Upload
jim-cahill
View
10.708
Download
2
Embed Size (px)
DESCRIPTION
Presented by Greg McMillan at Ascend Chocolate Bayou Plant Seminar March 10, 2011
Citation preview
1
Standards
Certification
Education & Training
Publishing
Conferences & Exhibits
pH Control Solutions
Ascend Chocolate Bayou Plant Seminar March 10, 2011
2
Introduction of Presenter• ISA Presenter
– Gregory K. McMillan ( E-mail: [email protected] ) – 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 and Advanced Temperature Measurement and Control - 2nd Edition. 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/
3
Source
4
ISA Automation Week - Oct 17-20
Call for PapersDeadline is March 28 !
5
Legends Cutler and Liptak Give Keynotes Legends Cutler and Liptak Give Keynotes
6
Key Benefits of Seminar
• Recognition of the opportunity and challenges of pH control• Alerts to important implementation considerations• Awareness of modeling and control options• Understanding the root causes of poor performance• Prioritization of improvements based on cost, time, and goal• Insights for applications and solutions
7
Section 1: pH Opportunity and Challenge
• Extraordinary Sensitivity and Rangeability• Deceptive and Severe Nonlinearity• Extraneous Effects on Measurement• Difficult Control Valve Requirements
8
Top Ten Signs of a Rough pH Startup
Food is burning in the operators’ kitchen The only loop mode configured is manual An operator puts his fist through the screen You trip over a pile of used pH electrodes The technicians ask: “what is a positioner?” The technicians stick electrodes up your nose The environmental engineer is wearing a mask The plant manager leaves the country Lawyers pull the plugs on the consoles The president is on the phone holding for you
9
Extraordinary Sensitivity and RangeabilitypH Hydrogen Ion Concentration Hydroxyl Ion Concentration0 1.0 0.000000000000011 ` 0.1 0.00000000000012 0.01 0.0000000000013 0.001 0.000000000014 0.0001 0.00000000015 0.00001 0.0000000016 0.000001 0.000000017 0.0000001 0.00000018 0.00000001 0.0000019 0.000000001 0.0000110 0.0000000001 0.000111 0.00000000001 0.00112 0.000000000001 0.0113 0.0000000000001 0.114 0.00000000000001 1.0
aH = 10-pH
pH = - log (aH)
aH = *g cH
cH * cOH = 10-pKw
Where:aH = hydrogen ion activity (gm-moles per liter)
cH = hydrogen ion concentration (gm-moles per liter)
cOH = hydroxyl ion concentration (gm-moles per liter)
g = activity coefficient (1 for dilute solutions)pH = negative base 10 power of hydrogen ion activitypKw = negative base 10 power of the water dissociation constant (14.0 at 25oC)
Hydrogen and Hydroxyl Ion Concentrations in a Water Solution at 25oC
10
5
5.5
6
6.5
7
7.5
8
8.5
9
0 10 20 30 40 50 60 70 80 90 100
Temp (C)
pH 5
pH 6
pH 6.5
pH 7
pH 7.5REAL pH 8
pH 9 pH 10
Measured pH
Effect of Water Dissociation (pKw) on Solution pH
11
Temperature Comp Parameters
Solution pH Temperature Correction
Isopotential Point Changeable for Special pH Electrodes
pH / ORP Selection
Preamplifier Location
Type of Reference Used
Ranging
AMS pH Range and Compensation Configuration
12
Reagent Titration Curves
0.00000000
2.00000000
4.00000000
6.00000000
8.00000000
10.00000000
12.00000000
14.00000000
0.00000000 0.00050000 0.00100000 0.00150000 0.00200000
Wt Frac Base
pH Calculated pH
Reagent Titration Curves
3.00000000
4.00000000
5.00000000
6.00000000
7.00000000
8.00000000
9.00000000
10.00000000
11.00000000
0.00099995 0.00099996 0.00099997 0.00099998 0.00099999 0.00100000 0.00100001 0.00100002 0.00100003 0.00100004 0.00100005
Wt Frac Base
pH Calculated pH
14
12
10
8
6
4
2
0
pH
Reagent / Influent Ratio
11
10
9
8
7
6
5
4
3
pH
Reagent / Influent Ratio
Despite appearances there are no straight lines in a titration curve (zoom in reveals another curve if there are enough data points - a big “IF” in neutral region)
For a strong acid and base the pKa are off-scale and the slope continually changes by a factor of ten for each pH unit deviationfrom neutrality (7 pH at 25 oC)
As the pH approaches the neutral point the response accelerates (looks like a runaway). Operators often ask what can be done to slow down the pH response around 7 pH.
Yet titration curves are essential for every aspect of pH systemdesign but you must get numerical values and avoid mistakessuch as insufficient data points in the area around the set point
Nonlinearity - Graphical Deception
13
Figure 3-1b: Weak Acid Titrated with a Strong Base
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
0.000 0.500 1.000 1.500 2.000
Reagent / Influent
pH Calculated pH
Weak Acid and Strong Base
pka = 4
Figure 3-1c: Strong Acid Titrated with a Weak Base
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
0.000 0.500 1.000 1.500 2.000
Reagent / Influent
pH Calculated pH
Strong Acid and Weak Base
pka = 10
Figure 3-1e: Weak 2-Ion Acid Titrated with a Weak 2-Ion Base
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
0.000 0.500 1.000 1.500 2.000
Reagent / Influent
pH Calculated pH
Multiple Weak Acids and Weak Bases
pka = 3
pka = 5
pka = 9
Nonlinearity - Graphical Deception
Slope moderatednear each pKa
pKa and curve changes with temperature!
Figure 3-1d: Weak Acid Titrated with a Weak Base
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
0.000 0.500 1.000 1.500 2.000
Reagent / Influent
pH Calculated pH
Weak Acid and Weak Base
pka = 4
pka = 10
14
Reagent Savings is Huge for Flat Part of Curve
pH
Reagent to Feed Flow Ratio
Reagent Savings Original
set point
Optimum set point
4
10
Oscillations could be due to non-ideal mixing, control valve stick-slip. or pressure fluctuations
15
Effect of Sensor Drift on Reagent Calculations
pH
Reagent to Feed Flow Ratio
4
10
6
8
FeedforwardReagent Error
FeedforwardpH Error
Sensor Drift
pH Set PointInfluent pH
The error in a pH feedforward calculationincreases for a given sensor error as theslope of the curve decreases. This resultCombined with an increased likelihood ofErrors at low and high pH means feedforwardCould do more harm than good when goingfrom the curve’s extremes to the neutral region.
Flow feedforward (ratio controlof reagent to influent flow) workswell for vessel pH control if there are reliable flow measurements with sufficient rangeability
Feedforward control always requires pH feedback correction unless the set point is on the flat part of the curve, use Coriolis mass flow meters and have constant influent and reagent concentrations
16
Double Junction Combination pH Electrode
W W
W
W
W
WW
W
W
W
Em
R10R9 R8R7
R6
R5
R4
R3
R2
R1
Er
E5
E4
E3
E2
E1
outer gellayer
inner gellayer
secondjunction
primary junction
solution ground
Process Fluid
silver-silver chlorideinternal electrode
silver-silver chlorideinternal electrode
potassium chloride (KCl) electrolyte in salt bridge between junctions
7 pH buffer
Ii
High acid or base concentrations can affect glass gel layer and reference junction potentialIncrease in noise or decrease in span or efficiency is indicative of glass electrode problemShift or drift in pH measurement is normally associated with reference electrode problem
Process ions try to migrate into porous reference junctionwhile electrolyte ionstry to migrate out
Nernst Equationassumes insideand outside gellayers identical
Measurementbecomes slow from a loss ofactive sites ora thin coatingof outer gel
17
Life Depends Upon Process Conditions
25 C 50 C 75 C 100 CProcess Temperature
Months
>100% increase in life from new glass designsfor high temperatures
High acid or base concentrations (operation at the extremes of the titration curve) decrease life for a given temperature. A deterioration in measurement accuracy and response time often accompanies a reduction in life. Consequently pH feedforward control is unreliable and the feedforward effect and timing is way off for such cases.
18
New High Temperature Glass Stays Fast
0
50
100
150
200
0 50 100 150 200
minutes
mV
New GlassOther
Glass electrodes get slow as they age. High temperatures cause premature aging
19
What is High Today may be Low Tomorrow
AB A
B A
BpH
time
Most calibration adjustments chase the short term errors shownbelow that arise from concentration gradients from imperfectmixing, ion migration into reference junction, temperature shifts, different glass surface conditions, and fluid streaming potentials.With just two electrodes, there are more questions than answers.
20
Control Valve Rangeability and Resolution
pH
Reagent FlowInfluent Flow
6
8
Influent pHB
A
Control BandSet point
BEr = 100% * Fimax * ---- Frmax
Frmax = A * Fimax
BEr = ---- A
Ss = 0.5 * Er
Where:A = distance to center of reagent error band on abscissa from influent pHB = 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) (%)
21
Speed of Response Seen by pH Loop
pH
Reagent FlowInfluent Flow
6
8
10
12
2
4
pH2
pH1
pH3
Fastest process response seen byLoop at inflection point (e.g. 7 pH)
Slow
Slow
(1) Excursion from pH1 to pH2 acceleration makes response look like a runaway to loop(2) Excursion from pH2 to pH3 deceleration is not enough to show true process time constant(3) Batch neutralizers without reagents consumed by reactions lack process self-regulation
that causes an integrating response aggravating overshoot from acceleration.
Apparent loss of investment in large well mixed volume can be restoredby signal characterization of pH to give abscissa as controlled variable!
22
Key Points
• pH electrodes offer by far the greatest sensitivity and rangeability of any measurement. To make the most of this capability requires an incredible precision of mixing, reagent manipulation, and nonlinear control. pH measurement and control can be an extreme sport
• Solution pH changes despite a constant hydrogen ion concentration because of changes in water dissociation constant (pKw) with temperature, and activity coefficients with ionic strength and water content
• Solution pH changes despite a constant acid or base concentration because of changes in the acid or base dissociation constants (pKa) with temperature
• Titration curves have no straight lines A zoom in on any supposed line should reveal another curve if there are sufficient data points
• Slope of titration curve at the set point has the greatest effect on the tightness of pH control as seen in control valve resolution requirement. The next most important effect is the distance between the influent pH and the set point that determines the control valve rangeability requirement
• Titration curves are essential for every aspect of pH system design and analysis• First step in the design of a pH system is to generate a titration curve at the
process temperature with enough data points to cover the range of operation and show the curvature within the control band (absolute magnitude of the difference between the maximum and minimum allowable pH)
23
Key Points
• Accuracy and speed of response of pH measurements stated in the literature assume the thin fragile gel layer of a glass electrode and the porous process junction of the reference electrode have had no penetration or adhesion of the process and are in perfect condition at laboratory conditions
• Time that glass electrodes are left dry or exposed to high and low pH solutions must be minimized to maximize the life of the hydrated gel layer
• Most accuracy statements and tests are for short term exposure• Long term error of pH measurements installed in the process is an order of
magnitude greater than the error normally stated in the literature • pH measurement error may look smaller on the flatter portion of a titration
curve but the associated reagent delivery error is larger • Cost of pH measurement maintenance can be reduced by a factor of ten by
more realistic expectations and calibration policies• Set points on the steep portion of a titration curve necessitate a reagent control
valve precision that goes well beyond the norm and offers the best test to determine a valve’s actual stick-slip in installed conditions
• Reagent valve resolution (stick-slip) may determine the number of stages of neutralization required, which has a huge impact on a project’s capital cost
• Approach to the neutral point looks like a runaway due to acceleration • Batch processes have less self-regulation and tend to ramp• Batch processes can have a one direction response for a given reagent
24
Section 2: Modeling and Control Options
• Virtual plant and imbedded process models• Minimization of capital investment• Cascade pH control• Online identification of titration curve• Batch pH control• Linear reagent demand control• Adapted reagent demand control• Smart split range point• Elimination of split range control• Model predictive control
25
Virtual Plant Setup
Advanced Control Modules
Process Module for Bioreactor or Neutralizer
Virtual PlantLaptop or Desktop
or Control System Station
Configuration and Graphics
26
Virtual Plant: http://www.processcontrollab.com/Recorded Deminars: http://www.modelingandcontrol.com/deminar_series.html
Virtual Plant Access
Free State of the Art Virtual Plant Not an emulation but a DCS
(SimulatePro) Independent Interactive Study Structural Changes “On the Fly” Advanced PID Options and Tuning
Tools Enough variety of valve, measurement,
and process dynamics to study 90% of the process industry’s control applications
Learn in 10 minutes rather than 10 years
Online Performance Metrics Standard Operator Graphics & Historian Control Room Type Environment No Modeling Expertise Needed No Configuration Expertise Needed Rapid Risk-Free Plant Experimentation Deeper Understanding of Concepts Process Control Improvement Demos Sample Lessons (Recorded Deminars)
A new easy fast free method of access is now available that eliminates IT security issues and remote access response delays
27
Case History 1- Existing Control System
Mixer
AttenuationTank
AY
AT
middle selector
AY
splitter
AC
AT
FT
FT
AT
AY
ATAT AT
AY
ATAT AT
Mixer
AY
FT
Stage 2Stage 1
middle selector
AC
Waste
Waste
middle selector
FuzzyLogic
RCAS RCAS
splitter
AY
filter
AYROUT
kicker
28
Case History 1 - New Control System
Mixer
AttenuationTank
AY
AT
middle selector
AY
splitter
AT
FT
FT
AT
AY
ATAT AT
AY
ATAT AT
Mixer
AY
FT
Stage 2Stage 1
middle selector
Waste
Waste
middle selectorRCAS RCAS
splitter
AY
filter
AYROUT
kickerAC-1 AC-2
MPC-2
MPC-1
29
Case History 1 - Opportunities for Reagent Savings
pH
Reagent to Waste Flow Ratio
Reagent Savings
2
12
Old Set Point
New Set Point
Old RatioNew Ratio
30
Case History 1 - Online Adaptation and Optimization
Actual PlantOptimization(MPC1 and MPC2)
Tank pH and 2nd Stage Valves
Stage 1 and 2 Set Points
Virtual Plant
Inferential Measurement(Waste Concentration)
and Diagnostics
Adaptation(MPC3)
Actual Reagent/Waste Ratio
(MPC SP)
ModelInfluent Concentration
(MPC MV)
Model Predictive Control (MPC)For Optimization of Actual Plant
Model Predictive Control (MPC)For Adaptation of Virtual Plant
Virtual Reagent/Influent Ratio
(MPC CV)
Stage 1 and 2 pH Set Points
31
Case History 1 - Online Model Adaptation Results
Adapted Influent Concentration(Model Parameter)
Actual Plant’sReagent/Influent
Flow Ratio
Virtual Plant’sReagent/Influent
Flow Ratio
32
Case History 2 - Existing Neutralization System
Water93%
Acid
50%
Caustic
Pit
Cation Anion
To EO
Final acid
adjustment
Final caustic
adjustment
AT
33
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
34
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
35
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
36
Really big tank and thousands of miceeach with 0.001 gallon of acid or caustic
or
modeling and control
Case History 2 - Choices Case History 2 - Choices
37
Case History 2 - Tuning for Conventional pH Control Case History 2 - Tuning for Conventional pH Control
38
Gain 10x larger
Case History 2 - Tuning for Reagent Demand Control Case History 2 - Tuning for Reagent Demand Control
39
Embedded Process Model for pH
40
Slope
pH
Titration Curve Matched to Plant
5/15/2008 Titration Curve Slope
0
5
10
15
20
2 3 4 5 6 7 8 9 10 11 12
pH
Slo
pe
Slope
41
Signal characterizers linearize loop via reagent demand control
AY 1-4
AC 1-1
AY 1-3
splitter
AT 1-3
AT 1-2
AT 1-1
AY 1-1
AY 1-2
middlesignal
selector
signalcharacterizer
signalcharacterizer
pH set point
Eductors
FT 1-1
FT 1-2
NaOH Acid
LT 1-5
Tank
Static Mixer
Feed
To other Tank
Downstream system
LC 1-5
From other Tank
To other Tank
Modeled pH Control System
42
Start of Step 2(Regeneration)
Start of Step 4(Slow Rinses)
One of many spikes of recirculation pH spikes from stick-slip of water valve
Tank 1 pH for Reagent Demand Control
Tank 1 pH for Conventional pH Control
Influent pH
Conventional pH versus Reagent Demand Control
43
Feed
Reagent
Reagent
ReagentThe period of oscillation (4 x process deadtime) and filter time(process residence time) is proportional to volume. To preventresonance of oscillations, different vessel volumes are used.
Small first tank provides a faster responseand oscillation that is more effectively filtered by the larger tanks downstream
Big footprintand high cost!
Traditional System for Minimum Variability
44
Reagent
Reagent
Feed
Reagent
Traditional System for Minimum Reagent UseTraditional System for Minimum Reagent Use
The period of oscillation (total loop dead time) must differ by morethan factor of 5 to prevent resonance (amplification of oscillations)
The large first tank offers more cross neutralization of influents
Big footprintand high cost!
45
Tight pH Control with Minimum Capital Investment
Influent
FC 1-2
FT 1-2
Effluent AC 1-1
AT 1-1
FT 1-1
10 to 20 pipe
diameters
f(x)
*IL#1
Re
ag
en
t
High Recirculation Flow
Any Old TankSignal
Characterizer
*IL#2
LT 1-3
LC 1-3
IL#1 – Interlock that prevents back fill ofreagent piping when control valve closes
IL#2 – Interlock that shuts off effluent flow untilvessel pH is projected to be within control band
Eductor
46
Linear Reagent Demand Control
• Signal characterizer converts 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 spaced 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 interface to:– Operator sees loop PV in pH and enters loop SP in pH– Operator can change mode to manual and change manual output– Operator sees both reagent demand % and PV trends
• 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)
47
Cascade pH Control to Reduce Downstream Offset
M
AT 1-2
Static Mixer
Feed
AT 1-1
FT 1-1
FT 1-2
Reagent
10 to 20pipe
diameters
Sum FC 1-1
Filter
Coriolis MassFlow Meter
f(x)
AC 1-1
AC 1-2
PV signalCharacterizer
RSP
f(x)
Flow Feedforward
SP signalcharacterizer
Trim of Inline Set Point
IntegralOnly
Controller
Linear ReagentDemand Controller
Any Old Tank
48
Full Throttle (Bang-Bang) Batch pH Control
Batch Reactor
AT 1-1
10 to 20 pipe
diameters
Filter
Delay
Sub Div
Sum
Dt
Cutoff
PastDpH
Rate ofChangeDpH/Dt
Mul
Total System Dead Time
Projected DpHNew pH
Old pH
Batch pH End Point
Predicted pHReagent
Section 3-5 in New Directions in Bioprocess Modeling and Control shows how this strategy is used as a head start for a PID controller
49
Linear Reagent Demand Batch pH Control
Batch Reactor
AC 1-1
AT 1-1
10 to 20 pipe
diameters
f(x)
Master Reagent DemandAdaptive PID Controller
Static Mixer
AC 1-1
AT 1-1
10 to 20 pipe
diameters
Secondary pHPI Controller
Signal CharacterizerUses Online
Titration Curve
FT 1-1
FC 1-1
FQ 1-1
FT 1-2
Online Curve Identification
Influent #1
Reduces injection and mixing delays and enables some crossneutralization of swings between acidic and basic influent. It issuitable for continuous control as well as fed-batch operation.
Influent #2
50
Adapted Reagent Demand Control
Neutralizer
AC 1-1
AT 1-1
10 to 20 diameters
f(x)
Master Reagent DemandAdaptive PID Controller
Static Mixer
AC 1-1
AT 1-1
10 to 20 diameters
Secondary pHPI Controller
Signal CharacterizerUses Online
Titration Curve
FT 1-1
FC 1-1
FQ 1-1
FT 1-2
Online Curve Identification
Influent
Reduces injection and mixing delays and enables somecross neutralization in continuous and batch operations
51
Recently Developed Adaptive Control
• Anticipates nonlinearity by recognizing old territory– Model and tuning settings are scheduled per operating region– Remembers what it has learned for preemptive correction
• Demonstrates efficiency improvement during testing– Steps can be in direction of optimum set point– Excess reagent use rate and total cost can be displayed online
• Achieves optimum set point more efficiently– Rapid approach to set point in new operating region
• Recovers from upsets more effectively– Faster correction to prevent violation– More efficient recovery when driven towards constraint
• Returns to old set points with less oscillation – Faster and smoother return with less overshoot
52
• For a first order plus deadtime process, only nine (9) models are evaluated each sub-iteration, first gain is determined, then deadtime, and last time constant.
• After each iteration, the bank of models is re-centered using the new gain, time constant, and deadtime
First Order Plus Deadtime Process
Estimated Gain, time constant, and deadtime
Multiple Model parameterInterpolation with re-centering
Changing process input
sKe DT
1
Gain
Tim
e C
onst
ant
Dead time
1
2
3
First Order plus Dead Time Model Identification
Changes in the process model can be used to diagnose changesin the influent and the reagent delivery and measurement systems
53
Model and tuning is scheduled based on pH
Scheduling of Learned Dynamics and Tuning
54
Adaptive Control Efficiently Achieves Optimum
hourly cost of excess reagent
hourly cost of excess reagent
total cost ofexcess reagent
total cost ofexcess reagent
pH
pH
55
Adaptive Control Efficiently Rejects Loads
hourly costof excess
hourly cost of excess
pH
total costof excessreagent
total cost of excessreagent
pH
56
Adaptive Control is Stable on Steep Slopes
pH
pH
57
Smart Split Range Point
222
111 S
KK
KKS
pv
pv
21 100 SGS
)100(2211
111 G
KKKK
KKS
pvpv
pv
G = split range gap (%)Kv1 = valve 1 gain (Flow e.u. / CO %)Kv2 = valve 2 gain (Flow e.u. / CO %)Kp1 = process gain for valve 1(PV e.u. / Flow e.u.)Kp2 = process gain for valve 2(PV e.u. / Flow e.u.)S1 = 1st split ranged span (PV e.u.)S2 = 2nd split ranged span (PV e.u.)
58
Smart Split Range Point
Neutralizer
AC 1-1
AT 1-1
PID Controller
Large(Coarse)
Small(Fine)
Splitter
Split RangeBlock
Reagent
For large valve 4x small valve flow:
PID Small LargeOut Valve Valve
0% 0% 0%20% 100% 0%20% 100% 0%100% 100% 100%
Smart in terms of valve gaincompensation but not smartin terms of valve sensitivity !
59
PID Valve Sensitivity and Rangeability Solution 1
Neutralizer
AC 1-1a
AT 1-1
PID Controlleror PIDPlus withsensitivity limit
Large(Coarse)
Small(Fine)
AC 1-1b
Proportional only Controlleror PIDPlus withsensitivity limit
Reagent
60
PID Valve Sensitivity and Rangeability Solution 2
Neutralizer
AC 1-1
AT 1-1
PID Controlleror PIDPlus withsensitivity limit
Large(Coarse)
Small(Fine)
ZC 1-1
Integral only Controlleror PIDPlus withsensitivity limit
Reagent
61
MPC Valve Sensitivity and Rangeability Solution
manipulated variables
Small (Fine)Reagent Valve SP
NeutralizerpH PV
Small (Fine)Reagent Valve SP
cont
rolled
va
riab
le
MPC Large (Coarse)Reagent Valve SP
cont
rolled
va
riab
le
null
Model Predictive Controller (MPC) setup for rapid simultaneous throttling of a fine and coarse control valves that addressesboth the rangeability and resolution issues. This MPC canpossibly reduce the number of stages of neutralization needed
http://www.controlglobal.com/articles/2005/533.htmlhttp://www.modelingandcontrol.com/2009/03/application_notes.html
62
MPC Valve Sensitivity and Rangeability Solution MPC Valve Sensitivity and Rangeability Solution
63
MPC Valve Sensitivity and Rangeability Solution
64
MPC Valve Sensitivity and Rangeability Solution
Successive Load Upsets Process Set Point Change Trim Valve Set Point Change
CriticalProcess Variable
CoarseValve
TrimValve
65
MPC Maximization of Low Cost Reagent
manipulated variables
High Cost FastFeed SP
Critical PV(normal PE)
Low Cost SlowFeed SP
(lowered PE)
contr
olled
vari
able
Maximize
MPC Low Cost SlowFeed SP
null
opti
miz
ati
on
vari
able
66
MPC Maximization of Low Cost Reagent
67
MPC Maximization of Low Cost Reagent
Riding Max SPon Lo Cost MV
Riding Min SPon Hi Cost MV
Critical CV
Lo Cost Slow MV
Hi Cost Fast MV
LoadUpsets
Set PointChanges
LoadUpsets
Set PointChanges
Low Cost MV Maximum SP Increased
Low Cost MV Maximum SP Decreased
Critical CV
68
MPC Maximization of Low Cost Reagent
manipulated variables
SupplementalReagent Flow SP
Cheap Reagent Flow PV
Neutralizer pH PV
Acidic Feed Flow SP
Supplemental Reagent
Valve Position
con
trolled
v
ari
ab
lecon
str
ain
t
vari
ab
le
MPC
disturbance variable
Acid Feed Flow SP
null
op
tim
izati
on
v
ari
ab
le
nullMaximize
Note that cheap reagent valve is wide open and feed is maximized to keep supplemental reagent valve at minimum throttle position for minimum stick-slip
69
Key Points
• More so than for any other loop, it is important to reduce dead time for pH control because it reduces the effect of the nonlinearity
• Filter the feedforward signal to remove noise and make sure the corrective action does not arrive too soon and cause inverse response
• The effectiveness of feedforward control greatly depends upon the ability to eliminate reagent delivery delays
• If there is a reproducible influent flow measurement use flow feedforward, otherwise use a head start to initialize the reagent flow for startup
• The reliability and error of a pH feedforward is unacceptable if the influent or feed pH measurement is on the extremities of the titration curve
• Use a Coriolis or magnetic flow meter for reagent flow control • Every reagent valve must have a digital valve controller (digital positioner)• Except for fast inline buffered systems, use cascade control of pH to reagent
flow to compensate for pressure upsets and enable flow feedforward • Linear reagent demand can restore the time constant and capture the investment
in well mixed vessels, provide a unity gain for the process variable, simply and improve controller tuning, suppress oscillations and noise on the steep part of the curve, and speed up startup and recovery from the flat part of the curve
70
Key Points
• Changes in the process dynamics identified online can be used to predict and analyze changes in the influent, reagent, valve, and sensor
• New adaptive controllers will remember changes in the process model as a function of operating point and preemptively schedule controller tuning
• Use inline pH control, mass flow meters, linear control valves, and dynamic compensation to automatically identify the titration curve online
• Use gain scheduling or signal characterization based on the titration curve to free up an adaptive controller to find the changes in the curve
• Batch samples should be taken only after the all the reagent in the pipeline and dip tube has drained into the batch and been thoroughly mixed
• Use a wide open reagent valve that is shut or turned over to pH loop based on a predicted pH from ramp rate and dead time to provide the fastest pH batch/startup
• Use online titration curve identification and linear reagent demand pH control for extremely variable and sharp or steep titration curve
• Use an online dynamic pH estimator to provide a much faster, smoother, and more reliable pH value, if the open loop dead time and time constant are known and there are feed and reagent coriolis mass flow meters
• Use linear reagent demand model predictive control for dead time dominant or
interacting systems and constraint or valve position control
71
Section 3: Plant Design and Maintenance
• Common Problems with Titration Curves• Effect of Measurement Selection and Installation• Options to improve accuracy and maintenance• Effect of piping design, vessel type, and mixing pattern• Implications of oversized and split ranged valves • Online Troubleshooting
72
Common Problems with Titration Curves
• Insufficient number of data points were generated near the equivalence point • Starting pH (influent pH) data were not plotted for all operating conditions• Curve doesn’t cover the whole operating range and control system overshoot• No separate curve that zooms in to show the curvature in the control region• No separate curve for each different split ranged reagent• Sequence of the different split ranged reagents was not analyzed• Back mixing of different split ranged reagents was not considered• Overshoot and oscillation at the split ranged point was not included• Sample or reagent solids dissolution time effect was not quantified• Sample or reagent gaseous dissolution time and escape was not quantified• Sample volume was not specified • Sample time was not specified• Reagent concentration was not specified • Sample temperature during titration was different than the process temperature • Sample was contaminated by absorption of carbon dioxide from the air • Sample was contaminated by absorption of ions from the glass beaker• Sample composition was altered by evaporation, reaction, or dissolution • Laboratory and field measurement electrodes had different types of electrodes • Composite sample instead of individual samples was titrated • Laboratory and field used different reagents
73
• Inherently ignores single measurement failure of any type including the most insidious PV failure at set point
• Inherently ignores slowest electrode• Reduces noise and spikes particularly for steep curves• Offers online diagnostics on electrode problems
– Slow response indicates coated measurement electrode – Decreased span (efficiency) indicates aged or dehydrated glass electrode– Drift or bias indicates coated, plugged, or contaminated reference electrode
or high liquid junction potential– Noise indicates dehydrated measurement electrode, streaming potentials,
velocity effects, ground potentials, or EMI• Facilitates online calibration of a measurement
Middle Signal Selection Advantages Middle Signal Selection Advantages
For more Information on Middle Signal Selection see Feb 5, 2010 post http://www.modelingandcontrol.com/2010/02/exceptional_opportunities_in_p_11.html
74
Online Diagnostics for Cracked Glass
Reference Electrode Glass
ElectrodeSolution Ground
3K 0-5 M
Cracked Glass!
Reference - Joseph, Dave, “What’s the Real pH of the Stream”, Emerson Exchange 2008
• Cracked Glass Fault– pH Glass electrode normally has high impedance of 50-150 Meg-ohm– Glass can be cracked at the tip or further back inside the sensor– Recommended setting of 10 Megohm will detect even small cracks
75
Online Diagnostics for Coated Sensor
Coated Sensor!
Reference Electrode
Glass ElectrodeSolution
Ground
40k 150 M
• Coated Sensor Detection Can– activate sensor removal and
cleaning cycle– place output on hold while
sensor is cleaned
Reference - Joseph, Dave, “What’s the Real pH of the Stream”, Emerson Exchange 2008
76
Temperature Comp Parameters
Solution pH Temperature Correction
Isopotential Point Changeable for Special pH Electrodes
pH / ORP Selection
Preamplifier Location
Type of Reference Used
Ranging
AMS pH Range and Compensation Configuration
77
Impedance Diagnostics On/Off
Reference Impedance
Warning and Fault Levels
Reference Zero Offset
Calibration Error Limit
Glass Electrode Impedance
Warning and Fault Levels
Glass Impedance Temp Comp
(Prevents spurious errors due to Impedance decrease with Temperature)
AMS pH Diagnostics Configuration
78
Live Measurements and Status
Calibration Constants from Last Calibrations
Buffer Calibration Type & Buffer Standard Used
Sensor Stabilization Criteria Zero Offset Beyond this Limit will create a Calibration Error
If you want to know more about Buffer Calibration, hit this button…
AMS pH Calibration Setup
79
AMS pH Auxiliary Variables Dashboard
80
Horizontal Piping Arrangements
AE
AE
AE
AEAEAE
5 to 9 fps to minimize coatings0.1 to 1 fps to minimize abrasion
20 to 80 degreesThe bubble inside the glass bulbcan be lodged in tip of a probe that is horizontal or pointed up or caught at the internal electrodeof a probe that is vertically down
pressure drop foreach branch mustbe equal to to keep the velocities equal
Series arrangement preferred to minimize differences in solids, velocity, concentration, and temperature at each electrode!
10 OD10 OD
20 pipe diameters
20 pipe diameters
static mixer or pump
flush
drain
flush
drain
throttle valve toadjust velocity
throttle valve toadjust velocity
81
Vertical Piping ArrangementsA
E AE
AE
AE
AE
AE
5 to 9 fpscoating abrasion
10 O
D1
0 O
D
0.1 to 1 fps
holeorslot
Orientation of slot in shroud
throttle valve toadjust velocity
throttle valve toadjust velocity
Series arrangement preferred to minimize differences in solids, velocity, concentration, and temperature at each electrode!
82
Options for Maximum Accuracy
• A spherical or hemi-spherical glass measurement and flowing junction reference offers maximum accuracy, but in practice maintenance prefers:– A refillable double junction reference to reduce the complexity of installation and
the need to adjust reference electrolyte flow rate – This electrode is often the best compromise between accuracy and maintainability.
– A solid reference to resist penetration and contamination by the process and eliminate the need to refill or replace reference particularly for high and nasty concentrations and pressure fluctuations – This electrode takes the longest time to equilibrate and is more prone to junction effects but could be right choice in applications where accuracy requirements are low and maintenance is high.
• Select best glass and reference electrolyte for process • Use smart digital transmitters with built-in diagnostics• Use middle signal selection of three pH measurements
– Inherent auto protection against a failure, drift, coating, loss in efficiency, and noise (see February 5, 2010 entry on http://www.modelingandcontrol.com/ )
• Allocate time for equilibration of the reference electrode• Use “in place” standardization based on a sample with the same
temperature and composition as the process. If this is not practical, the middle value of three measurements can be used as a reference. The fraction and frequency of the correction should be chosen to avoid chasing previous calibrations
• Insure a constant process fluid velocity at the highest practical value to help keep the electrodes clean and responsive
83
Wireless pH Transmitters Eliminate Ground Spikes
Wired pH ground noise spike
Temperature compensated wireless pH controlling at 6.9 pH set point
Incredibly tight pH control via 0.001 pH wireless resolution
setting still reduced the number of communications by 60%
84
Wireless Bioreactor Adaptive pH Loop TestWireless Bioreactor Adaptive pH Loop Test
85
Enhanced PID Algorithm PID integral mode is
restructured to provide integral action to match the process response in the elapsed time (reset time set equal to process time constant)
PID derivative mode is modified to compute a rate of change over the elapsed time from the last new measurement value
PID reset and rate action are only computed when there is a new value
If transmitter damping is set to make noise amplitude less than sensitivity limit, valve packing and battery life is dramatically improved
Enhancement compensates for measurement sample time suppressing oscillations and enabling a smooth recovery from a loss in communications further extending packing -battery life
+
+
+
+
Elapsed Time
Elapsed Time
TD
Kc
Kc
TD
http://www2.emersonprocess.com/siteadmincenter/PM%20DeltaV%20Documents/
Whitepapers/WP_DeltaV%20PID%20Enhancements%20for%20Wireless.pdf
Link to PIDPlus White Paper
86
Flow Response - Enhanced vs. Traditional PID
Traditional PID Sensor PV
Enhanced PID Sensor PV
87
pH Response - Enhanced vs. Traditional PID
Enhanced PID Sensor PV
Traditional PID Sensor PV
88
Failure Response - Enhanced vs. Traditional PID
Traditional PID Sensor PV
Enhanced PID Sensor PV
89
Everyday Mistakes in pH System Design
M
Mistake 7 (gravity flow)
AT 1-1
AT 1-3
AT 1-2
Mistake 4 (horizontal tank)
reagentfeed tank
Mistakes 5 and 6(backfilled dip tube &injection short circuit)
Mistake 11 (electrode in pump suction)
Mistake 8 (valve too far away)
Mistake 9 (ball valve with no positioner)
Mistake 10 (electrodesubmerged in vessel)
Mistake 12 (electrode too far downstream)
Mistake 3 (single stage for set point at 7 pH)
Influent (1 pH)
Mistake 1: Missing, inaccurate, or erroneous titration curveMistake 2: Absence of a plan to handle failures, startups, or shutdowns
90
Mixing Pattern and Vessel Geometry Implications
Stagnant Zone
Stagnant Zone
MFeed
Reagent
AT 1-3
Short Circuiting
Stagnant Zone
PlugFlow
91
Oversized Reagent Valves are a Big Problem
dead band
Dead band
Stick-Slip is worse near closed position
Signal (%)
0
Stroke (%) Digital positioner
will force valve shut at 0% signal
Pneumatic positionerrequires a negative % signal to close valve
The dead band and stick-slip is greatest near the closed position so valves that ride the seat from over sizing or split ranged operation create a large limit cycle
Dead band is 5% - 50%without a positioner !
Limit cycle amplitude is operating point dependent and can be estimated as:stick-slip (%) multiplied by valve characteristic slope (pph/%) and by titration curve slope (pH/pph)
92
Key Points
• The time that glass electrodes are left dry or exposed to high and low pH solutions must be minimized to maximize the life of the hydrated gel layer
• Most accuracy statements and tests are for short term exposure before changes in the glass gel layer or reference junction potential are significant
• The pH measurement error may look smaller on the flatter portion of a titration curve but the associated reagent delivery error is larger
• The cost of pH measurement maintenance can be reduced by a factor of ten by more realistic expectations and calibration policies
• The onset of a coating of the glass measurement electrode shows up as a large increase in its time constant and response time
• The onset of a non conductive coating of the reference electrode shows up as a large increase in its electrical resistance
• Non-aqueous and pure water streams require extra attention to shielding and process path length and velocity to minimize pH measurement noise
• Slow references may be more stable for short term fluctuations from imperfect mixing and short exposure times from automated retraction
• The fastest and most accurate reference has a flowing junction but it requires regulated pressurization to maintain a small positive electrolyte flow
• The best choice might not be the best technical match to the application but the electrode with the best support by maintenance and operations and vendor
93
Key Points
• For non abrasive solids, installation in a recirculation line with a velocity of 5 to 9 fps downstream of a strainer and pump may delay onset of coatings
• For abrasive solids and viscous fluids, a thick flat glass electrode can minimize coatings, stagnant areas, and glass breakage
• For high process temperatures, high ion concentrations, and severe fouling, consider automatic retractable assemblies to reduce process exposure
• When the fluid velocity is insufficient to sweep electrodes clean, use an integral jet washer or a cleaning cycle in a retractable assembly
• The control system should schedule automated maintenance based on the severity of the problem and production and process requirements
• pH measurements can fail anywhere on or off the pH scale but middle signal selection will inherently ride out a single electrode failure of any type
• Equipment and piping should have the connections for three probes but a plant should not go to the expense of installing three measurements until the life expectancy has been proven to be acceptable for the process conditions
• The more an electrode is manually handled, the more it will need to be removed• A series installation of multiple probes insures the electrodes will see the same
velocity and mixture that is important for consistent performance• Wireless pH control of static mixers with enhanced algorithm can provide a
exceptional setpoint response and measurement failure protection
94
Key Points
• A system considered to be well mixed may be poorly mixed for pH control • To be “well mixed” for pH control, the deviation in the reagent to influent flow
from non ideal mixing multiplied by the process gain must be well within the control band
• Back mixing (axial mixing) creates a beneficial process time constant and plug flow or radial mixing creates a detrimental process dead time for pH control
• The agitation in a vessel should be vertical axial pattern without rotation and be intense enough to break the surface but not cause froth
• The actual equipment dead time is often larger than the turnover time because of non ideal mixing patterns and fluid entry and exit locations
• Horizontal tanks are notorious for short circuiting, stagnation, and plug flow that cause excessive dead time and an erratic pH response
• The dead time from back filled reagent dip tubes or injection piping is huge• To provide isolation, use a separate on-off valve and avoid the specification of
tight shutoff and high performance valves for throttling reagent • Set points on the steep portion of a titration curve necessitate a reagent control
valve precision that goes well beyond the norm and offers the best test to determine a valve’s actual stick-slip in installed conditions
• Reagent valve stick-slip may determine the number of stages of neutralization required, which has a huge impact on a project’s capital cost