana saccharina Walker infestation in sugarcane. The simulation model comprises four distinct building blocks, namely a graphical user interface, the implementation of the life cycle and associated influence of temperature on E. saccharina, the mating process of E. saccharina and the oviposition and dispersal of E. saccharina. These building blocks are based on existing literature pertaining to the biology and behaviour of the pest and, in cases where the relevant literature is insu cient or non-existent, expert opinion and careful assumption. In order to select areas from existing sugarcane farms on which to execute location-speci c experiments, functionality which allows Geographic Information Systems (GIS) data importation is included as a platform on which to run the simulation model. These data inform the model in respect of the shape and characteristics of the underlying sugarcane elds in which the simulated E. saccharina population interacts. The model interacts and operates within an AnyLogic simulation software environment and, in so doing, aims to emulate the behaviour of a population of E. saccharina moths in sugarcane. It is anticipated that the model implementation may serve as a basis for facilitating future design and testing of control measures in order to suppress the pest and its consequent detrimental e ect to sugarcane through infestation and feeding on interior stalk nutrients. Numerous working mathematical models of the pest exist in the literature, but, in all previous cases, intricate aspects of the stalk borer's biology have been aggregated on a population level and average population changes have been a ected at discrete time steps. The resulting analyses therefore yield conclusions that do not necessarily re ect the continuous, changing nature of E. saccharina on a localised level. Using agent-based modelling, however, the pest's behaviour may be modelled in more detail so as to facilitate more thorough investigation of potential control strategies and their expected e cacy on the pest at di erent points in its life cycle. The agent-based simulation model designed in this dissertation is subjected to a number of veri cation and validation techniques. Furthermore, a pilot sensitivity analysis is conducted to identify the most in uential parameters in the simulation model. These parameters are then considered further in a comprehensive parameter variation analysis in order to illustrate the exibility and diversity of the model in terms of the variety of scenarios pertaining to E. saccha- rina population behaviour that it can accommodate. In some cases, simpli ed implementations of control measures are also imposed on the pest within the model in order to further illustrate its implementation capabilities, as anticipated for future model development and use. In light of this exibility, the model is also presented as a computerised decision support and analysis tool, including the ability to upload and recreate a speci c user's own sugarcane farm shape le, as well as to alter a set of available parameters. This may aid in simulating speci c behaviour in a simulation run in accordance with what has typically been observed by the user, or of hypothetical scenarios which require investigation. In turn, as the model is further developed and detailed control measures are included as part of the simulation execution, it is believed that an appropriate response pertinent to the characteristics of the geographical area under consideration and the corresponding E. saccharina population present in this area may be predicted, allowing for control measure alteration and redesign so as to optimise the associated parameters or actions prior to in- eld implementation. In order to further re ne the model and improve its accuracy, as well as ensure agreement between the existing modelling approaches and actual biological processes in nature, the entire simulation model of E. saccharina is subjected to an expert panel discussion. The experts comprising the panel encompass some of the key researchers pertaining to E. saccharina and other moth behaviour and population dynamics, both in South Africa and internationally. The simulation model is updated or adjusted according to suggestions made and new information shared by the expert panel in an attempt to simulate the pest as accurately as possible in accordance with the body of knowledge currently available. Although several other approaches to modelling E. saccharina populations have been adopted in the past, no existing models implement such a low level of abstraction with respect to the biology of the pest. In addition, previous models are often case-speci c, investigating speci c control measures that are imposed on an aggregate level on a population of the pest. By actively simulating E. saccharina's biological decision-making processes, intricate aspects pertaining to one or a number of interacting control strategies, as well as the manner in which they alter the pest's biology or behaviour, may easily be incorporated using an agent-based simulation modelling approach. Adopting a low level of abstraction also requires extensive information pertaining to the pest and, as such, areas where little understanding still exists with respect to the behaviour of E. saccharina have been highlighted and, consequently, may be prioritised for future entomological research by experts in the eld. Finally, numerous options for future investigation into this problem, including model re nement, control measure design and testing and comparison to existing models, pose positive possibilities for the eventual establishment of a functional, integrated pest management programme for E. saccharina.
  • Thumbnail Image
    Item
    A harvest and processing decision support system for table grape production
    (Stellenbosch : Stellenbosch University, 2023-02) Wium, Jolene; Van Eeden, Joubert; Bekker, James; Stellenbosch University. Faculty of Engineering. Dept. of Industrial Engineering.
    ENGLISH ABSTRACT: Harvesting and production decisions for large table grape producers are complex. Customers demand specific product requirements, based on the cultivar, packaging type, and grape quality. Harvest efforts by producers depend on the ripeness and volume of stock, which can change over time, justifying a constant modification in company-wide planning. This study concerned the harvest and processing decisions for table grape producer-exporters, using data obtained from a large South African producer exporter as a case study. A mixed-method—exploratory sequential design was employed in this study to support the engineering design process. User requirements were obtained from interviews with operational, tactical, and strategic decision-makers. The user requirements guided the design and build of a system, supporting decisions at each decision-making level in the company. The decision support system uses data and model driven methods. It comprises managing source data, an algorithm for prescribing suitable harvest and pack plans and, last, data visualisation, presenting results and information to the decision-maker. Data were extracted from source systems and stored in a data warehouse providing a stable, non-transactional environment, suitable for large database queries. The non-dominated sorting genetic algorithm II was used to prescribe weekly harvest, processing, and delivery plans. The two model objectives minimised deviation from the demand plan and travel distance between the orchard and the pack site to preserve grape quality. An operational model builds on the weekly tactical model, providing an in-week daily schedule for each pack site and farm. Users could access role-specific visualisations by providing insights into their deliverables. End-users and industry experts validated the decision support system. It was also compared to an existing human system. The developed model closely resembles the human model; however, it can provide a result in a much-improved time. The interconnectedness of the harvest supply and processing facilities justifies an update of the entire plan when an attribute of the plan changes. Changes to either processing or demand factors can easily be incorporated through a model, whereas the human system relies on heuristic methods for an end result. Automation through optimisation, therefore, supplies a solution in a constantly changing harvest and demand plan environment. The study produced a harvest and processing model, contributing to an omission in literature, therefore, focusing on table grapes. Specific handling techniques required for table grapes justifies the need for specific objective function values during modelling. It also presents a unique contribution through a geographically diverse case study focus, accommodating customers’ specific pack-to-order needs. Past studies established orders by cultivar. This study extends past work by including box type and the ability for a customer to reject a specific cultivar as an order.
  • Thumbnail Image
    Item
    New multi-objective ranking and selection procedures for discrete stochastic simulation problems
    (Stellenbosch : Stellenbosch University, 2018-03) Yoon, Moonyoung; Bekker, James; Stellebosch University. Faculty of Engineering. Dept. of Industrial Engineering.
    ENGLISH ABSTRACT: In stochastic simulation optimisation, several system designs are considered. These designs are ranked in order and the best is selected based on one or more performance measures. Any ranking and selection (R&S) procedure must ensure that the correct system design is chosen, and this is a challenging task in the stochastic environment. This dissertation discusses the design and development of a new multiobjective ranking and selection (MORS) procedure, called Procedure MMY, and two variants of it, called Procedures MMY1 and MMY2. Single-objective ranking and selection procedures endeavour to find the best system, i.e., the system with the minimum or maximum output, out of a limited number of feasible solutions. There are two important approaches in the single-objective R&S area: the indifference-zone (IZ) approach and the optimal computing budget allocation (OCBA) framework. While the OCBA procedure has been extended to the multi-objective domain, an MORS procedure with the IZ approach has not yet appeared in the literature. The MMY family procedures have been developed in an attempt to fill this gap, therefore they take the IZ approach. Indifference-zone procedures should guarantee that the probability of correct selection is at least a prespecified value P*, denoted by P(CS) * P*, where `correct selection' denotes the event that the system with the minimum output is selected for a single-objective minimisation problem. In the multi-objective context, Pareto optimality is employed to define `correct selection'. The concept of relaxed Pareto optimality is proposed in this research to accommodate the indifference-zone concept properly in the multi-objective domain. Thus, Procedure MMY guarantees P(CS) * P* considering the event of identifying a relaxed Pareto set as a correct selection. Procedure MMY1 tries to find the normal Pareto optimal set while Procedure MMY2 focuses on identifying Pareto optimal solutions with the IZ concept. The statistical validity of the MMY family procedures is proved through rigorous mathematical analyses in this dissertation. A Bayesian probability model was used in the P(CS) formulation in the proofs. Using a Bayesian model in the P(CS) formulation in IZ R&S procedures is a novel approach even in the single-objective context. The researcher therefore proposed a new single-objective R&S procedure, called Procedure MY, in addition to the multi-objective MMY family procedures. The MY procedure is discussed prior to the discussion of the MMY family procedures, verifying the effectiveness of the Bayesian model, thereby laying the theoretical foundation for employing it for the MMY family procedures. The performance of the proposed MMY family procedures was demonstrated using four simulation case studies. These simulation case studies provided various types of test beds to understand the behaviour of the proposed procedures. In all four cases the estimated probability of correct selection was observed to be greater than P* for all three procedures, proving the statistical validity of them empirically, too. In addition, the performance of the proposed MMY family procedures was compared to that of the MOCBA procedure, which is the only existing MORS procedure. The result showed the superiority of the MMY procedure over the MOCBA procedure in many cases.