In today's global economy the need for an efficient and optimised supply chain is increasing. Recent studies showed that supply chain management is one of the areas that have a great impact on the financial well being of an organization as well as customer satisfaction. The recognition of the importance of efficient and optimised supply chains has led to increasing investments in supply chain planning and execution systems. In order to compete in the global market place organizations want to develop systems that enable fast and effective on time delivery of products to customers. Therefore generating the necessary customer satisfaction. Today there are APS (Advanced Planning&Scheduling) systems available to help "manage" the supply chains. These tools were specifically designed to have the ability to rapidly and simultaneously plan and schedule customer demand while considering material and capacity constraints. Not only does these systems provide the ability to increase revenues, but it can also increase the customer service and cut costs by synchronized management of the complete supply chain. Although these systems help to improve the system, it is restricted to the static part and it does not incorporate the dynamic part. The result therefore is that a lot of "noise" still exists within the system once the results are achieved. This opened the way for solutions that can provide insight to the uncertainty and interdependency of processes and customer demand within the supply chain. One way of gaining insight into the system variation and interdependencies is through the use of simulation technology. This type of technology allows organizations to predict future behaviour and test future designs or do redesigns of their current supply chains. The scope of this dissertation is to develop a supply chain planning methodology, which will help to improve the understanding of the uncertainty and interdependency of processes within the supply chain. To design this methodology different steps are taken in order to introduce the final solution. Therefore, four main methods were used; literature research, market research, supply chain planning methodology development and a case study. The literature research brought to light the reasons for the inefficiencies and variations in supply chain planning and why the need for change exists. During the supply chain market research several supply chain planning and execution systems were under study. From this it was quite clear that the only way that organisations can ensure one optimal answer is when the demand is constant and there is a zero percent chance that it could change. In real world systems it is virtually impossible to accurately predict future demand 100 percent of the time, and therefore variability and randomness cannot be excluded from a supply chain solution. This paved the way for the introduction of simulation technology as a possible solution for this variability and randomness. The market research was concluded with the analyses of the current simulation solutions in the market. The next step in the design phase was the introduction of the new supply chain planning methodology. The main purpose of this new methodology is to use the power of modelling and simulation to improve the initial supply chain performance. This methodology focuses on initial supply chain design, analyses and optimisation. By introducing this methodology organisations are now able to compare current supply chains with an unlimited realm of possible future configurations .... and without disrupting the initial day-to-day operations of an actual supply chain. The methodology is also designed to help predict the supply chain performance in terms of throughput, tardiness, utilisation, profitability, and other key performance indicators ... In order to experience real-life supply chain problems a case study has been done. This case study is about the automotive industry, which will include the ordering of parts assembly of vehicles, warehousing and distribution of vehicles. Different problems and difficulties were experienced. In conclusion, this case study provided a better insight into the behaviour of a supply chain. The case study was used to evaluate the use of this new methodology and as a result certain inefficiencies were recognized. As a result of the evaluation certain improvements need to be made to the supply chain methodology in order to make it more suitable for the market. These improvements would focus on inventory planning, supply chain analysis as well as database integration. The result of the case study also showed that the supply chain planning methodology is now set to develop a supply chain solution on the lowest level. There is however a need to be able to grow this supply chain methodology from a low level to a relatively high level. These functions are among others higher-level planning modules, which focus on transportation, production, demand and distribution and performance measurements. The focus will be to introduce these functions as objects. Every object will have the ability to design a supply chain solution on a high level or low level depending on the detail and requirements. ••• I also believe that the one who adapts his policy to the times prospers, and likewise that the one whose policy clashes with the demands of the times does not. 11 Niccolo Machiavelli, 1525 / Dissertation (MEng (Industrial Engineering))--University of Pretoria, 2006. / Industrial and Systems Engineering / unrestricted

M.Comm. / In the last hundred years there has been a technological revolution that has forced people to change the way they live and run their organisations. This technological revolution has had a major impact on the business world. Coyle, Bardi and Langley (2003; 57) have suggested that “the rate of change has accelerated with consequent negative impacts if organisations do not change.” With today’s emphasis on cutting costs, streamlining expenses while at the same time trying to offer a competitive edge with regard to customer service, many organisations are looking to improve their bottom line and financial performance by implementing new technology into their supply chains. A popular way for organisations to speedily reap the benefits of having a more competent and competitive, technology-enabled supply chain, is by outsourcing their supply chain needs to a third party logistics organisations. The Star newspaper reported that, according to Brett Bowes, inefficiencies in the supply chain meant that fast-moving consumer goods manufacturers and retailers were losing R7 billion every year (The Star, 16 March 2007; 2). Although there are many auxiliary benefits and advantages of implementing software systems into the supply chain, the two critical benefits which justify the expense are reduced costs and improved customer service. Implementing technology is a costly, challenging and sometimes risky endeavor. This often results in an unwillingness to change until these organisations outgrow their systems, or the business environment becomes so complex, that they are forced to implement new technology. This hesitancy to introduce new technology timeously could hamper the progress and growth of these organisations, and could also affect their competitiveness in a highly competitive environment. The purpose of this case study is to compare the benefits achieved from existing older technology to new technology, based on a routing and scheduling case study in a large 3PL organisation.

CAPES / O segmento downstream da cadeia de suprimentos da indústria brasileira de petróleo é composta por bases de produção (e.g. refinarias), armazenamento (e.g. terminais) e consumo (e.g. mercados consumidores) e modais de transportes (e.g. oleodutos, navios, caminhões e trens). O planejamento da transferência de derivados de petróleo nesta rede multimodal é um problema complexo e atualmente é realizado para um horizonte de três meses com base na experiência de profissionais e sem auxílio de um sistema computacional de apoio à decisão. Basicamente, o problema pode ser visto como uma negociação para alocação de recursos disponíveis (tais como derivados de petróleo, tanques e modais de transporte) pelas diferentes bases envolvidas que necessitam enviar ou receber derivados de petróleo. Na literatura, alguns problemas semelhantes, porém mais voltados para o planejamento de redes formadas por um único tipo de modal de transporte, têm sido tratados por diferentes abordagens, com predominância da programação matemática. Estes trabalhos ilustram a difícil tarefa de modelar grandes problemas por meio desta abordagem. Geralmente, tais trabalhos consideram apenas um curto horizonte de planejamento ou apenas uma parte do problema original, tal como uma parte da rede petrolífera brasileira, gerando limitações importantes para os modelos desenvolvidos. Devido às características do problema em estudo, o qual envolve toda a rede de transporte e apresenta perfil de negociação entre as diferentes entidades envolvidas, surge o interesse da utilização do paradigma de sistemas multiagente. O paradigma de agentes tem sido aplicado a problemas de diferentes contextos, particularmente em problemas de gerenciamento de cadeias de suprimentos devido à sua correspondência natural com a realidade e, em geral, em problemas que envolvem a competição por recursos por meio de mecanismos de negociação com base em leilões. Este trabalho apresenta um novo protocolo de negociação baseado em leilões e aplicação deste protocolo em forma de um modelo multiagente na resolução do problema de planejamento em questão. Os agentes que formam a solução representam principalmente os locais de produção, armazenamento, consumo e os modais de transporte na rede petrolífera brasileira. O objetivo destes agentes é manter um nível de estoque diário factível de cada produto em cada local por meio de transferências de produtos pela rede petrolífera brasileira com preferível redução do custo de transporte. Por fim, este trabalho apresenta a satisfação destes objetivos por meio de experimentos em cenários fictícios e reais da rede brasileira de petróleo. / The Brazilian oil supply chain is composed by oil refineries, consumer markets, terminals for intermediary storage and several transportation modals, such as pipelines, ships, trucks and trains. The transportation planning of oil products in this multimodal network is a complex problem that is currently performed manually based on expertise, for a period of three months, due to the lack of a software system to cover the problem complexity. Such problem involves the negotiation of available resources such as oil products, tanks and transportation modals between different sources and consumption points. Similar problems, but more directed to the planning of single modes of transportation, have been treated by different approaches, mainly mathematical programming. Such works illustrate the difficult task of modeling large problems with this mechanism. Generally, they consider a short horizon planning or only part of the original problem, such as a part of the network, rendering important limitations to the models developed. Due to the characteristics of the problem in study where the full network needs to be considered and there exists negotiation amongst the different entities involved, the usage of multi-agent models seems to be worth to explore. Such models have been applied in different contexts such as to supply chain problems due its natural correspondence with the reality. Furthermore, in problems involving competition for resources, multi-agents negotiation mechanisms based on auctions are commonly applied. Thus, this thesis presents one auction-based solution formed by the cooperation among agents for them to achieve their goals. The agents involved in the auctions represent mainly the production, storage and consumption locations. Their goal is to maintain a daily suitable inventory level for each product by means of transportation through the multimodal network at a low transport cost. Finally, this paper presents the satisfaction of these objectives through experiments on real and fictional scenarios of Brazilian oil network.

CAPES / O segmento downstream da cadeia de suprimentos da indústria brasileira de petróleo é composta por bases de produção (e.g. refinarias), armazenamento (e.g. terminais) e consumo (e.g. mercados consumidores) e modais de transportes (e.g. oleodutos, navios, caminhões e trens). O planejamento da transferência de derivados de petróleo nesta rede multimodal é um problema complexo e atualmente é realizado para um horizonte de três meses com base na experiência de profissionais e sem auxílio de um sistema computacional de apoio à decisão. Basicamente, o problema pode ser visto como uma negociação para alocação de recursos disponíveis (tais como derivados de petróleo, tanques e modais de transporte) pelas diferentes bases envolvidas que necessitam enviar ou receber derivados de petróleo. Na literatura, alguns problemas semelhantes, porém mais voltados para o planejamento de redes formadas por um único tipo de modal de transporte, têm sido tratados por diferentes abordagens, com predominância da programação matemática. Estes trabalhos ilustram a difícil tarefa de modelar grandes problemas por meio desta abordagem. Geralmente, tais trabalhos consideram apenas um curto horizonte de planejamento ou apenas uma parte do problema original, tal como uma parte da rede petrolífera brasileira, gerando limitações importantes para os modelos desenvolvidos. Devido às características do problema em estudo, o qual envolve toda a rede de transporte e apresenta perfil de negociação entre as diferentes entidades envolvidas, surge o interesse da utilização do paradigma de sistemas multiagente. O paradigma de agentes tem sido aplicado a problemas de diferentes contextos, particularmente em problemas de gerenciamento de cadeias de suprimentos devido à sua correspondência natural com a realidade e, em geral, em problemas que envolvem a competição por recursos por meio de mecanismos de negociação com base em leilões. Este trabalho apresenta um novo protocolo de negociação baseado em leilões e aplicação deste protocolo em forma de um modelo multiagente na resolução do problema de planejamento em questão. Os agentes que formam a solução representam principalmente os locais de produção, armazenamento, consumo e os modais de transporte na rede petrolífera brasileira. O objetivo destes agentes é manter um nível de estoque diário factível de cada produto em cada local por meio de transferências de produtos pela rede petrolífera brasileira com preferível redução do custo de transporte. Por fim, este trabalho apresenta a satisfação destes objetivos por meio de experimentos em cenários fictícios e reais da rede brasileira de petróleo. / The Brazilian oil supply chain is composed by oil refineries, consumer markets, terminals for intermediary storage and several transportation modals, such as pipelines, ships, trucks and trains. The transportation planning of oil products in this multimodal network is a complex problem that is currently performed manually based on expertise, for a period of three months, due to the lack of a software system to cover the problem complexity. Such problem involves the negotiation of available resources such as oil products, tanks and transportation modals between different sources and consumption points. Similar problems, but more directed to the planning of single modes of transportation, have been treated by different approaches, mainly mathematical programming. Such works illustrate the difficult task of modeling large problems with this mechanism. Generally, they consider a short horizon planning or only part of the original problem, such as a part of the network, rendering important limitations to the models developed. Due to the characteristics of the problem in study where the full network needs to be considered and there exists negotiation amongst the different entities involved, the usage of multi-agent models seems to be worth to explore. Such models have been applied in different contexts such as to supply chain problems due its natural correspondence with the reality. Furthermore, in problems involving competition for resources, multi-agents negotiation mechanisms based on auctions are commonly applied. Thus, this thesis presents one auction-based solution formed by the cooperation among agents for them to achieve their goals. The agents involved in the auctions represent mainly the production, storage and consumption locations. Their goal is to maintain a daily suitable inventory level for each product by means of transportation through the multimodal network at a low transport cost. Finally, this paper presents the satisfaction of these objectives through experiments on real and fictional scenarios of Brazilian oil network.