Without seeing your data, some things to check first;
Crossplot your log Sw data (not populated Sw, as this would mean you trust how the geo/software has distributed the property) against log perm in a log-normal plot - See if you can find a reasonable relationship of Sw as a function of Log (Perm). What about if you colour code the points by height above free water level.... What about if you plot Sw vs SQRT(K/PORO)?
What we are looking for is a way to ditch the RMS distributed Sw's and come up with a relationship that Eclipse can calculate at run-time with OPERATE keyword.
In short, I recommend AGAINST using Geo distributed Sw's to initialize your model with
So in summary;
1. Find what relationship best fits observed (log) Sw's as a function of perm, Sqrt(K/Phi) etc
2. Use endpoint scaling
3. Use this function to initialize SWL's using OPERATE function
4. Calc other endpoints appropriately (ie SWU = 1-SWL, SWCR=SWL etc)
If you are in transition zone with mobile water initially, and need to factor in height above FWL, then it gets a bit more complex to do this at runtime, and you may have to back-calculate 'synthetic' J-Func curve in addition...
Note: if you go with a SQRT(K/PORO) type relationship, ENSURE in talking with your geo, that they have properly paired the distribution of poro and Perm through co-krigging etc, otherwise they may have simply distributed the PORO and PERM independently, meaning your nice SQRT(K/PHI) relationship is rubbish away from well control.... In fact you might even want to consider asking them to created a log property ofSQRT(K/Phi) (treat it like a continuous facie value - similar to FZI or flow zone indicator type analysis) and populate THAT property along with PORO, and then calculate appropriate PERM.... just a thought
Reply With Quote
Bookmarks