Flexible manufacturing cells may be plagued by a number of common problems. Some of these stem from design errors and are largely avoidable. Others are errors that occur during cell operation and require the design and implementation of contingency plans.
Design problems
It is unlikely that we can anticipate all of the problems associated with designing a cell but we list a few of the most common design pitfalls below:
· Flexibility versus Productivity - The components of the cell can be designed to maximize either of these criteria but not both. It is therefore important to decide from a production standpoint how flexible the cell should be and what its productivity should be. All components of the cell should then reflect this compromise. There is no point in designing extremely adaptable fixturing if the supervisory control system for the cell is not flexible, and vice-versa.
· Failures - Occasional failures must be anticipated. A machine may fail to operate correctly or a 32 procedure may fail to take place as expected. Failures can result from many causes including equipment breakage, faulty incoming parts and robot drift. It is probably acceptable to scrap a part (or in a severe case, a whole batch of parts) but the equipment in the cell should not be allowed to seriously damage itself.
· Critical Items - Evaluate how crucial each component is to the partial operation of the cell. If the lathe is out of commission then the cell can still do parts that require milling. If one of the robots is out of commission then parts will have to be fed by hand. If the supervisory computer is out of commission then the machine tools and robots can no longer function as a cell. In the case of very critical components, such as the host computer, there should always be spare parts on hand. Sensors, such as limit switches used to determinethe presence of a part, should be made redundant so that if one fails then the desired information will still be available.
· Cell Host Extensibility -- The designers of the cell host will certainly not think of everything the cell "should" do. They will probably avoid doing things that are just plain wrong, but experience in working with the cell will illuminate areas that need additional support. The programs running on the cell host must be easily changed. Their integration into what has been called the cell supervisor must be clear and well structured so individual programs can be easily modified and extended. In particular:
o There must be a straight-forward procedure for changing the way the cell supervisor sequences the machines in the cell. More than any other operational feature, the sequencing of the machines in the cell will need to be adjusted.
o The understanding of run-time errors will grow (probably exponentially, at first) with experience. The software to handle the errors should be designed to make it straightforward to add new routines and modify old ones. In addition, the hardware must allow for expansion of the software. A cell host that seems just adequate for the cell at design time will probably soon be bogged down with un-anticipated sensor handling routines.
o The type of data the Supervisor keeps track of will be fairly well understood at design time, but the individual pieces of data almost certainly won't. This is particularly true for the statistical data which is sent to the manufacturing system computer. Additional sampling of random run-time data should be easy to implement.
Runtime problems
Runtime errors will occur in any cell, although they can be reduced by proper design. The cell must be able to recover from minor errors and must at least be able to avoid damaging itself when confronted with serious errors. Runtime errors include broken cutting tools, dropped parts, misaligned parts, machine tool failures and computer crashes. The near-term cell described above is a minimum configuration cell and its response to most of these errors will be to allow one or more parts to be scrapped. One problem with a minimum configuration is that the consequences of any given error are more severe for the cell as a whole. Thus, it is especially important to design durable and perhaps redundant components for the cell.
