Saw this online... I wonder if they were ever able to tweak that model to match the dry cleaner pump test:
http://www.geo.sunysb.edu/lig/Conferences/abstracts99/White.pdf
For a while after moving on from my groundwater days, I thought about it, and if you can't simulate that kind of drawdown using a single fracture simulation in a equivalent porous media model alone, there is not a chance you can get that to occur in the complexities of a full scale, multi-layer flow model.
By the way, this is still one of the most hilarious ground water stories:
http://aquadoc.typepad.com/waterwired/files/freeze_gw.pdf