YES - on site reprocessing is indeed feasible.
That is one of the principal features of the Integral Fast Reactor [ IFR ] that I worked
on in the early part of my career when I was at Argonne National Laboratory.
Here is an interview my former boss, Dr. Charles Till; did with PBS's Frontline in a program
entitled "Nuclear Reaction" hosted by Pulitzer Prize winning author Richard Rhodes
[ "The Making of the Atomic Bomb" ]:
As Dr. Till mentions, the fuel for the IFR is in metalic form, as opposed to the oxide
ceramic found in most reactors. By keeping the fuel in metalic form, it is particularly
easy to do the reprocessing. Instead of needing a chemical processing plant, as one
needs with oxide fuel; IFR fuel is reprocessed using metallurgical techniques, namely
"halide slagging" followed by "electrorefining".
As Dr. Till mentions, the plutonium / actinide containing output of this process can NOT
be used as nuclear weapons fuel, but CAN be recycled back to the IFR to be burned.
Therefore the IFR doesn't impose a proliferation risk.
Because the metallurical process is much simpler than chemical reprocessing, the
reprocessing plant can be placed on-site, as you suggest. The plutonium / actinides
never leave the high-radiation region of the plant; so there is no opportunity for theft.
The following web page, courtesy of the Nuclear Engineering Department at the
University of California - Berkeley mentions the on site nature of the IFR's
reprocessing system under the heading of "Diversion":
In addition, the IFR was "inherently safe". It did not rely on engineered systems to
shutdown and cool the reactor. All that is needed to keep the IFR safe, is that the
laws of Physics work - which they ALWAYS do.
As stated in the article; the IFR is "almost too good to be true" as Richard Rhodes
phrased it. It's really too bad that President Clinton cancelled the IFR in 1994.
Dr. Gregory Greenman