PVSC Improves PLC Programming Using Real-Time Dynamic Simulation

1
PVSC Improves PLC Programming UsingReal-Time
Dynamic Simulation

• Paul Cavanagh, P.E.
• Passaic Valley Sewerage Commissioners
• ISA WWAC Symposium
• August 2009

2
Background
3
Passaic Valley Sewerage Commissioners

• Owns and Operates a 330 mgd wastewater treatment
  plant in Newark, NJ
• remove 93 of BOD
• remove 94 of TSS
• peak dry weather flows of 400 mgd
• peak wet weather flows of 550 mgd
• Treats about 25 of NJs Wastewater
• 15 industrial by volume
• 50 industrial by strength

4
PVSCs Control System Requirements

• Balance plant flow between select units
• Maintain a Constant Upstream Channel Level
• Respond quickly to flow disturbances and rain
  events

5
Mixed Liquor Channel Level

• Maintain a constant stable level for scum
  collector
• Quickly correct for disturbances and prevent
  overflows

6
Secondary Clarifier Settling Tanks

• Evenly Split the flow between the 12 units
• Ability to turn off control to select units
• Minimize the movements of the Flow Control Valves

7
The Problem
8
Existing PLC Flow Balance and Level Control was
not meeting its Requirements

• The controls were slow to correct for a
  disturbance such as a rain event
• Stability problems that got worse during peak
  plant flows
• Control to individual units could not be shut off
• Many Flow Control Valves were constantly opening
  and closing 1 to 3 every minute
• Tuning attempts only had a minimal effect

9
The Channel Level ControlResponds Slowly and is
Unstable
Worse During High Flows
10
Many Flow Control Valves WereConstantly Hunting
Back and Forth
11
As a Result of the Problems

• the System was kept in manual most of the time
• Distribution of solids was uneven making blanket
  levels more difficult to control
• The Scum Collector weir level would need to be
  frequently adjusted
• The risk of overflow into an empty tank was
  always present

12
But Modifying the Control Program on the Running
System is Problematic

• Need to wait for a rain event to test the changes
• Uncertainty would remain as not all rain events
  are exactly alike
• The new program could fail when no one is looking
• A units settling could be disturbed and solids
  would be discharged
• The channel could overflow

13
So PVSC chooses to try

• Real-Time Dynamic Simulation
• Build a model of that captures most of the
  real-time dynamic interaction between the channel
  level and the flow control valves and test the
  PLC program on the model
• Make changes to the PLC Program and after proving
  the program against simulated disturbances
  install the changes in the field

14
Developing the Dynamic Modelof the Channel and
Valves
15
How Do You Build a Dynamic Simulator?

• Identify the Dynamic Components Blocks
• Resistors
• Valves, Orifice, Restrictions
• Line Equation
• Capacitors (Integrators)
• Tanks, Reservoirs, Valve Actuators
• Integrates the flow
• Inductors (Differentiator)
• Pipes, Conduits,
• Differentiates the flow
• Transfer Function
• A convenient way of representing a dynamic system

16
Fluid Resistors

• A Fluid Resistors are like Electrical Resistors
  except
• For Electrical Resistors the Electric Current
  changes linearly with Electric Voltage (Ohms
  Law)
• For Fluid Resistors the Fluid Current typically
  changes non-linearly with Fluid Pressure
  (Bernoullis Equation)
• Flow Control Valves behave like variable fluid
  resistors
• Use a straight line equation for an ideal flow
  control valve

17
Fluid Capacitors

• A Fluid Capacitor is like and an Electrical
  Capacitor except
• An Electrical Capacitor stores electric energy
• A Fluid Capacitor stores fluid mass
• An open tank or reservoir is a fluid capacitor
• Integration of the flow entering/exiting the
  capacitor an initial valve produces a pressure
  or head

18
Fluid Inductors

• A Fluid Inductor is like an Electrical Inductor
  except
• An electric current through an electric inductor
  is sustained by its magnetic field.
• A fluid current through a fluid inductor is
  sustained by the inertia of the mass of fluid in
  motion.
• Long pipes or conduits are examples of fluid
  inductors.
• Differentiate the change in flow through the
  inductor to find the change in pressure across
  the inductor.

19
Transfer Functions

• Laplace Transform Transfer Function
• 1st Order
• 2nd Order
• nth Order
• Captures the input/output dynamics in one block

20
Using VisSim/OPC

• PVSC staff used VisSim software to build and run
  its dynamic simulations.
• VisSim is a computer software application that
  provides a visual block diagram language for
  modeling and simulation of complex nonlinear
  dynamic systems. Its fast execution lets you run
  models in real-time.
• The OPC (OLE for Process Control) is an add-on to
  VisSim. The VisSim model use OPC to read and
  write data on the PLC.
• With VisSim/OPC you can run a virtual plant for
  testing and developing the PLC code.

21
Modeling the Mixed Liquor Channel

• Modeled it like a fluid capacitor it integrates
  the difference between the flow in and flow out
  of the channel into the gallons in the channel.
• The surface area of the channel is used to
  convert the volume of gallons in the channel into
  an elevation level.
• Limit the Integration for the real world
  boundaries
• The Elevation of the Upstream Weirs (104 feet)
• The Elevation of the Downstream Weirs (99 feet)
• Use the channel level to supply the pressure
  across the valves
• 0 to 5 feet of water

22
PVSCs Channel Level Model in VisSim
23
Modeling a Flow Control Valve

• 3 Steps Were Used
• Determined the dynamic valve position by
  integrating the open and close signals and adding
  the result to the initial position
• Treated the valve like a variable fluid resistor
  that is effected linearly by the position and by
  the square root of the level
• Used straight line equation for steady state
  flows response to position
• Biased that equation by the square-root of the
  head pressure across the valve (creates the
  observed non-linear effect)
• Used a transfer function to capture the lag
  response

24
Step 1 - Valve Position Calculation
25
Step 2 - Steady State Flow Calculation
26
Valve Flow Increases With Channel Level
27
Step 3 - Lag Response Calculation
28
PVSCs Complete Flow Control Valve Model
29
The Complete Plant Model
30
The OPC Interface
31
Human Machine Interface to the Model
32
Improvements Made From Testing PLC Program with
the Dynamic Simulator
33
Changes in Channel Level Control Program

• Old Program

• New Program

• Moving Average Filter of Channel Level
• Channel Level Control Bias Multiplied Flow Set
  Point
• Proportional Integral Control for of channel
  level

• Exponential Average Filter of Channel Level
• Channel Level Control Bias Added to Flow Set
  Point
• Proportional Only Control of channel level
  (system is self integrating)

34
Changes to Flow Balance Control Program

• Old Program

• New Program

• PID block calculates Position
• Manipulates valve to a calculated position
• Large 2 position deadband
• Proportional-Integral Control of each Tanks Flow
• All operating tanks needed to operate
• Cannot Balance without level meter

• PID block calculates Change of Position
• Manipulates valve for a calculated time
• Small 1 second deadband
• Proportional Only Control of each Tanks Flow
• Any combination of tanks can be put in manual
• Can Balance without level meter

35
Comparison of Channel Level Program Response
Total Flow Changed from 360 to 600 mgd
36
Comparison of Flow Balance Level Response
Total Flow Changed from 360 to 600 mgd
37
Comparison of Valve Position Manipulation
Total Flow Changed from 360 to 600 mgd
38
The Simulation Shows thatthe New Program
Provides

• Much Faster and Very Stable Level Control
• prevents overflows of upstream weirs
• helps the operation of the channels scum
  collector
• Much Faster and Very Precise Flow Splitting
• produces more even solids distribution
• A Significant Reduction in Valve Movements
• saves on valve maintenance and repair

39
Running New PLC Program On the Real System
40
New PLC Program Worked Immediately!
41
Conclusions about Using Dynamic Simulation

• Dramatically improves PLC programming
• Can be faster in the long run than waiting for
  real disturbances
• Much Safer than Testing code on the actual system
• Provides new insight into the mechanics of the
  process which can lead to further improvements

42
Special Thanks to

• The Commissioners
• Bryan Christiansen
• Sheldon Lipke
• Phil Habrukowich
• Tom Wasilewski
• Jerry Oselador
• Loukas Koufodontes

43
Resources

• Books
• Modeling Engineering Systems by Jack W. Lewis
• Instrument Engineers' Handbook Process control
  and optimization  by Béla G. Lipták
• Engineer In Training Reference Manual by Michael
  R. Lindberg, PE
• Web Sites
• http//www.vissim.com/
• http//blog.prosig.com/

44
Demonstration
45
The End

• Paul Cavanagh, PE
• pcavanagh_at_pvsc.com
• http//www.pvsc.com/