next up previous
Next: 3.4 Diagnosis of Abrupt Up: 3 Bond Graphs for Previous: 3.2 Bond Graphs

3.3 Causality

A causal structure can be imposed on a bond graph model, using local constraint relations among the components associated with a junction.gif A systematic algorithm for causality assignment is the Sequential Causality Assignment Procedure (SCAP) [21]. This algorithm categorizes local constraints as (1) enforced causality, by sources tex2html_wrap_inline980 and tex2html_wrap_inline982 , (2) preferred causality, for energy storage elements, C and I, and (3) indifferent causality for resistances R. The constraints are applied in the order above, with an effort source always imposing an effort (e.g., pressure) causality on a junction and a flow source always imposing a flow (e.g., volume flow) causality on a junction. Opposite assignment of causality to a source indicates a physical incorrectness of the model, e.g., a shorted voltage source. The preferred causality for energy storage elements is for them to operate as integrators as opposed to differentiators. The integral relation establishes natural dependence among the effort or flow variables associated with a bond.gif For example, the implication of the integral form for a C element is that it prefers to deliver effort on a junction, i.e.,

(2) displaymath1031

An I element prefers to enforce flow on a junction, i.e.,

(3) displaymath1033

Resistive elements have no preference of causality, they conform with how they are driven.

Consider the bi-tank system and its bond graph in Fig. 4. Fig. 5 depicts the causality assignment procedure. The first step selects the flow source, tex2html_wrap_inline982 and assigns it flow causality which means it enforces flow on the adjacent 0-junction. By convention, the direction of effort is shown by a perpendicular stroke at one end of a power bond. Since flow is enforced on the 0-junction, effort is enforced on the flow source, depicted by the perpendicular stroke at the beginning of its connecting bond (1) in Fig. 5. Because the flow constraint on a 0-junction requires the sum of all flows to equal zero, the flow values of the two remaining bonds are still undetermined. Therefore, the assigned causality does not propagate any further. In the second step, the capacity tex2html_wrap_inline964 is selected and enforces effort on its adjacent 0-junction depicted by the perpendicular stroke at the end of bond (2). Because a 0-junction represents an equal pressure junction, this enforced pressure propagates across all other connected bonds, (3) and (4). So, tex2html_wrap_inline1008 determines the outflow of tex2html_wrap_inline964 given this pressure. Again, the effort causality on the connecting 1-junction does not propagate any further. In the third step, tex2html_wrap_inline966 is selected and its preferred causality enforces effort on the 0-junction through bond (7) which propagates across the other connected bonds (6) and (8). This results in tex2html_wrap_inline1018 determining the outflow of tex2html_wrap_inline966 , bond (8). By default the remaining bond of the connecting 1-junction (common flow) (5) has to impose effort causality on the connecting resistance, tex2html_wrap_inline1024 , which determines the flow between tex2html_wrap_inline964 and tex2html_wrap_inline966 .

  figure154
Figure 5: Algorithmic assignment of causality.


next up previous
Next: 3.4 Diagnosis of Abrupt Up: 3 Bond Graphs for Previous: 3.2 Bond Graphs

Pieter J. Mosterman
Tue Jul 15 11:26:35 CDT 1997