By

The design of an information system for an enterprise requires a complete understanding of the various enterprise operations, the information needs, and dynamics. Such an understanding is promoted by formal enterprise models covering the above facets of the enterprise. In addition to the design of information systems, enterprise models should be capable of facilitating functions such as business process reengineering and prediction of enterprise performance.

The model representing the operations of an enterprise tends to be large and complex; this model should also be continuously evolving to reflect the dynamic nature of the enterprise. To handle the size, complexity and evolution, the model should have the ability to scale-up and deal with changes in the enterprise. It should also use formalisms that are easy to learn, use and teach so that enterprise personnel themselves can create, enhance and maintain the model. A framework for enterprise modeling should facilitate development of a model fulfilling these requirements.

Enterprise Modeling Framework (EMF), discussed in this dissertation, consists of a methodology for modeling the three major facets of an enterprise, viz., function, information and dynamics, and software tools implementing the methodology. These facets of an enterprise have been broadly structured using the object-oriented paradigm into three EMF models, viz., the Activity Model, the Entity Model and the Knowledge Model. The EW Activity Model is a representation of the various functions performed in operating a manufacturing enterprise. The activities in this model are structured into aggregation, specialization and precedence hierarchies. The first two hierarchies are user-defined whereas EMF derives the relationships for the third. There are several types of entities in a manufacturing enterprise: operators, equipment, facilities, plans, orders and design are but a few examples of this rich set. The EMT Entity Model represents these entities using an object-oriented data model. The model consists of a specialization hierarchy of entity classes. The Knowledge Model comprises the knowledge and heuristics employed for carrying out different functions in a manufacturing enterprise. This knowledge can be maintained in the form of separate knowledge bases or can be integrated with

the EMF models in the form of constraints and triggers associated with various entities and their attributes. EMF is implemented in Common Lisp.

There is a need for combining structured analysis (based on functional decomposition) and object-oriented analysis for the following reasons: (i) enterprises are usually organized based on the functions (e.g., design, manufacture and marketing); and (ii) lack of isomorphism between function and object hierarchies. Based on the experience gained during modeling real world enterprises, a method for combining structured, and object-oriented analysis methods has been adopted.

A methodology has been developed for integrating constraints (assertions and Event-Condition-Action (ECA) rules) with the class hierarchy of the EMF Entity Model. The assertions of the class of an entity and all its generic classes are checked in the most generic class first order. Hence, entities of a more specific class must conform to the constraints imposed by the generic class. The action part of the ECA rules attached to all the classes in a hierarchy are carried out in decreasing order of genericity. Hence, instances of specific classes display all the dynamic behavior of their generic class and possibly expand upon it. EMF provides a form based user-interface for specifying the constraints at the conceptual level. The rules, in addition to maintaining integrity of the model, can be used to specify and simulate the dynamics of an enterprise.

EMF also allows the user to automatically generate SIMAN models from the information already present in the Activity and the Entity Models. The EMF approach to modeling the dynamics of an enterprise has several merits. It facilitates modeling by manufacturing personnel using higher level concepts through a form-filling interface. Dynamics modeling has been integrated with function and information modeling. This helps to avoid redundancy and inconsistency between models. The EMF graphical browsers help manufacturing personnel determine the completeness and correctness of the model. The hierarchical nature of the EMF Activity Model is used to generate simulation models at the desired level of detail.

The role of EMF in developing and implementing an information system in an apparel enterprise has been demonstrated. The integrated approach to creating a simulation model from activity and entity descriptions has also been illustrated. This model has been used for estimating the resource requirements and evaluating the shopfloor performance.

Based on the principal performance metrics for evaluating a modeling methodology, viz., learning time, model development time, workload on the user and completeness in terms of enterprise facets modeled, EMF represents a significant advancement over existing enterprise modeling methodologies. Recommendations have been made for enhancing the features of EMF and exploring wider applications for EMF.

Back