Operating CANDU PHWRs present significant challenges with respect to their ability to mitigate accidents that are beyond the envelope of design basis drafted over 40 years ago. Today, consideration of severe accidents is a public as well as a regulatory requirement whose implementation begs serious reconsideration in an international coordinated effort. The PHWR enhanced vulnerabilities to accidents such as a sustained loss of AC power, as in Fukushima, arise not only out of the inherent design features but also out of the institutional arrangements that surround their licensing. For one, the reactors will, in absence of a containing pressure vessel as in PWRs, put fission product activity directly into the containment, sport multiple potential containment bypass vulnerabilities and produce copious amounts of flammable gases due to presence of large amounts of Zircaloy in fuel channels and carbon steel in feeders. The relatively thin walled, stepped, welded Calandria vessel into which the disassembled core debris will rest has potential to mechanically fail early, causing explosive and energetic interactions of hot debris with enveloping water. This can catastrophically fail the reactor structures. For severe accidents the containments, well designed for design basis accidents, are either small and weak as in single unit plants or unable to practically take any significant over pressure in negative pressure multi-unit reactor buildings that depend upon a single vacuum building, too small for a multi-unit severe accident.
The paper presents analytical arguments in support of these observations, lists conclusions from a series of design reviews and discusses development of ROSHNI, a new generation PHWR dedicated computer code package for simulating an unmitigated station blackout scenario. It does not directly address the institutional issues that handicap a potent reduction of the residual risk posed by continued operation of these reactors without serious design upgrades but discusses the regulatory failures in this regard. It introduces ROSHNI, a newly developed severe accident simulation package that models the reactor core in a greatly enhanced detail necessitated by the variability amongst reactor fuel channels. For a single unit CANDU 6 reactor, the code simulates thermal-mechanical degradation of 4,560 fuel bundles in 380 diverse fuel channels individually (for a total of over 70,000 dissimilar fuel entities) and computes source terms into containment of flammable deuterium gas and fission products. A number of questions are raised about differences between Hydrogen source terms and mitigation measures that are being implemented for light water reactors and Deuterium specific reaction kinetics in generation and mitigation that must be clearly differentiated but ignored so far by PHWR operators. A discussion of effectiveness of certain severe accident specific design upgrade measures that have been implemented at some operating plants is also addressed. For example, potential for a smaller than optimum number (for severe accidents) of PARS units to actually cause Deuterium/Hydrogen explosions as an unintended consequence is discussed. Continued reluctance of CANDU utilities to address a long standing issue of inadequacy of reactor overpressure protection is also detailed.