I mean its not surprising honestly, the equations describe continuums which arent a real thing. I feel like its not surprising that it blows up even when "nice" boundary conditions are applied. To describe those cases better you just need to move a layer down and start accounting for particles instead of viewing it as a "goop".