Computeraided Hazard Identification

1
Computer-aided Hazard Identification

• Paul Chung (p.w.h.chung_at_lboro.ac.uk)
• Department of Computer Science

2
Hazards and Operability (HAZOP) Studies

• Established and widely used technique in the
  process industry for hazard identification
• Time consuming, labour intensive process
• Tedious
• Expensive

3
Computer-aided HAZOP

• Different levels of support
• Electronic report form
• Electronic data (on plant, on fluids, etc.)
• Automated Hazard Identification
• Continuous operation
• Batch operation

4
Automated Hazard Identification

• Continuous operation
• From basic research to commercial product HAZID
• Basic technology
• Signed directed graph (SDG) representation
• Fault propagation
• Go through a list of deviations systematically
  and identify the faults that cause the deviations
  and the consequences that result from the faults
  and deviations

5
HAZID Overview

• Automated extraction of plant design from a CAD
  system, e.g. Intergraph SmartPlant PID
• Convenient forms for adding any missing process
  specific information
• Tick boxes for selecting analysis options
• Deviations, e.g. more flow, less flow, etc.
• Items to HAZOP, etc.

6
HAZID Overview

• HAZOP style output in different output formats
• XML, with HTML web page view.
• Excel spreadsheet.
• Query facility for viewing analysis results
• e.g. viewing faults and consequences relating to
  a particular plant item
• e.g. viewing the propagation path between a
  particular fault and consequence
• Compare facility for viewing the difference
  between two HAZID runs
• Useful for after making a change to the design

7
7
Hazid Operation
Select SP PID From SmartPlant
Hazid maps SP icons to Hazop Process models
Run Hazid data wizard, Extract plant data from SP
database Plant Items Piping, valves,
fittings Controls Fluids information Temperatu
re Pressure Fluid name
Converts PI diagram to Analysis model
Hazid analyses plant and generates Hazop Report
8
7
Hazid Operation
Select SP PID From SmartPlant
Hazid maps SP icons to Hazop Process models
Run Hazid data wizard, Extract plant data from SP
database Plant Items Piping, valves,
fittings Controls Fluids information Temperatu
re Pressure Fluid name
Converts PI diagram to Analysis model
Hazid analyses plant and generates Hazop Report
9
(No Transcript)
10
7
Hazid Operation
Select SP PID From SmartPlant
Hazid maps SP icons to Hazop Process models
Run Hazid data wizard, Extract plant data from SP
database Plant Items Piping, valves,
fittings Controls Fluids information Temperatu
re Pressure Fluid name
Converts PI diagram to Analysis model
Hazid analyses plant and generates Hazop Report
11
(No Transcript)
12
7
Hazid Operation
Select SP PID From SmartPlant
Hazid maps SP icons to Hazop Process models
Run Hazid data wizard, Extract plant data from SP
database Plant Items Piping, valves,
fittings Controls Fluids information Temperatu
re Pressure Fluid name
Converts PI diagram to Analysis model
Hazid analyses plant and generates Hazop Report
13
8
Mapping SmartPlant to Hazid models
SmartPlant Database References
Hazid Model Types
Mapping created by user for all company, then
mapping is gt95 automatic
Automatic mapping by Hazid

Centrifugal Pump

Valve

????? Mixer Stripper Absorber Reactor
User makes choice
This User mapping is required only once for the
whole Project
14
9
Checking Automatic Nozzle Mapping
Hazid model is process function
But PID is piping
Vapour Out Port
Vapour/Liquid In Port
Vapour
Liquid
Liquid Out Port
User confirms nozzle functions N1 is
vapour/liquid in port, N2 and N3 are liquid out
ports, N4 is a vapour out port N5 is Cleanout
liquid in port.
15
(No Transcript)
16
(No Transcript)
17
7
Hazid Operation
Select SP PID From SmartPlant
Hazid maps SP icons to Hazop Process models
Run Hazid data wizard, Extract plant data from SP
database Plant Items Piping, valves,
fittings Controls Fluids information Temperatu
re Pressure Fluid name
Converts PI diagram to Analysis model
Hazid analyses plant and generates Hazop Report
18
Equipment Knowledge Base
Knowledge about behaviour of equipment
Interaction between Faults, Deviations and
Consequences
These links are called arcs in Hazid
Consequence
Deviation
Fault
Fault - can cause Consequence Fault
- can cause Deviation to process
variable Deviation can cause another
Deviation Deviation can cause Consequence
19
Equipment Knowledge Base
Knowledge about Faults and Consequences
This is mainly engineering knowledge and
experience
Fault Bearing failure

• Consequences
• Casing damage
• Seal damage and leakage
• Loss of discharge pressure
• Flow disturbance
• Pump stops

20
7
Hazid Operation
Select SP PID From SmartPlant
Hazid maps SP icons to Hazop Process models
Run Hazid data wizard, Extract plant data from SP
database Plant Items Piping, valves,
fittings Controls Fluids information Temperatu
re Pressure Fluid name
Converts PI diagram to Analysis model
Hazid analyses plant and generates Hazop Report
21
(No Transcript)
22
HAZOP Style Output from HAZID
23
HAZID Viewer Queries

• Standard format questions
• What causes could there be for a selected hazard?
• What consequences are there for a given failure
  mode?
• How is a given hazard realised?
• Show a path of deviations for propagation
• Display all hazards with a given severity rank or
  higher

24
(No Transcript)
25
Filtered Output showing Fault Paths for a given
Hazard
26
Automated Batch Plant Hazard Identification

• Batch operation
• Early research prototype (CHECKOP)
• Basic technology
• Action representation
• State-based simulation
• Go through a set of operation instructions
  systematically and identify potential
  ambiguities, operating problems and hazards
• Applying guidewords such as before, after, no
  action, etc.

27
CHECKOP System Overview
28
Simple Batch Reactor Example
29
CHECKOP Plant Description

• instance(tank101 isa tank, content info
  reactantA, outports info out is
  pump101,in).
• instance(pump101 isa pump, status is offline,
  outports info out is valve101,in).
• instance(valve101 isa valve, status is closed,
  outports info out is reactor101, in2).
• instance(reactor101 isa stirred_tank_reactor,
• outports info out1 is valve103,in, out2 is
  valve106,in,
• heatSink info hout is jacket101,hin,
• reaction info reaction_ab_p
• ).

30
Operating instruction format

• Natural language
• Easy for user
• Requires natural language processing
• Could be ambiguous
• Structured template
• Easy for computer to process
• Limited expressive power

31
Operating instruction format

• Object Action
• valve101 open
• Object Action until Condition
• mixer on until elapsed-time 20 minute
• Object1 Action Object2 Filler-word Fluid until
  Condition
• reactor101 fill-from tank101 with reactantA until
  volume 30 percent

32
CHECKOP Operating Procedure Input

• charge reactor101 with reactantA
• (1) valve101 open
• (2) pump101 start
• (3) reactor101 fill_from tank101 with
  reactantA until volume 30 percent
• (4) pump101 stop
• (5) valve101 close
• etc

33
Deviation Generation

• Combine each single action in the procedure with
  guide words, from
• No action Simple omission.
• Early/Late action Sequence of procedure changed
  (how many steps feasible?).
• Early/Late action termination until condition
  of action varied.
• Then, simulate the effect of executing the new
  procedure on the plant, detect hazards.

34
Batch HAZOP Output
35
Batch HAZOP Output
36
Future Work on CHECKOP

• A formal structured language for operating
  instructions
• More guide words
• Other Action change other variables of the
  Action model
• Rules for reasonable deviations
• What are the most likely mistakes in operation?
• Integrate with HAZID
• HAZID is strong on process hazards, CHECKOP
  better for operating errors, etc.
• HAZOP of start-up, shutdown, maintenance, etc.

37
Conclusions

• Automated hazard identification
• continuous operation
• commercial tool that can reduce the time of HAZOP
• batch operation
• promising area of research and development for
  identifying problems associated with human errors
  and operating procedures
• Benefits
• Doing HAZOP earlier, and on modifications.
• Consistent, repeatable, complete hazard
  identification method.

38
Acknowledgements

• The work described in this presentation is being
  funded by
• HAZID Technologies Ltd
• Engineering and Physical Sciences Research
  Council, UK
• Thanks are also due to my colleagues
• Dr Steve McCoy
• Mr Dingfeng Zhou