It was done decades ago.
Pacific Northwest National Laboratory conducted in-reactor experiments that involved total fuel failure in a controlled environment. The series of experiments took place in the Canadian research reactor NRU located at the Chalk River Laboratory in Ontario. There were a series of experiments over about a six year period in the 1980s.
Three Mile Island's accident was the trigger for this research program. There was financial support for the project from the US, Canada, Japan, Germany, and a consortium of around 20 other nations.
The most severe of the accidents that we simulated involved simulating a Loss Of Coolant Accident (LOCA) that resulted in fuel rod cladding failure (including melting in the worst cases) to try to recreate the near total blockage of coolant flow in the fuel bundle. There were around 200 thermocouples in the test rig, along with lots of flow meters, etc. The idea was to gather enough detailed data to allow the regulatory agencies to properly evaluate the computer programs developed and used around the world that would try to predict the test results.
We actually used full 12-foot commercial reactor sized fuel rods. The reactor had only a 3-meter long core so our experimental containment actually stuck out the top and bottom of the regular core. We had a tiny bundle of rods, fully instrumented, inside a specially designed containment and the whole thing was then inserted into a process tube inside the reactor.
You can do a Google (or other) search using the words "pnl nru loca" and you can find a lot of information.
I was the lead programmer for the data acquisition and control system for the experiments.