The thesis analyzes the problem of planning industrial product-oriented projects and specifies a set of modeling constructs to support the planning process. In an industrial project where the specific goal is to create a (physical) product, the structure of the project plan depends strongly (1) on the structure of the product and (2) on the way the product information is organized. The project plan must be built out of the services that the resources inside or outside the company can provide. Planning is the task of composing a configuration of available services - that is, a set of services and their mutual relationships - whose execution will result in the creation of the product. A plan is needed primarily to provide operational guidance during the execution of a project and secondly to support the estimation of performance properties of the project.
The execution of a project is characterized by continuous change. The information about the goal of the project, the current state of the execution, and the available services evolves all the way during a project. Some of the changes are planned, (e.g., those caused by design and information gathering activities) and some are spontaneous (e.g., exceptions, failures, or revised customer requirements). The plan should continuously maintain information on how to get from the current state to the goal state, based on the latest information that has been gathered. If the plan is not updated when changes occur, the quality of the operational guidance deteriorates and the interpretation of the plan becomes unsystematic. As a result, ad hoc planning decisions are required at the time of the execution when there is less freedom to tackle with problems. This thesis deals with the problem of maintaining a plan in a dynamic environment so that appropriate operational guidance would be continuously available.
The model developed in the thesis is specified in object-oriented way, using the mechanisms of constraint-based reasoning to specify the dependencies between properties of objects. The model includes a generic mechanism to represent the objects in multiple different worlds. This can support the analysis of the differences between the current state and the goal state of a project, the differences between the knowledge of different agents and the differences between alternative courses of action. A knowledge base built using the model could support simple reasoning to recognize conflicting sets of constraints.
The model describes the relationship of information objects (e.g., reports, drawings, and schedules) to product objects. Moreover, it provides a generic mechanism for the representation of structural objects through their compositional structure and the interactions among their parts. In addition, it includes a representation of the activities as transformations between the states of objects. Resources are regarded as pools of services, i.e., abstract activity specifications. The nature of the project enterprise - the virtual enterprise of a project defined by the contractual relationships - is analyzed. Its structure is described by the control relationships between the participating agents.
The dynamic influences within a planning process are analyzed. Planning is defined as decision making about future action. The decisions that a planner has to make depend on the possible executions that can happen in the executing organization. Additional constraints for the planning process are created by the properties of the command process, the observation process and the rationality requirements posed to plans. Finally a summary of the model is presented and its use is illustrated with the aid of an example.
Full text : PDF