Technology
Data Integration and Material Flow Data Model:
A Unifying Framework for Production Intelligence
Formulating a data analysis framework for large-scale industrial processes can seem overwhelming, especially in light of the engineering detail involved in the planning, construction and operation of industrial plants. By focusing material flow, the essence of industrial processes, Pattern Discovery Technologies has developed an object-oriented framework, Material Flow Data Modeling (MFDM). MFDM is our paradigm for industrial data analysis.
What is a Material Flow Data Model?
The essence of any industrial process is the transformation of raw material, through a series of identifiable steps, into the desired product. Material Flow Data Modeling focuses on abstracting the elements of an industrial process into Material Transformation Objects and Material Objects.
Material Transformation Objects
For data analysis purposes, Material Transformation Objects are abstractions of a stage within an industrial process. The key elements of that abstraction are outlined in Figure 1
Figure 1: A Material Transformation Object
Material Transformation Objects are characterized by:
• A set of primary and secondary material input(s)
• A set of primary and secondary material output(s)
• A set of measurement points that characterize either the process itself, or the state of material within the transformative step
• A set of control points through which the transformation step is managed by human and/or computer control
• A set of intrinsic attributes to that transformation
Material Transformation Objects can represent any stage of an industrial process to the level of granularity required by the needs of analysis applications. They can also encapsulate any necessary computational logic (e.g. modeling a chemical and/or mechanical process) required by an application.
Material Objects
In addition to the Material Transformation Objects, material flow models explicitly identify and characterize all material used throughout the industrial process. Each material's characterization is defined by the needs of the analysis, and may include mass and volume measurements, chemical composition, temperature, etc
The Material Flow Data Model
A Material Flow Data Model consists of a sequence of connected Material Transformation nodes whose primary Material output(s) from one stage become the primary Material input(s) to the next (see Figure 2). The Material Flow Data Model is defined, to the level of detail required, to represent the entire industrial process in an integrated, holistic manner.
Figure 2: A Material Flow Data Model
Why is the Material Flow Data Model Valuable?
The key value of the Material Flow Data Model is that it formally defines the context and the essential details of the distinct stages within an overall industrial process as well as the materials used throughout. This formal definition is critical to four aspects of Production Intelligence:
1. Communication Between Process Engineering and Application Development
One of the key benefits of a Material Flow Data Model is that is establishes a common vocabulary between the process engineering and software development teams, which in turn leads to better quality and more effective software. The two groups become collaborative in application development. With a common vocabulary, better user interfaces are developed, and computer-generated reports express results in contextual terms that are readily understandable.
2. Formal Application-Based Reasoning
The second benefit of a formal Material Flow Data Model is that it can be encoded in a computer-based representation (e.g. XML) that in turn can be used by analysis software to enforce constraints that reflect physical reality.
3. Improved Data Integrity Audits
The third benefit of a Material Flow Data Model is that it provides the foundation for a rigorous approach to data integrity. As the material flow model is being developed, one naturally begins to ask "How should we characterize the material coming into this stage of the process?" "Is that characterization sufficient, and if not, how could it be improved to provide the insight we genuinely need for our analysis?" Also, time synchronization issues (coordination of multiple independent clocks used in data acquisition) must be addressed at this stage. By asking these types of questions, the combined process engineering/software development teams can identify areas where instrumentation could be improved or human processes refined to collect data at appropriate frequency and accuracy.
4. Improved Data Repository Design
Lastly, the Material Flow Data Model serves as the reference point for the overall design of an integrated data repository that provides the operational foundation for analysis applications.
