Saturation height function

[swibow]

Hi Vinomarky,

What a great sharing, thanks a lot.
One simple question; how to get a1, a2, L1, L2, b1, b2 for lambda function and a, b, Alpha & Beta for unknown function?

Thanks !

[vinomarky]

Fit to your data

I usually use Excel and the Solver addin

[swibow]

Fit to your data

I usually use Excel and the Solver addin

Do you have any excel file as an example?
I you don't mind, please send it to my email; swibow@gmail.com.
I really want to learn about this stuff, so far only J function that I know.
Thanks before, I appreciate your help.

[vinomarky]

I make them as needed - it's not difficult, just give it a try - use the solver to minimize the error by changing the variables
Check the plots of measured data vs curve fit data and if you are happy with the results use it....

[swibow]

I make them as needed - it's not difficult, just give it a try - use the solver to minimize the error by changing the variables
Check the plots of measured data vs curve fit data and if you are happy with the results use it....

Though I still don't understand, I'll try it. Thanks.

[temr]

Dipak, please read carefully the Eclipse thread the initialization process has there quite detailed describtion
If you have any questions ask

[dipak_m]

Dear Temr,
There is an initialization chapter in your thread. But I want in details. What are the different methods of initialization. I want to know every details of each initialization process, from Xls file to Petrel and then Eclipse.
Thanks

[dipak_m]

Fit to your data

I usually use Excel and the Solver addin

Dear Vinomarky,
What is the Solver addin?

[reservoir_engineer]

Vinomarky
thanks for your great help,
do you have some books, articles or papers about SW distribution and generally model initialization?
it will be very helpful for us
thanks

[vinomarky]

There are many many books and article on this subject. The challenge is not to find the books, the challenge is to actually put down the books, get some raw data and work through the problems..... you'll learn 10x more by doing than by reading/skimming

In my opinion, one VERY important aspect that may be lost on a lot of younger (and indeed some older) engineers getting into simulation is that you really must firstly get a mental model of what you think is going on:
Where is there water,
How is it distributed,
Why,
Is it physically possible....

When people start talking about tilted contacts, perched water etc this becomes extremely important. If you have perched water, you need to address with your geomodel.. if you have dynamic aquifer then you need to ensure you have good perm and pressure differential across your model (and is one of the few reasons you'd need to let a model equilibrate after initializing)... if you have poor perm and/or weak aquifer and observed different producing OWC contacts is that from capillary pressure effects, or is it compartmentalization? They will ALL have significantly different (a) ways to initialize and (b) prediction characteristics - so simply adopting the approach of placing the water where it is observed, initializing and then allowing it to equilibrate (without thought as to the reason) is lazy engineering and poor simulation practice that will in all likelihood give you incorrect insights into the predicted behaviour.

As far as more tidbits in characterizing/grouping saturation families - I quite like the FZI (Flow Zone Indicator) approach... it does not always work well, but often does.

FZI is a simple calculated index which you can often use to discriminate/group families of similar characteristic rocks (think of it as a continuous flow facie indicator). If you plot up your lognormal perm vs poro data, and overlay colour codes of FZI, you quite often are able to identify better perm/poro relationships to apply - this is best done in conjunction with your geologist, so that they can possibly make the linkage between FZI and depositional facie and therefore populate the model with appropriately paired poro and perm properties for that family of rock. Similar FZI rocks usually have similar saturation/flow characteristics

FZI = 0.0314 x SQRT(K/PHI) / (PHI / (1-PHI))

The numerator is obviously a J-function type relationship, while the denominator is the ratio of porosity to non-porosity. Generally, the higher the FZI, the better (cleaner) the rock is.... thinking it through, for a given permeability, if you have a low porosity it means that you generally have less clays and other pore-filling material (which in turn generally means lower Swir's and more favourable rel perms etc) than if you had a higher porosity for the same perm.

Again - no silver bullet (there never is), but it is another useful approach to know about

[dipak_m]

Dear Vinomary,
Is it possible to post your XLS files with solver addin where your calculate Sw from cap data (different methods)?
Thanks

[vinomarky]

Post a set of cap pressure data, and I'll see what I can whip up