The atomic elephant...

The theme of this article basically states that because TEPCO was very sensitivity to the release of radiation that they delayed the venting of steam from the primary coolant loops (containing water), most likely to debate the consequences. Once you begin to release steam from the primary you would have some release of radioactive gasses but, you would also release pressure. This release of pressure would allow more water to convert to steam effectively cooling the reactor fuel. Without the cooling in the first 12 hours, the temperatures could reach values where fuel elements could crack and release even greater amounts of fission products and partial meltdown becomes more probable. Obviously, these are very hard decisions and I'm sure this one went to the very top of the Japanese government. Considering the aftermath of the earthquake included aftershocks of 7.0 and greater and many other issues relating to communication, the delay is understandable. I don't know the exact sequence of events yet, so this is just an educated guess: since the fuel got much hotter than expected, once venting occurred parts of the zirconium cladding became uncovered from water and exposed to steam. At these high temperatures, a chemical reaction with the zirconium metal with steam produces zirconium oxide (ZrO2) releasing hydrogen gas. This gas collected in the reactor secondary containment building producing the explosion once exposed to a spark. This is what occurred. The explosion(s) was not caused directly by the high pressure buildup, but rather indirectly from the delay in cooling by the venting process...at least this is what this article tends to point to.

What the article implies is that the real cause of this delay in venting is the public fear of radiation which produced officials being reticent in making this decision in an expeditious manner. In my opinion, these types of complex decision-making dynamics rarely are so one-dimensional.

Reference on ZrO2 + Steam:
High-temperature oxidation of zircaloy-2 and zircaloy-4 in steam
V.F. Urbanic and T.R. Heidrick
Journal of Nuclear Materials
Volume 75, Issue 2, August 1978, Pages 251-261

Abstract: At temperatures above the transformation temperature for zirconium alloys, steam reacts with ?-Zr to form a superficial layer of zirconium oxide (ZrO2) and an intermediate layer of oxygen-stabilized ?-Zr. Reaction kinetics and the rate of growth of the combined (ZrO2 + ?-Zr) layer for Zircaloy-2 and Zircaloy-4 oxidation in steam were measured over the temperature range 1050–1850°C. The reaction rates for both alloys were similar, obeyed parabolic kinetics and were not limited by gas phase diffusion. The parabolic rate constants were consistently less than those given by the Baker and Just correlation for zirconium oxidation in steam. A discontinuity was found in the temperature dependence of both the reaction rate and the rate of growth of the combined (ZrO2 + ?-Zr) layer. The discontinuity is attributed to a change in the oxide microstructure at the discontinuity temperature, an observation which is consistent with the zirconium-oxygen phase diagram.