Home
 
Search
   
 
Contact Us
   

PID Loop Optimizer

  
   
Sign up for an educational Webinar.

Sign up for Tuning Tips from our Newsletter.
     
 
 
LIBRARY MENU
Table of Contents
Performance Supervision
PID Loop Optimization
Basic PID Tutorials
 
MORE RESOURCES
Webinars
Training
Newsletter
   
 

Neutralize pH control system instabilities

Achieve good control with a well-designed physical plant, linearized signal and optimal PID tuning

by Greg Shinskey and John Gerry

March 01, 2002

  

Reprinted with permission of Plant Services. This article appeared in the March issue of Plant Services.
When you're under the gun to tune a three-mode pH controller, relying on guesswork and continuous adjustment can be stressful and downright futile. But, it doesn't have to be this way.

The challenges
Controlling a neutralization process can be challenging for many reasons. It usually involves mixing two materials having widely different volumes or flow rates. Unless the mixing process is intimate, all bets are off. Probe placement also is critical to getting a meaningful measurement. The product or effluent flow can change radically without notice. The sensitivity of the pH measurement to changes in the manipulated variable (or controller output) is usually extremely non-linear. A small change in the flow of one reagent might cause only a tiny pH change—or it can peg the meter. The whole business can be quite perplexing.

Meet the challenge
Good pH control starts with good physical design. Several factors must be considered when designing a neutralization vessel:
  • The volume should provide five minutes of holdup for soluble reagents, such as caustic, and 20 minutes for slurries, such as lime.
  • The dimensions of length, width and depth should be similar.
  • Reactants should enter at the top on one side and product should be withdrawn from the bottom of the other, a configuration that minimizes short-circuiting.
  • Measure pH at the exit, but still within the mixed volume. If a flowing sample must be withdrawn, keep tubing short and velocity high to minimize dead time.
  • Agitate the fluid with a high-speed (no gear reduction) propeller or axial turbine to maximize shear within the vessel. The mixer's pumping rate should be about 10 times the maximum feed rate. Use baffles to avoid vortex formation.
  • Injecting reagents at the feed point promotes uniform distribution and mixing. The transfer line from each reagent valve should have a loop seal to prevent it from draining. Flow should start as soon as the valve opens and stop as soon as it closes.
  • To avoid releasing off-spec effluent, provide downstream capacity to impound out-of-spec effluent for recycle until it meets specifications.

Valve selection
The reagent valves should be sized to deliver enough flow to neutralize the maximum anticipated load. Although the titration curve is nonlinear, characterizing the reagent valve will not compensate. It should have a linear characteristic.

In many installations, the feed pH oscillates on both sides of neutral and requires both acidic and basic reagents. The pH controller output then must operate the valves in split range, opening the acid valve as output moves from 50 to 100 percent, and opening the base valve from 50 to 0 percent output. Because both valves should fail closed, the base valve requires a reverse-acting valve positioner.

Because of the high slope of most titration curves near the neutral range, precise reagent delivery is important. Digital valve positioners are recommended, as they are both fast and precise.

If the feed is delivered by a pump that cycles on and off, configure the system to shut the reagent valves and switch the pH controller to the manual mode when the pump stops. This interlock prevents reagent flow when no feed needs treatment and prepares the system for bumpless return to automatic operation when the pump restarts.

Linearize, linearize, linearize
The strongest weapon for managing the extreme non-linearity inherent in pH control is the titration curve. It illustrates how the pH changes as a function of the volume of reagent added.


Figure 1. Most pH loops are non-linear.

When reagent flow first starts, the pH changes only minimally. This results in a low process gain. But, as more reagent is added, the pH suddenly changes by a large amount, resulting in a high process gain. The titration curve can work with the pH measurement to linearize the loop completely.

Use the titration curve to generate equations that linearize the loop in a piecewise manner. This is much easier to do with appropriate software. If you attempt to generate these equations manually, be sure they have a positive slope, so that the controller's action is the same with or without the linearization equations.

Linearizing the loop begins with asking your lab or quality assurance department for a titration curve for the feed material. Use data points from the curve to produce the linearizing equations for the pH signal.


Figure 2. Linearizing equations generated by analysis software.

Because the setpoint is the controller's target for the desired pH, both the pH signal and the loop's setpoint must be characterized. Linearizing pH equations assume the pH signal is scaled between 0 and 100 percent. If the controller had previously been scaled for pH, the controller must be reset. Check to be sure the controller's minimum and maximum are set to 0 and 100 percent.


Figure 3. Linearizing equations condition both the pH signal and the setpoint.

The linearized signal is an intermediate calculation that will look like gibberish to operations people. So, don't display it. Instead, display the actual pH signal before it passes through the linearization equations. The same is true with the setpoint entry. The setpoint value must be fed to the linearization equations. The outputs from these equations go to the controller's setpoint, but don't display the actual setpoint because it will appear strange.

PID tuning—the icing on the cake
The last step in achieving positive pH control is tuning the PID controller. If the previous steps are done properly, tuning becomes the easy part, especially when using appropriate software.

Induce a brief, intentional bump while operating in either the auto or manual modes. For example, move the setpoint up and then back down. Feed the resulting process data into the analysis software to determine the proper tuning variables. Analysis software lets you perform this test in automatic mode while causing the least amount of loop disturbance.

However, you probably won't have tuning parameters sufficient for the loop to operate in the automatic mode. In this case, with the loop in manual, jog the controller output using a pulse or doublet pulse.


Figure 4. Fast pulse test data taken in manual mode on a linearized pH loop.

For example:
  1. Let the pH signal stabilize in the manual mode.
  2. Decrease the controller output by 10 percent.
  3. Wait 15 seconds and increase the controller output by 20 percent of its original value.
  4. Wait another 15 seconds and decrease the controller output by 10 percent of its original value. This returns the controller output to where it started.
  5. Let the pH signal re-stabilize.
The analysis software processes the data to determine the optimal PID tuning variables. The advantage of this test is that the net change in the amount of reagent added is zero, which results in the smallest possible loop upset.

Greg Shinskey and John Gerry are with ExperTune Inc, Hartland, WI. They can be reached at +1 (262) 369 7711.

Figures: ExperTune Inc


 
© 1999–2008 ExperTune Inc. Lake Country Research Center 1020 James Drive, Suite A Hartland WI 53029-8305 USA
Telephone +1 (262) 369 7711 • Fax +1 (262) 369 7722