Estudio del consumo

The simulation of a large-scale AGV system is difficult and computational expensive [113]. In order to simplify the kind of simulation, it is necessary to identify a basic system layout that consists of all key elements in a real AGV system so as the performance of the large-scale AGV system can be inferred. The key elements should include stations for different purposes, paths connecting stations together, and a number of AGVs running on the system. Herein, for facilitating the development of the simulation model, a basic AGV system that consists of 9 stations (S1 to S9) and 12 bidirectional paths is considered which is shown in Figure 6.1. The size of the stations and the lengths of the paths are assumed to be the same. These settings are commonly used in large automated warehouses such as a Shentong (STO) Express sorting centre in Hangzhou, China, which is illustrated in Figure 6.2 [114]. More complex systems that consist of more AGVs, more stations, and differing length of paths are studied in later Chapters.

Figure 6.1 System layout

In the system layout shown in Figure 6.1, it is assumed that two different types of AGVs, i.e. single-load and multi-load, are able to operate in the system. However, it is assumed that only those AGVs with the same load-carrying capacity are allowed to run in the same system because using the same type of AGVs can ease management. A typical mission of the single-load AGVs is assumed to consist of five phases, i.e. (1)

mission allocation and route optimisation, (2) dispatch to the targeted pick-up station, (3) collect one item, (4) travel to the corresponding unload station, and (5) unloading. Different from the mission described in Chapter 4, the AGVs are not required to return to the ‘Base’ after unloading because it is assumed that the new mission will be allocated immediately after the completion of the previous mission. So, they can start their next mission directly. The time durations for phases 1, 3 and 5 are assumed based on expert knowledge as listed in Table 6.1.

Figure 6.2 Shentong (STO) Express sorting centre in Hangzhou

It is worth noting that compared to the phase length for phase 1 that was defined in Chapter 4 and 5, a different phase length is defined here for phase 1, which indicates that the system must use more advanced computer and software to achieve a more powerful central control and management system. The time for completing phases 2 and 4 is dependent on the distance or the paths used by the AGVs.

Table 6.1 Assumed phase lengths

Phase Phase Length

(hour) Phase 1: Mission allocation & route optimisation 0.005

Phase 3: Collect one item 0.02

The mission of the multi-load AGV is similar to the single load AGV but there would be more than one pickup and unload stations. The time taken to complete the movement from one station to a directly connected station is assumed to be 0.1 hours. The loading capacity of the multi-load AGV is defined as the maximum number of items it can load. This means the multi-load AGVs will pick up items from different stations and then transport them to the corresponding unload stations. Station S1 is defined as the AGV’s base where the AGVs are stored and charged. Therefore, pickup and unload will not happen at station S1. In every mission, the pickup and unload stations are randomly selected from stations S2-S9, thereby guaranteeing the actual operation of the AGV system can be simulated as closely as possible, and thus the added value of this research for optimising and managing the future AGV systems. The 12 bidirectional paths that connect the stations are assumed to have the same length so that the AGVs will take the same time when travelling on any of them. But for different applications, the layout of the system can be easily modified by varying the length of the paths, the number of stations, and the connectivity between them if necessary. In addition, the bidirectional paths are defined in the system layout, but their width only allows the AGVs that travel in the same direction to go through. As a consequence, all AGVs travelling on the same path must move in the same direction to prevent deadlock. Those AGVs that are going to travel in opposite direction have to wait in the station until the path is evacuated. This will lower the efficiency of the system to certain extent, but today it is still commonly used in practice due to the use of magnetic tape for navigation. The further discussion of this issue is conducted in later Chapter.

Finally, the capacity limit of the 9 stations in the defined system layout is not considered in this research, so that the AGVs that are already parked in the stations will not prevent other AGVs from entering the same station and perform tasks. But this is not true in reality. So, the impact of the station capacity on the system performance will be further investigated in Section 9.2.