The following text is the reply I posted in the SPE forum to a similar question:

Irreducible water saturation (Swir) is usually a strong function of the absolute permeability and that can be populated across the reservoir model in such a way. This type of correlation has been widely reported in the literature by, for example Amyx and Bass (Petroleum Reservoir Engineering, 1960) in page 151 figure 3-20. If you are able to derive a function Swir = f (k) from available kr lab data and/or Pc data, you’ll be able to populate Swir in your 3D Petrel model provided you have already populated rock permeability.

As a clarification, this methodology yields a good estimate of Swir across the reservoir, however, Capillary-Gravity equilibrium must be considered for initialization purposes whenever capillary forces cannot be neglected. In other words, if there is any transition zone (i.e. capillary forces cannot be neglected), then initial water saturation (Sw) is related to Swir as follows:

1. Within the Transition Zone, Swir < Swi, and Swi is determined by Cap.Gravity Eq.
2. Above the Transition Zone, Swir approx. equal to Swi. Note that this condition still comply with Cap.Gravity Eq.