Could the dynamic Casimir effect be used for propulsion without propellant? The dynamic Casimir effect has been demonstrated experimentally by Professor Wilson from Chalmers University in Sweden. Instead of moving mirrors at relativistic speeds, Professor Wilson circumvented the problem by varying the electrical properties of a mirror (using a SQUID), rather than moving it in space. 
For the device itself, the photons are emitted in correlated pairs, so the total momentum is zero, as before the production of the photons.
Assuming that Professor Wilson held the device in his hands, some of the photons carried momentum to his retina, others were lost in space. 
If we have a large number of such devices attached to a large material surface, 
some of the photons emitted (as a consequence of the dynamic Casimir effect) will carry momentum to the material surface, other photons will be lost in space. Is this a feasible method of propulsion, without using any propellant? A small amount of nuclear fuel (for example) must be "on board", in order to provide the energy input, required by the many devices, but no actual propellant would be necessary.
Note that this is not the controversial EM drive design, the physics behind it is well understood, and experimentally demonstrated. The question is related to feasibility, and the number of such devices required. In your answer, please give a numerical, order of magnitude estimation of the momentum transferred, when $N$ such devices are involved. Can the technology be improved?
 A: The answer is yes, but it's not interesting since there are much easier ways of doing the same thing.
Here is an easier way: on your spacecraft, get a torch, and shine it in the direction opposite to which you want to go: the photons carry momentum, and conservation of momentum means you move in opposite direction to which you are shining the torch.  If you want to make the situation more similar to yours, then put an omnidirectional light source behind your spacecraft and paint the back of it matt black to absorb the photons that hit it.  In fact, you can instead of painting it black, mirror it to reflect them and transfer more momentum ... and that's the torch I started with.
These things are often called 'photon drives' and they are common in science fiction, and physically quite possible (I don't know if any have been tested in space).  They are the most efficient form of propulsion you can have, but it's hard to design them so they have high thrust.  It's easy to know the momentum transferred: it's at most $E/c$ where $E$ is the energy input, since for a photon $E = pc$ (this assumes that all the photons go directly backwards, the spacecraft does not radiate any heat, &c &c).  You need to consume at least $p/c$ of mass for momentum transfer of $p$ using another famous equation.
Using the dynamic Casimir effect to do this would be, I think, merely a maximally Heath Robinson approach to solving the same problem.
