Abstract
As analysts still grapple with understanding core damage accident progression at Three Mile Island and Fukushima that caught the nuclear industry off-guard once too many times, one notices the very limited detail with which the large reactor cores of these subject reactors have been modelled in their severe accident simulation code packages. At the same time, modelling of CANDU severe accidents have largely borrowed from and suffered from the limitations of the same LWR codes (see IAEA TECDOC 1727) whose applications to PHWRs have poorly caught critical PHWR design specifics and vulnerabilities. As a result, accident management measures that have been instituted at CANDU PHWRs, while meeting the important industry objective of publically seeming to be doing something about lessons learnt from say Fukushima and showing that the reactor designs are oh so close to perfect and the off-site consequences of severe accidents happily benign.
Integrated PHWR severe accident progression and consequence assessment code ROSHNI can make a significant contribution to actual, practical understanding of severe accident progression in CANDU PHWRs, improving significantly on the other PHWR specific computer codes developed three decades ago when modeling decisions were constrained by limited computing power and poor understanding of and interest in severe core damage accidents. These codes force gross simplifications in reactor core modelling and do not adequately represent all the right CANDU core details, materials, fluids, vessels or phenomena. But they produce results that are familiar and palatable. They do, however to their credit, also excel in their computational speed, largely because they model and compute so little and with such un-necessary simplifications.
ROSHNI sheds most previous modelling simplifications and represents each of the 380 channels, 4560 bundle, 37 elements in four concentric ring, Zircaloy clad fuel geometry, materials and fluids more faithfully in a 2000 MW(Th) CANDU6 reactor. It can be used easily for other PHWRs with different number of fuel channels and bundles per each channel. Each of horizontal PHWR reactor channels with all their bundles, fuel rings, sheaths, appendages, end fittings and feeders are modelled and in detail that reflects large across core differences. While other codes model at best a few hundred core fuel entities, thermo-chemical transient behaviour of about 73,000 different fuel channel entities within the core is considered by ROSHNI simultaneously along with other 15,000 or so other flow path segments. At each location all known thermo-chemical and hydraulic phenomena are computed. With such detail, ROSHNI is able to provide information on their progressive and parallel thermo-chemical contribution to accident progression and a more realistic fission product release source term that would belie the miniscule one (100 TBq of Cs-137 or 0.15% of core inventory) used by EMOs now in Canada on recommendation of our national regulator CNSC. ROSHNI has an advanced, more CANDU specific consideration of each bundle transitioning to a solid debris behaviour in the Calandria vessel without reverting to a simplified molten corium formulation that happily ignores interaction of debris with vessel welds, further vessel failures and energetic interactions. The code is able to follow behaviour of each fuel bundle following its disassembly from the fuel channel and thus demonstrate that the gross assumption of a core collapse made in some analyses is wrong and misleading. It is able to thus demonstrate that PHWR core disassembly is not only gradual, it will be also be incomplete with a large number of low power, peripheral fuel channels never disassembling under most credible scenarios.
The code is designed to grow into and use its voluminous results in a severe accident simulator for operator training. It’s phenomenological models are able to examine design inadequacies / issues that affect accident progression and several simple to implement design improvements that have a profound effect on results. For example, an early pressure boundary failure due to inadequacy of heat sinks in a station blackout scenario can be examined along with the effect of improved and adequate over pressure protection.
A best effort code such as ROSHNI can be instrumental in identifying the risk reduction benefits of undertaking certain design, operational and accidental management improvements for PHWRs, with some of the multi-unit ones handicapped by poor pressurizer placement and leaky containments with vulnerable materials, poor overpressure protection, ad-hoc mitigation measures and limited instrumentation common to all CANDUs. Case in point is the PSA supported design and installed number of Hydrogen recombiners that are neither for the right gas (designed mysteriously for H2 instead of D2) or its potential release quantity (they are sparse and will cause explosions).
The paper presents ROSHNI results of simulations of a postulated station blackout scenario and sheds a light on the challenges ahead in minimizing risk from operation of these otherwise unique power reactors.