(441b) The Role of Ontologies in Process Modelling
AIChE Annual Meeting
Wednesday, November 6, 2013 - 8:55am to 9:20am
I want to convince the community that thinking and consequently working with ontologies is simply good practice because it imposes a thorough analysis of the problem's structure and thinking carefully through the problem-relevant basics. For process models this is the physics associated with
- the process structure
- the model's granularity
- the conservation principles
- the transport behaviour -- usually an event-dynamic assumption applied to a dynamic transport system
- the kinetics -- the isolated behaviour of the reactants
- the relation between the conserved quantities and the driving forces, the geometry, the material properties
- the assumed equilibria
Process ontologies form well-structured frameworks that build layers on layers, strata on strata defining the basic framework to which the process models must adhere.
Using mechanistic-based ontologies leads to process models that adhere to the basic laws. When used in its basic form, any converged numerical simulation will satisfy the basic balance equations thus providing a level of correctness, which overall is only limited by the validity of the structuring process leading to a certain level of granularity of the the model. Latter can be interpreted as the resolution with which the process is depicted.
The concept extents into the software-internal representation, the way the models are being stored and consequently handled during the process of preparing the computational tasks to be solved in association with a technical problem. The concept of ontologies provides a very useful thinking pattern that may even be considered essential to the design of tools and handling of data in the computational engineering domain.
The objective is always the same: it is the respective basics that are laid down first through a careful analysis of the state problem. Once verified it is applied and results conform resulting instantiated model or software components.