Process Control in Huntsman Petrochemicals

1
Process Control in Huntsman Petrochemicals

• Mike Oglesby

2
Improving the Assets

• The plants purchased from ICI had control systems
  nearing the end of their useful life
• Spares becoming difficult to get
• Increased failure rate of control room equipment
• Control rooms dont meet current safety standards

3
Replacing the Control System

• Existing heritage systems are control panel
  based pneumatic or electronic
• New systems are computer based Distributed
  Control Systems (DCS)
• Control Systems are very expensive
• Their replacement has to be justified on
  financial grounds

4
Justifying the Replacement

• Replacing a control system with one having the
  same functionality gives very little quantifiable
  benefit
• Big benefits come from exploiting the improved
  functionality of the DCS system
• For a viable project we need to combine the DCS
  with advanced process control (APC)
• DCS costs 80 of total cost gives 20 of
  benefit

5
Identifying the Areas for Improved Control

• Define plant objective
• Maximum production rate of on-spec. material
• Translate to local objectives for unit operations
• Run column at maximum reflux subject to
  constraints
• Constraints are
• Flooding, Reboiler limits, condenser limits

6
Estimating the benefits

• Use historic data
• How well did the existing control system
  operator manage to achieve the objective on
  average?
• What was the best ever achieved performance?
• What benefits would result if the best ever was
  achieved all the time?

7
Achieving the Benefits

• Approach followed depends on the plant and the
  Control opportunities identified
• Approaches can be very different
• Olefins plant has used on-line computer control
  for over 30 years plant very well understood
• 30 year old Paraxylene plant uses same control
  philosophy as when built interactions between
  areas of plant not fully understood

8
Olefins Plant

• Full plant on-line optimiser at the top level
• Sends down control strategy objectives to next
  level
• Runs every couple of hours
• Model Based Predictive Controllers
• Maintain control strategy on minute to minute
  basis between updates from optimiser
• Send setpoints down to PID loops at bottom of
  chain
• 23 Model Based Controllers co-ordinated
  operation
• Very large schemes 41 by 72 multivariable
  controller

9
Data Flow Diagram
Set Points
Plant data
10
Paraxylene Plant

• Benefits study here identified over 50 APC
  schemes
• Many based on running plant up against
  constraints
• Much simpler schemes than on Olefins plant
• Largest is 8 by 24 with little dynamic
  interaction
• Decision made to use PID based technology to
  achieve benefits

11
Implementing the APC schemes

• Stage 1 is to cut over from existing Control
  System to the DCS
• Operators have to get used to operating plant
  using new equipment
• To ease this, the DCS uses exactly the same
  control system as the old system
• Stage 2 is to install commission the new schemes

12
Project Plan

• DCS installation complete
• Early June
• Plant Available for APC implementation
• Early July
• APC implementation to be completed Plant
  Maintenance Shutdown Starts
• Mid September
• Project has to be Fast Track

13
APC Scheme Implementation

• Build scheme using standard DCS building blocks
• Test Offline
• Ensure there are no holes
• Ensure correct response on signal failure, etc.
• Operator Training Validation
• There are 6 Shifts each with 5 people to train on
  about 50 different schemes
• Scheme Tuning
• Needs plant testing to define process gains
  dynamics

14
Speeding up the Project

• Access to the plant is restricted due to fast
  track nature of whole project
• Need to do as much as possible off-line
• Interface Models to the APC schemes
• Simple models based on 1st order lag Dead Time
• Models based on dynamic chemical engineering
  models
• Use these models to reduce time on site
• Use plant tests to match models to plant
• Then use the models to tune APC schemes

15
Distillation Column

• Maximise reflux subject to
• Column Flooding Column PD
• Condenser Limit Pressure controller output
  lt100
• Reboiler Limits
• Fuel Gas Valve lt100
• Furnace Tube Skin Temperature Limit
• Low Stack Oxygen Limit

16
(No Transcript)
17
Commissioning

• All non-APC controller tuning checked
• Plant tested to measure responses of each
  constraint to reflux changes
• PI settings for each constraint controller
  identified
• Switch on with constraint setpoints closed in
• Force each constraint in turn to become active
  and check response
• As scheme proves its ability to control plant
  well move constraints out

18
Experience to Date

• All schemes commissioned to plan
• Delivering expected benefits
• Spin-off benefits achieved
• Distillation column with sharp temperature profile

19
Column with Sharp temperature Profile

• Problem for 30 years
• Process Gain at Setpoint VERY high but very low
  just a small distance away from the setpoint
• Difficult / Impossible to Stabilise
• Works well with Gain scheduled PID controller

Height
Control Tray
Setpoint
Temperature
20
(No Transcript)
21
Where Next?

• Paraxylene
• Now parts of the plant are tied down further
  opportunities can start to be seen
• Starting to make use of data collected by DCS to
  understand the process
• Olefins
• Expert system to provide early warning of
  potential plant upset
• Aromatics
• Next plant to have a control system upgrade
• Currently part DCS / part Pneumatic

22
The Future

• Supply Chain Integration
• Build on information platforms
• Provide timely, accurate data to all key Business
  functions
• Off plots Areas
• Movements Automation Tracking
• Benefits
• Could be 10-12 of Turnover