19 replies to this topic

[vikas nagpal]

We plan to sparge liquid nitrogen in a batch reactor to control an exothermic reaction. How to estimate loss of solvent during above operation.

 

[breizh]

May be a bit off, but why don't you consider a vent condenser with liquid returning to your reactor ?

You cannot release solvent to atmosphere ( safety , environment and probably cost)

 

Breizh

[fseipel]

You would first calculate the vapor pressure of the solvent at the reactor's vent temperature/lowest temp the exiting gas is exposed to.Then you'd divide by the operating pressure -- this result is the saturation concentration of the solvent in the Nitrogen passing out the vent.  This is referred to as Dalton's law.  You can then multiply concentration by Nitrogen consumed, either by weight change in dewar or based on level transmitter readings if using a bulk liquid tank.

 

I have used liquid Nitrogen both directly (on tube side of a 'u'-tube heat exchanger/condenser), and indirectly (cooling a secondary heat transfer fluid).  I would say this is very much a niche application even in specialty chemicals.  I prefer a u-tube liquid nitrogen (LN2) condenser, to straight tubes -- less chance of deformation/failure of tube sheet on thermal shock at start-up since one side of tubes is free to move.

 

I take it you require a cryogenic operating temperature here to promote selectivity of desired product or some other purpose?  What is the reactor volume and solvent and cycle time?  What operating temperature is required? The LN2 condenser is great IF you can use a solvent system that boils at cryogenic temperatures -- e.g. methane, ethane -- if you can go higher ammonia may be attractive.  The condenser will give you more heat transfer per unit area than jacket cooling.  For instance, if you were targeting -150 C operating temperature and could use propane as solvent, the vapor pressure is only 0.03 mmHg at that temperature, so vapor concentration would be 0.03/760 = 0.00003947 mole fraction propane in the Nitrogen.  Latent heat of LN2 is ~86 BTU/lb so from heat of reaction you can calculate approximately how much LN2 is req'd, neglecting sensible heat (or include it).

 

Typically LN2 is not run in jackets, though it HAS been done.  Usually LN2 is vaporized in an external heat exchanger, to cool a secondary heat transfer fluid that is pumped/circulated through the jackets.  This way you can operate the reactors over a wide range of operating temperatures.  Be sure to check minimum design metal temperature (MDMT) on all equipment for conformance as embrittlement can occur at cryogenic temperatures.  LN2 piping should be vacuum jacketed if possible or at least well insulated.  Note a special heat transfer fluid MUST be used, because obviously, most fluids thicken beyond where they can be pumped at ~-196 C.

 

The use of an expendable refrigerant (LN2) MAY have merit for cryogenic operations but long term, a cascade refrigeration system is often going to be cheaper.

 

You can also compress the vent stream and condense it against cooling water -- essentially running a refrigeration loop on the process.  Obviously if Oxygen is present this can present some serious hazards.  I'm not sure if you have a volatile solvent present so it's difficult to make recommendations.

 

Needless to say the gas companies are a good source on information on this topic that compares the merits of direct injection, semi-indirect, and indirect, e.g. http://www.airproduc...id-nitrogen.pdf

Edited by fseipel, 07 December 2013 - 08:42 PM.

[curious_cat]

Great answer @fseipel. 

 

I just hope the OP has verified a real need for Liq. N2 for his particular reaction. Like you say it's a niche application and in my anecdotal experience people just too often casually propose "liq. N2" as a cooling solution to  a problem that doesn't need it at all. 

 

@vikas nagpal:

 

Can you provide more details of your application? Is it a new project or a retrofit? 

[fseipel]

curious cat: Yes, it's a real energy hog.  LN2 is about $1/gallon in the midwest/United States (bulk delivery).  You only get 86 BTU/lb refrigeration from LN2 latent heat (neglecting cooling if operating above 196 C), so, say, for a very small, 5000 BTU/hr load that's 5000 BTU/hr / 86 BTU/lb / 6.7 lb/gal x $1.00/gal = $8.67 per hour operating cost. Compare that to a chiller @ 0.83kW/ton x 1 ton x 5000/12000 BTU x 1 hr x $0.12/kw-hr = $0.04/hr operating cost.  And this doesn't even account for the 'blast' of gaseous Nitrogen you first see when you start up a system at 20 deg C and need to cool it down to -90 or -100 -- that's the 'big gulp' initial consumption to cool the piping down.  Even on a small system that may amount to 50 gallons or more LN2 in the first operating hour.  Of course, the gas companies never tabulate that in their full-color brochures.  It's true you can use the vaporized gaseous Nitrogen produced from the LN2 cooler elsewhere in a plant, but how much use do you have for that in purging? Typically most will be vented.

 

I haven't seen it used where it wasn't needed in my experience.  Typically I've used it when we ran short-term projects and it took too long to get an air permit (the LN2 would get us under the local daily emission limits for which a permit was required).  I've seen it used for long-term projects where it was needed due to the chemistries.  In that case it was a bad choice due to high operating cost -- i.e. the low-temperature chiller would have been paid off in a year or two.  If a project becomes long-term, replacement of LN2 seems VERY attractive due to the high operating cost.  I have not seen a CHEAP alternative, that is, a pre-engineered, packaged, low-cost, cascade refrigeration system larger than a lab unit.  Or even a low cost chiller operating on hexane that can attain reasonably low temperatures in a single compressor. Edwards Engineering & others make them, but you have to be willing to pony up $100,000+.  Any recommendations on vendors for these from any posters are welcome!  I'm talking about 0.5 - 2 ton capacity, -100 C and below.

Edited by fseipel, 08 December 2013 - 04:26 PM.

[vikas nagpal]

The reaction is to be carried out at 5-10 deg celcius in a batch reactor of 3000 ltr capacity.The reaction is to be completed in two hours. Heat of reaction is 1,20,000 Kcal. The reaction volume is 1500 ltr. The heat transfer area available through reactor jacket is not sufficient to remove the heat.We do not have any other low temperature utility at site. The project is a retrofit and has to be done in next 2 months.We can not use external heat exchanger to remove heat of reaction due to safety concerns. We are evaluating installing internal coil in the reactor to improve upon heat transfer area.We are discussing with Praxair on using D Cool/ICool technologies.

[curious_cat]

@vikas

 

Thanks for details.

 

@fseipel

 

This was exactly what I was suspecting. The OP's reaction is to be carried out at 5 C. 

 

At this temperature can the economics ever work out to justify liq. N2? I don't know but I'm skeptical.

 

You might have some idea? 

[vikas nagpal]

@curious_cat

 

We have no option but to go for liquid nitrogen due to timelines of the project. We may shift to conventional referigeation system at a later date

[curious_cat]

@curious_cat

 

We have no option but to go for liquid nitrogen due to timelines of the project. We may shift to conventional referigeation system at a later date

 

Have you mentioned what your solvent is? Like @fspeiel showed above you need to know what solvent to estimate the loss. 

 

Have you tried calculating how much liq. N2 you'd need? What's your answer? I'm getting a fairly large amount. 

[vikas nagpal]

@curious_cat

 

The solvent is Tetrahydrofuran. Quantity of liquid nitrogen required is 2500 kg/hr

[curious_cat]

@curious_cat

 

The solvent is Tetrahydrofuran. Quantity of liquid nitrogen required is 2500 kg/hr

 

THF vapor pressure @ 10 C is ~110 mm Hg. That'd mean you are losing ~15% (v/v) in your N2 

 

You sure seem to need a vent condensor. 

 

What's your vent size BTW? That's a lot of N2 evolved. 

Edited by curious_cat, 12 December 2013 - 12:16 AM.

[fseipel]

curious_cat: Yes, I didn't realize LN2 was being proposed for such a high temperature -- I've never heard of using it for a temperature that is attainable with conventional refrigeration in a single compressor.  2500 kg N2/hr x 2 hr x 1 kg mole N2 / 28 kg x 0.15 kg-mole THF / kg-mole N2 x 72 kg THF/kg-mole THF = 1928 kg THF / 0.88 kg/Liter = 2,191 Liters = more THF than you started with in a 1500 Liter charge to a 3000 Liter vessel.  I think renting a portable air-cooled chiller, ~50 tons and circulating glycol to a vent (reflux) condenser is a good option.  At 2500 kg/hr x ~200 kJ/kg = ~500,000 kJ/hr x 0.9487 BTU/kJ / 12000 BTU/ton refrigeration = ~40 tons refrigeration.  This amounts to ~3294 lb ice/hr melted -- unfortunately quite a bit (dumping ice in a water tank circulated through a vent condenser would require a great deal of ice).

[curious_cat]

@fseipel

 

I had suspected something like this from his first post. Somehow people get attracted to liq. N2 when they see a large exotherm even when it doesn't need a cryogenic temperature. 

 

I think partly the liq. N2 manufacturers are pushing such applications to widen their markets. 

 

3294 lb of ice per hour isn't too bad when viewed against the fact that his other alternative was dumping ~5000 lb/hr of liq. N2, eh? Just think about the differential cost & safety & ease of material use etc. 

Edited by curious_cat, 12 December 2013 - 11:56 PM.

[vikas nagpal]

Sorry. I gave you wrong figure of 2500 kg/hr. It wd be ~ 1000 kg/hr

 

Refrigeration capacity of LIN = 47+1*0.25*(-190-(+10) ~90 k Cal.

Theoretical LIN requirement = 60,000/92~ 1000 kg/hr considering the process

efficiency of 65% (excluding the onetime cooling of the liquid nitrogen transfer line).

 

We have decided to go for an internal coil in the reactor with liquid nitrogen as utility. So no losses of solvent vapors

[curious_cat]

I have no idea what you mean by process efficiency. 

 

My calculation shows

 

Heat load in kJ / hr = 120,000 kcal / (2hrs) *4.18

 

Latent heat of liq. N2 = 199 kJ / kg

 

Liq. N2 needed = 1260 kg / hr

 

Can you explain your calculation? Are you taking credit for the specific heat of liq. N2? I'm not sure you can typically for a coil vaporizer. I may be wrong. 

[curious_cat]

PS. I've never tried coil cooling with liq, N2 but since THF has a -100 C freezing point how do you prevent a frozen sheet from forming on the coil surface? Is there anything else in your solvent system? Products? Water? Anything liable to freeze on the coil at those temperatures? 

[fseipel]

Coil cooling with LN2 works -- I've done it -- when used on a vent stream (to avoid permitting, keep emissions below 10 lb/day threshold), there was ice formation on the coils but very little in this application.  This is a different application, with coil submerged in liquid.  I am assuming tank is mixed, so that THF won't freeze as long as tank is agitated and there is some coil spacing to allow flow.

 

Of course, our LN2 coil made a huge noise when first started up until the coil cooled down.  In that case the the throttling valve was placed on the outlet so the entire line was filled with LN2.  It may be worth considering including a silencer.  Also be careful about the coil supports; if it is a helical coil, its contraction on cooling and subsequent growth on heating, may cause tubing rupture if rigidly supported at many points.  Try not to shock cool the coil.  May be better to have guides to allow expansion/contraction.  Even then, tubing can rub against guides causing failures.  If the coil tubing or pipe fails, can reactor vent handle the load and will there be liquid entrainment?  LN2 can condense O2 from the air, so don't run it until reactor is properly purged.  Liquid O2 can explode upon contact with organics.

[curious_cat]

I am assuming tank is mixed, so that THF won't freeze as long as tank is agitated and there is some coil spacing to allow flow.

 

Interesting. I'd intuitively have expected that with your coil surface ~100 C colder than your reactor content's freezing point it'd be hard to avoid a thin film freezing up unless you provided really high velocity mixing. 

 

Will a typical agitator on a 3000 L vessel be enough to avoid any stagnation & a film velocity high enough that the boundary layer does not initiate freezing? 

 

Of course, I'm only speculating so either experience of a detailed model would tell the true story. You obviously seem to have a lot of experience with this. 

Edited by curious_cat, 15 December 2013 - 06:57 AM.

[curious_cat]

 If the coil tubing or pipe fails, can reactor vent handle the load and will there be liquid entrainment? 

 

One other interesting failure mode to do a HAZOP on might be agitator failure. I suspect that'd really be between a rock and a hard place:

 

If you decide to  drain the liq. N2 / stop flow, you face a potential runaway since it's a highly exothermic reaction. 

 

OTOH, if you don't your coil surface probably quickly freezes solid. At which point you lost your heat transfer efficiency & worse might even damage your coil. 

 

If it's a semi batch, you might be lucky: stop a reactant flow.

 

If true batch with slow kinetics & a sustained exotherm then it's going to be tricky. 

[fseipel]

curious cat: You can control temperature with LN2, though in a vent condenser, icing can be a problem -- especially when a system is shut down if it draws in humid air -- if personnel don't monitor it, vent can be plugged -- typically, if vent condenser is up and running, condensed organic solvents melt the ice -- typically I've run them at -60 to -70 C -- above the solvent freeze points.  If you want tight temperature control, I tend to prefer an external media loop (e.g. Syltherm/AGA Cumulus system for cryogenic reactors).  With that scheme, you can dial in a setpoint temperature just above the freeze point of the organic solvent(s).  Some reactions do not produce the desired product distribution unless run at low temperatures, and other reactions, such as liquid ammonia, require very low condensing temperatures IF run atmospheric pressure.  Syltherm is a special silicone heat transfer fluid so it is pumpable at -100 C -- most heat transfer fluids are not and either get very viscous or freeze outright.  The downside is that it adds another pump and heat exchanger. And you have to cool all the Syltherm fluid down at start-up, so additional LN2 is consumed that isn't used by the process.  In my opinion, it's safer, though, because you don't have the coil under high pressure and you don't have to worry about the coil rupturing and venting into the reactor.  You also don't have to worry about freezing solvent and condensing LOX out of the air.  The [reaction mix/reactor] thermocouple or RTD should not be at the wall or near the coil.  In general, I prefer amperage monitoring to confirm agitators are running, and temperature monitoring be through bottom valves rather than baffles.  Reactant addition block valves, are then closed if agitation is lost.  Multiple temperature measurement points are even better still.  You also have to be very careful if running LN2 lines external to reactor, especially temporary ones -- if the lines rupture, someone can be frostbitten or deprived of Oxygen. I don't know how others 'meter' liquid Nitrogen, but I think it's easier to put a rotameter on the vapor side (vent) and/or weigh the dewar to know how much is being fed in to confirm it matches the exotherm.  You can check for frozen THF through a sight glass.  Again I don't think this is a great idea even for a short-term project, liquid Nitrogen ought to be used for temperatures you can't easily obtain with a single compressor chiller.  If it were me, I'd rent a small portable chiller (probably air-cooled) and attach it to a cheap plastic tank with brine or a glycol/water mixture and circulate the mix through reactor coil or jacket.  I'd probably undersize the chiller, and over-size the cooling fluid tank, to have a big thermal flywheel and keep the circulating chiller rent down.

You may have frozen THF on the coil, but with the liquid THF around it, it will melt quickly, akin to throwing ice in 100 C water bath.  This assumes you are not horsing too much LN2 through the coil AND you have agitation, hence the need to meter it.  Since exotherm is known, it can be converted into SCFM N2; with OP's estimate of 1000 kg/hr then the meter should be set for ~500 SCFM .  Of course, you need a long-stem valve (cryogenic service) where the packing gland doesn't freeze up to throttle LN2.  In the vent condenser application, I also included a temperature monitor on the coil outlet/vent.  If it gets down there to -200 or you see LN2 rather than vapor spewing out the vent, you need to throttle it back.  And I think you're right, it sounds like bulk gas suppliers ARE marketing this for inappropriate uses. Per my earlier post, operating cost is ~216 times higher for LN2 vs compressor (will vary with electricity rates but good ballpark figure). Relief also needs considered since if you shut valves at both ends of LN2 line you'll develop 2500 psi quickly.

Edited by fseipel, 15 December 2013 - 03:49 PM.