pH is the most common analytical measurement. pH is important for product purity and environmental compliance. The most stringent pH control requirements occur in bioreactors for biopharmaceuticals where a deviation of 0.05 pH can cause a noticeable degradation in mammalian cell growth rate and product formation rate. Fortunately such tight control is possible. For waste treatment systems the allowable deviation in pH is much larger but the degree of difficulty is much greater due to extreme nonlinearity and sensitivity and the consequential extraordinary control valve rangeability and resolution requirements. A lot has been recently learned to address challenging pH control applications.
The slides noted in this blog are in the ISA Automation Week 2011 tutorial "pH-Measurement-and-Control-Opportunities" The subject is worthy of a book. It just so happens I have written one. For a detailed understanding see the ISA book Advanced pH Measurement and Control - 3rd Edition
For pH control systems in chemical unit operations and in water and waste treatment, the minimization of deadtime is most important. The smaller the deadtime the smaller the excursions on the titration curve reducing exposure to the nonlinearity from the drastic change in slope of the titration curve.
The biggest source of deadtime in well-designed equipment is reagent injection delay. The normal mechanical design of feed addition that puts a dip tube down to the eye of the impeller introduces horrendous delivery delays of an hour or more on starting and stopping reagent flow due to the extremely small reagent flow of concentrated reagents and the relatively large dip tube volume due to the minimum diameter set for structural integrity. Liquid reagent delay can be reduced by reagent dilution, short dip tubes, and isolation valves mounted on the equipment or piping nozzle. For chemical reactors and neutralizers with recirculation, injection of reagent into the recirculation line just before entry into the vessel minimizes reagent injection delay.
The next biggest source of deadtime is fouled or aged electrodes that has causes the measurement time constant to increase from seconds to minutes. Slide 24 shows that the 86% response time of a pH electrode not designed for high temperature increased from 10 seconds to 50 minutes because of premature aging caused by high temperature exposure.
The slope of a titration curve for strong acids and bases can change by a factor of 10 for each pH unit deviation from the neutral point. To help an adaptive tuner, signal characterization is used to translate the controlled variable from pH to percent reagent demand. The titration curve shape can change with temperature and composition so the linearization is far from perfect but can reduce the nonlinearity by several orders of magnitude. An adaptive tuner compensates for changes and unknowns in the titration curve, changes in loop deadtime and time constant, and other nonlinearities such as the control valve.
The rangeability and precision requirement is incredible for strong acids and bases. Theoretically a 7 pH setpoint in a true strong acid and strong base system, an influent stream that can change between 0 to 6 pH and an influent flow that can change by factor of 10, would require a control valve rangeability of one million to one and resolution of one hundred thousandth of a percent for control within a 0.1 pH. Such valves don't exist so the strategy is to use several stages of neutralization in series, coarse and fine valves in parallel, and pray for some weak acids and weak bases. Slide 54 shows how to estimate the control valve rangeability and resolution requirement from the titration curve.
For bioreactors, the disturbances are slow and the titration curve is rather flat and changes in slope are greatly moderated by carbon dioxide and weak acids and bases. The main challenge is noise, discontinuity of the split range point, and unnecessary crossings of the split range point.
Noise can be mitigated by avoiding the predominant path of bubbles from spargers, not installing electrodes in flashing streams (e.g. pump suction), using middle signal selection of 3 pH transmitters, and installing wireless transmitters to eliminate ground loops. Slide 38 shows how a wireless transmitter eliminated the spike seen in the wired transmitter from electrical magnetic interference (EMI).
Unnecessary excursions of split range point can be reduced by the use of analog output setpoint rate limit (velocity limit) if the PID external reset feedback (dynamic reset limit) is enabled. An enhanced PID developed for wireless with a threshold sensitivity limit increased near the split range point can also reduce excursions. For bioreactors, it is particularly important to minimize the unnecessary addition of sodium bicarbonate because the accumulation of sodium increases cell pressure leading to cell membrane rupture and death.
The discontinuity of the split range point can be reduced by the use of a fine valve in parallel with a coarse reagent valve so that the transition at the split range point is between precise valves. An enhanced PID can be used here for more effective valve position control (VPC) where the VPC manipulates the coarse valve to keep the fine valve in good throttle range. As the pH controller approaches the split range point, the VPC naturally closes the big valve. An enhanced PID and a threshold sensitivity limit can help stop limit cycles. A directional AO setpoint rate limit and external reset feedback can help the VPC respond quicker when it needs to get away from the split range point.
Pulse width modulation of an inexpensive solenoid actuated ball valve can be used in series with a throttled valve to extend the rangeability requirement and prevent the control valve from plugging for extremely low flow requirements by a low output limit to the throttling valve (slide 69). For low viscous flows a roller diaphragm controller valve provides a rangeability of 1000 to 1 or more and enforces laminar flow so that the valve sizing and performance is predictable.
The following checklist is not intended to cover all the application requirements but some of the major details to be addressed.
•1. Are the items in the Control Talk Blog "Checklist for pH Measurement" addressed?
•2. Are wireless transmitters used to eliminate EMI spikes?
•3. Is the reagent injection delay minimized by a shorter dip tube, reagent dilution, or injection into a recirculation line?
•4. Is the vessel considered to be well mixed with a turnover time of 10 seconds or less for chemical systems and 1 minute or less for biological systems?
•5. Does the electrode location prevent bubbles from touching the electrode?
•6. Does the electrode location velocity prevent coating of the electrode in fouling streams by maintaining a velocity of 5 fps or more past the electrode?
•7. Does the electrode design prevent premature aging from high temperatures or chemical attack?
•8. Has a titration curve from individual samples at operating temperature with at least 20 data points in the control region been generated for all operating cases (not a composite sample)?
•9. Has the control valve rangeability and resolution requirement been estimated from the titration curve for the worse case condition of maximum flow, steepest slope, and furthest distance from the setpoint on the titration curve (extreme values of influent pH)?
•10. Can pulse width modulation be used to prevent valve plugging for extremely low reagent flows?
•11. Is a roller diaphragm control valve needed for small viscous reagent laminar flow?
•12. Are the items in the Control Talk Blog "Checklist for Control Valves" addressed?
•13. Are several stages of neutralization required for waste treatment based on titration curve?
•14. Can an inline pH control system be used for the first stage of neutralization?
•15. Are parallel fine and coarse valves and valve position control used to increase the rangeability and resolution and reduce the discontinuity at the split range point?
•16. Can an analog output directional rate limit (velocity limit) and external-reset feedback (dynamic reset limit) be used to reduce unnecessary excursions across the split range point?
•17. Can an enhanced PID be used for valve position control to eliminate reaction to limit cycles?
•18. For setpoints on steep part of a titration curve can the pH nonlinearity be reduced by signal characterization to convert the PID process variable (PV) from pH to the X axis of the titration curve scaled to be 0 to 100% (PV and SP scaled 0-100 % reagent demand)?
•19. For extremely nonlinear titration curves, are cascaded signal characterizers used to provide sufficient resolution for linearization?
•20. Is an adaptive tuner used to compensate for unknowns in the titration curve, valve nonlinearity, and changes in the process deadtime and time constant?
•21. Is an open loop backup for an inline pH control system needed to prevent RCRA pH violations as shown in slides 118-120 in the ISA Edmonton short course "Effective Use of Measurements, Valves, and PID Controllers"?
•22. Is pulse width modulation and proportional-derivative control needed to prevent overshoot for batch neutralization where the pH response is only in one direction (e.g., only one reagent and no reagent consumption or evolution)?
To see all of my checklists and more go to my Control Talk Blog main page
