Thank you Nicholas, you are most optimistic about the media.
Let us be pessimistic about the heat transfer.
I made another model, this time for 2 week old spent fuel, the maximum possible heat load (takes more than a week to boil off the water so less than 2 weeks seems impossible). It assumed 1600 fresh fuel rods (not possible of course but let's exaggerate to see max temps). To make things more realistic, I've reduced the heat flow from the sides, by adding insulation, 10 cm of 1 W.m/K material. This reflects the fact that conduction won't work as well horizontally as it works vertically, since there is no continuous zircalloy heat bridge horizontally.
Interestingly, the fuel rod surface area is about 1000x higher than the area of the sides of the assemblies if they are packed tightly. This means that even a choked laminar flow regime (reflecting debris and tight rod pitch) of just 0.5 W/m2/K would result in an effective thermal conductivity of about 500 W/m/K on the whole. This is 20x the thermal conductivity of zirconium!!!
Putting in 2 week old fuel making about 4x as much heat as the 3 year old fuel, we get the following:
Attachment:
dry spent fuel bay model fresh fuel with internal convection.png [ 129.88 KiB | Viewed 1005 times ]
The peak temperature hovers around 450 to 485 degrees Celsius (oddly enough the model doesn't reach equilibrium). This is even lower than the case without internal convection but with 4x lower decay heat load (current fuel).
This shows that even in a choked laminar flow regime, there won't be fuel failure. Suprising, not sure if I believe my own model here...