Flowpath Layout

Flowpath layout, also known as guidepath design, is one of the most important and priority issues to be considered in the design of an AGV system [10, 11]. It determines the route that the AGVs will follow during operation, thereby having a significant influence on the cost of material transportation. A flowpath layout defines the connections between stations, machines, target positions and other structures, which are usually assumed to be stationary or fixed and via them, the AGVs can reach desired

pick-up and delivery points. However, it is worth noting that there is a difference between ‘flowpath’ and ‘guidepath’, i.e. the moving directions of AGVs on guidepaths are always fixed while their moving directions on flowpaths can be modified as required [19]. A successful flowpath layout design will enable the AGV system to have short AGV traveling distance and high efficiency, thereby saving both time and cost [10].

At present, there are three types of flow topologies that are popularly adopted in layout design. They are respectively single loop, tandem configuration, and conventional topology, as shown in Figure 2.1. In each layout design, AGVs can travel along the paths in either one direction (i.e. unidirectional) or in two directions (i.e. bidirectional). Among these designs, single loop is the simplest case, in which the flowpath consists of a single loop only [20]. A typical single loop layout is illustrated in Figure 2.1(a) [21]. Where, an AGV picks up raw material from the input point ‘I’ and carries the material to machine ‘M1’ for accepting processing for the first time. After the material is processed by machine ‘M1’, the AGV will carry the processed material to machine ‘M2’ for further processing. Finally, the AGV will carry the material after further processing to the output point ‘O’ and drop off the material there. After dropping off the material at the output point ‘O’, it is regarded that the AGV has finished the allocated job in the present cycle. Then, it will return to the input point ‘I’ from the output point ‘O’ and start a new cycle of job by repeating the loading and unloading process. In 1992, Tanchoco and Sinriech’s established a method to find the optimal single-loop flowpath design [20]. An integer programming formulation was employed to identify an initial valid loop. Following that, all possible valid single loop guidepaths will be identified by using an enumeration procedure. Finally, the optimal locations of the pick-up/delivery stations along a given loop are obtained by using a mixed integer programming formulation based on a from/to material flow matrix. As opposed to using more complicated guidepaths, the benefit of using a simple flowpath under light and average shop workload is that it can be easily realised with the aid of a simpler controller. However, it was found that the single loop layout is inefficient, and it will require more space and AGVs when the scale of the system and the required

workload increase. Moreover, vehicle interference would be an issue when multiple AGVs travel at different speeds in the same single loop. Hence, the following two alternative flow topologies are developed in modern applications.

The tandem guidepath system was first introduced by Bozer and Srinivasan in 1991 [22]. According to [22], the tandem guidepath system usually consists of a few zones, but there is only one AGV operating in every zone. The zones are connected via transfer stations. In this way, the deadlock and conflict problems can be eliminated. Herein, deadlock refers to the situation where a part of the system stalls, i.e. the vehicle is blocked by other AGVs in an unsolvable situation [23]. Conflicts usually arise when several AGVs try to run on the same route or pass the same crossing points. These two factors have a huge impact on AGV speeds, expected travel time, and route planning. The concept of tandem guidepath system was further improved later on. For example, as shown in Figure 2.1 (b) [24], the tandem configuration consists of a few non- overlapping unidirectional loops. One or more AGVs can run in each loop and transfer between adjacent loops can be made through pre-designed transmission points [24]. Apparently, most deadlock and conflict problems can also be avoided in such a design. Farling et al. developed a typical tandem configuration, grouping together several single loops without overlapping [25]. In their work, the impact of different factors, including system size, number of loops in the tandem layout, load/unload time, failure rate of machines in the AGV system, and the tandem AGV systems performance with respect to the mean flowtime of AGVs, were investigated by the approach of simulation written using Arena® a popular simulator. The mean flow time is defined as the average amount of time it takes for jobs to complete their sequence through the system. The simulation results have disclosed that system size, load/unload time, and machine failure rate factors have significant influence on the operation of the AGV system. Moreover, it was found that the impact of the load/unload time on the performance of the systems considered in [25] can be mitigated by adding more loops to the traditional tandem configuration.

A conventional topology introduces more features, such as crosses and junctions, into the layout, as shown in Figure 2.1 (c) [26]. AGVs in conventional layouts are able to reach any targeted positions usually with more than one different routing options. The conventional flowpath system can be either unidirectional or bidirectional. The system with bidirectional paths is more complex but permits an improvement in efficiency, time and space saving [11]. However, the complexity of AGV systems with bidirectional networks increases enormously. Hence, problems like blocking, conflict, collisions, and deadlocks will arise more frequently in bidirectional applications [27]. The sudden occurrence of these problems may directly lead to the failure of the control system or even lead to physical hazards to the environment and humans. Deadlocks and collisions are usually not considered in the single loop tandem configuration because there is only one AGV operates in the zone.

(a) Single loop (b) Tandem configuration

(c) Conventional topology

More detailed description about the aforementioned flowpath layouts can be found from [21, 24, 26].

In 2007, Wu and Zhou studied the deadlock and blocking issues existing in bidirectional layouts [13]. They tried to construct the shortest route with the aid of coloured resource-oriented Petri Nets (CROPN) by prioritising the deadlock-free conditions. By setting the deadlock-free operation condition initially, the study provided a method to generate the route of AGVs without blocking. However, the model they constructed is a static model, which cannot precisely describe the operation of a real-life AGV system. To overcome this issue, Nishi and Tanaka tried a Petri Net decomposition approach in order to reach the same goal in dynamic environments [28]. The comparison with the results obtained by Wu and Zhou [13] has shown that the performance of the AGV system was indeed improved by optimising the task assignment and routing simultaneously.

The conventional bidirectional flowpath system is not popular in material handling systems because it adds more complexities to the control management problem. In 1992, Bozer and Srinivasan compared the performance of tandem and conventional AGV systems that consisted of 8 AGVs and 20 workstations [26]. It was found that a conventional AGV system performs better if there are only three or four vehicles in operation. However, the route optimisation of the conventional topology was very limited due to the assumption of single-AGV zones at that time. Yu and Egbelu developed a heuristic partitioning algorithm for a tandem AGV system with variable paths rather than loops [29]. A conventional layout was converted into a tandem AGV system layout. Then a comparison between the two layouts was conducted with respect to system performance under the same operating conditions. It was found that the modified tandem layout requires less AGVs to complete the same tasks in most cases. In addition, tandem layout required a shorter AGV usage time to complete the task than the conventional layout did. Also, the control problems related to vehicle routing and deadlock resolution in the tandem layout is relatively simpler due to the absence of

traffic congestion and conflict from the tandem layout.On the other hand, as the route optimisation was conducted previously for various purposes (e.g. minimising travel time, travel distance, and optimal number of vehicles required, etc.), it is not easy to carry out the comparison of different topologies [10]. For this reason, there is not an official standard today for specialising the layout design of AGV systems.