The capping of ABV based upon the maximum alcohol tolerance of the yeast used has been discussed already. (and this was also listed in the release notes for BS3.)
In looked at one of the recipes that joepjanvh posted above (really, if you want help, post the recipes in .bsmx format; screen shots tell us very little about how your process works).
In his Blonde recipe, he is using step mash with temperature rests at 140F and 158F. BS2 only counted the first temperature rest within the prime amylase rest zone from 147F to 160F and ignored the other rest temperatures used. Per Brad's release notes: "Corrected bug in FG estimation that underestimated attenuation of yeast when working with a mash profile that incorporated steps at both low and high mash temps."
While not perfect, compared to the results estimated versus actual for the Blonde, BS3 was much closer to the actual measured values for its predictions. Note that the other recipes included also show the same issue: Multiple rests in the fermentability range yields different results than a model that ignores the multiple rests.
Since this feature has been requested many times in the past, it should not be a surprise that it is included in the latest version. And listed in the release notes for users to see.