THIS MONTH'S PUZZLER
We're trying to resolve problems with our new two-stage countercurrent solvent extraction process. We use 2,000 lb/h of nitrobenzene (NB) to extract aniline from a 5,000-lb/h wastewater stream generated in the production of aniline. The process is supposed to transform the wastewater, which contains 5-wt.% aniline, into 4,750 lb/h of raffinate with 0.01-wt.% aniline, and give 2,250 lb/h of aniline-laden extract; our maximum contract discharge concentration is 0.25 wt.%. We're experiencing water carryover in the extract, which exits the process at the first stage. We can't achieve our contract limit, let alone our goal of 0.01 wt.%. We see about 1-wt.% aniline at the discharge. At operating temperature, the density of aniline is 1.018 g/cc, NB's is 1.12, and we use 0.98 for water. A stable rag layer has formed in both extractors; increasing the speed of the agitators has no effect. What can we do to achieve our goal and reduce the effects of the rag layers?
BREAK THE RAG LAYER
The rag layer is an emulsion formed in the extraction process. High speed agitation will create it. As you increase agitation, more emulsion will form. You need to figure out which speed gives you the desired extraction. You also need to break the rag layer. I would pass the blend through a fine inline bag filter to break the rag layer. You might have to experiment with the type of inline filter; a fine mesh steel filter pad might work. This would be the cheapest method and I would try it first.
Girish Malhotra, president
EPCOT International, Pepper Pike, Ohio
LOOK AT YOUR DESIGN BASIS
Based on a simplified solvent extraction model, I think you should re-check your lab data. Your partition coefficient, m, looks too optimistic: you need an m of 92.01 for 0.01%; you've got 26.7 for 1%.
First, measure m on a bench scale. Begin with preliminary experiments to fill out your test parameters. Then develop a block test matrix; this factorial approach is tedious but you won't miss much. Consider the following for the preliminaries: inversion, residence time, side-stream filtration, heating the water, and salting out. Confirm positive results with additional testing: find out why an improvement happened. (Some options only can be tested at full scale.) Select the approach that gets you the lowest raffinate concentration of aniline at the lowest capital cost.
One option would be to increase the solvent flow to 12 gpm from 4 gpm; this likely will cause an inversion, with water as the dispersed phase instead of nitrobenzene. You'll have to look at the residence times in the extractors to ensure this won't cause problems. If residence time is a problem, try reducing the feed flow or batch the feed through several times. A rough pass with Goal Seek with m and solvent flow fixed yields a feed flow of 1,230 lb/h or 3 gpm, and a recycle of four (5,000/1,230) as a minimum. (The equation used in the iteration was: Ψ = (P -1)/(Pn+1 - 1), where Ψ is the % of solute in the raffinate (water), n is the number of ideal stages, and P is m Θ where Θ is the ratio of feed to solvent and m is in lb-solute in feed/lb-solute in extract.
Now, let's talk about the rag layer. It's a stable suspension of solids. Engineers often are as confused about rag layers as they are about pump cavitation. The so-called rag layer could be a liquid layer or a mixture of gas, liquid and solids. Sample the rag layer(s) and centrifuge it to collect solids, then dissolve them in acid, crystalize and examine the crystals under an electron microscope. If you don't have this equipment, try a series of filter papers after decanting the foam. These tests will tell you if filtration is practical: the finest screen available won't capture particles smaller than 37 microns; a sock filter is limited to 0.5 micron. Do the tests immediately but set some samples aside to view in a few days. Also take further samples during the bench testing to check for changes in the process.
Options available in the solvent extraction bag of tricks are worth mentioning. One obvious one is to heat the aqueous phase, adding a thermal driving component. Another is to add salt to decrease the solubility of aniline in the aqueous phase. This is called "salting out." It works by lowering the dielectric constant favoring ion-pair formation. Salting out could ease capturing the rag layers by precipitation; MgSO4, Na2SO4 show promise.
You also could add the salt during a subsequent treatment of the extract; this is called drying. The salt must be chosen carefully because aniline is a strong base and because systems — including water, amines, hydrocarbons and some salts, like rust — can cause foaming. A molecular sieve also can handle drying.
Full-scale options for improving the efficiency of the process include changing the baffling in the extractor to promote back-mixing; pumping out and filtering the rag layer then returning it to extractor; and looking at wetting inside the extractor — plastics are better at wetting organics, metal wets water. Increasing the agitation to improve mass transfer is a method demanding caution; amines have been reported to foam with air, solids and water present. One last option worth mentioning is to add another stage. If space is a problem, add a column extractor as an additional stage.
Dirk Willard, consultant
Our new two-stage non-lubricated reciprocating recycle compressor that supplies 93% H2 to our diesel hydrotreater suffered a catastrophic event after a troublesome 20 hours. The valve disc plate on the second stage inlet broke. An outside-operated, port-type five-step unloader controls flow. Our operator was alerted when the safety valve on the first stage blew; this is the third event. After inspection, we discovered the paint on the cylinder was smoking. None of the alarms warned us of this event: the thermocouple in the damper indicated normal. Afterwards, we decided to add a flow meter in the discharge line to monitor the compressor output. The direct-drive motor is 1,000 hp and operates at 400 rpm. The compressor capacity is 6,700 scfm. The suction pressures and temperatures are: 242 psig, 86°F for stage 1 and 427 psig, 95°F for stage 2. The discharge conditions are: 434 psig, 210°F for stage 1 and 768 psig, 230°F for stage 2. The safety valve settings are 470 psig for stage 1 and 840 psig for stage 2. The valve is a concentric ring type. We have had a few previous events with this compressor: 1) a leak in the second inlet valve seat caused the first safety valve to open — the valve was replaced; 2) a copper gasket between the valve and the frame was destroyed by the cage locking the valve, which caused the first stage safety valve to open — this valve is difficult to install but we replaced it. What do you think caused these problems? How do we improve monitoring?
Send us your comments, suggestions or solutions for this question by December 13, 2013. We'll include as many of them as possible in the January 2014 issue and all on ChemicalProcessing.com. Send visuals — a sketch is fine. E-mail us at ProcessPuzzler@putman.net or mail to Process Puzzler, Chemical Processing, 555 W. Pierce Road, Suite 301, Itasca, IL 60143. Fax: (630) 467-1120. Please include your name, title, location and company affiliation in the response.
And, of course, if you have a process problem you'd like to pose to our readers, send it along and we'll be pleased to consider it for publication.