It is hard to imagine that in this day and age we are relying on "after the fact" process performance analysis on a gross scale. I wonder if we still use spreadsheets and costs sheets to judge a process unit and operator performance. We can see differences in setpoints at a shift change but have little idea as which shift's operating points provide the most production and least cost of goods (COGs). If pet food and ethanol manufacturers have online process metrics to compare shift and plant performance, why don't we see more of it in chemical plants? As COGs, energy conservation, and environmental emissions become of paramount importance, the benefits of online process metrics should be obvious. Since "seeing is believing" the use of wireless measurements moved from unit operation to unit operation can help develop and justify installations.

For example, if the utility flow and the inlet and outlet temperatures or pressures to coils, heat exchanger, and jacket are measured for each reactor, evaporator, and crystallizer, the energy use rate can be put online as an indicator of reaction, evaporation, or crystallization rate. The energy use per unit feed used or product produced can be monitored online for energy conservation. The overall heat transfer coefficient can be computed online to provide a performance indicator of fouling.

Inlet and outlet pressure measurements provide an indication of fouling (a heat exchanger specialist in my days in Monsanto's Engineering Technology said the pressure drop through the tubes of a heat exchanger is the best indication of fouling). Pressure measurements also show when a key unit operation is being starved for cooling supply especially since cooling tower and refrigeration systems often lagged beyond production increases in the 1990s. For example, a sold out fed-batch reaction at the top of a structure was periodically starved for coolant water by less profitable batch operations. Override control cut back the feeds to prevent a temperature trip but wireless measurements could have led to a more profitable scheduling of unit operations and improvement in coolant system design. Piping system and pump problems also extend to the process side.

An example that had all of the above problems big time is the stressed out production unit in Figure 3-6 in Advanced Control Unleashed The multiple lines of reactors, evaporators, and crystallizers converging and diverging were going up and down like Yo-Yos due to frosting (formation of crystals on surfaces) and subsequent defrosting (heating to melt the crystals). The plant was operating at double its original design capacity. Controller outputs would go to their high limit due to pipelines and pumps that were too small. Bigger control valves did not help much because the available pressure drop was insufficient. De-bottlenecking projects actually reduced capacity due to lack of measurements and the process knowledge they provide.

Thermowells on columns can find the most sensitive tray location by making a simple manual change in the flow manipulated by the temperature control (distillate flow). The best tray is the one with the largest and most symmetrical change in temperature for changes in the distillate to feed ratio (greatest and most linear process sensitivity or gain) as shown in the Figure 7-6 of the ISA book Advanced Temperature Measurement and Control Figure 7-7 shows the consequence of selecting the best tray and not so good tray for closed loop temperature control (Advanced-Temperature-Distillation-Column-Control-Excerpt.pdf). Three dimensional (3-D) profiles can provide a dramatic visualization of the change in the process gain with time as feed composition changes.

3-D plots generated of cross direction and machine direction sheet thickness profiles from a meter traversing the sheet provided the knowledge for optimizing the mixer, extruder, and sheet line as studied in the ISA 2001 conference paper "Constrained Multivariable Control of Plastic Sheets"

3-D plots are a natural for batch operations to show composition (e.g. cell density), pH, and temperature batch profiles for a series of batches.

For more information do a search for "metrics" on this website and you will find a variety of opportunities discussed for online process metrics and plots to help operations, process technology, and process control improvement initiatives. Wireless measurements and metrics can fill in the missing information for process simulations and vice versa.

In the old days we installed thermowells, pressure connections, and orifice flanges so we could move temperature and pressure gauges around to provide coarse local indication of equipment and piping system performance. Today portable wireless measurements can provide accurate indication in the control room offering opportunities that could be subject of a book or two let alone the blogs on this website.

Since efficiencies and most of the process gains for temperature and pressure are function of flow ratios (reactant/product, fuel/product, fuel/steam, steam/product, reagent/product, and coolant/product) flow is a key measurement. While a favor coriolis mass meters for process stream since they provide an accurate mass flow and density useful for inferential measurements of composition, which ultimately what you want to know, averaging annubars and thermowells can enable the use of wireless DP and temperature measurements for computing mass flow. Just knowing the material balance would be a big step forward as discussed in the Jan-Feb 2010 InTech article "Advances in Flow and Level Measurements Enhance Process Knowledge, Control"

Wireless pH and conductivity offers the ability to develop inferential measurements and prove the best electrode technology as revealed in the Jan-Feb 2010 InTech WEB Exclusive article "Opportunities for Smart Wireless pH, Conductivity Measurements"

You can click on the above to view and hear the recording of the Deminar.

Would you like to find tuning settings and develop a real time simulator for the more important loops in your plant in less than 10% of the time normally required? If this is of interest, check out Deminar #6. The test, triggered by a setpoint or output change, only needs to last about 3 deadtimes. Since the process time constant for the composition, pH, pressure, and temperature response of vessels and columns is 6 to 100 times the observed deadtime and the time to steady state is 4 time constants plus the deadtime, the time savings varies from 90% to 98%. The reduction in test time also minimizes the possibility of the test being disrupted by a disturbance. One of the problems we have with testing large columns to identify the dynamics for tuning or model predictive control is that the time to steady state is a day or more. Day to night temperature changes, feed changes, and shift changes usually disrupt the test of these columns. With the near-integrator approach the test time is a matter of hours and if there is a disruption, the test can be readily repeated. Also, the upset to the process from the test is significantly less because the excursion during the shorter test is much smaller.

The near-integrator gain parameter used to dramatically shorten the test time leads to a simpler expression for the controller gain that is just a function of the near-integrator gain and the observed deadtime. All of the tuning methods reduce to this same expression for maximum disturbance rejection as shown in "Appendix C - The Unification of Controller Tuning Relationships" in the ISA bookNew Directions in Bioprocess Modeling and Control. The controller gains differ by a factor that varies from about 0.5 for a Lambda tuning with a closed loop time constant equal to the process deadtime to 1.0 for the Ziegler Nichols Reaction Curve (ZNRC) method (not to be confused with the widely remembered and unpopular Ziegler Nichols ultimate oscillation method). Note that the ZNRC method requires an open loop test (change in manual output of the controller) and waits for the process to reach steady state to construct a tangent to the inflection point and find its intersection with the final value. The near-integrator method finds the maximum ramp rate for a step change in the controller output regardless of PID mode (e.g. triggered by a setpoint change or a remote output change for batch control).

What about the secondary time constants? If these time constants are much less than the primary process time constant, these secondary time constants result in an increase in the observed deadtime. Keying on a multiple of the observed deadtime self-compensates for this situation. For non-interacting secondary process time constants that approach the primary time constant (an interesting but relatively rare case), the search for the maximum ramp rate would need to be extended for several more deadtime intervals. The search can stop if the ramp rate is not increasing. For equal interacting time constants, the secondary process time constant is about 1/6 of the primary time constant. This methodology can be readily automated to identify the dynamics whenever there is a step change in controller output significantly larger than the final control element (e.g. valve) resolution limit.

For a simple real time process simulation that uses standard function blocks, the controller output and process variables from a scan or snapshot of the actual process for a representative relatively quiet operating point can be used to create deviation variables and provide a correction of the model.

The Deminar focuses on self-regulating processes that look like integrating processes because the process response ramps in the control region. The appearance can be caused by a time to steady state that is beyond the practical time range for observation or by a steady state that is beyond the operating limits of the equipment. For example, an increase in vessel pressure can force more flow out the vent valve but the vessel pressure required for the vent flow to balance the incoming or generated gas flows can be beyond the pressure relief valve setting. The time constant or ramp rate for gas pressure is generally order(s) of magnitude faster than for liquid temperature but the pressure loop deadtime is even faster. For example, the deadtime and time constant for a column pressure response might be 5 and 100 seconds, respectively whereas the deadtime and time constant for column temperature might be 5000 and 30,000 sec, respectively.

The near-integrator method can also be applied to true integrating processes which means level loops and composition, pH, pressure, and temperature loops in batch besides continuous processes can be rapidly tuned and simulated. Loops not suitable for this method are liquid pressure and flow loops and inline (pipeline) blending, pH, and temperature loops because the observed deadtime is comparable or even larger than the process time constant. However, the time to steady state for these loops is a matter of 2 to 20 seconds so that the test time is already fast and conventional methods can be employed.

You can click on the above to view and hear the recording of the Deminar.

In Deminar #5 we first show that for a self-regulating process, the process variable will line out (reach a steady state) when the controller is in manual unless there are continual disturbances. The self-regulating response is most commonly encountered response because there are more flow loops than any other type of loop. Liquid pressure loops and temperature control loops in continuous operations have a self-regulating response. Level normally has an integrating response but in the Deminar we show test results for a conical tank level with self-regulating response due to gravity discharge flow. The flow across the discharge valve is proportional to the square root of the liquid head as the level increases, the discharge flow increases and vice versa. The self-regulating or steady state process gain increases with level as a result. The significant increase in cross sectional area with level due to the conical shape causes a dramatic increase in the process time constant that creates a stabilizing effect. The process response at high level is much slower enabling the use of more aggressive tuning settings. However, the test results show these settings at low level cause excessive oscillation. The adaptive level controller is able to keep the set point response smooth and consistent over the level range. For more details you can check out the Control magazine article "Adaptive Level Control"

Most of the Deminar focuses on how an auto tuner, adaptive tuner, and adaptive controller can be used to improve the response of liquid flow and liquid pressure loops. The principle nonlinearities are the control valve characteristic for the flow loop and pump curve for the pressure loop.

You can click on the above to view and hear the recording of the Deminar. In Deminar #3 we explored the confusing situations that can develop for slow control valves and slow secondary loops. The loop can look fine but suddenly burst into oscillations and later go back to smooth sailing. The normal thought is "what changed in the process or the loop?" Well it turns out that nothing changed except the size of the upset or setpoint change. For large errors, the primary controller output starts changing faster than secondary loop or valve can respond. You could slow down the primary loop, but we know this correspondingly reduces control system performance as clearly quantified on slide 1 of EffectsLoopTuning&Dynamics-KPI.pdf. The best solutions of course are to make the valve faster and make the secondary measurement and tuning faster, but the "Quick Fix" that also offers long term protection is to enable the "Dynamic Reset Limit" option in the PID. Even if there is not a problem now, just simply turning on this option to protect against unforeseen deterioration in measurements or valves. For example, someone might try to make a secondary flow or pressure loop look smoother by the overzealous addition of a signal filter or transmitter damping setting in the middle of the night. Even more dramatically the time lag of an electrode in a secondary static mixer pH loop might go from 3 seconds to 300 seconds due to coatings or high temperatures. The Control magazine article "The Power of External-Reset Feedback" offers an excellent explanation of power of the "Dynamic Reset Limit" option which uses the PV of secondary loop as the external reset signal in the positive feedback implementation of the integral mode.

In the future, the Deminars will take only 45 minutes to reduce time commitments and audio and video recording file size. The current range of 8 to 10 demos per Deminar will be reduced to 4 to 5 demos per Deminar. Also, a "Quick Fix" will be discussed near the beginning for viewers who are short on time. The start time in June will be moved up from 1:00 pm to 10:00 am CDT to encourage European participation. I have added subtitles noting the process control improvement (PCI) in parentheses to the Deminars. Also, note that I have changed the topics for Deminars 6 through 8.

In particular, checkout the next Deminar on Thursday May 27 that shows how you can dynamically explore your own case histories and scenarios by you using the free online process control labs.

To attend the event, go to http://bit.ly/JC-LiveMeeting
Use the information below to connect (if you're not using the available computer audio):
• Toll-free: +1 (877) 771-7176
• Toll: +1 (225) 383-1099
• Participant code: 264679

(1) PID Control of Sampled Measurements (How to Eliminate Oscillations from Analyzers and Wireless Measurements with a PID Enhancement) - April 7, Wed 1:00 pm CDT

(2) PID Control of Valve Sticktion and Backlash (How to Eliminate Continual Oscillations with the "Integral Deadband" PID option) - April 21, Wed 1:00 pm CDT

(3) PID Control of Slow Valves and Secondary Loops (How to Eliminate Bursts of Oscillations with the "Dynamic Reset Limit" PID option) - May 12, Wed 1:00 pm CDT

(4) Web Lab Access and Use Instructions (How to Use Free Online Process Control Labs for Fun and Profit and Become Famous by Friday or at Least Saturday) - May 27, Thurs* 1:00 pm CDT (* - Thursday date is to avoid conflict with the World Batch Forum)

(5) PID Tuning for Self-Regulating Processes (How to Compensate for Nonlinearities in Flow and Liquid Pressure Loops) - June 9, Wed 10:00 am CDT

(6) PID Tuning for Near-Integrating Processes (How to Reduce the Tuning Time for Column and Vessel Temperature and Pressure Loops by 90%) - June 23, 10:00 am CDT

(7) PID Control of True Integrating Processes (How to Reduce the Batch Cycle Time for Temperature and pH Loops by 25%) - July 14, 10:00 am CDT

(8) PID Control of Runaway Processes (How to Improve the Performance of Exothermic Reactor Temperature Loops) - July 21, Wed 10:00 am CDT

You can click on the above to view and hear the recording of the Deminar. The second Deminar answers two questions. The first question "Why? (Why do I write so much stuff and why I am I doing these Deminars and setting up free worldwide access to generic loop and unit operation labs?) is answered on slide 4. The virtual plant used in these Deminars that creates a non DCS specific control room type experience is the most exciting thing I have done in years. This is either a commentary on my sedate existence or is an indication of the possibilities for an interactive opportunity assessment that could provide the knowledge and justification for process control improvements.

The answer to the second question that is actually a list of questions on slide 8 about the source of oscillations that cannot be tuned out is, as you might expect, the subject of the Deminar.

I think there are 8 main concepts not widely known that one can take away from this Deminar to provide guidance for a wide variety of applications.

(1) Valve stick-slip will create a limit cycle in any control loop where there are one or more integrators. The integrators can be via the integral action in the PID controller(s) or in the process (an integrating process type such as level and batch temperature). Some of the implications are as follows:

a. For a self-regulating process, integral action in any PID controller in the control loop will cause a limit cycle from stick-slip. In order to eliminate the limit cycle all PID controllers must have their integral action turned off either by a I-deadband setting bigger than the limit cycle amplitude or by using a structure with no integral action (e.g. "P on error, D on PV, no I").

b. For an integrating process, the limit cycle from stick-slip cannot be eliminated even if the integral action is turned off in all PID controllers.

(2) The limit cycle amplitude from valve stick-slip is set by the process gain and hence cannot be altered by changing the controller gain. For nonlinear processes and nonlinear valve characteristics, the amplitude changes with operating point.

(3) The limit cycle period from valve stick-slip is proportional to integral time. Slowing down the reset time will make the period larger. Thus to increase the filtering effect of process time constants in the primary loop or downstream processes, a tuning strategy would be to decrease reset time and if peak error for load disturbances is not important to decrease the controller gain to allow a further decrease in reset time.

(4) Valve deadband will create a limit cycle in any control loop where there are two or more integrators. The integrators can be via the integral action in the PID controller(s) or in the process (an integrating process type such as level and batch temperature). Some of the implications are as follows:

a. For a self-regulating process, a single loop with integral action will not develop a limit cycle from valve deadband. A cascade loop with integral action in both controllers will develop a limit cycle from deadband.

b. For an integrating process, the limit cycle from valve deadband can be eliminated if integral action is turned off as seen in slide 1 in: NonSelfRegulatingProcessDeadbandLimitCycle.pdf

c. For a runaway process (exothermic reaction) I expect the behavior to be similar to an integrating process but to a greater extreme (larger amplitude for limit cycle and larger offset for no integral action in PID controller) as seen in slide 2 of NonSelfRegulatingProcessDeadbandLimitCycle. The lack of process self-regulating in both integrating and runaway processes causes similar problems for a non-ideal valve response.

(5) The limit cycle amplitude from valve deadband is inversely proportional to controller gain.

(6) The limit cycle period from valve deadband is proportional to the integral time and is inversely proportional to the square root of the controller gain.

(7) The limit cycle amplitude in the primary process variable or in downstream process variables is proportional to the period of the limit cycle of the secondary process. The ratio of the primary or downstream amplitude to the secondary limit cycle amplitude is determined by the filtering effect of the time constant in the primary or downstream processes. When the period is smaller than the primary or downstream process time constant, the attenuation of amplitude can be approximated by the equation in: LimitCycleAmplitudeAttenuation.pdf

(8) The offset created by the use of I-deadband or selecting a structure with no integral action is less disruptive to downstream processes because a constant load upset is readily corrected by downstream loops. Periodic disturbances are more disruptive and can be amplified if the period is close to the period of loops. An offset rather than an oscillation causes less interaction between loops. One of the ways to reduce interaction is to remove integral action and decrease the gain in the least important controller.

The PID I-deadband setting should be larger than the maximum amplitude allowing for measurement noise. Note that the valve stick-slip and deadband will vary with time and operating point. The stick-slip and deadband is generally greatest near the closed position and when process material coats or corrodes the closure element seal, seat, and stem. Any addition of I-deadband or change in PID structure should be carefully monitored. Of course, the best solution is to correct the root cause of the problem and select a control valve per the "Best Practices for Valve Performance" on slide 27 of Deminar 2.

The next Deminar on "PID Control of Slow Valves and Secondary Loops" is set for May 12 Wednesday 1:00 pm Central Daylight Time.

When I first started teaching process control to junior and senior chemical engineers at Washington University in Saint Louis after retiring from Solutia, the students were less than receptive to my introduction of stuff they actually needed to know on the job. Except for the couple of students who were summer interns at Anheuser-Busch, my attempts of adding relevance were viewed as just being disruptive to the traditional task of learning frequency response and state space matrices. When I introduced the virtual plant for a weekly lab of hands-on learning, the attitude shifted from annoyance to enthusiasm. The skill and interest in using new computer tools and the fact the process simulations and graphics made the experience all seem real resulted in the labs becoming the highlight of the week. Several students decided to go on to careers in process control. One former student I met at Interphex became the manager of an automation group of a major pharmaceutical company. Even if the students didn't become process control engineers, the labs helped develop skills needed in industry. The distributed control system (DCS) is the window into the process and the ability to use and get the most out of the powerful tools and industrial standards in the DCS is important to anyone working in the process industry. This excitement and feeling that I was doing something significant to help students on "Day 1" of their prospective job, led me to think what can I do for bridging the gap between the leading edge research at universities and the opportunities for process control improvement in industry? The virtual plant to me seemed to be the way for universities and industry to get on the same page. This concept is summarized in the ACC 2009 paper ACC2009-BridgingtheGap.pdf.

The next step was to make labs as a self-learning experience available over the web with the idea that an employee could spend a few hours a month at a convenient time (e.g. lunch and learn) trying out the latest in PID control capability for various process and automation system designs and objectives. These labs provide a chance to find process control improvements by setting up scenarios that are of particular interest. Since the user interface employ operator graphics, knowledge of the particular DCS is not required. The capture of the last and best scores in terms of key performance variables (KPI) should help promote recognition and competitiveness for finding the best solutions.

I think we have barely scratched the surface of the true capability of today's PID controller with all of its features (e.g. structures, integral deadband, dynamic reset limit, and nonlinear gain). This spring and summer I will focus on generic control loops. This fall I will move on to the control of unit operations such as crystallizers, evaporators, extruders, neutralizers, and reactors. We hope users will twitter their results. The potential for learning and sharing is enormous and may be a way of getting the next generation of engineers to not only benefit from past expertise but take process control to a whole new level (see January 2010 Control article "The Future is Now")

I will conduct live seminars and demos twice a month to show how to use the labs. The connection and the topics and dates for the first 4 months are:

Recorded Live Seminar and Demo Series

To attend the event, go to http://bit.ly/JC-LiveMeeting
Use the information below to connect (if you're not using the available computer audio):
• Toll-free: +1 (877) 771-7176
• Toll: +1 (225) 383-1099
• Participant code: 264679

(1) PID Control of Sampled Measurements (How to Eliminate Oscillations from Analyzers and Wireless Measurements with a PID Enhancement) - April 7, Wed 1:00 pm CDT

(2) PID Control of Valve Sticktion and Backlash (How to Eliminate Continual Oscillations with the "Integral Deadband" PID option) - April 21, Wed 1:00 pm CDT

(3) PID Control of Slow Valves and Secondary Loops (How to Eliminate Bursts of Oscillations with the "Dynamic Reset Limit" PID option) - May 12, Wed 1:00 pm CDT

(4) Web Lab Access and Use Instructions (How to Use Free Online Process Control Labs for Fun and Profit and Become Famous by Friday or at Least Saturday) - May 27, Thurs* 1:00 pm CDT (* - Thursday date is to avoid conflict with the World Batch Forum)

(5) PID Tuning for Self-Regulating Processes (How to Compensate for Nonlinearities in Flow and Liquid Pressure Loops) - June 9, Wed 10:00 am CDT

(6) PID Tuning for Near-Integrating Processes (How to Reduce the Tuning Time for Column and Vessel Temperature and Pressure Loops by 90%) - June 23, 10:00 am CDT

(7) PID Control of True Integrating Processes (How to Reduce the Batch Cycle Time for Temperature and pH Loops by 25%) - July 14, 10:00 am CDT

(8) PID Control of Runaway Processes (How to Improve the Performance of Exothermic Reactor Temperature Loops) - July 21, Wed 10:00 am CDT

Knowledge of the flows and the accumulation of material in a unit operation are fundamental to the understanding and analysis of process and equipment performance. Flows are the primary way of affecting the process. Root cause analysis requires sensitive and repeatable flow measurements. I have seen costly expert systems fail to deliver benefits because of missing or inaccurate flows ("Drowning in Data, Starving for Information - 1").

The process gains of the more important process variables (e.g. composition, pH, and temperature) are best quantified and visualized in a plot versus a ratio of flows (e.g. coolant/feed, reactant A/reactant B, reagent/feed, reflux/feed, and steam/feed). If you are still into differential equations, you can checkout my Advanced Application Note 4 to see how process gains are dependent upon the ratios of flows.

The importance of flow ratios for affecting the process is seen in the prevalence of flow ratio control as detailed in my entries "What Have I Learned? - Flow Ratio Control" on this website.

The amount of time material spends in a unit operation is critical for crystallization and reaction. For continuous operation of well mixed volumes, the amount of time is the residence that is the fluid volume divided by the total throughput flow. Conversion is maximized by increasing volume or decreasing feed flows. For batch processes, the amount of time is the cycle time. Conversion is maximized by charging the feeds as fast as possible (increasing feed flows), to leave more of the batch cycle time for conversion.

In the direct material balance control scheme where the distillate flow is manipulated for overhead receiver level control, the sensitivity of the temperature and hence the composition control requires an exceptionally sensitive level measurement, low noise, and a high controller gain. Changes in distillate flow do not affect the column until there is a corresponding change in the reflux flow that maintains the material balance.

Then of course, there is the need to minimize the amount of storage of materials in the process. Ideally, storage tanks would be almost empty with just enough raw materials and intermediates to continually meet the flow demand of downstream operations and just enough products to continually meet the flow demand of customers.

For more information on how advances in flow and level measurements can improve material balance control, residence time control, inventory control, and process analysis and modeling, checkout "Advances in Flow and Level Measurements Enhance Process Knowledge, Control"