Focus on Russia | Safety
Stopping melt-through3 July 2012
An analysis performed during a European Union pre-accession instrument (PHARE project) in Bulgaria at Units 5&6 of the VVER-1000, V-320 Kozloduy Nuclear Power Plant discovered a vulnerability of this design consisting of early (one-hour) containment melt-through via ionization chamber channels situated around the reactor pit. An original technical solution has been developed and examined with thermo-mechanical analyses and experiments. By Dimitar Popov and Vladimir Yurukov
In a severe nuclear meltdown scenario, a rupture of the reactor pressure vessel (RPV) ends the in-vessel phase, and begins the ex-vessel phase. The melt falling down into the reactor pit begins to ablate the concrete in both axial and radial directions. An array of 27 ionization chambers (IC) circle the reactor at a distance of 145 mm from the wall of the reactor pit. They are used to house flux monitors and neutron physics experiments. For VVER-1000, V-320 it was found that the melt would ablate the side concrete wall of 145 mm thickness in about 45 min. after the melt release from RPV and would penetrate into IC channels .
The melt (mixed with ablated concrete) would breach and fall down the channels, which are plugged in the design by a protective layer of a biological shield of cloth laminate, with concrete and steel slabs beneath. As will be shown, the biological shield (insert) made in the original design from Teflon has a minimal margin to initial melting (TS˜327 °C). There are also small holes through these layers and slabs for the IC cables. Further ablation of the concrete and slabs over perhaps 15-20 minutes would lead to melt-concrete penetration into rooms designated for IC lifting apparatus beneath them, and outside containment.
To prevent the further penetration of the melt into these spaces, the authors propose a solution to plug the bottom of IC channels with high temperature-resistant materials.
The engineering solution for plugging the IC channels is illustrated in Fig. 1 and consists of a plug and a ball. The plug is designed to be made of Titanium Carbide, TiC, with melting point TM= 3170°C and density ?=4930 kg/m3. The ball is designed to be made of Tungsten Carbide, WC, with melting point TM= 2870°C and density ?=15800 kg/m3 . The melting points of the above materials are chosen conservatively to be higher than the melting point of UO2. It is also proposed to replace the original biological shield by Teflon or light concrete.
The principle of operation of the plugging device consists of the following. During a severe accident with the discharge of molten corium in the reactor pit, the melt ablates the side concrete wall, and melt mixed with concrete falls down into the IC channels. The temperature of the melt is high enough to burn through the power cable, which falls through the hole, and releases the ball (in normal conditions the 20mm-wide cable holds the ball). The ball, being almost twice as heavy as the melt (?UO2=8740±200 kg/m3 , but which also includes ZrO2, other light metals and concrete that lowers its average density), falls over the central orifice (designed to admit the IC power cable) and plugs it without risk of being dislodged.
Thermo-mechanical analyses were done by simulation of the real process of melt penetration into IC channels. Finite-element models (FEM) were built, which include the proposed plugging devices and the adjacent parts that would be influenced as well—the steel tube, an exterior steel tube, concrete of the containment wall, concrete fill between the two tubes, the biological shield below the plugs (made from Teflon) and the penetrating melt itself (Fig. 2).
The CAD-CAM package NISA II – DISPLAY III code was used to perform the FEM calculations by applying the updated HEAT-III module for phase transition and radiation heat transfer.
Taking into account the geometry of the construction and adjacent parts, which are rotationally symmetric about the vertical axis of the IC channel, as are their material properties and applied loads, FEM were developed by using 2D axisymmetric (pseudo-3D) elements, which are quadrilateral type with mid-side nodes.
The thermo-mechanical analyses included the consecutive solution of two connected tasks. First, a heat transfer task was solved to find out the temperature distribution, that is, the temperatures for each point of the modelled object for characteristic moments of time (Figs 5&6). The second task was to find the stress distribution for the chosen characteristic moments, taking for thermal loads the temperature distributions solved in the first task. Preliminary stress analysis of ceramic materials has shown that material properties of the proposed system depend on their method of fabrication, which is yet to be decided. As a result those analyses have been excluded from this article, but will be carried out after manufacture. In addition, other mechanical loads such as the weight of the melt and pressure were considered for static strength calculations of the system.
The general model scheme used for the thermo-mechanical analyses with modelled parts description are shown in Fig. 2. The 2D axisymmetric solid FEM mesh used for thermal resistance (thermal stability) analysis is shown on Fig. 3.
An equal-to-maximum-allowed design temperature of 150°C for the containment was postulated for all parts (nodes) of the system as initial condition for the heat transfer task. The initial temperature of the melt was conservatively postulated as 2850°C, the specific volumetric power generated by the melt was chosen according to  as V=1.3 MW/m3. The other parameters are taken from  and  and shown in Table 1.
For the clearance between the plug and the tube a process of radiation heat transfer was modelled using the Stefan–Boltzmann law, E(T)=s*e*T4, where s is the Stefan-Bolzmann constant and e is emissivity. For the purposes of FEM calculations, the accepted value of the emissivity of the external cylindrical surface of TiC plug is eTiC=0.75. Several sensitivity calculations were made for Teflon at conservative values e=0.3 and e=0.5, but also e=0.99 (black Teflon). Based on the obtained results, eTeflon=0.5 was chosen. The lack of any convective cool down of the plugging ball, the plug and the biological shield (insert) was also accepted as a conservative measure.
Twelve characteristic nodes in the FEM mesh were chosen to represent the heat-transfer process (shown in Fig. 2 and Fig. 4).
The temperature history of the representative 12 nodes for the first 600 seconds is given in Fig. 5a, while on Fig. 5b it is shown for the first 24 hours.
In Figs 6a-6d, the temperature distribution from FEM analysis are shown for four characteristic moments within first 24 hours after the melt pouring in the IC channel.
The analysis of the results has shown that the thermal stability of all parts of the proposed device and adjacent components is guaranteed.
Two experiments were performed to prove the operability of the proposed devices for plugging the IC channels under normal operation and severe accident conditions.
In a cold experiment, the device was modelled in M=1:1 scale, with a plug and plugging ball made of steel, embedded in a 5 mm steel tube of 108 mm diameter (the largest type of channel) and with a real IC power cable. The first aim of the cold experiment (at room temperature) was to prove that the plugging ball and the device as whole will not disturb the movement of the IC power cable during normal operation, in which the device would be expected to be lowered for replacement. The second aim of the cold experiment was to prove the plugging performance of the ball.
Accordingly, the experiment consisted of two phases. First, the cable was shifted up and down, and the force of the ball exerted on the cable was measured by a digital dynamometer. The success criterion consisted in not exceeding the theoretically-calculated effort (13.9 kg) by more than 10% (Fig. 7). The greatest peaks are related to the initial pull-up/pull-down of the cable.
In the second phase, the cable was left to fall down under its own weight through the central orifice of the plug to see whether it would squeeze through at its overall 3m length, thus imitating the melting cut of the cable. The success criterion was that the cable must not hinder the ball to plug fully the orifice. Both phases of the cold experiment were successful.
The purpose of the hot experiment was to prove that a melt simulant with similar thermodynamic properties to the real melt will freeze in the clearance of 2 mm between the plug and the steel tube without penetrating below the plug. So the success criterion of the hot experiment was if the melt, after being poured into a 5 mm-thick 108 mm-diameter steel tube, would freeze in the clearance within the height of the plug (that is, within less than 200 mm).
For this purpose a melt simulant consisting of mixture of Al and TiC in a 65:35 proportion by weight was used. The simulant was melted in an induction furnace at T~700°C. The estimated viscosity of the mixture, using , at this temperature is in the range of ?=0.01-0.1 Pa.s, which is consistent with the results published in  and . The calculated hidden phase transition heat of Al-TiC mixture is ?Hf˜ 235.2 KJ/kg, while for a mixture of 88%UO2+12%concrete (this proportion was calculated in ) is ?Hf= 228±13 KJ/kg.
A K-Al-F flux (9.6% by weight) was added to the simulant for better absorption (wettability) of TiC in the aluminium, based on the information in ,  and . A preliminary experiment of adding K-Al-F flux to improve the mixing of TiC and Al with smaller quantities at the same proportions was performed by electron microscopy observation and metallographic analysis.
The melt was poured into the tube of the model heated preliminary in a resistance furnace. It was found, after cooling down of the model and cutting of the tube and the plug, that the melt was frozen over a depth of 20 mm below the 2 mm gap, which is 10% of the height of the plug (Fig. 9). Thus, the hot experiment was successful.
In conclusion, the authors consider that the implementation of the proposed engineering solution will be effective for prevention of early containment failure due to melt-through in case of severe accident and will considerably reduce the value of the Large Early Release Frequency (LERF) calculated within PSA Level 2 for VVER-1000. The realization of the proposed design is a condition for the licences of NPP Kozloduy Units 5&6 and is planned for 2014.
This solution can be also of interest for other designs of existing reactors that can have ionization chamber channels close to the wall of their reactor pit.
This article was first published in the June 2012 issue of Nuclear Engineering International
Dimitar Popov, Kozloduy NPP Plc., 3321 Kozloduy, Bulgaria, and Vladimir Yurukov, ATP-AtomToploTroekt Ltd, Sofia, Bulgaria.
This topic was presented at the International Experts’ Meeting on Reactor and Spent Fuel Safety in the Light of the Accident at the Fukushima Daiichi Nuclear Power Plant, IAEA, Vienna, Austria, 19-22 March 2012.
The hot experiment was conducted in the Aluminium factory “Alumina”, Pleven, Bulgaria, which provided all necessary I&C equipment, induction and resistance furnaces and crucibles. The cold experiment was performed at NPP Kozloduy. The metallographic works were conducted in the Technical University of Sofia.
 B.R. Sehgal, T.N. Dinh, R.S.Rao, VVER-1000/V320 Plant Safety Evaluation Report. Severe Accident Risk Assessment and Severe Accident Management, EC PHARE Project No BG 01.10.01, 2005.
 D. M. Considine, Van Nostrand's scientific encyclopedia, New York, 1988
 T.G.Theofanous et al., In vessel coolability and retention of a core melt, DOE/ID-10460 (Appendix L), Oct.1996.
 M. Ramacciotti, et al, Viscosity models for corium melts, Nuclear Engineering and Design 204 (2001) 377-389.
 http://www.insc.anl.gov/matprop/ uo2/index.php
 T. Sevon, Molten Core-Concrete Interactions in Nuclear Accidents. Theory and Design of an Experimental Facility. VTT Research Notes 2311, Espoo 2005.
 A.T.Dinsdale and P.N.Quested, The Viscosity of Aluminium and its alloys: a review of data and models, NPL Materials Centre, National Physical Laboratory, Teddington, UK TW11 0LW
 A.R.Kennedy, A.E.Karantzalis, The Incorporation and Redistribution Behaviour of Ceramic Partiles in Liquid Al and the Relation to Their Wetting Characteristics, Proc.International Conference High Temperature Capillarity, 29.06-02.07.1997, Cracow, Poland.
 A.R.Kennedy, Reinforcement Selection for MMCs Based on Wetting Information, Proc.International Conference High Temperature Capillarity, 29.06-02.07.1997, Cracow, Poland.