You need fully mature steel, aluminum, and electronics industries on Mars before you can think about UF6 centrifuges. The rocket equation just murders any idea anyone might have about shipping any kind of factory-in-a-box to any other rock in the Solar system--except one that takes in the local regolith and produces a copy of itself, or a similar factory-in-a-box with a different combination of inputs and outputs.
You're trying to jump right from putting the ore in at one end, and getting reactor fuel out at the other, but you apparently need at least a North Korea-sized economy inside the black box to do that, and Mars currently has an economy smaller than Sealand. There are no humans on Mars, and the robots currently there don't mine or farm anything in excess of their own immediate needs.
So a 4MW u-battery would use around 8kg fuel per year.
I did not count in burnup ratio. A simple thermal neutron reactor can only use <1% of fission energy in its fuel.
Burnup could still be limited due to fission products poisoning the reaction.
For conventional reactors it is around 3% and you are left with around 3% fission products.
Natural uranium has around 0.7% U-235. But I0m pretty sure that you would enrich the uranium befor you send it to mars. The 16TJ/kg is for enriched uranium with 20% U-235 content.
Anyway, energy density of nuclear fuel is extremly high. Even if you pay 10000 USD/kg for the transport of fuel to mars, you'd pay around 1 cent/kwh for the transport.