’s Fukushima Corium Research Experiment Results

In our research we have tried to find ways to help explain some of the behaviors seen in meltdown accidents. A picture can be better than just explaining it. A full on experiment can be even better.

We have been investigating the processes of the melted fuel escaping the reactor vessel and how it behaves. Below the reactor vessel is a round concrete pedestal that helps hold up the reactor vessel. This round structure has a doorway type opening at one side to allow workers and equipment into the pedestal. There has been some research on how melted fuel (corium) can leave the pedestal via the doorway. The existing research shows that this doorway will allow the melted fuel to flow out of the pedestal across the containment floor. This melted fuel can burn through the steel containment liner. In some locations on containment side walls there is no concrete structure to back up the steel liner. The downcomer pipes are just such a location and are a known risk for melted fuel to escape containment in the Mark 1 reactor design used at Fukushima Daiichi.


Cut away view of the reactor vessel sitting in the concrete pedestal of a GE Mark 1 BWR, pedestal doorway missing in diagram


Pedestal dimensions for a BWR Mark 1 Reactor, similar to Fukushima Daiichi unit 1


Interior of the reactor pedestal and doorway, unit 5 Fukushima Daiichi (taken pre-disaster)


We wanted to gain a better understanding how the melted fuel might behave as it escapes the pedestal and how it might act on the edge of the containment side wall where it meets the floor. member (and retired INL research reactor manager) Dean Wilkie developed an experiment that would give a good visual representation of the melted fuel leaving the pedestal. Below is an overview of the experiment and what we found.

Equipment & Methods:


BWR Reactor Pedestal inner radius 10.125 ft (3.0861 meters)
BWR Reactor Drywell floor radius (inner) 22.852 ft (6.9652896 meters)
BWR Reactor Pedestal doorway about 4 ft (1.2192 meters) with 18 degree angle opening

Model scale: roughly 1:60

Flower pot size: 4 inch (10.16 cm) clay flower pots ( 1inch – 5 feet roughly)
Top pot drain hole:  1 inch (2.54 cm) diameter hole
Bottom pot: 4 inch (10.16 cm) hole cut in top
Bottom pot doorway: 1 inch (2.54cm) cut into side of pot

Kiln: Glass fusing kiln Kiln interior base material:
Coated clay Kiln temps: 1650F (898C) for the melt stage where they opened the door, 1700F (926C) maximum temperature
Fire time: Up to maximum temperature,  hold for 2 hours then cool.
Glass brand: Bullseye (
Glass specs:


Fusions Glass Studio in Eagle Idaho assisted with the project and advised on selection of the glass and kiln run. (

Clear and red glass used for art glass fusing manufactured by Bullseye Glass ( was used in the project. The combination of clear and red was used to attempt to find flow paths in the glass melt as it flowed out of the simulated pedestal area.

Glass was selected as the representation of the fuel due to the availability and ease of conducting the experiment. Many of the academic studies of reactor core melts have been done with materials such as lead and other metals. Fusing glass melts at a fairly low temperature and makes a viscous flowing substance with enough similarities to molten fuel for our purposes. Glass was finely cut up and packed tightly into the upper pot used to resemble the reactor vessel core region. The height of the pot is roughly the height of the fuel core.

experiment1Glass stacked in top pot
experiment2Glass filled top pot over the lower pot, ready for firing

The upper pot is suspended over the lower pot that represents the reactor pedestal using fire brick and clay kiln supports. Pots sit on the coated clay kiln slab.experiment3

Pots staged in the kiln before firing


The upper pot has a 1 inch (2.54 cm) diameter hole. The lower pot had the bottom cut off leaving it open. A one inch (2.54cm)  hole, scaled to represent a 4 foot pedestal doorway, was cut in the lower pot to represent the pedestal doorway..experiment4

Detail of the lower pot with pedestal doorway cut out


The kiln is programmed to ramp up to the required heat level and run with the door closed. Once the kiln reached 1650F (898C), the point where the glass was expected to have begun melting, the kiln door is quickly opened to photograph the experiment.experiment5Kiln before loading


Photos and video were briefly taken during the opening of the kiln door. Then the experiment was allowed to run the rest of the melting process. The maximum temperature of 1700F (926C) was held for 2 hours, then allowed to slowly cool in the kiln. This follows standard methods used in glass fusing art.


Photo of the kiln door being opened briefly while the temperature was at 1650 F  (898 C). Molten glass can be seen leaving the simulated pedestal doorway onto the containment floor



After the project was properly cooled it was removed from the kiln for further study. What was found was quite stunning, the red glass showed very distinct flow patterns in the melted material as it left the pedestal opening. See a video of the finished melt here and here.

Pedestal pot and glass melt after firing and cooling


The glass that remained inside the pot was about .5 inches (1.3cm) with a residue mark about 1.37 inches (3.5cm) up the inner wall of the pot. This indicates that as some point during the melt process more of the melted fuel representation remained in the pedestal.

A distinct step down in the glass was found at the pedestal doorway opening with a smooth concave drop towards the glass that flowed out of the pedestal. Directly outward from the pedestal doorway was a lower thickness of glass that was almost completely clear with wide spaced bands of red further out. This radiates out from the pedestal indicating this was a path of high flow rate for the melted glass. The lower thickness rises as the melt reaches the edge of the flow pattern.

experiment8Pedestal opening showing glass levels

experiment9Detail of the melt striations and low area in the glass


Concentric bands can be seen in the almost round flow of glass out of the pedestal. Bands were the tightest at either side of the pedestal opening with widening bands as the melt flowed back around the pedestal. The outer edges of the glass melt were about .19 inches  (.5cm) with a fairly uniform thickness except for the region near the pedestal doorway.

experiment10Overhead view of the melt outside of the pedestal 


A line was drawn to show the physical edge of the containment wall. In this experiment the fuel melt clearly reached the edge of the simulated containment and would have had considerable contact with the steel liner and wall along the area directly facing the pedestal doorway. In the actual reactor units at Fukushima Daiichi this area is roughly where the #8 torus downcomer pipe is located. In the floor plan below the glass melt can be seen vs. containment structures. Red boxes indicate areas of known high radiation. Some of the suspect areas have not been investigated or reported by TEPCO yet.


Floor plan of unit 1 with pedestal doorway and melt overlay


The interior of the upper pot that represents the reactor core, a noticeable residue was left in the pot. Most established research into reactor core melts assume a similar behavior in an actual meltdown, where some residual fuel melt will remain inside the reactor.
experiment12Upper pot with residue


The glass melt showed a much wider flow of melted fuel out of the pedestal. Studies using computer models at Sandia National Laboratory assume a 130 degree flow from out of the pedestal that would radiate out in a fan shape. Ours showed a more round flow from the pedestal. One factor that may be different with the glass melt is the overall ambient temperature within the kiln vs. a more localized heat within the melting substance itself in an actual core melt. This may not be a deciding factor since the understanding is limited of the actual ambient temperatures within the pedestal and small containment structure of a BWR reactor during an actual meltdown. A faster hotter meltdown and reactor vessel failure could potentially achieve something similar to what was seen in the glass experiment.

The existing academic research on fuel melts show that “classic fluid dynamics” are involved. The fuel melt spreading is governed by gravity, inertia and viscosity. Two other known behaviors of melted fuel were not represented in the glass melt and are absent in many of the academic models. The reaction of the melted fuel and the ability of it to burn the concrete floor of the containment structure is one. The other being the ability of the melted fuel to reheat itself in certain conditions. Since these are not part of many of the existing established research studies on melted fuel we opted to not include them in this effort.

Questions also arose about the density of the glass compared to the density of a core melt. When reactor fuel melts it also melts other materials such as fuel cladding, control rods and portions of the steel structures and reactor vessel. As was seen at Chernobyl, the weight and density of melted fuel is considerable. One such example from Chernobyl is the finding of a crushed but not melted access ladder in a steam pipe surrounded by melted fuel. It is assumed the weight of the melted fuel mass rather than the actual heat collapsed the ladder.

To account for some of the potential density issues a similar experiment using lead as the fuel melt representation is being planned. Brookhaven National Lab and some other fuel melt experiments used lead as their fuel melt representation. This experiment should give some additional data for the purposes of density involvement and comparison against the glass melt experiment.



Fukushima Daiichi Accident Study – Sandia National Laboratory

Investigations On Melt Spreading And Coolability In A LWR Severe Accident
Doctoral Thesis (2001)
Maxim J. Konovalikhin

Photo gallery of the corium experiment can be found here with larger images:

Some of our ongoing research into melted fuel behaviors at Fukushima Daiichi can be found here:

This article would not be possible without the extensive efforts of the SimplyInfo research team
Join the conversation at

© 2011-2023, All Rights Reserved Content cited, quoted etc. from other sources is under the respective rights of that content owner. If you are viewing this page on any website other than (or it may be plagiarized, please let us know. If you wish to reproduce any of our content in full or in more than a phrase or quote, please contact us first to obtain permission.



Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: