Supporting Geotechnical Design with Petri-Net-Based Process Patterns

The paper deals with process patterns based on Coloured Petri-Nets as a process model to coordinate planning in structural engineering, e.g. geotechnical engineering. Coloured Petri-Nets are mathematically well-founded, and are often used to solve problems concerning concurrency in processes in other engineering domains. The paper introduces a new method to represent geotechnical information in Coloured PetriNets, thus enabling a content-based control over design process. Elementary process patterns are introduced, which serve as smallest modules in a bottom-up approach to the more complex process models. C o n st ru ct io n I n fo rm at ic s D ig it al L ib ra ry h tt p :/ /i tc .s ci x. n et the output place. Thus firing a transition transforms the state of the system to the subsequent state. The basic form of a Petri-Net is defined as a tuple PN = (P, T, F, M0), where P is a finite set of places P = {p1, p2, ..., pm} T is a finite set of transitions T = {t1, t2, ..., tn} ( P ∪ T ) = ∅ and ( P ∩ T ) ≠ 0 F is a set of arcs as flow relations F ⊆ (P ×T) ∩ (T × P) M0 is the initial marking M0: P → {0, 1, 2, 3, ... } Petri-nets have a graphical representation as is shown in figure 1. The figure also depicts the connection between Petri-nets and the corresponding elements of the planning process. Fig. 1: Graphical representation of Petri-nets and 2.1 Formal requirements As Petri-nets provide a formal framework, they also provide the possibility to analyze them and to define properties of the net, which are appropriate to the given problem. With respect to the requirements of process modelling of business processes, v.d. Aalst gives a formal definition of Petri-nets as workflow nets [v.d. Aalst 1996, 1997, 1998, 2002], defining as well a soundness criterion. According to it, a workflow net (wf-net) has one single input place I and one single output place O. When the wf-net is short-circuited by a transition t*, it is strongly connected. When the output place O is marked with a token, all other places in the wf-net must be empty, so no other activities can be carried out (no transitions are enabled to fire). This prevents, that a task has erroneously not been carried out, although the process should already be terminated. In this case the process has not been modelled correctly. Figure 2 shows basic routing primitives in Petri-nets. Fig. 2: Basic routing primitives in Petri-nets In civil engineering the design process is partly unknown at the beginning but evolves with the proceeding planning. It is necessary to adapt the underlying process-model. This is achieved by the hierarchical substitution of transitions in the net as shown in figure 3. In the coarse net the transition with the adjacent socket-places is substituted by an underlying, finer net with the port places. Fig. 3: Hierarchy by substitution 2.2 Petri-nets with individual tokens For the representation of information in the Petrinets it is necessary to introduce Coloured Petri-nets, where different tokens are distinguishable. Within the activities the planning information is generated or modified. Depending on the actual planning state the further planning activities have to XOR-Split XOR-Join Iteration AND-Split AND-Join Sequence Token Information Arc A ∈ F Dependency Transition t ∈ T Activity Place p ∈ P State Petri-Net Planning Process