You are on page 1of 27

Simulation of Model

Predictive Control for a


Coal Fired Power Plant
Standards
Certification
Education & Training
Publishing
Conferences & Exhibits
Presenters

• Phillip D. Schnelle Jr. - Principle Consultant, Process


Control - E. I. DuPont de Nemours Co. Inc.

• Paul S. Fruehauf - Principle Consultant, Fruehauf


Engineering Services, Inc.

2
Agenda

• Background – Paul
– Literature Survey
– Power Plant Unit Management
– Dynamic Simulation
• Discussion - Dave
– ARC vs. MPC
– Simulation Step Testing
– MPC Configuration
– Simulation System
– Results
– Operator, Maintenance and Engineer Training
– Rapid Prototyping
• Conclusions

3
Literature Survey

• Advanced Controls Not Widely Applied in Power Industry


(Rees And Lu, 2002).

• 70% Improvement in Ramp Rate Using MPC On A Coal


Fired Power Plant. (Stump and Williams 2009)

• Application of DMC to Power Plants (Aurora, et. al.


2004), (Moon and Lee 2009)

• Additional Measurements to Improve MPC


– Turbine First Stage Pressure (LU and Hogg, 1997)
– Flame Radiation Intensity Signal (Wang et. al. 2007)

4
Literature Survey (Flame Radiation)

Taken from (Wang et. al. 2007)

Steam Pressure – Coal Radiation Intensity – Coal


Step Increase Step Increase
• Note: Deadtime and Time Constant Reduction
• Input to MPC to Further Improve Control

5
Background

• 500 Megawatt Power Plant

6
Background CFB Boilers

7
Background Turbine Generator

8
Background

• Unit Management Controls

– Need to Control Megawatts and Main Steam Pressure


– We Can Manipulate Turbine Valves and Coal Rate
– It is Often Presented as a 2 x 2 Control Problem

9
Background

• Two Single Input Single Output (SISO) Variable Parings


• Turbine Follow Mode

• Advantage - Stable Operation


• Disadvantage – Imprecise Megawatt Control

10
Background

• Boiler Follow Mode

• Advantage - Precise Megawatt Control


• Disadvantage – Less Stable

11
Background

• Coordinated Control

• Retains the Advantages of Both SISO Schemes


• Faster Megawatt Ramp Rates
• Implemented in Past Using ARC

12
Dynamic Simulation

• Boilers and Turbine

13
Regulatory APC Approach: Overrides to protect steam header pressure, drum level. Interaction
compensation between pressure and drum level. 3 element drum level, deadtime compensation for coal feed.
Control Objective:
Decoupling / 3 element Network
- Control Drum Level (inverse response behavior, interactions)
- Control Header Pressure (deadtime wrt coal feed, interactions)
- Control MW (interaction, maximize) +
+
ARC Control thoughts: Overrides to protect steam header pressure and drum level, interaction compensation between pressure
(T2drum
and drum level, 3 element s+1) level, deadtime
(T2s+1) compensation for coal feed, feedforward for load upsets…
1/(T2s+1)
(T4s+1) (T4s+1)
Feedforwards
How
Smith to Implement in ARC? What is minimum essential?
Predictor 1/(T1s+1)
SPPC Overrides Controllers
LC
e-Ls LC
+ FF
+ FF KWC
PC

LS

LT
FC

FC

Boiler Feed Water


Regulatory APC Approach: Overrides to protect steam header pressure, drum level. Interaction
compensation between pressure and drum level. 3 element drum level, deadtime compensation for coal feed.
Decoupling / 3 element Network

+
+
(T2s+1) (T2s+1)
1/(T2s+1)
(T4s+1) (T4s+1)
MPC Feedforwards
Smith Predictor 1/(T1s+1)
SPPC Overrides Controllers
LC
e-Ls LC
+ FF
+ FF KWC
PC

LS

LT
FC

FC

Boiler Feed Water

How to Implement in MPC


Model Predictive Control
How does it work? Why is it better?

CVs

MVs

t=0
Model Step Test for MPC Development
MPC Configuration

20
Simulation / Training System

Sever

Web based remote monitoring


at site or any location
OPC connection

First Principles Modeling


MPC Software

For DuPont Use Only


Simulation Results
Operator HMI Prediction Plots

Controller is ON making moves

Generating moves for MVs

Predicts it can achieve Targets


at steady state
Operator HMI Prediction Plots
Model Predictive Control Rapid Prototyping

• No Capital Cost, No Hardware or Software to Install


• Quantitative Analysis of Benefits
• Real-Time Comparison of Model to Actual Process
Rapid Model Development Real-Time View of Model Performance
• Hours to Implement
Establish Relationship
Adjust this
between Manipulated
Process MPC Controller
Variables and Control Actual Process
Condition Adjustments
Variables. Conditions

MPC “Drives”
Process to these
Conditions

Quantify Benefits by
Comparing Steady
View Impact on
State to Actual
other Process
Model Development Measurement
Conditions
Rapid Prototyping System

Control Concept Remote Monitoring


At Site or Any Location
-

Determine Benefits of Actual


Site PHD Historian
to Model
Site IP21 Historian
Site IP21 Historian
2. Operate at
1. MPC will lower Reactor
Real-time Process Data increase rate Pressure

3. Run at Chest
Pressure
4. Quantify Benefits Constraint
by Comparing
Actual (Blue) to
Steady State
For DuPont Use Only (Green)
How to estimate Benefit from Prototyping

Controller is OFF just calculating not making moves

Would be moves for MVs


Not actually made

Predicts it can achieve Targets


Or hold Limits at steady state
How to estimate Benefit from Prototyping

MPC would dive pressure to higher limit, so


benefit would be integral between blue and
solid green line.

Blow up
Conclusions

• Significant benefits to be obtained by using MPC for coordinated boiler / turbine control.
• MPC provides better regulation to target because of the natural deadtime compensation and
model based feed forward action.
• Better regulation allows the coordinated control to push and hold tighter to
the upper bounds of target limits which allows for more efficient operation.
• The Simulation and MPC test environment provides for an efficient means for evaluating and
demonstrating the benefits of advanced control.
• This environment also provides a good platform for engineer, maintenance and
operator training.
• Rapid prototyping can be very helpful in conceptualizing and promoting the benefits of MPC.

You might also like