Blackoil through e300

[Dana]

Hello!

I am currently working on a black oil simulation which is required to be performed in Eclipse 300, and of course, I am having some issues.

First of all-gas is injected, and at some point, there is timestep chopping. I tried numerous PVTs, relative permeabilities etc.

But, I noticed something unphysical: when gas is injected, it seems that there is no transition between oil and gas saturations (like: gas saturation is increasing and oil saturation is decreasing), but instant exchange of the fluids: Swc is for example 0.25, oil in that grid cell is 0.75, and in the very next timestep, after gas injecting, situation is like: Sw=0.25, Sg=0.75 and So=0.
I used several relative permeabilities, but no results.
I suppose the problem (and the solution) is in some keyword in runspec of data file, since black oil simulation in E300 is a specific case, I guess.
I would appreciate if someone could send me example data file with black oil in E300, or at least give me some ideas.

Thank you in advance!

[Bang Gaol]

Have you made the specific PVT for gas as well? It's important. How about Rs? I've tried the same case as yours long time ago, however I didn't face anything like yours.
Do triple checks with your data. Hope you can mention more abou the project.

Regards

[Shakespear]

Take a look at CASE3.DATA in the E300 DATA directory. There you have a case of running E300 for a Black Oil problem. It might give you some hints regarding your problem.

Try this. Use the CASE3.DATA PVT properties on your problem and see what happens. This may tell you where to look. You could do the same with Rel. Perms :-)

[link Point to another website Only the registered members can access]

[Dana]

Oh, I am sorry, I forgot to mention maybe the most important part, and that is that we have black oil, but we use compositional export of oil from PVTi (so we don't use BLACKOIL keyword in RUNSPEC section).
No matter, we tried different pvts, the ones from other fields that we are positive they work, relative permeability...
Now we handled gas injection by defining the composition of an injected gas, but still, that gas is injected until it wipes out all oil (Soil=0).
We also tried in one run injection of gas in one well and water injection of several others. In Floviz, it can be clearly seen that around water injectors, Soil is as it should be (Soil=0.3) and around gas injector it is as I already said, zero.

Anyways, thanks for the answers!

[Bang Gaol]

So, how's the result? Can you share with us? I think when somebody is already finished with his study, it'd be nice if it can be shared.

Regards

[Dana]

Well, like I said in previous post, we still don't have results, no matter what pvt and relative permeability we use, oil is swept until Soil=0, which is unacceptable. We are working on it, once we handle the problem, I will share that solution with you.

[OICURMT!]

Oh, I am sorry, I forgot to mention maybe the most important part, and that is that we have black oil, but we use compositional export of oil from PVTi (so we don't use BLACKOIL keyword in RUNSPEC section).
No matter, we tried different pvts, the ones from other fields that we are positive they work, relative permeability...
Now we handled gas injection by defining the composition of an injected gas, but still, that gas is injected until it wipes out all oil (Soil=0).
We also tried in one run injection of gas in one well and water injection of several others. In Floviz, it can be clearly seen that around water injectors, Soil is as it should be (Soil=0.3) and around gas injector it is as I already said, zero.

Anyways, thanks for the answers!

Are you implying that you are using an EoS? If so, then you need to be specific when equilibrating the system, as E300 determines what is oil and what is gas based on their fugacities [LATE EDIT: Actually, each components fugacities must be equal in both phases]. You need to think of Blackoil simulation as a simulation of two components, which is analogous to two equivalent phases, oil and gas. In compositional simulation, which you are obviously performing, as you stated that you are using "compositional export of oil from PVTi", each individual hydrocarbon component is tracked, with the user supplying what is oil and what is gas and under what conditions, via an Equation of State.

With respect to you question as to why the simulation drive the process to a zero residual saturation, you will need to determine the minimum miscibility pressure for the system. It may be that your injection gas is ideally miscible with the oil in the reservoir and that the reservoir pressure has a achieved a vaporizing miscible drive (i.e. the pressure is so high that all of the oil vaporizes into the gas). Without direct lab data to support this mechanism, your results will probably be wrong.

As for time-step chopping, you must know how simulation is actually performed. In the case of E300 and compositional simulation, the end of each time step (even time step 0, which is initialization) a stability test is performed for each cell to determine if more than one hydrocarbon phase exists in the cell. If there are two phases, then the standard K-value is assigned from the previous iteration and the process continues until a final solution (i.e. convergence) is achieved. If the pressure in the cell where you are injecting radically changes pressure to the point where you instantly move along the phase diagram from an liquid to a vapor, then Eclipse 300 must reduce the timestep (chop) until is can smoothly transition the cell from a liquid into an alternate form (same liquid, two phase or vapor) and converge while still satisfying the stability tests.

OIC!