Chapter 7: DISCUSSION AND CONCLUSION
7.1 Ontology Updates
As already stated in the previous chapter the Ontology we developed should be updated in order to fulfil the requires of the manufacturer. In particular it is necessary to monitor the market in order to find new solutions for our present processes. Sometimes when new technologies come up, could be effective, from the cost’s point of view, to make change in the assembly cycles. Moreover new technologies could lead to new business of the company, and it entails an huge effort.
Summarizing, the needs of an upgrade surface when:
• The study of a new product shows lacks or un-effective way of work. • A new technology can be apply to a current process.
The General Ontology includes the whole Know-How of the company about the present assembly processes: for this reason it is a dynamic set of files that grows more and more with the manufacturer’s experience. Of course, refreshing the Ontology means, not only adding new classes, but also removing the useless, obsolete ones.
All the considerations that lead us during the Ontology reviews are explained in the third chapter: in particular all the choices about the linkage have been deeply discussed.
7.2 The Electrolux® Test case
Both the “Handle” and the “Electric Valve Group” belongs to a well defined Class of product. We chose, of course, the two most suitable single product among the possible set and we tried to adapt them for an Automatic Assembly System by the evaluation with the DFA2 scheme and by several modifications. Anyway the nature of the products does not allow to have a proper automated process. For example in the Electric Valve Group cycle there is the need of Oil to lubricate the tube and such an activity is not recommended for an Automatic Assembly Process.
Our main Goal was to prove the possibility to change Assembly System in few minute (instead of months). First of all the products we studied are very different, therefore they are very suitable to show how the system “evolves” in order to cope heterogeneous assembly cycles.
This objects are included in the same object, an Electrolux Washing Machine and this is useful to demonstrate the economical sustainability for a Small-Medium sized Enterprise of the “Evolvable” approach.
7.3 Emergent Behaviour
The Emergent Behaviour is one of the most important strong point of this new approach. The Evolvability of a System comes from the interactions among its components. Of course the higher is the granularity, the higher is the Emergent Behaviour. Anyway it is important to highlight that a fine Granularity costs and therefore it is necessary to solve the trade-off between economical and technical aspects. As long as an higher granularity ensures a profit it could be implemented. Of course, it depends from the specific situation: sometimes it is enough to have an agentified gripper, sometimes even the single agentified fingers.
In our test-case, we found two different situations where it is possible and economically opportune to exploit the Emergent Behaviour.
In the “Handle” Cycle it is necessary to push the sub-assembled “Latch-Spring” into the Plastic Handle. The Emergent Behaviour of the gripper allows to perform this action without a specific cylindrical Tool, and therefore without a Tool exchange. Each device is designed to have a basic Skill, but as we have seen above other skills can surface, especially because of the cooperation among the components. This example shows the importance to consider all the existing feature of the devices (Two locked finger have the same function of a specific tool to push!)
In the “Electric Valve Group” Cycle we used the Clamp Ring to grab the tube, instead of a specific gripper: we save the time for the gripper exchange and the money for the development of such a gripper. This example is very suitable to show that is possible to get an useful
Emergent Behaviour not only by the cooperation among the devices, but also by the integration of devices and Product.
7.4 Future Developments
All the aspects shown in the § 2.3 Limitations, should be object of further studies. One of the main issue for an Evolvable System will be the development of suitable interfaces for all the components. These interfaces must provide all the necessary connections ( Mechanical, Electrical, Pneumatic, Signal....).
Proper interfaces make the Agentification process easier. Moreover the interfaces allow the plugability. In order to reduce the emplacement time we recommend to use lever mechanism instead of screws or other complex way to plug the modules up.
The Agentification of all the modules included in our layouts should be carried out in order to have very task-specific devices. Of course this is a stepwise process: first of all it is necessary to agentify the most important component such for example robots, feeders, conveyors, pallets etc.. and then the others. Particular attention has to be paid to control issues and programming efforts: the best hardware is useless if it cannot be efficiently programmed.
We did not take care of the integration between the automated and the manual station, but this is a very important issue for an assembly System. In the first period of a product lifecycle the volume uses to be very low, so it is not convenient to develop a fully automated Assembly System: if the product goes wrong the investment could cause problems for the company. Therefore, at the beginning it is advisable to use human workers and then if it is necessary (during the ramp-up phase) just change them with automated stations.
Creating assembly systems for specific products leads to dedicated installations which are extremely expensive and are rarely amortisable in today’s market situation, where a product is only manufactured during a very limited time. Globalisation brings strong concurrence from all over the world, but especially from eastern low-wedge countries such as China and India. To stay or become again competitive, we need short Time-to-Market, quick Ramp-up, many product variants in medium or small lot sizes, fast reactivity and low cost.
date there is a current research about EMS (Evolvable Manufacturing System) in Lisbon UNINOVA .
As already stated above the importance of the Assembly System leads us to put the focus on the Process and Modularity, instead of the Product and Flexibility. The previous approach tried to put the flexibility in the devices, we attempt to have a flexible concept.
The EAS Module’s main feature is the interface that gives its the possibility to join the module society in order to cooperate to accomplish the task. Of course each module must be acknowledged by the Control System, and for these reasons we included in our Ontology the Verification Units.
In natural system the lower a component is positioned in the hierarchical structure, the more flexibility is necessary to ensure a certain flexibility to the whole system; this means even a system with low flexibility needs a very flexible Control (see fig.2). If otherwise, all the components are task-specific they can cooperate to reach the common goal, and through the emergent behaviour they can accomplish an unexpected amount of different activities.
Summarizing, the main points are:
• Analysis of processes, development of ontologies • True modularity
• Agent-based control systems
With this work a first attempt to make a process-oriented Ontology has been made, and we provided a test-case, which, if developed in the right directions is suitable to prove the effectiveness of such an approach.
