How autonomous control can improve the performance of logistics networks - a simulation experiment

In this thesis the application of autonomous control concepts to logistics networks is studied by means of a simulation model. This simulation model is based on an actual outbound bulk product supply network of a commodity company. Logistics planning and operation is facing growing challenges, such as increasing complexity and distribution, driven by Megatrends such as globalisation and integration. Decentralisation through autonomous control seems to offer to a promising approach to address these challenges. The idea for the supply network at hand is therefore, to enable individual transportation units to autonomously take operational decisions, thus shifting control of the supply network from a central to a local perspective. In surveying the literature and the academic discussion on autonomous control in logistics, software agents are identified as a suitable and well-studied approach to implement such a concept. Therefore, a multi-agent-based simulation model of the supply network is constructed to execute and test the solution. The model is built using data based on empirical observations and offers a full-scale simulation of the actual supply network. In the model, software agents represent the individual transportation units, allowing them to communicate and interact autonomously, effectively decentralising operational control. A comparative simulation experiment is designed and carried out, contrasting several different control scenarios. The simulation results obtained show, that autonomous control can positively impact the performance of this supply network. Autonomous control scenarios require a lower number of trucks to achieve full order delivery and help to increase robustness of the supply network regarding the impact of environmental factors. Additionally, the more efficient use of transportation capacity may lead to a reduction in cost for transportation. The findings are verified with an industry subject matter expert and potential barriers on the path towards implementation are described.