Transient Thermal Storage Tank Model

[Boghi1990]

Hi,

A 24 feet dia, and 40 feet cone shaped roof uninsulated storage tank, filled with 80% crude oil. The fluid in the tank is stationary and the tank is exposed to ambient air. The intial mean temperature of the crude oil in the tank is 20C.The data available is the daily min. and max. temperatures over 30 days(July & August)and daily wind speed data.

Required is a transient thermal analysis(ALGOR Fluid Flow & Heat Transfer Analysis), carried over a period of 21 days to determine when the maximum bulk temperature of the crude oil is reached.

What boundary conditions do I need to assign on the outside surface of the base plate of the tank ? Do I need to model the soil beneath the tank ?
As for the outside surface walls of the tank, what boundary conditions are most appropiate to use? Are convection coefficients OK as boundaries assigned on those surfaces?

If there is anyone out there with some experience on tyhis subject, please send me some suggestions.


Regards,

Bogdan

 

[GBor]

Bogdan,

Sorry I let this one go, but I attempted to answer in the other post...guess it was chosen to delete that one and keep this one. In general, double posting is not allowed.

Generally, the ground is a constant temperature, so the boundary condition for it doesn't involve any convection or conduction...just place a constant temperature on that surface. You can allow conduction into the fluid, assuming you model the fluid as solid elements with an initial temperature. Yes to convection on the sides, but the convection constant will change with the wind speed.

 

[Boghi1990]

Hi GBor,

Thanks for your reply. Yes, the 80% fluid has been modeled as solid elements and an initial temperature is applied on the nodes. The buoancy effect will be considered also because this will affect the temperature distribution in the fluid.I will also consider the temperature dependet convection coefficient acting on the exterior surface wall sides and exterior surface roof wall of the tank. Although the ambient temperature is time dependent and a curve will be generated for that.

I have two more questions about this model. First, I am not sure how could I simulate the 20% air inside the tank.I am refering the air which in on top of the crude oil and fills the remaining space in the tank to the roof. There will be a buoancy effect within that part of the tank. I have not modeled the air.

Do I need to apply natural convection coefficients on the vertical inside surface walls of the tank(dry area of the verical wall and the roof). If that is the case, I have to consider those formulas(Nusslet , Grashof and Rayleigh numbers)to calculate the hi acting on the vertical and horizontal surfaces. Am I right about this approach?

I am not sure on what boundary (convection coefficient?) needs to be assigned on the top surface of the fluid? There is natural convection at the interface between the air in the tank and the top surface of the fluid..How should I take into account this? I have attached here the tank model(as neutral files step & iges format).

Looking forward to hearing from you.

Regards,

OnePoint

 

[GBor]

This is getting pretty complex. Are you sure FEA is the right approach? I realize it will be hard to get a peak temperature without some transient approach, but with all of the considerations that you have, I'm afraid you are out of my league.

I will offer this: I once analyzed a wall undergoing fire testing. I modeled a 2-D cross-section and followed those results with a 3-D full analysis. I modeled the air in the cores as brick elements and used conduction instead of convection since the air was "trapped" inside the wall. That actually worked pretty well. You should be able to do the same with the air that is trapped.

The buoyancy is where you lose me...I've seen it done, but never done it myself in a thermal analysis. You approach seems reasonable, but I also have to mention radiation...are you accounting for it during the day?

By the time you put all of this in a single model, any error in input or assumptions may significantly alter your final results. This is why I question FEA.

Perhaps someone with a stronger heat transfer background will jump in. I questioned site management for deleting the thread in the FEA forum and leaving this one. The reason this was done was because you specifically mentioned Algor in the post. The FEA forum is for more general questions regarding process and procedure. If you don't get a response in this forum, go to the FEA forum and post a link to this thread asking for those with thermal backgrounds for assistance.

To post the link to the thread, there is a thread number at the top of this page that can be copied and pasted into your post.

Good Luck!