Just like the survey above the grating on April 23, 2014, TEPCO's workers inserted the pan-tilt camera from the floor above to identify the location of the leak in the MSIV room in Reactor 3 on May 15, 2014. They used a hook to lift a portion of the grating so that the camera could go below the grating.
They found a leak, and it was from the expansion joint that connects to the Main Steam Pipe D, one of the four Main Steam Pipes.
From TEPCO's photos and videos library, 5/15/2014:
Leak was found from the expansion joint to the Main Steam Pipe D, which is in fact hidden from view. What you see in the foreground is the Main Steam Pipe C and the Expansion Joint C. Leak is marked by red circles in the photo (English labels are by me):
So, the MSIV (Main Steam Isolation Valve) itself may have worked exactly as it is designed for, which is to shut off the steam from the reactor in case of a severe accident that would necessitate the scram (we don't know for sure, unless someone enters the MSIV room and investigate), but the leak is from the joint right before the MSIV.
Here's the video, showing the leak (look very closely after 33 seconds):
Bad news, I think, for reactors around the world, but no one in the media has picked up on the significance of the news so far.
This is the typical reporting that I've seen in the Japanese media:
"A leak was found on the Containment Vessel of Reactor 3 for the first time."
Clearly, reporters don't bother to know what MSIV - Main Steam Isolation Valve - is. They don't seem curious either to ask questions like:
How did the expansion joint get damaged?
Was it by the earthquake?
Was it by pressure that exceeded the spec?
Was it by high heat that exceeded the spec?
If it was high pressure, what caused the high pressure?
If it was high heat, what caused the high heat? What are the implications for other nuclear power plants in Japan (and the world) that use this type of joints for many critical safety components in the plants?
Instead, they'd rather write about "nosebleed" in Fukushima Prefecture and how TEPCO's employees "disobeyed" the order and fled from Fukushima I NPP on March 15, 2011 (that's according to fanciful Asahi Shinbun). (More later, maybe.)
Below are the tweets from "Happy", who was hoping that the leak would be from one of the auxiliary pipes that comes through the Containment Vessel, such as the drain pipe:
The leak in the Reactor 3 MSIV Room was from the bellows [expansion joint], not from the drain pipe. A focus is on how to stop the leak from PCV (Primary Containment Vessel), but I think it's a serious stuff if this is a rupture of the pipe because of the earthquake, and the restart of nuclear power plants in Japan should be halted...
The reason why it is a big problem if the leak from the bellows in the Reactor 3 MSIV Room was from the rupture of the pipe caused by the earthquake is because this type of expansion joints are used everywhere on important pipes, PVCs, and RPVs (Reactor Pressure Vessels), etc. It's no exaggeration to say that every single nuclear power plant in Japan uses the same type of expansion joints.
The basic design of a bellows (expansion joint) takes into account the expansion and contraction of the pipe by the heat, rather than the seismic resistant capacity. If [the damage] is from the effect of the earthquake, the design would need to be revised. What the "nuclear village" is most afraid of is the rupture of pipes due to an earthquake.
That's why I think the on-site investigation hasn't been done, ostensibly due to high radiation. I expect the reason for the MSIV leak that the national government and TEPCO will come up with will be "due to the pressure that exceeded the design pressure, not the earthquake" or "due to extremely high temperature," citing some numbers.
Well, even if they say the damage was due to high pressure or high temperature, I would assume the critical component like this that connects to the super-critical component (MSIV) should not fail in a severe accident that would probably generate high pressure and high temperature that would exceed the design specs.