Help with Custom Relative Permeability Function

[Ethan]

I am using TOUGH2 with ECO2N EOS and I would like to input custom relative permeability functions such that they better match a set of experimental data for a supercritical CO2-brine system. None of the default functional forms in PetraSim are adequate, specifically because the endpoint relative permeability to gas is always 1. This is not the case for the data that I am trying to match. It seems as though creating a customized function is the way to approach this problem, however, it unclear to me what the input parameters in the custom relative permeability dialog box represent (i.e. IRP and RP(1-7)). How can I define these input parameters to correspond to the desired functional form that I am trying to match? By the way, I do have a functional form, which is similar to the Corey expressions, however, the exponent values are different and there is a multiplicative scaling factor as well.

Any insight is much appreciated,

Ethan

[Charlie Thornton]

The Custom mode is really only useful if you have a custom compile of the simulator. It doesn't give you any capabilities that the others don't, beyond removing some of the validations and giving access to all of the input fields. For users that have a source code license from Berkeley and a FORTRAN compiler, they can use this to get the inputs they want via PetraSim.

If the options you want are just not possible via one of the built-in functions, we may be stuck. Could you post the details about the function you're after? Somebody might have a clever idea for a workaround.

- Charlie

[Ethan]

Here are the functional forms I am trying to model:

krw = [(Sw - Swr)/(1 - Swr - Sgr)]^nw

krg = alpha*[1 - (Sw - Swr)/(1 - Swr - Sgr)]^ng

where:

Sw = wetting phase saturation

Swr = residual wetting phase saturation

Sgr = residual non-wetting phase saturation

nw = wetting phase expnonent = 2.28

ng = non-wetting phase exponent = 1.69

alpha = scaling factor = .4558

After reviewing the TOUGH2 manual, it seems like I may just have to work with functions that have endpoint values of 1. I will probably just choose the model that yields the smallest error through the range of effective saturations.

Thanks,

Ethan