PMA - Production Engineering, Machine Design and Automation
How to refer to this page:
Bongaerts, 1998, "Integration of Scheduling and Control in Holonic Manufacturing Systems", Ph.D. Thesis PMA/K.U.Leuven, Chapter 3.

Process Control

The process planning and execution holarchy -- called process control -- takes care of the functions ranging from product design, and process planning to on-line production, processes and monitoring. Most work in holonics does not include the entire design process, but interfaces the holonic production system with existing CAD systems. Process planning includes the sequence planning of operations (also called route planning), the process or machine selection, the set-up planning, fixture design and production, tool selection, and the specific operation planning itself (including tool paths, and process parameters, like cutting speeds).

A typical process planning and execution holarchy could look like this (Kruth, 1996): products and resources contain the basic information and methods to plan and execute all processes. However, when a product is created (because there is an order for it in an engineer-to-order environment or because it was initiated by marketing), it first gets itself a geometry via the CAD system and the (human) designers. Then, it gets itself analysed such that technological features (CAPP-features) are extracted. Using this information, the sequence planning holon (a staff holon) defines the necessary operations and the possible operation sequences. The off-line process planning holons (that can differ for different types of operations) extract the CAM-features (the machine dependent features) for each CAPP-feature. They thereby consider the resources to use (like machines, fixtures), the operation types, and the workstations types able to execute the operation. If necessary, they can split up CAPP features into several CAM-features or join several of them to one CAM feature. These sequence planning and off-line process planning holons should co-operate with the resource holons to account for their abilities, available capacity and status. The final process planning step is done on-line inside the machine, defining the exact operation parameters like the tool, the cutting parameters, the tool path, etc. The operation is started however when triggered by orders in the resource allocation holarchy (Wyns, 1996).

At PMA/K.U.Leuven, the following results on holonic process control were obtained. Van Zeir (1997) presents a feature based process planning system, developed in the COMPLAN project, and its extension to a blackboard-based, distributed process planning system, where expert modules co-operate to combine process and enterprise specific knowledge to come to an overall process plan (sequences, process and machine selections, fixture design, set-up planning, tool selection, quality control, selection of process parameters, time and cost calculations, and so on). One of the major advantages of this approach is the interactive nature of the process planning process. Moreover, it is possible to incorporate the process planning knowledge in small autonomous and reusable units. This enables the fruitful co-operation of human and computer experts, and makes it easier to integrate process planners for different process types (e.g. machining, heat treatment and grinding). It also allows the system to combine the advantages of early and late process planning. It is possible to postpone process planning decisions as long as possible, and come back on taken decisions if disturbances cause replanning. However, if some decisions need to be taken on time, for instance to prepare the necessary auxiliary resources, this approach enables the expert to do so without taking all decisions at that early instant in time.

Van Ginderachter (Kruth, 1998) is developing a feature based NC machine, using product holons, sequence planning staff holons, and resource holons. The NC machine can autonomously react to disturbances because it understands the meaning of its actions. Reactive process planning can take place at different levels, balancing the severity of the problem with the appropriate action. Products and orders negotiate -- order specific and during production -- about appropriate actions.

Tanaya (1995) and Kruth (1996) present a holonic NC controller. It is an open NC-controller, based on the device driver concept (Valckenaers, 1993), but extended to allow multilevel and in-process reaction to disturbances. Tanaya therefore incorporates a hierarchy of instructions, and analyses the causes and costs of an interruption to take the appropriate decision.

Copyright 1996, Katholieke Universiteit Leuven
Page design: Jo Wyns
Information provider: PMA Division KULeuven
Comments for the authors: Jo Wyns
http://www.mech.kuleuven.ac.be/pma/project/goa/procctrl.htm