Mixing Storage Tank By Circulation

[BenjaminM]

We have a storage tank for a finished product. We may mix a couple of different batches in the tank. We circulate the tank with a pump to make sure the different lots are homogeneous before samplings or packing out to tank trucks. Our production manager wants to circulate the tank for a hour. Our QC manager wants to circulate the tank for a day. I am pretty sure the actual needed answer is somewhere in the middle.

I've got the volume of product in the tank, and the circulation pump rate.

I can appreciate the viscosity of the material should have some affect on the mixing. In this case it is thick (think hot maple syrup).

Are there any standard methods for determining how many times the tank should be turned over / duration of circulation for good mixing?

 

[jari001]

What are the volumes of the tank? Are there particles that need to be fully dispersed or is the issue making sure that nothing separated because of variations in density/temperature? How different are the possible batches that go into the tank in terms of properties and operating conditions?

Also checkout thread407-402919 and thread124-108085, maybe there are some answers there.

 

[EdStainless]

I presume that the pump is moving the product from one side of the tank to the other.
The scientific way to do this is to add a very small amount of a tracer chemical and then take samples at regular intervals to see when it becomes uniform. This is a common method in some processes.
If you want a place to start then how long would it take your pump to move the entire tank volume 7 times?
We all have a gut feel that the correct answer here is something like a few hours. Now prove it.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy

 

[SuperSalad]

If you are maintaining the multiple lots for the final product, does it even matter if the mixing is perfectly uniform?

Or is part of the goal here to create a new, larger single lot of material with one set of results?

If yes to the latter, I'd think you are better off pulling periodic samples during a set time period and tracking the results until they become consistent. Then base all future batches on that test case.

It is a less sophisticated method and will take a little time for sure, but the accuracy of the results will be fairly unquestionable and there's no guesswork or modelling to be relied upon.

Andrew H.

http://www.mototribology.com

 

[georgeverghese]

Duration for circulation will depend on viscosity and circulation rate for this high viscosity maple syrup. Circulation rate at the pump may be affected by viscosity also. Viscosity will depend largely on bulk tank temp. Is bulk tank temp at the start of mixing the same for all cases?
Generally speaking, trying to achieve a homogenous mix by recirculation is not as effective, time and energy wise, as running a built in motor operated paddle mixer.

 

[Latexman]

No data. No answer.

Why don't y'all compromise and recirculate it for a shift?

Good Luck,
Latexman

 

[LittleInch]

Are there any standard methods for determining how many times the tank should be turned over / duration of circulation for good mixing?

In short not that I'm aware of as there are just too many variables and also most people recognise it isn't a great way to mix a tank.

If you have jet mixers, either tank side mounted or central you might stand a chance of getting a mix within hours not days, but simply dragging liquid out of one nozzle ( at the bottom?) and then pumping it back into another nozzle (also at the bottom?) with relatively low velocity, really won't mix your tank up, even if you did it for days. Especially with a viscous substance you could easily find that the amount of "mxiign " is actually quite low and only a small cone withint he liquid really gets mixed.

"I've got the volume of product in the tank, and the circulation pump rate." - That's good - Do you mind sharing it with us? Also nozzle size and location.

But to get any sort of "mixing" you need to input more energy (velocity) and variable jets like this:

https://www.transvac.co.uk/pdf/Tank_Jet_Mixers.pdf

or

https://www.gea.com/en/products/liquid-jet-mixer.jsp

https://www.alfalaval.nl/products/fluid-handling/mixing-equipment/tank-mixers/

The vendors will have some fancy CFD analysis to show how fast and how many of their jet nozzles you need.

You can also get centre mounted spray nozzles which rotate.



Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[3DDave]

I agree with georgeverghese. Isn't this the reason mixers usually use mixing paddles?

 

[Compositepro]

Is it possible to pump the various batches into the storage tank simultaneously through one pipe so that they blend as the tank is filling?

 

[SuperSalad]

Yes, we can all agree and agitator would be more efficient and better in just about every way, but the reality, I will have to assume due to lack of response from the OP up to now, is that the company probably doesn't want to spend a dime up front to get this result. It'll cost more in the long run: CAPEX and OPEX of Pump vs. Agitator Mixing; but hey, that is just how it goes a lot of the time.

General rule I've seen/used is 3-7 times total volumetric turnover time, but that is highly, highly dependent on the fluid properties, which is why I suggest actually monitoring this process a few times to determine the minimum mix time directly.

Set up a bet between the production and QC manager on who's prediction is closer to reality and you can all have a laugh at the end of it too.

Andrew H.

http://www.mototribology.com

 

[jari001]

@Latexman - I've heard of SOPs and other such process docs codify a 8/12/24 hr mixing time and then when someone asks, the oldest operator chuckles and says "20 years ago, the last engineer and I let it mix for said time because we had to go take care of X,Y,Z and forgot about the tank on recirc, but it passed QC so we sent it!". So letting it run for a shift sounds pretty good to me!

I've asked around my office and I got 10 volume changes, but I think that's because 10 is a nice, round number.

 

[MJCronin]

The question you ask is in the "Black Magic" realm of Process Engineering .....

MJCronin
Sr. Process Engineer

 

[Latexman]

If we knew the viscosity, we'd have a "feel" for the problem.

If we had all the data, we might be able to perform "magic".

OP, search the site using "jet mixing" or "jet mixer". There's a little knowledge out there.

Good Luck,
Latexman

 

[BenjaminM]

WOW thank you all for the quick replies. My apologizes for now following up sooner.

SuperSalad: We are wanting to combine multiple lots and issue a new lot number. Presumably all of the product is within specification, but on occasion we may prepare a batch slightly higher on one specification or another to balance out a batch that is on the low end.

Jari001: Roughly the the tank is 13,000 gallons, but might only be 75% full.

It's a cone bottom tank, and we are pumping from the bottom center of the tank to a flange on the roof offset from the center (closer to the wall), no dip tube.

I will have to double check pump rates, but for now I'll say 200 gpm.

Compositepro: We produce discrete batches in a single reactor. Sometimes it is packed directly to tank trucks, sometimes it is set to the bulk tank.

Latexman: I will have to follow up on the exact viscosity. At 70*C it flows fairly well, at 20*C it's just goo.

Jari001: 10 hours may be close.

George / 3DDave: I absolutely agree an agitator would be far superior. This is a thin wall atmospheric tank though, never mind the economics of the process.

Ultimately I wouldn't consider this to be a critical process other than wanting to issue a new lot number for the tanks contents. I'd like to make a recommendation rather than just pulling numbers out of the air.

I will have to follow the links provided later this weekend, then flow up with actual data rather than guessing, forgive me for searching for a method first.

Thank you all again!

 

[EdStainless]

Your return should either come down the side wall or have some kind of diffuser on it or you will entrain a lot of air.
If I use 13,000gal, 200gpm, and a mix factor of 7 I get about 8 hours.
But testing is the only way to validate that.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy

 

[Latexman]

Splash entry. I deduce this application uses no surfactant, right?

I have a spreadsheet for jet mixing using a lower side entry ending with a standard BW reducer to generate the high velocity jet and it’s aimed up at an angle where the normal liquid level meets the opposite side.

If only we had all the data.


Good Luck,
Latexman

 

[geesamand]

My company performs direct research on this problem. Both large (API 65) and small vessels. Long and the short of it: there are no good rules of thumb. 5 turnovers, 10 turnovers, assumed entrained flow, etc are all valid methods but they are limited to very similar problems only.

Achieving homogeneity depends on the variance in viscosity and density of the two parts. If you have to break a layering effect the total energy required (and value of a properly designed system) increases. If they are nearly the same in the first place, then you don't need much time or energy.

In any case you have the system you have, and it will surely be easier to collect data and derive an exact method for your product and finishing tank than it is to try and 'math' the optimum.

David