Wednesday, February 13, 2013

Process Wednesday: pH adjustments on scale

Adjusting pH in a 2 liter flask in the lab is easy, but what about doing it in real time in the plant? Francis X. McConville has something to say about that in "The Pilot Plant Real Book":
For example, take a look at the lag time in the response of acid/base pumps or valves, and the real-time titrant delivery rates... If pH adjustment in an agitated vessel is the goal, ensure that agitation and turnover rates are sufficient in comparison to the rate of titrant addition. It may be necessary to add the acid/base to a separate closed circulation loop to ensure better distribution in very large tanks. It may simply be that the adjustment has to be undertaken slowly to allow sufficient mix time. pH control of slurries is particularly difficult.  
The relative location of the pH probe and titrant addition lines will also have a major effect on the operation of the control loops. Poor placement can cause long response delays or, worse, huge pH fluctuations. Try to determine the actual time required for a complete feedback cycle to be completed and compare this to the flowrates of all the streams involved. And finally, pay attention to the titration curve of the system you are trying to control. pH titration curves are not linear, and if you are trying to control pH on the steepest part of the curve, where small adjustments can mean major pH swings, it will only magnify your control problems. 
I have had some small experience observing the adjustment of pH in large stirred vessels (50 gallon) and above. Patience, I think, is a real virtue in these situations -- back-titrating after you've rocketed past the pH you were trying to hit is really annoying. Slurries, as Mr. McConville notes, are even more problematic and slow to respond.

I think it points to the general undesirability of pH adjustments on scale, especially ones that rely on judgment and metering in small amounts of titrant. It's probably just better to have a process where you add a known amount of titrant and let the agitator do its work.

I'm sure I'm being wrong-headed here somehow -- thoughts?


  1. Selecting the sampling point in tanks that contain 100s or 1000s gallons of solution is crucial in getting an accurate pH measurement. In a 1000-gal tank with a good recirculating pump, our minimum wait time to get a homogeneous solution was 15 min.

  2. pH adjustments at large scale can be problematic at times where often entails straightforward engineering issues that are mentioned. However the process chemist may be able to help modulate certain control elements by good selection of the acid or base and possibly concentration variables during lab studies. HCl and NaOH are favored, for legitimate reasons, but on occasion a weaker acid or base choice can provide refined control that may be more forgiving. I have used H3PO4 or AcOH as acids and K2CO3 as base (CO2 evolution manageable with little foaming) in plant settings to avoid over shooting a critical pH range that was hard to achieve with more common choices (with Patience was a luxury did not have).

  3. Would it not be advantageous for the larger reactors to have multiple dosing ports located at different sections/depths of the reactor (all fed through a single loop)? Maybe this is already the case. 15 min. feedback on a 1000 gal tank doesn't seem too bad, but still. How big of a concern is localized concentration at such scale?

  4. Biopharma does this at scale (2500 - 12000L). You actually want to minimize number of ports in a pressure vessel to minimize the number of leaks and in the case of biopharma, the points of entry where contaminants can skunk your batch.

    It ends up being a mixing problem more than anything, agitator design helps but is expensive and you have to know the CFD software reasonably well. Then shear is an issue, in biopharma we add shear protectants and antifoam to deal with it. Blessedly I have not had to deal with titration of anything particularly viscous.

    Depending on how you are titrating, if you are using carbonate based buffering you can have some % CO2 in the overlay gas to help control a bit.