Deadlock detection and resolution in automated material handling system

• المؤلف الرئيسي: Huang, BianXun.
• مؤلفون آخرون: School of Mechanical and Aerospace Engineering
• التنسيق: Final Year Project
• اللغة: English
• منشور في: 2013
• الموضوعات: DRNTU::Engineering::Industrial engineering::Operations research
• الوصول للمادة أونلاين: http://hdl.handle.net/10356/54187

Adoption of automated guided vehicles (AGVs) by container terminals has been rapid in recent years. This is especially true in Singapore, with PSA International Pte. Ltd seeking to develop AGVs prototype in bid to increase their efficiency and throughput. However, there is a possibility where the quay cranes, stacking cranes and AGVs request for each other directly or indirectly to start a loading or unloading operation. This leads to a halt among the affected resources and creates a condition call deadlock. In this project, a method to detect deadlocks when it occurs is presented. The method first introduced a way to represent resource allocation graph via a matrix. Matrix operations to reflect real time updates from the system is then proposed, and deadlock can be identified when the matrix detects a cyclic relationship among the resource.A simulation study is done to create the condition of deadlock and test the viability of the detection method proposed. To replicate a realistic scenario present in a real world container terminal, a scenario generator is written to provide input data to the simulation model.Two approaches are discussed to resolve the deadlocks that occur in automated container terminal. The appropriateness under which approaches should be called is discussed, where the approach with minimum impact on the terminal operation should always be called for if condition allows. These resolution procedures aim to modify the handling sequence of the resources involved in the deadlock so that the cyclic request relationship can be broken off.Finally, the performance of the system is analysed and a parameter sweep on the number of AGVs is done to study how performance of terminal is affected by its change.