Tank hydrotest, will the water freeze? 2

[Mkotop]

Good day to all!
First time poster with a problem.

We have a 557 m3 vertical cylindrical steel tank, D=9,3m, h_fill=8,2m tank wall is cca 50 mm.
It will be filled with water from the nearby river in 3 stages. In every stage 1/3 of the volume and left to settle for 48 hours. After it's full it will be drained 1/3 of the volume every 24 hours. The filling part for each stage is roughly 4 hours, emptying rate unknown.

Temp of the river is 3°C, expected outside temperatures are -5°C at night and 4°C during the day. Wind speed not given

Question is will the water freeze at any point without the addition of antifreeze?

water 3°C c=4,21 kJ/kgK
steel c=0,49 kJ/kgK; k=25 w/m2K
Air -5°C c=0,717 kJ7kgK
Tank wall surface 237 m2

Heat loss of a full tank Q=kAdT=25*237*8=47,4 kW

If I calculated ok, the heat contained in 3°C water is (m*cp*T)

at 1/3 of Vmax: 2,344,970.00 kJ
at 2/3: 4,689,940.00
at Vmax=7,034,910.00

I know that the water will not be 3°C when the 2. and 3. stage of fill begins but don't know how to approximate.

So, this is all I got. Where do I go from here with the calculaions? I am without the clue how to add the time dimension in all of this.
Thank you!

 

[zdas04]

First, I'd like to thank you providing complete information and a clear statement of the problem. We see too many threads that we have to pull teeth to figure out what the question really is.

Isn't 1 kJ = 1 kW-sec? If it wasn't for that pesky latent heat of fusion, the answer would be either 2,344,970 / 47.4 (I'm assuming that heat loss through the surface into the empty tank is about the same area as the full tank, may or not be, you'll have to look at that) or an iteration with smaller time steps.

I think I would iterate on 1/2 hour time increments until the temperature reached 0C. Then you have to get rid of the heat of fusion the same way.

My gut feeling is that the surface of the water and the water adjacent to the tank walls will freeze several cm in (which is enough to cause all sorts of problems). The second 1/3 will melt the ice to some extent, but at the cost of cooling the new water very close to 0C. Next day it will probably be frozen harder than the first 1/3. The last 1/3 may not even melt the ice in the tank, but it will cool off quickly.

I expect that after the test is done you'll have to wait for summer to get this stuff out of the tank.

David

 

[IRstuff]

Using the numbers given, the best case tank temperature at 48 hr is ~0.15°C. Since circulation is definitely going to poor, the water next to the walls will probably freeze as David indicated above. Thermal conductivity of ice is much better than water's, so there will be massive freezing in the second period.

TTFN
faq731-376

 

[vpl]

I'd strongly suggest adding anti-freeze.

Patricia Lougheed

******

Please see FAQ731-376: Eng-Tips.com Forum Policies for tips on how to make the best use of the Eng-Tips Forums.

 

[zdas04]

Pat,
I started to recommend anti-freeze, but then realized that you can't put the water back into the river or on the ground if it has anti-freeze in it.

David

 

[saplanti]

Another point: The test water temperature is very important for the material selection. Have you checked/considered the charpy impact test requirement from the code for the material selected. If not you may cause fructure in the shell wall during the test, it may be disaster for the entire job.

You'd better check against the code you are running under.

Regards,

Ibrahim Demir

 

[saplanti]

I forgot to write the nature of the fructure; Brittle fructure.

Ibrahim Demir

 

[IRstuff]

While ethylene glycol (EG) is definitely out, propylene glycol (PG) might be worthwhile, although the amount of PG involved would be substantial. A 35/65 PG would drop the freezing point to around -18°C.

Also, according to its MSDS, PG is biodegradeable in both soil and water

TTFN
faq731-376

 

[CarlB]

What's the concern if a few centimeters freeze against the wall?
Many water storage tanks in south central Alaska are not insulated, and can function fine with several inches of ice on the walls and surface, sometimes even 2 feet thick.

 

[Compositepro]

I have an above ground pool of about 17,000 gallons. Even when temperatures have fallen to 0F, ice has never formed on the walls, only on the top surface. Large bodies of water only freeze on top due to convection. To get freezing on the walls you need much colder temperatures, like in Alaska in winter, not mild freeze/thaw conditions.

I would expect a small amount of ice to form at night and to melt during the day, given the conditions described. A covered tank would be much less prone to freezing than an open one.

 

[zdas04]

Experiments are always better than analysis. One question, how much chlorine is in the pool?

David

 

[jharris3]

If possible you might add some salt to the water instead of antifreeze? Lower the freezing point a little. Heat some water up, or put some gas heaters next to the tank. With the size of tank you are dealing with that shouldn't be too difficult.

There are also some alternatives to a full height hydrotest even for tanks.

Wind: You should be able to find weather statistics for the general location of the tank on the internet. Use average windspeed for the time of year together with the average temperature. Then lookup the forcast and see if you need to be more conservative (ie is this year colder than average?)

I usually assume that air has a convection coefficient somewhere between 5 & 50 W/m2 K, depending on the windspeed used. If there is any wind you can start at 15 W/m2 K.

You may want to calculate heat energy contained at 3.0C and subtract the heat energy contained at 0.0C(liquid). This is all the heat that is removable before ice begins to form. At that point any additional heat removed will create some ice somewhere in the water and as the liquid-solid quality is reduced more ice develops.

Then you take the difference and divide by your cooling rate to obtain the amount of time you have before that much heat is removed. Latent heat doesn't matter so much since you seem to want to avoid any ice at all.

In pools, lakes, etc the ice doesn't form except on top of the water because of 4.0C water is more dense than 0.0C water. Ice can form on your walls because of convection outside, but it will also freeze on the surface inside first.

Though ice has a better conductivity than water, as the ice forms there will be increasing distance between the water and the outside surface where the heat will be removed. At the ice-water interface the contact resistence will not change much from the steel-water interface. But again you are trying to avoid any ice at all...

 

[Mkotop]

Hello, first of all thank you for all of your input.
I believe I solved the problem.

We have around 47 kW of heat losses with the full tank of 3°C water.

We multiply kg of water with the latent heat of fusion for water:

557000[kg]*334[kJ/kg]=186 038 000 [kJ]

And then we divide the above heat taken from water to form ice with the heat losses of the tank

186 038 000 [kJ]/47 [kW]/ 3600 =1 099 hours to fully freeze

For 1/3 of the tank that would be 366 h and the whole hydrotest will last 216 h.

This is calculated for the worst case scenario (-5°C), when we take into account that the avg day temperature is actually cca 0°C the time to freeze will be even longer.
Some ice may form on the shell of the tank due to convection but I believe that won't be a problem. Tank has a roof so the surface of water probably also will not freeze.

All of this was pretty simplified at the end, but I think it is correct.

@jharris3 could you please elaborate on the alternatives to a full height hydrotest for tanks ?





Marko

 

[Compositepro]

Is there some new physics? You calculate that all this water will freeze but there will be no ice on the walls or the top surface. Where will it be?

 

[MiketheEngineer]

Wait til Spring??

 

[IRstuff]

"For 1/3 of the tank that would be 366 h and the whole hydrotest will last 216 h"

This is a misleading calculation. While you are correct is thinking that ALL of the water will take at least 366 hr to freeze, that doesn't mean that some of it won't freeze considerably sooner. Statistically, it's likely that 216/366 = 59% of the water will be frozen while the remainder is on its way there.

TTFN
faq731-376

 

[jharris3]

"@jharris3 could you please elaborate on the alternatives to a full height hydrotest for tanks ?"

I have been able to use additional NDE on tanks to avoid hydrotesting to a full height. However, i must admit it has been awhile since I dug into those options. There are some alternatives but i am not absolutely certain they apply, AND your client may not accept those alternatives anyway.

Thoroughly consult API 650 and 653.

 

[zdas04]

What you cannot do is a pneumatic test. Filling a 20 ft tall tank puts almost 9 psig on the bottom weld. Pressuring an empty tank to 9 psig will blow the lid into the next county (a 15 ft diameter tank has a lid that is 25k square inches, so the up force on the lid is almost 230,000 lbf).

David

 

[MintJulep]

Just because propylene glycol is degradable doesn't mean that it's ok to dump mass quantities of it back into a river or spill it onto the ground.

Arrange a circulating pump system to keep the water in the tank agitated.

The movement, combined with the pump energy added will probably be sufficient to prevent freezing.

If the river water is a constant 3C you could arrange to bleed off water constantly and pump in from the warmer river to maintain a constant level.

 

[ornerynorsk]

Salt may not be an acceptabel anti-freeze, as with the glycols, but what about sugar?

It is better to have enough ideas for some of them to be wrong, than to be always right by having no ideas at all.