Modeling and Analysis of Workflows Using Petri Nets

  • Authors:
  • Nabil R. Adam;Vijayalakshmi Atluri;Wei-Kuang Huang

  • Affiliations:
  • Center for Information Management, Integration and Connectivity (CIMIC) and MS/IS Department, Rutgers University, 180 University Ave., Newark, NJ 07102.;Center for Information Management, Integration and Connectivity (CIMIC) and MS/IS Department, Rutgers University, 180 University Ave., Newark, NJ 07102.;Center for Information Management, Integration and Connectivity (CIMIC) and MS/IS Department, Rutgers University, 180 University Ave., Newark, NJ 07102.

  • Venue:
  • Journal of Intelligent Information Systems - Special issue on workflow management systems
  • Year:
  • 1998

Quantified Score

Hi-index 0.00

Visualization

Abstract

A workflow system, in its general form, is basically a heterogeneousand distributed information system where the tasks are performed usingautonomous systems. Resources, such as databases, labor, etc. are typicallyrequired to process these tasks. Prerequisite to the execution of a task isa set of constraints that reflect the applicable business rules and userrequirements.In this paper we present a Petri Net (PN) based framework that (1)facilitates specification of workflow applications, (2) serves as a powerfultool for modeling the system under study at a conceptual level, (3) allowsfor a smooth transition from the conceptual level to a testbedimplementation and (4) enables the analysis, simulation and validation ofthe system under study before proceeding to implementation. Specifically, weconsider three categories of task dependencies: control flow, value andexternal (temporal).We identify several structural properties of PN and demonstrate theiruse for conducting the following type of analyses: (1) identify inconsistentdependency specifications among tasks; (2) test for workflow safety, i.e.test whether the workflow terminates in an acceptable state; (3) for a givenstarting time, test whether it is feasible to execute a workflow with thespecified temporal constraints. We also provide an implementation forconducting the above analyses.