[CRAGIO]

Hi,

Trying to model the convection caused by the density difference between water and water that has been dissolved with carbon dioxide. This occurs in carbon capture storage systems. The carbon dioxide is injected into deep saline aquifers which are areas of porous rock surrounded by impermeable rock. the porous medium is saturated in water. The injected co2 begins to dissolve in the water creating a solution with a higher density than the surrounding water. This of course causes density/buoyancy driven convection.

Does anyone know how to go about modelling this? the only things i can find about modelling natural convection which is primarily driven by heat transfer.

Any help would be really appreciated.

Thanks

[jamalf64]

Hi dear friend
do you solve this problem?
I have same problem. can you help me?

[ComputerGuy]

Originally Posted by CRAGIO

Hi,

Trying to model the convection caused by the density difference between water and water that has been dissolved with carbon dioxide. This occurs in carbon capture storage systems. The carbon dioxide is injected into deep saline aquifers which are areas of porous rock surrounded by impermeable rock. the porous medium is saturated in water. The injected co2 begins to dissolve in the water creating a solution with a higher density than the surrounding water. This of course causes density/buoyancy driven convection.

Does anyone know how to go about modelling this? the only things i can find about modelling natural convection which is primarily driven by heat transfer.

Any help would be really appreciated.

Thanks

This is exactly a natural convection problem, which although occurs a healthy amount in heat transfer-related problems, is driven by density differences as you describe them.

As far as modelling this, I suspect it will be a simple single phase, multicomponent model. That's not the hard part, however.

There are two issues I see:
1) Fluid properties
2) Boundary conditions

I suppose I'd first get your fluid properties correct first. CO2 can be a tricky one. Is this liquid CO2, dense phase, etc? What's the size of your domain? Will pressure change significantly due to elevation? Does the geothermal gradient come into play such that you have to worry about temperature too? Is the water entering the CO2 saturated with salt, or is it undersaturated such that you won't have to worry about it?

The boundary conditions on your problem depend on how concerned you are about the draw down of water from the surrounding rock, or, rather, how constantly the porous media stays filled with water. This sounds like it's a wall boundary condition on all sides (except the inlet perhaps, which is a pressure boundary with CO2 mass fraction=1). The temperature, depending on the size of the domain, is either isothermal or perhaps T(z) due to the geothermal gradient. Perhaps you want to worry about locally cooling the rock, but at some point in your numerical domain (10m away from the CO2/Rock interface, perhaps), you're going to assume isothermal/T(z). To implement temperature as a function of depth is a simple UDF. Search the site or ask for help specifically on this.

The inlet water flow is the toughest part. There are a number of ways you could do this. Keeping with the wall boundary conditions, I suppose what I'd do is simply loop around all the cells at the walls and adjust their mass fraction of water based on some known diffusion rate you have every step. If I recall, water/liquid CO2 diffusion is quite high (D ~ 2 × 10−7 m2/s). Assuming your fluid properties are robust, as well as your CO2<-->H2O diffusion value, fluent will handle the rest.

I'd start with the simplest case, see if it's giving you what you want, then come back once you've solved everything and want to move to something more complex.

Let us know -- good luck!
ComputerGuy

[CRAGIO]

Originally Posted by ComputerGuy

This is exactly a natural convection problem, which although occurs a healthy amount in heat transfer-related problems, is driven by density differences as you describe them.

As far as modelling this, I suspect it will be a simple single phase, multicomponent model. That's not the hard part, however.

There are two issues I see:
1) Fluid properties
2) Boundary conditions

I suppose I'd first get your fluid properties correct first. CO2 can be a tricky one. Is this liquid CO2, dense phase, etc? What's the size of your domain? Will pressure change significantly due to elevation? Does the geothermal gradient come into play such that you have to worry about temperature too? Is the water entering the CO2 saturated with salt, or is it undersaturated such that you won't have to worry about it?

The boundary conditions on your problem depend on how concerned you are about the draw down of water from the surrounding rock, or, rather, how constantly the porous media stays filled with water. This sounds like it's a wall boundary condition on all sides (except the inlet perhaps, which is a pressure boundary with CO2 mass fraction=1). The temperature, depending on the size of the domain, is either isothermal or perhaps T(z) due to the geothermal gradient. Perhaps you want to worry about locally cooling the rock, but at some point in your numerical domain (10m away from the CO2/Rock interface, perhaps), you're going to assume isothermal/T(z). To implement temperature as a function of depth is a simple UDF. Search the site or ask for help specifically on this.

The inlet water flow is the toughest part. There are a number of ways you could do this. Keeping with the wall boundary conditions, I suppose what I'd do is simply loop around all the cells at the walls and adjust their mass fraction of water based on some known diffusion rate you have every step. If I recall, water/liquid CO2 diffusion is quite high (D ~ 2 × 10−7 m2/s). Assuming your fluid properties are robust, as well as your CO2<-->H2O diffusion value, fluent will handle the rest.

I'd start with the simplest case, see if it's giving you what you want, then come back once you've solved everything and want to move to something more complex.

Let us know -- good luck!
ComputerGuy

Thanks for the advice. Im just wondering if CFX would be better? Its just that I have more experience in CFX as opposed to ANSYS but all the text I had read so far focused on fluent.

[mshahriar]

Hello, did you figure this problem out? I want to simulate this exact thing but can't seem to do it after trying out a lot.

I use Multiphase VOF model, use CO2 as inlet (set is at mass flow inlet). But I can't get the finger pattern it makes when dissolving.

Please help me out, if you get the chance.