"Perpetual Force" Air Molecule Motor Part II

Modelling of further cases

I modelled ten more cases of the foil in a container with air molecules, as discussed previously. The foil was held stationary for a further 1s, 2s, etc up to 10s, before being released, in order to obtain a different random starting distribution of molecules for each case. 

Results

In all of these cases also, (summarised in the spreadsheet below) the foil moved generally upwards in the container, reaching velocities between 0.02138 and 0.09513m/s just before hitting the upper boundary. On only one occasion did it move to (just) within one foil thickness of the lower boundary, and even then I didn't notice any molecules impacting directly between that boundary and the foil. Boundary effects do play a part in accelerating the foil near the end of the simulation, but only because it has already moved to within about half a foil thickness from the upper boundary.

Fig 4. Spreadsheet of results for foil in container with air molecules

Discussion

Is this a credible result, i.e. would such a device really work? Could it really deliver a perpetual net force, and energy, from nothing more than "thin air"? 

The probability of the foil moving to the upper boundary 11 times in a row by chance is the same as tossing a coin to get 11 heads in a row, i.e. 1 in 2^11 = 1 in 2048 = 0.0004883. Still, these eleven "spot-check" results are not fully conclusive. The three main reasons for that are the unrealistic modelling of the air molecules, boundary effects, and the 2D rather than 3D analysis.

The physical data for the air molecules used in the model are:—

radius 2mm, mass 0.261 gram, initial velocity 1.414 m/s. 

These data are vastly different from the (approximate) data for real air molecules (mostly N2 with some O2):—

radius 0.00000025mm, mass 4.8 × 10^-23 gram, rms velocity 500m/s.

Also the foil mass, modelled at 1kg, is far heavier than would be the case for a foil of the correct scale interacting with real air molecules.

Nevertheless, I find these results so far to be interesting. They do suggest that further work on this idea would be worthwhile.