«Item 7b Severe Accidents Related Issues Preliminary Monitoring Report Report to the Federal Ministry of Agriculture, Forestry, Environment and Water ...»
• The calculations have considerable uncertainty associated with them.
• The calculations did not include modelling combustion of the obtained sensitive hydrogen cloud.
• The calculations were for a generic WWER 1000 and were not entirely specific to Temelín due to lack of access to detailed plant documentation and physical access to the plant to confirm geometric details.
18.104.22.168.4 Implications - The GASFLOW II Results in Perspective The small LOCA sequence was selected for modelling because it was recognized that the containment atmosphere would be potentially conducive to hydrogen combustion (i.e., it would not be steam inerted), and because the hydrogen could be released into the steam generator boxes, which present a relatively confined release area that does not communicate vertically with the containment dome. The conditions in other types of sequences would be different.
If the release point is from the barbotage tank, for example, as it would be in most transient sequences, the hydrogen would not be trapped in the SG box but would instead communicate easily with the upper containment and be dispersed in the greater volume of the containment. In large LOCA core melt sequences, the in-vessel hydrogen release is smaller and the release rate is lower because most of the water in the primary system is lost out the break and is not available to produce steam in-vessel to react with the fuel cladding and produce hydrogen.
For the ex-vessel portion of accidents, in most cases (as a result of the introduction of a design change discussed at the Specialist Workshop), the reactor cavity door would be open and the core debris would spread over a larger area. The hydrogen release rate would (except for the initial phase of MCCI) be lower than for the peak in-vessel period, and the release would be to a less confined area of the containment, which communicates easily with the upper containment along the periphery of the containment. Thus, formation of sensitive hydrogen clouds would be much less likely for the ex-vessel portion of severe accidents in the VVER-1000/320 configuration.
The contribution of small LOCA sequences to the risk profile of Temelín then becomes important. It was identified during the Specialist Workshop that a very conservative initiating event frequency was used for small LOCA, which is about a factor of ten higher than other PSA studies. If a more typical small LOCA initiating event frequency were to be used, the contribution of small LOCA sequences to core damage frequency (CDF) in the updated Temelín PSA would drop from about 3,3×10-6/a (representing 22,1% of CDF) to about 3,3×10-7/a (representing 2,7% of CDF). At such a low CDF, in order for early containment failure to result from energetic hydrogen combustion to pose a significant hazard to Austria,
there would have to be a very high conditional probability of all of the following conditions:
• A sensitive cloud would have to form. Whether a sensitive cloud forms in the SG box depends on the location of the small LOCA along the primary piping (not all small LOCAs are alike in their potential to form a sensitive cloud because some locations allow much more ready communication of the released cloud with the upper containment, and thus easier mixing with the larger containment volumes).
96 ETE Road Map - Preliminary Monitoring Report – Item 7b: Severe Accidents Related Issues
• The sensitive cloud would have to include a sufficient amount of hydrogen and be in an unfavourable geometry at the time of ignition (such that a detonation capable of threatening containment integrity could result). Not all small LOCAs have the same potential in this regard (small LOCAs range in size from 20÷50 mm effective diameter, and could in principle occur anywhere along the primary coolant piping and attached pipes).
• An ignition source would have to be available during the time that the sensitive cloud persists in the required concentration and quantity of hydrogen and in the specific unfavourable geometry. The presence of ignition sources during the transient period when a sensitive cloud is present is difficult to predict, however it should be noted that the PARs themselves could become ignition sources (Fischer 2003). (The precise locations of the PARs are not known to us (although the compartments in which they are located are known), and our calculations have used locations established by expert judgment rather than actual design information.)
• And even with all of these conditions, the cloud would have to undergo flame acceleration and DDT or direct detonation. It is important to understand that sensitive clouds do not always detonate – even with an ignition source present. Deflagrations or series of deflagrations can also be an outcome. This has been well demonstrated by experiments.
In the end, the GASFLOW calculation has showed that the MELCOR code can miss at least some conditions in which a sensitive (i.e., potentially detonable) cloud can form (the MELCOR calculation itself identified only marginally flammable conditions and no detonation potential was identified). Neither the MELCOR nor GASFLOW codes (nor other codes of which we are aware), however, have probabilistic models for combustion allowing for the possible presence or absence of an ignition source and the possible occurrence or non-occurrence of a detonation.
In other types of VVER-1000/320 severe accident sequences than small LOCAs, the potential for formation and ignition of a sensitive cloud appears to be less than in the small LOCA sequence, however this is an expert judgment rather than the result of a series of extensive calculations. The formation of a sensitive cloud (i.e., a cloud containing hydrogen which could upon ignition undergo flame acceleration and deflagration-to-detonation transition) as a result of a severe accident in a WWER 1000 appears as a result of the GASFLOW calculation to be physically possible, but its likelihood is uncertain. The specific sequence modelled has a low absolute frequency (estimated to be of the order of 3×10-7 per year), but this is not the only potentially affected sequence.
4.2.4 LB LOCA Sequences
The case of LB LOCA has very small probability of occurrence, made even smaller in view of the ‘Leak Before Break’ strategy implemented in most NPPs nowadays. In fact, according to the estimates provided in the 2002 Temelín PSA level 1, the frequency of core damage due to LB LOCA is about 3E-8/year, i.e. 100 times less than the frequency of core damage due to medium primary to secondary leakages. In addition the radiological consequences of LB LOCA are much less than those of PRISE severe accidents due to the protection provided by containment in case of LOCA events. Nevertheless, LB LOCA accidents are considered as the sequences with the fastest development of core damage processes and the highest decay heat releases at the time of RPV failure.
Within PN7 project LB LOCA sequences have been studied, and two cases – with molten corium contained within the reactor cavity and with molten corium spreading – have been analyzed. In the latter case the calculations were repeated after the Prague Workshop using the base mat concrete compositions provided by the Czech side and the updated PAR capacity. The main area of interest for this analysis was the challenge of the combustible gases generated to the integrity of the containment. The scenario was chosen because of its duration so that to allow estimation of the oxygen depletion due to operation of the PARs in the containment.
ETE Road Map - Preliminary Monitoring Report – Item 7b: Severe Accidents Related Issues 97 The scenario was assumed to start with a break in the cold leg 200 mm ED (break in the surge line to the pressuriser) with EFW available, HPIS and LIPS not available. To study possible effects of operator errors it was assumed that containment spray system CSS is recovered and actuated at the most challenging containment conditions with and without hydrogen deflagration. This made it possible to evaluate the hazards to containment integrity in case of unplanned hydrogen burn at the worst possible conditions (calculated with MELCOR code).
To study the effectiveness of the strategy of base mat protection against penetration, analyses were made for the case with corium contained inside the reactor cavity and with corium spreading outside cavity over 12 m2 and 100 m2.
The calculations showed that in the case of LB LOCA the penetration of RPV bottom head occurs at 4,21 h, with ejection of about 145 tons of corium and start of MCCI in the reactor cavity. In the case of molten corium remaining within the reactor cavity (steel door closed) the upper concrete layer (serpentinite) is ablated within 14 hours after the accident and the corium starts interaction with the base concrete.
The mass of gases produced during MCCI is high and for hydrogen it exceeds the amount of H2 produced during zirconium oxidation in the core. The mass of hydrogen produced during MCCI in the first concrete layer – serpentinite concrete is about 1100 kg, and in the lower layer of base concrete about 1900 kg. Duration for the base concrete ablation is about 33,6 h. The total time needed for penetration of the base mat is about 43 h.
In the case of corium spread out of reactor cavity the calculations confirmed that corium spreading extends considerably the time till baseman penetration by more than 26 h, so that the baseman is penetrated after about 74 h since the start of the accident.
For the case with spreading of the corium outside cavity the corium stratifies into 3 layers metal layer (MET), heavy oxide layer (HOX) and light oxides (LOX) but very fast transformation takes place and practically from the onset MCCI is with 2 layers - metal and light oxides.
According to the predictions of CORCON code (part of MELCOR) after HOX disappears MCCI becomes more vigorous and intensifies the generation of gases. Therefore this case with large spreading area leads to rather vigorous MCCI. The corium interaction with the base concrete produces 2940 kg of H2 Long term pressure inside containment reaches about 0,55 MPa.
The temperature of the “cavity” compartment is rather low – about 560 K. The temperature of the rest of the containment compartments is around 480 K. The containment pressure increases quite fast and 45 hours after the accident reaches 4,5 bar. The peak recombination rate of 1 PAR for H2 is about 0,18 g/s, and for CO – 0,19 g/s. The total recombined masses of H2 and CO by 1 PAR during the calculated time period of 43 hours are 22,2 kg of H2 and 20,8 kg of CO.
During the first 42 hours the atmosphere is not inerted. The H2 concentration increases above 10 % at time 7,81 h. The amount of oxygen inside the containment is sufficient for deflagration (above 5 %). Therefore at this time deflagration of hydrogen is possible in the containment dome. The depletion of O2 below 5 % due to the operation of the recombiners takes place 32,3 h after the accident.
The total amount of H2 generated (both in-vessel Zr oxidation and from MCCI) is 3240 kg. Of this amount only 518 kg, representing about 16 %, is recombined by the PARs. The rest stays inside the containment posing a threat to the containment integrity in case of deflagration. Without deflagration the maximal H2 concentration reached is about 19 vol.%. First conditions for deflagration appear in the ECCS compartment. Here again we can observe lower steam concentration in these compartments due to larger amount of surfaces per unit free volume and more extensive condensation in these volumes.
The first deflagration takes place at time 7,2 h in the ECCS compartment. The deflagration propagates to the annular corridor, to the SG compartments and to the containment dome.
98 ETE Road Map - Preliminary Monitoring Report – Item 7b: Severe Accidents Related Issues The resulting peak pressure is 4,79 bar. A second deflagration follows at 10,5 h. In this case the initiation of the burning is in the “cavity”, which for this analysis is part of the annular corridor. Then the deflagration propagates to the SG boxes, to the annular corridor and the containment dome. The second pressure peak is higher – 5,55 Bar, but still it does not exceed the design pressure of the containment. The peak temperature reached as a result of the deflagrations is about 1000 OC.
After the second deflagration the amount of O2 is below 1,18 %. Thereafter the recombiners slowly deplete the oxygen. The hydrogen concentration increases due to MCCI and is around 12 vol.%. The steam concentration stabilizes around 57-58 vol.%.
If the steam in the containment is condensed, the concentration of oxygen can increase above 5 vol. %. In the calculations it was assumed that the CSS was started, though the pressure inside the containment does not require such action. It was found that the operation of the spray increases insignificantly the O2 concentration, but no deflagration follows and there is no threat to the containment integrity.
The general conclusion from the analyses is that the hydrogen and oxygen concentrations may pose a challenge to the containments of Temelín NPP, but the situation is far from dramatic as the peak pressure remained within the limits below the design pressure of the containment. It can be noted that the operator is instructed not to actuate CSS at the time when atmosphere is not inerted and wait with spraying until the hydrogen and oxygen concentrations become lowered by action of recombiners below flammability limits. The Czech calculations recognize that the case of an erroneous actuation of sprays the loss of containment integrity would be possible. They stress that the operators are well trained to avoid such errors, and the likelihood of such a scenario is very low.
On the other hand if the RPV fails and the molten corium is released to the reactor cavity, the hazard of basemat penetration is significant.
The calculations with MELCOR code showed that assuming implementation of SAMGs the containment integrity is not threatened by increases of pressure due to long-term gas generation or hydrogen burns. The main hazard consists in the possibility of containment baseman penetration.
4.3 Discussion of Phenomenological Issues