wireless measurement and control opportunities

Distributed with permission of author(s) by ISA 2011 Presented at ISA Automation Week Technical Conference; http://www.isa.org

Wireless Measurement and Control Opportunities

Greg McMillan (CDI Process & Industrial) Key words: conductivity measurement, electrode testing, electrode diagnostics, flexible manufacturing, inferential measurement, online metrics, online diagnostics, pH calibration, carbon dioxide absorption, pH measurement, process optimization, wireless control, wireless measurement ABSTRACT Wireless measurements can reduce maintenance and noise problems in addition to installation costs by the elimination of wiring problems and electromagnetic interference. Less recognized are the opportunities afforded by wireless measurements for troubleshooting and optimizing measurement locations as well as developing and prototyping process control innovations. However, battery life and network integrity raise reliability, security, and maintenance questions. Communication interruptions and discontinuous updates can cause oscillations for traditional PID controllers. This paper addresses these concerns and discusses the potential use of wireless pH measurements for minimizing noise, maximizing sensor performance, selecting sensor technology, predicting sensor life, and developing inferential measurements. An example of the use wireless conductivity and pH measurements as inferential measurements of solvent and carbon dioxide is given to enable the optimization of absorber operation. The advantage of using spare wireless transmitters instead of lab meters for communicating test data for inferential measurements and calibration data from standardization methods with grab samples is offered. A simple enhancement of the PID algorithm for wireless control to extend battery life is explained and test results are presented for measurement failures, setpoint response, load upsets, and valve stiction. The benefits for valve position control used to eliminate split range control and to optimize unit operations is offered. The effect of wireless transmitter settings such as “default update rate” and “trigger level” on control loop performance is estimated for unmeasured disturbances in terms of the additional deadtime added by wireless settings. The equations show the importance of a threshold sensitivity setting for reducing the impact of noise. The positive effect of the threshold sensitivity setting on increasing the controller gain and reducing deadtime from valve backlash and sticktion is quantified. The paper finishes with a discussion of extensive opportunities offered by large scale sensor networks for flexible manufacturing, plantwide feedforward, online diagnostics and metrics, and Live Process Flow Diagrams.


INTRODUCTION The use of conductivity and pH in water and waste water covers a wide range of application conditions ranging from nearly pure water to wastewater with an incredible variety of chemicals. For condensate and de-ionized water, the low fluid conductivity poses special problems in terms of velocity effects, continuity, and sensitivity. For waste water streams chemicals can attack the electrodes and solids can coat the electrodes. Process and electrical noise are problems in all applications. The use of conductivity and pH in the measurement and control of carbon dioxide absorption is an emerging opportunity with special considerations due to the relatively high solvent concentration. Wireless measurements can play a key role in the design, installation, maintenance, and performance of conductivity and pH, particularly in demanding applications. Wireless measurements can provide remote location access and the coordination of production units and utility systems particularly important for plants that are highly integrated. Just in time flexible manufacturing and logistics require more information from tank farms and delivery systems. Less recognized are the opportunities afforded by integral mounted wireless measurements for troubleshooting and optimizing unit operations and measurement locations as well as developing and prototyping process control innovations and realizing the best measurement technology. Wireless conductivity and pH meters offer the opportunity for more intelligent lab tests and field calibrations with primary and auxiliary variables readily historized and analyzed in the DCS. The development of an enhanced PID algorithm offers an excellent setpoint response and stability for slower update rates extending wireless control to 99% of the loops in a typical process. The only loops not suitable for wireless control have an extremely small deadtime and large fast unmeasured disturbances (e.g. compressor surge control, liquid polymer line pressure control, and phosphorous furnace pressure control). DESIGN Wireless technology offers the opportunity to select the best electrode for an application. Spare wireless transmitters can be used to test electrodes in lab solutions set to mimic the concentrations and temperatures of the process. The process lab offers a controlled environment to vary the process conditions of interest and keep other conditions constant. The first through fourth variables of smart wireless transmitters can be readily made available for historization and analysis in the distributed control system (DCS). The slope, offset, and resistance of the electrodes can be studied. The accuracy and response time can be measured over an extended period of time to determine the performance and life of various electrode designs. For example, new high temperature glass formulations increase the life expectancy (Figure 1) and prevent the deterioration of electrode response time (Figure 2) from premature aging of the glass electrode.


Figure 1. The life expectancy that decreases with temperature can be doubled by the use of new high temperature glass formulations.

Figure 2. The onset of premature aging of the glass that greatly increases the response time can be delayed by the use of new high temperature glass formulations


Figure 3. The change in water dissociation constant will decrease the pH with temperature by an amount that increases as the pH approaches the dissociation constant Process temperature changes the dissociation constants and mobility of the ions and thus the actual solution pH and conductivity. The best way to determine the effect of process temperature on solution pH is to vary the process sample temperature over the expected operating range. The change in the water dissociation constant (pKw) with temperature causes a change in solution pH that increases as the pH approaches the pKw (Figure 3). Smart wireless transmitters have the ability to configure solution temperature compensation. However, lab meters often cannot provide more than the standard temperature compensation for the electrode per the Nernst equation. Wireless transmitters could be used instead of lab meters for standardization with grab samples enabling historization and analysis in the DCS. Standardization is safer and more effective since it does not require removal of the process electrodes eliminating the associated risk of personnel exposure and damage to the glass, and disruption of the reference junction equilibrium. Conductivity can provide an inferential measurement of the concentration of an acid, base, or salt by measuring the concentration and mobility of ions. Plots of conductivity versus ion concentration will increase from zero concentration to a maximum as the number of ions in solution increases. The conductivity then falls off to the right of the maximum as the ions get crowded and start to interact or associate (group) reducing the ion mobility (Figure 4). The peak must be found from process samples and measurements made for all possible operating conditions to insure operation on one side or the other of


the peak. If the conductivity approaches the peak, the inferential measurement can lose sensitivity and change sign invalidating it for process control.

Figure 4. Conductivity as an inferential measurement of concentration can be used if the operating range can be verified to be always on one side of the peak. For the maximum efficiency of carbon dioxide reduction in powerhouse stacks, the solvent and carbon dioxide concentration in the absorber must be measured and controlled. The absorber is never at steady state but is in a constant state of transition because boiler emissions vary significantly with weather and day to night load swings. Conductivity and pH were studied with wireless transmitters in the University of Texas (UT) Separations Research Facility with a reactive absorption system and solvent recovery column (Figure 5). The operation of the UT pilot plant boiler was varied to emulate the type of load changes experienced in a powerhouse. Wireless conductivity and pH transmitters were used in a process sample UT Separations Research lab representative of absorber operation (Figure 6). The use of a new glass formulation and reference design was found to increase the life of the pH electrode from 2 weeks to over 6 months. The relationships between pH and conductivity and the solvent and carbon dioxide concentrations were quantified by tests at concentrations representative of the range of powerhouse operation.


Figure 5. UT Separations Research Facility for Carbon Dioxide Recovery

Figure 6. Wireless Test Setup of Conductivity and pH Transmitters for CO2 Recovery


The use of conductivity as an inferential measurement of solvent concentration turned out to be not viable due to operation at the peak in the conductivity versus concentration plot (Figure 7). The maximum was a rounded hump rather than a sharp peak causing poor sensitivity (small change in conductivity for a change in concentration). Even more problematic was the reversal of the conductivity response that would lead to saturation of a controller output due to reversal of the direction of the process action. Conductivity was found to be a possible inferential measurement of CO2 loading if proper process temperature compensation is applied (Figure 8). Coriolis density and viscosity measurements also showed sufficient sensitivity to CO2 loading. However, wireless conductivity is more portable and affordable. Since smart wireless conductivity like pH transmitters offer solution temperature compensation and auxiliary variables, these transmitters are preferable to lab meters for standardization with grab samples. Tests of the wireless pH transmitter in the UT lab showed a linear relationship between pH and solvent concentration for a given CO2 concentration and temperature with good sensitivity (Figures 9 and 10). The Y axis is solvent concentration free of CO2. Again solution pH temperature compensation was important. The change in pH with solvent is due to the change in the concentration and the mobility of the hydrogen ion with the change in water concentration.

Figure 7. Conductivity for Inferential Measurement of Solvent Concentration had Poor Process Sensitivity and Reversal of Process Action Sign


Figure 8. Conductivity for Inferential Measurement of CO2 Loading Showed Promise if Proper Temperature Compensation is Developed and Automated.

Figure 9. pH can be an Inferential Measurement of Methyl Ethyl Amine (MEA) Solvent Concentration if the Solution pH is Compensated for CO2 Loading and Temperature


Figure 10. pH can be an Inferential Measurement of Piperazine (PZ) Solvent Concentration if the Solution pH is Compensated for CO2 Loading and Temperature INSTALLATION AND MAINTENANCE pH measurements are susceptible to spikes from the start of motors and the change in speed of variable frequency drives. The fact these spikes are not observed when the electrode is connected to a laboratory meter is clue that transmitter output wiring may be a significant contributing factor to ground loops. When a wireless pH transmitter was used on a “single use bioreactor” (Figure 11), the spikes in the wired transmitter left in service did not appear in the wireless transmitter trend chart (Figure 12). Wireless pH measurements eliminate the inevitable questions as to whether there is a problem with the wiring or terminations in troubleshooting the measurement. Also, wireless pH transmitters with automated solution temperature compensation and measurements of electrode resistance can be used to provide more intelligent grab sample calibrations by statistical analysis in the DCS. Most pH systems suffer from over calibration with calibration adjustments chasing previous adjustments. If you compare the readings of two pH electrodes inline over several days, what may be high to day may be low tomorrow due to concentration gradients (Figure 13). The use of 3 wireless measurements with middle signal selection can ride out a failure of any type.


Figure 11. Wireless pH Transmitter on a Single Use Bioreactor

Figure 12. Wireless pH Transmitters Do Not Exhibit Spikes Seen in Wired Transmitters


Figure 13. The pH Electrode that is High Today may be Low Tomorrow The portability of wireless conductivity and pH measurements offer the opportunity to find the best location by plant trials. If several process connections are provided, the connection that offers the most representative measurement with the least noise and deadtime can be found. Noise is minimized by finding the location with the best mixing and least bubbles. Deadtime is minimized by reducing the transportation delay and increasing the velocity at the electrode, decreasing the electrode lag and coatings. CONTROL Wireless measurement devices have a “default update rate” (time interval for periodic reporting) and a “trigger level” (sensitivity limit for exception reporting) set as large as possible to conserve battery life. The integral mode in the traditional PID will continue to ramp while the PID is waiting for an updated measurement from a wireless device. Also, when an update is received, the traditional PID considers the entire change to have occurred within the PID execution time interval. If derivative mode is used, the rate of change of the measurement is the difference between the new and old measurement divided by the PID execution time interval. The result is a spike in the controller output. The non-continuous update scenario occurs for many applications besides wireless devices. During the time when a measurement is not updated due to a failure, resolution limit, sensitivity limit, or backlash, the PID output continues to ramp from the integral mode. Failures, resolution limits, and sensitivity limits can originate in an analyzer, sensor, transmitter, communication system, or control valve. Analyzers also have a time interval between updates determined by the sample time and cycle time.


The enhanced PID for wireless executes the PID algorithm as fast as wired devices. A change in setpoint, feedforward signal, and remote output translates immediately (within PID execution time interval) to a change in PID output. However, integral action does not make a change in the output until there is an update. When an update occurs, the elapsed time between the updates is used in an exponential calculation that mimics the action of the filter block in the positive feedback implementation of integral action. If derivative action is used, the elapsed time rather than the PID execution time interval is used to calculate the rate of change of the process variable. The integral and derivative calculations are executed only once upon a change in setpoint or measurement compares a simplified block diagram of the traditional PID to the enhanced PID (Figure 14). A traditional PID will have to be detuned to prevent instability for a large increase in the time between updates. The enhanced PID will continue to be stable for even the longest update time interval. For a measurement update time interval larger than the process response time and a fast reset time, the enhanced PID controller gain can be set equal to the inverse of the open loop gain (product of valve, process, and measurement gain) to provide a complete correction for setpoint change or update. The fast reset time prevent the proportional mode from reacting to the return of the measurement to setpoint. Tests show the enhanced PID can suppress oscillations from a wide variety of sources. This reduction in variability results from the suspension of integral action and the wait in feedback correction till there is a more complete response. To achieve these benefits, the user simply enables the enhanced PID option in the PID block, which automatically enables the dynamic reset limit option. No retuning is necessary to achieve a smooth response but if the update time is larger than the process response time the enhanced PID can be tuned with a much higher gain. For cascade control of static mixer pH to reagent flow control, the static mixer is able to make a single correction for a load disturbance (Figure 15) and a setpoint change (Figure 16) compared to the slow response of traditional PID. The enhanced PID eliminated the ramp off toward the output limit by a traditional PID for a communication failure or broken sensor or wire (Figure 17). Stick-slip, shaft wind-up, backlash, and resolution causes limit cycles because integral action in a traditional PID continues to ramp when there is an offset. The exceptional sensitivity of the pH measurement can result in extremely large limit cycles on the steep portion of the titration curve. The stick-slip is also greatest at the split range point because the reagent valves are operating near the seat where the friction from seating and sealing is greatest. The enhanced PID inherently prevents the limit cycles from stick-slip and backlash if a threshold sensitivity setting is used to prevent an update to the PID pH due to measurement noise (Figure 18). This capability is important for valve position controllers used to optimize unit operations by keeping a process controller’s valve at a maximum throttle position.


Figure 14. Enhancements of PID for Wireless Prevent the Ramping from Integral Action and the Spikes from Derivative Action for Discontinuous Updates

Figure 15. For a Wireless pH Loop Load Upset, the Enhanced PID Makes a Single Reagent Correction whereas the Traditional PID Slowly Makes a Series of Corrections


Figure 16. For a Wireless pH Loop Setpoint Change, the Enhanced PID Makes a Single Reagent Correction whereas the Traditional PID Slowly Makes a Series of Corrections

Figure 17. For a Wireless pH Loop, the Enhanced PID Rides out a Communication Failure or Broken Electrode whereas the Traditional PID Ramps to its Output Limit


Figure 18. The Enabling of the Enhanced PID option for Wireless Control Eliminates Limit Cycles from Valve Stick-Slip and Backlash Split range control has often been used to achieve greater control valve rangeability by throttling a small valve in parallel with a large valve when the demand for flow is low. For a low flows, the small valve starts to open and the big valve freezes at minimum position to prevent riding the seat of the big valve which is the point of greatest stick-slip and wear. However, unless the big valve threshold sensitivity, resolution, and backlash in percent of stroke is much less than the small valve, the limit cycles at high flow demand are much larger. A better solution is to eliminate split range control and use instead valve position control where the process PID manipulates the small valve and a valve position control (VPC) PID manipulates the big valve. Traditionally, integral only control is used in the VPC with an integral time that is 10x the product of the process PID gain and reset time to eliminate interaction. However, this action is sometimes too slow for large disturbances. For measured load upsets, feedforward control should be added. Here the feedforward signals for the process PID and VPC PID would be intelligently limited to keep the small valve in a good a throttle range and avoid unnecessarily positioning of the big valve with its larger stick-slip. The enhanced PID with a large update trigger level for the small valve position can be used for VPC instead of an integral only controller. The enhanced PID for VPC suppresses limit cycles from stick-slip and backlash and reduces interaction with the process PID while offering more aggressive action for large disturbances.


The crossing back and forth and discontinuity of the split range control can also be reduced for opposing media by the temporary transition to VPC of the valve not being throttled by the process PID instead of completely closing the valve. In this case the VPC provides a smooth transition by a small opening of the valve normally shut by split range control in order to keep the valve manipulated by the process PID from dropping below some minimum throttle position. There is a small temporary waste of energy or raw materials while the opposing media valves are both open but this may be less than the loss of efficiency from the crossing back and forth and discontinuity of the split range point. The enhanced PID can be used for this VPC to reduce interactions and provide more aggressive tuning. Valve position control is also used for the prioritization of the manipulation of multiple flows (multiple PID outputs) and for decreasing energy use and increasing production rate. Valve position control (VPC) is used to optimize energy use by decreasing the pressure of boilers and compressors and increasing the temperature of refrigeration units and cooling towers by working the furthest open user valve towards its maximum throttle position. The controlled variable for the VPC is the high signal selector output of user valve positions. The VPC PID output is the pressure or temperature setpoint of the utility. Similarly VPC can be used to optimize production rate by increasing feed rate to columns, crystallizers, evaporators, and reactors by working the furthest open utility and feed valve towards its maximum throttle position. The controlled variable for the VPC is the high signal selector output of utility or feed valve positions. The VPC PID output is the feed flow setpoint. The enhanced PID can reduce interactions, prevent limit cycles, and prevent the throttle valve from losing sensitivity from being on the flat upper portion of installed characteristic or hitting a high output limit. The enhanced PID can handle large step disturbances not typically possible with traditional slow integral only VPC. Wireless measurements can be added to quantify benefits of these strategies. LOOP PERFORMANCE For a step change in an unmeasured disturbance that would cause an open loop error of (Eo), the minimum peak error (Ex) (Equation 1) is proportional to the ratio of the total loop deadtime to the 63% response time (T63). The minimum integrated error (Ei) (Equation 2) is approximately the peak error multiplied by the total loop deadtime that is the original deadtime (θo) plus the additional deadtime from a wireless measurement (θw) and a control valve (θv). The 63% response time (Equation 3) is simply the sum of the original loop deadtime plus the wireless measurement deadtime and the open loop time constant (τo), which is hopefully the primary process time constant. Note that the valve deadtime from stiction and backlash does not appear in the 63% open loop step response time because the step change occurs within the controller execution and must be larger than the valve resolution, sensitivity, and deadband to create a response. Consequently,


on open loop step response trend charts, the deadtime from valve stiction and backlash is not seen. However, in closed loop operation, the controller output is ramping for an unmeasured disturbance and the time the controller takes to ramp through the valve resolution, sensitivity, and deadband is additional deadtime.

ovwo

x ET

E ∗++

=63

θθθ (1)

ovwo

i ET

E ∗++

=63

2)( θθθ (2)

owoT τθθ ++=63 (3)

The additional deadtime from a wireless measurement (Equation 4) is the smallest of the deadtimes from the wireless “default update rate” (update time interval) for periodic reporting (ΔTw), and the wireless “trigger level” (threshold sensitivity) for exception reporting (Sw). The deadtime from periodic reporting (θΔT) (Equation 5) is one half of the update time interval. The deadtime from exception reporting (θS) (Equation 6a) is one half of the sensitivity setting divided by the maximum rate of change of the % process variable (%ΔPV/Δt)max. Half of the interval and sensitivity settings are used because on the average the disturbance starts halfway in the interval and halfway in the sensitivity. The rate of change is a maximum during the beginning of the disturbance before the control loop has had any effect. This maximum rate of change is dictated by the size of the disturbance and process dynamics. The maximum ramp rate (Equation 6b) can be approximated by a near integrator gain multiplied by the equivalent change in controller for the disturbance. The equivalent change in controller output is the open loop error (Eo) divided by the open loop gain (Ko). The near integrator gain (Equation 6c) is the open loop gain divided by the open loop time constant (τo). The substitution of Equation 6c into Equation 6b cancels out the open loop gains giving the maximum ramp rate (Equation 6d) as simply the open loop error divided by the open loop time constant. The substitution of Equation 6d into Equation 6a yields Equation 6e where we see the deadtime from the sensitivity setting decreases as the size of the disturbance increases and the open loop time constant decreases, both of which cause a faster rate of change.

),( STw Min θθθ Δ= (4)

wT TΔΔ ∗= 5.0θ (5)


max)/%(5.0

tPVSm

S ΔΔ∗

=θ (6a)

)/()/%( max ooi KEKtPV ∗=ΔΔ (6b)

o

oi

KKτ

= (6c)

o

oE

tPVτ

=max)/%( ΔΔ (6d)

o

omS E

S τθ ∗∗=

5.0 (6e)

Similarly, the deadtime from valve threshold sensitivity (Sv) can be estimated (Equation 7a) as half of the valve threshold sensitivity divided by the maximum rate of change of the controller output that is initially mostly due to the proportional mode. The valve threshold sensitivity is the root mean square of the sum of the valve stick-slip, shaft windup, linkage backlash deadband, actuator resolution, and positioner resolution. The rate of change of controller output (Equation 7b) is the rate of change of the process variable multiplied by the controller gain (Kc). The controller gain (Equation 7c) can be approximated as the Ziegler Nichols gain multiplied by a detuning factor (Kx) to provide robustness and a smoother response (first term in Equation 7b). However, the gain should not inflict disturbances from noise induced fluctuations in the PID output exceeding the control valve threshold sensitivity (second term in Equation 7b). An enhanced PID threshold sensitivity setting or wireless trigger level expressed as measurement sensitivity (Sm) can reduce this noise allowing a higher controller gain. If the PID threshold sensitivity or wireless trigger level setting makes the second term smaller than the first term in Equation 7b, the full disturbance rejection capability of the PID with a smooth response is attained. The detrimental effect of an increase in loop deadtime from a wireless trigger level or PID threshold sensitivity setting may be less than the beneficial effect from an increase in controller gain by reducing the response to noise. The substitution of Equation 7c into Equation 7b cancels out the open loop time constant (Equation 7d). The substitution of Equation 7d into Equation 7a yields Equation 7e where the deadtime is proportional to the product of the open loop gain and original deadtime and inversely proportional to the product of the detuning factor and open loop error. This last equation is based on the enhanced PID threshold sensitivity or wireless trigger level preventing noise amplified by the controller gain causing the fluctuations in the controller output from noise larger than the valve threshold sensitivity making the valve to move.


max)/%(5.0

tCOSv

v ΔΔ∗

=θ (7a)

maxmax )/%()/%( tPVKtCO c ΔΔΔΔ ∗= (7b)

[ ]⎥⎦⎤

⎢⎣

⎡−∗

∗=

002.0),(max,min

mm

v

oo

oxc SN

SKKK

θτ (7c)

oo

x

KEK

tCO o

θ∗∗

=max)/%( ΔΔ (7d)

ox

oovv EK

KS∗

∗∗∗=

θθ 5.0 (7e)

Ei = integrated error for unmeasured load disturbance (% sec) Ex = peak error for unmeasured load disturbance (%) Eo = open loop error (loop in manual) for unmeasured load disturbance (%) Kc = controller gain (dimensionless) Ki = near integrator process gain (% per % per sec) Ko = open loop gain (product of valve, process, and measurement gains) (dimensionless) Kx = detuning factor for controller gain (dimensionless) Nm = measurement noise (%) Δ%CO/Δt = rate of change in PID % controller output (% per sec) Δ%PV/Δt = rate of change in PID % process variable (% per sec) ΔTw = wireless default update rate (update time interval) (sec) Sm = wireless measurement trigger level (threshold sensitivity) (%) Sv = valve sensitivity, resolution, or deadband (%) T63 = 63% process response time (sec) θo = original loop deadtime (sec) θΔt = additional deadtime from default update rate (sec) θs = additional deadtime from wireless trigger level (sec) θv = additional deadtime from valve (sec) θw = additional deadtime from wireless measurement (sec) τo = open loop time constant (largest time constant in loop, hopefully in process) (sec)


FLEXIBLE MANUFACTURING Just in time flexible manufacturing requires the minimization of inventories and better knowledge of production capability and transitions between product grades and production rates. Wireless measurements offer opportunity to find and employ what information is needed to meet real time production requirements. A plant can go to a new production rate within minutes by plant wide feedforward control. If all of the key flows (feed, utility, recycle, and vent flows) are set to the values on the Process Flow Diagram (PFD), the plant can move to a new production rate without waiting on composition, level, pressure, pH, and temperature loops to respond. Wireless measurements can provide a wealth of details on how to improve the efficiency (reduce raw materials, utilities, and waste), flexibility, maintainability, operability, profitability, and safety, all of which ties to together in terms of compliance and competitiveness. Plant wide feedforward can help achieve all of these objectives by not waiting on feedback correction to realize these goals while meeting a demanding market in terms of production rate and product grade flexibility. The PFD values may not be a simple ratio of the new to old production rate although this could be a good starting point. Better would be sets of flows captured at different operating rates and ambient conditions (e.g. summer versus winter operation) for the flow feedforwards. Process PID loops would as usual trim through feedback correction the feedforwards. The concept can be expanded to product grade changes and fed-batch operation for faster and more adaptive transitions and batches. The plant may have cascade control and several layers of loops but ultimately what is manipulated is nearly always a flow by means of a control valve, damper, or variable speed drive. Flows are the inputs and outputs of unit operations in the process industry. While Coriolis meters provide the ultimate knowledge in terms of mass flows and compositions via densities, wireless multivariable DP flowmeters with averaging pitot tubes can get an initiative prototyped. Wireless measurements can be put on every flow in a unit operation and the concept demonstrated to automatically move the unit operation to a new production rate by means of feedforward. When the benefits are realized, the wireless measurements can be moved on down the line. Wireless portable instruments can offer online tests historized and analyzed in the DCS of how much faster and smoothly you can move to new production rates and grades. If a plant can respond quickly to changes in raw material supply and market demands, the inventories can be reduced. If a plant can increase rates when utility costs are low (e.g.


off peak time rates) and ambient temperatures are low for cooling water, the energy efficiency can be increased. The advantages of adding flow measurements extend way beyond feedforward. The opportunities for plant knowledge, online process metrics, modeling, and data analytics greatly improve through measurement of unit operation inputs and outputs, the flows. An accurate material balance of the product tank farm enables the minimization of inventory and the coordination of upstream operations. A material balance requires measurement of level and the production flow required to keep the level at its optimum. The production rate requirement then becomes immediately implemented by the use of a plantwide feedforward system to rapidly achieve the new feed and manipulated flows. The plantwide feedforward puts each flow on the process flow diagram (PFD) at the right value for the new production rate. The proper flows are achieved by feedforward gains and biases corrected by feedback control. Level and pressure loops keep liquid and gas holdups constant. If the keeping the residence time constant is important for product formation or separation, proportional only (P-only) level control is used to correct the feedforward. If keeping the holdup constant is important for equipment operation such as heat transfer, proportional plus integral (PI) level control is used to correct the feedforward. Temperature and analytic loops are used to correct the feedforward for composition control. Normally the correction is a bias to eliminate scaling problems and the introduction of nonlinearity by multiplication of the feedforward [1]. The enhanced PID can be used with wireless measurements in the feedback loops to help suppress oscillations from feedforward timing errors by ignoring short term excursions. For example, if the temperature control point is closer to the top of a distillation column than the feed entry point, the feedforward correction of reflux flow may affect the control point before the feed flow causing an inverse response. A delay inserted in the feedforward may not be accurate for all operating conditions. If the delay is set too large, a secondary disturbance is created after the onset of the primary disturbance. A wireless temperature measurement and enhanced PID can help ignore residual timing errors. For long term feedback corrections representative of an error in the feedforward relationship, a feedforward input bias is adapted by an enhanced PID controller developed for wireless that seeks to make the bias correction from feedback control nearly zero. The smarter input bias keeps the feedforward gain more consistent by recognizing the manipulated flow may not be zero for zero feed flow due to process requirements and flow measurement rangeability and accuracy limitations. The use wireless multivariable DP flowmeters and averaging pitot tubes as insertion flow elements can provide a low cost and quick solution for missing flow measurements for plant wide flow feedforward [3]. If the wireless DP, pressure, and temperature transmitters are direct mounted on pipe connections, the flowmeter is


portable. This enables the prototyping of the concept and demonstration of the benefits for justification of a plant wide solution. The wireless flow can be easily moved from stream to stream. For product grade transitions, the concept is similar. In this case, the flows are set on the basis of what is defined by the PFD for the particular product grade. Temperature and analytical setpoints may also change as a result of the change in product composition. Setpoint feedforward and the enhanced PID can help these composition loops. DISTURBANCES The first flow that has significantly changed or started to oscillate is normally directly or indirectly associated with the root cause of a disturbance. If the original disturbance is temperature or concentration, the correction in the manipulated flow of a pressure, temperature, or concentration loop is the key to focusing on the right unit operation and loop. The most frequent unidentified disturbance is a change in raw material composition. Often this changes the utility or vent flow for temperature or pressure control, a reagent flow for pH control, and a recycle, reflux, or reactant flow for composition control. The deployment of wireless flow measurements can help track down disturbances, since the predominant inputs to processes are manipulated flows. The most common readily preventable disturbances are feed rate changes by the operator. Often these are seen at the start of a shift control room as operators go for their “sweet spot.” Undersized surge tanks, equipment performance problems, equipment constraints, manual startups, and manual actions are conducive to setting up a continual transient in plant operation. The automation of all manual and corrective actions to make them repeatable and predictable coupled with plantwide flow feedforward control can move process diagnostics from the basic to the advanced level by looking at changes in the flow ratios as indicative of changes in raw materials, operating condition, utilities, and ambient conditions. These disturbances and front end production rate changes typically propagate downstream so the furthest upstream loop with an oscillation is normally the culprit. However, when the flow is set to centrifuges and dryers, level control on slurry tanks or evaporators may manipulate a flow into the vessel. In this case, a change in back end production rate propagates upstream. Whether level or pressure control manipulates an inlet or outlet stream, determines whether a disturbance is propagated upstream or downstream. Since the disturbance is a readily observable flow change, the upset is easy to spot. In fact the best way of finding the source of any disturbance is the first flow with a significant change on a trend chart of all of the PFD flows, achievable with wireless measurements and an expansive trend or trajectory visibility on an improved operator interface. The time range of the trends (trajectories) of flows must be consistent and long enough to spot the initiating event.


Control valve stick-slip, shaft windup, deadband, and resolution can be quantified from the response of sensitive flow measurements. The fix may be as simple and quick as a digital positioner tuned for the application or as expensive and time consuming as replacing a fundamentally poor valve choice. If plant oscillations stop or their amplitude is reduced when the controller is put in manual, the suspect controller is confirmed as the culprit. Steps are then made in the controller output in both directions starting with a minimum size, such as 0.25%. The steps are increased in size until there is a consistent, uniform, and timely process variable response. Temperature and level measurements are too slow. Many pressure loops cannot be put in manual long enough to use pressure as a measurement to judge valve response. Many temperature and gas pressure loops have an integrating response like level loops, making identification of the size of the valve response problematic. Flow measurements provide the fastest and most direction indication of the valve response. If a flow measurement is not installed, a portable wireless flowmeter can be the solution for the tests. Since stick-slip can increase at lower production rates (operation closer to seat) and from stem corrosion and excessive tightening of the packing, periodic testing may be warranted. If the flow measurement is used to provide a flow loop, the stick-slip or resolution can be monitored in closed loop operation as being simply the amplitude of the saw tooth oscillation in the flow controller output. Noise, poor actuator-positioner performance, disturbances, and cascade control can complicate the identification of the limit cycle making tests in manual useful. EQUIPMENT PERFORMANCE Plants often consist of many production units and trains of equipment. The production units may produce raw materials as intermediates in a series of processes that lead to a predominant product. Often intermediates or byproducts are also sold. Trains of equipment feed into a common tank that may feed other trains of equipment. Recycle streams further complicate the process. The production usually culminates as a feed to an area for final processing, such as purification, drying, and packing. Equipment diagnostics and the automation of corrective actions are essential for these plants. The equipment diagnostic of widest applicability involves the measurement of heat transfer. The measurement of heat transfer can be done by a flow measurement and inlet and outlet temperature measurements of a principle stream to compute the net gain or loss in energy of the stream. For vessels, the stream is a utility such as cooling water or steam. For heat exchangers without reactions, measurements of a process stream can be used. For synchronization in the heat transfer calculation, the inlet temperature should be passed through a deadtime block with the deadtime equal to the residence time before the difference between the inlet and outlet temperatures is computed. The residence time is simply the volume between the inlet and outlet temperatures divided by the flow. The


heat transfer rate can be used as an inferential measurement of the rate of condensation crystallization, evaporation, drying, and reaction for monitoring unit operation performance. The integral of the rate is representative of the total amount of material condensed, crystallized, dried, and reacted. The rate is integrated over the residence time for continuous unit operations and the pertinent portion of the cycle time for batch unit operations. Probably, the largest opportunities occur in the monitoring of crystallizations and reactions. These rates may be limited by the heat transfer coefficients particularly if the process surfaces can be coated or fouled. The computation of the overall heat transfer coefficient (U) is complicated by the driving force for heat transfer being the log mean temperature difference between the two sides of the heat transfer surface and the variable heat transfer surface area covered in a vessel. While theoretically 4 temperature measurements for inline or plug flow equipment and 3 measurements for back mixed vessels are needed along with knowledge or computation of the heat transfer area (A) covered from level, an inferential measurement of UA can simply use the minimum difference between the temperatures of the process and utility stream known as the "approach temperature." For a relatively constant process load and utility supply temperature an increase in the "approach temperature" or an increase in the utility flow is indicative of a decrease in UA. Fouling can generally be measured by differential pressure measurements across a portion of a unit operation. An increase in differential pressure at a given flow is indicative of an increase in fouling of coils, filters, tubes, and trays. Fouling of heat transfer surfaces from chemicals in the process or biological growth in cooling tower water that decreases the heat transfer coefficient can show up as simply an increase in pressure drop on the fouled side. A more exact calculation would use the sizing equation for the flow coefficient (Cv) of a control valve. For liquid flow, the equivalent flow coefficient would use the square root of the pressure drop divided by fluid density as the numerator and the mass flow as the denominator. For vapor flow, the computation gets more complex. The rate of decrease in the flow coefficient is indicative of fouling rate. ONLINE PROCESS METRICS Process efficiency measurements involve the measurement of process flows and utility or raw material flows. The energy efficiency depends upon utility temperatures and heat transfer coefficients. The raw material efficiency depends upon feed concentrations and the desirable reaction rate to product and the undesirable reaction rate to a byproduct. For simply changing a stream’s temperature or pressure, the measurement of exit stream flow is sufficient. For changes in composition, the composition of exit stream composition is needed besides flow. The efficiency is proportional to the ratio of the exit stream component (e.g. product) flow to utility or raw material flow (e.g. kg product per kg


fuel gas or kg product per kg reactant). For synchronization, the utility or raw material flow should have dynamic compensation by means of an adjustable deadtime and filter time. For heat exchangers, inline blenders, dryers, kilns, and plug flow reactors just a deadtime equal to the residence time is sufficient. For columns, a deadtime and filter time equal to 15% and 85% of the residence time, respectively, is a good first approximation to the interactive time constants in series corresponding to equivalent trays. For well mixed vessels with negligible injection and conversion dynamics, a simple filter time constant equal to the vessel residence time can be used. Change of state dynamics can be negligible for condensation, evaporation, and drying. The conversion dynamics in crystallization and reaction can be significant with product or byproduct formation continuing downstream. Dynamic compensation can be adjusted manually by observing the response. If the raw material efficiency first decreases below its final value for an increase in raw material feed rate (inverse response), the deadtime or filter time should be increased. A notably sharp inverse response is indicative of insufficient deadtime. The result should be averaged to eliminate temporary excursions from timing mismatch. In sustainable manufacturing energy efficiency is often referred to as energy productivity. Raw material efficiency is often cited as yield. Wireless measurements enable the prototyping and evaluation of productivity, yield, and other online process metrics. LIVE PROCESS FLOW DIAGRAMS The first document on a project is typically a process flow diagram (PFD). The PFD defines the process. It is the ultimate source of information and sets the plant performance and design. It is interesting that we really don’t know how well existing operations match the PFD. In the PFD and in chemical engineering courses, the plant is assumed to be at steady state. Of course this does not work for batch processes. Less obvious is that it doesn’t work well for continuous processes with merging and diverging trains of equipment and recycle streams. Even if a plant was at steady state, the plant is rarely within 10% of the PFD design conditions on all of the PFD process variables due to non ideal and unknown effects in the process calculations or simulations that generated the PFD. To date process engineers manually update personalized spreadsheets in attempts to close the material and energy balances defined on the PFD. Coriolis flowmeters offer a true mass flow measurement that does not depend upon composition, density, velocity profile, Reynolds number, or viscosity. The physics of the measurement afford a rangeability and accuracy of mass flow that is unexcelled. Coriolis


also provides a direct density measurement, a tube temperature measurement, and when coupled with an accurate differential pressure transmitter (DP) for viscous fluids, an inferential viscosity measurement. Wireless multivariable DP flowmeters provide a more affordable and portable solution. For utility streams these wireless flowmeters are often sufficient and provide an accurate computation of mass flow since the composition is fixed and known. On process streams, these wireless flowmeters may provide enough diagnostic capability and if there is not enough justification to install a Coriolis meter. When a Coriolis meter is put on a stream, the only process variable missing for a live online PFD is pressure, which could be easily added via a wireless pressure transmitter. For streams with acids and bases, wireless conductivity and pH transmitters could provide additional information on stream composition. For example, in absorbers for CO2 capture, wireless pH and conductivity measurements in concert with a Coriolis density and temperature measurement can provide inferential measurements of solvent concentration and CO2 loading important for optimizing absorber flow distribution. A live PFD would open the door to much better recognition by maintenance, operations, process technology, and research of the actual performance of the plant and the source of abnormal operation and equipment problems by the use of data analytics. The additional information from wireless measurements enables more extensive and accurate inferential measurements of compositions from first principle, statistical (projection to latent structures), and neural network models. CONCLUSION The portability and connectivity of smart wireless transmitters creates new possibilities of using spare transmitters to get data into the DCS from lab testing and field calibration and for developing inferential measurements and selecting the best sensor technology and location. The elimination of wiring reduces the susceptibility to electrical noise especially for wireless pH. An enhancement to the PID algorithm provides protection for delays and interruptions in communication and offers a nearly perfect setpoint response when the default update rate is slower than the process response time and the reset time is decreased to be about the process response time. The PID can eliminate oscillations from valve backlash and stiction and loop interactions. Faster tuning is possible for valve position controllers used for the elimination of split range and the optimization of unit operations. The effect of wireless settings on the peak and integrated errors for unmeasured disturbances can be estimated. Opportunities for large scale sensor networks for making plants more flexible, maintainable, and efficient are extensive. The portability of wireless enables the demonstration of opportunities for process control improvements including online diagnostics, metrics, and live process flow diagrams.


REFERENCES

1. Essentials of Modern Measurements and Final Elements in the process Industry, ISA, 2010

2. “Opportunities for Smart Wireless pH, Conductivity Measurements”, InTech web exclusive, Jan-Feb, 2010 http://bit.ly/lY6q75

3. “Wireless – Overcoming Challenges of PID Control & Analyzer Applications”, InTech, July-Aug, 2010 http://bit.ly/jwDEgl

4. PID Control in the third Millennium: Lessons Learned and New Approaches, Industrial Applications of PID Control, Springer, 2011

5. “Feedforward control enables flexible, sustainable manufacturing”, InTech, March-April, 2011 http://bit.ly/emoSZg

6. Essentials of Modern Measurements and Final Elements in the Process Industry, ISA, 2010

7. “Advances in flow and level measurements enhance process knowledge, control”, InTech, Jan-Feb, 2010 www.isa.org/intech/20100204

8. ISA standard ISA-75.25.01-2000 (R2006), “Test Procedure for Control Valve Response Measurement from Step Inputs”, ISA, 2006

9. ISA technical report ISA-75.25.02-2000 (R2006), “Control Valve Response Measurement from Step Inputs”, ISA, 2006

10. “Is Wireless Process Control Ready for Prime Time”, Control, May, 2009 http://www.controlglobal.com/articles/2009/WirelessProcessControls0905.html

http://bit.ly/lY6q75�

http://bit.ly/jwDEgl�

http://bit.ly/emoSZg�

http://www.isa.org/intech/20100204�

http://www.controlglobal.com/articles/2009/WirelessProcessControls0905.html�

wireless measurement and control opportunities

Documents