EPRI Control

Power Plant Control System Tuning Short Course Notes

1003740

Power Plant Control System Tuning Short Course Notes 1003740

Technical Update, March 2004

EPRI Project Manager R. Torok

EPRI • 3412 Hillview Avenue, Palo Alto, California 94304 • PO Box 10412, Palo Alto, California 94303 • USA 800.313.3774 • 650.855.2121 • [email protected] • www.epri.com

DISCLAIMER OF WARRANTIES AND LIMITATION OF LIABILITIES THIS DOCUMENT WAS PREPARED BY THE ORGANIZATION(S) NAMED BELOW AS AN ACCOUNT OF WORK SPONSORED OR COSPONSORED BY THE ELECTRIC POWER RESEARCH INSTITUTE, INC. (EPRI). NEITHER EPRI, ANY MEMBER OF EPRI, ANY COSPONSOR, THE ORGANIZATION(S) BELOW, NOR ANY PERSON ACTING ON BEHALF OF ANY OF THEM: (A) MAKES ANY WARRANTY OR REPRESENTATION WHATSOEVER, EXPRESS OR IMPLIED, (I) WITH RESPECT TO THE USE OF ANY INFORMATION, APPARATUS, METHOD, PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS DOCUMENT, INCLUDING MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, OR (II) THAT SUCH USE DOES NOT INFRINGE ON OR INTERFERE WITH PRIVATELY OWNED RIGHTS, INCLUDING ANY PARTY'S INTELLECTUAL PROPERTY, OR (III) THAT THIS DOCUMENT IS SUITABLE TO ANY PARTICULAR USER'S CIRCUMSTANCE; OR (B) ASSUMES RESPONSIBILITY FOR ANY DAMAGES OR OTHER LIABILITY WHATSOEVER (INCLUDING ANY CONSEQUENTIAL DAMAGES, EVEN IF EPRI OR ANY EPRI REPRESENTATIVE HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES) RESULTING FROM YOUR SELECTION OR USE OF THIS DOCUMENT OR ANY INFORMATION, APPARATUS, METHOD, PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS DOCUMENT. ORGANIZATION(S) THAT PREPARED THIS DOCUMENT EPRI Instrumentation and Control Center

This is an EPRI Technical Update report. A Technical Update report is intended as an informal report of continuing research, a meeting, or a topical study. It is not a final EPRI technical report.

ORDERING INFORMATION Requests for copies of this report should be directed to EPRI Orders and Conferences, 1355 Willow Way, Suite 278, Concord, CA 94520. Toll-free number: 800.313.3774, press 2, or internally x5379; voice: 925.609.9169; fax: 925.609.1310. Electric Power Research Institute and EPRI are registered service marks of the Electric Power Research Institute, Inc. EPRI. ELECTRIFY THE WORLD is a service mark of the Electric Power Research Institute, Inc. Copyright © 2004 Electric Power Research Institute, Inc. All rights reserved.

CITATIONS This report was prepared by EPRI I&C Center 714 Swan Pond Road Harriman, TN 37748 Principal Investigators C. Taft This report describes research sponsored by EPRI. The report is a corporate document that should be cited in the literature in the following manner: Power Plant Control System Tuning Short Course Notes, EPRI, Palo Alto, CA: 2004. 1003740.

iii

ABSTRACT Power plant control system tuning is a continuing challenge for many power producers. To help address the problem, EPRI began a project in 2002 to investigate improved methods for control system tuning. One of the tasks undertaken in that project was the development of a power plant control system tuning short course. This report provides approximately 100 slides that make up the bulk of the course material. In addition, tuning demonstrations using computer simulation software will be a part of the course. The course notes provided here include information on process response concepts, control system fundamentals, PID tuning, and boiler control system applications examples. It will be useful for plant engineers and technicians involved with control system tuning and plant responsiveness.

v

TABLE OF CONTENTS

1 INTRODUCTION.................................................................................................................1-1 2 COURSE OUTLINE.............................................................................................................2-1 3 COURSE NOTES .................................................................................................................3-1

vii

1 INTRODUCTION

Power plant control system tuning is a continuing challenge for many power producers. To help address the problem, EPRI began a project in 2002 to investigate improved methods for control system tuning. The project has already produced several short reports on power plant control system tuning, including one on tuning assessment and one on PID controller tuning software programs (see Section 3 for complete reference list). The latest effort was the development of a short course on fossil power plant control system tuning. This technical update provides the latest information on the course. Section 2 contains the course outline, and Section 3 provides approximately 100 slides that make up the bulk of the course material. In addition, tuning demonstrations using computer simulation software will be a part of the course, but are not included in these notes. The two-day course will be offered at the EPRI Instrumentation and Control Center in Harriman, TN. The course notes provided here include information on process response concepts, control system fundamentals, PID tuning, and boiler control system applications examples. It will be useful for plant engineers and technicians involved with control system tuning and plant responsiveness.

1-1

2 COURSE OUTLINE 1.

2.

3. 4.

5.

Course scope and objectives 1.1. Conventional control system tuning (PID -based) 1.1.1. PID tuning 1.1.2. Cascade control 1.1.3. Feedforward signals 1.1.4. Focussed on boiler control systems 1.1.5. Applicable to other systems 1.2. Make plant engineers and technicians more knowledgeable about control system tuning methods 1.3. Not a control design course 1.4. References What is tuning? 2.1. Adjustment of control system parameters to make plant respond as desired 2.2. Tuning is not the same as design 2.2.1. Design means selecting a strategy and logic 2.2.2. Tuning means adjusting settings in current design 2.3. There are other adjustments besides PID settings 2.4. Linearizers (f(x)), lead/lags, a few others 2.5. Tuning terminology and definitions 2.5.1. System 2.5.2. Process 2.5.3. Response 2.5.4. Controller 2.5.5. Closed loop 2.5.6. Open loop, etc. Control system diagramming 3.1. SAMA diagrams 3.2. Block diagrams Process response concepts 4.1. Common responses 4.1.1. First order lag 4.1.2. Second order under-damped 4.1.3. Integrator 4.1.4. Deadtime 4.2. Parameters, e.g., Gain, time constant,, time delay 4.3. Second order comments Control system fundamentals 5.1. Feedback 2-1

Course Outline

5.2. 5.3. 5.4. 5.5. 5.6. 5.7. 5.8. 6.

7.

8.

9.

2-2

PID Controllers Cascade Control Feedforward control Trim controllers Loop error or deviation Continuous control Digital control 5.8.1. Sampled data 5.8.2. Aliasing Performance measures 6.1. Speed vs. stability 6.2. Overshoot 6.3. Rise time 6.4. Settling time 6.5. Setpoint tracking vs. disturbance rejection PID tuning concepts 7.1. Empirical methods 7.1.1. Example 7.2. Analytical Methods 7.2.1. Example 7.3. Automated programs System linearity issues 8.1. Why linearity is important 8.2. Characterization 8.3. Gain scheduling Boiler control system tuning 9.1. Approach to boiler control tuning 9.2. The hierarchy of tuning 9.3. Actuator setup and tuning 9.4. Feedwater flow/drum level 9.5. Furnace pressure control 9.6. Air flow control 9.7. Fuel flow control 9.8. Boiler follow throttle pressure control 9.9. Steam temperature control 9.10. Front end control, throttle pressure and megawatts

3 COURSE NOTES

This section contains the slides that make up the course notes.

3-1

Power Plant Control System Tuning Short Course

Course Notes by Cyrus W. Taft, PE EPRI I&C Center

Course Objectives • Present an overview of control system tuning principles and practices for fossil power plants. • Provide detailed guidance on: – – – –

Proportional-integral-derivative (PID) controller tuning. Feedforward tuning Major boiler control loop applications Tuning assessment

• Intended audience is plant engineers and technicians, technical specialists, support engineers. • Course is not intended as control design course. • Students should have some experience with power plant control systems and typical boiler processes.

2

Copyright © 2004 Electric Power Research Institute, Inc. All rights reserved.

1

Tuning References

• EPRI Reports – Tuning Guidelines for Utility Fossil Plant Process Control, EPRI, Palo Alto, CA: TR-102052, Volumes 1-4, 1993-1994. – Power Plant Control System Tuning, EPRI, Palo Alto, CA: TE113653, 1999 – Automated Control System Tuning: Issues, Available Solutions, and Potential for Improvement, EPRI, Palo Alto, CA: 1004067, 2001 – Control System Tuning Assessment Guidelines, EPRI, Palo Alto, CA: 1004425, 2002 – Review of State-of-the-Art PID Controller Tuning Software Programs, EPRI, Palo Alto, CA: 1004080, 2004

3


Tuning References

• Books – Astrom, K.J. and Hagglund, T., PID Controllers: Theory, Design and Tuning, ISA, Research Triangle Park, NC: Second Edition, 1995. – McMillan, G. K., Tuning and Control Loop Performance, ISA, Research Triangle Park, NC: Second Edition, 1983. – Corripio, A. B., Tuning of Industrial Control Systems, ISA, Research Triangle Park. NC: 1990. – Dukelow, S. G., The Control of Boilers, ISA, Research Triangle Park. NC: Second Edition, 1991. – Levin, W., Editor, The Control Handbook, CRC Press, Boca Raton, FL: 1995.

4


2

Tuning References

• Papers – Morse, R. H., et al, Aspects of Tuning a Boiler Control System - A Strategy for Optimization, ISA IPI 76462, pp 121-142, ISA POWID Symposium Proceedings, 1976. – Hubby, R. N., Pulverizer Control, A Tutorial Review, ISA, ISBN155617-213-3, pp 157-164, ISA POWID Symposium Proceedings, 1991. – Taft, C. W. and McFarland, G., A Guide to Boiler Control System Startup and Checkout, ISA, First Joint ISA POWID/EPRI Controls and Automation Conference Proceedings, 1991. – Heuszel, C. B., Methods of Accurate Control for Environmental Compliance, Tutorial presented at ISA/94 Conferences in Philadelphia, PA, and Anaheim, CA., 1994

5


What is Control System Tuning? • Tuning is the adjustment of various control system parameters to make the process respond as desired. • Sometimes viewed as a “black art.” • Mostly based on science with a little art thrown in. • Not as difficult as you may think. • More than just PID tuning • Tuning is not the same thing as design in the conventional control world but, in advanced control, the two have much overlap.

6


3

Tuning Terminology

•

Process - The system being controlled, e.g., the feedwater flow system or the air flow system. Also referred to as the plant.

•

Controller - the device or algorithm which is manipulating the process input to effect a change. Most common is PID controller but there are many other possibilities.

•

Actuator - The device which moves the final control element, such as a valve or damper, in response to command signals from the controller. May use pneumatic, electric or hydraulic power.

•

Feedback - A type of control in which a measurement of the process is used by the controller to adjust the final control element. The concept of feedback is used in all major control functions in a power plant.

•

Controlled variable - The process parameter being controlled by the control system. Also called the process variable.

•

Manipulated variable - The controller output signal.

7


Tuning Terminology, cont’d

•

Open loop - The system has no feedback in service. Also means the control loop is in manual mode.

•

Closed loop - Feedback from the process is being used by the controller. Control loop is in automatic mode.

•

Gain - The output of a system divided by its input. A measure of the change in signal size as it passed through a system.

•

Frequency response - The output of a system in response to a sine wave input signal at many different frequencies. Usually specified as a gain and phase shift.

•

Step response - The output of a system in response to a step change in the input signal.

•

Impulse response - The output of a system in response to an impulse change in the input signal. In theory an impulse is an infinitely large and infinitely short pulse.

8


4

Tuning Terminology, cont’d

•

Time response - The output of a system as a function of time.

•

Time constant - A measure of the speed of response of a system. Sometimes refers to the time for a first order system to reach 63% of its final step response.

•

Stable - In theory, a type of system whose output is bounded for all bounded inputs. In practice, a system whose output settles out in response to a disturbance.

•

Stability margin - A measure of how close a system is to instability.

•

Disturbance - Any upset that occurs in a system. Often implies an unmeasured upset that the control system is trying to eliminate.

•

Continuous control - An analog control system, i.e., one in which all signals are continuous functions of time.

•

Discrete control - Also called digital control. A digital control system, i.e., one in which the controller signals are only updated at distinct points in time. Also referred to as a sampled data system.

9


Control System Diagramming

• There are two methods commonly used in power plant control analysis – SAMA (Scientific Apparatus Makers Association) Diagrams – Block Diagrams • SAMA diagrams functional diagrams in which on block represents one control system function, such as a summer. They are normally used by control system vendors to describe how the system works. They contain no information about the process being controlled. • Block diagrams are more mathematically oriented and are used widely in control system analysis. They show the control system and the process being controlled. • Both types will be used in this course.

10


5

SAMA Diagram Example

Feedwater Flow

Drum Level

Drum Pressure

Steam Flow

FT

LT

PT

FT

Density Compensation

∆

∆

PI

PID

Circle - Measurement or readout device

Measurement

Κ

Rectangle Automatically processed function

Control Participation

Σ

Σ

−Κ

Σ

f(x)

A

T

f(t)

f(x)

A

A

T

Output

A Diamond - Manually controlled function

11

f(x)

f(x)

Boiler Feed Pump A

Boiler Feed Pump B

Trapezoid Final control element


Block Diagram Example

Disturbance Setpoint

+

8___ 38s + 1

Kp + Ki/s Error

Controlled Variable or Process Variable

Manipulated Variable

1___ 4s + 1

12


6

Process Response Concepts • Real power plant processes are quite complex and include: – Fluid flow through pipes, valves, turbines, fans, pumps. – Heat transfer by conduction, convection, radiation. – Combustion – Fluid phase changes, boiling, condensation. – Material transport and graining, pulverizers. • These are all mechanical or chemical type processes. • All processes must obey the laws of physics, including conservation of mass, energy and momentum.

13


Process Response Concepts • An important characteristic of all processes is their speed of response. This is quantified as either a time or a frequency which are reciprocals of each other. • Another important parameter is the gain of the system. It is important to understand that the gain of a system changes depending on the speed (or frequency) of the input signal. • No real processes respond instantly to a change in the input signal. All have some delay or deadtime in their response. This is very important in tuning. • In the following examples, the input signal chosen is a step change. It is important to remember that the characteristics of the process are independent from the type of input signal. The step response is just a convenient signal to visualize.

14


7

Common Process Time Responses to Step Inputs First Order Lag 2 e d ut i gn a M

5

10

15Order Second

20

25

1 0 0

5

10

15 Integrator

20

25

30

0

5

10

15 Deadtime

20

25

30

20

25

30

4

e d ut i ng a M

•

Most common is 1st order with deadtime – Characterized by 3 parameters: gain, time constant, deadtime – Gain - ratio of final change in output to change in input. – Time constant - time to 63% of final response. – Deadtime - time before any response begins. – No overshoot in step response

30

2

e d ut i gn a M

Four common process models – First order lag, second order under-damped, integrator, and deadtime

Time constant

0 0

e d ut i ng a M

• 63%

1

2 0

1 Deadtime 0 0

5

15

10

15 Time (seconds)


First Order Lag Plus Deadtime Process Examples • Boiler pressure response to a step change in fuel input. • Pulverizer outlet temperature response to a step change in hot air damper position • Superheat outlet temperature response to a step change in spray valve position

16


8

Integrator Plus Deadtime Process Examples • Drum level response to a step change in feedwater flow rate. • Feedwater heater level response to a step change in drain valve position.

17


Comments on Second Order Systems • Processes that have an under-damped second order type response are rare in power plants. • However, many second order responses occur in closed loop control systems. • The combination of a PI controller and a first order process produces a second order closed loop system. • A system must be at least second order to have any overshoot in its step response. • A second order system can be over-damped and have no overshoot in its step response.

18


9

Closer Look at the Process

• Every process contains many smaller processes

Disturbance Setpoint

+


8___ 38s + 1

Kp + Ki/s Manipulated Variable

Error

1___ 4s + 1

• Total response is combination of all blocks Process


I/P

Booster

Diaphragm

Valve

Desuperheater


Thermocouple What Operator Sees 19


Control System Fundamentals • Feedback is the foundation of automatic control • PID controllers are the workhorse • Cascade control and feedforward control are important in boiler control applications • All new control systems today are digital • Field devices, transmitters and actuators, are very important components in the control system. • Must know control system objective before tuning

20


10

Feedback

• Process measurement is sent back to the control system • Negative feedback because measurement is subtracted from setpoint

Control system

Field

Process Variable Process Under Control

• Foundation of automatic control • Vastly improves accuracy of control • Introduces possibility of instability

Measurement Device Feedback from process to control system

• When feedback is used, it is called closed loop control

21


Proportional-Integral-Derivative (PID) Controller • Dominant control algorithm by far. • Three possible control modes: – Proportional – Integral – Derivative • Most common implementation is PI control, but P, I, and PID are also quite common.

∆ K

d dt

• All PID controllers are conceptually the same, but each vendor has own implementation. • Where the tuning action is. • Tuning settings have a variety of names. 22


11

PID Controller Step Responses

PID Control Step Responses

• Proportional Output = Kp * Input

3 2 p or P

• Integral Output = Ki * Int(Input)

1 0

• Derivative Output = Kd * d/dt(Input)

0

5

10

15

20

25

30

0

5

10

15

20

25

30

0

5

10

15 Time (seconds)

20

25

30

15 10 t nI

5 0 2 1.5

vi r e D

1 0.5 0

23


Forms of PID Controllers • There are several forms of PID controllers and these two are quite common today: • Standard (ISA) • Kp * (e)[ 1 + 1/Ti * Int(e) + Td * d/dt(e)] – Note how Kp affects integral and derivative setting. • Parallel • Kp * (e) + Ki * Int(e) + Kd * d/dt(e) – Kp, Ki, and Kd are all independent

24


12

Proportional Only Control (Single-mode control) • Simplest form of control. • Only one tuning parameter. • Usually does not produce zero steady state error ( has an offset). • Application examples: – Turbine governor speed regulation – Emergency drain valve on feedwater heaters – Boiler feedpump minimum flow control • Tuning parameter may be specified two ways: – Gain, – Proportional band • Gain = 100/PB; PB = 100/Gain

25


Proportional plus Integral Control (Two-mode control) • Most common form of control. • Two tuning parameters. • Provides zero steady state error, no offset. • Examples: – Feedwater flow control – Air flow control – Furnace pressure control • Integral tuning parameter has several possible names: – Integral gain – Reset rate, repeats per minute – Reset time, minutes per repeat – Integral time, minutes

26


13

Proportional plus Integral plus Derivative Control (Three-mode control) • Not used as much as it should be. • Three (or four) tuning parameters. • Can improve the response of slow processes. • Derivative mode can amplify process noise. • Application examples: – Superheat Temperature Control – Pulverizer Outlet Temperature Control • Derivative tuning parameter has two names: – Derivative gain (Rate) – Derivative time, minutes

27


Cascade Control

•

Two feedback loops nested together with the output of the primary loop controller acting as setpoint for the secondary loop. This control scheme requires that the secondary loop be much faster than the primary loop.

•

Example: – Drum level control (slow primary or outer loop) – Feedwater flow control (fast secondary or inner loop)

Primary Control Variable

Secondary Control Variable

TT

FT

Setpoint

∆

+

A

PID


Secondary Error

-

Primary Controller

+

-

Secondary Controller

Primary Error

Secondary Process

Measurement

∆ PID

Primary Process

Inner Loop Outer Loop

Measurement

To Actuator

28


14

Cascade Control

• Advantages: – Disturbances arising within the secondary loop are corrected by secondary controller before they influence primary controller. – Secondary loop linearizes the process response improves the speed of response of the primary loop. • Disadvantages: – More complex strategy, more tuning adjustments. – Control of each variable is assumed single-loop and is designed to operate satisfactorily, Instability may occur when both loops are closed (in auto). – Cascade control requires that an intermediate process variable can be reliably measured. – Secondary loop must be significantly faster than primary loop.

29


Feedforward Control • A control strategy in which a signal other than the controlled variable is added directly to the controller output to improve a control loop’s response. • For a feedforward signal to be advantageous, it must provide some intelligence about a process change before the controlled variable detects it. • Can be dynamic (e.g., kicker) or static (e.g., load index). • Example: – Steam flow as a feedforward to drum level control. • Advantages: – Can dramatically improve the response of slow loops. • Disadvantages: – Adds complexity to loop. – Can be difficult to tune. 30


15

Feedforward Control

Feedforward

Process Variable

FT

TT

∆

A

PID

K

f(x)

Σ

To Actuator

31


Trim Controller • A controller whose output is used to bias the main control demand signal. • The trim controller’s output is normally zero centered with a +/20% to 30% range. • Example: – Drum level controller in a three element control strategy. – Excess oxygen controller • Main control signal is usually a feedforward signal.

32


16

Trim Controller

Demand Signal

Process Variable

FT

TT 0 to 100%

∆

A

PID Trim Controller -20% to +20%

Σ 0 to 100%

Main Signal Flow

To Actuator 33


Loop Error or Deviation • Difference between setpoint and controlled variable. • To better understand a control loop’s performance, watch the error signal. – Especially on setpoint tracking loops – Easy to spot offset or slow return to setpoint.

34


17

Loop Error Step Response

1.2

∆

1

0.8

K

Magnitude

Loop Error

d dt

0.6

0.4

0.2

0

-0.2

0

5

10

15

20

25

30

Time (seconds)

35


What is Continuous Control? • Also called Analog Control • Control signals are continuous with time. • Pneumatic control systems and analog electronic control systems are continuous. • No computer involved.

36


18

What is Digital Control? • Also called Discrete Control or Sampled Data Systems. • Implies a computer is involved. • Control actions are not continuous with time, they only occur at certain intervals, called the sample time. • Requires discrete mathematics to analyze. • No knowledge of the process exists between samples. • Aliasing of data can occur if sample rate is not fast enough. • If sample rate is very fast, digital control is very similar to analog control.

37


Digital Control Diagram

Digital Controller

Field Transmitter

A/D Converter

Controller

D/A Converter

Analog

Actuator Analog

Plant Being Controlled Analog

38


19

Digital Data Sampling S te p R e s p o n s e 2 .5

Amp litu d e

2

1 .5

1

0 .5

0 0

5

10

15

20

25

Time (s a mp le s ) 39


Aliasing • Aliasing occurs when a waveform is sampled too slowly. • The sampled data does not represent the actual waveform. The apparent frequency of the sampled data may be much lower than the actual frequency of the original signal. • A waveform must be sampled at least twice per cycle to prevent aliasing. • Aliasing can also be prevented by low pass filtering the signal before sampling. This eliminates high frequency components in the signal. • Some DCS systems don’t adequately filter the analog input signals before sampling to prevent aliasing.

40


20

Aliasing

1

0 .5

0

-0 .5

-1

0

41

20

40

60

80


Control System Performance Requirements • Need to know desired performance before loop can be properly tuned. • Performance is quantified in two categories: – Speed • Time taken for the process to respond to a specific input (usually a step) • Measurables: time constant, delay time, rise time, bandwidth, natural frequency

– Stability • Theory - The output response is bounded for all bounded inputs • Practice - Response does not oscillate too much • Measurables: damping, overshoot, settling time, offset

• Faster speed of response generally leads to less stability. • Most power plant control systems do not have well defined performance goals

42


21

Control System Performance Measurements

1.8

1.6

1.4

Overshoot

1.2

Magnitude

1

0.8

0.6

0.4

Rise Time 0.2

Settling Time

0

0

5

10

15

20

25

30

Time (seconds) 43


Power Plant Control System Performance Goals • Usually expressed in terms of maximum deviation allowed for a given transient • For example: – Maintain the final superheater outlet temperature within +/-15 degF of setpoint during a load ramp at 3% per minute. • With a goal like this, how would you tune the superheat temperature loop? – Probably would sacrifice response time to get more stability to prevent large overshoots.

44


22

Power Plant Control System Performance Goals • Important to have some overshoot in a step response. – Shows that gain is not much too low. • Good general performance goal: – Make response as fast a possible but keep overshoot less than 25%.

45


Power Plant Control System Performance Goals • Another goal to consider in loop performance is whether the loop’s main function is: – Setpoint tracking (variable setpoint) • Air flow control • Megawatt control

– Disturbance rejection (constant setpoint) • Also called a regulator • Drum level control • Steam temperature control

• Most setpoint tracking loops also do some disturbance rejection. • Performance tests should match main function. • In other words, don’t use a setpoint step response to test the drum level performance. 46


23

PID Tuning Concepts • Many tuning methods available, two main categories – Empirical (trial and error) – Analytical • Manual • Computer-aided

• Empirical is used much more than analytical. – Doesn’t mean it is better – Many tuners don’t know analytical methods • Should understand both

47


Simple PID Loop Empirical Tuning Procedure • Start with proportional mode. • Adjust integral and derivative gains to 0.0 or very small. • Set proportional gain small enough to ensure little control action. • Make a setpoint change and observe response. • If response is not significant, double the gain and make another setpoint change. • Repeat until response is significant (noticeable). • If loop is designed for disturbance rejection, introduce a repeatable disturbance and check response. If loop is designed for setpoint tracking, continue with setpoint steps. • Continue increasing gain until desired response is obtained. Doubling gain at each step is reasonable but finer steps will be necessary to finish tuning. 48


24

Proportional Tuning Example

Proportional Gain Tuning

•

Notes:

1.4

•

Offset decreases as gain increases

1.2

Doubling gain does not cause problems

•

“Best” gain between 0.2 and 0.4

•

No need to do test at a gain of 1.6 because 0.8 was already to high.

•

Could do another test at 0.3

•

Large offset at “best” gain.

Setpoint

1 Controlled Variable

•

Gain = 1.6

0.8

Gain = 0.8

0.6 Gain = 0.4

0.4 Gain = 0.2 Gain = 0.1

0.2

0

49

0

5

10

15 Time (seconds)

20

25

30


Integral Tuning Steps • Very similar to proportional tuning • Select a slightly conservative proportional gain setting. • Select a very low integral setting for first try. • Make setpoint step change and watch response. • If response is too slow, double integral gain and make another step in the setpoint. • As response get reasonable, change to a disturbance input if the loop is so designed. • Fine tune gain until response is as desired.

50


25

Integral Tuning Example

Integral Tuning Response 1.6

•

Notes:

•

Integral eliminates offset

•

Doubling gain does not cause problems

•

“Best” gain close to 0.2

•

Could try 1.5 and 2.5 as a final check.

•

Int Gain = 0.2

With proportional gain at 0.2 and integral gain at 0.2, loop is well tuned. Overshoot about 14%.

1.2 Controlled Variable

•

Int Gain = 0.4 1.4

1

0.8

0.6

Int Gain = 0.1

0.4 Int Gain = 0.05 0.2 Prop Gain = 0.2 0

0

5

10

15

20

25

30

Time (seconds)

51


Derivative Tuning Steps • If using a PID controller, tune P first and then D, then I. • Set D and I to zero or very small values. • Tune P as before. • Add derivative action to reduce oscillations. • Increase P to restore previous response. • Continue adding derivative and increasing P to maintain the desired response. • Stop adding derivative when controller output becomes too noisy. • Add integral to eliminate offset.

52


26

Derivative Tuning Example Derivative Tuning Example

•

Notes:

•

Derivative improves stability

•

Reduces overshoot

•

Proportional gain increased from 0.2 to 0.8 with no increase in overshoot.

•

Kd = 0.0

Simulation results not always realistic with derivative because there is no noise. No integral gain shown in this plot.

Kd = 0.2

0.8 Controlled Variable

•

Setpoint 1

Kd = 0.4 0.6 Kd = 0.8 0.4

0.2 Kp = 0.8 0

-0.2

53

0

5

10

15 Time (seconds)

20

25

30


Empirical Tuning Comments • It’s a trial-and-error process. • Requires some experience and skill to do efficiently. • Can be very time consuming on slow responding loops. • On fast loops, can be done in minutes.

54


27

Analytical Tuning Methods • Many methods are described in the references. • All require a simple process model to start. • Basic procedure – Perform open loop test on process, typically a step test. – Examine controlled variable response and identify process model – Use parameters from process model such as gain and time constant to compute PID tuning parameters according to formulas in a table.

• Zeigler-Nichols described two methods in 1942 – Ultimate gain method – Open loop step response method

• Zeigler-Nichols does not give good results for process control; too oscillatory. 55


Analytical Tuning Methods • Many others have developed new methods for process control systems. • Some are: – Lambda tuning – Internal model control – Dominant pole placement

• Some methods are designed for a particular type of process. • No single method is best for all processes. • Modern computer tuning programs provide many of these methods.

56


28

Analytical Tuning Example • Perform open loop step response test on process. • Draw straight line tangent to point of greatest slope. • Determine the value of the two parameters, a and L as shown below. 1

0.75

0.5

0.25

0

L = 0.8 a = 0.22

L

-0.25

a

-0.5 0

57

1

2

3

4

5

6

7

8

9

10


Analytical Tuning Example • From previous plot, L = 0.8, a = 0.22. • Use table of tuning parameters below to determine setting for P, I, and D. • Tuning should give maximum speed of response with no more that 20% overshoot.

Controller

K

Ti

P

0.7/a

PI

0.7/a

2.3L

PID

1.2/a

2L

Td

0.42L

Table from Chien, Hrones and Reswick disturbance response method as shown in Astrom and Hagglund, p. 150. 58


29

Analytical Tuning Example • From previous table, for a PID controller the tuning would be: • K = 1.2/a = 1.2/0.22 = 5.45 • Ti = 2.0L = 2.0*0.8 = 1.6 • Td = 0.42L = 0.42*0.8 = 0.336

• With this tuning, the disturbance rejection response would be as shown below. Disturbance Rejection

1

0.8 0.6

Controlled Variable

0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1

0

5

10

15

20

25

30

Time (seconds)

59


Analytical Tuning Comments • Analytical tuning can yield good results. • Requires open loop step response test to quantify process characteristics. – Some methods only use two parameters – Others use three (more is usually better) • Must know the PID controller form and units. • If PID form does not match tuning table, must convert values to proper units. • Notice from the table, that use of a PID controller allow a higher P gain than with a PI controller.

60


30

PID Tuning Software • Several popular packages available for PID controller tuning. • Most connect to a DCS to collect live loop data. • Identify process model using a variety of test signals such as step, pulse, double pulse, binary sequence. • User selectable PID tuning algorithm. • For supported DCS platforms, the programs understand the details of the PID algorithm and make sure units are correct. • Allow simulation of new tuning parameters before putting them in service. • EPRI Report 1004080 provides additional information. 61


System Linearity Issues • A linear system is easier to control than a non-linear system. • Linearity definition: – If input 1 produces output 1, and input 2 produces output 2 then (input 1 + input 2) produces (output 1 + output 2).

• No power plant systems are strictly linear, but some come close. • Some common sources of non-linearity include: – – – – 62

Valve and damper flow characteristics Centrifugal pump curves Pressure drop vs. flow relationship. Backlash in actuators Copyright © 2004 Electric Power Research Institute, Inc. All rights reserved.

31

System Linearity Issues • Easy to check for linearity using an input/output • Plot manipulated variable vs. controlled variable at different loads under steady state conditions. • If data is not in a straight line, the system is non-linear. • Slope of line at any point indicates steady state process gain at that point. • Change in slope of the line indicates degree of non-linearity. • Easy to compensate for this type of non-linearity using function generators in the DCS. • Better to make field devices as linear as possible first.

63


System Linearity Example System Linearity Examples 120

Controlled Variable (Output)

100

80

Not Too Bad 60

Pretty Bad 40

20

0 0

20

40

60

80

100

120

Manipulated Variable (Input) 64


32

Characterization Example Linear

Non-linear Non-linear

Setpoint

+

PID

F(x)

Actuator

Process

Controlled Variable

-

Measurement

Actuator

Result 100

90

90

80

80

80

70

70

70

60

+

50 40

=

60 50 40

Controlled Variable

100

90

Controlled Variable


F(x) 100

60 50 40

30

30

30

20

20

20

10

10

0

0

0

20

40

60

80

100

65

10

0

Controller Output

20

40

60

80

100


0 0

20

40

60

80

100

Controller Output


Gain Scheduling • Gain scheduling is another way to deal with non-linear loops. • Modern DCSs allow controller gains to be defined as a function of another variable. • Enables consistent tuning throughout the load range. • Drawback is that it requires more tuning tests. • Tough enough to get a loop well tuned at one load, let alone 3 or 4 loads.

66


33

Boiler Control Loops • Unit Master - Megawatts and Throttle Pressure • Fuel Control • Air Control • Furnace Pressure Control • Feedwater/Drum Level Control • Excess Oxygen Control • Steam Temperature Control • Pulverizer Control • Miscellaneous Control Loops – Oil Temperature Control – BFP Minimum Flow Control – APH Cold End Temperature Control 67


Approach to Boiler Control Tuning on a New Unit • Verify proper manual operation of all final actuators. • Tuning a boiler is an iterative process that requires a few iterations. • Initial tuning is done to provide stable operation with little concern for responsiveness. That will come later. • Perform initial tuning on single-element drum level and furnace pressure control first, since these are the most troublesome to the operator. • At this point, operator should be able to load unit manually. • Characterize fuel flow, air flow, steam flow, and feedwater flow at several load points. Temperatures, pressures and excess oxygen should be at design and unit must be at steady state.

68


34

Approach to Boiler Control Tuning on a New Unit, continued • Do initial tuning of flow loops, air, fuel and feedwater. • Tune three-element drum level control. • Tune pulverizer controls. • Tune boiler-follow throttle pressure control. • Tune steam temperature control • Tune coordinated control • At this point, load on the unit can be ramped automatically. • Go back to beginning and check each loop’s tuning using load ramps as evaluation criteria.

69


Control Tuning Hierarchy Unit Master

Throttle Pressure

Megawatts

Turbine Master

Boiler Master

Excess Oxygen

70

Fuel Control

Air Flow Control

Feeder Control

...

Feeder Control

FD Fan Control

FD Fan Control

Actuators

...

Actuators

Actuators

Actuators


35

Actuator Setup and Tuning • Start at the bottom and work up. • Actuator response very important to overall control loop performance. – – – –

Should be linearized in the actuator and linkage if possible. No slop in linkage Pneumatic actuators have stiction Electric actuators have deadbands

• Electric actuators have position controllers that may need tuning to get good response.

71


Feedwater/Drum Level Control Feedwater Flow

Steam Flow

Drum Level DP

Drum Pressure

f(x)

f(t)

A

Σ

PID Drum Level Controller

FW Flow Controller

PID 3-Element Control T

A

BFP Demand 72


36

Drum Level Tuning Notes • Single element is useful at low loads when reliable feedwater flow measurements are not available. • Calibrating compensated drum level measurement can be a challenge. • PID controller tuning will be mostly proportional. • Very little integral can be used without causing excessive oscillations. • If level measurement is not too noisy, derivative may improve response somewhat. • Don’t try to get response too tight at first. Err on the side of stability. • May need to lag steam flow feedforward for about 30 seconds to prevent problems due to inverse response of some drums. 73


Furnace Pressure Control Furnace Pressure

Furnace Pressure Controller

Air Flow Demand

f(x)

PID

Σ

T

A

Spray Valve Demand 74


37

Furnace Pressure Control Tuning Notes • Furnace pressure measurement is very noisy. • Filtering is necessary to deal with it. – Gap on error signal is one way. – Time lag not the best due to impact on loop response time. • Simple PI tuning. • Check linearity of ID fan dampers. • Feedforward from FD fans helps dynamic response. • Don’t use air flow as feedforward because ID fans influence air flow. • Lots of additional interlock logic on the ID fans to prevent furnace implosions.

75


Air Flow and Excess O2 Control Air Flow

Fuel Flow

Boiler Demand

Excess Oxygen

Steam Flow

f(x)

>

Cross Limiting

PID

Air Master

T

Programmed O2 Setpoint

f(x)

× Air Flow Controller

< PID

T

O2 Controller

A

A

FD Fan Demand 76


38

Air Flow Control Tuning Notes • Simple PI control tuning • If control mechanism is inlet vanes, check linearity over the range of the fans. • Air flow measurement will probably be too noisy to allow effective use of derivative. • Key to good air flow control is proper calibration of the demand signal because oxygen analyzers are quite slow. • Biases should be added to cross-limiting signals to prevent unnecessary blocking action. • Oxygen controller is a ratio trim controller with a range of +/- 30% typically. 77


Fuel Control Total Fuel Flow Feeder Speeds or Heat Release

Boiler Demand Air Flow Cross Limiting

< PID

A

T

Fuel Master (Optional)

Feeder Demand to Feeders

78


39

Fuel Control Tuning Notes • Oil and Gas – Simple PI flow control tuning. • Coal – Initial tuning should just use feeder speed as the fuel measurement. Almost trivial. – Later tuning should use “heat release.” – Heat release is rate of change of drum pressure plus steam flow. – Inferential measurement of fuel flow in the furnace. – Much slower than feeder speed but much more accurate.

79


Boiler Follow Unit Master Megawatts

First Stage Pressure Dispatch

Throttle Pressure

A

÷

Ramp Up/Ramp Down/Hold

×

PID

A PID

Unit Demand

Σ

Σ

Turbine Master

T

A

Turbine Demand 80

A

T

Boiler Master

Boiler Demand


40

Throttle Pressure Control Tuning Notes • Initial tuning of pressure controller for stability at steady load. • Feedforward is an important element of good pressure response. • Don’t use steam flow as a feedforward on a coal-fired unit. It is regenerative (positive feedback). • Pressure ratio is very good feedforward signal. Represents effective turbine valve area. • Important to have dynamic compensation on feedforward signal (kicker) to overcome energy storage changes with load.

81


Steam Temperature Control Desuperheater Outlet Temp

Final Steam Temp

Feedforward Signals

f(x) A PID Desuperheater Outlet Controller

PID

T Spray Valve Demand 82

Final SH Outlet Controller

Σ A

Typical Feedforward Signals: Steam Flow, Drum Pressure, Air Flow, Burner Tilt Position


41

Steam Temperature Tuning Notes • Steam temperature is usually the toughest loop to tune. • Many sources of disturbances combined with very slow process response. • Classic cascade control structure. • Tune inner loop first with about 20% overshoot. • Tune outer loop with only slight overshoot. Derivative can be used quite effectively here. • Best feedforward signal(s) must be determined by test. Determine which other process variables have the most impact on steam temperature. • Make sure reset limiting is working properly on outer controller. Normal built in capability usually won’t work correctly.

83


Typical Parallel Loading Unit Master Megawatts

Operator Dispatch

Throttle Pressure

A

Ramp Up/Ramp Down/Hold

A

PID

PID Unit Demand

Σ

Σ

Turbine Master

A

T

Turbine Demand 84

A

T

Boiler Master

Boiler Demand


42

Unit Master Control • Also called Front End Control. • Load setpoint from operator or load dispatch system. • Throttle pressure setpoint fixed or sliding. • Outputs are Boiler Demand and Turbine Demand. • Several strategies in use, most called Coordinated Control. • Several modes, Coordinated, Boiler Following, Turbine Following. • Also handles runbacks and rundowns. • Lots of logic involved.

85


Unit Master Tuning Notes • This is a two input, two output system with a lot of interaction between the megawatt control and the pressure control. • Turbine valves provide fast response while firing rate provides much slower response. • Dispatcher would like tight megawatt control while plant usually favors tight throttle pressure control. • Most implementations allow for trade-off between the two loops to achieve desired balance.

86


43

Boiler Control Notes • Boiler control uses many cascade loops which complicates tracking and anti-reset windup. • Many loops have multiple final drives. Check participation and gain changing.

87


Turbine Control Systems • Provide speed control during startup. • Provide frequency control when on-line. • May require special I/O card for interface with hydraulic actuators. • Fast response needed for overspeed protection. • Enhancements such as: – Bumpless single/sequential valve transfers. – Automatic startup • Speed control tuning usually done by OEM. • MW control is a part of the boiler control system.

88


44

Thank you and The End !!

45

About EPRI EPRI creates science and technology solutions for the global energy and energy services industry. U.S. electric utilities established the Electric Power Research Institute in 1973 as a nonprofit research consortium for the benefit of utility members, their customers, and society. Now known simply as EPRI, the company provides a wide range of innovative products and services to more than 1000 energy-related organizations in 40 countries. EPRI’s multidisciplinary team of scientists and engineers draws on a worldwide network of technical and business expertise to help solve today’s toughest energy and environmental problems. EPRI. Electrify the World

© 2004 Electric Power Research Institute (EPRI), Inc. All rights reserved. Electric Power Research Institute and EPRI are registered service marks of the Electric Power Research Institute, Inc. EPRI. ELECTRIFY THE WORLD is a service mark of the Electric Power Research Institute, Inc. 1003740 Printed on recycled paper in the United States fA i EPRI • 3412 Hillview Avenue, Palo Alto, California 94304 • PO Box 10412, Palo Alto, California 94303 • USA 800.313.3774 • 650.855.2121 • [email protected] • www.epri.com

EPRI Control

Recommend Documents