DuPont Activity Modeling Tool

Overview - Screen Shots

Presentation given on October 11, 2000 at
IEEE International Systems, Man, and Cybernetics (SMC 2000) Conference.

Purpose

There exists a need to glue together complex, heterogeneous systems of data acquisition, processing, and presentation.  Information within one system should be used in other systems.  Configuring these interconnections on a system-to-system basis is not efficient.  Finding a means to map these interfaces onto a standardized pipeline guarantees flexibility in the advent of unforeseen interconnection possibilities or the addition or subtraction of new systems.  Model Integrated Computing (MIC) helps creating systems where an extensible linking of components occurs through system modeling.  We want to solve our domain-specific issues of system integration and synthesis in exploiting MIC techniques.  The resulting MIC environment should allow for extreme ease of target reconfiguration and expose the least possible number of details associated with whatever target is synthesized.  The target is a run-time network consisting of data acquisition, processing, and presentation systems; these form complex activities.

Method

A core signal flow paradigm has been created and utilized for data distribution in the context of both modeling and run-time environments.  A modeling environment has been chosen allowing for the creation of models that govern the characteristics of the signal flow network, and the data acquisition, processing, and presentation systems.  The creation of these models allows for the specification of target applications.  Information relevant to each of the system interfaces must be collected and migrated into a form capable of generic interconnection.  Upon modeling paradigm creation, model specification allows for complete configuration and synthesis of environments satisfying the system integration requirements at hand.

Results

A prototype has been installed and is being evaluated at the DuPont Chemical Corporation’s Old Hickory DMT plant.  This prototype includes a modeling paradigm inspired through needs of interconnecting a Process Monitoring and Control (PM&C) Variable Database, chemical plant simulation software, remote graphical user interfaces, and processing nodes.  This paradigm results in the creation of a target environment application that assists operators in making informed decisions based on temporal plant conditions.  System modelers familiar with the domain produce such environments that are rapidly created and migratable to serve current plant operational needs.

Breakthrough Aspect of Work

A high level approach and framework has been specified for which environments consisting of complex, heterogeneous systems can quickly be created, modified, and extended.  The modeling interface remains unchanged regardless of the complexity or multiplicity of the underlying entities connected to the signal flow network.  Modeled networked presentation interfaces are instantiated on demand in response to the needs of the signal flow network.  Environment instances that independently would require excessive amounts of development time can be created through this common framework.  Individuals capable of such environment creation need only have a background consistent with the specific domain, instead of being forced to acquire new integration skills.  These new skills previously crucial in integrating entities are unrelated to the understanding of the functionality of such components, and unrelated to the solutions that such system integration problems solve.

Summary

A system integration tool has been developed allowing for specifying how different systems interact with one another.  These activities have their characteristics represented in a model.  The efforts of creating models to represent the environment design and the synthesis of the target environment yield results far superior to “manual” integration strategies.  Furthermore, frequent modifications that are inherent to our specific domain of interest are trivial.  The individual designing the environments can focus on the semantics of the constituent systems rather than their interconnection syntax.  This is due to the efforts placed in moving the difficulties associated with system integration into capturing environment characteristics and using this information to synthesize a target environment.