Tuning pH control loops

The titration curve is the problem

pH is a logarithmic measurement of a process whose response is anything but linear. Near the equivalence point a tiny amount of reagent moves the pH enormously; away from it, large additions barely register.

That means the process gain the controller sees changes by orders of magnitude depending on where the process is sitting. A single tuning set cannot be right across that range — it will be sluggish at the ends and wildly unstable in the middle.

Get the actual titration curve for your stream if you possibly can. It tells you where the difficult region is and how wide it is, and no amount of trend-watching substitutes for it.

What usually fixes a pH loop

• Better mixing, so the measurement sees a representative sample rather than an unmixed stream.
• A shorter, faster sample path, since sample dead time sets the performance limit.
• Correctly sized reagent valves, often two in a coarse and fine arrangement, because a single valve cannot cover the required turndown.
• Signal linearisation against the titration curve, so the controller sees something closer to a constant process gain.
• Staged neutralisation across more than one vessel where the swing is large.

Questions that come up

Why does my pH loop hunt no matter how I tune it?

Most likely because the process gain where it operates is far higher than where it was tuned, or because reagent dead time exceeds what the tuning assumes. Both are structural. Detuning enough to stop the hunt usually leaves a loop too slow to hold pH at all, which is the trade-off that makes these loops so frustrating.

Related