Exceptional Process Control

Opportunities - An Interactive
Exploration of Process Control
Improvements - Day 3

ISA Saint Louis Short Course Dec 6-8, 2010

Standards
Certification
Education & Training
Publishing
Conferences & Exhibits
Welcome
• Gregory K. McMillan
– Greg is a retired Senior Fellow from Solutia/Monsanto and an ISA Fellow.
Presently, Greg contracts as a consultant in DeltaV R&D via CDI Process &
Industrial. Greg received the ISA “Kermit Fischer Environmental” Award for pH
control in 1991, the Control Magazine “Engineer of the Year” Award for the
Process Industry in 1994, was inducted into the Control “Process Automation
Hall of Fame” in 2001, was honored by InTech Magazine in 2003 as one of the
most influential innovators in automation, and received the ISA Life
Achievement Award in 2010. Greg is the author of numerous books on process
control, his most recent being Essentials of Modern Measurements and Final
Elements for the Process Industry. Greg has been the monthly “Control Talk”
columnist for Control magazine since 2002. Greg’s expertise is available on the
web site: http://www.modelingandcontrol.com/
Improving Loops - Part 2

Top Ten Reasons I use a Virtual Plant

• (10) You can’t freeze, restore, and replay an actual plant batch
• (9) No separate programs to learn, install, interface, and support
• (8) No waiting on lab analysis
• (7) No raw materials
• (6) No environmental waste
• (5) Virtual instead of actual problems
• (4) Bioreactor batches are done in 14 minutes instead of 14 days
• (3) Plant can be operated on a tropical beach
• (2) Last time I checked my wallet I didn’t have $100,000K
• (1) Actual plant doesn’t fit in my suitcase
Improving Loops - Part 2

Fed-Batch and Startup Time Reduction - 1

• PID on Error Structure

– Maximizes the kick and bump of the controller output for a setpoint change.
– Overdrive (driving of output past resting point) is essential for getting slow loops,
such as vessel temperature and pH, to the optimum setpoint as fast as possible.
– The setpoint change must be made with the PID in Auto mode.
– “SP track PV” will generally maximize the setpoint change and hence the kick and
bump (retaining SP from last batch or startup minimizes kick and bump)
• SP Feedforward
– For low controller gains (controller gain less than inverse of process gain), a
setpoint feedforward is particularly useful. For this case, the setpoint feedforward
gain is the inverse of the dimensionless process gain minus the controller gain.
– For slow self-regulating (e.g. continuous) processes and slow integrating (e.g.
batch) processes, even if the controller gain is high, the additional overdrive can
be beneficial for small setpoint changes that normally would not cause the PID
output to hit a limit.
– If the setpoint and controller output are in engineering units the feedforward gain
must be adjusted accordingly.
– The feedforward action is the process action, which is the opposite of the control
action, taking into account valve action. In other words for a reverse control
action, the feedforward action is direct provided the valve action is inc-open or the
analog output block, I/P, or positioner reverses the signal for a inc-close.
Improving Loops - Part 2

Fed-Batch and Startup Time Reduction - 2

• Full Throttle (Bang-Bang Control) - The controller output is stepped to it output
limit to maximize the rate of approach to setpoint and when the projected PV
equals the setpoint less a bias, the controller output is repositioned to the final
resting value. The output is held at the resting value for one deadtime. For more
details, check out the Control magazine article “Full Throttle Batch and Startup
Response.” http://www.controlglobal.com/articles/2006/096.html
– A deadtime (DT) block must be used to compute the rate of change so that new values of
the PV are seen immediately as a change in the rate of approach.
– If the total loop deadtime (θo) is used in the DT block, the projected PV is simply the current
PV minus the output of the DT block (∆PV) plus the current PV.
– If the PV rate of change (∆PV/∆t) is useful for other reasons (e.g. near integrator or true integrating
process tuning), then ∆PV/∆t = ∆PV/θo can be computed.
– If the process changes during the setpoint response (e.g. reaction or evaporation), the
resting value can be captured from the last batch or startup
– If the process changes are negligible during the setpoint response, the resting value can be
estimated as:
– the PID output just before the setpoint change for an integrating (e.g. batch) process
– the PID output just before the setpoint change plus the setpoint change divided by the process gain
for a self-regulating (e.g. continuous) process
– For self-regulating processes such as flow with the loop deadtime (θo)
approaching or less than the largest process time constant (τp ), the logic is
revised to step the PID output immediately to the resting value. The PID output is
held at the resting value for the T98 process response time (T98 = θo + 4∗ τp ).
Improving Loops - Part 2

Structure, SP Feedforward, & Bang-Bang Tests

Structure 3 Structure 1 Structure 1 + SP FF Structure 1 + Bang-Bang

Rise Time = 8.5 min Rise Time = 1.6 min Rise Time = 1.2 min Rise Time = 0.5 min
Settling Time = 8.5 min Settling Time = 7.5 min Settling Time = 6.5 min Settling Time = 0.5 min
Overshoot = 0% Overshoot = 1.7% Overshoot = 1.3% Overshoot = 0.2%
Improving Loops - Part 2

Fed-Batch and Startup Time Reduction - 3

• Output Lead-Lag
– A lead-lag on the controller output or in the digital positioner can kick the signal
though the valve deadband and sticktion, get past split range points, and make
faster transitions from heating to cooling and vice versa.
– A lead-lag can potentially provide a faster setpoint response with less overshoot
when analyzers are used for closed loop control of integrating processes When
combined with the enhanced PID algorithm (PIDPlus) described in:
– Deminar #1
http://www.screencast.com/users/JimCahill/folders/Public/media/5acf2135-38c9-422e-9eb9-3

– White paper
http://www.modelingandcontrol.com/DeltaV-v11-PID-Enhancements-for-Wireless.pdf
• Deadtime Compensation
– The simple addition of a delay block with the deadtime set equal to the total loop
deadtime to the external reset signal for the positive feedback implementation of
integral action described in Deminar #3 for the dynamic reset limit option
http://www.screencast.com/users/JimCahill/folders/Public/media/f093eca1-958f-4d9c-9
.
– The controller reset time can be significantly reduced and the controller gain
increased if the delay block deadtime is equal or slightly less than the process
deadtime as studied in Advanced Application Note 3
http://www.modelingandcontrol.com/repository/AdvancedApplicationNote003.pdf
Improving Loops - Part 2

Deadtime Compensator Configuration

Must enable dynamic reset limit !

Insert

deadtime

block
Improving Loops - Part 2

Deadtime Myths Busted in Deminar 10

• Deadtime is eliminated from the loop. The smith predictor, which created a PV
without deadtime, fools the controller into thinking there is no deadtime. However, for
an unmeasured disturbance, the loop deadtime still causes a delay in terms of when
the loop can see the disturbance and when the loop can enact a correction that
arrives in the process at the same point as the disturbance. The ultimate limit to the
peak error and integrated error for an unmeasured disturbance are still proportional to
the deadtime, and deadtime squared, respectively.
• Control is faster for existing tuning settings. The addition of deadtime
compensation actually slows down the response for the existing tuning settings.
Setpoint metrics, such as rise time, and load response metrics, such as peak error,
will be adversely affected. Assuming the PID was tuned for a smooth stable response,
the controller must be retuned for a faster response (see slide 11). For a PID already
tuned for maximum disturbance rejection, the gain can be increased by 250%. For
deadtime dominant systems where the total loop deadtime is much greater than the
largest loop time constant (hopefully the process time constant), the reset time must
also be decreased or there will be severe undershoot. If you decrease the reset time
to its optimum, undershoot and overshoot are about equal. For the test case where
the total loop deadtime to primary process time constant ratio was 10:1, you could
decrease the reset time by a factor of 10, smaller than what was noted on slide 11.
Further study is needed as to whether the ratio of the old to new reset time is
comparable to the ratio of deadtime to time constant and whether the PID module
execution time (0.5 sec) is the low limit to the reset time for an accurate deadtime.

For access to Deminar 10 ScreenCast Recording or SlideShare Presentation go to

http://www.modelingandcontrol.com/2010/10/review_of_deminar_10_-_deadtim.html
Improving Loops - Part 2

Deadtime Myths Busted in Deminar 10

• Compensator works better for loops dominated by a large deadtime. The
reduction in rise time is greatest and the sensitivity to per cent deadtime modeling
error particularly for an overestimate of deadtime is least for the loop that was
dominated by the process time constant. You could have a deadtime estimate that
was 100% high before you would see a significant jagged response when the process
time constant was much larger than the process deadtime. For a deadtime estimate
that was 50% too low, some rounded oscillations developed for this loop. The loop
simply degrades to the response that would occur from the high PID gain as the
compensator deadtime is decreased to zero. While the magnitude of the error in
deadtime seems small, you have to remember that for an industrial temperature
control application, the loop deadtime and process time constant would be often at
least 100 times larger. For a 400 second deadtime and 10,000 second process time
constant, a compensator deadtime 200 seconds smaller or 400 seconds larger than
actual would start to cause a problem. In contrast, the deadtime dominant loop
developed a jagged response for a deadtime that was high or low by just 10%. I think
this requirement is unreasonable in industrial processes. A small filter of 1 second on
the input to the deadtime block in the BKCAL path may have helped.
• An underestimate of the deadtime leads to instability. In tuning calculations for a
conventional PID, a smaller than actual deadtime can cause an excessively oscillatory
response. Contrary to the effect of deadtime on tuning calculations, a compensator
deadtime smaller than actual deadtime will only cause instability if the controller is
tuned aggressively after the deadtime compensator is added.
• An overestimate of the deadtime leads to sluggish response and greater
stability. In tuning calculations for a conventional PID, a larger than actual deadtime
can cause an excessively slow response. Contrary to the effect of deadtime on tuning
calculations, a compensator deadtime greater than actual deadtime will cause jagged
irregular oscillations.
Improving Loops - Part 2

Fed-Batch and Startup Time Reduction - 4

• Feed Maximization
– Model Predictive Control described in Application Note 1
http://www.modelingandcontrol.com/repository/AdvancedApplicationNote001.pdf
– Override control (next slide) is used to maximize feeds to limits of operating constraints via
valve position control (e.g. maximum vent, overhead condenser, or jacket valve position with
sufficient sensitivity per installed characteristic).
– Alternatively, the limiting valve can be set wide open and the feeds throttled for temperature
or pressure control. For pressure control of gaseous reactants, this strategy can be quite
effective.
– For temperature control of liquid reactants, the user needs to confirm that inverse response
from the addition of cold reactants to an exothermic reactor and the lag from the
concentration response does not cause temperature control problems.
– All of these methods require tuning and may not be particularly adept at dealing with fast
disturbances unless some feedforward is added. Fortunately the prevalent disturbance that
is a feed concentration change is often slow enough due to raw material storage volume to
be corrected by temperature feedback.
• Profile Control
– If you have a have batch measurement that should increase to a maximum at the batch end
point (e.g. maximum reaction temperature or product concentration), the slope of the batch
profile of this measurement can be maximized to reduce batch cycle time. For application
examples checkout “Direct Temperature Rate of Change Control Improves Reactor
Yield” in a Funny Thing Happened on the Way to the Control Room
http://www.modelingandcontrol.com/FunnyThing/ and the Control magazine article
“Unlocking the Secret Profiles of Batch Reactors”
http://www.controlglobal.com/articles/2008/230.html .
Improving Loops - Part 2

Identified Responses for Fed-Batch Profile

Model Predictive Control (MPC)
Improving Loops - Part 2
Model Predictive Control (MPC) of
Growth Rate and Product Formation Rate

Dissolved Oxygen

Biomass Growth rate

Product Formation Rate

Substrate
Improving Loops - Part 2
Model Predictive Control (MPC)
Reduces Fed-Batch Cycle Time

MPC in Auto
Estimated Product
Concentration (AT6-6)
pH (AT6-1)

Estimated Biomass
Concentration (AT6-5)

Dissolved
Oxygen (AT6-2) Estimated Substrate
Concentration (AT6-4)

Estimated Biomass Growth

Rate (AY6-11)

Estimated Net Production

Rate (AY6-12)

Batch Basic Fed-Batch Batch APC Fed-Batch

Inoculation Inoculation
Improving Loops - Part 2
Model Predictive Control (MPC)
Improves Batch Predictions

Predicted Final
Product Yield (AY6-10E)
MPC in Auto
Predicted Batch
Cycle Time (AY6-10B)

Current Product
Yield (AY6-10D)

Predicted Batch
Cycle Time (AY6-10B)

Current Batch
Time (AY6-10A) Predicted Final
Product Yield (AY6-10E)

Predicted Cycle Time

Improvement (AY6-10C)

Predicted Yield
Improvement (AY6-10F)

Inoculation Batch Basic Fed-Batch Batch APC Fed-Batch

Inoculation
Improving Loops - Part 2

Batch Sequence Time Reduction

• Reduce wait times, operator attention requests, and manual actions by automation.
• Reduce excess hold times (e.g. heat release can confirm reaction start/end).
• Improve charge times and accuracy by better sensor design (e.g. mass flow meters
and valve location (e.g. minimize dribble time and holdup).
• Minimize acquire time by improved prioritization of users (e.g. unit operation with
biggest effect on production rate gets access to feeds and utilities).
• Reduce failure expression activation by better instruments, redundancy and signal
selection, and more realistic expectations of instrument performance.
• Improve failure expression recovery by configuration and displays.
• Eliminate steps by simultaneous actions (e.g. heat-up and pressurization).
• Increase feed and heat transfer rate by an increase in pump impeller size.
• Minimize non constrained processing time by all out run, cutoff, and coast.
• Minimize processing time by better end point detection (inferential measurements
by neural networks and online or at-line analyzers).
• Mid batch correction based on adapted online virtual plant model or batch analytics
projection to latent structures (PLS) and first principle relationships.
Improving Loops - Part 2

Open Loop Backup Configuration

SP_Rate_DN and SP_RATE_UP used to insure fast getaway and slow approach

Open Loop Backup Configuration

Open loop backup used for prevention of

compressor surge and RCRA pH violation
Improving Loops - Part 2

PID Controller Disturbance Response

Improving Loops - Part 2

Open Loop Backup Disturbance Response

Open Loop Backup

Improving Loops - Part 2

Conductivity Kicker for Evaporator

Improving Loops - Part 2

pH Kicker for Waste Treatment (Pensacola Plant)

MPC-1
MPC-2

Waste
RCAS RCAS
middle selector
ROUT
AC-1 AC-2 AY kicker AY

splitter splitter AT AT AT
AY AY AY
middle selector middle selector filter
FT FT
AY AY Attenuation
Tank
Stage 1 AT AT AT Stage 2 AT AT AT
Mixer Mixer FT
Improving Loops - Part 2

Virtual Plant Opportunities Beyond

Operator Training Systems (OTS)
• Dynamic simulations offer the opportunity to explore, quantify, demonstrate,
detail, and prototype process control improvements (PCI).
• However
– The investment in software and time to learn and develop simulations typically limits the
creation of models to specialists who have significant simulation and DCS expertise.
– Process deadtime, measurement dynamics, and valve response is often not modeled (not
understood by traditional process simulation software suppliers)
– The emulation of the basic and advanced control in a DCS by process simulators is unrealistic
• What is needed is a virtual plant that uses the actual DCS with all of its
capability and uses dynamics of all parts of the process and automation
systems in a friendly control room environment by the use of the DCS
operator interface
• The virtual plant should be useable by any one who wants to learn the best of
the practical control technologies for the process industry and to find,
demonstrate, estimate, and convince people of the benefits of PCI
– Automation Engineers
– Local Business Partners
– Process Engineers
– Students
– System Integrators
– Suppliers
• The virtual plant offers the ability to develop, prototype, and demo the
dynamic advantages of solutions, products, and services
Improving Loops - Part 2

Virtual Plant Synergy

DCS batch and loop

configuration, displays,
and historian
Embedded
Embedded
Advanced Control Tools
PAT Tools

Dynamic Loop Monitoring

Process Model Virtual Plant And Tuning
Laptop or Desktop
Personal Computer
Or
DCS Application
Station or Controller
Online Model Predictive
Data Analytics Control

Process Knowledge
Improving Loops - Part 2

PCI and OTS Virtual Plants

Virtual
DeltaV
VIM
ProPlus Sensors
Actual Virtual MiMiC Virtual
I/O Module
DCS OTS Process
Virtual
Configuration
Valves Dynamic
Graphics
Process
Trends
Simulators
Virtual
DeltaV

SimulatePro Sensors
Virtual Virtual MiMiC Virtual

DCS I/O PCI Process

Virtual

Valves
Improving Loops - Part 2

Virtual Plant Essentials

DeltaV Simulate Product Family

A_Vlv1
Condenser

CW Out Cooling water Fcw

Vnt
V_D1

A_VD1 Lc, Vc_out

V_D

Reflux Drum

Reflux L_R

Heavy liquid L_HvLiq

Feed 2
Distillate product L_D
MiMiC Simulation Software
Side withdraw 2

Feed 1

Side withdraw 1

V_B
Reboiler
A_v Heating steam

HE condensate

L_B + V_B Buttom product L _B

Improving Loops - Part 2

Smart Bang-Bang Lab

• Objective – Show how to reduce batch and startup time by a full
throttle setpoint response (bang-bang control)
• Activities:
– Go to Main Display and select Single Loop Lab01
– Click on PID faceplate and click on magnifying glass icon to get Detail display
– Enter tuning settings: Gain = 1.7, Reset = 210 sec, Rate = 2 sec
– Click on any block in block diagram and then on Process tab detail
– Set primary process Delay = 9 sec, Lag 2 Inc & Lag 2 Dec = 100 sec
– Set primary process Type = Integrating
– Enable setpoint metrics
– Make PID setpoint change from 50% to 60%
– Wait for setpoint response to complete and note metrics
– In PID detail, set Bang-Bang Bias = 4%
– Make PID setpoint change from 60% to 50%
– Wait for setpoint response to complete and note metrics
Improving Neutralizer pH Control

Nonlinearity - Graphical Deception

Reagent Titration Curves

14.00000000
14

12
12.00000000

10
10.00000000 For a strong acid and base the
pKa are off-scale and the slope
8
8.00000000 continually changes by a factor
pH of ten for each pH unit deviation
pH

Calculated pH

6
6.00000000 from neutrality (7 pH at 25 oC)

4
4.00000000

Reagent Titration Curves

2
2.00000000
11
11.00000000

0
0.00000000
10
10.00000000
0.00000000 0.00050000 0.00100000 0.00150000 0.00200000
/ Influent
ReagentWt Frac Base Ratio
9
9.00000000

8
8.00000000

Despite appearances there are no straight lines in a pH 7

pH
7.00000000

titration curve (zoom in reveals another curve if there

are enough data points - a big “IF” in neutral region) 6
6.00000000

5
5.00000000

Yet titration curves are essential for every aspect of pH system 4.00000000
4
design but you must get numerical values and avoid mistakes
such as insufficient data points in the area around the set point3.00000000
3 0.00099995 0.00099996 0.00099997 0.00099998 0.00099999 0.00100000 0.00100001 0.00100002 0.00100003 0.00100004 0.00100005

Reagent / Influent
Wt Frac Base Ratio
Improving Neutralizer pH Control

Effect of Acid and Base Type

Figure 3-1c: Strong Acid Titrated with a Weak Base
Strong Acid and Weak Base
Figure 3-1b: Weak Acid Titrated with a Strong Base
Weak Acid and Strong Base
14.00 14.00

12.00 12.00

pka = 10
10.00 10.00

8.00 8.00

pH
Calculated pH
pH

Calculated pH

6.00 6.00

pka = 4
4.00 4.00

2.00 2.00

0.00 0.00
0.000 0.500 1.000 1.500 2.000 0.000 0.500 1.000 1.500 2.000

Reagent / Influent
Slope moderated Reagent / Influent

Figure 3-1d: Weak Acid Titrated with a Weak Base near each pKa ! Figure 3-1e: Weak 2-Ion Acid Titrated with a Weak 2-Ion Base
Weak Acid and Weak Base Multiple Weak Acids and Weak Bases
14.00 14.00

12.00 12.00

10.00 10.00
pka = 9
8.00 8.00
pH

Calculated pH Calculated pH
pH

6.00 6.00
pka = 5
pka = 4
4.00 4.00

pka = 3
2.00 2.00

0.00 0.00
0.000 0.500 1.000 1.500 2.000 0.000 0.500 1.000 1.500 2.000
Reagent / Influent Reagent / Influent
Improving Neutralizer pH Control

Effect of Mixing Uniformity and Valve Resolution

10
pH

pH Set Point

4
Reagent to Feed
Flow Ratio

Control valve resolution (stick-slip) and

mixing uniformity requirements are
extraordinary on the steepest slope

Fluctuations or Oscillations
In Flows or Concentrations
Improving Neutralizer pH Control

Control Valve Size and Resolution

pH

8
Set point Control Band
6

B
Er = 100% ∗ Fimax ∗ −−−−
Frmax Influent pH
B Reagent Flow
Influent Flow
Frmax = A ∗ Fimax A

B Most reagent control valves are oversized,

Er = 100% ∗ −−−− which increases the limit cycle amplitude
A from stick-slip (resolution) and deadband
(integrating processes and cascade loops)
Ss = 0.5 ∗ Er

A = distance of center of reagent error band on abscissa from origin

B = width of allowable reagent error band on abscissa for control band
Er = allowable reagent error (%)
Frmax = maximum reagent valve capacity (kg per minute)
Fimax = maximum influent flow (kg per minute)
Ss = allowable stick-slip (resolution limit) (%)
Improving Neutralizer pH Control

Traditional System for Minimum Variability

Reagent
The period of oscillation (total loop dead time) must differ by more
than factor of 5 to prevent resonance (amplification of oscillations)
Reagent
Reagent

Feed

Small first tank provides a faster response Big footprint

and oscillation that is more effectively filtered and high cost!
by the larger tanks downstream per Eq. 5-3j
Improving Neutralizer pH Control

Traditional System for Minimum Reagent Use

Reagent
The period of oscillation (total loop dead time) must differ by more
than factor of 5 to prevent resonance (amplification of oscillations)
Feed Reagent
Reagent

Big footprint
and high cost!
The large first tank offers more cross neutralization of influents
Improving Neutralizer pH Control

Tight pH Control with Minimum Capital Investment

IL#1 – Interlock that prevents back fill of

reagent piping when control valve closes

IL#2 – Interlock that shuts off effluent flow until

vessel pH is projected to be within control band
Eductor
High Recirculation Flow

Reagent
Any Old Tank
Signal
Characterizer
LC LT
1-3 1-3

*IL#2 FT f(x)
1-1
Effluent AC
1-1
FC
1-2 AT
1-1
*IL#1
FT
Influent
1-2
10 to 20
pipe
diameters
Improving Neutralizer pH Control

Linear Reagent Demand Control

• Signal characterizer translates loop PV and SP from pH to % Reagent
Demand
– PV is abscissa of the titration curve scaled 0 to 100% reagent demand
– Piecewise segment fit normally used to go from ordinate to abscissa of curve
– Fieldbus block offers 21 custom space X,Y pairs (X is pH and Y is % demand)
– Closer spacing of X,Y pairs in control region provides most needed
compensation
– If neural network or polynomial fit used, beware of bumps and wild extrapolation
• Special configuration is needed to provide operations with pH interface to:
– See loop PV in pH and signal to final element
– Enter loop SP in pH
– Change mode to manual and change manual output
• Set point on steep part of curve shows biggest improvements from
– Reduction in limit cycle amplitude seen from pH nonlinearity
– Decrease in limit cycle frequency from final element resolution (e.g. stick-slip)
– Decrease in crossing of split range point
– Reduced reaction to measurement noise
– Shorter startup time (loop sees real distance to set point and is not detuned)
– Simplified tuning (process gain no longer depends upon titration curve slope)
– Restored process time constant (slower pH excursion from disturbance)
Improving Neutralizer pH Control

Case History 1- Existing Control System

Fuzzy
Logic

Waste
RCAS RCAS middle selector
ROUT
AC AC AY kicker AY

splitter splitter AT AT AT
AY AY AY
middle selector middle selector filter
FT FT
AY AY Attenuation
Tank
Stage 1 AT AT AT Stage 2 AT AT AT
Waste
Mixer Mixer FT
Improving Neutralizer pH Control

Case History 1 - New Control System

MPC-1
MPC-2

Waste
RCAS RCAS middle selector
ROUT
AC-1 AC-2 AY kicker AY

splitter splitter
AT AT AT
AY AY AY
middle selector middle selector filter
FT FT
AY AY Attenuation
Tank
Stage 1 AT AT AT Stage 2 AT AT AT
Waste
Mixer Mixer FT
Improving Neutralizer pH Control

Case History 1 - Opportunities for Reagent Savings

12

pH

Old Set Point

New Set Point 2 Reagent to Waste
Flow Ratio
New Ratio Old Ratio

Reagent
Savings
Improving Neutralizer pH Control

Case History 1 - Online Adaptation and Optimization

Model Predictive Control (MPC)

For Optimization of Actual Plant

Stage 1 and 2 Set Points

Optimization Actual Plant
(MPC1 and MPC2)
Tank pH and 2nd Stage Valves

Inferential Measurement
(Waste Concentration) Stage 1 and 2 Actual
and Diagnostics pH Set Points Reagent/Waste Ratio
(MPC SP)
Virtual
Reagent/Influent Ratio
(MPC CV)
Adaptation
Virtual Plant (MPC3)
Model
Influent Concentration
(MPC MV)
Model Predictive Control (MPC)
For Adaptation of Virtual Plant
Improving Neutralizer pH Control

Case History 1 - Online Model Adaptation Results

Actual Plant’s
Reagent/Influent
Flow Ratio

Virtual Plant’s
Reagent/Influent Adapted Influent Concentration
Flow Ratio (Model Parameter)
Improving Neutralizer pH Control

Case History 2 - Existing Neutralization System

93% 50%
Water
Acid Caustic

AT
Cation Anion
To EO

Final caustic
Final acid
adjustment
adjustment Pit
Improving Neutralizer pH Control

Case History 2 - Project Objectives

• Safe
• Responsible
• Reliable
– Mechanically
– Robust controls, Operator friendly
– Ability to have one tank out of service
• Balance initial capital against reagent cost
• Little or no equipment rework
Improving Neutralizer pH Control

Case History 2 - Cost Data

• 93%H2SO4 spot market price $2.10/Gal
• 50% NaOH spot market price $2.30/Gal

2k Gal 5k Gal 10k Gal 20k Gal 40k Gal

Tank $20k $30k $50k $80k $310k
Pump $25k $35k $45k $75k $140k
Improving Neutralizer pH Control

Case History 2 - Challenges

• Process gain changes by factor of 1000x
• Final element rangeability needed is 1000:1
• Final element resolution requirement is 0.1%
• Concentrated reagents (50% caustic and 93% sulfuric)
• Caustic valve’s ¼ inch port may plug at < 10% position
• Must mix 0.05 gal reagent in 5,000 gal < 2 minutes
• Volume between valve and injection must be < 0.05 gal
• 0.04 pH sensor error causes 20% flow feedforward error
• Extreme sport - extreme nonlinearity, sensitivity, and
rangeability of pH demands extraordinary requirements
for mechanical, piping, and automation system design
Improving Neutralizer pH Control

Case History 2 - Choices

Really big tank and thousands of mice

each with 0.001 gallon of acid or caustic
or
modeling and control
Improving Neutralizer pH Control

Case History 2 - Demineralized pH Titration Curve

pH

Slope
Improving Neutralizer pH Control

Case History 2 - Demineralized pH Control System

AY signal
pH set point
1-3 characterizer
Signal characterizers linearize loop
via reagent demand control
AC
1-1
LC LT
1-5 1-5
signal splitter
characterizer AY
1-4
Feed
AY
1-2
middle
NaOH Acid To other Tank
signal FT FT
Tank AY
selector 1-1 1-2
1-1

AT AT AT
Eductors 1-1 1-2 1-3
Static Mixer From other Tank

To other Tank
Downstream system
Improving Neutralizer pH Control

Case History 2 - Tuning for Conventional pH Control

Improving Neutralizer pH Control

Case History 2 - Tuning for Reagent Demand Control

Gain 10x larger

Improving Neutralizer pH Control

Case History 2 - Process Test Results

One of many spikes from

stick-slip of water valve

Tank 1 pH for Reagent Demand Control

Tank 1 pH for Conventional pH Control

Start of Step 4
Start of Step 2 (Slow Rinses)
(Regeneration)
Improving Neutralizer pH Control

Case History 2 - Control Logic

• If Tank pH is within control band, reduce tank level rapidly to minimum. (CL#1a). If
Tank pH is out of control band, close valve to downstream system and send effluent
to the other tank if it has more room (CL#1b).
• For caustic reagent valve signals of 0-10%, set control valve at 10%, pulse width
modulate isolation valve proportional to loop output, and increase loop filter time
and reset time to smooth out pulses (CL#2)
• If reagent valves are near the split range point, periodically (e.g. every 5 minutes)
shut the reagent valves and divert feed to other tank for 15 seconds to get tank pH
reading (CL#3).
• Coordinate opening and closing of reagent isolation valves with the opening and
closing of reagent control valves (CL#4)
• If feed is negligible and tank pH is within control band, shut off the recirculation
pump (CL#5)
Improving Neutralizer pH Control

Case History 2 - Dynamic Model in the DCS

Streams, pumps,
valves, sensors,
tanks, and mixers
are modules from
DeltaV composite
template library.

Each wire is a pipe

that is a process
stream data array
(e.g. pressure, flow,
temperature, density,
heat capacity, and
concentrations)

First principle
conservation of
material, energy,
components,
and ion charges
Improving Neutralizer pH Control

Case History 2 - Summary

• Study shows potential project savings overwhelm reagent savings
• Modeling removes uncertainty from design
– First principle relationships show how well mechanical, piping, and automation
system deal with nonlinearity, sensitivity, and rangeability
• Modeling enables prototyping of control improvements
– Linear reagent demand control speeds up response from PV on flat and
reduces oscillations from the PV on steep part of titration curve
– Control logic optimizes pH loops to minimize downtime and inventory to
maximize availability and minimize energy use
– Pulse width modulation of caustic at low valve positions minimizes plugging
– Recirculation within tank and between tanks offers maximum flexibility and
continuous, semi-continuous, and batch modes of operation
– Periodic observation of tank pH to determine best mode of operation
Neutralizer pH Control Lab
• Objective – See how optimizing setpoint can reduce reagent use
• Activities:
– Go to Main Display, select pH Lab02b
– Set Desired Run time = minimum run time
– Change from Explore to Run Mode
– Note process metrics when done
– Click on AC1-1 PID Faceplate and change pH setpoint from 7 to 4.5 pH
– Change from Explore to Run Mode
– Note process metrics when done
The Top Ten Signs You are
Ready for a Hawaiian Vacation
• (10) You give your boss the “hang loose” hand gesture
• (9) You day dream about hula dancers in hardhats
• (8) Your cubicle has a mosquito net with tropical sounds
• (7) You bring a kayak to the company’s waste pond
• (6) You ask “where is the company’s pupu stand”?
• (5) You tell your secretary she is wearing a nice muumuu
• (4) You play a ukulele in your office
• (3) You show up to a meeting in a Hawaiian shirt, shorts and sandals
• (2) You start answering your phone saying "Aloha“
• (1) You wear a snorkeling mask instead of glasses
Improving Reactor Temperature Control

Reactor Control Strategies

Improving Reactor Temperature Control

Reactor Cascade Control

Improving Reactor Temperature Control

Exothermic Reactions
Improving Reactor Temperature Control

Reactor Valve Position Control

Improving Reactor Temperature Control

Reactor Equilibrium Control

Improving Reactor Temperature Control

Reactor Rate of Change Control

A low
Improving Reactor Temperature Control

Reactor Override Control

Improving Reactor Temperature Control

Reactor Temperature Control Lab

• Objective – See how optimizing setpoint can reduce coolant use
• Activities:
– Go to Main Display, select Temperature Lab02a
– Set Desired Run time = minimum run time
– Change from Explore to Run Mode
– Note process metrics when done
– Click on TC1-1 PID Faceplate and change pH setpoint from 35 to 40 deg C
– Change from Explore to Run Mode
– Note process metrics when done
Improving Unit Op Temperature Control

Heat Exchanger Coolant Control

Improving Unit Op Temperature Control

Heat Exchanger By-Pass Control

Improving Unit Op Temperature Control

Heat Exchanger Feedforward Control

Improving Unit Op Temperature Control

Column Control by Manipulation of Distillate

Improving Unit Op Temperature Control

Column Control by Manipulation of Reflux

Improving Unit Op Temperature Control

Column Control by Manipulation of Steam

Improving Unit Op Temperature Control

Column Control by Manipulation of Bottoms

Improving Unit Op Temperature Control

Kiln Feedforward and Valve Position Control

Improving Unit Op Temperature Control

Kiln Differential Temperature Control

Improving Unit Op Temperature Control

Kiln Oxygen Control

Improving Unit Op Temperature Control

Crystallizer Control
Improving Unit Op Temperature Control

Extruder Specific Energy Control