Testing Automation Software With Dynamic Process Models

1

• Testing Automation Software With Dynamic Process
  Models
• Anne Mueller

2
Outline

• Introduction
• Creating the Dynamic Process Model
• Expanding the Model with UniSim Operations
• Connecting the DCS
• Conclusion and Outlook

3
Personal Data

• Anne Mueller
• Studying Process Engineering at the Technical
  University of Dresden, Germany, since 2004
• Project developed during work experience
  placement at Linde-KCA-Dresden GmbH
• Supported by the Department of Process Automation
  by Prof. Dr.-Ing. habil. Wolfgang Kloeden

4
Introduction

• Modern industrial plants are very complex and
  require extensive control schemes
• Control typically provided by a Distributed
  Control System (DCS)
• Dynamic process models allow the accurate
  prediction of plant behaviour for various process
  conditions
• The connection of both leads to a powerful means
  to test the implemented control algorithms of the
  DCS
• In this case, the control system PCS7 by Siemens
  is connected to a USD dynamic model

5
Introduction

• Ready-made solutions to connect a USD process
  model to common DCS already exist for Experion,
  SimC300, SimACE and more
• So far, no interface between UniSim Design and
  PCS7 available
• This work demonstrates the steps required to
  establish communication between the two systems

6
Creating The Dynamic Process Model

• Process CO2-Purification and Liquefaction
• Used to separate CO2 from flue gas
• Operating conditions up to 20 bar and between
  -50C and 100C

7
Creating The Dynamic Process Model

• Components include CO2, O2, N2, Ar, CO, H2S, SO2,
  SO3, NO, NO2, H2O and NH3
• Peng-Robinson EOS recommend for cryogenic gas
  processes
• Dynamic model created from scratch with a steady
  state model as reference
• Inlet stream as starting point for the model
• Pressure-driven simulation ? pressure
  specifications on every boundary stream

8
Creating The Dynamic Process Model

• Piecewise addition of unit operations downstream
  ? integrator run after each new operation
  to update stream data
• Advantage errors found immediately
• Disadvantage separate steady-state calculation
  necessary for initial values for the distillation
  column
• Contains
• 1 distillation column
• 1 plate-fin heat exchanger
• 3 adsorber
• 4 compressors and 1 pump

• 7 vessels
• 11 shell-and-tube heat exchangers
• about 180 transmitters and control units

9
The Dynamic Process Model - Challenges

• Keeping numerical stability
• Process showed periodical oscillation
• Stability increased by changing the integrator
  time step from 0.5 s to 0.3 s

10
The Dynamic Process Model - Challenges

• Insert pipe segment between units without
  pressure-flow relation or holdup (e.g. Tee or
  Mixer)

Holdup volume NOT modelled
Holdup volume modelled
11
The Dynamic Process Model - Challenges

• Modeling adsorber regeneration
• adsorber used for removing moisture
• USD adsorber model does not support regeneration
• spreadsheet and integrator logic used
• ? counts adsorbed water
• ? automatic switch of water split fraction when
  maximum capacity is reached
• ?counts back to zero during regeneration

? implementation of complicated adsorber control
possible
12
The Dynamic Process Model - Challenges

• Parametrising the plate-fin heat exchanger
• Estimation of heat transfer coefficients (U)
• Difficult for mixed gas-liquid flow and
  condensation / vaporisation
• Assumption of constant U-value
• Only rough correlation for U on mass flow
• Experimental estimation by comparison with
  steady-state temperatures
• Results UGas2080 W/m²K Ucond20002800 W/m²K
• Problem with temperature crosses solved with new
  LMTD approach

13
The Dynamic Process Model - Challenges

• Temperature cross of hot and cold streams

• Desired temperature profile

14
Setting Up The Control Scheme

• Control units necessary to keep desired process
  conditions, e.g. for plant pressure, levels and
  temperatures
• Dynamic model allows efficient controller tuning
  by monitoring the step response
• Different tuning rules used, e.g. Ziegler and
  Nichols
• Parameter testing possible
• Reduction of controller tuning time at startup

15
Setting Up The Control Scheme

• Step response of a pressure controller to
  identify the parameters of the system

(CHIEN, HRONES and RESWICK)
16
Setting Up The Control Scheme

• Using the stability limit to find controller
  parameters with Ziegler-Nichols

Kkr11,9
Tkr
17
Setting Up The Control Scheme

• Focus on level taps
• used to observe level in certain boundaries
• level normalised to a desired scale
• scale depends on the scale of the transmitter
• provides input for level controller

18
Preparing The DCS Connection

• Basic dynamic model ready and tuned
• Addition of all DCS tags besides the controller,
  i.e. transmitters for pressure, temperature, flow
  and composition as well as switches and state
  transmitters
• Making tag values available to external devices

19
Preparing The DCS Connection
Transmitter
Level control switch
Pressure specification on boundary stream
20
Expanding The Model With UniSim Operations

• OPC (OLE for Process Control) communication
  standard of real-time plant data between control
  devices from different manufacturers
• UniSim Operations provides an advanced operator
  training solution plus capabilities for control
  system checkout
• Only the OPC-Server is used for the project

21
Expanding The Model With UniSim Operations

• Create a reference text file with desired tags
  for OPC access
• ? UsdTags.dat
• Maintain a certain folder structure
• Load the dynamic model into UniSim Operations

22
Expanding The Model With UniSim Operations

• Successful loading indicated by the message
• Tags automatically added to MV-table
• Data now available over OPC

23
Testing the OPC-connection

• Test with a free OPC-client from Rockwell
  Software International (RSI)
• All values can be read
• Analogue values can be written by the OPC-client
• Writing error occurs with digital values (states)

24
Connecting the DCS - Software

• PCS 7 Distributed Control System by Siemens
• handles all automation and controlling tasks
• OPC-client NOT included
• Simit platform for running and testing
  S7-applications
• includes process simulation library
• includes Emulation Platform to emulate the
  automation hardware
• equipped with an optional OPC-Client package
• ? bridge between USD model and PCS 7

25
Connecting the DCS - Structure

• DCS structure
• Automation hardware emulated as SoftPLC on a PC
  with the Simit Emulation Platform
• Same performance and S7-code as real AS
• Signal communication between dynamic model and
  DCS

26
Connecting the DCS Test Network

• Test-network on three virtual machines

27
Connecting the DCS Platform Configurator

• Setting up a new project with the platform
  configurator
• ? Basis for overall connection
• Specifying all network resources (IP-address,
  shared folder, etc.)
• Specifying the distribution of the resources ? On
  which computer?

28
Connecting the DCS Platform Configurator

• Import PCS 7 hardware configuration SoftPLC
  emulation
• Specify communication connections (to SoftPLC, to
  operator interface (WinCC), to USD model)
• Generate emulation environment
• ? Simit project file with signal interface
  automatically generated
• Contains all tags from the DCS
• Tag names in the model must equal the DCS tags
• Do NOT use spaces in tag names

29
Connecting the DCS Platform Configurator
30
Connecting the DCS Inside Simit

• Import the generated Simit project
• Already contains the signal interface (connection
  between emulation and USD model)
• Additions necessary (Insert ? Coupling ? )
• Data link to emulation
• OPC-Client
• (Activate Take signals from plans)
• Compile Simit project
• ? Start simulation

31
Connecting the DCS Inside Simit
Signal interface hierarchy
Data from the model
32
Connecting the DCS Inside Simit

• Opening simulation starts the SoftPLC on the
  emulation PC

go online (CPU accessible to PCS7)
initialised running
33
Connecting the DCS Inside PCS7

• Model values visible in CFC plans in PCS 7 (test
  mode)

34
Connecting the DCS - Summary

• Communication successfully established

USD Model Process Values
TCP/IP
Simit Analogue / Binary - Switch
OPC
PCS 7 SoftPLC CFC - Plan
WinCC Operator Interface
USD Model Controller Output
TCP/IP
Simit Analogue / Binary - Switch
OPC
35
Conclusion and Outlook

• UniSim Operations together with Simit allows the
  connection of a UniSim Design Dynamics model and
  the control system PCS 7
• Proof given by model values in PCS 7 CFC-plans
• Signal interface easy to generate with Simit
• Difficulties
• Failure to communicate digital values
• ? Cause apparently in USO OPC-server
• Connection to operator interface inside PCS 7 not
  established
• (no effect on model DCS interaction)
• Creating UsdTags.dat reference file is
  time-consuming

36
Conclusion and Outlook

• Possible uses
• Testing the control system
• First estimates for controller parameters
• Operator training with the real DCS
• Reduce time and cost of startup
• Increase safety and stability of the plant
• latest USO release R320 already includes an
  interface to Siemens PCS7 Simit

37

• Thank you for your attention!