Next: Mathematical Ontology Up: The PHYSSYS Ontology Previous: Components

## Process Ontology

Our physical process ontology specifies the behavioural view on physical systems. In the general case it is quite difficult to formalize what the notion of a dynamic process precisely entails. Fortunately, for a certain part of physics this has been done to a level where one can define really primitive process concepts. The approach we take here is known in engineering as system dynamics theory, which also forms the theoretical background of the bond graph method [Karnopp et al., 1990]. The basic idea behind this theory is that the dynamics of a system can always be captured by looking at the change of different kinds of stuff. This change of stuff is also called flow. For instance, in electrical systems, dynamic behaviour consists of the change of electrical charge, i.e. electrical current. Likewise, in the mechanical domain the stuff is called location and change of location is velocity. The thing required to bring about a flow is called effort. Table 1 lists the types of stuff, flow and effort of some of the physical domains defined in the ontology.

Table 1: Some examples of physical domains. In each domain, dynamic behaviour is described as flow, i.e. change of stuff. Effort is that what is required to bring about a flow.

The interesting aspect about this table is that the product of a flow with its related effort has the dimension energy / time, i.e. such a pair defines an energy flow. Physical behaviour can therefore be defined in terms of energy flows. The process ontology introduces physical mechanisms which are applications of physical laws or principles to one or more energy flows. An important feature of these mechanisms is that they exploit in detail the analogies that exist between different physical domains. For example, the principle of conservation of momentum in mechanics is analogous to induction in the electrical domain. Many more of these analogies exist. This approach is valid for standard classical, deterministic physics, covering such diverse fields as mechanics, electricity and magnetism, hydraulics, acoustics, and thermodynamics.

Figure 7: Excerpt from the process ontology. This ontology formalizes a large part of physics. It defines how process descriptions can be formed by making a network of domain-independent mechanisms and energy flows. The ontology is relatively simple because systems theory is used for the definition of the networks.

Complex process descriptions can be formed by making a network of mechanisms, linked by energy flows. This abstraction is used to construct the process ontology. The process ontology includes systems theory and specializes the system theoretic concepts to processes. Just like the component ontology, the process ontology defines relatively simple concepts and relations onto which the system ontology is projected. This can be seen in Figure 7. Mechanisms are defined as simple mereological individuals. Energy flows, which have a certain direction, flow from one mechanism to another. The projection is performed by stating that an energy flow is a topological connection that connects the two mechanisms. A process description can now simply be defined as a system of mechanisms. The definition of physical domains as depicted in Table 1 is not shown in Figure 7.

Figure 8: The taxonomy of physical mechanisms. The properties discriminating between the classes after branching are printed above the branch points. The classes on the right give some examples of mechanisms in the electrical and mechanical domain.

Figure 8 shows the taxonomy of mechanisms as defined in the process ontology. The classes in the figure are present as classes in the ontology and class-subclass relations are defined for every line in the figure. The discriminating properties used to construct this taxonomy are the number of energy flows linked by a mechanism (connectivity), the mechanism type, whether effort or flow plays the most important role with respect to the mechanism type and the physical domain (e.g. mechanics, electricity, hydraulics, thermodynamics). Note that some discriminating properties are not useful for certain types of mechanisms.

The order in which the discriminating properties are applied here is the opposite of the order used in the typical engineering education. There, the distinction between physical domains is made first: there are separate courses in mechanics, electrical engineering and thermodynamics. Only when students have mastered all courses, they are able to see the analogies between the domains that makes the process ontology as compact and elegant as it is.

Next: Mathematical Ontology Up: The PHYSSYS Ontology Previous: Components

Pim Borst
Fri Sep 27 13:28:43 MET DST 1996