A micron-sized laminar model leading to an unnatural solution
 

My propeller model is 30 microns in diameter and 10 microns in height, and rotates counterclockwise at an angular speed of 50 rad/s. Each blade is even in thickness and is of a twisted surface that twists gradually in counterclockwise direction from hub to edge. What has been puzzling me is that the CFD results using the laminar model (because of Re<<1) are just the opposite to reality. The flow direction is from the top down, whereas it should have been from the bottom up because the blades push the liquid up rather than pull it down. What's more, the static pressure on the upwind surfaces is negative. The results are as follows.
http://ww2.sinaimg.cn/mw600/64d22f47tw1dpm0msm0ruj.jpg
this figure is the propeller geometry
http://ww1.sinaimg.cn/mw600/64d22f47tw1dpm0n6yhjhj.jpg
this figure shows the velocity vectors in the y-z plane.
http://ww1.sinaimg.cn/mw600/64d22f47tw1dpm0momuuvj.jpg
the velocity vectors in z=5 plane (perpendicular to the rotating axis) attest to the counterclockwise direction in which the propeller rotates.
http://ww4.sinaimg.cn/mw600/64d22f47tw1dpm0mwj5tdj.jpg
this figure shows the static pressure on the propeller surfaces. Note the pressure on the upwind surfaces, which is negative that contradicts reality.

I have rechecked my settings for times to find no clue that accounts for the odd result. However, I can get an acceptable result when calculating the turbulent model of a centimeter-sized propeller which is 10k times bigger in extent and several times higher in rotation speed.
I have refined the grid, tried slowly increasing the rotation speed from a small value, and using a more precise solver and discretization method, but all have failed to be a solution. Please help if you're interested. Any help would be much appreciated.