Effective Use of PID Controllers ISA New Orleans Standards Certification Education & Training Publishing Conferences & Exhibits

Transcription

1 Effective Use of PID Controllers ISA New Orleans Standards Certification Education & Training Publishing Conferences & Exhibits 1

2 Presenter Greg is a retired Senior Fellow from Solutia/Monsanto and an ISA Fellow. Greg was an adjunct professor in the Washington University Saint Louis Chemical Engineering Department Presently, Greg contracts as a consultant in DeltaV R&D via CDI Process & Industrial and is a part time employee of Experitec and MYNAH. 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 Greg is the author of 20 books on process control, his most recent being Advanced Temperature Measurement and Control. Greg has been the monthly Control Talk columnist for Control magazine since 2002 and has started a Control Talk Blog. Greg s expertise is available on the Control Global and Emerson modeling and control web sites: 2

3 Resources 3

4 Top Ten Ways to Impress Management with Trends (10) Make large setpoint changes that zip past valve dead band and nonlinearities. (9) Change the setpoint to operate on the flat part of the titration curve. (8) Select the tray with minimum process sensitivity for column temperature control. (7) Pick periods when the unit was down. (6) Decrease the time span so that just a couple data points are trended. (5) Increase the reporting interval so that just a couple data points are trended. (4) Use really thick line sizes. (3) Add huge signal filters. (2) Increase the process variable scale span so it is at least 10x control region (1) Increase the historian's data compression so changes are screened out 4

5 Contribution of Each PID Mode Proportional (P mode) - increase in gain increases P mode contribution Provides an immediate reaction to magnitude of measurement change to minimize peak error and integrated error for a disturbance Too much gain action causes fast oscillations (close to ultimate period) and can make noise and interactions worse Provides an immediate reaction to magnitude of setpoint change for P action on Error to minimize rise time (time to reach setpoint) Too much gain causes falter in approach to setpoint Integral (I mode) - increase in reset time decreases I mode contribution Provides a ramping reaction to error (SP-PV) to minimize integrated error if stable (since error is hardly ever exactly zero, integral action is always ramping the controller output) Too much integral action causes slow oscillations (slower than ultimate period) Too much integral action causes an overshoot of setpoint (no sense of direction) Derivative (D mode) - increase in rate time increases D mode contribution Provides an immediate reaction to rate of change of measurement change to minimize peak error and integrated error for a disturbance Too much rate action causes fast oscillations (faster than ultimate period) and can make noise and interactions worse Provides an immediate reaction to rate of change of setpoint change for D action on Error to minimize rise time (time to reach setpoint) Too much rate causes fast oscillation in approach to setpoint

6 Contribution of Each PID Mode kick from filtered derivative mode %CO 1 Signal (%) step from proportional mode seconds/repeat %CO 2 = %CO 1 repeat from Integral mode %SP PID structure with proportional, integral, and derivative action on error Time (seconds) Contribution of Each PID Mode for a Step Change in the Set Point Structure of PID on error (β=1 and γ=1)

7 Effect of Gain on P-Only Controller Red is 150% of maximum, Green is 100% of maximum, Purple is 50% of maximum of Gain Setting

8 Effect of Reset Time on PI Controller Red is 150% of maximum, Green is 100% of maximum, Purple is 50% of maximum Reset Time

9 Effect of Rate Time on PD Controller Red is 200% of maximum, Green is 100% of maximum, Purple is 0% of maximum Rate Time

10 Proportional Mode Basics Note that many analog controllers used proportional band instead of gain for the proportional mode tuning setting. Proportional band is the % change in the process variable ( %PV) needed to cause a 100% change in controller output ( %CO). A 100% proportional band means a 100% %PV would cause a 100 % %CO (a gain of 1). It is critical that users know the units of their controller gain setting and convert accordingly. Gain = 100 % / Proportional Band Proportional Mode Advantages Minimize dead time from stiction and backlash Minimize rise time Minimize peak error Minimize integrated error Proportional Mode Disadvantages Abrupt changes in output upset operators Abrupt changes in output upset other loops Amplification of noise 10

11 Integral Mode Basics Note that many analog controllers used reset settings in repeats per minute instead of reset time for the integral mode tuning setting. Repeats per minute indicate the number of repeats of the proportional mode contribution in a minute. Today s reset time settings are minutes per repeat or seconds per repeat which gives the time to repeat the proportional mode contribution. Often the per repeat term is dropped giving a reset time setting in minutes or seconds. Seconds per repeat = 60 / repeats per minute Integral Mode Advantages Eliminate offset Minimize integrated error Smooth movement of output Integral Mode Disadvantages Limit cycles Overshoot Runaway of open loop unstable reactors 11

12 Derivative Mode Basics Nearly all derivative tuning settings are given as a rate time in seconds or minutes. The effective rate time setting must never be greater than the effective reset time setting. The effective settings are for an ISA Standard Form. The advantages and disadvantages of the derivative mode are similar to that of the proportional mode except the relative advantages is less and the relative disadvantages are greater for the derivative mode. Seconds = 60 minutes Derivative Mode Advantages Minimize dead time from stiction and backlash Minimize rise time Minimize peak error Minimize integrated error Derivative Mode Disadvantages Abrupt changes in output upset operators Abrupt changes in output upset other loops Amplification of noise 12

13 Reset Gives Operations What They Want Should steam or water valve be open? TC-100 Reactor Temperature CO PV SP temperature steam valve opens SP 50% PV water valve opens? time

14 Open Loop Time Constant (controller in manual) Signal (%) %CO Controller is in Manual Open Loop Error E o (%) 0.63 E o %PV %SP 0 θ o τ o Time (seconds) Dead Time (Time Delay) Open Loop (process) Time Constant (Time Lag)

15 Closed Loop Time Constant (controller in auto) Signal (%) %CO Controller is in Automatic %SP %SP 0.63 %SP %PV 0 θ o Dead Time (Time Delay) τ c Closed Loop Time Constant (Time Lag) Lambda (λ) Time (seconds)

16 Top Ten Signs Loops Need to be Tuned (10) Lots of trials and errors. (9) When asked what the controller gain setting is, the answer is given in %. (8) When asked what the controller reset time setting is, the answer is in repeats/min. (7) The data historian compression setting is 25%. (6) There is more recycle than product. (5) Valves are wearing out. (4) Tempers are wearing thin. (3) Operators are placing bets on what loop will cause the next shutdown. (2) The output limits are set to keep the valve from moving. (1) Preferred mode is manual. 16

17 Conversion of Signals for PID Algorithm Final Control Element SP SCLR % % SUB % %PV PID % %CO SCLR OUT (e.u.) AO Control Valve MV (e.u.) Process Equipment PV (e.u.) SCLR PID DCS PV - Primary Variable SV - Second Variable* TV - Third Variable* FV - Fourth Variable* AI Smart Transmitter PV (e.u.) Measurement Sensing Element * - additional HART variables The scaler block (SCLR) that convert between engineering units of application and % of scale used in PID algorithm is embedded hidden part of the Proportional-Integral-Derivative block (PID) To compute controller tuning settings, the process variable and controller output must be converted to % of scale and time units of dead times and time constants must be same as time units of reset time and rate time settings!

18 Series Form Form in analog controllers and early DCS available as a choice in most modern DCS β Gain proportional %SP Inverse Reset All signals are % of scale in PID algorithm but Time inputs and outputs are in engineering units filter γ Rate Time Filter Time = α Rate Time integral Σ %CO filter derivative %PV filter Σ Switch position for no derivative action 18

19 Parallel Form Form in a few early DCS and PLC and in many control theory textbooks β Proportional Gain Setting proportional %SP Integral Gain Setting All signals are % of scale in PID algorithm but inputs and outputs are in engineering units filter integral Σ %CO γ Derivative Gain Setting derivative %PV filter 19

20 ISA Standard Form β Default Form in most modern DCS Gain proportional %SP Inverse Reset Time All signals are % of scale in PID algorithm but inputs and outputs are in engineering units filter γ Rate Time Filter Time = α Rate Time integral Σ %CO filter derivative %PV filter 20

21 Positive Feedback Implementation of Integral Form for Enhanced PID developed for wireless Gain * Back out positive feedback of Feedforward (*FF) and ISA Standard Form of Proportional (*P) and Derivative (*D) modes with β and γ factors %SP filter + β For reverse action, Error = %SP - %PV + All signals are % of scale in PID algorithm but inputs and outputs are in engineering units Positive Feedback For zero error Out1 = 0 Out1 Σ Out2 P = (β 1) Gain %SP Filter Time = Reset Time Σ P D Feedforward Σ *P *FF FF %CO %PV γ filter + Rate Time filter Filter Time = α Rate Time filter derivative Filter Time = Reset Time *D E-R Switch position for external reset feedback E-R is external reset (e.g. secondary %PV s ) Dynamic Reset Limit 21

22 Conversion of Series to ISA Form To convert from Series to ISA Standard Form controller gain: K c = T ' i + T T ' i ' d K ' c To convert from Series to ISA Standard Form reset (integral) time: ' ' Ti + Td ' ' T i = Ti = Ti + T T ' i To convert from Series to ISA Standard Form rate time: T d = T ' i T ' i + T ' d T ' d ' d Interaction factor Primed tuning settings are Series Form Note that if the rate time is zero, the ISA Standard and Series Form settings are identical. When using the ISA Standard Form, if the rate time is greater than ¼ the reset time the response can become oscillatory. If the rate time exceeds the reset time, the response can become unstable from a reversal of action form these modes. The Series Form inherently prevents this instability by increasing the effective reset time as the rate time is increased. 22

23 Anti Reset Windup (ARW) and Output Limits For digital positioners and precise throttling valves ARW & Out Lo Lim = 0%, ARW & Out Hi Lim = 100% For pneumatic positioners & on-off heritage valves Lo Lim = -5%, Hi Lim = 105% ARW set inside output limits to get thru zone of ineffective valve response (stick-slip, shaft windup, & poor sensitivity) For primary PID in cascade control, limits are set to match secondary setpoint limits in engineering units

24 Checklist for PID Migration - 1 There are many features and parameters that vary with the DCS supplier. It is imperative the DCS documentation and supplier expertise be fully utilized and all migrations tested by a real time simulation for stability. Note the default of 0% low and 100% high output and ARW limits do not change to match changes made in output scale or engineering units. For cascade control did you set the output scale of the primary PID in engineering units of the PV scale of the secondary loop? For cascade control did you set the primary PID low and high output limits in engineering units to match setpoint limits of secondary PID? Did you set the anti-reset windup (ARW) limits to match the output limits using same units as output limits unless there is some special need for ARW limits to be set otherwise? Did you convert controller gain setting units (being especially aware of the inverse relationship between proportional band and gain)? Did you convert reset units setting (being especially aware of the inverse relationship between repeats per minute and seconds per repeat)? Did you convert rate units setting and make the alpha setting the same for the rate filter? If rate time is not zero and ISA Standard Form is used, did you convert Series Form gain, reset, and rate settings to corresponding ISA Standard Form settings? 24

25 Checklist for PID Migration - 2 For override control if the positive feedback implementation of integral mode is used, did you remove the filter on external reset signal used to prevent walk-off since this filter is already there? For cascade control, id you turn on external reset feedback (dynamic reset limit) and use PV of secondary loop for external reset feedback to automatically prevent burst of oscillations from violation of cascade rule that secondary loop must be 5x faster than primary loop? For slow or sticky valve, did you turn on external reset feedback (dynamic reset limit) and use a fast PV readback for external reset feedback to automatically prevent burst of oscillations from violation of cascade rule that positioner feedback loop must be 5x faster than primary loop and to prevent limit cycles from stick-slip? Did you realize the PV readback must normally be faster than a secondary HART variable update time? For wireless control and at-line or on-line analyzer, did you use an enhanced PID developed for wireless that suspends integral action between updates (PIDPlus option) and uses elapsed time in the derivative action. The external-reset option should automatically be turned on? Did you make sure the BKCAL signals are connected properly paying particular attention to the propagation of the BKCAL settings for intervening blocks for split range, signal characterization, and override control? 25

26 Top 10 Things You Shouldn't Say When You Enter a Control Room (10) Does this hard hat make my butt look big? (9) At the last plant I was in we always did it this way. (8) I added alarms to each loop. (7) Does that flare out there always shoot up that high? (6) Ooooh! Did you mean to do that? (5) Can't somebody do something about all those alarms? (4) We just downloaded the version released yesterday (3) Here, I will show you how to operate this plant. (2) Are you ready to put all your loops in Remote Cascade? (1) We want a "lights out" plant! 26

27 Triple Cascade Loop Block Diagram Process Primary Controller Secondary Flow Controller Digital Valve Controller DCS Valve Positioner Process SP PID Flow SP External Reset BKCAL PID CO External Reset BKCAL AO PID* Drive Signal I/P Relay Position (Valve Travel) Control Valve Flow Meter Process PV AI PV AI Position Loop Feedback * most positioners use proportional only Process Sensor Secondary (Inner) Loop Feedback Primary (Outer) Loop Feedback

28 Effect of Slow Secondary Tuning (cascade control) Secondary loop slowed down by a factor of 5 Secondary CO Primary PV Secondary SP Secondary CO Secondary SP Primary PV

29 External Reset Feedback (Dynamic Reset Limit) Prevents PID output changing faster than a valve, VFD, or secondary loop can respond Secondary PID slow tuning Secondary PID SP Filter Time Secondary PID SP Rate Limit AO, DVC, VFD SP Rate Limit Slow Valve or VFD Use PV for BKCAL_OUT Position used as PV if valve is very slow and readback is fast Enables Enhanced PID for Wireless Stops Limit cycles from deadband, backlash, stiction, and threshold sensitivity or resolution limits Key enabling feature that simplifies tuning and creates more advanced opportunities for PID control

30 PID Structure Options (1) PID action on error (β = 1 and γ = 1) (2) PI action on error, D action on PV (β = 1 and γ = 0) (3) I action on error, PD action on PV (β = 0 and γ = 0) (4) PD action on error, no I action (β = 1 and γ = 1) (5) P action on error, D action on PV, no I action (β = 1 and γ = 0) (6) ID action on error, no P action (γ = 1) (7) I action on error, D action on PV, no P action (γ = 0) (8) Two degrees of freedom controller (β and γ adjustable 0 to 1)

31 (1) PID action on error Fastest response to rapid (e.g. step) SP change by Step in output from proportional mode Spike in output from derivative mode can be made more like a kick by decreasing gamma factor (γ <1) Zero dead time from deadband, resolution limit, & stiction Burst of flow may affect other uses of fluid Operations do not like sudden changes in output Fast approach to SP more likely to cause overshoot Setpoint filter & rate limits eliminate step & overshoot

32 (2) PI action on error, D action on PV Slightly slower SP response than structure (1) Still have step from proportional mode Spike or bump from derivative mode eliminated Decrease in SP response speed is negligible if Output hits output limit due to large SP change or PID gain Rate time is less than total loop dead time Alpha factor is increased (α > 0.125) (rate filter increased) Setpoint filter & rate limits eliminate step & overshoot Most popular structure choice

33 (3) I action on error, PD action on PV Provides gradual change in output for SP change Slows down SP response dramatically Eliminates overshoot for SP changes Used for bioreactor temperature and ph SP changes (overshoot is much more important than cycle time) Used for temperature startup to warm up equipment Generally not recommended for secondary loops

34 (4-5) No Integral action Used if integral action adversely affects process Used if batch response is only in one direction Must set bias (output when PV = SP) Highly exothermic reactors use structure 4 because integral action and overshoot can cause a runaway 10x reset time (T i > 40x dead time) to prevent runaway Traditionally used on Total Dissolved Solids (TDS) drum and surge tank level control because of slow integrating response and permissibility of SP offset. Low controller gain (K c ) cause slow rolling oscillations due to violation of inequality for integrating process. The inequality is commonly violated since K i (integrating process gain) is extremely small on most vessels (K i < %/sec/%). Most common problem is use of too small of a reset time for vessel batch composition and temperature, level, and gas pressure control causing violation of following rule K c 2 * Ti > K i

35 (6-7) No Proportional Action Predominantly used for valve position control (VPC) Parallel valve control (VPC SP & PV are small valve desired & actual position, respectively, & VPC out positions large valve) Optimization (VPC SP & PV are limiting valve desired & actual position, respectively, & VPC out optimizes process PID SP) VPC reset time > 10x residence time to reduce interaction VPC reset time > K c T i of process PID to reduce interaction VPC tuning is difficult & too slow for fast & large disturbances Better solution is external reset feedback & SP rate limits

36 Improvement in Batch Temperature by Elimination of Integral action degrees C Batch temperature response in a single ended temperature control. Integral action causes overshoot. Typical Batch Temperature Time (min) Setpoint PV CO% degrees C Batch temperature response in a single ended temperature control. PD on error. No I action. Batch Temperature (new tuning) Time (min) Setpoint PV CO% 36

37 (8) Two Degrees of Freedom β and γ SP weighting factors are adjusted to balance fast approach & minimal overshoot for SP response Alternative is using SP lead-lag with lag = reset time and lead = 20% of lag to achieve fast SP response with minimal overshoot

38 Effect of Options on SP Response

39 Top Ten Reasons to Use a DCS for Your BBQ (10) Automated recipes (9) Predicted BBQ times (8) Five-course meal no problem (7) Don't have to watch cooking shows (6) Feed-forward control (5) Process control comes home (4) Children want to become automation engineers (3) Spouse finally appreciates your expertise (2) Griller not grilled (1) More time to drink beer 39

40 Fed-Batch and Startup Time Reduction - 1 PID on Error Structure Maximizes the step and kick 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 step and kick (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 increase-open or the analog output block, I/P, or positioner reverses the signal for a increase-close.

41 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 dead time. For more details, check out the Control magazine article Full Throttle Batch and Startup Response. A dead time (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 dead time (θ o ) 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/θ o 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 dead time (θ o ) approaching or less than the largest process time constant (τ p ), 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 T 98 process response time (T 98 = θ o + 4 τ o ).

42 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 stiction, 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 White paper Wireless.pdf Dead Time Compensation The simple addition of a delay block with the dead time set equal to the total loop dead time to the external reset signal for the positive feedback implementation of integral action described in Deminar #3 for the dynamic reset limit option 96b7-9229e4a6b5ba. The controller reset time can be significantly reduced and the controller gain increased if the delay block dead time is equal or slightly less than the process dead time as studied in Advanced Application Note 3

43 Fed-Batch and Startup Time Reduction - 4 Feed Maximization Model Predictive Control described in Application Note 1 Override control 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 and the Control magazine article Unlocking the Secret Profiles of Batch Reactors

44 Dead Time Compensator Configuration Must enable dynamic reset limit! Insert deadtime block

45 Dead Time Myths Busted Dead time is eliminated from the loop. The smith predictor, which created a PV without dead time, fools the controller into thinking there is no dead time. However, for an unmeasured disturbance, the loop dead time 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 dead time, and dead time squared, respectively. Control is faster for existing tuning settings. The addition of dead time 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. For a PID already tuned for maximum disturbance rejection, the gain can be increased by 250%. For dead time dominant systems where the total loop dead time 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 dead time to primary process time constant ratio was 10:1, you could decrease the reset time by a factor of 10. Further study is needed as to whether the minimum reset time is a fraction of the underestimated dead time plus the PID module execution time where the fraction depends upon the dead time to time constant ratio For access to Deminar 10 ScreenCast Recording or SlideShare Presentation go to

46 Dead Time Myths Busted Compensator works better for loops dominated by a large dead time. The reduction in rise time is greatest and the sensitivity to per cent dead time modeling error particularly for an overestimate of dead time is least for the loop that was dominated by the process time constant. You could have a dead time estimate that was 100% high before you would see a significant jagged response when the process time constant was much larger than the process dead time. For a dead time 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 dead time is decreased to zero. While the magnitude of the error in dead time seems small, you have to remember that for an industrial temperature control application, the loop dead time and process time constant would be often at least 100 times larger. For a 400 second dead time and 10,000 second process time constant, a compensator dead time 200 seconds smaller or 400 seconds larger than actual would start to cause a problem. In contrast, the dead time dominant loop developed a jagged response for a dead time 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 dead time block in the BKCAL path may have helped. An underestimate of the dead time leads to instability. In tuning calculations for a conventional PID, a smaller than actual dead time can cause an excessively oscillatory response. Contrary to the effect of dead time on tuning calculations, a compensator dead time smaller than actual dead time will only cause instability if the controller is tuned aggressively after the dead time compensator is added. An overestimate of the dead time leads to sluggish response and greater stability. In tuning calculations for a conventional PID, a larger than actual dead time can cause an excessively slow response. Contrary to the effect of dead time on tuning calculations, a compensator dead time greater than actual dead time will cause jagged irregular oscillations.

47 Top Ten Reasons Why Automation Engineers Makes Great Spouses or at Least a Wedding Gifts (10) Reliable from day one (9) Always on the job (8) Low maintenance (minimal grooming, clothing, and entertainment costs (7) Many programmable features (6) Stable (5) Short settling time (4) No frills or extraneous features (3) Relies on feedback (2) Good response to commands and amenable to real time optimization (1) Readily tuned

48 General PID Checklist - 1 Does the measurement scale cover the entire operating range, including abnormal conditions? Is the valve action correct (increase-open for fail close and increase-close for fail open)? Is the control action correct (direct for reverse process and reverse for direct process if the valve action is set)? Is the best Form selected (ISA standard form)? Is the obey setpoint limits in cascade and remote cascade mode option selected? Are the external reset feedback (BKCAL) signals correctly connected between blocks? Is the PV for BKCAL selected in the secondary loop PID? Is the best Structure selected (PI action on error, D action on PV for most loops)? Is the setpoint track PV in manual option selected to provide a faster initial setpoint response unless the setpoint must be saved in PID? 48

49 General PID Checklist - 2 Are setpoint limits set to match process, equipment, and valve constraints? Are output limits set to match process, equipment, and valve constraints? Are anti-reset windup (ARW) limits set to match output limits? Is the module scan rate (PID execution time) less than 10% of minimum reset time? Is the signal filter time less than 10% of minimum reset time? Is the PID tuned with a proven tuning method or by an auto-tuner or adaptive tuner? Is the rate time less than ½ the dead time (the rate is typically zero except for temperature) Is external-reset feedback (dynamic reset limit) enabled for cascade control, analog output (AO) setpoint rate limits, and slow control valves or variable speed drives? Are AO setpoint rate limits set for blending, valve position control, and surge valves? Is integral deadband greater than limit cycle PV amplitude? Can an enhanced PID be used for loops with wireless instruments or analyzers? 49

50 Feedforward Applications Feedforward is the most common advanced control technique used - often the feedforward signal is a flow or speed for ratio control that is corrected by a feedback process controller (Flow is the predominant process input that is manipulated to set production rate and to control process outputs (e.g. temperature and composition)) Blend composition control - additive/feed (flow/flow) ratio Column temperature control - distillate/feed, reflux/feed, stm/feed, and bttms/feed (flow/flow) ratio Combustion temperature control - air/fuel (flow/flow) ratio Drum level control - feedwater/steam (flow/flow) ratio Extruder quality control - extruder/mixer (power/power) ratio Heat exchanger temperature control - coolant/feed (flow/flow) ratio Neutralizer ph control - reagent/feed (flow/flow) ratio Reactor reaction rate control - catalyst/reactant (speed/flow) ratio Reactor composition control - reactant/reactant (flow/flow) ratio Sheet, web, and film line machine direction (MD) gage control - roller/pump (speed/speed) ratio Slaker conductivity control - lime/liquor (speed/flow) ratio Spin line fiber diameter gage control - winder/pump (speed/speed) ratio Feedforward is most effective if the loop deadtime is large, disturbance speed is fast and size is large, feedforward gain is well known, feedforward measurement and dynamic compensation are accurate Setpoint feedforward is most effective if the loop deadtime exceeds the process time constant and the process gain is well known For more discussion of Feedforward see May 2008 Control Talk

51 Feedforward Implementation - 1 Feedforward gain can be computed from a material or energy balance ODE * & explored for different setpoints and conditions from a plot of the controlled variable (e.g. composition, conductivity, ph, temperature, or gage) vs. ratio of manipulated variable to independent variable (e.g. feed) but is most often simply based on operating experience * Plots are based on an assumed composition, pressure, temperature, and/or quality For concentration and ph control, the flow/flow ratio is valid if the changes in the composition of both the manipulated and feed flow are negligible. For column and reactor temperature control, the flow/flow ratio is valid if the changes in the composition and temperature of both the manipulated and feed flow are negligible. For reactor reaction rate control, the speed/flow is valid if changes in catalyst quality and void fraction and reactant composition are negligible. For heat exchanger control, the flow/flow ratio is valid if changes in temperatures of coolant and feed flow are negligible. For reactor temperature control, the flow/flow ratio is valid if changes in temperatures of coolant and feed flow are negligible. For slaker conductivity (effective alkali) control, the speed/flow ratio is valid if changes in lime quality and void fraction and liquor composition are negligible. For spin or sheet line gage control, the speed/speed ratio is valid only if changes in the pump pressure and the polymer melt quality are negligible. Dynamic compensation is used to insure the feedforward signal arrives at same point at same time in process as upset Compensation of a delay in the feedforward path > delay in upset path is not possible

52 Feedforward Implementation - 2 Feedback correction is essential in industrial processes While technically, the correction should be a multiplier for a change in slope and a bias for a change in the intercept in a plot of the manipulated variable versus independent variable (independent from this loop but possibly set by another PID or MPC), a multiplier creates scaling problems for the user, consequently the correction of most feedforward signal is done via a bias. The bias correction must have sufficient positive and negative range for worst case. Model predictive control (MPC) and PID loops get into a severe nonlinearity by creating a controlled variable that is the ratio. It is important that the independent variable be multiplied by the ratio and the result be corrected by a feedback loop with the process variable (composition, conductivity, gage, temperature, or ph) as the controlled variable. Feedforward gain is a ratio for most load upsets. Feedforward gain is the inverse of the process gain for setpoint feedforward. Process gain is the open loop gain seen by the PID (product of manipulated variable, process variable, and measurement variable gain) that is dimensionless. Feedforward action must be in the same direction as feedback action for upset. Feedforward action is the opposite of the control action for setpoint feedforward. Feedforward delay and lag adjusted to match any additional delay and lag, respectively in path of upset so feedforward correction does not arrive too soon. Feedforward lead is adjusted to compensate for any additional lag in the path of the manipulated variable so the feedforward correction does not arrive too late. The actual and desired feedforward ratio should be displayed along with the bias correction by the process controller. This is often best done by the use of a ratio block and a bias/gain block instead of the internal PID feedforward calculation.

53 Linear Reagent Demand Control (PV is X axis of Titration Curve) 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 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 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) 53

54 Output Tracking for SP Response Head-Start logic for startup & batch SP changes: For SP change PID tracks best/last startup or batch final settling value for best/last rise time less total loop deadtime Closed loop time constant is open loop time constant (λ f =1) Not as fast as Bang-Bang (PID OUT is not at output limit) Bang-Bang logic for startup & batch SP changes: For SP change PID tracks output limit until the predicted PV one deadtime into future gets within a deadband of setpoint, the output is then set at best/last startup or batch final settling value for one deadtime Implementation uses simple DT block (loop deadtime) to create an old PV subtracted from the new PV to give a delta PV that is added to old PV to create a PV one deadtime into future Works best on slow batch and integrating processes

55 Output Tracking for Protection - 1 Open Loop Backup to prevent compressor surge: Once a compressor gets into surge, cycles are so fast & large that feedback control can not get compressor out of surge When compressor flow drops below surge SP or a precipitous drop occurs in flow, PID tracks an output that provides a flow large enough to compensate for the loss in downstream flow for a time larger than the loop dead time plus the surge period. Open Loop Backup to prevent RCRA violation: An excursion < 2 ph or > 12 ph for even a few sec can be a recordable RCRA violation regardless of downstream volume When an inline ph system PV approaches the RCRA ph limit the PID tracks an incremental output (e.g. 0.25% per sec) opening the reagent valve until the ph sufficiently backs away Open Loop Backup for evaporator conductivity

56 Open Loop Backup Configuration - 2 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

57 Output Tracking for Protection - 3 Feedback Action

58 Output Tracking for Protection - 4 Open Loop Backup

59 RCRA ph Kicker Optimization of ph filter and kicker increment saved $50K in reagent costs MPC-1 MPC-2 Waste RCAS AC-1 RCAS AC-2 ROUT AY Kicker middle selector AY splitter AY splitter AY AY AT AT AT FT Stage 1 middle selector AY AT AT AT FT middle selector AY Stage 2 AT AT AT Filter Attenuation Tank Mixer Mixer FT

60 Evaporator Conductivity Kicker Conductivity spike WBL Flow Kicker

61 Setpoint Filter PID SP filter reduces overshoot enabling fast tuning Setpoint filter time set equal reset time PID SP filter coordinates timing of flow ratio control Simultaneous changes in feeds for blending and reactions Consistent closed loop response for model predictive control PID SP filter sets closed loop time constant PID SP filter in secondary loop slows down cascade control system rejection of primary loop disturbances Secondary loop must be > 4x faster than primary loop Primary PID must have dynamic reset limit enabled Setpoint Lead-Lag minimizes overshoot and rise time Lag time = reset time Lead time = 20% lag time

62 Setpoint Rate Limits AO & PID SP rate limits minimize disruption while protecting equipment and optimizing processes Offers directional moves suppression Enables fast opening and slow closing surge valve VPC fast recovery for upset and slow approach to optimum AO SP rate limits minimize interaction between loops Less important loops are made 10x slower than critical loops PID driving AO SP or secondary PID SP rate limit must have dynamic reset limit enabled so no retuning is needed PID faceplate should display PV of AO to show rate limiting

63 Top Ten Reasons to do APC from your Home (10) Can immediately implement an inspiration. (9) Can watch the ball game on one of your screens. (8) Get to wear shorts and sandals. (7) Get to listen to music rather than alarms. (6) Lose weight from not eating doughnuts. (5) Can BBQ while solving control problem. (4) No more lonely nights and meals. (3) Your kids start to recognize you. (2) Your kids want to become automation engineers. (1) Your spouse starts to offer you advanced process control. 63

64 Enhanced PID for Wireless Features Positive feedback implementation of reset with external-reset feedback (dynamic reset limit) Immediate response to a setpoint change or feedforward signal or mode change Suspension of integral action until change in PV Integral action is the exponential response of the positive feedback filter to the change in controller output in elapsed time (the time interval since last update) Derivative action is the PV or error change divided by elapsed time rather than PID execution

65 Flow Setpoint Response Enhanced PID Sensor PV Traditional PID Sensor PV

66 Flow Load Response Enhanced PID Sensor PV Traditional PID Sensor PV

67 Flow Signal Failure Response Enhanced PID Sensor PV Traditional PID Sensor PV

68 ph Setpoint Response Enhanced PID Sensor PV Traditional PID Sensor PV

69 ph Load Response Enhanced PID Sensor PV Traditional PID Sensor PV

70 ph Sensor Failure Response Enhanced PID Sensor PV Traditional PID Sensor PV

71 Stop Limit Cycles Traditional PID Enhanced PID PID PV PID Output Limit Cycles from Valve Stick-Slip

72 Benefits Extend Beyond Wireless - 1 The PID enhancement for wireless offers an improvement wherever there is an update time in the loop. In the broadest sense, an update time can range from seconds (wireless updates and valve or measurement sensitivity limits) to hours (failures in communication, valve, or measurement). Some of the sources of update time are: Wireless update time for periodic reporting (default update rate) Wireless measurement trigger level for exception reporting (trigger level) Wireless communication failure Broken ph electrode glass or lead wires (failure point is about 7 ph) Valve with backlash (deadband) and stick-slip (resolution) Operating at split range point (no response & abrupt response discontinuity) Valve with solids, high temperature, or sticky fluid (plugging and seizing) Plugged impulse lines Analyzer sample, analysis cycle, and multiplex time Analyzer resolution and threshold sensitivity limit To completely stop a valve limit cycle from backlash or stick-slip, measurement updates must not occur due to noise

73 Benefits Extend Beyond Wireless - 2 Enhanced PID executes for a change in setpoint, feedforward, or remote output to provide an immediate reaction based on PID structure The improvement in control by the enhanced PID is most noticeable as the update time becomes much larger than the 63% process response time (defined in the white paper as the sum of the process deadtime and time constant). When the update time becomes 4 times larger than this 63% process response time ( 98% response time frequently cited in the literature), the feedforward and controller gains can be set to provide a complete correction for changes in the measurement and setpoint. Helps ignore inverse response and errors in feedforward timing Helps ignore discontinuity (e.g. steam shock) at split range point Helps extend packing life by reducing oscillations and hence valve travel Since enhanced PID can be set to execute only upon a significant change in user valve position, this PID as a valve position controller offers less interaction and cycling for optimization of unit operations by increasing reactor feed, column feed or increasing refrigeration unit temperature, or decreasing compressor pressure till feed, vent, coolant, and/or steam, valves are at maximum good throttle position. Website entries on Enhanced PID Benefits

74 Why over 100 PID Tuning Rules? Aidan O Dwyer s Handbook of PI and PID Controller Tuning Rules - 2 nd Edition has over 500 pages of rules The originators all think their rules are best due to Gamesmanship Diverse sources of change Diverse objectives Diverse dynamics Diverse metrics

75 Convergence of Tuning Rules The most popular PID rules converge to the same equations for 99% of temperature, composition, level, and gas pressure loops despite diversity of metrics, dynamics, objectives, and sources of change if the following is used: Tuning to minimize the effect of unmeasured disturbances Tuning to maximize absorption of variability (e.g. surge tank level) Dead time block in identification of process dynamics Primary PID Setpoint Lag = reset time and Lead = 20% of Lag Analog output setpoint rate limit and PID external-reset feedback Enhanced PID developed for wireless with threshold sensitivity For the remaining cases: For drastic deceleration from dead time dominance decrease gain, reset time, and rate time For severe acceleration from runaway reaction, increase gain, reset time, and rate time

76 Diverse Sources of Change Raw material and recycle composition and impurities Weather (temperature, humidity, snow, rain) Utility temperature and pressure Operators (production rate changes and manual actions) Interactions and Optimization Batch sequences and on-off control Startups, transitions, and shutdowns Measurement and process noise Limit cycles

77 Diverse Process Objectives Maximize safety Prevent activation of relief devices and Safety Instrumented Systems (SIS) Maximize equipment, environmental, & process protection Minimize product variability Minimize limit cycles Minimize oscillatory loop response Minimize interaction between loops Maximize coordination between loops Maximize process capacity and efficiency Increase production rate and decrease raw material and utility use

78 Diverse Process Objectives Automated Risk Reduction SIS PID

79 Diverse Process Objectives Maximize Protection Eliminate temperature shock and water hammer Slow action of control valve in direction of causing shock Eliminate compressor surge Slow closing of surge valves and downstream user valves Fast opening of surge valves Eliminate flare stack emissions Fast opening of runaway reactor coolant valves Eliminate RCRA ph Violations Fast opening of base reagent valve when approaching 2 ph Fast opening of acid reagent valve when approaching 12 ph

80 Diverse Process Objectives Minimize Product Variability Minimize cycling from valve discontinuities Suspension of integral action when valve is not moving or for an impending unnecessary crossing of the split range point Minimize oscillatory response Slow approach to setpoint and suspension of integral action between updates from analyzers and wireless transmitters Minimize interaction between loops Slow and fast action of less and more critical loop, respectively Maximize coordination of loops Identical ratioed rates of change of feeds particularly for plug flow reactors, and inline systems, such as blenders and static mixers

81 Diverse Process Objectives Maximize Efficiency and Capacity Use PID for valve position control (VPC) to increase feed or reduce raw material or energy use for valve constraint. Slow approach by VPC to optimum to avoid upsetting loops Fast getaway by VPC for upset to avoid running out of valve Suspension of integral action in VPC for valve that is not moving or whose movements are inconsequential

82 Key PID Features for VPC Feature Function Advantage 1 Advantage 2 Direction Velocity Limits Limit VPC Action Speed Based on Direction Prevent Running Out of Valve Minimize Disruption to Process Dynamic Reset Limit Limit VPC Action Speed to Process Response Direction Velocity Limits Prevent Burst of Oscillations Adaptive Tuning Automatically Identify and Schedule Tuning Eliminate Manual Tuning Compensation of Nonlinearity Feedforward Preemptively Set VPC Out for Upset Prevent Running Out of Valve Minimize Disruption Enhanced PID (PIDPlus) Suspend Integral Action until PV Update Eliminate Limit Cycles from Stiction & Backlash Minimize Oscillations from Interaction & PV Update Delay

83 Examples of Optimization by VPC Optimization VPC PID PV VPC PID SP VPC PID Out Minimize Prime Reactor Feed Max Throttle Position Compressor or Pump Mover Energy Flow PID Out Pressure SP Minimize Boiler Fuel Cost Steam Flow PID Out Max Throttle Position Boiler Pressure SP Minimize Boiler Fuel Cost Equipment Temperature PID Out Max Throttle Position Boiler Pressure SP Minimize Chiller or CTW Energy Equipment Temperature PID Out Max Throttle Position Chiller or CTW Temperature SP Minimize Purchased Reagent or Fuel Cost Purchased Reagent or Fuel Flow PID Out Min Throttle Position Waste Reagent Or Fuel Flow SP Minimize Total Reagent Use Final Neutralization Stage ph PID Out Min Throttle Position First Neutralization Stage ph PID SP Maximize Reactor Production Rate Reactor or Condenser Temperature PID Out Max Throttle Position Feed Flow or Reaction Temperature SP Maximize Reactor Production Rate Reactor Vent Pressure PID Out Max Throttle Position Feed Flow or Reaction Temperature SP Maximize Column Production Rate Reboiler or Condenser Flow PID Out Max Throttle Position Feed Flow or Column Pressure SP Maximize Ratio or Feedforward Accuracy Process Feedback Correction PID Out 50% (Zero Correction) Flow Ratio or Feedforward Gain

84 Liquid Reactants (Jacket CTW) Liquid Product Optimization ratio calc FY 1-6 LY 1-8 ZC1-4 OUT reactant A residence time calc CAS FC 1-1 FT 1-1 FC 1-2 CAS LC 1-8 LT 1-8 PT 1-5 TT 1-3 FT 1-5 PC 1-5 ZC1-4 is an enhanced PID VPC FC 1-1 CAS vent TC 1-3 ZC 1-4 reactant B FT 1-2 TT 1-4 TC 1-4 return Valve position controller (VPC) setpoint is the maximum throttle position. The VPC should turn off integral action to prevent interaction and limit cycles. The correction for a valve position less than setpoint should be slow to provide a slow approach to optimum. The correction for a valve position greater than setpoint must be fast to provide a fast getaway from the point of loss of control. Directional velocity limits in AO with dynamic reset limit in an enhanced PID that tempers integral action can achieve these optimization objectives. AT 1-6 FC 1-7 AC 1-6 makeup CTW FT 1-7 product 84

85 Liquid Reactants (Jacket CTW) Gas & Liquid Products Optimization ratio calc FY 1-6 LY 1-8 ZY1-1 OUT reactant A residence time calc CAS FC 1-1 FT 1-1 FC 1-2 CAS LC 1-8 LT 1-8 TT 1-10 PT 1-5 W TT 1-3 FT 1-5 PC 1-5 product TC 1-10 TC 1-3 ZC 1-5 ZY-1 IN1 ZC 1-10 reactant B FT 1-2 TT 1-4 TC 1-4 return ZY-1 IN2 FC1-1 CAS ZC-5 OUT low signal selector ZY ZC OUT ZC-4 OUT CAS AT 1-6 FC 1-7 AC 1-6 makeup CTW ZC 1-4 ZY-1 IN3 ZC1-4, ZC-5, & ZC-10 are enhanced PID VPC FT 1-7 product 85

86 Innovative PID System to Optimize Ethanol Yield and Carbon Footprint Corn Production Rate Enhanced PID setpoint AC 1-4 SC 1-4 AY 1-4 AT 1-4 XY 1-4 NIR-T Fermentable Starch Correction Average Fermentation Time Enhanced PID XC 1-4 Feedforward DX 2-4 Slurry Solids Enhanced PID DC 2-4 RCAS FC 1-5 Dilution Water FT 1-5 FC 1-6 Backset Recycle FT 1-6 DT 2-4 Coriolis Meter Slurry Tank 1 Slurry Tank 2 Lag and Delay DY 2-4 Predicted Fermentable Starch 86

87 Loop Block Diagram (First Order Approximation) Delay Lag Gain θ d τ L Load Upset K d DV Delay <=> Dead Time Lag <=>Time Constant Delay Lag Gain Secondary Delay Secondary Lag Primary Delay Primary Lag Gain θ v τ v Valve K v = slope of installed flow characteristic K v F v θ s τ s θ p τ p K p Process τ o is the largest lag in the loop (hopefully τ p ) For self-regulating processes: K o = K v K p K m PV For near integrating processes: K i = K v (K p / τ p ) K m PID % %CO K c T i T d Local Set Point %SP % K m = 100% / span ½ of Wireless Default Update Rate % %PV Delay Lag Gain Lag Delay Lag τ c2 Lag θ c Controller τ c1 K m τ m2 θ m2 τ m1 θ m1 Measurement Delay First Order Approximation: θ ο θ v + θ s + θ p + θ m1 + θ m2 + θ c + Y τ v + Y τ s + Y τ m1 + Y τ m2 + Y τ c1 + Y τ c2 (set by automation system design for flow, pressure, level, speed, surge, and static mixer ph control) 87

88 Open Loop Response of Self-Regulating Process Response to change in controller output with controller in manual % Process Variable (%PV) or % Controller Output (%CO) K o = %PV / %CO Self-regulating process gain (%/%) %CO %CO Maximum speed in 4 dead times is critical speed %PV 0.63 %PV %PV Noise Band observed total loop dead time θ o τ o ideally τ p Self-regulating process open loop negative feedback time constant Time (seconds) 88

89 Open Loop Response of Integrating Process Response to change in controller output with controller in manual % Process Variable (%PV) or % Controller Output (%CO) K i = { [ %PV 2 / t 2 ] [ %PV 1 / t 1 ] } / %CO %CO Integrating process gain (%/sec/%) %CO Maximum ramp rate in 4 dead times is used to estimate integrating process gain %PV ramp rate is %PV 1 / t 1 ramp rate is %PV 2 / t 2 observed total loop dead time θ o Time (seconds) 89

90 Open Loop Response of Runaway Process Response to change in controller output with controller in manual % Process Variable (%PV) or % Controller Output (%CO) K o = %PV / %CO Runaway process gain (%/%) %CO Acceleration For safety reasons, tests are terminated within 4 dead times before noticeable acceleration %PV 1.72 %PV Noise Band observed total loop dead time θ o τ o must be τ p runaway process open loop positive feedback time constant Time (seconds) 90

91 Diverse Loop Metrics Peak and integrated errors for load disturbances Rise time for setpoint change (time to reach setpoint) Overshoot for setpoint change Settling time for setpoint change Standard deviation of oscillations

92 Diverse Metrics Peak and Integrated Error The use of a setpoint lead-lag with the lag equal to the reset time and the lead 20% of the lag will provide a fast setpoint response with minimal overshoot despite tuning for maximum load rejection

93 Ultimate Limit to Loop Performance Peak error is proportional to the ratio of loop dead time to 63% response time (Important to prevent SIS trips, relief device activation, surge prevention, and RCRA ph violations) Total loop deadtime that is often set by automation design E x θo = ( θ + τ ) o o E o Largest lag in loop that is ideally set by large process volume Integrated error is proportional to the ratio of loop dead time squared to 63% response time (Important to minimize quantity of product off-spec and total energy and raw material use) E i = 2 θo ( θ + τ ) o o E o For a sensor lag (e.g. electrode or thermowell lag) or signal filter that is much larger than the process time constant, the unfiltered actual process variable error can be found from the equation for attenuation

94 Effect of Disturbance Lag on Integrating Process Periodic load disturbance time constant increased by factor of 10 Adaptive loop Baseline loop Adaptive loop Baseline loop Primary reason why bioreactor control loop tuning and performance for load upsets is a non issue!

95 Practical Limit to Loop Performance Peak error decreases as the controller gain increases but is essentially the open loop error for systems when total dead time >> process time constant E x = (1 + 1 K K o c ) E o Open loop error for fastest and largest load disturbance Integrated error decreases as the controller gain increases and reset time decreases but is essentially the open loop error multiplied by the reset time plus signal delays and lags for systems when total dead time >> process time constant E i = T i + t K o x K + τ c f E o Peak and integrated errors cannot be better than ultimate limit - The errors predicted by these equations for the PIDPlus and deadtime compensators cannot be better than the ultimate limit set by the loop deadtime and process time constant

96 Implied Dead Time from Slow Tuning Slow tuning (large Lambda) creates an implied dead time where the loop performs about the same as a loop with fast tuning and an actual dead time equal to the implied dead time (θ i ) θ =.5 ( λ + θ ) i 0 o For most aggressive tuning Lambda is set equal to observed dead time (implied dead time is equal to observed dead time) Money spent on improving measurement and process dynamics (e.g. reducing measurement delays and process dead times) will be wasted if the controller is not tuned faster to take advantage of the faster dynamics You can prove most any point you want to make in a comparison of control system performance, by how you tune the PID. Inventors of special algorithms as alternatives to the PID naturally tend to tune the PID to prove their case. For example Ziegler-Nichols tuning is often used to show excessive oscillations that could have be eliminated by cutting gain in half

97 Disturbance Speed Effect of load disturbance lag (τ L ) on peak error can be estimated by replacing the open loop error with the exponential response of the disturbance during the loop dead time For E i (integrated error), use closed loop time constant instead of dead time E = (1 e ) θo /τ L L E o For a load disturbance lag much larger than the dead time, the load error in one dead time Is very small, allowing a very large implied dead time from slow tuning. In other words, tuning and control loop dynamics are not important in terms of disturbance rejection. The focus is then on the effect of tuning and dynamics on rise time (time to reach a new setpoint)

98 Setpoint Response Rise Time Rise time (time to reach a new setpoint) is inversely proportional to controller gain T r = K i min ( % SP % CO,( K max c + K ff ) % SP ) + θ o Rise time can be decreased by setpoint feedforward and bang-bang logic that sets and holds an output change at maximum ( %CO max ) for one dead time until future PV value is projected to reach setpoint. The fastest possible rise time is: T r = K i % SP % CO max + θ o

99 Basic Lambda Tuning (Self-Regulating Processes) Self-Regulation Process Gain: % PV K o = % CO K c Controller Gain = K o Ti λ + θ ) T ( o Lambda (Closed Loop Time Constant for Setpoint Response) λ = λ f τ o Controller Integral Time i = τ o Lambda tuning excels at coordinating loops for blending, fixing lower loop dynamics for model predictive control, and reducing loop interaction and resonance

100 Fastest Lambda Tuning (Self-Regulating Process) For max load rejection set lambda equal to dead time λ = θ o K c = 0. 5 K o τ o θ o Ti = τ o

101 Basic Lambda Tuning Integrating Processes Lambda (closed loop arrest time in load response) λ = λ f / K i Integrating Process Gain: % PV2 / t2 % PV1 / t1 K i = % CO K Controller Gain: c = K i i [ λ + θo 2 ] Controller Integral (Reset) Time: Ti T = 2 λ + θ Controller Derivative (Rate) Time: T = τ secondary lag d s o

102 Fastest Lambda Tuning Integrating Processes For max load rejection set lambda equal to dead time λ = θ o Controller Gain: K c = K i 3 4 θ o Controller Integral (Reset) Time: Ti = 3 θ Controller Derivative (Rate) Time: T = τ secondary lag d s Check for prevention of slow rolling oscillations: K c o 2.25 * Ti = K i

103 Often Violated Criteria for Integrating Processes To prevent slow rolling oscillations: K c * T > i 2 K i 103

104 Near Integrator Approximation (Short Cut Method) For Near Integrating gain approximation use maximum ramp rate divided by change in controller output Ko Ki = = Max [( % PV / t) / % CO] τ p Compute maximum ramp rate as maximum delta between input (new %PV) and output (old %PV) of dead time block divided by the block dead time and finally the change in controller output (block dead time is total loop dead time) K i = PV θo % CO % max Estimate open loop gain as the difference between current operating point and original operating point K o = % PV % CO % PV % CO o o

105 Fastest Controller Tuning (short cut method) For self-regulating processes: K c =. 4 τ o 0 Ti = 4. 0 θo Td = τ s Ko θo Near integrator (τ o >> θ o ): K c = 0.4 K i 1 θ o Dead time dominant (τ o << θ o ): K c = K o Ti = 0. 5 θ o T d = 0 For integrating processes: 1 Kc = = 4 Td = τ s 0.6 Ti θo Ki θo For runaway processes: τ ' p Kc = 0.8 K θ K c = 0.8 K o Near integrator (τ p >> θ o ): i 1 θ o o Ti = 40 θ Td = 2 τ s o 1.0 for Enhanced PID if Wireless Default Update Rate > Process Response Time! These tuning equations provide maximum disturbance rejection but will cause some overshoot of setpoint response unless a setpoint lead-lag is used

106 Top Ten Things Missing in University Courses on Process Control (10) Control valves with stick-slip and deadband (9) Measurements with repeatability errors and turndown limits (8) Volumes with variable mixing and transportation delays (7) Process input load disturbance (6) Control action (direct & reverse) & valve action (increase-open & increase-close) (5) PID algorithms using percent (4) PID structure, anti-reset windup, output limits, and dynamic reset (3) Industry standards for function blocks and communication (2) Control Talk (1) My books

107 Ultimate Period and Ultimate Gain Measurement (%) Ultimate Period T u If τ o >> θ o then T u = 4 θ If τ o << θ o then Τ u = 2 θ Set Point Ultimate Gain is Controller Gain that Caused these Nearly Equal Amplitude Oscillations (K u ) 0 Time (min)

108 Damped Oscillation - (Proportional-Derivative) Measurement (%) Quarter Amplitude Period T o Offset Set Point 110% of θ o 0 Time (min)

109 Damped Oscillation Tuning Method 1. Put the controller in auto at normal setpoint. 2. Choose largest step change in controller setpoint that is safe. Increase the reset time by a factor of 10x for test. 3. Add a PV filter to keep the controller output fluctuations from noise within the valve deadband. 4. Step the controller setpoint. If the response is non-oscillatory, increase the controller gain and step the controller setpoint in opposite direction. Repeat until you get a slight oscillation (ideally ¼ amplitude decay). Make sure the controller output is not hitting the controller output limits and is on the sensitive part of the control valve s or variable speed drive s installed flow characteristic. 5. Estimate the period of the oscillation. Reduce the controller gain until the oscillation disappears (½ current gain), set the reset time equal to ½ the period, and the rate time equal to ¼ of the reset time. If the oscillation is noisy or resembles a square wave or the controller gain is high (e.g. > 10), set the rate time to zero. The factors are ½ the ultimate period and twice the ultimate gain factors because the controller gain that triggered the ¼ amplitude oscillation is about ½ the ultimate gain and the ¼ amplitude period is larger than the ultimate period. 6. If a high controller gain is used (e.g. > 10) use AO setpoint rate of change (velocity) limits if a big kick in the controller output for setpoint changes is disruptive to operations from PD action on error (enable external reset feedback). 7. Make setpoint changes across the range of operation to make sure an operating point with a higher controller gain or larger process dead time does not cause oscillations. Monitor the loop closely over several days of operation.

110 Traditional Open Loop Tuning Method 1. Choose largest step change in controller output that is safe. 2. Add a PV filter to keep the controller output fluctuations from noise within the valve deadband. 3. Make a change in controller output in manual. 4. Note the time it take for the process variable to get out of the noise band as the loop dead time. 5. Estimate the process time constant as the time to reach 63% of the final value. 6. Estimate the process gain as final change in the process variable (%) after it reaches a steady state divided by change in the controller output (%). 7. To use reaction curve tuning, set the controller gain equal to ½ the process time constant divided by the product of the process gain and dead time. 8. If the process lag is much larger than the loop dead time, set the reset time setting equal to 4x the dead time and set the rate time setting equal to the dead time. If process lag is much smaller than the loop dead time, set the reset time to 0.5x the loop dead time and the rate time to zero. 9. If a high controller gain is used (e.g. > 10) use setpoint rate of change (velocity) limits if a big kick in the controller output for setpoint changes is disruptive to operations for PD on error (enable external reset feedback). 10. Make setpoint changes across the range of operation to make sure an operating point with a higher controller gain or larger process dead time does not cause oscillations. Monitor the loop closely over several days of operation.

111 Short Cut Ramp Rate Tuning Method 1. Choose largest step change in controller output and setpoint that is safe. If the test is to be made in auto, increase the reset time by factor of 10x for test. 2. Add a PV filter to keep the controller output fluctuations from noise within the valve deadband. Measure the initial rate of change of the process variable ( PV 1 / t). 3. Make a either a change in controller output in manual or change in set point in auto 4. Note the time it take for the for the process variable to get out of the noise band as the loop dead time. 5. Estimate the rate of change of the process variable ( PV 2 / t) over successive dead time intervals (at least two). Choose the largest rate of change. Subtract this from initial rate of change of the process variable and divide the result by the step change in controller output to get the integrating process gain. 6. Set the controller gain equal to the inverse of the product of integrating process gain and loop dead time multiplied by 0.4 (self-regulating), 0.6 (integrating), and 0.8 (runaway) If the inverse of the integrating gain is much larger than the loop dead time, set the reset time setting equal to 4x the process dead time and set the rate time setting equal to the process dead time, otherwise set the reset time to 0.5x the process dead time and the rate time to zero 7. If a high controller gain is used (e.g. > 10) use setpoint rate of change (velocity) limits if a big kick in the controller output for setpoint changes is disruptive to operations from PD action on error (enable external reset feedback). 8. Make setpoint changes across the range of operation to make sure an operating point with a higher controller gain or larger process dead time does not cause oscillations. Monitor the loop closely over several days of operation.

112 On-Demand Tuning Algorithm Set Point Signal (%) Ultimate Gain 4 d K u = π e Ultimate Period T u n a d 0 Time (min) e = sq rt (a 2 - n 2 ) If n = 0, then e = a alternative to n is a filter to smooth PV

113 Adaptive Tuning Algorithm Changing Process Input Multiple Model Interpolation with Re-centering K Pure Gain Process Initial Model Gain = G1 Estimated Gain For each iteration, the squared error is computed for every model I each scan Ei( t) = ( y() t Yi() t ) Where: yt ( ) is the process output at the time t is i-th model output Yi( t) A norm is assigned to each parameter value k = 1,2,.,m in models l = 1,2,,n. N kl Ep ( t) = γ klei( t) = 1 γ kl i= 1 kl if parameter value p is used in the model, otherwise is 0 2 G2-Δ G2 G2+Δ G3-Δ G3 G3+Δ Multiple iterations per adaptation cycle For an adaptation cycle of M scans M kl kl sumep = Ep () t f k1 t= 1 F = kl sumf k 1 F kl = sumep kl The interpolated parameter value is p a p f p f p f k 1 ( ) = k 1... kl... kn k + + kl + + kn

114 Broadley-James Corporation Bioreactor Setup Hyclone 100 liter Single Use Bioreactor (SUB) Rosemount WirelessHART gateway and transmitters for measurement and control of ph and temperature. (pressure monitored) BioNet lab optimized control system based on DeltaV

115 Bioreactor Adaptive Control Performance

116 Bioreactor Adaptive Tuning Setup

117 Bioreactor Adaptive Model Viewing

118 Bioreactor Adaptive Learning Setup

119 Bioreactor Adaptive Tuning Gain 40 Reset o C overshoot Output comes off high limit at 36.8 o C

120 Bioreactor Adaptive Tuning Gain 40 Reset 5, o C overshoot Output comes off high limit at 35.9 o C

121 Bioreactor Adaptive Tuning Gain 40 Reset 10, o C overshoot Output comes off high limit at 36.1 o C

122 Bioreactor Adaptive Tuning Gain 40 Reset 15, o C overshoot Output comes off high limit at 36.4 o C

123 Bioreactor Adaptive Tuning Gain 80 Reset 15, o C overshoot Output comes off high limit at 36.1 o C