L16 - Model Predictive Control Within the Logix Controller Lab · Metal strip. DV2. Combustion gas exhaust. Metal strip. Hot air inlets. Suction flow. Pressure (constant) DV2. DV1

PUBLIC

Copyright © 2018 Rockwell Automation, Inc. All Rights Reserved. 1

L16 - Model Predictive Control Within the Logix Controller Lab

Optimization in the Logix Chassis For Easy Deployment

PUBLIC Copyright © 2018 Rockwell Automation, Inc. All Rights Reserved. 2

Agenda

Course Feedback

Lab Overview – Go!

What is Model Predictive Control?


What Makes MPC Different

We do this because it makes processors money:Higher production, yields, efficiency and product quality

SINGLE INPUT - SINGLE OUTPUT CONTROLLER

PID MPC

CONTROL BASED ON CURRENT ERROR

POOR ABILITY TO HANDLE PROCESS DELAYS, & NON-LINEARITIES

POOR ABILITY TO HANDLE DIFFERENT TYPES OF DISTURBANCES AND SETPOINT SIGNAL FORMS

POOR ABILITY TO HANDLE CONSTRAINTS

MULTIPLE INPUT - MULTIPLE OUTPUT CONTROLLERA centralized approach. All variables are simultaneously considered

PREDICTIVE CONTROLController action based on current and anticipated future PV deviations from target

COMPENSATES FOR PROCESS DELAYS & NON-LINEARITIES

OPTIMAL CONTROL FOR ALL TYPES OF DISTURBANCESAND SETPOINT SIGNAL FORMS

PREDICTIVE HANDLING OF CONSTRAINTS


MPC Opportunities

Coating oven Distillation column Compressor Chemical reactor Boiler Furnace Dryer Drilling rig and many more


Making Control Scheme Smarter

6

M

PID left in the scheme or direct output control

M

Trad

ition

al C

ontr

ol A

rchi

tect

ure

Easier to design and maintain

Arch

itect

ure

with

MPC


now

CV

MV

∆MV

1 Horizon

CVSP

MVSP

0

How a Receding Horizon Works

measuredcalculated (predicted)calculated and sent out as MV

CVConstraints

CV MVProcess

ControllerModel

CV predictionMV prediction

MPC

Setpoints

( )

( )

( )∑

∑

∑

−

=

−

=

=

∆+

−+

−=

1

0

2

1

0

2

1

2

)(

)()(

)()(

Horizon

kMoveSupp

Horizon

kSPCoef

Horizon

kSPCoef

kMVMV

kMVkMVMV

kCVkCVCVJ

The Task: Minimize J by calculating MV over the Horizon

0

1 Horizon0

Horizon0 1


PID Versus MPC Variables

CVCVSP

MVSP

DVDV

MV

CV

unmeasured disturbance

MPCProcess

PVSP CV

PV

unmeasured disturbance

PIDProcess

CV – Controlled (Process) Variables: objective and/or constraints

MV – Manipulated Variables: PID targets or outputs that MPC adjusts to meet objectives

DV – Disturbance Variables: affects CVs, but is independent of controller


• Design the MPC Application:• Controlled Variables: What are we trying to do better, what is in the way/limits this today?• Manipulated Variables: What can we adjust to affect this result today?• Disturbance Variables: What do you watch out for that causes/forecasts a change?

• Perform Plant Testing (set up trends/logs for above variables)• Individually step each MV and DV (as possible) within safe limits to see changes in CVs

• Develop MPC models based on identification of above data• Develop any calculations necessary to support MPC goals• Implement/integrate and test application, train operators and turn-on!

Measure improvement!

A MPC Project


MPC in Integrated Architecture®

10

FactoryTalk® ProductionCentre®, FactoryTalk® Historian, Pavilion® Real-Time Optimization™ software…

Pavilion8® MPC, Software CEM®, VOA®, Asset Management, …Rockwell Software® Studio 5000® environment,PlantPAx® MPCBuilder, PlantPAx® ModelBuilder

• PlantPAx® MPC• IMC,CC,MMC• FuzzyLogic• Soft Sensor® (AOI)• PID,PIDE• Motion• Discrete


PlantPAx® MPC Lab Setup

Controller 1756-MPC Module with firmware

Lab Files:

Oven.ACD

OvenDataSet.CSV

Oven_ModelReady.mpc

Oven_SimulationReady.mpc

Oven_ExportReady.mpc

0 1 2

• PlantPAx® MPCBuilder

• Studio 5000® environment

• RSLinx® Classic

• FactoryTalk® View SE Client

192.168.1.112

192.168.1.113


Dynamic Identification


Combustion Gas Suction

MV0

DV3

CV0Oven_Temp

CV1Oven_Press

Ambient_Temp

MV1

HeatingChamber

rpm constantCombustion gas circulation

Atmospheric pressure (constant)

Curing Oven MPC – 2 MVs, 4DVs, 2CVs

Metal strip DV2

Combustion gas exhaust

Metal stripHot air inlets Suction flow

Pressure(constant)

DV2

DV1

MV0

MV1

DV0

CV1

Air_Flap_Valve

Gas_Valve

Suction_Flow

Hot_Air_Temp

Hot_Air_Press

Oven_Temp

Oven_PressAmbient_Temp

DV3

CV0

Hot Air Supply

Gas Supply

Hot_Air_Temp

Hot_Air_PressDV0DV1 Air_Flap_Valve

Gas_Valve


MPC Configuration & Settings

Oven

DV2

DV1

MV0

MV1

DV0

CV1

Air_Flap_Valve

Gas _Valve

Suction_Flow

Hot_Air_Press

Hot_Air_Temp

Oven_Temp

Oven_PressAmbient_Temp

DV3

CV0310 °C

340 °C

60 50 260 s

2 °C/s

21 65

max -1 Pa

min -20 Pa

≤ 100%

0%

100%

0%

max 600 °Cmin 550 °C

max 250 Pamin 200 Pa

max 40 °Cmin 27 °C

max 0.3 m3/smin 0.2 m3/s

keep pressure within limits

track temperature setpoint

OversampleDT – MPC frequency 0.5 seconds

SPValueReq - TrueSPProg 0% ROC 10%/sScale 100%

ROC 10%/sScale 100%

SPTrajectoryReq - TrueCoef 100Scale 100 °C

ZoneReq - TrueCoef 10Scale 10 Pa

CVSPTrajectory0 s 310 °C6 s 310 °C

21 s 340 °C50 s 340 °C65 s 310 °C

260 s 310 °C

ZoneHiLimit -1 PaZoneLoLimit -20 PaDV2


Do Not Skip Faceplates


PlantPAx® MPC Configuration Faceplates MPC Step Test AOI MPC Model Quality AOI MPC Control Status AOI

PP MPC Process Library AcceleratorsTrack 2 AOI and Faceplates: if you have already completed Track 1

PlantPAx® MPC Faceplates

PlantPAx® MPC Status

PlantPAx® MPC Step Test SupportPlantPAx® MPC Model Quality

www.rockwellautomation.com

PUBLIC

Copyright © 2018 Rockwell Automation, Inc. All Rights Reserved. 17

Thank You!

Documents

L16 - Model Predictive Control Within the Logix Controller Lab · Metal strip. DV2. Combustion gas exhaust. Metal strip. Hot air inlets. Suction flow. Pressure (constant) DV2. DV1