Scheduling factories of the future

Abstract

The trend toward factories of the future, indicates that world class manufacturing systems will have no more than 20 highly productive machines with an automated material handling system. These factories will operate in Just-In-Time mode under computer control. The machines, in general, will be more versatile. It is also expected that by the year 2000, cutting speed will be up to 10 times faster than today's standards. Reduced lot sizes will dramatically increase the demand on material handling; cooperative workstations will have to be close to each other. The numerical control code will not be generated until the specific routing for a certain order is known. A typical factory order will require only a few units up to less than a hundred. Every order will be produced in exact quantity Just-In-Time for shipping, hence with no inventory. We strongly believe that the virtual cell concept, extensively documented by McLean, etc., and others, will meet the requirement for the factory of the future. This concept has the potential for making flexible manufacturing systems (FMS) even more efficient, because of its inherent capability for sharing resources. Contrary to the group technology cell configuration, a virtual cell is not identifiable as a fixed physical grouping of workstations, but as data files and processes in a controller. Upon selection of a job order, a virtual cell controller is created. The controller takes over the control of a set of workstations capable of processing the job for which they have been selected. This temporary grouping of workstations is called a virtual cell. Once all resources have been requested and awarded, processing can begin. Parts are moved from workstation to workstation similar to a flow line sequence. We have extended the virtual cell concept proposed by McLean and we have developed a scheduling algorithm that permits the creation of virtual cells and schedule them under workstations and tools availability constraints. Equally important, the algorithm is characterized by its polynomial time complexity. The size of the model grows quickly, but linearly, and remains reasonable for a typical size factory. It is efficient enough to be executed in real time for scheduling any virtual cell based system within hundred of seconds with current computer technology. The scheduling model has many advantages that makes it suitable for scheduling factories of the future. Firstly, it permits consideration of critical tools. Secondly, it provides a quick reaction to unexpected events. Finally, it enables a coherent planning of resource utilization, potentially reducing possibilities of delays and deadlock, which are very common in fast pace manufacturing systems.