~

Computers and Chemical Engineering Supplement (1999) S747-S750

Pergamon e 1999 Elsevier Science Ltd. All rights reserved
PII: S0098· 1354/99/00200-8

Integrating Operating Procedure Synthesis

and Hazards Analysis Automation Tools for
Batch Processes
Shankar Viswanathan, Jinsong Zhao and Venkat Venkatsubramanian
Laboratory of Intelligent Process Systems, School of Chemical Engineering
Purdue University
Linas Mockus, Jonathan Vinson, Al Noren, Prabir K. Basu
Searle, a Monsanto Company, Skokie, IL
Abstract:

In this paper we address the problem of integrating two systems iTOPS, an InteIIigent Tool for Operating Procedure
Synthesis and BHE, BatchHAZOPexpert that are respective used for generating operator instructions and
performing process hazards analysis for batch processes. The architecture of the integrated system is described and
its application to an industrial case study is presented.

an operation like charge a reactor or heat a reactor. A

INTRODUCTION group of operations constitute a unit procedure like
Several activities need to be performed prior to start of reaction or distiIIation and a set of unit procedures make
production in a batch processing facility. These include up a procedure. Two systems, InteIIigent Tool for
operating procedure synthesis (OPS), process hazards Operating Procedure Synthesis(iTOPS) (Viswanathan
analysis (PHA), scheduling, and generation of batch et. al, 1998) and BatchHAZOPExpert (BHE)
control code. OPS generates the sequence of operations (Srinivasan and Venkatasubramanian, 1998), that use
and their corresponding instructions, that an operator such a process-oriented approach, for modelling batch
performs to manage a batch process. PHA determines process information. have been developed to automate
the hazards that could occur when performing this OPS and PHA respectively. The advantage of
sequence of operations and devcIops safety integrating these two systems is that based on BHE's
recommendations for the process based on this analysis. PHA the operating procedures synthesized by iTOPS
Scheduling optimally aIIocates resources and equipment can be reworked to incorporate the safety critical
for each operation in the sequence generated by OPS information and appropriate recommendations into
and batch control code generation implements the them. To integrate iTOPS and BHE into one unified
sequence in a procedural control system. All these system it is necessary to pursue two goals - resolve the
involve considerable amount of time, effort and different "views" that underpin the models in the
expertise from plant personnel which make automation individual techniques and provide user interface and
of these tasks very desirable and important. Significant database support to allow for flow of information across
progress has been made recently in techniques and the two applications.
software for automating each of these individual The basic framework for iTOPS uses grafchart-based
functions. concepts to model the process at the four levels
Integration of these software tools is the next step in this identified in S88. It uses a hierarchical planning
direction and wiII lead to use of consistent process and technique to synthesize the operating procedures.
plant information across these activities thereby iTOPS starts with information about the process
eliminating redundancy. In this research effort we materials, equipment, and chemistry to synthesize the
address the issues involved in integrating two software procedure, unit procedures, operations, and phases for
tools geared towards OPS and PHA into one unified the process. The operating procedures are then
system and present results of using such an integrated generated from the phases.
system on a practical industrial case study. The basic framework for BHE uses a two-layered petri-
SYSTEM FRAMEWORK AND ARCHITECTURE net (PN) based model of the process. The top layer
caIIed the recipe petri net includes information at the
A process-oriented approach is adopted in our level of unit procedures and the bottom level referred to
framework because, as specified in the S88 batch as the task petri net consists of information at the level
standards, the information required for the various of operations. In addition to the two-layered process
functions such as OPS and PHA should be organized at model BHE also requires information about the process
four process-based levels - phases, operations, unit materials, equipment, and chemistry. Using this process
procedures. and procedure. Phases are activities such as information coupled with a signed digraph (SDG) based
open valve, start agitator. etc. A set of phases makes up
S748 Computers and Chemical Engineering Supplem ent (1999) S747-S750

causal model BHE performs hazard and operability information is stored in the inferred knowledge
(HAZOP) analysis . component of the iTOPS knowledge base.
In the integrated system it is envisioned that the user At this stage the process information translator and the
inputs the information about the process - the materials. product recipe generator in the iTOPS-BHE interface
the chemistry, and the equipment through the iTOPS are used to generate the process specific knowledge and
interface. iTOPS then uses this information to the product recipe information required by the BHE
synthesize the top three process levels - procedure, unit knowledge base using the declarative knowledge and
procedures, and operations. The unit procedures and the generated inferred knowledge in iTOPS.
operations are then converted through an interface from
a grafchart-based to a petri net-based representation to Once the interface has completed the translation, iTOPS
be used in BHE. BHE then performs HAZOP using this and BHE complete their respective tasks. iTOPS
generates the phases and the nominal operating
r--;::====;---,
IiTOPSKnowledge Bose I I ;TOPSInforen« Engine I
Procedural
I ITOPSInterface I Knowledge
Gnfctw1 Procedures
fotUllitProC"cdum.,. In(cmd
ath Searrh
Equipment
Dialog. menu-based, ~lioltJand Knowlc:dgc Assignmentand
Plu>es Preliminary
and grapbic3lcdllors Unil
Proctdurn(Block Sequencing
!'SD).
pPcnlionJ.{PSD).
~MOR).lnJ ( FtnalSequencing I
9Pcrnin,
PrOcedures Oper:uing
Procedure
Synrbcsls

Process-SpeciflC Product Recipe

Dial0i:. nxnu ·bascd. KDO\~ik'dgl: Information
ond &Olphical cduo"
('rcnu MacerWJ.
EquipmellC. OwmU'J ~;r:::~~&

Process-Generic ]
Know1cdge
r Sut:w.tDiiTlrt,J

I BHE Know1<dge Base I II BHE InferenceEngine I

Figure t: Architecture of Integrated System

information. Meanwhile iTOPS synthesizes the phases. procedures which is stored as part of iTOPS' inferred
After BHE has performed HAZOP analysis and iTOPS knowledge.
has synthesized the phases the details as to which
BHE's inference engine determines the potential
processing steps are safety critical are transferred from
hazards in the process, using the translated product
BHE through the interface back to iTOPS. Safety
recipe information, process specific information, and
instructions corresponding to these are then created and
process generic subtask digraph models .
incorporated along with the basic instructions generated
by iTOPS in the final batch record document provided The integrated system was implemented in the G2
to the operator. environment. The system architecture ensures that
consistent process information is used across the
The architecture of the integrated system that was
applications while ensuring minimal intervention
developed based on the above outlined concept is
between applications.
shown in Figure 1. It has three parts:
CASE STUDY
• iTOPS components
The integrated system was tested on 2 case studies
• iTOPS-BHE interface provided by Searle. The details of the how the system
• BHE components was used to perform OPS and HAZOP for one of the
case studies is described below.
The iTOPS and BHE components each include an user
interface, a knowledge base, and an inference engine The process under consideration is an intermediate step
The iTOPS user interface allows the user to edit the during manufacture of a drug. There are 8 materials and
declarative knowledge or the process specific 5 process equipment that are involved in the process.
information which is stored in the iTOPS knowledge The chemistry of the process consists of a reaction
base in an object-oriented fashion.The iTOPS inference followed by filtration and drying steps. The
engine uses the declarative knowledge and the specification of the the material, equipment and
procedural knowledge (modeled using grafchart chemistry information in iTOPS is shown in Figure l.
procedures) in a hierarchical planning strategy to The iTOPS inference engine based on the material,
generate the unit procedures and operations. This equipment and chemistry information generates the
sequence of operations. also called the process sequence
 Computers and Chemical Engineering Supplement (1999) S747-S750 S749

diagram or PSD for the process. The PSD for this Currently we are developing a common model manager
particular case study, consisting of 29 operations is that links to external databases and provides the
shown in Figure 3. required process specific information to both iTOPS
and BHE.
The iTOPS-BHEinterface then utilises the generated
PSD to create the product recipe informationfor BHE. REFERENCES
The generation of the BHE product recipe information
Viswanathan, S., Johnsson, C., Srinivasan, R.,
based on the iTOPS PSD is shown in Figures 4 and 5.
Venkatasubrarnanian, V., and Arzen, K., (1998)
Once the iTOPS-BHE interface completes the Automatingoperating procedure synthesis for batch
translation iTOPS generates the operating procedures processes - II Implementation and Application,
also referred to as the Master Operations Record or Computers and Chemical Engineering. 22(11) 1686·
MOR. The MOR for this process is shown in Figure 6. 1698
The BHE system uses the translated process specific
Srinivasan,R., Venkatasubramanian, V.,Automating
and product recipe information to perform HAZOP
HAZOP analysis of batch chemical plants: Part II
analysis as shown in Figure 7.
Algorithms and Application, Computers and Chemical
CONCLUSIONS AND FUTURE WORK Engineering. 22(9) 1357- 1370
The basic.architectureof a system that performs OPS
and PHA for a batch process was presented and its use
in an industrial case study described. The system had
the advantage of ensuring use of consistent information
while preserving the primary structure of the individual
components.

r~"' 1

:..;-;: [iJ"
1- :"" 1

..... 1Wa.ul
I I

Figure 3: G:nerated PSD in iTa'S

K·.'Ol

~
~

Figure4: GeneratedRecipePetri Net for BHEfromiTOPSPSD

5750 Computers and Chemical Engineering Supplement (/999) S747-S750

)(· 303

FigureS: Generating Task Petri Nets for BHE from iTOPS PSD

... - ~
,-
-- ....
......... - .-
tolMTt R ort".AA't'IOf'CS R[Q)1tD

S.•" o ~ """"".... Itt-

I ~~rth1' ~l Il
I FJIlI1R )(.)()] A a.,
2
I
) I itdct IlODWp.((J)IA (Dn""" "'d ·", rl K· lOl

•
~
R.e~ 10DWP-lOJIB
Refer10DWP«l)1 0
ScIJD K ·:.o),lnJ lC·~

0-", kJoIA IOKo»)

Rtter10DWP«Ql Prtulft lV"f K·103

figure 7: Perfcrming HAZOP in BHE

Figure6: Generating MORin ITOPS