Meet the Experts, Lafayette Hill, PA, Sept.

18, 2012

APC and Advanced Services

© ABB Service SM2011-180

February 11, 2009 | Slide 1
 What’s APC?

 APC – Advanced Process Control

 PID – Proportional, Integral, Derivative
 SISO – Single Input, Single Output
 MPC – Model Predictive Control
 MIMO – Multiple Inputs, Multiple Outputs
 State Space – Linear Algebraic form
 Constraints – Process or Physical Limits
 Linear Programming – Cost Minimization
 IMP – Inferential Modeling Platform
 Scheduling and Batch optimization

© ABB Group
September 25, 2012 | Slide 2
 1 Control Performance Issues
Half life of Process Controllers
Given: a 100 PID loops all tuned at once. Simple PID Utilization
Then: it is estimated that with in 6 months,
50 of these loops will have a degradation in
performance.
Loop performance

100 15% 30%

degradation

Manual Operation
Output Out of range
50 Increasing Variability
35% Improving process
25%
0
6 Months

PID Controllers are designed to: However, ABB is finding that PID Automation:

• Regulate the process • PID loops are not being maintained

• Reduce product instability • PID loops have degraded

• Improve operations • PID loops are standing in the way of

production and performance.

© ABB Group
September 25, 2012 | Slide 3
 ABB’s APC Project Approach

Process
APC Process
Interaction
Stabilize Project Interaction
Matrix
Process - Delivery Matrix Monitor
Identification -
Loop Identification - Performance
APC
Performance APC SCAN LoopSCAN
Fingerprint
Fingerprint

Service Service
Installation and
Pre Study + Implementation commissioning Periodic Service

Quick Customer Value No results erosion

Project delivery improved Continuous Improvement

Proven Approach
© ABB Group
September 25, 2012 | Slide 4
Goal: Continuous Improvement Action Plan

1
2
3
4
Tuning
Process Loops
1
2
3
Actuator 4
Prioritize
and
Categorize Implement
1
2
Signal 3
4

Stand Alone Tools

 Signal Analyzer 1
 LoopTune Logic 2
3
 Loop Analyzer 4
Continuous Tools
• ServicePort
Signal Analysis
LoopTune

Supports
 Self Regulating
 Non-Self Regulating
 Auto Model Identification

 Results stored as a LoopTune channel in ServicePort

 Provides long term process model tracking.
LoopTune: Tuning/Simulation

Supports
 800xA Controllers
 Harmony/Infi90 Controllers
 Mod300 Controllers
 Generic industry standard controllers
 Plug in modules for custom controllers
LoopTune: Loop Tuning Report

Automatically generates
Loop Tuning Reports
Outline

 Assuring Optimal Control Performance

 1) Loop Performance – Dupont
 2) Batch Optimization – BASF
 3) Boiler Fuel Savings – Arkema
 4) QCS Success Story - Paper

 APC – Advanced Process Control

 1) Powerhouse Optimization

 Sustaining Results

© ABB Inc.
September 25, 2012 | Slide 18
 ABB Automation & Power World - May 18-20, 2010

CCH-101-1 11:00 Tuesday, room 351C

Doug Reeder, Ted Matsko
Loop Performance Fingerprint for a
DuPont Monomers Plant
© ABB Inc.
September 25, 2012 | Slide 19
LoopAnalyzer Tool: Control Loop Diagnoses

(stiction)
(flow loops)
(aggressive)
(sluggish)

FCE = Final Control Element

© ABB Inc.
September 25, 2012 | Slide 20
Loop Performance Fingerprint
LoopScan Analysis – Performance Calculations

Navigation Loops Ranked

options for selected
Diagnosis
Diagnosis
results
in tree
Navigate by
clicking on
bars

© ABB Inc.
September 25, 2012 | Slide 21
Diagnosis results in frame for selected loop
Process Fingerprint
LoopScan Analysis – More Plots
Power Spectrum

Plot types

Histogram
Quick comparison plots, any plot
© ABB Inc.type, one tag frozen on top half
September 25, 2012 | Slide 22 Group Trend
Process Fingerprint
Oscillating Loops
 When a loop oscillates in
automatic mode and there is
not evidence of an external
disturbance, over tuning is a
possible cause.
 Power spectrum shows two
frequencies of interest
 This loop is in the TFE
Synthesis area

© ABB Inc.
September 25, 2012 | Slide 23
Process Fingerprint
Final Control Element Problem
 This is a flow controller that is the inner loop of a cascade.
 Exhibits classic stiction
 Controller output ramps up and down in triangular pattern
 Process variable moves in square wave

© ABB Inc.
September 25, 2012 | Slide 25
Process Fingerprint
Report
 The report highlights some loops, as shown in the previous
slides and summarizes the results in tables. This table is
for the TFE Synthesis section of the plant.

© ABB Inc.
September 25, 2012 | Slide 26
Process Fingerprint
PCA Cluster Example
 Find signals with
similar patterns,
probably due to
disturbances
 Not looking for
oscillating signals
 Here all signals are
in two columns that
are adjacent

© ABB Inc.
September 25, 2012 | Slide 30
Process Fingerprint
Tuning Parameter Cluster Analysis

 Yellow box shows normal area

for good tuning for flow loops
 Overall, tuning is a little
conservative, but may be
satisfactory for process with
slower time constants
 Some “outliers” need to be
checked.

© ABB Inc.
September 25, 2012 | Slide 31
 2 Workbench: Implementation improvements

Visualization
Analysis
and Setup

Standard Reporting Tool Workbench

Identification

Collection
Tuning and Simulation

© ABB Group
September 25, 2012 | Slide 33
 ABB Automation & Power World: April 23-26, 2012

C549I
11:15 Wednesday, April 25, 2012 (room 371D)
Batch Process Optimization
BASF Polymerization Reactor
© ABB Inc.
September 25, 2012 | Slide 34
Introduction
BASF Greenville
 Site History

© ABB Inc.
September 25, 2012 | Slide 35
BASF Batch Reactor Optimization
Polyester – What is it?

 Polyester is produced in our Greenville Ohio site “I”

Reactor as an automotive component product.

 Polyester is made by a solvent process in a batch reactor

 Followed by a fractionating column to remove trace

amounts of water.

 The processing time can be unacceptably long.

© ABB Inc.
September 25, 2012 | Slide 36
BASF Batch Reactor Optimization
Polyester – Key Equipment in “I” Reactor System
 5,000 gallon reactor with Therminol-66 heating system
 Fractionator column
 Condenser and Decanter for water collection

© ABB Inc.
September 25, 2012 | Slide 39
BASF Batch Reactor Optimization
Polyester – ABB Study of Process Efficiency
 SymBatch / ProduceIT help ensure
consistency on operational steps.

 However, batch cycle time variability

came from several factors.

 A multifunctional team of Chemists,

Engineers, Operators, and Operations
Management continually made
improvements, but outside resources
were needed.

 ABB offered to help with their

Fingerprint service.

© ABB Inc.
September 25, 2012 | Slide 40
BASF Batch Reactor Optimization
Temperature Control Loop Performance

 There are six main control loops, only four are operative
1. Reactor hot oil temperature (inner loop)
2. Reactor temperature (outer loop)
3. Reactor-Coil T override (outer loop)
4. Condenser Reflux Rate (not in use)
5. Fractionator overhead temperature
6. Reactor Pressure (saturated during vacuum)

 Oscillations observed in reactor temperature and

fractionator temperature

 Hot Oil loop override active and saturated output

 Pressure loop saturates

© ABB Inc.
September 25, 2012 | Slide 41
BASF Batch Reactor Optimization
Temperature Control Loop Performance

 Looking at standard data, the oscillations are difficult to see

because of the wide range of operation typical in batch
processes and different starting times
 Define steps (events)
 View by event

© ABB Inc.
September 25, 2012 | Slide 42
BASF Batch Reactor Optimization
Temperature Control Loop Performance

 Looking at standard data, the oscillations are difficult to see

because of the wide range of operation typical in batch
processes and different starting times
 Define steps (events)
 View by event

© ABB Inc.
September 25, 2012 | Slide 43
BASF Batch Reactor Optimization
Temperature Control Loop Performance

 Looking at standard data, the oscillations are difficult to see

because of the wide range of operation typical in batch
processes and different starting times
 Define steps (events)
 View by event

© ABB Inc.
September 25, 2012 | Slide 44
BASF Batch Reactor Optimization
Temperature Control Loop Performance

 Looking at standard data, the oscillations are difficult to see

because of the wide range of operation typical in batch
processes and different starting times
 Define steps (events)
 View by event

© ABB Inc.
September 25, 2012 | Slide 45
BASF Batch Reactor Optimization
Temperature Control Loop Performance

 Looking at standard data, the oscillations are difficult to see

because of the wide range of operation typical in batch
processes and different starting times
 Define steps (events)
 View by event

© ABB Inc.
September 25, 2012 | Slide 46
BASF Batch Reactor Optimization
Temperature Control Loop Performance

 Use Loop Analyzer to look at reactor temperature cascade

 Amplitude plot shows oscillation frequency
 Implication is that outer loop is the problem

© ABB Inc.
September 25, 2012 | Slide 47
BASF Batch Reactor Optimization
Process Economics - Quality Control and Production Rate

Process Economics is tied to two things

 Quality Control
 Decrease cost, i.e. lower energy
 Increase yield at same quality
 Reduce offspec losses
 Value increases with quality

 Production Rate
 Hold fixed costs constant
 Increase production rate, revenue
 For a batch process, production rate means cycle time

 For this polyester product, Step 4 dominates

© ABB Inc.
September 25, 2012 | Slide 48
BASF Batch Reactor Optimization
Process Economics - Quality Control and Production Rate
 Quality variance is very low for this product
 Batch held until all specs met increase time
 Lab tests repeated, manual adjustments
 This plant is Production Rate
limited on this product
 Long cycle time
 5 day x 24 hr work week
 120 hrs working time
 60 hrs = 2; 40 hrs = 3 batches
 Opportunity
 Reduce variance
of step times

© ABB Inc.
September 25, 2012 | Slide 49
BASF Batch Reactor Optimization
Process Economics - Quality Control and Production Rate
 Quality variance is very low for this product
 Batch held until all specs met increase time
 Lab tests repeated, manual adjustments
 This plant is Production Rate
limited on this product
 Long cycle time
 5 day x 24 hr work week
 120 hrs working time
 60 hrs = 2; 40 hrs = 3 batches
 Opportunity
 Reduce variance
of step times

© ABB Inc.
September 25, 2012 | Slide 50
BASF Batch Reactor Optimization
Reducing Batch Cycle Times
 This plot confirms conclusion about vacuum and batch
cycle time
 Real data is not always pretty (scatter)
 Due to lab measurement, operator manual
operations, unknown contaminants

© ABB Inc.
Step4 Elapsed Time(Vacuum)
September 25, 2012 | Slide 52
BASF Batch Reactor Optimization
Reducing Batch Cycle Times
 This plot confirms conclusion about vacuum and batch
cycle time
 Real data is not always pretty (scatter)
 Due to lab measurement, operator manual
operations, unknown contaminants

Conclusion:
Investigate cost to
improve vacuum

© ABB Inc.
Step4 Elapsed Time(Vacuum)
September 25, 2012 | Slide 53
BASF Batch Reactor Optimization
Polyester Batch Reaction - RESULTS
 Difficult analysis: Improvements ongoing
 Reactor:
 Evaluate more driving force on oil loop.
 Batch Cascade loop oscillation (1oF) causing heating oil Loop
oscillation (10oF) needs tuned.
 Fractionator:
 Temperature Control could use a feed-forward
 Identified Fractionator temperature loop failure
 Identified saturated transmitter
 BASF Advance Process Control (APC) group also analyzed the
reactor from a theoretical perspective and arrived at much the
same conclusions.
 WE GOT THE CYCLE TIME IMPROVEMENTS !
 Related Harmony Fingerprint Analysis determined BRC100
© ABB Inc.
September 25, 2012 | Slide 57
memory issues needing resolved.
 Dwight Stoffel, Bob Horton
ABB Optimization Services

ABB Automation & Power World - April 18-20, 2009

CSE-102-1: Boiler Fingerprint

Success Story: How Arkema Saved
$300,000 per year on Energy

© ABB Inc.
September 25, 2012 | Slide 58
Boiler Fingerprint : Value
 Energy Savings
 More Responsive to Process Steam
Demands
 Extended Operating Range
 More Reliable
 Improved Safety
 Reduced Carbon Footprint

© ABB Inc.
September 25, 2012 | Slide 59
Boiler 2: Simplified Schematic

Steam

Gas

Feed Water

Air

© ABB Inc.
September 25, 2012 | Slide 60
FD Fan Control – Combustion Air
Raw Data
A-FIC-201(PV) A-FIC-201(SP) A-FIC-201(C)

33

32

31

30

52.5

50.0

47.5

45.0

2300 2400 2500 2600 2700 2800 2900 3000 3100 3200 3300 3400 3500 3600
Sample Number, Ts = 5 , Total Samples = 10801

© ABB Inc.
September 25, 2012 | Slide 61
FD Fan Control – Combustion Air
Raw Data
A-FIC-201(PV) A-FIC-201(SP) A-FIC-201(C)

33

32

31

30

52.5

50.0

47.5
Positioner and Dampers = Suspect
45.0

2300 2400 2500 2600 2700 2800 2900 3000 3100 3200 3300 3400 3500 3600
Sample Number, Ts = 5 , Total Samples = 10801

© ABB Inc.
September 25, 2012 | Slide 62
Boiler Hardware Issues

Forced Draft Control Drive and fan Induced Draft Control Drive and fan

• Positioner drives not operating smoothly.

• Cylinder/Piston assemblies should be rebuilt or replaced.
• Motors are oversized.

© ABB Inc.
September 25, 2012 | Slide 63
Hysteresis – Air and Fuel
H ys te re s is in F u e l a n d A ir C o n tro ls
14-M ay -08 P M D ata S et

A -FIC -33 0(P V ) A -FIC -33 0(S P ) A -FIC -33 0(C O ) A -FIC -33 0(C O ) A -FIC -30 0(P V )

A -FIC -30 0(S P ) A -FIC -30 0(C O ) A -FIC -30 0(C O ) A -A T-3 15 A -A T-3 15
100 0
Fuel

900
800
14
O u tp u t

12
10
35
30
A ir

25
20
25.0
O u tp u t

22.5
20.0
17.5
15.0
8
6
O 2

4
300 400 500 600 700 800 900 100 0 110 0 120 0
D ata P o ints

© ABB Inc.
September 25, 2012 | Slide 64
Load Test
Raw Data
A-FRD-200(PV) A-FRD-200(SP) A-FIC-230(PV) A-FIC-230(SP) A-FIC-201(PV)

A-FIC-201(SP) A-PT-263 A-AIC-215(PV) A-AIC-215(SP) A-AIC-215(C)

45
40
35 Steam
30
2000
1500 Gas
50
40 Air
30
170.0
Boiler Pressure
167.5

5
4 % O2
3
2.5
Trim Output
0.0

1250 1500 1750 2000 2250 2500 2750 3000 3250

Sample Number, Ts = 5 , Total Samples = 3601

© ABB Inc.
September 25, 2012 | Slide 65
Boiler Fingerprint Recommendations
 Hardware
 Repair FD and ID control drives
 Resolve O2 transmitter reading issues. Check calibration, find leak, change location.
 Adequately seal all doors
 Recalibrate steam flows
 Add blowdown flow monitoring
 Adjust, clean, or replace sight glass for drum level
 Control Logic
 Perform full combustion test to fine tune steam to air curves, esp. for oil
 Update control logic to current implementation standards
 Adjust logic to know when oil/gas is off
 Excess Air calculation needs to be updated
 Tuning
 Retune loops to be less aggressive.
 Reduce output surge and ringing tendencies.
 Add a small filter to level measurement. This will reduce feed water chatter.
 Reduce filter on old steam flow measurement.

© ABB Inc.
September 25, 2012 | Slide 67
 LoopScan: Cluster Analysis
Loop Parameter Evaluation
AI Setup Parallel FC 19 Standard
P I D Gain Integral time
Tag Description Zero Span Range 100/range S5 S6 S7 S8 S9 S10 Kc Ti (sec)
A-FRD-200 Boiler 2 Steam Flow control 0 50 50 2.00 2 0.75 0.5 0 50 10 0.75 90.0
A-FIC-201 Boiler 2 Air Flow Control 0 60 60 1.67 1.7685 0.5 4.25 0 100 25 0.5 7.1
A-FIC-230 Boiler 2 Natural 0 3000 3000 0.03 0.0333 0.35 2.5 0 50 0 0.35 8.4
A-FIC-260 Blr 2 Feedwater Flow Control 0 50 50 2.00 2 0.25 1.75 0 100 0 0.25 8.6
A-FIC-201 Boiler 2 Air Flow Control 0 60 60 1.67 1.7685 0.5 4.25 0 100 25 0.5 7.1
A-PIC-211 Boiler 2 Furnace Pressure Control -3 3 6 16.67 16.6 0.2 3.5 0 25 0.2 3.4
A-FIC-201 Boiler 2 air Flow Control 0 60 60 1.67 1.768 0.5 4.25 0 100 25 0.5 7.1
A-FIC-260 Blr 2 Feedwater Flow Control 0 50 50 2.00 2 0.25 1.75 0 100 0 0.25 8.6
A-AIC-268 Blr 2 Blowdown Conductivity 0 5000 5000 0.02 0.02 1 7 0 100 -5 1 8.6
A-RSP-262 Level 5 2.75 0.2 2.75 825.0
A-PIC-570 Header Pressure (PV) -25 25 50 2.00 0.5 10 0.9 0 100 0 10 666.7

Industrial Example: Cluster Analysis Before vs After implementation

Flow Control Tuning Level Control Tuning

120 400
350
100
300
80 Integral 250
Integral

60 Original 200 Original

New 150 New
40
100
20
50
0 0
0 1 2 3 4 0 5 10 15
Proportional Proportional

© ABB Inc.
September 25, 2012 | Slide 68
 Updated Air Curves
Gas Fuel for Air Curve Oil Fuel for Oil Curve

1.7 1.7

1.6 1.6

1.5 1.5
Demand for Air Flow

Demand for Air Flow

1.4 1.4
Air (Original) Air (Original)
Air (new) Air (new)
1.3 1.3

1.2 1.2

1.1 1.1

1 1
10 15 20 25 30 35 40 45 50 10 15 20 25 30 35 40 45 50
Steam Flow (KLB/Hr) Steam Flow

Steam to Trim O2 Curve

10

7
O2 Setpoint

O2 Trim (Original)
6
O2 Trim (New)
5

2
10 20 30 40 50
© ABB Inc.
September 25, 2012 | Slide 69
Steam Flow
Boiler #2: Customer results
Boiler 1 O2 vs Load
1 Hour Avgs.

6
% Oxygen

5 March 2008
February 2009
February All Auto

2
15 20 25 30 35 40
Steam Load

© ABB Inc.
September 25, 2012 | Slide 70
Boiler #2: Customer results
Boiler 1 O2 vs Load
1 Hour Avgs.

6
% Oxygen

5 March 2008
February 2009

Fuel Savings February All Auto

2
15 20 25 30 35 40
Steam Load

© ABB Inc.
September 25, 2012 | Slide 71
Financial Impact – Boiler #2, Summer 2008

Boiler #2 rated at 40 klb steam/hr

Savings range of 2% to 3% achieved
without major capital
Approximate value = $75K to $100K for
Boiler No. 2 alone
Savings for all four boilers = $300,000

© ABB Inc.
September 25, 2012 | Slide 74
 ABB Automation & Power World - May 18-20, 2010

Optimization Services
Paper machine Fingerprint

© ABB Inc.
September 25, 2012 | Slide 75
 Variability – Paper Process
Headbox

• Machine Direction (MD)

• Cross Direction (CD)

MD

CD

© ABB Inc.
September 25, 2012 | Slide 76
System Reel Report

2-Sigma’s

Basis Wt 1 LBS Quality Analysis

Target Actual Efficiency RES MDL CD TOT
38.28 38.28 100.00 1.54 0.34 0.81 1.77

Moisture 1 PCT
Target Actual Efficiency RES MDL CD TOT
3.70 3.70 100.00 1.08 0.30 0.59 1.26

© ABB Inc.
September 25, 2012 | Slide 77
Optimization Services - Service Areas
Process Optimization
Applied to: All Paper Making Systems
Maintenance

Control Application Utilization

Applied to: Controls and System usage

Steam
approach
Dilution
Stuff
Valve
Stock Box
approach
CT FT
Stock
Machine Thick Stock Valve
Chest Consistency

Support System Availability

Applied to: Hardware
© ABB Inc.
September 25, 2012 | Slide 78
 Fingerprint: Paper Machine
Provides information on:
Evaluates: Cyclic content of Weight and
• Controllable Energy
Moisture in: Cross Direction and Machine
• Mechanical Pulsations of vibrations
Direction from High Frequency up 500Hz
• Benchmark of machine stability
Product down to Low Frequency of 5 hours.

Evaluates: On control
Variability
Performance of: Total Head,
Fingerprint: Weight and Moisture
Provides information on:

COV as a percent of total

100

Thick Stock Flow, Thick Stock

80
60
40
weight
moisture
• Start up time
Consistency, Machine Chest Stock 20
0 • Grade Change recovery
Machine

CD

6
• Disturbance reduction

de

de

de

de

de

de
ca

ca

ca

ca

ca

ca
Level Approach

De

De

De

De

De

De
COV Weight = 5.17
COV Moisture = 36.45 Response • Sheet break recovery
Stability Cyclic response • Responsiveness
Poor regulation

History Slow response

Offset from setpoint
Provides information on:
(VPA)
• Controllable Energy Fast response
• Stock Approach performance Good regulation
Profiling Evaluates: Automatic and
• Tuning Quality Capability Manual mode operation of:
• Oscillation sources
Actual vs. Calc. Profiles for 1 inch Zones

3
Actual Calculated
Weight, Moisture, and Total
2

1 Head.
0

-1

-2
Profile

-3

-4
Sigma

-5

Evaluates: On control Provides information on:

-6

-7

• Will CD control improve the profile?

0 5 10 15 20 25 30 35 40 45 50 55 60 65

Performance and model for: Zones

Current Potential

Weight, Moisture, and Caliper • Is current CD control optimized?

© ABB Inc.
September 25, 2012 | Slide 79
Fingerprint: Benchmarking

How does this paper machine compare to others in its class?

Where should improvement efforts be focused?

Index Total Acceptable

Product Variability 24.8 < 15

Fiber Stability 0.63 < 0.3

Machine 7.3 < 3.0

Response
Profile Capability 133 to 162 < 30

What is the best control that can be achieved on this machine?

© ABB Inc.
September 25, 2012 | Slide 80
Return on Investment

Every dollar invested in reducing variability,

results in a return of 5 to 10 times!

Mill Investment Annual Savings Ratio

Medium 65k 325k 5 to 1

Fine Writing 95k 610k 6.4 to 1

Newsprint 20k 250k 12.5 to 1

Linerboard 70k 650k 9.3 to 1

Tissue 75k 560k 7.5 to 1

Pulp 120k 2,270k 18.9 to 1

Average 74k 778k 10.5 to 1

© ABB Inc.
September 25, 2012 | Slide 81
 AGP400: High Speed Data Collection

Portable
-82
OS
 AGP400: High Speed Data Collection
-83
OS
High Frequency Results
Single
Amplitude Spectrum Comparison Point Frame data
0.3
P oints = 131072
collected at 100 Hz and
1000 Hz for complete high
0.2
frequency picture.
Basis Weight

0.1

0.0
0.75 Provides insight into
0.50 pulsation and vibration in
Moisture 1

0.25
both weight and moisture
0.00
0.4
Moisture2

0.2

0.0
10-5 10-4 10-3 10-2 10-1 X1000 Hz
A B C D oint
Cycles/P E F G
Amplitude
Frequency Period BW MT1 MT2 diameter length Potential Source
A 0.061 16.38 0.27 0.8 0.44 265 832 Machine Chest Level?
B 0.36 2.78 0.048 0.33 0.358 45.1 141.7 Top Felt
C 0.71 1.39 0.046 0.29 0.25 22.6 70.8 Wire?
D 1.84 0.54 0 0.298 0 8.76 27.5 ?
E 3.15 0.317 0.07 0.38 0 5.13 16.1 After Section Dryer
F 6.3 0.159 0.09 0.55 0 2.56 8.06 After Section Dryer
G 10.3 0.1 0.07 0.26 0.12 4.9 1.56 fan pump?
(Yankee speed = 3049fpm)

© ABB Inc.
September 25, 2012 | Slide 84
PM Fingerprint: Product Variability
Provides a summary of
the cyclic energy in the
Fingerprint Comparison - % Total COV sheet over a
corresponding frequency
Weight Moisture band.
80 QCS DCS Mechanical Includes Cross direction,
70 Weight and moisture up
60 to 100 hertz.
Comparison

50

40

30

20

10

0
CD Decade 1 Decade 2 Decade 3 Decade 4 Decade 5 Decade 6
Weight 25.3743210 50.5483932 6.1163764 1.8690249 6.7068100 3.6419485 5.7431183
Moisture 2.2876263 84.7719421 3.6153731 0.5106572 0.8314567 2.8962073 5.0867424

© ABB Inc.
September 25, 2012 | Slide 85
Data Logger 300: OPC Data Collection

 Conforms to OPC open standards

 Supports OPC DA V1 and later
 Log all OPC supported data types
 Access data from different OPC
vendors
 Log data from OPC servers on the
network
 Log data from multiple servers at the
same time
 Log scalar and array data
 Run multiple logging sessions with
different tags in each session

OS -86
Data Logger 300: OPC Data Collection

 Supports multiple logging start-stop

conditions
 Manual
 Time
 Number of samples
 Multiple OPC tag value condition
statements
 Provide data buffering before the logging
start event, which is beneficial for root-
cause analysis
 Sheet break
 Grade change
 Process value exceeding a limit
 High variability in a process value

OS -87
  Loop Analysis

Statistical

Analysis

Spectral Modeling

Analysis Analysis

OS -88
  CD Profile Analysis

Statistical

Analysis &
Model

Analysis

Spectral Analysis
3D

Analysis

OS -89
 Transition, AGC, SB Recovery
-90
OS
 PM1 Grade Change Time
ROI $445K - $526K
(2260 – 2660 Tons) AGC Coordination
Disabled
Grade Change Time (avg 10)
Optimization Period

50
45
40
35
30
Time (avg 10)
min

25
Moving Avg
20
15
10
5
0 Average grade change time
before optimization was
35.9 min and is currently
down to 21.6 to 19.3 min

Average Grade Change

Each data point is average of 10
Time reduced by
grade changes
14.3 to 16.6 min
(194 grade changes analyzed)
-91

 PM3 - Sheet Break Recovery Time

ROI $65K - $86K

Average Sheet Break Recovery Time (1300 – 1725 Tons)

50

45

40

35
Time (min)

30

25

20

15

10
April 2009 Sept-Oct 2009 Feb-May 2010 June-Sept 2010 Oct-Nov 2010

Oct-Nov 2010 sheet break recovery data showed a 20.9 min reduction in
sheet break recovery time (38.5 down to 17.6 min)
-92

 Optimization Summary – Achieved ROI
 Total Optimization Savings 2009 : $1.2M – $1.4M

 PM1
 Grade change time improvement:
 14.3 to 16.6 minutes: ROI $445K - $526K (2260 – 2660 Tons)
 PM2
 Grade change time improvement:
 9.5 to 10.0 minutes: $365K - $387K (2080 – 2200 Tons)
 MD weight and moisture variability reduction improvement
 Weight target shift: $232K - $287K, (488 – 604 Tons)
 Moisture target shift: $74K - $147K, (155 – 309 Tons)
 PM3
 Recovery Time: ROI $65K - $86K (1300 – 1725 Tons)
-93
 ABB’s APC Project Approach

Process
APC Process
Interaction
Stabilize Project Interaction
Matrix
Process - Delivery Matrix Monitor
Identification -
Loop Identification - Performance
APC
Performance APC SCAN LoopSCAN
Fingerprint
Fingerprint

Service Service
Installation and
Pre Study + Implementation commissioning Periodic Service

Quick Customer Value No results erosion

Project delivery improved Continuous Improvement

Proven Approach
© ABB Group
September 25, 2012 | Slide 94
APC Example: Zellstoff Celgar
Supervisory Control Architecture
 Multi-Layered Solution

LOAD FORECASTING AND PRODUCTION PLANNING ENERGY

MONITORING
OPTIMIZATION
ENERGY LAYER
SHORT-TERM OPTIMIZATION AND INVENTORY CONTROL REPORTING

Advanced Process Control (APC)

ADVANCED
POWER PLANT
COORDINATED HEADER CONTROL PERFORMANCE CONTROL AND
ASSESSMENT MONITORING
LAYER

APC AND OPTIMIZATION INTERFACE POWER BOILER COS

AUTORECOVERY BASE
WOODWARD
EVAPORATOR LIME KILN SLAKER
BALANCE OF PLANT
GOVERNOR CONTROL
CONTROL CONTROL CONTROL
LAYER
ABB INFI-90 DCS TURBINE #3 CONTROL TURBINE #2 CONTROL

SENSORS AND ACTUATORS

WBL HBL

RB
G G
PROCESS
WL GL

© ABB Inc.
September 25, 2012 | Slide 95
ABB’s APC Solution: Predict and Control

Features
 True Multivariable Control
 State Space, Model Predictive Control (MPC) Structure
 Allows Constraints
 Inferential Model Prediction (IMP) of hard to measure
variables

© ABB Group
September 25, 2012 | Slide 96
APC - Predict and Control

 Allows safe operation closer to constraints

 Smoothes operation by predicting effects of control moves
and compensating
 Works best when base level controls are optimized
© ABB Group
September 25, 2012 | Slide 97
Predict and Control: Engineering Tool

 Allows process model development

 Shows interaction effects among variables
 Allows simulation of benefits

© ABB Group
September 25, 2012 | Slide 98
APC Fingerprint

Features
 Uses Predict and Control Engineering
Tools
 Quantifies critical process interactions
 Determines $ benefits from APC
 Suggests best structure
 Suggests required MV’s, CV’s, DV’s
 Determines additional measurement
requirements and value
 Allows customer to get the best APC
product for their plant

© ABB Group
September 25, 2012 | Slide 99
MPC Simulation Results
 Shell Heavy Oil Fractionator
(HOF) Problem
 3 MVs, product flows
 2 DVs (used as unmeasured
disturbances)
 2 CVs w/SP, product quality
 1 CV constraint, temperature
 4 PVs available
 Soft constraint on an MV for
process optimization
Optimize IT Predict & Control Performance
 3 simulation runs of the HOF problem

Comparing
typical DMC
treatment of
disturbances to
state feedback,
using the Shell
Heavy Oil
Fractionator
problem with
unmeasured
disturbances and
modeling error
Optimize IT Predict & Control Performance
 3 simulation runs of the HOF problem

Comparing
1st example
typical
DMC likeDMC
tuning,
treatment
compositions of
disturbances
(blue and red) to
state feedback,
have large
using the Shell
deviations
Heavy
causedOil by
Fractionator
unmeasured
problem with
cooling duty
unmeasured
disturbances,
disturbances
(magenta line, and
modeling error
lower grid)
Optimize IT Predict & Control Performance
 3 simulation runs of the HOF problem

Comparing
1st example
2nd example,
DMCtypical
likeDMC
P&C tuning,
with
treatment
compositions
Kalman Filterof
disturbances
(blue forto
and red)
estimate
state feedback,
have large
disturbances.
using the Shell
deviations
Large
Heavy reduction
Oil
caused by
inFractionator
PV deviation.
unmeasured
problem with
cooling duty
unmeasured
disturbances,
disturbances
(magenta line, and
modeling error
lowergrid)
Optimize IT Predict & Control Performance
 3 simulation runs of the HOF problem

Comparing
3rdexample
1st example,
2nd example,
P&C
DMCtypical
P&ClikeDMC
with Kalman
tuning,
with
treatment
Filter
Kalmanestimate
compositions of for
Filter
disturbances
disturbances.
(blue forto
and red)
estimate
statehavefeedback,
large
disturbances.
Using additional
using the Shell
deviations
Large
PVs
reduction
Heavy
caused Oil
by
inFractionator
PV deviation.
Further
unmeasured
problem
reduction with
cooling duty in PV
unmeasured
deviation.
disturbances,
disturbances
(magenta line, and
modeling
lowergrid)error

 3 Loop Optimization helps sustain customer results
Goal: Maintain improved performance level
 Adjust maintenance operating procedures Proactive and collaborative
 Adjust standard operating procedures
 Remote process monitoring
3
Service Sustain
(Scan/Track)
 Specifics are a function of the
Implement phase 2 Implement
(HandsOn)
 Periodic monitoring of key
process indices utilizing local 1
or remote expertise Diagnose
Increased
(Fingerprints) performance

© ABB Group
September 25, 2012 | Slide 105
 Remote Optimization Services
ABB’s Proactive Architecture
Customer Request
On Site On Site

Reactive Scheduling Collection Analysis Resolution

On Site Remote On Site

Proactive Audit
Collection Analysis Resolution Long lead time

On Site
Remote On Site
Data Pool Periodic Fingerprint
Collection Analysis Resolution Modules
ServicePort

Short Lead time

On Site KPI
Data Pool Trending On Site

Collection Analysis Resolution Scan

On Site Condition
Data Pool Monitoring Remote

Collection Analysis Resolution

Track No Lead time

OS -106
Event Trigger
Service Infrastructure
Remote-enabled services improve access, cut costs
Service Capabilities

Event Optimization Support Software System Remote

Notification Control Tuning Services Services Support Health Check Troubleshooting

Remote
Local
Secure Access
Customer-Defined Access

HoistScan/Track
ServicePro LoopScan/Track DriveScan/Track

System Scan/Track Troubleshooting APC Scan/Track

Services Services

Customer-Defined Access

Secure Tunnel

Instruments/
Engineering Stations Operator Stations Drives OCS Historian
Actuators
ABB ServicePortTM
Secure, remote delivery for Scan and Track

 Secure portal residing at customer site

through which plant personnel and
ABB experts can access:
 Configuration tools
 Diagnostic applications
 Improvement activities
 Performance-sustaining
troubleshooting
 Scanning software that deploys
agreed actions.
 ABB can connect to any system
through ServicePort and implement
fixes to diagnosed problems.
Full Size

© ABB Group
September 25, 2012 | Slide 108
Desk top
 ABB ServicePort Channels
Main Menu Value Map Maintenance Tracking

Loop Performance Disturbance Analysis

Platform Performance Alarm/Event Traffic

Application channels
• Platform
• Process
• Control
• Maintenance
© ABB Group
September 25, 2012 | Slide 109
 Sustain: track KPIs to ensure improvement
 Q1  Q2  Q3  Q4

Production
increase!

Continuous
KPI Tracking

Improvement
Variability
decrease!
 Total

 RES
1

 CD

 MDL

© ABB Group
September 25, 2012 | Slide 110
© ABB Group

 Delivery Schedule
September 25, 2012 | Slide 110
© ABB Group
September 25, 2012 | Slide 111