Emerson Exchange 365

The post, Why Batch Processes are Difficult , first appeared on ControlGlobal.com's Control Talk blog. Batch process control can seem like another world compared to continuous process control. In batch operations, process conditions are constantly changing and control loops are going in and out of service. PID control may take a back seat to sequential scheduling of manipulated flows. Here we look at why batch processes are so different and challenging and what can we do wherever possible to apply the power of PID control and Model Predictive Control (MPC). The ultimate objective here is to take advantage of the lessons learned in both worlds and enable the process control engineer to move effectively between the two worlds. Like a perpetual startup and shutdown. Equipment is continually coming into and out of service in batch operations. Target operating conditions must be reached. As soon as these are reached, the operation moves on to the next task possibly using different equipment and different measurements and valves as quickly and as repeatable as possible. Every phase can be a process onto itself. A batch operation may have pressurization, heating, cooling, conversion, and stripping phases. The control system must move seamlessly from one phase to another. Controllers may have different functions and objectives and certainly different tuning settings. Wide spectrum of product grades and formulations is common. The start of each batch offers the opportunity to make a new product or at least a different grade or formulation. This can lead to complex changes in recipes that the operator and control system must deal with. In continuous processes, transitions to a different product grade or formulation are minimized or not made because the transitions take appreciable time, require extensive process knowledge and can create off-spec product that is often difficult to recycle. Extensive sequencing and operator involvement is required. There is usually never a dull moment with a batch. Something is always changing. There is no chance to sit back and see relatively constant valve positions like when a continuous process runs at a steady state. Dynamic response is non-self-regulating (non-stationary and no conventional steady state). The batch response of concentration, pressure, and temperature is integrating or runaway. PID tuning rules and MPC strategies based on self-regulating processes do not generally work without some modification or translation. Integrating process models and tuning rules are needed. Extreme rangeability of manipulated variables is often needed. The cooling rate, vent rate, and feed rates for biological and chemical reactions and crystallization exponentially change with batch time. For dissolved oxygen (DO) control the oxygen uptake rate may increase by several orders of magnitude as the batch goes from the pre-exponential growth to near the end of the exponential growth phase. For fermenters, DO control is split ranged between air sparge flow, agitator speed, and vessel pressure. For bioreactors, DO control is split ranged between air and oxygen sparge and overlay flows. Dynamic response is nonlinear. Changes in liquid volume, heat transfer surface area and coefficient, operating conditions (e.g., concentrations, pressures, and temperatures) and in manipulated variables (e.g., split ranged variables) causes changes in integrating process gains and secondary time constants and dead times. Dynamic response may be unidirectional. For batch processes where there is only heating or only cooling and no endothermic or exothermic reactions or changes in phase, the temperature response goes only in one direction. For batch neutralizers where there is only an acid or only a base addition and no consumption of reagent in a reaction or changes in phase, the pH response goes only in one direction. For cell and product concentration in fermenters and bioreactors, the concentration only increases assuming death rate and hydrolysis is negligible. For these processes, integral action cannot be used to control temperature or pH at a specific setpoint. A PID structure of Proportional on error and Derivative on process variable (P on E, D on PV, no I) is used. Setpoint overshoot is problematic. The integrating or runaway response makes setpoint overshoot more likely. Tuning becomes difficult and counter intuitive in that a higher PID gain may be needed to prevent the lingering overshoot from integral action. For unidirectional response, there is no return to back to setpoint. For sensitive biological processes, overshoot of a few tenths of degree or a few hundredths of a pH is undesirable. Here a very slow approach to setpoint that eliminates overshoot is desirable because the increase in time to reach setpoint is very small compared to batch time. A two degree of freedom (2DOF) structure with beta and gamma set to zero (equivalent to an I on error, and PD on PV structure) and conservative tuning may be the best choice particularly since disturbances from cell growth and product formation are so slow. Window of allowable PID gains exists. Oscillations develop for a PID gain that is too small as well as too large due to the non-self-regulating response. For runaway processes, the process response can accelerate to a point of no return if the PID gain becomes less than the open loop positive feedback gain. Temperature controllers on highly exothermic batch reactors have this threat to an extreme where the controller cannot be put in manual for open loop bump tests and integral action is not permitted. Contaminants, impurities, and inhibitors are trapped in the batch. Since there is no liquid discharge flow till the batch is done, concentrations of undesirable components will build up as the batch progresses. At-line analyzer and off-line analyzer results are often too late. Analysis results are often not available until the phase or batch is completed. Variability is trapped in batch endpoint. There is no inherent attenuation from a continuous flow through a volume. You are stuck with a bad batch requiring possibly scrapping the whole batch unless you are blending a whole lot of parallel batch trains downstream. For bioreactors, the loss of a multimillion dollar multiday batch is a huge hit to the bottom line. Batch process yield, production, quality, and repeatability are interrelated. There may be a tradeoff in extending a batch time to gain yield or improve quality versus losing production rate. Also the ability to improve operating conditions depends upon batch repeatability. This is why batch data analytics first tries to identify what batches differ from the average batch and why. The ability to make better decisions and improvements is often related to batch repeatability just as with any measurement used for control. Data exclusion frequently needed for batch analysis . Since equipment and associated controls are continually going in and out of service, data and alarms must be intelligently excluded. We can learn from batch operations how to better automate the startup and shutdown of continuous processes. We can use the tuning rules for integrating processes used extensively in batch processes and the awareness of the window of allowable PID gains to tune the PIDs for continuous concentration, temperature, pH and pressure control of vessels and columns (e.g., near integrating response). We can learn about the buildup of contaminants, impurities, and inhibitors in batch processes to prevent a similar occurrence in continuous processes with extensive recycle (e.g., snow balling effect). We can use the adaptive scheduling of tuning settings needed for batch operations in continuous operations to deal with startup, shutdown, split ranged operation, and the catalysis degradation and heat transfer surface fouling with time. We can translate the controlled variable from batch concentration, pH or temperature to a rate of change of concentration, pH or temperature for control of the desired batch profile to provide a pseudo steady state and a bi-directional response. This enables the use of integral action and MPC offering a smoother and tighter control of the batch profile. We can use some of the strategies developed for continuous reactor control for fed-batch reactor control including the use of valve position control to maximize reactant feed rate as detailed in the ISA 2015 book Advances in Reactor Measurement and Control . It also facilitates a greater improvement by the use of an enhanced PID for the large and variable update times of at-line and off-line analyzers per the 7/06/2015 Control Talk Blog. We can also use inferential measurement techniques developed for continuous processes to provide concentration measurements between corrections by analysis results for closed loop control of batch concentration. We can head off bad batches and better develop the Projection to Latent Structure (PLS) prediction of batch end point by data analytics software to make mid batch corrections. For a summary of the challenges and opportunities in batch process control see the Chemicals & Petrochemicals Plant Automation Congress 2015 presentation “ Batch Process Control Strategy ”.

The post How to Make Transitions Between Cooling and Heating Smooth and Fast first appeared on the ISA Interchange blog site. The following insights are part of an occasional series authored by Greg McMillan , industry consultant, author of numerous process control books and a retired Senior Fellow from Monsanto. Editor’s Note : This is Part 3 of a four-part blog series on smart automation of vessel heating and cooling. Click these links to read Part 1 or Part 2 . In this post I look at how we can eliminate oscillations associated with the discontinuity and changes in phase between heating and cooling and make the response of the secondary temperature loop faster. Most temperature loops tend to develop an oscillation across the split range point during the transition between heating and cooling. The transition creates a discontinuity resulting from the change in utility fluid and the increase in friction and loss of sensitivity in valve seating or sealing. The use of a vessel PID with external reset feedback and setpoint rate limits on the manipulated valves or flows for cooling and heating can provide directional move suppression to help eliminate unnecessary crossings of the split range point. The setpoint rate limit in the direction of approaching split range point when close to the split range point is set slower. The result is directional move suppression. The vessel PID does not try to change the setpoint faster than the flow will respond by external reset feedback of the Analog output or flow loop process variable (PV). The external feedback of actual valve position or flow can also eliminate the oscillations from deadband and in some cases stick-slip. Given that crossings of the split range must occur in the transition from heating to cooling, we need to recognize and address the challenges to enable tight vessel temperature control. The worst case is often the transition between steam and water coolant due to the huge difference in temperatures and the creation of steam bubbles in water coolant or water droplets in steam. Steam valves tend to have higher seating and sealing friction due to the higher temperatures and pressures. Hot and cold liquids and tempered water systems offer a much smoother transition than between steam and water coolant and reduce nonlinearities for coil and jacket temperature control. The switch from heating to cooling creates oscillations and an irregular response due to the discontinuities including stick-slip at the split range point and changes in phase, such as bubbles in the transition to cooling and condensate drops in the transition to heating. A lead may precede the lag in the response of the jacket inlet temperature. Steam injection and tempered water systems can provide tighter jacket temperature control and the elimination of cold and hot spots. Tuning and special PID options can be used to minimize unnecessary crossings of the split range point and deal with lead-lag response. See Greg McMillan’s new ISA book, Advances in Reactor Measurement and Control, for an extensive view of practical opportunities for designing control strategies to achieve product quality and maximize yield and capacity in different types of fermenters, bioreactors, and chemical reactors. Steam does not provide uniform heating in coils or a jacket. Steam typically collects in the top of the jacket and condensate collects in the bottom. Hot spots can develop around inlets. Thermal shock and steam hammer can damage glass lined vessels. The time required to drive steam completely out of the jacket before cooling water is introduced introduces a considerably delay in the control system. Improper trap design or operation can cause condensate buildup. The addition of hot water instead of steam directly into the coil or jacket provides a more uniform heat distribution, a dramatically smoother transition between heating and cooling, and a more efficient and maintainable system. For rapid heating, the use of direct steam injection heaters and pressurized water as shown in Figure 1 can provide hot water temperatures well above 100 o C. If the injection heater has hundreds of small orifices, the bubbles are extremely small and are rapidly and quietly mixed into the water. Variable orifice steam injection heaters are not as quiet and the mixing is not as complete. The use of jacket outlet temperature reduces the possibility of bubbles hitting the temperature sensor. Insight: The use of hot water instead of steam for heating eliminates the discontinuities and noise in the transition between heating and cooling. A steam injector can be used to provide a smooth and rapid transition from cold water to varying degrees of hot water. Figure 1 – Steam injection heaters to create hotter water offer rapid heating and tight control of coil or jacket temperature with smoother split ranged transitions. Heat exchangers in the coil or jacket recirculation system are used to provide a tempered water system. This allows the use of colder or hotter utility streams as inputs. As the inputs are to heat exchangers rather than being direct inputs to coil or jacket, temperature extremes are moderated that would cause heat transfer surface coating or product degradation from localized cold and hot spots thermal shock that could crack a glass lining. Mammalian bioreactors are particularly vulnerable to temperature extremes because of metabolic sensitivity. There is also a greater propensity for localized temperature variations due to less mixing from the reduction in agitation to avoid cell rupture. Mammalian cells, unlike bacterial cells, have membranes rather than cell walls, making them more susceptible to damage. Insight: Tempered water systems can eliminate cold and hot spots in heat transfer surfaces. Figure 2 – Heat exchangers with split ranged steam and chilled water have less self-regulation, more nonlinearity, and a slower response than direct steam injection and blending of hot and cold water. Separate exchangers are used for steam and chilled water. The coil or jacket temperature response to a change in controller output may exhibit a temporary initial change (lead) followed by a ramp (integrating response) before approaching a final steady state value (self-regulating response). The integrating response originates from the recycle of jacket water. For example, an increase in heat exchanger outlet temperature makes a loop through the coil or jackets and comes back to the heat exchanger as an increase in inlet temperature. The ramp rate (integrating process gain) increases as the coil and jacket volume decreases. A lead in the opposite direction of the response to a change in steam flow can be caused by a temporary change in the condensing rate of steam. This lead causes the temperature to temporarily increase when the steam flow is decreased. There is a thermal lag from the UA of the heat exchanger and an increase in process gain and dead time at low utility flow in the self-regulating. Coil and jacket temperature control by the manipulation of a cooling or heating utility stream to a heat exchanger has a slower and more nonlinear and irregular response than the direct steam injection and blending of hot and cold water in a constant coil or jacket circulation flow. Insight: The use of heat exchangers in a utility recirculation system can introduce a temporary initial change (lead) followed by a ramp (integrating response) before getting into the heat transfer time constant (lag) of the exchanger volume and heat transfer area. The coil or jacket exchanger thermal lag can be passed by the manipulation of an exchanger bypass flow creating a faster temperature loop that is easier to tune. A valve position controller (VPC) can be used to reduce cooling or heating liquid utility flow during low loads and to increase temperature loop turndown. Steam heat exchanger bypass control is not used because of steam blowing into the condensate system at low heating requirements. Insight: The throttling of a heat exchanger bypass flow instead of the cooling and heating flow makes the secondary jacket temperature control faster by eliminating the exchanger thermal lag. Use a PID with external reset feedback and directional moves suppression to help eliminate unnecessary crossings of the split range point and oscillations from the increased stick-slip near the closed position. Avoid if possible the switch between steam and coolant to the jacket by the use of steam injector or a tempered water system. When heat exchangers are used, the manipulation of a bypass flow instead of a utility flow provides a faster jacket temperature response. A valve position controller can be used to help ensure a minimum utility flow at low loads.

The post, Batch and Continuous Control with At-Line and Offline Analyzers Tips , first appeared on the ControlGlobal.com Control Talk blog . What if you could use at-line analyzers and even off-line analyzers for control of batch and continuous processes without the need to retune the PID despite long and variable cycle times? What if the lab analyzers could be used for closed loop control without having to be concerned about the PID becoming oscillatory or going out to lunch as the cycle time increases or when analyzer results are not available? Here is a simple solution and some guidance on how to get the most out of this opportunity. The dead time from analyzers can be confusing and disruptive to say the least. If simulation tests have the disturbance arriving just as the analyzer gives an update, which tends to be the case in the module setup for testing, the full effect of the dead time is not seen and some may not even believe the additional dead time exists. In reality, the disturbance can arrive anytime in an analyzer cycle time. On the average it arrives in the middle of the cycle time, creating a dead time that is ½ the cycle time. The corresponding phase shift has been confirmed by frequency response analysis. You also need to add the time to get an analysis result once a sample arrives at the analyzer inlet. This latency is additional dead time. For chromatographs where the analysis result is available at the end of the analyzer cycle time, the result is a dead time of 1.5 times the analyzer cycle time. For a chromatograph with a 20 minute cycle time, the additional loop dead time from the analyzer is 30 minutes. Sample processing and transport and multiplex times create additional dead time as quantified by equations in my comprehensive chapter on the effect of measurement dynamics in Tuning and Control Loop Performance – 4th Edition Momentum Press, 2015. Now just imagine what the dead time could be from lab samples. The dead time is huge (e.g., hours to days). If there is not an enforced schedule of taking samples, doing the analysis, and entering the results, the dead time is extremely variable. This has typically excluded the direct use of off-line analysis for closed loop control. The creation of neural network, step response, and first principle models that is updated by lab results can be a viable solution here. However, if the model is not very good resulting in large and variable corrections, we are back to an instability problem caused by excessive dead time from an analyzer. A simple solution enables the use of a PID where the process variable comes directly from an at-line or for even off-line analyzer that has a large and variable time between analysis results. The PID must have the positive feedback implementation of integral action where the integral contribution is via a filter whose input is the PID output or external reset feedback. The filter output (integral contribution) is added to the proportional mode contribution (hence positive feedback). If the contribution of the filter is computed as a first order exponential response to the change in PID output or external reset feedback when there is an analysis update, we have an enhanced PID that was originally developed for wireless applications that I think has even a more significant future for analyzer applications. The derivative mode contribution is also computed based on the elapsed time between updates, but for at-line and offline analyzers, the updates are too discontinuous to enable the use of much if any derivative action. See the InTech July/August 2010 feature article “ Wireless: Overcoming challenges in PID control & analyzer applications ” for an overview. Tests have shown that if the dead time from the analyzer is greater than the 63% process response time (i.e. process dead time plus process time constant), the tuning of enhanced PID does not change as the time between analyzer results change. The tuning in fact simplifies to be a maximum PID gain that is the inverse of the dimensionless open loop gain (e.g., product of valve or VFD gain, ratio gain, process gain and measurement gain). The measurement gain is a simple function of the measurement span (measurement gain = 100%/span). If the valve or VFD gain and process gain is known and used to update the maximum PID gain, the actual PID gain can be set relatively close to the inverse. A PID gain as much as 50% larger than this maximum will not be excessively oscillatory for the enhanced PID. If the PID reset time is set to equal the analysis dead time, the PID can make a single correction upon an update that will bring the controlled variable back to setpoint within the repeatability of the analyzer. This capability only holds true for processes with a steady state (self-regulating process response). While an improvement in control exists for a batch process (integrating process response) by the use of an enhanced PID, the results are not as dramatic and excessive oscillations can develop for a large analysis time. So what can we do to extend these remarkable benefits we see for continuous processes to batch concentration control? If you translate the controlled variable from concentration to rate of change of concentration that is the slope of the batch concentration profile, you have created a pseudo steady state and self-regulating process. This also has the benefit of creating a bidirectional response (the slope can decrease as well as increase) whereas many batch processes have a unidirectional response (concentration can only increase) that excludes the use of integral action in the PID structure. Thus, the use of batch profile slope as the controlled variable provides the benefits seen in the application of the enhanced PID for continuous processes. The only major requirement is that the slope be computed and updated only when an analysis result is available. The slope is simply the change in analysis results divided by the time interval between the analysis results. While I don’t suggest you be so aggressive as to set the PID gain to be the inverse of the open loop gain, you will not need to retune the PID for extremely large and variable times between analysis results as long as you have a reasonable knowledge of the open loop gain. This is pretty exciting to me because ultimately we want to control concentrations in the process. We almost always have a lab analysis. At-line analyzers may be viewed as too expensive and difficult to support and online analyzers may not specifically and accurately measure the component of interest. Of course more frequent analysis results is desirable to detect and correct for disturbances sooner, but at least we can take advantage of analysis results as they become available giving much more repeatable and accurate compensation than what you would gain from manual correction by operators. Of course analysis results would be screened and if strange values are detected, the analysis that is sent to the PID would not be updated. The enhanced PID has no problem waiting longer for a valid result. My August 2010 white paper PID-Enhancements-for-Wireless gives test results for various applications focusing on wireless opportunities. I understood the potential advantage for analyzers when I wrote the paper but the excitement was all about wireless at the time. While wireless has a great future in terms of reducing installation costs and enabling portable diagnostics, I think the greater opportunity in terms of immediate process performance improvement is the use at-line and offline analyzers for batch profile and continuous process concentration control by checking a PID options box that turns on an enhanced PID and simply setting the PID gain to be less than the inverse of the open loop gain.

The post How to Optimize Cascade Control Systems in Vessel Heating and Cooling first appeared on the ISA Interchange blog site. The following insights are part of an occasional series authored by Greg McMillan , industry consultant, author of numerous process control books and a retired Senior Fellow from Monsanto. Editor’s Note : This is Part 2 of a four-part blog series on smart automation of vessel heating and cooling. Click here to read Part 1 , on critical aspects of jacket and coil design in vessel heating and cooling. In this post, I look at how to make sure the cascade control system is intelligently designed to meet process requirements in terms of vessel and heat transfer surface temperatures. Whether inlet temperature or outlet temperature is used depends upon the source of upsets and constraints on heat transfer surface temperature. In either case the jacket or coil temperature control loop must be fast. Cascade temperature control is the most prevalent strategy applied where the primary vessel temperature PID provides a setpoint to a secondary jacket temperature PID for the throttling of hot and cold fluids (Figures 1 and 2). The use of cascade control offers considerable performance improvements in terms of reducing the peak and integrated absolute error (IAE) from disturbances. The jacket or coil temperature controller can correct for disturbances to the jacket before they affect the vessel temperature. The jacket or coil temperature loop also isolates process and valve gain nonlinearity from the vessel loop. For a negligible increase in the heat of reaction compared to the heat transfer capability, the process gain for the vessel temperature loop approaches unity. Insight: Jacket or coil temperature loops prevent the vessel temperature loop from seeing utility system and valve nonlinearities and disturbances. Figure 1: Control of Jacket Inlet Temperature Figure 2: Control of Jacket Outlet Temperature A jacket or coil temperature loop can isolate the effects of process and valve gain nonlinearity from the vessel temperature loop. Jacket or coil inlet temperature control provides quicker correction of utility disturbances and better prevention of cold and hot spots. Outlet temperature control offers a smoother response by attenuation of mixing disturbances and phase discontinuities. A change in outlet jacket or coil temperature for the same production rate or batch cycle time and same vessel temperature can provide a warning of an increase in heat transfer surface coating or fouling. The jacket or coil temperature control must be fast and responsive to the demands of the vessel temperature controller. See Greg McMillan’s new ISA book Advances in Reactor Measurement and Control for an extensive view of practical opportunities for designing control strategies to achieve product quality and maximize yield and capacity in different types of fermenters, bioreactors, and chemical reactors. For a faster and more uniform measurement the temperature sensor should be in the pipeline about 20 pipe diameters downstream of the outlet of an exchanger or jacket rather than in the equipment for a faster and more uniform measurement of the heat transfer fluid. Pipeline velocity and turbulence reduce the sensor time constant and process noise from the splashing of level, hot and cold spots, and mixed phases. The sensor tip should be near the centerline of the pipe with a sufficient immersion length to make conductive heat loss negligible. For highly exothermic fluidized reactors boiler feed water (BFW) is added under level control and the reactor temperature PID output is the jacket outlet steam pressure PID setpoint or coils are switched in and out of service. The jacket temperature control schemes are suitable for batch as well as continuous operation. Coils generally offer a faster temperature response than a jacket by a decrease in the volume of the heat transfer fluid and an increase in the velocity. Both of these work to decrease the process dead time, which is the coil volume divided by the utility flow rate. The increase in velocity increases the heat transfer coefficient but this is partially offset by the decrease in surface area. An increase in the product (UA) of the overall heat transfer coefficient (U) and surface area (A) will decrease secondary process lag in the thermal response. The transition in split range operation is faster with a coil than a jacket which is useful for a valid transition between hot and cold utility streams but can be problematic for unintended transitions from valve stick-slip and an integrating response in the process or controller. Insight: The more aggressive temperature effect and faster transition between heating and cooling by coils help deal with the highly exothermic and fast gas reactions but the switching of coils and split range transitions are more disruptive to vessel temperatures. Whether the secondary loop uses coil or jacket inlet or outlet temperature is often a matter of tradition for a particular company or process industry. The dynamic response of the cascade control system to vessel disturbances such as feed and reaction rate are the same for coil or jacket inlet and outlet temperature control. Coil or jacket inlet temperature control will correct for changes in cooling or heating utility supply temperature and pressure sooner than the transportation delay through the coil or jacket. The coil or jacket loop process dead time is also less by the amount of this delay, allowing a faster reset time setting and faster correction of valve nonlinearities. For crystallizers, cold spots can cause the formation of fine crystals that coat the coiling surface causing an excessive heat transfer lag and an upset to the population balance of desired crystals by not growing into the proper crystal size. For biological operations and sensitive products local hot spots can decrease capacity and quality. At the coil or jacket inlet the mixing of the recirculation with the hot or cold makeup flow may be incomplete and the discontinuity in the transition from hot to cold may be more abrupt. The location of the temperature sensor on the jacket outlet offers time for mixing and volume for smoothing transitions. Less measurement temperature noise can translate to a higher controller gain and less overreaction to the discontinuity at the split range transition. Insight: The use of the jacket inlet for the secondary control loop can correct for utility disturbances more quickly but is more susceptible to noise from mixing and phase changes. The difference between the vessel and the jacket outlet temperature (approach temperature) can provide an inferential measurement of the heat transfer coefficient (U) for a constant jacket circulation flow, a given production rate, and the heat transfer area (A). The approach temperature increases as UA decreases. For residence time control in liquid vessels the increase in level will increase the heat transfer area covered by reactants and product offsetting the increase in heat release with production rate and eliminating the need for production rate and level correction. For fed-batch vessels and continuous vessels without residence time control, a correction for level is needed to compute U from UA. Insight: The difference between vessel temperature and jacket outlet temperature (approach temperature) can be used to compute the heat transfer coefficient for a constant jacket flow. For the coil or jacket temperature control to provide rapid adjustments of cooling and heating for disturbances to the coil or jacket and setpoint changes from the vessel temperature control, the coil or jacket PID response needs to be fast, which may be achieved by a fast sensor and fast tuning. A tight fitting sensor bottomed in a thermowell made from a high thermal conductivity metal with a tapered tip near the pipe center line provides a fast measurement. Spring loading can ensure that the sensor sheath is bottomed. The clearance between the sheath outside diameter and the thermowell inside diameter must be minimized and the fluid velocity must be maximized. While grounded thermocouple sensors are a few seconds faster in responding than resistance temperature detector (RTD) sensors, the difference is insignificant compared to the effects of thermowell design and fluid velocity. The greater sensitivity and lower drift of an RTD is important for jacket as well as vessel temperature measurements. The lower drift reduces maintenance and the higher sensitivity provides faster recognition. The use of RTDs also facilitates more accurate online heat transfer computation for process diagnostics and inferential measurements of reaction rate. Thermocouples are preferred for temperatures above 400 o C where RTD insulation resistance and sensor integrity become problematic. A high PID gain provides a faster setpoint response. The jacket temperature controller has a self-regulating process response with a maximum PID gain that is about half of the open loop time constant divided by the product of the open loop steady state gain and dead time. The minimum reset setting is about 4 times the loop dead time for the jacket temperature PID. The elimination of offset from setpoint in the secondary jacket temperature PID is not as important as an immediate response to setpoint changes from the vessel temperature PID. Consequently, proportional action is more important than integral action and robustness to changes in dynamics is better achieved by increasing the reset time. External reset feedback in the primary vessel temperature PID prevents the manipulated jacket or coil secondary PID setpoint from changing faster than the jacket or coil temperature can respond but a faster secondary PID response dramatically improves the tightness of vessel temperature control. Insight: RTDs with a tight fit in a tapered thermowell at the pipeline centerline and a PID tuned with more proportional action than integral action provide the fastest secondary loop response. Determine whether the major source of jacket or coil temperature disturbances originates from changes in the utility temperature or pressure or from mixing and discontinuities and accordingly choose whether the jacket or coil temperature should be controlled. Make sure the temperature measurement is fast and representative of the true jacket or coil temperature. Tune the jacket or coil temperature controller for a fast setpoint response to immediately start to meet the demands of the vessel temperature controller.

The post What are the Critical Aspects of Jacket and Coil Design in Vessel Heating and Cooling? first appeared on the ISA Interchange blog site. The following insights are part of an occasional series authored by Greg McMillan , industry consultant, author of numerous process control books and a retired Senior Fellow from Monsanto. Editor’s Note : This is Part 1 of a four-part series on smart automation of vessel heating and cooling. The heating and cooling system is critical for vessels since temperature plays such a huge role in determining product quality. The limit as to what can be done to increase production rate depends upon the capability of the heating and cooling system. The system can also be the source of discontinuities and nonlinearities. Most temperature control problems beyond tuning can be traced back to deficiencies in the vessel coil or jacket or heat exchanger mechanical, process, piping and control system design. In this post, I assess the critical aspects of the design of the jacket and coil. Heating is required for endothermic reactions (reactions that consume energy), for vaporizing liquids that are being vaporized, or for bringing a vessel up to operating temperature. Cooling is needed for exothermic reactions (reactions that product heat), condensing vapors that are being condensed, and to bring a vessel down to operating temperature. Often both heating and cooling are often required. Split ranged control is used to go back and forth between heating and cooling. In split ranged control for an exothermic reactor with a fail open cooling valve, as the temperature controller output increases from 0% to the split range point, the coolant valve goes from wide open to completely closed. As the temperature controller output increases from the split range point to the 100%, the heating valve goes from closed to wide open. The split range point is traditionally set at 50% but would be better set so that the change in temperature for a change in flow is about the same for each valve (e.g., open loop process gains are the same for each valve). Cooling and heating can be applied directly to the vessel by heat transfer surfaces in the vessel exterior (e.g., jackets) or interior (e.g., coils) or indirectly via a process recirculation stream. Heat exchangers are used for some jacket temperature control systems and for all process recirculation temperature control systems (discussed in Parts 3 and 4 next month). The use of a recirculation system for coils and jackets can eliminate the increase in dead time, process gain, heat transfer lag, and fouling for low heating and cooling loads leading to oscillations and potentially unsafe operation. The use of precise throttling valves and external reset feedback can eliminate many of the oscillations normally prevalent at the split range point. See Greg McMillan’s new ISA book Advances in Reactor Measurement and Control for an extensive view of practical opportunities for designing control strategies to achieve product quality and maximize yield and capacity in different types of fermenters, bioreactors, and chemical reactors. Coils offer faster and more aggressive heating and cooling. Jackets offer a more even heat transfer distribution However, jackets have a lower utility velocity at the heat transfer surface that can increase fouling rate and require good mixing to ensure that the jacket has a uniform effect on the entire liquid contents. Coils offer excellent heat transfer to fluids in the interior but generally not in a dished bottom. At low liquid levels, the process liquid may not be in contact with the coils. For highly exothermic plug flow reactors such as gas reactors, coils provide more effective cooling from direct process contact in the core and high fluid velocities inside and outside the coils. Insight: Coils offer faster and more aggressive cooling and do not require a well-mixed process fluid but do not offer good heat transfer for low liquid levels. Many of the temperature control considerations for coils and jackets are the same. For example, both coil and jacket systems benefit from keeping the utility flow through the coil and jacket constant by the use of a recirculation pump and piping to return utility flow from the outlet back to the inlet of the coil or jacket and from eliminating split range discontinuities by control valve and control system design. Insight: Coils and jackets both benefit from a recirculating utility flow and better split range control. A constant jacket flow provided by recirculation and the throttling of a makeup flow is preferred to the throttling of total flow to the jacket. The process gain and dead time for the jacket temperature controller are inversely proportional to flow. At low jacket flows the combination of high process gain and high process dead time can cause sustained oscillations. In addition, at low jacket flows the heat transfer coefficient is greatly reduced and the fouling rate is greatly increased decreasing the capability of the vessel temperature control system to do its job. Fouling can create a large secondary time constant that can is particularly detrimental in the near-integrating and true integrating response of vessel temperature. For runaway reactors, the secondary time constant can cause dangerous unsafe operation from a narrowing of the window of allowable reactor temperature controller gains. Insight: Throttling a utility makeup flow instead of the jacket or coil flow helps to keep the jacket or coil dynamics constant and reduces fouling of the utility side of heat transfer surfaces. Mechanical design needs to prevent mixed phases in the jacket or coil (e.g., steam bubbles in coolant and coolant droplets in steam) after a transition between cooling and heating. Control valves must be precise (e.g., minimum backlash and stick-slip) and properly sized to prevent limit cycles and abrupt changes as one valve closes and another valve opens. Valves designed for tight shutoff have excessive seating and sealing friction near the closed position. Isolation valves are not throttling valves and vice versa. To prevent leakage of a utility flow, an isolation valve should be added in series with the throttling valve to open and close on an on-off basis as the throttling valve strokes open or closed. Insight: Mechanical design and control valve design should prevent mixed phases and abrupt changes in heat transfer medium, temperature or flow rate at a transition between heating and cooling. All of the control schemes can benefit from the use of external reset feedback (dynamic reset limit) in the vessel and jacket or coil PID controllers. External reset feedback can prevent the primary vessel temperature PID from trying to change the secondary temperature PID setpoint faster than the jacket or coil temperature can respond. External reset feedback with a fast utility valve position readback can help prevent limit cycles from valve backlash that are particularly problematic near the split range point. The reset time can adapted to reduce the overshoot by a providing a smarter time for the valves coming off output limits for large setpoint changes in batch operation and continuous operation transitions and startups. For highly exothermic reactors, external reset feedback offers the ability to reduce the chance of a runaway by the addition of directional rate setpoint limits to provide a faster response to a demand for cooling. Setpoint rate limits on the analog output block to valves or on the PID for a secondary flow loop provide directional move suppression to help reduce unnecessary crossings of the split range point. The setpoint rate limit is slower in the direction of moving to the split range point. External reset feedback of the actual response of the valve or secondary flow is used to prevent the jacket or coil temperature PID output from trying to change faster than the setpoint rate limits will allow. Insight: External reset feedback and directional move suppression can help eliminate most sources of oscillations from split ranged valves. Get involved in the mechanical design of the coil and jacket system to head off potentially severe problems at low heating and cooling loads. Make sure the utility valves have minimal backlash and stiction especially near the closed position that is the split range point. Use external reset using a fast read back of actual valve position and directional move suppression to help prevent oscillations.

The post, Essential Feedforward Control and Operator Interface Tips first appeared on ControlGlobal.com's Control Talk blog . All feedforward control systems can be reduced to a common form that enables a better understanding and recognition that leads to the best performance and the best interface for the operator. For “smart controls” to be fully appreciated and utilized, the operator needs to know what is going on and how to participate. Here we show how to make the advantages of feedforward control more achievable and recognizable. 99.99% of the feedforward control systems for PID control in process industry actually involve ratio control where the numerators and devisors of the ratio are flow, energy, power, or speed. Consider the following examples: Blend composition control - additive/feed (flow/flow) ratio Column temperature control - distillate/feed, reflux/distillate, reflux/feed, steam/feed, and bottoms/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 We have a desired ratio setpoint (e.g. feedforward gain) at a given process operating point based on process and operational knowledge. The primary process controller (e.g., composition, level, pH, or temperature) corrects this ratio that is then multiplied by the devisor (feedforward signal) and sent as the setpoint to a secondary controller (e.g., flow, speed, or power). Besides understanding that fundamentally we are essentially dealing with ratio control corrected by a primary process controller, we need to give the operator an interface to foster recognition and participation. The operator must be able to see the actual ratio (corrected ratio) and be able to set the desired ratio. Almost always the operator must be given the ability to put the primary controller in manual and operate on ratio control by simply setting the desired ratio. This capability is critical during startup of many unit operations (e.g., distillation columns) and for composition control when an analyzer is providing extraneous values and during subsequent analyzer servicing, replacement, or recalibration. There must be a bumpless transition between manual ratio control and corrected ratio control. Process engineers also appreciate the interface because they know that most process variables important for process performance are a function of a ratio of an input to manipulated flow or speed. The scaling for the feedback correction must be readily observable and adjustable. I can’t emphasize enough how important this setup and interface is in terms of longevity and performance of the feedforward control system. The desired ratio setpoint can be computed from a material or energy balance as detailed in the online White Paper " First Principle Process Relationaships " and 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 the 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 revealing why we need feedback correction by a primary controller. Here are some examples of assumptions: 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. The correction by the feedback controller for plug flow systems, sheets, extruders, and spin-line is best done by the controller providing the ratio (feedforward multiplier) because the input flow or speed multiplication compensates for the inverse relationship between the process gain and the input flow or speed. For well mixed volumes (crystallizers, evaporators, columns, neutralizers, and reactors), the decrease in feed flow increases the residence time and primary time constant which offsets the increase in process gain in terms of primary controller tuning. Here a bias correction of the input after multiplication by the desired ratio (feedforward summer) is best. The feedforward summer is also easier to scale and corrects for measurement drift and offset. A bias correction has a long history of being a robust correction of predictions of Model Predictive Controls, Neural Networks, and Inferential Measurements. The ratio setpoint can be optimized by simply adding an integral-only valve position controller whose controlled variable is the current bias correction, whose setpoint is a zero bias and whose output is the desired ratio. For dynamic compensation of the feedforward signal, the input flow, speed, or power is simply passed through a dead time and lead-lag block. The objective is for the effect of the manipulated variable to arrive in the process at the same point and the same time as the feedforward input. If the input feedforward and manipulated variable enter the process at the same point (e.g. blend and reaction unit operations), a setpoint of a speed or flow controller is used for the feedforward input. The controllers for the feedforward input (e.g., leader) and the manipulated variable (e.g., follower) can be tuned for the same closed loop time constant. If pressure disturbances are a consideration, the controllers can be each tuned for the fastest response and a filter applied to each setpoint so that the changes in flow are synchronized. Unfortunately, this strategy often does not work well for neutralization because reagent flow can have a much longer injection delay than the effluent flow (injection delay that is dip tube volume divided by injected flow is incredibly large for conventional dip tube size and length and an extremely small reagent flow). Consequently the effluent flow setpoint must be delayed and the setpoint before the delay used as the feedforward for the pH controller. For a glimbse of the enormous potential opportunity, see the March-April 2011 Intech feature article " Feedforward control enables flexible, sustainable manufacturing "

The post Simple Strategies for Maximizing Biopharmaceutical Product Quality first appeared on the ISA Interchange blog site. The following insights are part of an occasional series authored by Greg McMillan , industry consultant, author of numerous process control books and a retired Senior Fellow from Monsanto. Bioreactor batches can take days or weeks to complete. Each batch may be worth millions of dollars. Preventing bad batches and achieving repeatable high quality product is critical. The use of key PID features, tuning objectives and feedforward control can meet the stringent requirements for biopharmaceutical production. Interaction between DO and pH control can be greatly reduced by sparger design. Remaining interaction can be dealt with by an enhanced PID for DO and pH control with a threshold sensitivity setting for ignoring noise and a decoupling feedforward signal that is the opposing controller’s gas addition rate. Since tight pH control is more important than DO control, a feedforward of DO PID output (e.g., air flow setpoint) can be added to the pH PID output to help the pH controller immediately correct the carbon dioxide (CO 2 ) flow setpoint. The feedforward does not need to correct the pH PID output when manipulating the sodium bicarbonate setpoint. This simple half decoupler of air to CO 2 flow setpoint is usually sufficient unless there is a sparger design problem. Problems occur when air and CO 2 flow share the same sparger, the spargers are too close to each other, or there is no tight pressure control of air or CO 2 . Insight : Good sparger design can eliminate most of the interaction between pH and DO control. What remains can be reduced by a half decoupler where the air flow setpoint from the DO PID output is added to the pH PID output that is the CO 2 flow setpoint. The utilization of key PID features and straightforward feedforward and ratio control can enable bioreactor batches to be more predictable and consistent deliver higher quality product. See Greg McMillan’s new ISA book Advances in Reactor Measurement and Control for an extensive view of practical opportunities for designing control strategies to achieve product quality and maximize yield and capacity in different types of fermenters, bioreactors, and chemical reactors. High concentrations of carbon dioxide (CO 2 ) can be avoided by preventing overreaction of the pH controller to pH shifts (pH setpoint changes to promote product formation) by the use an enhanced PID and a secondary CO 2 flow PID with setpoint rate limits. High CO 2 concentrations will dissipate with time. Inert gases are sometimes added to help sweep CO 2 out of the batch. Mammalian bioreactor batches take 10 or more days. The exceptionally slow kinetics translates to extremely slow changes in reagent, oxygen, and cooling demand. Exceptionally slow load disturbances and slow integrating process gains result in negligible errors from disturbances. The objectives of tuning and control loop performance for pH, dissolved oxygen, and temperature control are in terms of setpoint response and optimization rather than disturbance rejection. The recognized opportunity is getting to a setpoint with no overshoot for a temperature shift and pH shift (setpoint change) to move from enhancing cell growth to promoting product formation. A PID structure that has proportional action on the process variable rather than on error and no derivative is useful for preventing overshoot. Getting to setpoint faster is not a consideration since the minutes saved are inconsequential in terms of the batch cycle time. Insight : Mammalian cell kinetics are extremely slow with batch cycle times of 10 or more days. The errors from load disturbances are so slow that tuning for disturbance rejection or getting to setpoint faster is not the goal. Most of the focus is on the prevention of overshoot. As previously mentioned, excessive accumulation of sodium bicarbonate causes a high sodium concentration leading to a high osmotic pressure that can contribute to cell membrane rupture and cell death. Unlike carbon dioxide that can escape in the vent gas, sodium bicarbonate is trapped in the media. Excessive crossing back and forth of the split range point for pH control over the course of the batch can lead to high osmotic pressure. A split range gap or deadband complicates tuning and does not stop the eventual excursion from the integrating process response. External reset feedback (dynamic reset limit) in the primary pH PID and setpoint rate limits in the secondary flow PID can provide the directional move suppression needed to slow down a movement into the direction of adding sodium bicarbonate reducing unnecessary split range excursions without requiring special tuning. New at-line analyzers can measure osmolality (osmotic pressure) by freeze point depression to help in optimization of the setpoint rate limits. Insight : External reset feedback in the pH PID and setpoint rate limits in the CO 2 and the sodium bicarbonate flow controllers can provide directional move suppression to eliminate unnecessary crossings of the split range point and an increase in osmotic pressure. Online and at-line analyzers enable substrate concentration control with glutamine feed ratioed to glucose. An OUR feedforward anticipates changes in glucose utilization rate. New online and at-line analyzers for measuring glucose and glutamine open the opportunity for improved concentration control of this sugar and this amino acid. At present, glucose and glutamine addition are scheduled in anticipation of utilization rates. A move from sequenced batch charges to fed-batch closed loop control can compensate for disturbances and consistently maintain a concentration (see figure). The glutamine feed setpoint can be ratioed to the glucose feed setpoint. A feedforward signal for glucose addition rate based on oxygen uptake rate (OUR) can provide preemptive correction for the acceleration of substrate utilization in the exponential growth phase. The sampling rate for at-line analyzers doesn’t need to be more frequent than once per 4 hours because the changes in mammalian cell metabolism are slow. An enhanced PID can provide tight control for relatively large sample times and can suppress reaction to noise and feedforward timing errors by a deadband (threshold sensitivity) setting. Insight : New online and at-line analyzers of glucose and glutamine concentration offer the opportunity for optimizing the ratio of these concentrations for better cell performance (higher cell growth rate and product formation rate and better product quality). The extensive use of external reset feedback is critical for preventing oscillations whether due to cascade control, split range discontinuities or discontinuous signals as discussed in the March 26, 2012 Control Talk Blog “ What is the Key PID Feature for Basic and Advanced Control .” This feature is an essential part of the enhanced PID originally developed for wireless that offers extensive advantages for analyzers as described in the July/August 2010 InTech article “ Overcoming Challenges of PID Control & Analyzer Applications ” All of these improvements can be explored, prototyped and tested by means of a high fidelity dynamic simulation of bioreactors I developed running much faster than real time that is easy to setup and use as a virtual plant. The use of key PID features and feedforward for bioreactors can provide more repeatable and higher quality batches. Take advantage of this opportunity by learning about the use of external reset feedback, an enhanced PID and a high fidelity model that are easy to configure.

The post Simple Strategies for Optimizing Ethanol Plants first appeared on the ISA Interchange blog site. The following insights are part of an occasional series authored by Greg McMillan , industry consultant, author of numerous process control books and a retired Senior Fellow from Monsanto. Corn is the highest cost factor in ethanol production. The addition of three PID controllers in an ethanol plant can improve yield and the ability to match front end production to back end purification. These controllers can make the most out of online and off-line analyzers providing recognition and reduction of the sources of variability. Yeast fermenters to produce ethanol traditionally only have temperature control. The subscribed permissible pH range is quite large (e.g. 4 to 6 pH). These fermenters typically don’t have pH control. A gradual decrease in pH during ethanol fermentation is normal due to acetic acid and lactic acid formation. A greater than normal drop in pH is indicative of a bacterial contamination so pH may be monitored. The main opportunity for maximizing production is taking advantage of a decrease in the time to reach the desired alcohol batch concentration endpoint. The inhibition effect of alcohol concentrations above the endpoint dramatically slows down and suspends further conversion to alcohol. Consequently, longer batch times do not translate into greater ethanol concentrations. Production rate can be increased by reducing the batch time to when the endpoint is reached. An online near infrared analyzer or an at-line chromatograph can provide a measurement of ethanol concentration. Most batch cycle times are much longer than the average time to an endpoint because of the variability in time to the endpoint. If batch cycle times are fixed, a consistent early endpoint can be used to reduce corn feed rate to improve ethanol yield, which is typically expressed as gallons per bushel. Corn is the largest cost for ethanol production. An improvement in yield taken as a decrease in corn feed rate significantly decreases cost and decreases the carbon footprint, which increases revenue. The causes of changes in production of the ethanol are often unknown. Corn quality may be the biggest unknown factor. Front end production may be the bottleneck or may not be coordinated with back end purification. The solution and resulting optimization involves a simple addition of PID controllers. See Greg McMillan’s new ISA book Advances in Reactor Measurement and Control for an extensive view of practical opportunities for designing control strategies to achieve product quality and maximize yield and capacity in different types of fermenters, bioreactors, and chemical reactors. The front end of the ethanol plant consists of slurry, liquefaction, cooling and saccharification-fermentation (SSF) areas. The front end is mostly continuous with batch operation of the fermenters in the SSF area. The basic control system for the front end of an ethanol plant controls the temperature and pH of the fermenter feed. The fermenter may have temperature control as well. The control system developed for optimization of ethanol yield in the front end uses three enhanced PID controllers, an at-line corn feed analyzer and an at-line multiplexed ethanol analyzer for the fermenters (See Figure). The PID controllers optimize the corn feed rate and solids in the slurry tanks based on predicted fermentable starch and actual SSF batch time to reach the ethanol end point. The enhanced PID controllers originally developed for wireless offer extensive advantages for analyzers as described in the July/August 2010 InTech article “ Overcoming Challenges of PID Control & Analyzer Applications ” Insight : Ethanol plant yield can be increased by the use of an at-line corn feed analyzer and enhanced PID to optimize corn feed rate, slurry solids concentration, and batch time. Insight : Enhanced PID controllers can optimize ethanol plant yield by the use of at-line analyzers. While the use of at-line analyzers is described, if only off-line analyzers are available, the same strategies can be used with minimal tuning correction. The enhanced PID controller will be stable if the PID gain is simply reduced by a factor of despite a variable and large analysis cycle time. The downside is that changes in front end will not be detected quickly and corrective action will be delayed and be more gradual. The first enhanced PID AC1-4 controller is on the first slurry tank. The PID is a production rate controller with its PV and SP scaled to ethanol production rate. The PV is computed based on corn feed rate and the percent of fermentable solids, with an inferential prediction of ethanol yield being provided by an at-line analyzer taking corn feeder samples. External reset feedback (Dynamic Reset Limit) is enabled so that the PID does not try to drive its output faster than the corn feeder can respond. The output of the PID manipulates corn feeder speed. For an increase in the inferential prediction of yield, the PID will immediately cut back on corn feed rate, reducing corn use. The simple change of the PID setpoint enables operations personnel to change front end production to match the back end separation and purification capability. Insight : An enhanced PID uses as its controlled variable a front end production rate prediction computed from corn feeder speed and an at-line corn analyzer prediction of yield. The PID can immediately translate an improvement in corn yield capability to a reduction in corn use. If the change in analyzer output from the last significant change is less than the threshold sensitivity setting of X gpm for production rate, there is no update to the PV and no change in PID output. Since the PID will wait till the effect of its feedback correction is seen and will ignore noise, the PID gain can be set equal to the inverse of the process gain to provide a full immediate correction in feeder speed to a significant new analysis result or a new production rate setpoint. In order to prevent an oscillatory response from this great increase in gain, the reset time must also be decreased to be about equal to the loop dead time. This decrease in reset time is counterintuitive because normally one would think of increasing the reset time to suppress oscillations. The increased reset action suppresses the proportional step when the measurement is reported as returning to setpoint. Changes in analyzer sample time or communication latency do not affect the tuning because PID action is suspended until there is an update. For changes in predicted yield (fermentable starch), the PID makes a single adjustment in corn feed rate bringing the production rate back to setpoint. The figure also shows that production setpoint changes can be easily made to match corn supply, market demands and back end capacity constraints. Insight : An ethanol plant production rate controller can make a single correction for an at-line analyzer update of a change in corn yield that will return the production rate to setpoint. The immediate and nearly exact correction is the result of being able to set the enhanced PID gain to be the inverse of the open loop self-regulating process gain. The second enhanced PID DC2-4 controller is on the second slurry tank. This PID is a slurry solids concentration controller with its PV and SP scaled to be 0-40% weight percent fermentable solids. Slurry solids concentration is inferred from a Coriolis meter density measurement in a recycle stream. The output of the PID manipulates the dilution water to the first slurry tank. Since the manipulation is to the first slurry tank while the solids being controlled are in the second slurry tank, there is an equivalent dead time from the residence time of the two slurry tanks plus mixing and injection delays. A feedforward signal is computed for the dilution water based on corn feed rate, steam injection, and backset recycle. To ignore feedforward timing errors, an update time of 12 seconds is set on the inferential measurement of fermentable solids. A dead time block in the external feedback path from the analog output to the PID provides dead time compensation. A threshold sensitivity setting is used to ignore measurement noise. Since the PID compensates for dead time, ignores noise, and waits out feedforward timing errors, the PID gain can be increased to provide a faster setpoint response. The feedforward signal from the change in Slurry Tank 1 prevents the changes in production rate from affecting the solids concentration in Slurry Tank 2. The feedforward signal is the production rate setpoint minus the water flow from other sources such as backset. The remote cascade (RCAS) setpoint is the desired solids concentration multiplied by a fermentable starch factor predicted by the corn analyzer. The prediction passes through a delay and lag set equal to the dead time and residence time, respectively, so the change in setpoint coincides with the change in solids measurement from a change in corn feeder speed from the production rate controller. Insight : An enhanced PID for slurry solids control in the front end of an ethanol plant is improved by addition of dead time compensation, a feedforward of dilution water and steam injection rate and a correction of setpoint based on corn feeder speed. The PID uses a Coriolis meter in a slurry tank recirculation line to provide an inferential measurement of percent solids. The third enhanced PID XC1-4 controller uses a High Performance Liquid Chromatograph (HPLC) measurement of ethanol profiles in simultaneous saccharification and fermentation (SSF) batches. Saccharification uses an enzymatic reaction to convert maltose, maltriose and higher polymer dextrins to glucose. Fermentation employs yeast to produce ethanol from glucose. In older plants, the saccharification and fermentation are done in separate vessels. The HPLC provides an inferred measurement that is a running average of the fermentation time (batch time to ethanol endpoint) in the most recent SSF batches. When the HPLC indicates that the ethanol has reached the desired endpoint, the running average of batch times is updated. Note that the actual batch cycle time is fixed. The average is the time to the desired endpoint. The enhanced PID output biases a correction to analyzer measurement of predicted fermentable solids, which changes the inferred production rate. The production rate controller correspondingly adjusts the corn feeder speed to bring the production rate back to setpoint. An unmeasured increase in fermentable solids will show up as an early achievement of the batch end point resulting in a cutback in corn feed rate providing an immediate improvement in yield. Insight : An enhanced PID controller can gradually correct the inferential prediction of ethanol yield used by the front end production rate controller by averaging the time that fermenter batches take to reach the desired ethanol endpoint. All of these improvements can be explored, prototyped and tested by means of a high fidelity dynamic simulation of bioreactors I developed running much faster than real time that is easy to setup and use. The addition of PID controllers for front end production rate, slurry concentration, and average fermentation time can decrease the cost and increase the production rate of an ethanol plant. Take advantage of this opportunity by learning about the use of an enhanced PID and a high fidelity model that are easy to configure.

The post, Best Control Valve Flow Characteristic Tips , first appeared on ControlGlobal.com's Control Talk blog . Often arguments as to whether a linear or equal percentage trim is best are based on the theoretical inherent flow characteristics. Valve rangeability is often stated as simply a deviation of the catalog flow characteristic from the theoretical characteristic. Here we will see how the consideration of the changes in process dynamics, available valve pressure drop, and control valve dynamics can alter what you consider as the best flow characteristic. The simple rule based on theoretical inherent flow characteristics and types of processes is that for flow, pressure, and level loops a linear trim is best because the process gain and valve gains are linear and for temperature processes an equal percentage trim is best because the process gain is inversely proportional to flow, which is the opposite of the valve gain. The real story requires a greater understanding of the effects at play. The best trim characteristic is the one that minimizes changes in PID tuning, minimizes the effect of backlash and stiction, offers the greatest rangeability, and prevents abnormal conditions. First of all, the actual installed flow characteristic of linear and equal percentage trims become more like a quick opening and linear characteristic, respectively as the ratio of the valve pressure drop to system pressure drop at maximum flow decreases as shown in the Installed-Control-Valve-Characteristics-Figures . Due to an increased emphasis on energy savings and an attempt to show variable speed drive valves do not save as much energy as touted, some may say that only 5% of the system pressure drop needs to be allocated at maximum flow (pressure drop ratio = 0.05). While hopefully this gamesmanship is not taken seriously, allocating less than 25% of the system drop can appreciably make the linear trim gain nonlinear and the equal percentage gain no longer proportional to travel losing the assumed benefits of both trims for the processes commonly cited. Additionally, the pressure drop ratio decreases as the frictional losses in the system increase. Fouling of inline equipment can make this a deteriorating situation with time. Hunter Vegas said as much in his reply to the Feb 26, 2015 Control Question “ Linear or Equal Percentage Valves: When Should I Use Which ? Not realized as well is the increase in backlash and stiction as the valve approaches the closed position. Tests are typically not done for positions less than 20%. Furthermore, the effect on process flow depends upon the valve gain since deadband and resolution limit are given as a percent of valve stroke. The error in flow for a given deadband or resolution is 4 times larger for a linear valve than an equal percentage valve at a 10% valve position. If you consider the real rangeability is the maximum flow divided by the minimum controllable flow and take into account the installed flow characteristic and limit cycles introduced by deadband and resolution, the equal percentage valve has a better rangeability. This makes more practical sense to me than the official definition of rangeability based on deviation of the catalog inherent characteristic from the theoretical characteristic which in most cases would then make a linear trim the best choice for rangeability. I have been saying this for over 20 years without much traction except for the satisfaction that the users in the plants who have said “Thank You - Now I know why an equal percentage characteristic gives me better control.” While it is true the process gain of temperature processes increases at low loads, the process time constant also increases at low loads for volumes with some degree of mixing (e.g. vessels and columns) because the process time constant is proportional to residence time (volume/flow). The controller gain is proportional to the ratio of the process time constant to process gain. Thus, for these volumes the effect of flow cancels out in terms of controller tuning and a true equal percentage characteristic would be introducing a nonlinearity that is actually detrimental in terms of tuning. Furthermore, most temperature loops benefit from a secondary flow loop to compensate for pressure upsets and to enable flow feedforward. Here the temperature loop is isolated from the nonlinearity of the valve. These same considerations apply to concentration control loops because the process gain and process time constant are both inversely proportional to flow and secondary flow loops are beneficial particularly for reactors and neutralizers when high rangeability flow meters are used (e.g., Coriolis and magnetic flow meters). If the pressure drop ratio approaches one (available pressure drop is relatively constant), the installed flow characteristic is the inherent characteristic. If the deadband and resolution is negligible, the minimum controllable flow is solely dependent on the uniformity of the flow characteristic near the seat. For these big “Ifs”, linear trim is best for flow, level, and pressure. For the prevention of surge and gas pressure relief, a linear trim is best regardless of “Ifs” because a more immediate response is essential even if it somewhat like a quick opening characteristic. For the control of inline temperatures (e.g., heat exchangers) and concentration (e.g., static mixers) control where the lack of back mixing means the process time constant is negligible and the big “Ifs” are true, an equal percentage characteristic will help keep the tuning less dependent upon load (e.g., feed flow). The advantage of using signal characterization is not straightforward. The characterizer must be based on the installed characteristic that is often unknown and dependent upon frictional losses in the system. Also, whether linearization of the valve gain is beneficial depends upon how the process gain and time constant change with flow, which is often not well understood. Finally, the effect of signal characterization is a mixed bag. For a given change in controller output, the changes in valve signal after characterization are smaller and larger for operating points on the steeper and flatter portion of the installed characteristic, respectively. Small changes approaching the actuator/positioner threshold sensitivity limit will increase the stroking time. Small changes less than the dead band or resolution limit of the valve will increase the dead time associated with the delay until the accumulated changes is large enough to get the valve to start to move. There is an opportunity for the pressure sensors to measure the valve pressure drop and make a fast online calculation of the installed characteristic and flow from knowledge of the valve drop, actual valve position, and inherent flow characteristic. The valve signal then becomes a desired flow providing a linear valve gain. It is not precise enough to replace an accurate flow meter (e.g., Coriolis or magnetic flow meter) but can be used to extend the rangeability of flow meters that get noisy or erratic at low fluid velocities (e.g., differential head or vortex meters). External reset feedback of the fast flow calculation should be used to prevent oscillations from a secondary flow loop trying to change the valve flow faster than the valve can respond. We can summarize what we have learned as follows: Signal characterization must be done cautiously with knowledge of the installed flow characteristic. An opportunity exists for a fast online calculation of flow and installed flow characteristic. Accurate high rangeability flow measurements in secondary loops are the best way of isolating valve nonlinearities from upper loops, compensating for pressure upsets and providing feedforward control. For applications with a precise valve (e.g., negligible deadband and resolution limit), nearly constant valve drop (e.g., high valve to system drop ratio), and negligible changes in process gain with load (e.g., flow, level, or pressure), or that need an immediate significant flow response to prevent abnormal conditions (e.g., surge control and pressure relief prevention), a linear trim may be best. Otherwise, an equal percentage trim is generally best for many practical reasons reasserting what most technicians and engineers in the field have realized from many years of field experience. Maybe this is a lesson to listen more to these practitioners on the front line. I had already learned this decades ago when I found out why a technician did not want to replace a positioner with a booster on a surge valve as discussed in the 2/09/215 Control Talk Blog “ Secondary Flow Loop and Valve Positioner Tips ”

The post Optimizing Multiphase Reactors with a Single Phase Product first appeared on the ISA Interchange blog site. The following insights are part of an occasional series authored by Greg McMillan , industry consultant, author of numerous process control books and a retired Senior Fellow from Monsanto. For both batch and continuous reactors, completion control seeks to provide a complete conversion of all the reactants and consequently no excess accumulation of a reactant in a particular phase. If the reactants are in different phases and the product is a single phase, inventory control can be used for reaction completion control. The product must be a gas, liquid, or solids with no recycle or co-products in the other phases. Inventory control is possible if there is a discharge flow in addition to a feed flow to the phase. Gas volumes have a continuous discharge flow in batch as well as in continuous reactors by means of an off-gas flow going to a vent or recovery system. Liquid and solid volumes only have a discharge flow in continuous operations unless there is a phase separator physically built-in to continuously separate solids from liquids during batch operation (e.g., a hydroclone in a recirculation line). Insight : Completion control seeks to provide for batch and continuous reactors a complete conversion of all reactants. Multiphase reactors offer an inherent ability to optimize excess reactant concentration by the proper use of a simple pressure or temperature control loop. The remaining optimization opportunity is in terms of setting the best purge flow. See Greg McMillan’s new ISA book Advances in Reactor Measurement and Control for an extensive view of practical opportunities for designing control strategies to achieve product quality and maximize yield and capacity in different types of reactors. Using these concepts one can determine whether inventory control provides completion control. For a liquid product, excess gas reactant is inherently prevented by pressure control for both batch and continuous processes. For a gas product, excess liquid reactant is inherently prevented by level control for continuous processes and possibly solids control for all processes. Insight : For multiple phase reactants and a single phase product, inventory control can provide reaction completion control. Batch processes do not have liquid inventory control (level control) to prevent the accumulation of reactants in the liquid phase. For a liquid product, a pressure controller provides continuous completion control by increasing gas feed for a decrease in pressure from a deficiency of gas reactant and by decreasing gas feed for an increase in pressure from an excess of gas reactant (Figure 1). The gas phase reaction is normally fast enough for the gas reactant to be totally consumed in the reaction so that only the inerts are left in the overhead vapor space. An off-gas purge flow prevents the accumulation of inerts. A level controller maintains the liquid material balance by manipulating the liquid product discharge flow. The residence time control by the level loop shown for a single phase liquid reaction (Figure 4) could help maintain the residence time in the gas phase as well the liquid phase by keeping the bubble rise time constant. The bubble rise time is the bubble path length set by level divided by the bubble velocity set by sparge rate. For pure batch and fed-batch reactors there is no level control and hence no residence time control but otherwise the control scheme is applicable. Insight : An off-gas purge of inerts enables gas pressure control to provide reaction completion control in the gas phase for a liquid product. Figure 1: Basic Control of Multiphase Reactor with Liquid Product For a gas product, a level controller provides continuous completion control by increasing liquid feed for a decrease in level for a deficiency of liquid reactant and decreasing liquid feed for an increase in level from an excess of liquid reactant (Figure 2). A purge flow from the bottom prevents the accumulation of inerts in the liquid phase. A gas pressure controller maintains the gas material balance by manipulating the gas product discharge flow. For residence time control, the liquid controller setpoint computed as the liquid reactant flow multiplied by the desired residence time would need a lag inserted because the level controller manipulating the liquid reactant flow forms a positive feedback loop. For fed-batch reactors this strategy is not applicable because there is no level control or purge. Insight : A bottom liquid purge of inerts enables liquid level control to provide reaction completion control in the liquid phase for a gas product. Production rate can be maximized in both cases by the use of a valve position controller (VPC) monitoring coolant valve position. The VPC setpoint is the maximum desirable valve position, and the VPC process variable is the temperature controller output. The output of the VPC trims the setpoint of the liquid and gas reactant flow controller for liquid and gas products, respectively. An enhanced PID with dynamic reset limiting for the VPC eliminates limit cycles, reduces interaction between the VPC and temperature controller, and enables smoother optimization with faster correction for disturbances. Figure 2: Basic Control of Multiphase Reactor with Gas Product The purge flow for both types of reactor presents an opportunity for optimization. An excess accumulation of inerts reduces the residence time and reaction rate. Inhibiters can reduce yield and initiators of undesirable compounds can cause poor product quality besides reducing yield. First principle models can provide an inferential measurement of inert, inhibiter, or side reaction initiator accumulation. The predicted accumulation is corrected by at-line or off-line analyzers of the concentrations in raw materials (reactants) and phases within the reactor. It is important that research studies be conducted on the type of inerts, inhibiters, and initiators of undesirable reactions that can be found in the reactant feeds and their effect on reaction rate and yield. For a single liquid product, use tight gas pressure control to prevent excess reactants in the gas phase and optimize a gas purge. For a single gas product, use tight liquid level control to prevent excess reactants in the liquid phase and optimize a liquid purge. In both cases use research studies and measurements of components in raw materials and reactant phases that adversely affect reactor yield and capacity to optimize the purge rates.

This post, Good Tuning Rule of 5 Tips , first appeared in the Control Talk blog on ControlGlobal.com . Most of the control literature focuses on minimizing the integrated absolute error (IAE) for a step disturbance, often in a linear system. In the process industry, there are many other objectives and complications that require special attention. Here we quickly review when and how to achieve minimum IAE and see how a rule of 5 comes into play for applications with different challenges. Step disturbances are rare if the automation system is well designed and complete. The fastest disturbances are liquid pressure and flow. Often a pressure change shows up as a flow disturbance. If flow controllers and throttling valves are installed rather than on-off valves discretely opened and closed on all the streams, the flow loops correct for pressure upsets and provide a closed loop time constant for changes in flow. Pressure controllers can be tuned for tight control (maximum transfer of variability from controlled variable to manipulated variable) to reduce the propagation of pressure upsets. Level controllers can be tuned for maximum absorption of variability (minimum transfer of variability) to reduce the propagation of level upsets as detailed at the end of the white paper “ So Many Tuning Rules, So Little Time ”. http://www.controlglobal.com/whitepapers/2014/so-many-tuning-rules-so-little-time/ Tight control (minimum IAE and/or peak error) is important for pressure and many loops for product formation (e.g. reaction) and purification (e.g. distillation). If the system is linear and the dynamics are well known (Big “Ifs”), the PID gain could approach being about twice the PID gain from Ziegler-Nichols reaction curve method turning. This PID gain setting corresponds to a closed loop time constant for self-regulating and an arrest time for near-integrating and true integrating processes that is about equal to the dead time. For runaway reactors, the PID gain may be further maximized for worst case dynamics by an arrest time equal to half of the dead time. Lambda is the closed loop time constant in self-regulating tuning rules for processes with a time constant to dead time ratio less than 4 and Lambda is the arrest time in integrating process tuning rules for all other processes (near-integrating, true integrating, and runaway). In all cases, Lambda rather than a Lambda factor is used and Lambda is set relative to the dead time. This makes sense in that dead time is the primary fundamental limitation to loop performance and the ultimate period is a factor of the dead time (e.g., approximately 4 dead times for balanced and 2 dead times for dead time dominant self-regulating processes). For many challenging situations, the rule of 5 is useful. Table D-1 Lambda Tuning Solutions for Difficult Situations and Different Objectives shows that to minimize the effect of resonance, inverse response, and nonlinearities, Lambda should be greater than 5 dead times. To prevent the violation of the cascade rule, the upper (primary) loop Lambda should be at least 5 times larger than the lower (secondary) loop Lambda. To minimize interaction, the slower loop Lambda should be at least five times larger than the faster loop Lambda. In these last two cases, it is in general preferable to make the lower and faster loop Lambda smaller rather than making the upper and slower loop Lambda larger. Note that in these situations, what may seem as a small disturbance can be amplified and propagated when these rules are violated. While the use of external reset feedback can help minimize consequences inherently for cascade control and by the use of directional move suppression for interactions, paying attention to the rule of five helps provide better control. If pressure disturbances are an issue, equal setpoint filters rather than the equal lambdas shown in Table D-1 can be used for the coordination of flow loops to maintaining a blend or reactant feed stoichiometric ratio. This allows each loop to be tuned for the fastest response to pressure upsets. For a perceptive, concise and direction approach to choosing tuning settings for a large variety of applications see the 2015 version of Good Tuning: A Pocket Guide – A Pocket Guide - Fourth Edition . For more information on how to minimize the sources and adverse consequences of disturbances, see the October 2013 Control Talk Blogs “ Disturbance Dynamics Recommendations ” and “ Effect of Disturbance Dynamics Perspective ”.

The post Getting the Most Out of Fluidized Bed Reactors first appeared on the ISA Interchange blog site. The following insights are part of an occasional series authored by Greg McMillan , industry consultant, author of numerous process control books and a retired Senior Fellow from Monsanto. Fluidized bed reactors are extensively used in the petrochemical business. Often these are high volume plants with multiple lines of reactors. Even a small percentage improvement in yield and production rate can translate to millions of dollars a year in additional profit. Fortunately, the product rate can be inherently maximized by a straight forward temperature control loop. The challenges of noise and sensor lag can be addressed by sensor installation and signal averaging and selection. For gas reactants and a gas product, a pressure loop controls the material balance and provides the time available for reaction at a given production rate by manipulating the discharge product flow to balance the total feed flow (see the figure below). The residence time for these fast reactions is small (e.g., a few seconds) but still must be kept above a low limit. A fluidized catalyst bed is used to promote reaction rate. If a temperature loop manipulates the leader gas reactant flow, the production rate is automatically maximized by the temperature and pressure controllers for a given cooling rate established by BFW flow and the number of coils in service as discussed below. Valve position control is not needed unless a control strategy is added to manipulate the BFW. Direct manipulation of feed rate by the temperature control is possible in gas reactors because the additional time lag for composition response is negligible due to the small residence time and the inverse response at the control points is negligible due to the fast reaction, high heat release, and catalyst heat capacity. Insight : Temperature control of fluidized bed gas reactor by manipulation of feed rates will inherently maximize feed rate for a given cooling capability with no appreciable inverse response or composition response lag. The fast kinetics and composition response of a fluidized bed reactor enable the direct manipulation of reactant feed rate by temperature control to inherently maximize production rate to the coolant capacity. However, channeling of flows, hot spots and thermowell lags pose challenges. Here we look at some of the simple solutions. See Greg McMillan’s new ISA book Advances in Reactor Measurement and Control for an extensive view of practical opportunities for designing control strategies to achieve product quality and maximize yield and capacity in different types of reactors. A gas reactor with a fluidized catalyst bed may develop hot spots from localized high reactant concentrations due to a non-uniform flow distribution and no back mixing. Numerous separate cooling coils are used so operations personnel can switch coolant coils in or out of service to deal with hot spots and changes in production rate. However, the switching causes a disturbance to the temperature controller as fast as the BFW on-off valves can move. Numerous thermowells each with multiple sensors traverse the reactor. Special designs can maximize the contact between the sensor tip and the thermowell wall to minimize the sensor lag. Since the gas velocity is usually quite high, the sensor time constants can be small if the insulating effect of air in the thermowell at the sensor tip is minimized. Note that this sensor time constant is the largest time constant in the loop and can mislead one into thinking a slower measurement is better due to the filtering effect and the larger maximum allowable temperature controller gain as detailed in the Dec. 2, 2014 Control Talk Blog “ Measurement Attenuation and Deception Tips. ” The average temperature is computed for each traverse with the highest average selected as the control temperature. Only three  thermowells and BFW coils are shown in the figure due to pictorial space limitations. A feedforward signal can provide preemptive correction for the disruption of coil switching by means of a gain and rate of change (velocity) limit set to match the BFW on-off valve installed characteristic slope and stroking time. Insight : Fluidized bed reactors use the highest of the average bed temperature at various distances in the flow direction to control hot spots. Insight : A feedforward signal based on the installed characteristic and stroking time of the BFW on-off valves can be used to reduce the temperature upset from the switching of coolant coils . The plug flow of reactants through the reactor provides a tight residence time distribution. To provide a residence time greater than the reaction time for the greatest production rate, the gas volume must be large enough and the flow distribution uniform enough for the reaction to go to completion. Changes in discharge composition are mainly due to errors in the flow measurement, hot spots triggering side reactions, or insufficient radial mixing. An excess of one reactant is often used to ensure complete conversion of other reactants. An at-line analyzer on the gas product can be used to correct the gas reactant ratio to improve yield by reducing excess reactant. First principle dynamic models have also been used to provide a fast inferential measurement of excess reactant concentration for well-defined reaction kinetics. The models are corrected by taking a fraction of the difference a validated analyzer result and the inferential measurement delayed to be synchronized with the analyzer response as a bias correction to the fast inferential measurement without the analyzer delay. The leader reactant flow multiplied by a ratio factor is the feedforward signal for the composition controller. A feedforward summer is used even though a feedforward multiplier can compensate for the inverse relationship between process gain and flow because of scaling and analyzer reliability issues and the predominant error seen is a bias rather than a span error in the flow measurements. The composition loop trims the feedforward signal. An enhanced PID with a threshold sensitivity setting helps deal with the analyzer sample and cycle time and the noise from poor mixing. Insight : Plug flow reactors have a tight residence time distribution but no opportunity for residence time control by level control. The reactor volume must be large enough and the flow distribution uniform enough to provide enough residence time at the highest production rate. Insight : An at-line analyzer and inferential measurement can reduce the excess concentration of a reactant used to ensure the complete conversion of other reactants. Use good temperature sensor installation practices to minimize temperature lag, averaging of signals to minimize noise, and high signal selection to deal with hot spots for tighter temperature control. Based on temperature sensor trends, find possible sources of poor flow distribution. Increase coolant capacity to increase production rate. Use rate of change limits and an inferential calculation of cooling based on installed valve characteristic to match the slewing rate of the BFW valve and the actual BFW flow as part of feedforward of coolant capacity changes to assist the temperature controller. Use an inferential measurement of excess concentration in the reactor discharge corrected by an analyzer to adjust the ratio of reactants to improve yield.

This post, How MPC will Take Over More of the Role of PID Tips , first appeared in the Control Talk blog on ControlGlobal.com . The power of the PID largely remains untapped. I have recently documented the extensive capability of the PID but being a realist, I expect MPC is going to take over more and more of the role of the PID. Here we look at the reasons why there is a brighter future for MPC and if there is an opportunity to reverse the decline in PID expertise. When I taught process control at Washington University in St. Louis to Chemical engineers after retiring from Solutia-Monsanto in 2002, I used MPC more than PID control in the lectures and labs because the application of MPC is more automated and the principles more relatable in terms of the process response than PID control. While a process model is the first step and basis in getting good PID tuning settings, the hundreds of tuning rules and extensive disagreements between the developers of the rules are confusing and discouraging as noted in the white paper “ So Many Tuning Rules, So Little Time ”. First let’s clear up misconceptions. I have nothing to sell. I am largely retired. I am not trying to sell my services. The royalties from a book are good for a night out. I write because I don’t want expertise lost. So here is a quick perspective revealing that the PID can do more than recognized and the MPC can learn from the extensive flexibility and capability of the PID. The PID can do dead time compensation by the simple insertion of a dead time block in the external reset path of the PID. Only a dead time parameter needs to be set whereas the Smith Predictor also required the identification and setting of the process gain and time constant. This dead time can be written to. The online calculation and setting of the dead time is critical particularly when the source is a transportation delay. Furthermore the normal application thought to be of greatest benefit (dead time dominant loops) does not see as much improvement as in lag dominant loops and can cause oscillations for just a 10% overestimate of the dead time. This is counter intuitive since we are normally concerned about underestimates of the dead time. The PID reset time can be greatly reduced if the dead time setting is accurate. The benefit of dead time compensation is not seen until the reset time is reduced to much lower values than found from tuning rules. The PID can provide directional move suppression that is also computable and settable online by the use of setpoint rate limits in the secondary loop or analog output block and external reset feedback (e.g., dynamic reset limit). The PID can also inherently prevent oscillations from violation of the cascade rule by the use of external reset feedback of the secondary loop PV and fast digital valve controller (DVC) readback of actual valve position. The PID can stop limit cycles from backlash and stick-slip by the integral deadband or an enhancement of the PID where integral action is suspended if there is no appreciable change in the process variable. The enhanced PID can also suppress oscillations and eliminate the need for detuning when an analyzer cycle time is larger than the process dead time and process time constant. The PID can do feedforward control and decoupling with dynamic compensation so the preemptive correction signal arrives at the same place at the same time in the process as the disturbance. Different PID structures can be chosen, some critical (e.g., no integral action due to unidirectional response). All of these features help in the implementation of valve position control to provide a gradual optimization, a fast getaway for upsets, and prevention of limit cycles. Auto tuners, adaptive controllers, and the “Rule of Five” scheduled for my April 12 Control Talk Blog make the PID able to deal with an extensive spectrum of difficult situations and different objectives. Sounds great, but the expertise required is largely undocumented and not automated. My latest book Tuning and Control Loop Performance – 4th Edition , attempts to remedy this situation but even after 556 pages, there are still gaps. Besides, who has time to read and study that much material? I just finished the 4th edition of my Good Tuning Pocket Guide available in April to help provide a more concise directed view. Still, your best bet is to get a consultant onsite. Meanwhile, the MPC implementation is very much automated and the tuning often simplifies to just setting a move suppression parameter (e.g., penalty on move). In other cases another parameter is set to provide more or less emphasis on a controlled variable or constraint (e.g., penalty on error). The dynamics of decoupling, feedforward, optimization for multiple variables are inherently addressed and signal characterization can be used for gain nonlinearities. If MPC move suppression and model dead time and remaining nonlinearities can be adapted online, and something akin to external reset feedback is available the lone remaining advantages of the PID are largely gone and come down to the applications where a special PID structure (e.g., unidirectional response), high PID gain and rate time for open loop unstable processes (e.g., highly exothermic reactors), or where a fast PID execution rate due to process dead times and times constants less than 1 second (e.g., compressor and polymer pressure control) is needed. For more on what MPC and PID can do, see the following Control Talk Blogs and the referenced Control Talk Columns with MPC and PID experts. MPC Best Practices for ISA Certification of Automation Professionals When do I use MPC rather than PID for Advanced Regulatory Control? When do I use PID, MPC or FLC for Basic Control? Checklist for Best PID Performance PID Structure Tips PID Form Trick or Treat Tips

The post Maximizing Production Rate of Single Phase Liquid Reactors first appeared on the ISA Interchange blog site. The following insights are part of an occasional series authored by Greg McMillan , industry consultant, author of numerous process control books and a retired Senior Fellow from Monsanto. The control scheme commonly used by gas reactors that inherently maximizes production rate cannot be effectively used by liquid reactors. The optimization of production rate can be achieved by the simple configuration of a PID loop that uses existing signals. However, special PID features must be used to prevent oscillations and to deal with unmeasured disturbances. The simplest control scheme for maximizing production rate in liquid reactors would be simply setting the coolant valve completely open and having the temperature controller bring in as much feed as the maximized coolant system can handle. However, the manipulation of reactant feeds by a temperature controller for maximization of production rate in liquid reactors causes inverse response for feeds colder than the reaction temperature and introduces a large lag that is the composition time constant of the vessel volume. For fast gas reactions, the residence time and consequently the composition time constant are small enough for this scheme to be effectively used, particularly when reactions are highly exothermic with a large heat generation that overwhelms any cooling effect of feeds. Insight: The manipulation of reactant feeds for temperature control in liquid reactors does not work due to excessive inverse response and lag in the liquid composition response. The control scheme is productive for fast gas reactions. In a first order plus dead time approximation for the temperature response, all time constants smaller than largest time constant are converted to equivalent dead time. The thermal process time constant for temperature control should be the largest time constant and the primary process time constant in the loop. The direct manipulation of feed rate for temperature control coupled with a slow reaction rate can cause a composition response time constant larger than the thermal time constant. The result is that a fraction of the thermal time constant becomes effectively dead time, causing a decrease in the maximum allowable PID gain and an increase in the minimum reset time and the peak and integrating errors for load disturbances. Insight: A composition response in series with the thermal response in liquid reactors creates excessive dead time and deterioration in loop performance. In a liquid reactor a coated temperature sensor or a sensor tip that does not extend past the nozzle into the vessel contents or a sensor hidden behind a baffle, could cause a measurement time constant larger that is than the thermal time constant. A sensor in a ceramic protection tube in a gas reaction phase or a sensor in a baffle with a glass coating in a small liquid volume will also have an excessive measurement lag due to the poor surface thermal conductivity of the sensor installation. The slow composition response from manipulation of liquid reactant feeds causes slow correction and the slow sensor causes slow recognition of a disturbance. In either case, thermal time constants become effectively dead time and control severely deteriorates. Insight: A sensor that is coated or is installed in a stagnant zone (e.g., behind a baffle) will add an excessive measurement lag. The temperature sensor location is also important for the secondary jacket temperature loop. The sensor should be located in the coolant recirculation line rather than in the jacket. The higher velocities and turbulence in the pipeline provide a faster measurement with fewer fluctuations from level and phase changes and cold or hot spots from product sticking on the reactor wall. Inverse response and a slow composition response in series with the thermal response prevent the use of a temperature controller to directly manipulate reactant feed to inherently maximize production rate.   However, a simple PID loop can be added to maximize production rate to the limit of cooling capability. See Greg McMillan’s new ISA book, Advances in Reactor Measurement and Control, for an extensive view of practical opportunities for designing control strategies to achieve product quality and maximize yield and capacity in different types of reactors Production rate can be maximized by the use of a valve position controller (VPC) monitoring coolant valve position. The VPC setpoint is the maximum desirable valve position, and the VPC process variable is the jacket temperature controller output. The use of actual valve position is unnecessary if the coolant valve has a digital positioner. The maximum throttle position setpoint keeps the coolant valve near a point on the installed characteristic that has sufficient slope (valve gain) to correct for disturbances. The output of the VPC trims the setpoint of the “leader” reactant flow controller. An enhanced PID with external reset feedback (dynamic reset limiting) for the VPC eliminates limit cycles from coolant valve deadband, reduces interaction between the VPC and the jacket temperature controller, and enables smoother optimization with faster correction for large disturbances by directional move suppression. Directional move suppression by means of rate limits on the manipulated feed flow setpoints enables a gradual optimization and fast correction for abnormal conditions. A reactant feed flow set point rate limit should be fast for correcting an increase in valve coolant position to prevent “running out of valve” for high heat releases. The setpoint rate limit should be slower for the opposite direction to provide a more gradual optimization. External reset feedback in the PID automatically prevents integral action from driving the VPC PID output faster than rate limits allow or than the feed flow can respond. To see the increase in production rate in fed-batch operation from a higher feed rate, either the batch cycle time must be allowed to decrease or the batch mass allowed to increase. Insight: A VPC, whose setpoint is the maximum coolant throttle position, can maximize feed rate smoothly with less interaction and better response to disturbances by the use of an enhanced PID and reactant feed setpoint rate limits to provide directional move suppression. The reactant feeds could be fresh reactant streams from raw material storage or recycle streams when excess reactants are recovered in downstream separators. For the recycle case, the level in the corresponding downstream separator volume must be controlled by the manipulation of makeup reactant flow. For an excess reactant that is a heavier or lighter component (has a lower or higher boiling point) than the product, the separator volume that is the source of the reactant recycle flow, is a column sump or distillate receiver, respectively. There must be a flow controller somewhere in the path of a recycle stream from reactor discharge to reactor feed to prevent the “snowballing effect”, an escalating divergence of excess reactant concentration from any upset. The flow controllers on the reactant feeds serve this purpose when these are recycle streams. I nsight: When excess reactants are recovered and recycled to the reactor from downstream separators, a flow controller must exist in each recycle path to prevent the “snowballing effect”. The production rate in well mixed single phase liquid reactors can be achieved by the addition of a valve position controller that pushes a coolant valve to its maximum throttle position. External reset feedback ideally by means of an enhanced PID should be used along with directional setpoint rate limits on the manipulation of reactant feed rate to provide a slow optimization but fast correction of unmeasured disturbances. This approach minimizes oscillations from coolant valve backlash and stick-slip and from interaction with the temperature loop. When reactants are recycled, a flow controller must be present in each recycle path to prevent a “snowballing effect” of excess reactant concentration.

The post Achieving Tight Control of Single Phase Liquid Reactors first appeared on the ISA Interchange blog site. The following insights are part of an occasional series authored by Greg McMillan , industry consultant, author of numerous process control books and a retired Senior Fellow from Monsanto. Well-mixed liquid reactors have dynamics that offer the possibility of extremely tight control of product composition but lack some of the self-regulation in multiple phase reactors achieved by inventory control. For a reactor to achieve its full potential requires an intelligent design of the control strategy. Here we look at how to setup up the control loops based with some guidance as to how the loops change for batch, fed-batch, or continuous reactors. For reactions that are all in one phase (gas, liquid, or solid), inventory control (pressure or level control) cannot automatically adjust a reactant flow for changes in conversion (amount of product or byproduct formed per unit of reactant added) to prevent an excess or deficiency of reactant in a product phase (e.g., liquid) that is the opposite of the reactant phase (e.g., gas). Consequently, for single phase reactors, the use of analyzers is more important to control reactant stoichiometry for maximizing yield beyond what tight temperature control can do. Insight: Single phase chemical reactors have a greater dependence on analyzers to maximize yield since there is not the inherent prevention of excess reactants by pressure or level control. For well mixed single phase reactors the largest sources of improper reactant concentration leading to excess reactant are errors in reactant flow measurement and changes in reactant composition. Note that a deficiency in one reactant concentration creates an excess of another reactant. Coriolis meters can be used to provide the greatest mass flow measurement precision and rangeability with density correction for any changes in reactant feed concentration. Consequently, Coriolis meters on reactant feeds eliminate most of the sources of reactant unbalances if the mass flow ratios are correct and coordinated to maintain reaction stoichiometry. Insight: Coriolis meters in reactant feeds are the best way to ensure correct stoichiometry if the feed setpoints are changed in unison with the correct ratio and the same closed loop response. If the density of the excess reactant is significantly different than the density of the other components in the reactor, a Coriolis meter in the recirculation line can provide an inline inferential measurement of the excess reactant concentration to provide a fast and accurate correction of the ratio of reactants. Insight: If changes in excess reactant concentration cause density changes greater than 1 percent and are at least 100 times larger than other density changes, a Coriolis meter in a recirculation line can be used as an inline analyzer for excess reactant concentration. For continuous reactors with liquid reactants and liquid product, a level loop controls the inventory and thus the time available for reaction by manipulating the discharge product flow. If the desired reactor residence time multiplied by the total reactant flow is scaled and used as the setpoint of the level controller, the level loop helps maintain a constant residence time important to provide enough time for slow reactions. The reactor discharge flow is not used in the residence time calculation, to avoid the creation of a second loop within the level loop where a change in level PID output would cause a change in level PID setpoint which would cause a change in level PID output and so forth. The reactant feed setpoints are used instead of reactant flow measurements for the same reasons that setpoints are used in feedforward flow signals. Well-mixed single liquid phase reactors have an excellent process time constant to dead time ratio but lack the ability for inventory control to inherently optimize excess reactant concentration.   The use of accurate ratio control of reactant flows is critical. See Greg McMillan’s new ISA book, Advances in Reactor Measurement and Control , for an extensive view of practical opportunities for designing control strategies to achieve product quality and maximize yield and capacity in different types of reactors. Batch reactors don’t have a level loop and the concept of a residence time is replaced with a batch time for available time for reaction. The rising level of a batch reduces any self-regulation that might exist; changes in heat or mass transfer with temperature and concentration affect the open loop responses of temperature and product composition necessitating the use of higher controller gains to provide sufficient closed loop regulation. Integral action is also more likely to produce overshoot of setpoint. Insight: A liquid level controller with a setpoint proportional to production rate can help provide the necessary residence time for liquid reactants in a continuous reactor. Temperature loops control the energy balance and the reaction rate through the Arrhenius equation for chemical reaction kinetics. A cascade temperature control system offers the greatest linearity and responsiveness to coolant pressure and temperature upsets. The reactor temperature PID manipulates the setpoint of a jacket inlet temperature control that in turn manipulates makeup coolant flow. The coolant exit flow, such as cooling tower water (CTW) return flow, equals the coolant makeup flow to the jacket by piping design and in some case by pressure control in the recirculation system. The resulting constant jacket coolant flow eliminates the increase dead time, process gain and fouling from a decrease in jacket flow. An enhanced PID is used for the jacket inlet temperature loop to prevent limit cycling from valve stick-slip. Insight: Temperature control determines the reaction rate via the Arrhenius equation. Insight: Cascade control of reactor to jacket temperature isolates the reactor temperature PID from most of the nonlinearities and disturbances associated the jacket temperature response. For batch reactors, as previously mentioned there is no level control and hence no residence time control but otherwise the temperature and gas pressure control systems shown for continuous liquid reactors are generally applicable. The tuning is even similar since a large primary process time constant enables continuous liquid reactor composition and temperature loops to be treated as having a near-integrating response that is similar to the integrating response of batch reactors. Insight: The temperature and gas pressure control systems and PID tuning for batch reactors and liquid continuous reactors are similar. For pure batch operations where flows are sequenced and scheduled based on batch times and totals the concentration endpoint is achieved by accurate charging of reactants and by a batch cycle time long enough to insure complete reaction. The simultaneous feeding of reactants by flow loops whose setpoints can be manipulated termed fed-batch control or semi-continuous control opens up opportunities for the use of concentration control as the batch progresses. The control schemes for concentration control of excess reactants in continuous liquid reactors, by correction of the reactant ratio shown in this chapter can be extended to fed-batch reactors. However, for non-reversible reactions the concentration control of product, as the batch progresses requires the translation of the controlled variable because the product concentration response is only in one direction. A translation to a rate of change of product concentration; slope of the batch product concentration profile provides a controlled variable that can increase and decrease based on feedback action. The computation provides an indication of product formation rate and yield and enables a prediction of batch endpoint. Insight: Reactant concentration control systems for continuous liquid reactors can be extended to fed-batch reactors. However, product concentration control of non-reversible reactions requires a translation of the controlled variable to be the product concentration profile slope. Analyzers and inferential measurements can provide composition control to correct the ratio of reactants. The lead reactant flow controller setpoint is multiplied by a ratio factor and added to the composition controller output. The setpoint is used instead of the actual flow measurement to reduce noise from the process and sensor (e.g. pressure and velocity profile fluctuations in differential head meters) and cycling from valve backlash or stick-slip and to enable better timing of the flow adjustments for production rate changes. However, the valve must be able to reach the flow setpoint by proper valve sizing and a valve design that prevents the valve from getting stuck or plugged. The reactant flow controllers are tuned for the same closed loop time constant to eliminate transient unbalances in the stoichiometric ratio of reactants added. A feedforward summer is used because both the process gain and the process time constant have an inverse relationship to reactant flow. Since the maximum controller gain is proportional to the ratio of process time constant to process gain, the effect of flow cancels out in the PID tuning. At-line analyzers can be multiplexed with automated sample systems so one analyzer can service several reactors to help justify the cost. Inferential measurements can make the concentration measurement faster and smoother and less susceptible to missing or spurious results and noise. Inferential measurements except possibly for those from Coriolis meters must be periodically corrected by analyzer measurements ideally in the process but possibly in the lab. An enhanced PID is used in the composition loop to prevent excessive reset action and cycling from the large sample time and analyzer cycle time of at-line analyzers. Insight: Analyzers and inferential measurements can provide concentration control to correct the ratio of reactants to account for unknowns in the reactants and reaction chemistry. Insight: Inferential measurements can provide smoother and faster concentration measurements but do not eliminate the need for analyzers somewhere except for those measurements based on Coriolis meters. Well-mixed single phase liquid reactors can achieve tight composition control by tight temperature and composition control. Accurate measurement and tight ratio control of reactants is critical. For continuous reactors tight level control is also important for slow reactions with a setpoint adjusted based on production rate to keep the residence time constant.

The blog post, Succinct Field Instrumentation Guidance for the Automation Generalist Tips first appeared on the ControlGlobal.com Control Talk blog . The amount of information that the generalist needs to know is staggering. The knowledge needed is buried in a hundred thousand pages of publications and presentations that are oriented toward the specialist. Here I provide the essentials for the best field instrumentation system. The emphasis in this guidance is on performance rather than cost. I think we short change ourselves in terms of implications of supposed upfront cost savings in terms of long term cost of maintenance and process variability and reliability especially when it comes to measurements and valves. The biggest mistakes in my career have been the result of attempts to save on hardware cost as exemplified by the case history noted in the last blog where valve positioners were omitted or replaced by boosters on fast loops. I don’t have time to reference all my articles, blogs, books, and columns that provide more details. I suggest you peruse the subjects in my ISA Interchange Insights , Control Talk Blogs , Control Magazine articles and Control Talk columns , InTech articles and ISA books , Momentum Press books , Emerson Deminars , and MYNAH Seminars . (1) Use radar sensors for maximum level measurement accuracy for inventory and material balance control (e.g., reactor residence time and distillate level control). (2) Use smart transmitters with separate compartments for wiring and electronics for maximum reliability and maintainability. (3) Use integral mounted transmitters or close coupled transmitters to eliminate sensor wires, filled capillary systems, and impulse lines if the sensor connection is safely accessible, surface temperature is less than 50 degrees C, and connection vibration is negligible. Avoid like the plague the direct wiring of temperature sensors to a DCS or PLC input card. (4) If impulse lines must be used, minimize length and ensure the fill is a single phase of constant density. (5) If capillary systems must be used and there is adequate pressure range and protection from excessive vacuums, make the seals and capillary lengths the same, minimize the length of the filled line for best response time, maximize the size of the seal for best sensitivity and ensure the lines are secured (do not move in wind) and are protected from the sun and rain. (6) Use tip sensitive and vibration durable 4-wire Resistance Temperature Detectors (RTDs) for maximum sensitivity and accuracy and minimum drift for process temperatures less than 500 degrees Centigrade. (7) Use ungrounded sheathed premium thermocouples (TCs) of appropriate type for process temperatures greater than 500 degrees Centigrade. (8) Use spring loaded compression fitting to ensure RTD and TC tip touches thermowell bottom with minimal annular clearance to minimize response time from air acting as insulator. (9) Use tapered thermowell to reduce thermal conduction error and vibration failure. (10) Specify and purchase the TC or RTD assembled in thermowell with integral mounted transmitter via a pipe union as an assembly from the supplier’s factory with a transmitter calibration to match sensor nonlinearity. (11) Install the thermowell in an elbow pointed into the flow at the pipe center line to minimize response time and error from cross sectional variation in temperature profile. (12) Use shrouded spherical high temperature pH glass and a replaceable reference junction for best accuracy, reliability, and maintainability. (13) Install pH electrode pointing 45 degrees down with tip in the centerline of a pipe whose velocity is 5 to 10 fps to minimize coatings and response time. (14) Use sliding stem valve for smaller lines and non-abrasive and non-plugging applications from a manufacturer whose heritage is control valves and not piping valves or on-off valves. (15) Use v-ball or contoured butterfly valves for larger lines and applications best served by a rotary valve from a manufacturer whose heritage is control valves and not piping valves or on-off valves. (16) Use a diaphragm actuator instead of a piston actuator unless the valve size or pressure drop cannot be handled by even the higher air pressure diaphragm actuators available today. (17) Always use a tuned digital positioner. Add a booster to the positioner output(s) with an integral bypass valve slightly opened if the valve needs to be faster. (18) For tight shutoff use an isolation valve from a piping or an on-off valve manufacturer that is coordinated to automatically work in conjunction with the control valve for throttling. (19) Use equal percentage trim if the valve pressure drop is less than 50% of the total system pressure drop to provide better rangeability and linearity. Note that surge control valves, pH reagent valves, and gas pressure vent valves tend to have a valve to system pressure drop ratio approaching one and benefit greatly from a linear valve trim. (20) Design the valve drop at maximum flow to be greater than 20% of the total system drop to prevent excessive loss of rangeability. (21) Use a pulse width modulated variable frequency drive (VFD) with high resolution input cards, torque to speed control system in the drive, minimum deadband and speed rate limiting, noise resistant cables, ventilated and if necessary fan cooled motors with higher service factor if energy savings and a more precise and faster response is justifiable provided the static head is less than 10% of the total system pressure drop. (22) Use Coriolis meters for smaller line sizes to maximize accuracy and rangeability and to minimize drift and maintenance. (23) Use magmeters for liquid flow in moderately large line sizes. (24) Use vortex meters in moderately large line sizes where magmeters cannot be used with straight run requirements similar to orifice meters (25) Use flow nozzles in very large line sizes to minimize permanent pressure loss and change in meter coefficient with wear and piping conditions with dual smart transmitters to maximize rangeability. Make sure straight run requirement is based on actual piping installation. (26) For maximum reliability and on-stream time, use three sensors with transmitters and middle signal selection. This setup is almost essential for any pH measurement used for process control.

The post Temperature Sensor Installation for Best Response and Accuracy first appeared on the ISA Interchange blog site. The following insights are part of an occasional series authored by Greg McMillan , industry consultant, author of numerous process control books and a retired Senior Fellow from Monsanto. The installation of the sensor can introduce errors, noise, and dynamics causing poor measurement and control loop performance. Here we look at best practices to get the most out of the inherent capability of the sensor. The next post will provide guidance on the communication of the sensor signal to the control room to provide the best total installation. Thermowell Length To minimize conduction error (error from heat loss along the sensor sheath or thermowell wall from tip to flange or coupling), the immersion length should be at least 10 times the diameter of the thermowell or sensor sheath for a bare element. Thus, for a thermowell with a 1 inch (2.54 cm) outside diameter, the immersion length should be 10 inches (25.4 cm). For a bare element with a ¼ inch (6.35 mm) outside diameter sensor sheath, the immersion length should be at least 2.5 inches (63.5 mm). This is just a rule of thumb. Computer programs can compute the error and do a fatigue analysis for various immersion lengths and process conditions. For high velocity stream and bare element installations, it is important to do a fatigue analysis because the potential for failure from vibration increases with immersion length. The choice of thermowell length, location, and construction determines whether the temperature measurement is representative of the process, how much process noise is seen, how much delay and error is introduced, and the potential failure rate. This post provides some general guidance. This post provides some general guidance. For more details including the equations to predict eight sources of measurement error see Greg McMillan’s ISA book Advanced Temperature Measurement and Control, Second Edition Thermowell Location The process temperature will vary with process fluid location in a vessel or pipe due to imperfect mixing and wall effects. For highly viscous fluids such as polymers and melts flowing in pipes and extruders, the fluid temperature near the wall can be significantly different than at the centerline (e.g., 10 to 30°C; 50 to 86°F). Often the pipelines for specialty polymers are less than 4 inches (101.6 mm) in diameter, presenting a problem forgetting sufficient immersion length and a centerline temperature measurement. The best way to get a representative centerline measurement is by inserting the thermowell in an elbow facing into the flow (position 1 in the figure below). If the thermowell is facing away from the flow, swirling and separation from the elbow as can create a noisier and less representative measurement (position 2 in figure). An angled insertion (position 3 in figure) can increase the immersion length over a perpendicular insertion (position 4 in figure) but the insertion lengths shown for both are too short unless the tip extends past the centerline. A swaged or stepped thermowell can reduce the immersion length requirement by reducing the diameter near the tip. The distance of the thermowell in a pipeline from a heat exchanger, static mixer, or desuperheater outlet should be optimized to reduce the transportation delay but minimize noise from poor mixing or two phase flow. Generally 25 pipe diameters are sufficient to ensure adequate mixing from turbulence if there is a single phase, turbulent flow, and no great differences in the viscosity of streams being combined. Two phases exist for desuperheaters, split ranged transitions from cooling water to steam in jackets, the use of lime ammonia as a reagent for pH control due to flashing and whenever slurries are involved. The transportation delay will increase with distance adding more dead time to the loop. Consequently, there is a compromise between getting enough mixing to achieve a representative low noise measurement and creating too much additional dead time. In general, the transportation delay should be less than 10% of the PID reset time setting. Insight : Generally a distance of 25 pipe diameters between the equipment outlet and the temperature sensor is sufficient to provide a relatively uniform temperature profile of a single phase fluid. The presence of different phases (e.g. bubbles or solids in liquids and droplets in steam) and high viscosity fluids will require longer distances. For desuperheaters, the distance from the outlet to the thermowell depends upon the performance of the desuperheater, process conditions, and the steam velocity. To give a feel for the situation there are some simple rules of thumb for the length of piping from the desuperheater to the first elbow known as straight piping length (SPL) and the total piping length from the desuperheater outlet to the sensor known as sensor total length (TSL). Actual SPL and TSL values depend on the quantity of water required with respect to the steam flow rate, the temperature differential between water and steam, the water temperature, pipe diameter, steam velocity, model, type, etc. and are computed by software programs. SPL (feet) = Inlet steam velocity (ft/s) x 0.1 (seconds residence time) SPL (m) = Inlet steam velocity (m/s) x 0.1 (seconds residence time) TSL (feet) = Inlet steam velocity (ft/s) x 0.2 (seconds residence time) TSL (m) = Inlet steam velocity (m/s) x 0.2 (seconds residence time) Typical values for the inlet steam velocity, upstream of the desuperheater range from 25–350 ft/s (7.6 to 107 m/sec). Below 25 ft/s there is not enough motive force to keep the water suspended in the steam flow. Water tends to fall out and run down the pipe to a drain. When this happens the water no longer cools the steam and the system thinks it needs to add more water, which compounds the problem. Problems can also include pipe wall erosion and high thermal stress gradients in the pipe wall (i.e., a hot top and cold bottom, which can crack welds or warp the pipe to an egg-shaped cross-section). Current technology has an inlet velocity limitation of 350 ft/s (107 m/sec). Velocities higher than 350 ft/s cause the desuperheater to vibrate and damage the unit to the point where it breaks apart. Thermowell Construction The stem of a thermowell is the part that is inserted into the process stream. Stems can be tapered, straight, or stepped. The performance of a thermowell varies with its stem design. In general, a tapered or stepped stem provides a faster response, creates less pressure drop, and is less susceptible to conduction error and vibration failure. If the thicknesses of the thermowell walls and the fit of the sensing element are identical, thermowells with straight stems have the slowest time response because they possess the most material at the tip (largest diameter). Thermowells with stepped stems have the fastest time response because they possess the least material at the tip (smallest diameter). A small diameter also results in the least amount of drag force. Thermowells with stepped stems also provide the maximum separation between the wake frequency (vortex shedding) and the natural frequency (oscillation rate determined by the properties of the thermowell itself). If the wake frequency is 80% or more of the thermowell natural frequency, resonance and probably damage can occur. Generally, thermowells with tapered stems are slightly more expensive as a result of a more complicated manufacturing process. Insight : Swaged, stepped, and tapered thermowells offer a faster response, lower pressure drop, and less possibility of vibration damage from resonance with wake frequencies. The tip of the sensor must touch the bottom of the thermowell. Spring loaded sensor designs help ensure this is the case despite different installation practices and orientation. The fit of the sensor should be as tight as possible to reduce the annular clearance since air acts as insulator. The sensor lag can increase by an order of magnitude for a sloppy fit. For liquid systems, the additional lag effectively becomes an additional equivalent dead time in the measurement. Insight : The tip of the temperature sensor must touch the bottom of the thermowell and the fit must be tight to prevent introducing a large sensor lag due to the low thermal conductivity of air. Take advantage of general guidelines on thermowell insertion length, location, construction, and fit to make sure the sensor is seeing the actual process temperature with a low probability of vibration failure and minimal noise, delay and lag.

This post by Greg McMillan, Secondary Flow Loop and Valve Positioner Tips first appeared in the Control Talk blog on ControlGlobal.com . A lot of time and money can be spent deciding which valves need positioners and which flows need measurement. We tend to look at short term costs such as hardware and not the cost of troubleshooting and the implications as to plant performance. Most of my big mistakes were the result of trying to save upfront costs. Here we look at common misconceptions as to whether flow and valve position control should be used. Most of the controversy centers on the violation of the cascade rule where a lower loop is not at least 5 times faster than a higher loop. The rule “valve positioners should not be used on fast loops” still haunts us. While here we focus on the manipulation of flow which is the most common manipulated process input, the next blog will offer more general and concise guidance but with the same theme of not short changing yourself or the automation system. In my book all feedback loops manipulating plant flows should have a valve positioner and a flow measurement. The use of external reset feedback and the positive feedback implementation of the integral mode will prevent an upper PID output from changing faster than a lower PID (e.g., secondary flow) can respond and will also prevent this lower PID output from changing faster than the control valve can respond (given fast readback). Thus, the violation of the cascade rule does not cause excessive oscillation. I have never seen a case where a positioner should not be used. The high gain smart digital valve positioner reduces the amplitude of deadband from backlash, eliminates the confusing offset during manual operation, deals with the integrating response of actuator pressure, prevents positive feedback from a booster and diaphragm actuator (potentially dangerous situation), provides ability to tune for different types of actuators, and gives readback of actual valve position for visibility and external reset feedback. The addition of a volume booster on the output of the positioner can reduce valve dead time to less than 0.1 second and increase slewing rate to 100% per sec. If this is not fast enough, then a pulse width modulated variable frequency drive with speed to torque cascade control in the drive. There are many design considerations with a VFD not commonly recognized. Most overlooked is the severe loss in turndown when the static pressure approaches the total head. I am an advocate of flow loops as well. The flow loop compensates for pressure disturbances, provides better regulation of process stoichiometry leading to better composition control by the use of lower flow loops and coordinated flow ratio control, and more accurate feedforward control to preemptively correct for feed and utility disturbances (e.g. flow and temperature changes). For startup, the cascade control system can be operated with just the lower loop in service (e.g. flow ratio control) until operating conditions are reached. However, a secondary flow loop may cause the PID output response to be slower than desired for gas and liquid pressure control. Here the PID output should go directly to a fast valve (booster on the output of a positioner) or a VFD. I would still have a flow measurement for better process knowledge. Flow measurements enable closure of material and energy balances, tracking down disturbances, better inputs for data analytics and verification of process simulations, online process metrics, and an accurate relative gain analysis (RGA) for interactions. What I learned the Hard Way In my 46 years of experience some of which was in E & I construction, I have never seen a case where a positioner should not have been installed. My recommendation for the last 40 years is any loop with a control valve should have a positioner. I have had several cases where the omission of a positioner has caused serious and potentially dangerous situations. However, not all positioners are created equal. Some valve designs can cause misleading position feedback and some positioner designs can have extremely poor threshold sensitivity (poor response to small changes). Here is my story. I started out in E & I construction in 1969. In the four plant installations and startups over the next 2 years all the valves had positioners but the control valves from a piping valve manufacturer used spool type positioners with a pulley system, on-off piston actuators and quarter turn on-off plug valves. The control was horrible with these piping valves. The control valves supplied by a control valve manufacturer all were originally designed for throttling and had a high gain sensitive pneumatic relay positioner. The pneumatic positioner would lose its calibration but the valves responded to small changes and the calibration offset was compensated by the process PID with an output that could go below 0% to make sure the valve was shut despite the calibration offset. About 5 years later as lead engineer for the world’s largest acrylonitrile plant with latest electronic analog controllers, the engineering contractor said he saved me a lot of money by not putting positioners on fast loops. Not ever having seen a valve try to do its job without a positioner, I agreed. During startup we found some controller outputs were 40% and the valve had still not opened. There was very little correlation between PID output and valve position. This was particularly confusing during startup and manual positioning of the valve. We ended up putting positioners on all valves during startup. About 5 years later as I started on the path of improving compressor surge control systems I realized I needed to use a booster to make the valve faster. I read a paper by a prominent technologist that unequivocally concluded after a theoretical study using Nyquist plots that a booster instead of a positioner should be used on fast loops. The instrument technician objected but agreed to replace the surge valve positioner with a booster. That night the compressor shutdown and the cause was found to be the surge valve slamming shut despite the PID output asking for the valve to be opened. When I went up to the valve, the technician showed me how he could grab the stem of the 24 inch valve and move it at will. He then showed me how he could not do this to another 18 inch valve that still had a positioner. I had the positioner reinstalled and put the booster on the outlet of the positioner. I adjusted the booster bypass valve to eliminate high frequency oscillations from the positioner and booster in series. All of the surge valves ended up with this configuration and control was great. When I ordered control valves for a phosphorous furnace pressure loop that could ramp off scale in a couple of seconds, I went with one of the best distributed control systems with a special option for a 0.03 second execution time making it about as fast as an electronic analog controller. I got a special transmitter with a minimum damping setting of 0.05 sec (0.05 sec time constant). I put a performance requirement that the valve stroking time was less than a second that would be witnessed by me by a test at the control valve suppliers factory. When I got there I saw boosters but no positioners on the 18 inch valves. I walked up to one valve and showed the guy how I could grab the shaft and move the valve. An old timer then arrived and gave out a 1958 article by C. Mamzic “Improving the Dynamics of Pneumatic Positioners” in ISA Journal 5, No.8 that said you should put the booster on the outlet of the positioner and slightly open the bypass valve. We did this and the valves worked fine and the combination of the fast controller fast transmitters and fast valves reduced the number of electrode seal blows by over 90%. Since then this has been my strategy on numerous fast loops. The high threshold sensitivity of a diaphragm actuator and booster’s outlet port is desirable but creates a positive feedback situation enabling a person to stroke a valve or a pressure unbalance on a butterfly disc cause the valve to slam shut. Boosters have poor input port sensitivity and significant deadband, which is only an issue if not used in conjunction with a positioner. My other main special area of expertise was pH control. Here I found the stick-slip and backlash from control valves was causing oscillations outside of the allowable control region in systems with relatively steep titration curves. In fact the main limit as how much you could reduce the number of stages of neutralization and the size of the volumes depending upon the precision of the reagent valves. These pH systems were extremely good at showing valve precision. I became sensitive to valve sensitivity. For a huge grass roots facility in Asia, I was asked to provide guidance on what valves could have positioners omitted to save money. A Fellow and recognized expert on valves advocated only putting positioners on slow critical loops. I said all loops should have positioners. I have never seen where the violation of the cascade rule (secondary loop not sufficiently faster than primary loop) by putting a positioner on a fast loop has caused a problem. I think the way we tune these PID loops with more integral than gain action and the speed of a proportional only analog positioner and the flexibility of tuning in a modern digital positioner eliminates the problem. In subsequent years I was plagued over and over again with piping valves posing as throttling valves and attempts to use on-off valves needed for tight shutoff as throttling valves. The positioner feedback of actuator shaft of excellent positioners said everything was good in that the shaft responded to 0.5% changes in signal. Testing the valves in the shop with travel gauges added to the disc or ball showed that the disc or ball did not move until the changes in signal were greater than 8%. The positioner was fooled into thinking everything was OK by a misleading position feedback. The shaft moved but the disc or ball did not, a symptom of shaft windup aggravated by poor shaft to stem and stem to disc or ball connection design. The high gain digital valve positioner reduces the amplitude of deadband from backlash, deals with the integrating response of actuator pressure, prevents positive feedback from a booster and diaphragm actuator combination, provides ability to tune for different types of actuators, provides readback of actual valve position for visibility and external reset feedback to prevent a PID output from changing faster than a valve can respond and enables a whole spectrum of valve performance diagnostics. The misconception still exists. An Automation Hall of Fame member in the last 6 months publically made the recommendation not to use positioners on fast loops when asked. If a positioner is a problem on a fast loop, you probably should not be using a pneumatic actuator or a digital controller execution time greater than 0.03 seconds. For such fast unusual applications you can go to speed control of the prime mover with a fan cooled motor, special cables, a pulse width modulated variable frequency drive with local speed to torque control, a high resolution input card, and zero rate limiting or deadband in drive setup. Even with all of these best practices, a VFD only has sufficient turndown in applications where the static head is small compared to pressure drop from system resistance, a fact not well recognized. Benefits of Cascade Control The process control benefits of cascade loops are numerous. Here is a list of the ones that come to mind. Lower loop isolates nonlinearities (e.g. valve and process) and stream and utility disturbances (e.g. pressure and temperature) from the upper loop. The lower loop encloses secondary time constant that converts the secondary time constant that would have been a detrimental term in a single loop to being beneficial term as the largest time constant in a lower loop. The cascade upper loop ultimate period is smaller than the original single loop enabling a faster upper loop and better rejection of disturbances in the upper loop. The peak error in the upper loop for a lower loop disturbance can be reduced to be as small as 12% for self-regulating, 2% for integrating, and 1% for runaway of the peak error for a single loop for a lower to upper dead time ratio of 0.6. For lower dead time ratios the improvement would be even more impressive. The best reduction in peak error for a given dead time ratio is achieved for time constant ratio approaching one where the secondary time constant was as large as the primary time constant. This latter relationship is often not recognized because it is counter intuitive and contradicts the cascade rule. The user must realize the cascade rule pertains to the closed loop response and not the open loop response. Since the ultimate period and lambda is a factor of the total loop dead time, the loop dead time sets the limits on the closed loop response. Better regulation of process stoichiometry leading to better composition control by the use of lower flow loops and coordinated flow ratio control. More accurate feedforward control to preemptively correct for feed and utility disturbances (e.g. flow and temperature changes). For startup, the cascade control system can be operated with just the lower loop in service (e.g. flow ratio control) until operating conditions are reached (e.g. distillation columns). The feedforward should be configured to be active with the upper loop in manual, which means the lower loop stays in cascade mode. For an upper loop measurement (e.g. analyzer) failure, the cascade control system can be operated with just the lower loop in service (e.g. flow ratio control) until the measurement is fixed. Benefits of Flow Measurements The process knowledge benefits of cascade control are not discussed much in the literature but can be just as important and more extensive. The improvement in the recognition and identification of relationships and the creation of models can translate to more intelligent setpoints and operator and process engineers understanding of confusing situations. An increase in process knowledge can be far reaching. Since nearly all manipulated process inputs are flows, the addition of flow measurements for lower flow control loops offers many advantages. A control loop transfers variability in the process variable to the manipulated variable (e.g. flow). If the upper loop (e.g. level, composition, pH, or temperature) is tightly controlled, nearly all of the variability caused by changes in the process is seen in the manipulated flow rather than the process variable. The process variable in these loops stay right at setpoint. The process knowledge is in the size and pattern of changes in the manipulated flows. All process simulations need to be compared to the plant and corrected. Process simulations have a difficult time getting the pressure drops and hence the installed flow characteristics right because of the incredible amount of detail on the geometry and characteristics of piping and valve systems needed besides the changes in interior surfaces (e.g. roughness and coatings). The control valve positions in a simulation (e.g. virtual plant) will not match up with the actual plant. The only way to improve simulations that are both doing a good job of control at setpoint is to match up the flows. The addition of flow measurements and flow control in the actual plant enables more accurate process simulations and hence process analysis and improvements. For more on these benefits see the Jan/Feb 2010 InTech article “ Advances in flow and level measurements enhance process knowledge, control ” Here is a summary of the benefits of flow measurements beyond cascade control. More linear, accurate, and representative inputs for data analytics and neural networks particularly when loops are tightly controlled. Better analysis of whether correlations represent causes and effects or coincident occurrences. More accurate process simulations for better process analysis and improvements. Greater recognition of the source and path of disturbances and abnormal operation to reduce the consequences and frequency of disturbances and failures. More accurate online process metrics for process efficiency and capacity. Adaptation of parameters in a virtual plant synchronized with an actual plant. More effective interaction analysis by the use of manipulated flows instead of PID outputs in the computation of relative gains for a RGA Much of what I have learned is shared in my Control Talk Blogs and Columns on the Control magazine website and in my ISA books 101 Tips for a Successful Automation Career and Advanced pH Measurement and Control and my Momentum Press books Axial and Centrifugal Compressor Control and Tuning and Control Loop Performance – 4 th Edition .

The post Temperature Communication for Best Accuracy and Reliability, and Least Noise first appeared on the ISA Interchange blog site. The following insights are part of an occasional series authored by Greg McMillan , industry consultant, author of numerous process control books and a retired Senior Fellow from Monsanto. Communication from the temperature sensor to the control room should be designed to provide the greatest accessibility, flexibility, reliability, and accuracy. What was traditionally thought to be the least expensive solution has severe ramifications and hidden costs. Here we look at how the advent of smart transmitters and wireless communication has opened the door to better communication and a lower installed and life cycle cost and process performance improvements. The most common process measurement rivaling or exceeding flow in the number of sensors in a process plant is temperature. Distillation columns may have 4 or more temperature sensors to find the best tray for temperature control and fluidized bed reactors may select the highest of 10 or more temperature sensors to deal with hot spots. In the past the most inexpensive hardware was used for the communication of the temperature to the control room not realizing the impact on measurement accuracy, noise, and reliability. In fact early Distributed Control Systems (DCS) were partly justified on the basis of eliminating the cost of temperature transmitters by wiring thermocouples (TCs) directly to a DCS TC input card that could handle 16 or 32 inputs. The results were disastrous in terms of introducing steps (analog to digital convertor noise) of 0.35 o F due to a 12 bit wide range input card and offset of 10 o F or more due to differences in sensors and lead wire errors. Noise from Electromagnetic Interference was prevalent. Derivative action could no longer be used and the PID gain had to be decreased. Column and reactor temperature errors and consequently composition errors caused poor process performance and off-spec product. In many cases, temperature setpoints were adjusted to compensate for the errors and a signal filter was added to reduce noise that increased the total loop dead time. Technically advanced temperature communication can decrease installation costs and reduce errors and noise. The advent of the smart transmitter makes mounting the transmitter directly on the thermowell more attractive. The combination of integral mounting with the wireless option creates an incredible opportunity for process performance improvements. This post provides some general guidance. For more details including the equations to predict eight sources of measurement error see Greg McMillan’s ISA book Advanced Temperature Measurement and Control, Second Edition Communication To be useful for control, safety, or monitoring applications, a temperature measurement signal must be communicated from the point of measurement to the control system of the process. The two most common ways are: Transmitter – the sensor is wired a short distance to the transmitter or connected directly to the transmitter, where its signal is converted to a digital, 4-20 mA, or wireless signal. The converted signal output is then communicated to the control system through transmitter wire or a wireless network.   Wired direct – the sensor’s lead wires are wired the entire distance to the control system. No signal conversion takes place along the route. The benefits of using a temperature transmitter over wiring directly to thermocouple and RTD input cards of control system are: A more robust signal is delivered – the 4–20 mA or digital signal output from the transmitter is much more robust than a sensor signal that is wired direct. Noise interference has less impact on 4–20 mA or digital signals.   Measurement accuracy is optimal – Transmitters offer improved measurement accuracy over wiring direct. For example, sensors can be matched to transmitters (transmitter sensor matching), which improves the accuracy of the temperature measurement. The temperature span can be narrowed to match the process operating range (significant for older DCS with 12-bit input cards).   Lower Installation and Maintenance Cost – Transmitter installation is often less expensive than wiring a sensor direct because of savings from cabling costs and installation (sensor wire, especially TC wire, is relatively expensive). Also, a robust signal and accurate measurements produce time and money savings through increased functionality and diagnostic capabilities of the transmitter.   Better Process Performance – The greater accuracy and reliability and lower noise reduces process variability and increases on-stream time. Insight : The use of transmitters instead of TC or RTD input cards is highly recommended to greatly improve accuracy and maintainability, by matching the calibration and nonlinearity compensation to sensor, narrowing the span, reducing noise, and offering diagnostics. The use of digital communications allows the additional flexibility of using a single transmitter to make more than one temperature measurement and communicate these back to the control system. There are temperature devices designed to specifically take advantage of this capability, providing the ability to measure four, or eight, or potentially more individual temperatures. The most common communication techniques are the HART® (including WirelessHART™), FOUNDATION™ Fieldbus and Profibus PA standard protocols. The reliability, security, and ease of setup of WirelessHART (Highway Addressable Remote Transducer) networks combined with increased battery life from new communication rules and PID enhancements have made wireless communication an excellent option. The temperature changes in most processes are quite slow, the refresh time can be set longer than for other types of loops, extending battery life. Also, the noise amplitude and period in temperature loops is usually quite small compared to other loops unless there are two phases (e.g., liquid and gas) or poor mixing (e.g., poor uniformity—increased variability due to insufficient agitation), decreasing the number of exception updates triggered by noise, which also extends battery life. Field-Mount Transmitters Field-mount transmitters are the most rugged of all transmitter styles. Their robust housings protect against corrosion and humidity. Some field mount transmitters house the electronics in dual-compartment housings, which completely isolates them from the effects of humidity. Dual-compartment transmitters are the best design for use in harsh environments. Field-mount transmitters can be integrally or remotely mounted. In integral mounting, the transmitter is installed directly on the thermowell by a threaded pipe nipple and a pipe union fitting to allow easy removal of the sensor. Since today’s smart transmitters are extremely reliable and have extremely low drift rates reducing calibration intervals to more than 5 years, diagnostics can be viewed remotely, and calibration can be done remotely , the need for a transmitter to be at ground level has greatly diminished. The integral mounting of transmitters reduces installation costs and eliminates errors and noise introduced by lead wires and additional terminations. Integral mount – the transmitter is threaded onto the sensor directly (mounts directly to U.S. style [1/2-inch NPT] sensor preferably with a pipe union fitting). Remote mount – the transmitter is mounted on a pipe stand or other support near the sensor. Remote mount is preferred when the measurement point is inaccessible or when the process environment is too harsh for the transmitter to be installed directly on top of the sensor. Insight : The integral mounting of smart transmitters where permitted by accessibility and temperature, improves measurement accuracy and reliability. The use of integral mounting and wireless transmitters provides flexibility and portability for monitoring unit operation efficiency and finding the most representative and sensitive measurement location with the least process dead time. Where ever there is a pipe connection or equipment nozzle and a line of sight to the Gateway device’s access point or nearby wireless transmitter for device hopping, the sensor with the wireless transmitter can be installed on a test basis and the benefits explored and quantified. Improvements in data analysis, equipment monitoring, and unit operation control can be prototyped and the “before” and “after” cases documented. Insight : The integral mounting of a wireless transmitter enables flexibility and portability for online process and equipment performance metrics and optimization of measurement location. Head-Mount Transmitters Head-mount transmitters are small, puck-shaped transmitters. They are typically housed in a protective enclosure – a connection head for direct mounting or a junction box for remote mounting. Rail-Mount Transmitters Rail-mount transmitters are designed to be attached to a DIN-rail (G-rail or top-hat rail) or directly screwed onto a wall. Rail-mount transmitters are also designed for compact mounting, which allows for a number of transmitters to be mounted very closely together.  Wiring Direct As mentioned, wiring direct refers to wiring the sensor’s lead wires back to the control system. Because the sensor’s lead wire (and original signal) is traveling the entire distance from the point of measurement to the control system, care must be taken to avoid two key problems: Noise – TCs are especially sensitive to noise interference and extension wires must be routed around such sources as generators and motors. Heat sources – a large change in the ambient temperature can affect the sensor’s signal as it travels to the control system. Lead wire resistance – the total resistance of RTD lead wire and the difference in resistance of RTD lead wires will affect the RTD accuracy. A 3-wire transmitter input can compensate for total lead wire resistance. A 4-wire transmitter input can compensate for differences in lead wire resistances. Use field transmitters for all temperature measurements important for process analysis and control. If not prohibited by accessibility or temperature, the transmitter should be integral mounted on the thermowell preferably with a pipe union to improve removability of the sensor. Consider the possibility of finding the best measurement location and exploring and prototyping the potential benefits for additional temperature measurements for process performance improvement by the use of portable wireless integral mounted transmitters.

This post, Why Tuning Tests are Not Repeatable Tips , by Greg McMillan first appeared in the Control Talk blog on ControlGlobal.com . Each process test will typically give a different result in the process dynamics identified and consequential tuning settings calculated. Here we look at the sources of this lack of repeatability, the implications, and what can be done to improve tuning tests. Valve deadband from backlash and resolution from stiction will reduce the actual step change in valve position for a given step change in PID output. This will lead to the identification of a smaller than actual valve gain and the calculation of a larger than desirable PID gain. The effect is not repeatable because the valve dead band and resolution vary with stroke and time and the effect depends upon the last response. The effect of deadband is not seen until the direction is reversed. The degree of the effect of resolution depends upon whether the step is an exact multiple of the resolution limit and whether the valve position just changed (slip) or is stationary (stick). Most valve installed characteristics are nonlinear and change as system resistances and static head change. The valve gain (one of the factors of the open loop gain) is the change in flow divided by the change in valve position. Thus, the gain is the slope of the line connecting the two points on the valve characteristic curve. Only for small changes in the signal is the slope of this line the slope of curve. However, we do not want the change in signal to be small for many reasons. Consequently, the slope of the line will change with step size even if the starting point is the same unless the installed flow characteristic is linear. The process gain (other major factor of the open gain) is nonlinear for most composition, pH, and temperature loops. The process gain for all of these loops is fundamentally inversely proportional to the production rate. The other part of the process gain is best found on a plot of composition, pH, and temperature versus the ratio of the manipulated flow to the feed flow. Just as with the valve, the nonlinearity of the curve seen depends upon the size of the change in manipulated flow. The process gain is the slope of the line connecting the starting point and end point of a step test and is only the slope of the process curve for small changes in manipulated flow. The slope of a temperature curve for distillation depends upon tray and operating conditions besides ratio. The slope of the pH curve (titration curve) depends upon feed and reagent concentrations besides ratio. The sensor time constant of thermowells and electrodes depends upon the velocity, coatings, process conditions, and the direction and size of the process change. pH electrodes are the most sensitive of the sensors to these changes besides the age and condition of the glass surface. Measurement time constants become the secondary time constant or dead time in liquid volumes. For gas plug flow volumes where feeds are manipulated for temperature or composition control, the measurement time constant may become the primary time constant since the process time constant is so small. This is confusing to say the least because a larger sensor time constant leads to more aggressive tuning and a smoother looking process trend as extensively discussed in the 12/2/2014 blog Measurement Attenuation and Deception Tips. Valves have a second order velocity limited exponential response that depends upon the stroking time and dynamics and tuning of the positioner. The apparent size of the valve time constant will depend upon the size and direction of the change in signal. The process time constant is generally inversely proportional to production rate for well mixed volumes. The process time constant also changes with the direction of the change. Decreases in temperature are generally slower than increases in temperature due to the smaller temperature difference acting as the driving force especially for ambient cooling. Decreases in vessel pressure are generally slower than increases in pressure due to smaller pressure drop for a vent flow especially vessel pressures close to atmospheric. Decreases in reactant concentration are slower than increases in reactant concentration where reaction rates are relatively slow. Decreases in substrate concentration (e.g. glucose) are much slower than increases due to the incredibly slow cell growth rates and slow product formation rates in bioreactors. The process dead time is generally inversely proportional to production rate but is also affected by transportation delays. The injection delay of small reagents flows in dip tubes is horrendously large and variable due to the dip tube transportation delay. Poor mixing patterns and short circuiting of manipulated flows to discharge flows causes erratic measurement dead time in vessels and measurement noise. The measurement dead time due to discontinuous signals depends upon when the disturbance or output change arrived in the discontinuous signal processing time interval. On the average, the disturbance arrives in the middle of the time interval. Consequently the dead time from digital and wireless devices is ½ the update time (e.g. default update rate) assuming negligible latency. Changes larger than the threshold sensitivity (e.g., trigger level) can cause an earlier update. The average dead time from analyzers is the sample transportation delay plus 1 ½ the analyzer cycle time assuming the analyzer result is available at the end of the analysis cycle time. There are many sources of noise . Most are associated with the changes in the axial and radial concentration, temperature, and velocity as determined by sensor type, location and installation. The minimum straight length requirement for differential head meters and vortex meters often does not take into account the actual piping system details that necessitate longer lengths. Sensors inserted in a pipeline (e.g., thermowells, electrodes, and insertion flow meters) do not take into account the radial profile and the changes due to less than ideal mixing. The more traditionally recognized source of noise is electromagnetic interference (EMI). Variable frequency drives (VFD) were the source of many EMI problems when less expensive drives and installations were used and strict practices on the type of VFD cables and the isolation from signal cables were not followed. Poor shielding and grounding practices will lead to many EMI problems. Disturbances are a big source of changes in the test result. If there were no disturbances, you would not need feedback control. The only thing for certain is the process conditions on the Process Flow Diagram (PFD) are not the conditions in the plant at any given time. The test time should be made as short as possible to reduce the probability of an occurrence of a disruptive disturbance. This is particularly important for distillation columns that may take one or more shifts to reach a steady state. The tuning test time can be minimized by using a large step size and using a tuning test and method that looks at the dead time and ramp rate of an excursion in the right direction for processes that take a long time to reach a steady state or do not have a steady state in a reasonable time frame (e.g., near-integrating and true integrating processes). Don’t get discouraged. Simply get realistic. Do not expect results to be repeatable to more than 2 significant figures even in the best scenarios. Use best practices to deal with the consequences of test variations. Four or more tuning tests should be done at each setpoint and plant production rate. Day and night tests are advisable if ambient conditions or shift operations make a difference. The tests should use steps in both direction. The step size should be large enough to make the effect of measurement noise band, valve dead band and valve resolution negligible and to reduce the test time. The operator should not be making feed changes when the tests are being done. Maintenance work should not be in progress. Process engineers should not be asking for setpoint changes. A quiet time (mostly likely a weekend) is best when the control room is not crowded. The step change should be made in the PID output rather than the PID setpoint so that the effects of PID structure and setpoint lead-lags used to improve setpoint response do not obscure the identification of the process dynamics. This can be done by injecting a step change by software automatically adding an increment or decrement or momentarily putting the PID in remote output or manual mode. The test results with the maximum open loop gain, minimum primary time constant, maximum secondary time constant, and maximum dead time should be used. This rule can be moderated to the degree that the maximums and minimums are physically not possible to be coincident and nonlinearities are compensated by signal characterization, adaptive control, and scheduling of tuning. Of course, tuning should not be a cover up. If a test is not repeatable due to poor equipment, piping, or automation system design and installation, the problem should be fixed not only for better tuning but for greater knowledge and control of the process. After all, we want our processes to be as repeatable as possible.

The post What are the Inherent Differences in the Performance of Pure Batch, Fed-Batch and Continuous Reactors? first appeared on the ISA Interchange blog site. The following insights are part of an occasional series authored by Greg McMillan , industry consultant, author of numerous process control books and a retired Senior Fellow from Monsanto. In pure batch operations, changes in feed conditions, disturbances, or demand show up as changes in the profile of key process variables. For a bioreactor, variability in the seed culture, indeterminacy of cell response and changes in the demand for air, substrate (e.g. glucose), and reagent cause the variation in the batch profile for dissolved oxygen and substrate concentration and pH. Process engineers can be ingenious at changing the timing and flow rate of air, substrate, and reagent via sequential scheduling. However process knowledge is far from perfect and unknowns change the profiles of the key process variables. Feedback control is needed because the process changes from minute to minute and the process response is not totally repeatable besides not being entirely predictable. Even model predictive control relies on feedback control to bias the predicted future response trajectory. Insight : The sequential scheduling of batch feeds in pure batch operations presumes complete knowledge and a lack of variability in raw materials and batch behavior. The fed-batch control system enables the dissolved oxygen and substrate concentration and pH to be more constant throughout the batch. The variability in these controlled variables is transferred to the manipulated air, substrate, and reagent flows. At first this can be disquieting because operations personnel have now lost the ability to force these flows to follow a fixed pattern. The definition of a controlled variable and a manipulated variable is based on the latter serving the needs of the former and not vice versa. The final resting value of manipulated variables can be optimized by slowly adjusting controlled variables. This is in fact the strategy of valve position controllers that seek to increase production rate by keeping a coolant valve at the maximum controllable position by manipulating a feed flow. One cannot keep constant both the controlled variable and the manipulated variable. Insight : The feedback control key process variables by the manipulation of batch feeds in fed-batch will automatically correct for unknowns and disturbances. The optimized fed-batch control system makes adds a higher mode of control. A model predictive controller whose controlled variables are the slope of the biomass concentration profile (cell growth rate) and the slope of the product concentration profile (product formation rate) manipulates the dissolved oxygen and substrate concentration setpoints. The result is a batch profile of these higher controlled variables that more closely matches their optimums. Insight : The addition of an upper mode of control by the addition of changes of compositions and temperature with respect to time as controlled variables can optimize batch profiles. For a given size reactor, the production rate is higher for a continuous reactor than for a pure batch or fed-batch reactor. After startup a continuous reactor is continually producing whereas a pure batch or fed-batch reactor takes time to charge reactants, bring the contents to the proper temperature and pressure, wait for the reaction to complete, and then empty the reactor. In some cases the reactor may need to wait on shared resources such as feed weigh tanks or on operator actions or lab results from sequential actions. Batch reactors are particularly vulnerable to an attitude that every flow must be scheduled based on process knowledge. The use of feedback control to automatically set these flows based on actual process response is critical. The addition of successively higher level of control transfers variability from more important process outputs to the flows that are process inputs. See Greg McMillan’s new ISA book, Advances in Reactor Measurement and Control, for an extensive view of practical opportunities for designing control strategies to achieve product quality and maximize yield and capacity in different types of reactors. Continuous processes cost less in terms of investment and utilities for a given capacity but generally require greater process research, development, and design. Mature high capacity products (e.g., oil, gas, and petrochemicals) tend to use continuous reactors whereas new high value processes (e.g., specialty chemical and biological) primarily use batch and fed-batch reactors. As volumes increase and the profit margins decrease (the product becomes a commodity), there is increased emphasis on developing a continuous process to provide a higher capacity and lower operating cost. Candidates for continuous reactors are products with a low profit margin, a high volume requirement, fast reactions, minimal adverse reactions, preventable buildup of inhibitors and inactive components, and extensive Research and Development (R&D) history leading to a deep fundamental knowledge of kinetics. Oil, gas, chemical intermediates, petrochemicals, and commodity chemicals use continuous reactors. Extensive integration of unit operations for energy recovery and recycle of materials offers complex opportunities for optimization. Insight : Continuous reactors offer more capacity for a given size or a lower capital cost and greater energy efficiency for a given capacity but require greater knowledge and effort in research, development and design. The fastest and simplest implementation is pure batch with quantities charged sequentially mimicking lab experiments much like ingredients in a recipe. As knowledge is gained batch reactors can move to become fed-batch reactors and eventually continuous reactors if there is enough demand and if reaction chemistry permits a variable residence time. In contrast to continuous reactors candidates for batch reactors are products with a high profit margin, low volume requirement, slow reaction, significant side effects, and minimal knowledge of kinetics. Specialty chemicals, food, beverages, and especially biopharmaceuticals are produced by batch reactors. For new biopharmaceuticals with the patent expiration clock ticking and extraordinarily high prices (e.g. > $1000 per gram), the time to market supersedes any consideration of process efficiency or capital cost. These processes have extremely slow kinetics, requiring days to weeks for the cells to produce product. Insight : Batch operations require less process knowledge and customization of equipment and offer a fixed and tight residence time distribution enabling a more certain result and a faster project when a new product is commercialized. The buildup of dead and mutated cells, toxins, inhibitors, viruses, and bacteria can be prevented in batch processes by emptying, cleaning, and sterilizing the equipment and piping after each batch. The adverse results of contamination to food, beverages, and drugs can be extreme in terms of the loss of capacity, cost of retrieval, and most of all the consequences to the customers. Insight : Batch processes enable the periodic emptying, cleaning and sterilization of equipment and piping to prevent the buildup of contaminants in products that end up in a person or animal. Some mature biopharmaceuticals are produced by “perfusion” processes operating in the continuous mode for months at a time with extensive recycle due to the slow kinetics. Product is removed from the discharge stream and the recycle stream is purified before being returned to the bioreactor. These processes are periodically shut down for total elimination of mutated cells, inhibitors and alien organisms and decontamination. R&D and process design will determine whether a reactor will be batch or continuous based on process knowledge, product capacity requirements, profit margin, and impact of time to market. Once this decision is made, the process control engineer should promote the automatic transfer of variability from the successively higher loops to lower loops. Essential is the understanding that variability must be transferred or absorbed and cannot simply entirely disappear and processes inherently have a significant degree of non-repeatability and unpredictability.

The post An Overview of the Control Loops For Batch, Fed-Batch and Continuous Reactors first appeared on the ISA Interchange blog site. The following insights are part of an occasional series authored by Greg McMillan , industry consultant, author of numerous process control books and a retired Senior Fellow from Monsanto. Nearly all reactors have temperature and pressure control since these affect reaction rate. Concentration control is desirable but depends upon the feasibility, accuracy, and reliability of analyzers and inferential measurements. For at-line analyzers, there is the additional consideration of the dead time added to the loop. To date offline analyzers have only been used to correct inferential measurements or to make manual adjustments. The use of an enhanced PID opens the door to the possibility of using off-line analyzers directly for control. A large and variable analysis time does not cause instability in a loop with an enhanced PID. The time between off-line analysis results still needs to be made as short as possible for the PID to correct for disturbances since the PID only makes a correction when there is an update. Insight : Nearly all reactors need temperature and pressure control and would benefit from concentration control. There is generally no level control except perhaps in terms of a high level override or high level shutdown of feeds in pure batch and fed-batch reactors. In batch reactors, the reactants are fed sequentially and shutoff when charge tank weight measurements or flow totals indicate that the total amount charged is complete. On-off isolation valves are used to start and stop feeds. Flow controllers and control valves are sometimes used to prevent vapor system overload and provide a repeatable feed cycle time. In fed-batch operation, the reactants are fed simultaneously at a rate determined by a control system. As a minimum there are flow controllers for each reactant. Often common and special opportunities are not recognized due to preconceptions of control loops used for a type of reactor. See Greg McMillan’s new ISA book, Advances in Reactor Measurement and Control, for an extensive view of practical opportunities for designing more innovative control strategies to achieve product quality and maximize yield and capacity in different types of reactors. Many of the same control systems used for continuous reactors are applicable to fed-batch, except that there typically is no level and residence time control. In some cases a level override of reactant feeds is used to push a maximum level constraint to increase batch size for a reactor with a gas coproduct. The level override control maintains a minimum vapor space and prevents liquid carryover into the gas coproduct recovery system. The literature does not lead one to understand that the pressure, temperature, and composition controllers used in continuous reactors to manipulate reactant flow setpoints can be used for fed-batch. The use of these cascade control systems in fed-batch may even be more important because of the opportunity of these systems to provide batch profile control. Insight : Many of the control systems used for continuous reactor control can be used for fed-batch control. The exception is that level control is not needed except possibly as an override. Relatively fast and tight temperature control is possible for gas reactors by the manipulation of reactant feed rate due to the lack of inverse response and the small process dead time. The process dead time from the transportation delay of reactants is small compared to the time constants from catalyst heat capacity and thermowell design. Insight : Relatively fast and tight temperature control is possible for gas reactors due to a small process dead time by the manipulation of reactant feed rate rather than coolant. There is an optimum profile of temperature, physical properties and composition with respect to length for a plug flow reactor. For a batch vessel there is an optimum profile with respect to batch time. For extruders, the decrease in temperature and the increase of viscosity with length can be optimized. Optimum profiles may also be applicable with respect to startup time and transition time of continuous reactors. There are more opportunities for profile control and optimization than presently realized in the process industry. Insight : Optimum profiles exist with respect to time for batch reactors, during the startup and transition of continuous reactors and with respect to length for plug flow reactors. Since intermediate and final product composition generally go in one direction only with time for nonreversible reactions, the control of an absolute value whose setpoint varies with time can result in a unidirectional response preventing the use of integral action. The use of the slope of a composition profile as the controlled variable enables a bidirectional response allowing the use of a standard PID structure and tuning. This strategy can be applied to intermediate and final products and byproducts. For example, bioreactor cell growth rate is controlled early in the batch and lactate formation and product formation rate is controlled late in the batch. Insight : The slopes of key compositions with respect to time should be controlled rather than a concentration value whose setpoint is varied with time. By understanding the underlying principles in reactor control we can apply advancements in reactor control loop design for one type of reactor to other reactors. By not treating each reactor as a special case, we can build our knowledge base and show how control loops can make significant improvements in reactor performance gaining greater recognition for the value of process control.

This post by Greg McMillan first appeared in the Control Talk blog on ControlGlobal.com . There are some simple diagnostic checks and rules of thumb on tuning adjustments that can be used to find out if there is a problem with the PID tuning and what is the solution. This guidance in conjunction with good tuning software can reduce process variability introduced or aggravated by improper PID tuning. The ultimate period of 99% of the loops is about 4 dead times. Loops with a large secondary time constant will have a larger period (e.g., 6 dead times) and loops with the dead time much larger than the primary time constant will have a shorter period (e.g., 2 dead times). The ultimate period corresponds to the critical frequency. Since the dead time is rather easy to visually see (time before the start of a response for a change in PID output or setpoint), the checks and test described here are based on comparison of the actual oscillation period to an ultimate period that is 4 times the dead time. Here we assume the data historian update time is less than 10% of the dead time and the compression is small enough to see the start of the response. If the period of oscillation is more than 6 times the dead time, the culprit is too small of a reset time. If the period is more than 10 times the dead time, the product of the PID gain and reset time is too low. This last situation occurs frequently in level, gas pressure, and liquid composition and temperature control loops. These loops have a ramping open loop response as described in the 3/19/2013 blog Processes with No Steady State in the PID Time Frame (Conclusion). Often the reset time is two orders of magnitude too large due to the propensity to not use the proportional mode because it can cause large abrupt changes in the PID and to rely more on the integral mode because it provides a more gradual ramping type of action. Also, the integral mode provides the type of control that humans would do manually where the direction of a change in output is not reversed until the error changes sign. The classic case I have often cited is where a split ranged steam valve rather than coolant valve is expected by the operator to be open when a temperature is below setpoint. Whether the problem is too small of a PID gain or too small of a reset time, the solution is simple and relatively safe for near-integrating, true integrating, and runaway processes . Increase the reset time by a factor of 100 and see if the oscillation goes away or the decay rate and period is faster. If the oscillation goes away, you can decrease the reset time gradually to see when the oscillation starts again after a change in the PID setpoint or output. See the 4/11/2013 blog How to Avoid a Common Tuning Mistake for more details . If the oscillation period is close to 4 times the dead time, the problem is most likely too large of a PID gain. The gain could be causing instability or amplifying a load disturbance due to resonance. In either case, decrease the PID gain until the oscillation goes away or decays more quickly. If the oscillation persists but has a smaller amplitude, resonance is likely. If derivative action is used and the oscillation period is about 3 times the dead time, the rate time could be too large. The rate time should be less than the dead time and less than ¼ the reset time for the ISA Standard Form PID. If there is no possibility of a runaway response, set the rate time equal to zero and see if the oscillation dies out. If the oscillation starts to increase, immediately restore the rate time. The PID should be put in manual if permissible (e.g., if there is no possibility of a runaway reaction or a fast pressure excursion) to see if the oscillation persists with the same amplitude or disappears. If the oscillation has the about the same amplitude in manual or automatic, there is not much to be gained by PID tuning. For level control, the tuning may need to be adjusted to maximize the absorption of variability so the load oscillations are not passed on to a level control valve on the discharge flow. An enhanced PID and external reset feedback and setpoint rate limits in an analog output block and secondary loop can help reduce the response to the disturbance oscillation to prolong valve packing life. The tuning problem and fix should be verified by tuning software tests and observing the response to a load disturbance. Note that several tests should be made because of the continual changes in automation system and process conditions and dynamics. There are continual disturbances and changes in gains, time constants, and dead times. The question is not whether these changes exist but now much do they affect the tuning. A process response is not repeatable for these and many other reasons (the subject of the next blog). The introduction of a small load disturbance for testing the tuning is more desirable than a setpoint change because of the effect of PID structure on the setpoint response and the need to see how a PID deals with a load disturbance, the basic objective of feedback control. If there were no disturbances, we would not need PID control. The PID should be first tuned for a good load response. The desired setpoint response (e.g., minimal overshoot) should be obtained by the choice of PID structure including the more flexible “2 Degrees of Freedom” (2DOF) or a setpoint lead-lag. 1. The tuning for a load disturbance can be tested by momentarily putting the PID in manual, making a step change in the PID output, and then putting the PID in automatic. The step size should be about the same size as a typical total correction in the PID output to deal with disturbances but at least 5 times larger than measurement noise, valve backlash, and any resolution limit. The dead time can be estimated as the time it takes from the step change in PID output till a noticeable change in the PID process variable. 2. For a load disturbance, if the return to setpoint is more than twice as slow as the initial excursion, the reset time is too large. If the process variable oscillates with a period much greater than 6 times the dead time, the reset time is too small. For a setpoint change, an overshoot of the new setpoint is probably caused or aggravated by too small a reset time. 3. For a load disturbance, if the return to setpoint oscillates with a period between 4 and 6 times the dead time, the PID gain is too large. For a setpoint change, a hesitation in the approach to the new setpoint is indicative of too large a PID gain. 4. For a load disturbance, if the return to setpoint oscillates with a period less than 4 times the dead time, the rate time is too large. For a setpoint change, an oscillation in the approach to the new setpoint is indicative of too large a PID rate time.

The post What are the Implications of Reactor Type on Control? first appeared on the ISA Interchange blog site. The following insights are part of an occasional series authored by Greg McMillan , industry consultant, author of numerous process control books and a retired Senior Fellow from Monsanto. This article was originally published at the ISA Interchange blog site. Reactors involving liquid reactants can be categorized as continuous stirred tank reactors (CSTR), batch, fed-batch (semi-batch or semi continuous), and plug flow. In continuous reactors, there is a continuous discharge of product flow. A level controller is used in the CSTR to maintain a level that provides enough reaction time for a given production rate. In batch and fed-batch reactors, the discharge valve is closed until the batch is ready to transfer. A gas product or byproduct may be continuously generated during the batch and condensed and accumulated in an overhead system. A liquid and/or solids phase reaction without a continuous liquid and/or solids discharge flow is the distinguishing characteristic of batch and fed-batch. The implications as to dynamics, yield, and inventory control are significant. The table below summarizes how the dynamic response, residence time, reaction rate and controlled variables vary with reactor type as discussed in the following sections. The reactor types considered are batch, fed-batch, plug flow, gas flow, and continuous stirred reactors. Reactor Type Dynamics and Control (*additional control besides temperature and pressure control) A continuous stirred reactor has a slow near-integrating response due to the residence time becoming a large primary time constant from good back mixing. A batch and fed-batch reactor has a slow true integrating response. A continuous plug flow reactor has a moderate self-regulating response for temperature control by manipulation of coolant due to the thermal time constant introduced by heat transfer surfaces. Inline polymerization reactors, static mixers, extruders, and gas reactors (e.g. fluidized bed reactors) are considered to be plug flow because there is negligible back mixing (negligible axial mixing). In gas reactors (e.g. fluidized bed reactors) feeds can be manipulated for temperature control because the reaction kinetics are so fast that there is little or no inverse response at the points chosen for temperature control. For these systems, coolant coils are switched in and out of service to set capacity. The process response is still moderate self-regulating when temperature manipulates feed due to the thermowell time constant. All of these reactors can develop a runaway response when the increase in reaction heat release with temperature exceeds the cooling capability. Insight : Temperature control of a continuous stirred reactor has a slow near-integrating response, a batch reactor has a slow true integrating response, and a continuous plug flow reactor has a fast moderate self-regulating response. All reactors can develop a runaway response if the exothermic reaction heat release rate exceeds the cooling rate capability. The use of an analyzer with a large sample, cycle, and/or multiplex time almost guarantees dead time dominance (total loop dead time larger than open loop time constant) for composition control of continuous, plug flow, and gas reactors. Inferential measurements, online analyzers and fast at-line analyzers are used to minimize the introduction of excessive dead time into concentration control. Insight : The use of slow at-line analyzers for concentration control of continuous reactors can create a dead time dominant loop. The introduction of a slow temperature sensor or slow analyzer can be disastrous for reactors that can develop a runaway response. The dead time or secondary time constant must be much smaller than the primary positive feedback time constant otherwise the window of allowable controller gains closes causing instability for all tuning settings. Fouling of heat transfer surfaces for coolant and for thermowells add an excessive secondary time constant. Slow analyzers add excessive dead time. Insight : The fouling of heat transfer surfaces and a long sample transportation cycle time or multiplex time can cause a reactor with runaway a response to be unstable for all tuning settings. Sometimes fed-batch is called semi-batch or semi-continuous. However, there is an important distinction between fed-batch and continuous reactors; in that in fed-batch as in batch reactors there is no liquid product discharge flow until the end of the batch. The lack of a liquid discharge flow is an integrating response and a rise in level with reactant addition. The increase in level from the start to the finish of the batch and the lack of self-regulation as a result of the closed discharge valve affects the dynamic response of temperature and composition requiring tuning methods and settings not commonly discussed in the control literature. For example, a temperature controller gain of 50 or more is possible for fed-batch reactors. The measurement and control of biological and chemical reactors is the key to product quality and the yield and production rate of most processes in the process industry. See Greg McMillan’s new ISA book Advances in Reactor Measurement and Control for an extensive view of practical opportunities for building and effectively using online estimators to improve process knowledge and control. Residence time is critical for reactors; it determines the amount of time available for reaction by virtue of the reactants being in contact with each other. The reaction time must be less than the residence time for the reaction to go to completion. If the residence time is too long, undesirable reactions and product degradation may occur. For maximum yield, the residence time should be slightly larger than the reaction time with a tight residence time distribution (statistical plot of each component population versus residence time should be a narrow peak). For maximum capacity and minimum capital cost, the residence time should be as small and the distribution as tight as possible since residence time is proportional to cycle time and volume. Insight : For maximum process efficiency and capacity and minimum capital cost, the residence time should be slightly larger than reaction time with an extremely tight residence time distribution. For batch operations the batch cycle time is extended past the feed phase via a hold phase to ensure the reaction is complete. The residence time is the average volume divided by average feed rate during the feed phase plus the time in the hold phase. For pure batch operations where all of the feeds are added quickly and simultaneously, all of the components are in the reactor for about the same time making the residence time distribution tight. In fed-batch operations the feeds are added via a cascade control and ratio control system. Typically, a primary loop for concentration control is manipulating the setpoint of a secondary loop for the leader feed flow. The other feed flows have setpoints ratioed to the leader feed flow setpoint. The rate of feed addition for fed-batch operation is slower and nonlinear. The difference from one batch to another is not seen in the concentration profile but in the profile of the manipulated feed flows if the cascade control system is doing a good job. The residence time of components added early in the feed phase is longer adding a slight degree of uncertainty in the residence time distribution. Batch operations are considered here to be well-mixed so that turnover time is negligible. The downside of batch operations is a greater equipment cost to achieve the same capacity as continuous operations. Insight : Batch operations assure a residence time greater than the reaction time and an extremely tight residence time distribution but offer less capacity than continuous operations. The overall residence time for continuous operations is the current operating volume divided by the total flow rate passing through the volume. Some tuning methods for surge tank level control inadvertently use a definition of residence time as the maximum volume divided by the maximum change in flow set by equipment size, level connection location, and PID scale ranges. At high production rates, the residence time gets shorter for reactions unless the level setpoint is increased accordingly. Insight : The residence time for continuous operations is the operating volume divided by the total flow. A plug flow reactor has no back mixing (axial mixing). Polymer plug flow reactors use inline equipment such as pipes or tube(s). In static mixers baffles create some radial mixing. In extruders, a screw pushes a viscous liquid and solids to the exit. Since there is no back mixing all of the components in the reaction mass exit at the same time that is equal to the transportation delay. All of the residence time becomes a process dead time. A plug flow reactor is similar to a fed-batch reactor in that there is generally no level control and the residence time distribution is extremely tight. If you consider a subsection of the reaction mass moving from the entrance to the exit, there is no discharge from this volume until the subsection approaches the exit. Insight : The residence time in plug flow reactors is a transportation delay and the residence time distribution is tight. Polymerization processes often use batch and plug flow reactors operating with high viscosities. For highly viscous products the residence distribution is not as tight. Product can accumulate on the vessel walls in batch reactors. The velocity near the pipe wall in plug flow reactors is slower from frictional drag causing the product near the wall to arrive later. The increase in time available for reaction increases the viscosity of the fluid near the wall aggravating the problem. Gas phase reactors are generally always continuous plug flow reactors. Many use fluidized catalyst beds. There is little to no back mixing except from turbulence. Consequently, the residence time distribution is tight for even catalyst distribution, the dynamic response to changes in reactant feed is a transportation delay similar to the plug flow liquid reactor except the reaction times are faster and the residence time shorter from a much higher velocity. Channeling of gas flows through the catalyst bed can result in some gases arriving at the outlet sooner reducing residence time and conversion for some portions of the exiting gas product. Insight : The residence time of gas phase reactors is fast and tight for an even flow distribution. In a continuous stirred reactor some of the reactants appear in the discharge within one turnover time, which in general is less than 1/20 of the residence time in a well-mixed vessel. The residence time becomes a large primary time constant causing a near integrating response. The amount of time spent by a given component in the reactor varies resulting in a looser residence time distribution and a loss in yield in the CSTR. Reactor designs with a liquid height about equal to the diameter, baffles to prevent swirling, axial agitation pattern that breaks the surface, and a dip tube design to bring reactants into the eye of the impeller; away from discharge nozzles to avoid short circuiting have a tighter residence distribution. Several CSTR in series or a reactant recovery and recycle system can be used to improve the yield of the production unit particularly for slow reactions and high purity requirements. However the use of a recycle system creates an integrating response and increases the propensity for buildup of inerts and impurities requiring the optimization of recycle flow and a purge flow for the best yield. Insight : The residence time in continuous stirred reactors becomes a primary time constant and the residence time distribution is loose but can be tightened by good equipment designs The implications of reactor type should be recognized. The effect on dynamic response, allowable measurement lags and delays, controller tuning, and control strategies is significant.

When is a controller in automatic not able to do anything to reduce an oscillation? When will a controller amplify an oscillation? In both of these cases, the controller is doing more harm than good by wearing out valves and upsetting other loops. Here we look at how to tell if an oscillation is likely to cause either of these scenarios and what we can do to reduce the detrimental effect on the process. For oscillation periods between ½ and 2 times the ultimate period (e.g. 2 to 8 dead times), resonance can be occurring. Figure 8.1 in the file Attenuation-and-Resonance-Amplitude-Ratio shows the general effect of resonance where feedbacks control action becomes in phase with the disturbance causing an amplification of the oscillation amplitude. The amplification becomes greater as the PID is more aggressively tuned. Note that abscissa (X axis) of this plot is the log of the ratio of oscillation period to ultimate period whereas the literature uses an abscissa that is the log of the ratio of the oscillation frequency to the natural frequency. An abscissa in the time domain enables a better visualization from trend chart oscillation periods and estimating the ultimate period as simply 4 times the dead time but the result is horizontal flip of what is normally seen in the literature. For oscillations periods much less than the ultimate period, the process provides significant attenuation by the process time constant acting as a filter. The equation in the previous blog Measurement Attenuation and Deception can be used to estimate the attenuation by the process. The PID provides no feedback correction for a fast load oscillation as seen in Figure 8.4a in the file Attenuation-and-Resonance-Test-Results . A trend of the PID in manual eliminates the feedback correction so that only the effect of the attenuation by the process time constant is seen. These extremely fast oscillations can be effectively considered to be noise and the best thing the PIUD can do is ignore it. While the oscillation is not made worse by PID control, Figure 8.4b shows how more aggressive tuning cause unnecessary extra movement of the valve that can prematurely wear out packing and can upset other loops particularly when there is no large process time constant acting as a filter of the process variability introduced by the oscillations in the manipulated flow. For oscillation periods between ½ and twice the ultimate period, the aggressive tuning causes amplification from resonance where the feedback correction oscillation gets in phase with the load oscillation. Here the PID is clearly doing more harm than good and the more aggressive tuning is clearly detrimental. When the load oscillation period becomes greater than twice the ultimate period the most aggressive tuning settings noticeably decreases the amplitude of the process variable oscillation compared to the PID in manual oscillation. The oscillation in the PID output is larger for the most aggressive settings. For oscillations periods much greater than the ultimate period the process provides no attenuation. Here PID feedback correction can be significant. More aggressive tuning provides a greater reduction in the oscillation amplitude as seen in the Figure 8.5a test results for a slow load oscillation. The benefit from aggressive tuning is not as noticeable as the load oscillation period increases toward the point where PID control can make the oscillation in the process variable disappear to the point of being lost in noise and measurement or final control element precision limit cycles. The oscillation will be totally visible in the PID output. Tight PID control will almost completely transfer variability from the process variable to the manipulated variable as seen in the valve movement in Figure 8.5b. This transfer has many implications in terms of tracking the path of variability propagation and what variables are chosen for developing neural network models and projection to latent structures (partial least squares) models. The benefit of more aggressive tuning will not be so clear for oscillations with a period that is incredibly long (e.g. 1000 times the ultimate period), which might be the case for day to night effects on operating conditions. So what can we do about fast oscillations? The best thing to do is mitigate the source of the oscillation. See my 5/30/2013 blog Causes and Fixes for Fast Oscillations   for some tips. We can also judiciously add a signal filter just large enough to prevent wearing out the valve and upsetting other loops. For a slow process (e.g., liquid temperature), a velocity limit can screen out noise without adding a measurement lag. See my 12/12/2012 blog What are Good Signal Filtering Tips? for some ideas. Note that for wireless transmitters, the filtering is best done via a transmitter damping adjustment to reduce unnecessary updates. The wireless trigger level should be set to be larger than noise amplitude after the judicious use of transmitter damping. An enhanced PID in conjunction with a threshold sensitivity limit can stop reaction to noise. Given that we are stuck with an oscillation that is causing resonance, the next step is to decrease the PID gain. This can be done by increasing lambda to 3 or more dead times.