Development of a discrete event simulation tool with genetic algorithms for the design and optimization of automated warehouses

The automatization and optimization of production systems is central in the new industrial development, and a big part of production efficiency reside in logistics. Automated warehouse can aid in creating smarter and more flexible storage systems. while using less space and energy then classic systems. They provides a fully digital controllable environment where it is easier to apply responsive policy to enhance the functionality not only of logistics, but of the entire production system. However, due to the number of variable and the interdependence of operations,it is not trivial to asses the best combination of technologies, infrastructures and strategy of operation for a given situation. The goal of this Master Thesis is to develop a proof of concept for a tool which able to check the different options and than select the optimal for the specific application. For the evaluation of the performance of a warehouse i used a discrete event simulation, capable of simulating the operation of a specific situation given the parameters about the warehouse and the task it has to perform, and output a set of statistics about operations. The simulator support different technologies, different size, infrastructures setup and strategies of operation. The simulator is written in Python using the SymPy framework. The simulator is then used to optimize the warehouse parameter against a fitness computed on the output statistics. The optimizer ,also written in Python, skims different set of parameters, running the simulations and evaluating the result the find the optimal solution based on throughput, energy consumption, cycle time, infrastructure usage and other metrics, weighted at user will. The optimization uses a genetic algorithm, a optimization method which is inspired by biological evolution. Some case study are utilized and discussed to present the software functionality.