Industrial Automation: Situation and Trends

©A

BB

Gro

up -

1-

26-A

ug-0

6

Industrial Automation:Situation and Trends

IFAC 50th AnniversaryHeidelbergSeptember 15, 2006

Peter TerwieschChief Technology Officer

ABB

©A

BB

Gro

up -

2-

26-A

ug-0

6Overview

� Industry needs

� A visual history of industrial automation system evolution

� Industrial application examples

� Summary and outlook

©A

BB

Gro

up -

3-

26-A

ug-0

6Industry Needs

Global competition: � Improve

productivity� Deliver flexibly,

better, more …

� … with less: economically, and ecologically

� Focus capital spending,improve asset utilization

Time

Throughput

Lost opportunity

Plant upset

Startup

Partial shut-down

Design limit

Loss

Pro

fit

Shut-down

Break Even

Operatingtarget

©A

BB

Gro

up -

4-

26-A

ug-0

6

©A

BB

Gro

up -

5-

26-A

ug-0

6

©A

BB

Gro

up -

6-

26-A

ug-0

6

©A

BB

Gro

up -

7-

26-A

ug-0

6

©A

BB

Gro

up -

8-

26-A

ug-0

6

©A

BB

Gro

up -

9-

26-A

ug-0

6

New technology enables ...• Standard Communications• The ability to interact with a device and know

the process better• Distributed intelligence• Reduced cost per function

Time

Func

tiona

lity

Fieldbus

Fieldbus

More valuesMulti variable

High resolutionDiagnostic dataQuality signal

Status De-central FunctionsDistributed Control

Bi-directional Asset Optimization

Graphics

SMART

HART

ValueDevice

Parameter

4-20mA

Traditional

Value

Advanced diagnostics

and Asset management +

multi systems in

tegration

Intelligent Field Devices

©A

BB

Gro

up -

10-

26-A

ug-0

6Automation Industry Consolidation

©A

BB

Gro

up -

11-

26-A

ug-0

6IT: Challenge and Opportunity

Customizedsolutions

Standardizedproducts

Build on fastdevelopment in

IT and electronics

Well-proven technology and 15-20 years life time

©A

BB

Gro

up -

12-

26-A

ug-0

6Process trends: increased integration

Process buffers at all levels get trimmed: supply chain, company, site, plant, unit

©A

BB

Gro

up -

13-

26-A

ug-0

6Process trends: tightening constraints

$$$

$$

Constraints and boundaries tighten, consequences of violation

become increasingly severe

©A

BB

Gro

up -

14-

26-A

ug-0

6Trend: broader scope, more complex models

DI542I271

T2

HC2

HC3

P-60

Press7+Blekeri2

P-112 P-113

P-117

V-1

P-123

P-144

Vatten

E-99

P-154

P-167

V-11P-214

LC542I286 lab

542Q110

FI421I209

FCA51424I312

LI424I450

LI424I460

FI424I301

FI424I302

FI424I303

O626

CI221I328

221Q105

©A

BB

Gro

up -

15-

26-A

ug-0

6

� 620,000 ton� Three fiber lines� Six paper machines

� 620,000 ton� Three fiber lines� Six paper machines

ContainerboardSackKraftMarket pulp

Trend: broader scope, more complex models

542Q 124

O799

411 Q 101

412Q 101

54 2Q 126

DI54 2 I271

54 2Q129

542 Q 128

LV A02541 I901

LV A 06542I901

LV A01541 I901

FI541 I6 47

TLB

TLL

KV

SU LF

SP LO X

SV L

VL

M ESA

KAL K

G L 3

G L2

G L1

TJL T5

TJ L

U TSK

H C 4

N S M

K VM

BT

T1

H C 1

T18

T 3

T2

H C 5

T1 7

H C 2

H C 3

R L

SK 2

P -2

P -3

Tvätt+S ileriT vättvatten

SK2

P -9P-10

P-11

P -12

O 2

P -13

NaO H

Tv.vä tskaBV 24

P-22

P -23

Flis

V itlu t

Indu nst.

P -28

P-33

P -34

P-35

SK 1

P -36

BT

Tvätt+S ileri

P-37

Na/S fö rluste rI m assan

pp

B lekeri1

Vatten

P-43

P -44

P -45 Skrubberstark gas-

de str

P -51

P -53

P -54

Rö kgaser

P -56

P-57

P D 11

P -58

P -59

P -60

M ix-P D 31

P -61

P -62

V a tten

M esaugn+Filter

P-6 4

P -65

V a tten

P -70

P -71

Sp i ll, d iv

SO 2

P -75

Vitlu tsO x

P -77

Svavelu gn+A bs.to rn

P -79

S vave l M ake-up

NaO H M ake-up

P -83

P -84

N S

P -8 5

Flis

P -88P-8 9 P M 6

P -90

P-91

P -92

P -9 3

H arts-kokeri

TO

BO

P -104

P -106

Björko lja til l M es augn sfö rb ränning

P-108

P-109P -110

T vättvattenSK1

P-111

Press7+B lekeri2

P -112 P -113

P-11 7

P -118

T19V-1

T ryck-d iffusör

O blekt f ibe r

P -120

P-121

P-123P -124

P-125

V -2

P -126

V-4

P-128

P -129

P -130

P -131

V -5

P -134

P -135

P -136

V -6P-140

P -141

Flis

P -144

Va tten

BU L

BL L

P -148

E -87

E lfite raska

E -88

E-89

V-7

P -150

P -152

P-151

E -91

E-92

E -95

E-96

P -135

E -97

E-99

P -154

P -131E -100

P-15 6

E -103

P -60

E -104P -57

E -10 6

NaO H

H 2S O4

P -161

V itlu t til l Ha rtskokerisk rubber

P -167

Filter

S ile r ifö rlus t

S ile r ifö rlus t

P-171

S urva tten ti ll C lO 2

Ka lkg rus

P-17 4

S vave l i va rm va ttenBLB

P -179

P -180

P-182

E -112

P -34

E -113

Ta l lo lja fö r avsa lu

M gS O 4

P -1 88

P -189

P -189

P -191

B V25

P-192E -124

P -161V-8

P -193

P -181

Kondensa t

P-194

P-195

P-1 96

V -9

S lam

V-10

P -193

E -12 6

P -200

P -201

P-203

NaO H

P -205

Soda-p anna+

sk rubber

P -206

P -207

Vatten

P -210P -211

P -212

M eO H

V -11

P-214

P -215

P -216

P -217

FC A01542 I216

FC A01542 I212

FC A01542 I220

L C542 I548

P-218

F I542 I263

F C A 0154 3 I301

F C542 I291

LC5 42 I293

LC542 I286

F I523 I626

LC542 I285

FI523625

FC542 I282

L I543I3 26

L I543 I328

F C542 I55 0

FI542 I506

LC542 I223

L I542 I247

LI542 I24 9

DCA 0 1523 I62 1

F C54 2 I220

F C54 2 I212

F C54 2 I216

LV A04542 I901

LV A04542 I90 4

LV A 04542 I907

54 2Q 123

F C543 I301

DC543 I302

AI543 I360

AI543 I322

P-219

F CA 02543I315 F C A01

54 3I315

F C5 43 I306

FI523I627

LC542 I457

LC542I458

lab542Q102

lab542Q 1 01

lab542 Q 110

lab52 3Q 101

lab52 3Q 102

lab52 3Q 113

A I5 23 I531

A I5 21 I532

A I5 21 I535

FC52 3 I513

LC541 I621

FCA 01541 I62 0

F C541 I620

A I541 I614

AI501 I644

F C541 I643

F C541I645

FC721 I101

S C721I2 47

FD A 02523 I577

A02523 I577

DI523 I5 83

FIA0 1523 I7 22

L I5 13 I597

L I5 13 I592

LC523 I5 40

L I513I124

LC51 3 I718

L I512 I179

L I512I177

FC A 01512 I3 15

FCA 01543 I315

FC51 3 I721

D C51 3 I717

L I501 I8 70

L889L887

FC512 I183 P -220

F C513 I735

FC51 3 I709

FC5 13 I714

F I421 I499

FI513 I705

F C721 I159

M VI4 95

F C513I706

FIA 01512I187

FI51 3 I723

FC513 I109F C

513 I733FI

513 I733LV

513 I911

D I5 13 I437

LV A015 13 I912

LV513 I903

L I513 I710

L I513 I736

L V513 I901

L I51 3 I707

A01-SP721 I127

A01-SP721 I129

TC -S P721 I127

TC -E G U721 I127

S C721I373

FI721 I327

L I731 I197

A01721 I331

A01721 I341

A 01721 I351

FC73 1I166

L I731 I177

FC731 I113

L I731 I101

6 ICO .A RTO N H

FI303 I301

SC A 51421 I162

F C542 I4 41

F C421 I37 5

F C421 I409

FC4 21 I410

F C421 I464

F C421 I8 60

E -130

P -22 1

P -222

FC421 I103

FC421 I441

M 51O 472

P -2 23

F I42 2 I641

FC422 I510

FC422 I509

F C422 I6 18

F C422 I6 54

FCA 5 5422 I55 9

FC A5 1422I55 9

F CA 5 1422 I565

FC A5 3422I56 6

FC51422 I258

FCA 51422 I255

L I4 22 I501

L I4 22 I506

FC A51421I441 F C A51

422 I642

F C A51422 I622

L I42 2 I670

LC422 I6 29

LC422 I63 9

F CA 5 2422 I6 42

F CA 51422 I618

FC A5 1422 I65 4

T 101

F C421 I428

LC422 I552

T 114

F C 05142 1 I103

LC4 21 I445

FC A52421 I428

T1 21T178

A 51T107

T114

A 51T 103

T 111

T 105A 51

T 119

A51T171

A51T 233

A5 1T25 8

T226

L I424 I309

va tten

T269FI

421 I2 09

F C A 5142 4 I312

L I424 I450

L I424 I460

F I4 24 I301

FI42 4 I302

FI424 I3 03

LI611 I6 11

O 7 79

O 626

FI304 I409

O 101

M 51O 101

A51O 258

O 258

O 271

F C411 I11 8

FC411 I25 8

O 331

O 425

M 51O 333

O 401

O 334 O 333

O 404M 51

O 143O 602

FC412 I749

L IC 625

F C413 I468

C526

L I413I571

FA 01413 I468

611 .dry ou t

P -22 9

54 2Q 114

DC513 I762

DC513 I761

L I513 I7 01

DC513 I7 34

LC51 3 I731

L I513 I738

DI513I34 5

LC51 3 I772

D I513I8 23

L I513 I798

L I513 I792

A IM V 1523 I578

FCA 03542 I2 12

421Q 105

421Q 103

M 8 56

FC A0 1523I28 4

542Q 119

FC A 52423I105

A51423 I119

X D423Q 113

CI22 1 I328

L V A0154 2 I903

L V A0154 2 I906

L V A0154 2 I909

M IHA S T523T0 27

F I523 I5 10

F I5 01 I764

413Q 109411Q 10 6

41Q 105412Q102

O 473

O 648

O 775

42 2Q 104

LC541 I658

lab721 I90 1

L I721 I18 6

541Q 103

541Q 101

LV721 I90 2

541Q 102 54 1Q 106

54 1Q107

54 1Q 109

541Q 104

541Q 105

A05721 I101

A01721 I101

FC541 I628

FC501I683

F C501 I761

F C523 I514

FI512 I1 63

541I6 13

P -230

FC A51421I428

22 1Q 105

P -23 1

Q C413 i467

FC411 I260

M VO 110

A I501 I644

AI501I633

Example:Pulp Mill

©A

BB

Gro

up -

16-

26-A

ug-0

61960s: Production optimization at Gruvön in

� Licentiate thesis by Bengt Pettersson

� Figures courtesy of Karl Johan Åström

Example:Pulp Mill

©A

BB

Gro

up -

17-

26-A

ug-0

6

AIP Server

( ) kkkk wuxfx +=+ ,1

Connection of units, pipes and measurements

FC542I441

SQL Server

Par

amet

ers

( ) kkk vxgy +=MeasurementMeasurement

1st order 1st order massmassbalancebalance

Today: Nonlinear Dynamic Model

SK1

Dynamicmass balance

inletPulp

inletSteam

inletWL

inletWater

outletNCG

Lockmannkolonn

outletBL

outletCondensate

outletPulp

67

8

Diffusören är inbakad i kokarenFlash 1-spliten är en intern loop

25 Production units38 Buffer tanks250 Streams250 Measurements2500 Variables

Example:Pulp Mill

©A

BB

Gro

up -

18-

26-A

ug-0

6

State EstimationOptimization

Simulation

Each cycle approx. 30 minutes

On-line operationExample:Pulp Mill

©A

BB

Gro

up -

19-

26-A

ug-0

6

Configuration of a heat recovery steam generation plant

Superheatersubsystems

NMPC: Minimizing Thermal Stress at Boiler Start-up

Thick-walled steam header (cylindric type, 6.4 cm thick)

Stretching of the material because of temperature gradients lead to lifetime reduction

Example:Boiler Startup

©A

BB

Gro

up -

20-

26-A

ug-0

6Exemplary Water-Steam Cycle

SPW-Druck vor HD-VorwärmerRL10P002

HD-Vorwärmer

Verdampfer

SPW-PumpeEco

Abscheider4

EK I/1EK I/2

Dampftemp vorEK I/2 NA12T007

pD Anfahrflasche NB01P003

Temp Speisesewasser vorVerdampfer NA01T001

Temp Speisew vor ECORL15T004

ECO Speisew.-Menge RL15F001

Niveau AnfahrflascheNB01L901

UmwälzmengeNB05F901 Summe Menge

Notabläufe RK10F001

EK IV/4

EK II/4

Dampftemp nachEK II/4 NA24T008

Dampftemp vor EKIII/4 NA34T004

RV EK II/4RL24S002

Dampftemp. vorEK IV/4 NA44T004

RV EK IV/4RL44S002

Dampftemp nachEK IV/4 NA44T008

Dampftemp nachÜ V RA04T004

Ü V

Ü IV

Ü III

Dampftemp vor EK II/4NA24T004

Ü II

EK IV/1

EK II/1

Dampftempnach EK II/1

Dampftempvor EK III/4

RV EK II/1

Dampftempvor EK IV/1

RV EK IV/1

TempDampftemp

EK IV/1

Dampftempnach Ü V

RA01T004Ü V

Ü IV

Ü III

Dampftempvor EK II/1

HD

Stellungsbefehl Y HDU 1...4 RA08C000

HD-Einlaß-RV TurbineSA11S016-46

FD-MengeRA07F901

Ü I

DT Sammler II/4 (Innen- und MittelfaserNA24T010I und NA24T010M)

DT Sammler III/4 (Innen- und MittelfaserNA34T010I und NA34T010M)

DT Sammler IV/4 (Innen- undMittelfaser NA44T010I und NA44T010M)

FD-DruckRA07P901

SPW-Behälter

ND-Vorwärmer

Kondensatpumpe

MD/ND G

MD-RVTurbine

Zwischenüberhitzer (ZÜ)

MDU

Druck kalte ZÜRC12P007

FD-TemperaturRA07T901

GeneratorleistungSP10E927

RV EK I/1RL11S002

RV EK I/2RL12S002

RadkammerdruckSA11P020

ZÜ I

ZÜ II

DT Kugelstück (Innen- und Mittelfaser03RA05T002I und 03RA05T002M)

Anordnung ist 4-strängig(4 HDUs, 4 HD-Turbinen-

RV)

DT Hosenstück (Innen- und Mittelfaser03RA07T003I und 03RA07T003M)

7

5

4

3Kombination Strahlung/Konvektion

6

2

1

8

Dampftemp n EK I/1NA11T007

zu den Abscheidern 1 - 3Dampftemp nach

Abscheider 4 NA16T005

Hauptentwässerungen

RA 80

VerdampfermengeNA01F001

RG-Temp v Ü IV NR04T010 und 20

RG-Temp v Ü VNR05T010 und 20

1 8 rauchgasseitige Reighenfolgeder Überhitzer


©A

BB

Gro

up -

21-

26-A

ug-0

6Object-oriented modeling

� Graphical flowsheet modeling

� Combine commodity models with proprietary knowledge

� Automatic generation of stand-alone executable code


©A

BB

Gro

up -

22-

26-A

ug-0

6Dynamic optimization problem

niTtTT

tY

qqqqqq

xxtYtYtqtxftx

ts

dtw

qtq

wptp

wTtT

J

iii

HPB

FFFFFF

AWHPBF

tYtq

t

t q

setmm

p

set

T

set

HPBF

f

m

,...,1,)(

1)(0

,

)0()],(),(),(),([)(..

min])([])([])([

max,min,

max,min,max,min,

)(),(0

2

2,

2

22

2

=∆≤∆≤∆

≤≤≤≤≤≤

==

→−

+−+−= �=

��

�

General startup optimization problem:Optimal transition to new operating point (T,p,qm) subject to control bounds and

state constraints (esp. thermal stresses).

� Process model

� bounds on fuel

� bounds on valve positions

� thermal stresses


©A

BB

Gro

up -

23-

26-A

ug-0

6Solution Approach

� Parameterize control u(t) = u(uk), k=0,1,…,K-1

� Convert dynamic optimization problem to discrete-time optimal control problem:

0)(

1,,0,0),(

1,,0),,(

..

min),()(

1

,

1K

0k00 0

≥−=≥

−==

→+=

+

−

=�

KK

kkk

kkkk

ux

kkK

xc

Kkuxc

Kkuxfx

ts

uxfxfJk

�

�

� Solve as large-scale nonlinear programming problem with vector of optimization variables v:

��

�

�

��

�

�

=

−1

1

0

Ku

u

u

v�

�

��

�

�

��

�

�

=

−

−

K

K

K

x

u

x

u

x

u

x

v

1

1

1

1

0

0

�


©A

BB

Gro

up -

24-

26-A

ug-0

6Organizing Information With Aspect ObjectsTM

AspectObject

Aspects

RealObject

Aspect Systems

Inventory reportMicrosoft

Excel

MicrosoftWord

SAP, IFS . . .

Dynamic Optimization

Operationalprocedure

Productionschedule

Productspecification

MechanicalDrawing

TransitionOptimization

AutoCAD


©A

BB

Gro

up -

25-

26-A

ug-0

6NMPC Integration with Control System

NMPC with 800xA� Seamless integration with

operator station � Data serving (Aspect

Server, Basic History)� Connectivity to process

databases� Connectivity to major

control systems� Further synergies,

e.g. alarm handling


©A

BB

Gro

up -

26-

26-A

ug-0

6Conventional ramp start-up

� Classical start-up: apply ramp on fuel flow rate q_F

� Thermal stress violates constraint (-60 instead of –25)


©A

BB

Gro

up -

27-

26-A

ug-0

6Optimized start-up

� Result: achieve faster startup and fulfill constraints

� Seamless integration with Trend&Historian subsystem


©A

BB

Gro

up -

28-

26-A

ug-0

6Thickness Control in Cold Rolling Mills

� Conventional control:� Single control loops

� Feed forward strategies

� Limited performance

� Task and objectives:� Improve tolerance over

the whole strip length

� Improve quality during ramp-up / ramp-down

� Better disturbance rejection to allow higher speed

Roll position

Uncoiler torque

Coiler torque

Output thickness

Uncoiler tension

Coiler tension

Input thickness

Rolling stand

Roll position

Uncoiler torque

Coiler torque

Output thickness

Uncoiler tension

Coiler tension

Input thickness

Rolling stand

Example:Cold Rolling

©A

BB

Gro

up -

29-

26-A

ug-0

6New Advanced Thickness Control

Dynamically decoupled MIMO control with� Online estimation of time

varying process parameters

� Online MIMO controller parameter adaptation using an in-built process model

� Dynamic feed-forward strategies support disturbance rejection

� Supervision layer to monitor and track the control quality

Pass schedule, Set-up and adaptation

MIMOController

Controllerdesign/adaption

On-line PlantModel

On-lineEstimation

Controllerparameter

Control objectivesSet point

PIDs DynamicDecoupler

DynamicDisturbance FF

Parameters

Pass schedule, Set-up and adaptation

MIMOController

Controllerdesign/adaption

On-line PlantModel

On-lineEstimation

Controllerparameter

Control objectivesSet point

PIDs DynamicDecoupler

DynamicDisturbance FF

Parameters


©A

BB

Gro

up -

30-

26-A

ug-0

6

Conventional control: h0 = 575µm, h1 = 400µm, width = 1244mmAdvanced control: h0 = 575µm, h1 = 400µm, width = 1285mm

Conventional control• Standard deviation:

• Body 3σ = 1,18%• Total 3σ = 1,32%

Conventional control• Standard deviation:

• Body 3σ = 1,18%• Total 3σ = 1,32%

Comparison: Conventional vs Advanced

New Advanced Control• Standard deviation:

• Body 3σ = 0,49%• Total 3σ = 0,66%

New Advanced Control• Standard deviation:

• Body 3σ = 0,49%• Total 3σ = 0,66%


©A

BB

Gro

up -

31-

26-A

ug-0

6

Recent innovations:� Multi-move: robots that collaborate� Force control: robots that feel� Vision: robots that see� Easy programming

Outlook:� More sensors, more sensor fusion� Closer human-robot collaboration

Control: Also at the heart of RoboticsExample:Robotics

©A

BB

Gro

up -

32-

26-A

ug-0

6Economic Process Optimization

� Different control systems, office environment, and enterprise systems can all talk to each other in real time

� Live external data feeds also easily integrated, e.g., spot market prices for feedstock, energy, product, weather information …

� Can we use this information together with modeling and a feedback control approach to online optimize whole plants from an economic perspective?

©A

BB

Gro

up -

33-

26-A

ug-0

6Hybrid: Mixed Logic Dynamic Systems

� Framework for handling continuous and discrete dynamics

� Describes both plant constraints and objective function

� Collaboration with ETH Zürich

©A

BB

Gro

up -

35-

26-A

ug-0

6Electrical Energy Management: Benefits

� Power cost down by 2-3%� Automatic rescheduling in case of unexpected events� Strict contractual and equipment constraint satisfaction

Example:Minerals

©A

BB

Gro

up -

36-

26-A

ug-0

6Modeling and model re-use

Control, equipment design optimization

Operating procedure optimization

Many others

Soft sensing

Yield accounting

Decision support

Diagnosis and troubleshooting

Optimization (ss + dynamic)

Linear models

[A,B,C,D]

MPC

Linearized models

Linearization

Advanced MPC

EstimationInitial model

Fitted modelOffline

Plant data

Up-to-date model

Reconciled plant

information

OnlineRaw plant data

Plant

Data Reconciliation Parameter Estimation

Outlook

©A

BB

Gro

up -

37-

26-A

ug-0

6

Reference ActuationSensing & Estimation

Context, e.g. market price, weather forecast

EconomicOptimization

Economic life-cycle cost estimate

Life-cycle optimization: The Next Loop?

Control Life-CycleCost ModelProcess

Outlook

©A

BB

Gro

up -

38-

26-A

ug-0

6Automating automation: The engineering data chain

P&IDFUP...

Offer Process Design Control Design Installation

Outlook

Engineering� Ideally, feed in only process model and objective

� Plant topology from P&ID, economic context from ERP

� Hardware and software components from vendor

� Hardware realization/configuration automatic

� Still a long journey, but economic importance high

©A

BB

Gro

up -

39-

26-A

ug-0

6Other open frontiers

� Methods� Bring mathematical assumptions and physical reality closer together

� extend hybrid optimizers to larger problem sizes� guaranteed properties for real-world closed-loop systems?

� Robustness of solvers towards unsupervised closed-loop operation� Dealing with significantly different time horizons

(asset dynamics tend to be much slower than process dynamics, e.g., wear, corrosion, crack propagation)

� Problem solving� Deal with abnormal, unexpected, situations

(e.g., interruption of communication, or alarm bursts etc) � Deal more readily with changing constraints

and changing configurations� Exploit distributed problem-solving capabilities

� distributed problem solving, e.g., agent approaches?� gain flexibility, but loose predictability? system properties?

Outlook

©A

BB

Gro

up -

40-

26-A

ug-0

6Summary and Acknowledgments

Summary

� Control system capability ceases to be the active constraint regarding computing power, memory, flexibility, and ease of interfacing

� Is the theory-practice gap narrowing?

Acknowledgments

� To customers, colleagues and university partners for sharing their insights

� R. Franke, A. Vollmer, A. Isaksson, E. Gallestey for providing examples

Industrial Automation: Situation and Trends

Documents

Transcript of Industrial Automation: Situation and Trends