«Item 7b Severe Accidents Related Issues Preliminary Monitoring Report Report to the Federal Ministry of Agriculture, Forestry, Environment and Water ...»
On the other hand the penetration of the basemat due to MCCI propagates and the basemat failure can be expected before 40 hours. Thus the main hazard in case of LOFW would be connected with containment basemat melt-through in a situation where the concentration of H2 in the cavity atmosphere is high.
18.104.22.168 SB LOCA
22.214.171.124.1 Evaluation of Depressurisation Behavior in Case of 15 mm LOCA In order to find out the possibilities of RCS depressurization by Temelín systems in case of a LOCA with smaller leaks, the consequences of SB LOCA 15 mm were calculated. In these analyses total loss of FW and failure of all 3 HPIS trains was assumed. In addition to the loss of HPIS, also the loss of high-pressure boron injection pumps was assumed. The 4 SITs and all 3 LPIS trains were assumed to be available. Thus, to establish water injection to the primary system, it was necessary to achieve depressurization down to the SITs pressure and thereafter – down to the LPIS injection pressure 92 ETE Road Map - Preliminary Monitoring Report – Item 7b: Severe Accidents Related Issues AM procedure investigated was RCS depressurization using the emergency gas removal system (EGRS) lines from the top of the RPV and from the pressurizer. At late stages of the accident progression it was necessary to depressurize the primary system also by opening of the pressurizer’s PORV in order to establish stable LPIS injection and cooling of the core.
The criterion for initiation of the SAM procedure was core exit temperature exceeding 650 °C.
The calculations showed that the Emergency Gas Removal System allows to depressurize the RCS and prevent RPV failure. This - as ETE stated - is the case only, if the criterion for entry into SAMG is set low (650 °C) and the achievable flow rate though EGR is high (above 30 kg/s).
However, in order to reach the stable safe state it was necessary also to open PORV in the last stage of the accident. This indicates that the EGR system may be unable to depressurize RCS by itself, without using PRZ PORV. Moreover, in the calculations the flow rates of EGR system were distinctly higher than the upper limit established in rough estimates for that system, namely 30 kg/s.
Therefore, one more calculation was run with the same initial conditions, but with the maximum EGR system capacity limited to 20 kg/s and with entry into SAMG criterion established as 650 oC. In this case the operation of EGRS was not sufficient to depressurize the primary system. The accident evolved into a severe accident with core melting.
Finally, one more case was studied, namely with total loss of FW (both normal and emergency) and thus – loss of heat sink, but with all 3 HPIS trains available, all 4 SITs and all 3 LPIS trains available. As before the loss of high-pressure boron injection pumps was assumed. The entry point to SAMGs was set at core exit T 650 OC.
The calculations showed that with entry point 650 OC, with EGR capacity limited to 20 kg/s and with HPIS available it is possible to succeed in a partial recovery and cooldown of the core. However, also in this case it will be necessary to depressurize with PORV to achieve stable cooling of the core.
126.96.36.199.2 Severe Accident Calculation for 50 mm Small LOCA A severe accident scenario involving a small LOCA was also modeled in the PN7 project.
This scenario, which is identified in the PSA as the most likely severe accident sequence (but see comments below in this regard), results from a small LOCA with common cause failure of high and low pressure injection and containment sprays. The primary and secondary pressure relief valves (PORV and BRU-A valves) were assumed to be available as was the charging pump system (consistent with the PSA definition of the scenario), so the accident progression was modeled with correct operator action to depressurise the reactor coolant system. Thus, at the time of vessel failure, the pressure in the primary coolant system was low and high pressure melt ejection from the vessel was avoided.
The small LOCA was modeled as occurring in the steam generator box where the pressurizer is located, and the break was modeled as occuring on the primary loop which includes the pressurizer (this was done in order to evaluate the effect of the break location on the "loop seal" in the affected loop).
The operators were modeled as correctly connecting the TB10 tanks to the charging pumps.
This provides additional charging pump water inventory to extend injection from the charging pumps. The charging pump inventory was calculated to be exhausted by 5,2 hours. As a result of operator action to depressurize the reactor coolant system, the hydroaccumulators (SITs) also discharged their inventory to the reactor coolant system. This action was completed within 3,2 hours of the start of the sequence. At 5,9 hours, continued loss of coolant inventory from the break results in the onset of core uncovery. Metal-water reaction leading to the production of hydrogen commences at 7,2 hours. Reactor pressure vessel failure was calculated to occur at 8,3 hours. Reactor coolant system pressure remains low through the time of vessel failure as a result of actions by the operators (consistent with the EOPs and SAMGs) to depressurize the reactor coolant system.
ETE Road Map - Preliminary Monitoring Report – Item 7b: Severe Accidents Related Issues 93 The small LOCA sequence was also run without operator action to depressurize the reactor coolant system In this case, the core damage and vessel failure events were considerably later (with vessel failure after 13 hours). However, the primary system pressure is borderline with respect to melt ejection, and since more of the energy in the primary system is transferred through the break to the containment (instead of to the environment via the BRU-A through depressurization efforts by the operators), the containment pressure is considerably higher in this case. The containment pressure was calculated by MELCOR to be near 0,8 MPa at the time of vessel failure. As estimated by CEZ [Sýkora 01 a], at a pressure of 0,8 MPa the conditional probability of containment failure is about 5 % (one chance in 20).
188.8.131.52.3 Detailed Hydrogen Distribution Modeling With GASFLOW II The in-vessel MELCOR 1.8.5 modeling of the small LOCA sequence (base case) was used as input to the GASFLOW II code to allow for a more detailed modeling of hydrogen combustion potential by estimating local, heterogeneous gas distribution within the containment. It is important to understand how MELCOR models hydrogen combustion, in order to appreciate how the GASFLOW code identifies conditions which are so different from those encountered with MELCOR.
MELCOR is a "lumped parameter code" – it does not model hydrogen distribution in a mechanistic manner. The MELCOR code requires the code user to define calculational nodes, which are assumed to be homogeneous. Such "nodalization" is an art, more than a science, because a calculational cell, which has homogeneous conditions in one part of an accident may not be homogeneous in another part of the same accident scenario. If the nodes are made too large, anomalous results can occur (if the nodes are made too small, calculation times become very long).
For example, take the case where a small opening into a large compartment conveys a concentrated hydrogen gas stream. In the MELCOR code, the amount of hydrogen conveyed into the large room during the time step of the code is calculated. Then that hydrogen is instantaneously and homogeneously mixed by the code into the entire volume of the compartment. Then the code checks to see if minimum combustion criteria are met. If not, no burn occurs. If so, then the code models a burn as occurring, irrespective of whether an ignition source is present. Moreover, the code can only model deflagrations or diffusion flames - the code is inherently incapable of modeling flame acceleration, deflagration-to-detonation transition, or global detonation.
What may happen in the real (three-dimensional) world is as the hydrogen gas enters the compartment; a region of combustible or maybe even detonable mixture is created. When homogeneously "smeared" over the entire compartment volume by the MELCOR code, noncombustible conditions can result, but in reality some fraction of the volume is combustible and perhaps even detonable.
But any time the MELCOR code calculates that minimum combustible criteria are obtained in a time step, a burn is initiated. The hydrogen burning is in reality a stochastic process. Hydrogen does not always burn when combustible conditions exist. Hydrogen does not always detonate when detonable conditions exist. But MELCOR is only capable of identifying that the conditions are met, then numerically deflagrating the hydrogen or flagging the attainment of a detonable concentration. Experiments have been conducted in which decidedly detonable conditions existed (e.g., 24 vol.% hydrogen with sufficient oxygen to support combustion and insufficient steam to prevent combustion) without a detonation occurring.
If a code user gets results from a lumped parameter code that indicate that a number of mild deflagrations occur over a period of minutes or hours and no pressure threatening containment was calculated, what is the code user supposed to deduce from this result? One must be very cautious in interpreting hydrogen combustion results from lumped parameter codes, because burning the hydrogen each and every time that combustible conditions exist - which 94 ETE Road Map - Preliminary Monitoring Report – Item 7b: Severe Accidents Related Issues is precisely what MELCOR does - is the functional equivalent of assuming that an ignition source is always present, when in fact there may be no ignition source.
GASFLOW II allows the prediction of local, heterogeneous gas behaviour during severe accidents - something which lumped parameter codes such as MELCOR (which assume homogeneous conditions within each calculation node) are incapable of doing. The consequences of deflagration at any given time can then be modelled with a code appropriate to the conditions at the time (e.g., a code which can model DDT, detonation, etc.).
The actual calculation of hydrogen combustion was not performed based on the GASFLOW II modelling. Rather, the goal was to understand whether, based on 3D CFD modelling, conditions conducive to energetic hydrogen combustion modes could occur in the WWER 1000 design, and to gain an understanding of where such conditions could develop and over what time period they might persist.
Two GASFLOW II calculations were performed for the ’in-vessel’ portion of the accident. This was considered to be the best use of the available resources and time. Extended calculations covering the full duration of the accident were not possible within the budget and time duration available for the calculations. It should be noted that the in-vessel portion of the accident is associated with the highest hydrogen release rate to the containment and the formation of a sensitive cloud is most sensitive to hydrogen release rate rather than to the total quantity released.
The ’base case’ calculation used the hydrogen source to the containment from the WWER 1000 small LOCA sequence calculated with MELCOR 1.8.5. The base case showed a peak hydrogen release rate of 0,2 kg/s into the steam generator box containing the pressuriser.
Another "sensitivity" case was also performed, using a synthetic hydrogen source from another design. In this case, the hydrogen was released from a primary loop in the other steam generator box (i.e., the SG box without the pressuriser opening to the upper containment).
This was done to investigate whether this SG box could be more sensitive to hydrogen mixtures, and also to understand whether a larger hydrogen release rate (a peak rate of 0,8 kg/s) could yield sensitive mixtures over a larger volume and longer time period.
The base case calculation was based on the MELCOR 1.8.5 prediction of steam and hydrogen source rates. The result was that a very small, transient sensitive hydrogen cloud was formed in the immediate vicinity of the release point from the primary system, but that the sensitive mixture rapidly dissipated and no large, sensitive hydrogen cloud formed in the SG box. One interesting result is that over the time frame covered by the calculations, the passive autocatalytic recombiners (PARs) in the lower part of the containment do not "see" the hydrogen. Thus, recombination rates are quite limited for these recombiners. In addition, as the peak hydrogen release rate extends for only limited periods of time (of the order of 500÷1000 seconds), even the PARs, which do "see" the hydrogen don't have much of an effect in terms of depleting the hydrogen cloud of its inventory.
At the time this calculation was done, only the MELCOR 1.8.5 results of source rates were available. Other more detailed, mechanistic code calculations were being done with the SCDAP-RELAP code in order to ’qualify’ the MELCOR code results, but these calculations were not available at the time. Once the results were obtained from the first GASFLOW calculation, it was decided to run the code again with a higher hydrogen release rate. As this release source was from a different type of reactor, it was deemed to be a ’synthetic’ hydrogen source and the calculation was deemed to be a sensitivity calculation. This synthetic hydrogen source had a peak hydrogen release rate about 4 times higher than the base case.
In the sensitivity case, a rather large sensitive hydrogen cloud formed and persisted for several hundred seconds. The cloud occupied the upper portion of the SG box, and extended into one portion of the containment dome near the release point from the SG box to the containment annulus (which communicates between the upper and lower containment around the inner edge of the containment).
ETE Road Map - Preliminary Monitoring Report – Item 7b: Severe Accidents Related Issues 95 After these calculations were completed, the results of the SDCAP-RELAP calculation became available. These results indicate that the 0,8 kg/s release rate, which was used in the ’sensitivity case’ using a synthetic hydrogen source, was not as conservative as previously believed. The SCDAP-RELAP calculation showed several periods in which peak hydrogen release rates exceeded 0,8 kg/s. Accordingly, the 0,8 kg/s release rate must be regarded as at least plausible since it is based on mechanistic code calculations for the WWER 1000 reactor coolant system.
The obtained results from the GASFLOW II calculations need to be understood in proper