A hierarchical model-based intelligent systems framework for synthesis of safe batch operating procedures

Shankar Viswanathan, Purdue University

Abstract

This dissertation proposes a two-tier methodology for automating the synthesis of safe operating procedures for batch processes. The first tier combines Grafcet, a discrete event modeling with hierarchical planning technique to generate nominal operating procedures. Starting with process specific knowledge including the process description and plant information, represented using an object-oriented methodology, the nominal operating procedures are incrementally generated over four stages of planning. The stages involved are path search & equipment assignment, preliminary sequencing, final sequencing, and operating procedure synthesis. The latter three use process generic information, that models the typical operations of a batch process, using Grafcet, to generate the nominal procedures. Each stage adds a layer of batch process operation information and these are also represented using Grafcet constructs. In the second tier, the operation information from the first tier is converted to a two level petri-net based representation. Combining these with digraph-based qualitative causal models, qualitative hazard analysis is performed using an automated methodology for HAZOP (Hazard and Operability Analysis). The HAZOP results are appended to the nominal operating procedures to provide safety-related insights for each operation. Then the critical scenarios from the results are identified and rigorous dynamic math models are formulated for these. Using these models, an optimization problem is developed that identifies alternative operating policies that prevent the occurrence of the scenario or mitigate its effects. The operating procedures that are finally provided to the operator include the nominal operating procedures, suitably modified based on the hybrid qualitative and quantitative analysis of the second tier. This two-tier methodology has been prototyped and successfully tested on several real-life industrial case studies using a set of software tools. iTOPS, an Intelligent Tool for Operating Procedure Synthesis, is an implementation of the knowledge representation and hierarchical planning strategies of the nominal operating procedure synthesis framework of tier I. The second tier is implemented using BatchHAZOPexpert for the qualitative HAZOP analysis component and a combination of gPROMS, a dynamic simulator and gOPT, an optimization utility for the quantitative modeling/optimization component.

Degree

Ph.D.

Advisors

Venkatasubramanian, Purdue University.

Subject Area

Chemical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS