An experimental study on pressure drop and dryout heat flux of two-phase flow in packed beds of multi-sized and irregular particles

This paper is concerned with debris bed coolability in a postulated severe accident of light water reactors, where the debris particles are irregular and multi-sized. To obtain and verify the friction laws predicting the hydrodynamics of the debris beds, the drag characteristics of air/water single- and two-phase flow in a particulate bed packed with multi-sized spheres or irregular sand particles were investigated on the POMECO-FL test facility. The same types of particles were then loaded in the test section of the POMECO-HT facility to obtain the dryout heat fluxes of the particulate beds heated volumetrically. The effective (mean) particle diameter is 2.25 mm for the multi-sized spheres and 1.75 mm for the sand particles, determined from the Ergun equation and the measured pressure drop of single-phase flow through the packed bed. Given the effective particle diameter, both the pressure drop and the dryout heat flux of two-phase flow through the bed can be predicted by the Reed model. The experiment also shows that the bottom injection of coolant improves the dryout heat flux significantly and the first dryout position is moving upward with increasing bottom injection flowrate. Compared with top-flooding case, the dryout heat flux of the bed can be doubled if the superficial velocity of coolant injection is 0.21–0.27 mm/s. The experimental data provides insights for interpretation of debris bed coolability (how to deal with the multi-sized irregular particles), as well as high-quality data for validation of the coolability analysis models and codes.     

Experimental results on the coolability of a debris bed with multidimensional cooling effects

Within the reactor safety research, the removal of decay heat from a debris bed (formed from corium and residual water) is of great importance. In order to investigate experimentally the long term coolability of debris beds, the scaled test facility “DEBRIS” (Fig. 1) has been built at IKE. A large number of experiments had been carried out to investigate the coolability limits for different bed configurations ( [Rashid et al., 2008], [Groll et al., 2008] and [0055]). Analyses based on one-dimensional configurations underestimate the coolability in realistic multidimensional configurations, where lateral water access and water inflow via bottom regions are favoured. Following the experiments with top- and bottom-flooding flow conditions this paper presents experimental results of boiling and dryout tests at different system pressures based on top- and bottom-flooding via a down comer configuration.A down comer with an internal diameter of 10 mm has been installed at the centre of the debris bed. The debris bed is built up in a cylindrical crucible with an inner diameter of 125 mm. The bed of height 640 mm is composed of polydispersed particles with particle diameters 2, 3 and 6 mm. Since the long term coolability of such particle bed is limited by the availability of coolant inside the bed and not by heat transfer limitations from the particles to the coolant, the bottom inflow of water improves the coolability of the debris bed and an increase of the dryout heat flux can be observed. With increasing system pressure, the coolability limits are enhanced (increased dryout heat flux).     

The effect of lateral flooding on the coolability of irregular core debris beds

The coolability of ex-vessel core debris is an important issue in the severe accident management strategy of, e.g. the Nordic boiling water reactors. In a core melt accident, the molten core material is expected to discharge into the containment and form a porous debris bed on the pedestal floor of a flooded lower drywell. The debris bed generates decay heat which must be removed by boiling in order to stabilize the debris bed and to prevent local dryout and possible re-melting of the material. The STYX test facility which consists of a cylindrical bed of irregular alumina particles has been used to investigate the effect of lateral coolant inflow on the dryout heat flux of the particle bed. The lateral flow was achieved by downcomers attached on the sides of the test rig. The downcomers provide coolant into the lower region of the bed by natural circulation. Both homogenous and stratified bed configurations have been examined. It was observed that the dryout heat flux is increased by 22-25% for the homogenous test bed compared to the case with no lateral flooding. For the stratified configuration with a fine particle layer on top of the bed, no significant increase in the dryout heat flux was observed. The experiments have been analyzed by using the MEWA-2D code. Models which include explicit consideration of gas-liquid friction were used in the calculations in order to realistically capture the lateral flow configuration.     

Dryout Heat Flux for Core Debris Bed, (I)

For gaining basic data on decay heat coolability of debris bed in the post-accident heat removal, measurement of dryout heat flux was made, with stagnant water as coolant, in a 50mm I. D. pyrex glass cylinder vessel. The fuel debris bed subjected to decay heat was simulated by steel ball particles which were inductively heated with a power supply of 20 kHz and 30 kW. The bed was made of homogeneous size particles. An emphasis was placed on the influence of system pressure and particle size. The experiment covered the ranges over the steel ball diameters of 0.3–4.0 mm and the system pressure of 0.02–0.5 MPa.

The experimental results, as a whole, agreed fairly well with the prediction based on Lipinski's 0-D model with respect to the dependence of dryout heat flux both on pressure and on particle size. In detail, however, the dryout heat flux deviates toward a lower value at a higher pressure while to a higher value for a smaller size particle bed. Comparison of the results between the free and fixed beds suggests that the deviation to the higher side will be attributed to the channeling and/or levitation.     

Dryout Heat Flux for Core Debris Bed, (III)

Mixture of cylindrical steel pellets and Al₂O₃ balls, which simulated the intact fuel pellets and fragmented claddings, respectively, was inductively heated in a 50 mm I.D. pyrex glass cylinder filled with water, to investigate the coolability of TMI-2 type degraded core debris bed. The size of steel pellets was 11 mm dia. × 11 mm for BWR, 8 mm dia. × 12 mm for PWR and 5.5 mm dia. × 9 mm for FBR and Al₂O₃ balls were about 2 mm in diameter. The height of the debris bed was 25 cm or lower.

The dryout heat flux does not level off up to a bed height of 25 cm or over for the TMI-2 type bed while 8 cm or so in the bed of only steel balls. The dependence of dryout heat flux on the system pressure agrees with the Lipinski's 0-D model by adopting a proper equivalent diameter. When a simple number-weighted average is used as the equivalent diameter, the prediction gives a fairly good agreement with the experiment for FBR type bed but underestimations for the PWR and BWR type beds. It should be noted, that the small balls of less fraction, not the large pellets, substantially govern the dryout. When the coolant flow is allowed from the bottom, however, the dryout heat flux is enhanced up to the level for the complete vaporization of coolant, and small amount of mass flux or circulation head can greatly improve the coolability.     

Boiling experiments for the validation of dryout models used in reactor safety

The coolability of fragmented corium is a major issue in reactor safety. Since the long-term coolability of such particle beds is limited by the availability of coolant inside the bed and not by heat transfer limitations from the particles to the coolant, the pressure field inside the debris has a strong effect on the cooling potential in multi-dimensional cases as expected in severe accidents in light water reactors (LWR). Therefore, the determination of the pressure field for two-phase flows in porous media is one central point of interest.In this context simulation models and in particular dryout models were developed for reactor safety analyses which have to be validated by reliable experimental data. Therefore, basic experimental investigations have been carried out with inductively heated steel balls of 6 or 3 mm diameter to provide a database for the validation and modification of the friction laws included in these dryout models.The performed boiling and dryout experiments show clearly that models without the explicit consideration of the interfacial drag cannot predict the pressure distribution inside a boiling particle bed, not even qualitatively. Against it, models with an explicit consideration of the interfacial drag can describe the distribution of pressure inside a boiling particle bed.     

Basic laws and coolability of particulate debris: comments on the status and present contributions

As already indicated by the contributions in Part 1 of this issue, an enhanced coolability can be expected especially from multi-dimensional coolant flows in non-uniform particle beds, compared with earlier analyses based on one-dimensional configurations under top flooding and related dryout heat fluxes (DHF). However, assured basic local laws of friction and heat transfer in such beds are required to really evaluate the coolability potential with multidimensional computer codes. Thus, as a common subject, this part of the present issue refers to experimental investigations and related modelling on basic laws of friction and heat transfer in particle beds. Since such laws are often investigated mainly with respect to dryout phenomena (“dryout heat flux”: DHF) this connection is also addressed here. A general impression is that the approaches are widely spread, as concerns the specific experimental aims and methods, the range of conditions analysed, the phenomena emphasised, the empirical laws derived and the conclusions indicated for coolability, especially under reactor conditions. Thus, it appears to be important to confront these approaches with one another in order to find a common line of major subjects and of derivation or validation of relevant empirical laws. Certainly, this cannot be finally reached here and by the present issue. A first step is undertaken here, trying to discuss remaining problems and tasks for future work.     

SILFIDE experiment: Coolability in a volumetrically heated debris bed

The SILFIDE facility was designed at EDF R&D to study the coolability of a debris bed in multidimensional configurations. The choice of induction heating mode was motivated by the necessity for the heat to be generated primarily within the solid particles instead of using external heaters. Particles were simulated by steel beads with diameters ranging from 2 to 7 mm. Coolability is significantly better in terms of CHF values in comparison with the Lipinski 1D formulation applied to the conditions (especially when considering vertically integrated power) where the dryout occurs first. The experiments also show that bottom coolant injection is at least two times more efficient than top coolant injection. With increasing thermal power, steady temperature overheats up to 200 °C above saturation were observed, and the bed was still coolable. Combined phenomena considered as responsible are discussed: steam flow cooling with or without entrained liquid droplets in post-dryout regime, and preferential paths of fluid in porous media. As a consequence, the critical heat flux definition under such conditions must be considered with care. Especially, a failure of coolability cannot necessarily be concluded only from the occurrence of dry zones.     

Experimental characterization of the effective particle diameter of a particulate bed packed with multi-diameter spheres

This paper is concerned with uncertainty reduction in coolability analysis of a debris bed formed in fuel-coolant interactions (FCI) during a postulated severe accident of LWRs. A test facility named POMECO-FL was designed and set up to investigate the friction laws of adiabatic single and two-phase flow through particulate beds which have the characteristics of the prototypical debris bed, such as packed with particles of multiple sizes or irregular shapes. The emphasis of the present study is placed on quantification of effective particle diameter of a particulate bed composed of multi-diameter spheres. Pressure drops are measured for water/air flow through the particulate beds packed with various combinations of spheres, and the effective particle diameters of the beds are obtained based on the pressure gradients and the Ergun equation. The results show that at low flowrate (Re < 7) the effective particle diameters can be represented by the area mean diameters of the particles in the beds, while at high velocity (Re > 7) the effective particle diameters are closer to the length mean diameters. If the area mean diameters are chosen as the effective particle diameters, the frictional pressure drops of two-phase flow in the beds can be predicted by the Reed model with good agreements.     

Experimental investigation of coolability behaviour of irregularly shaped particulate debris bed

In case of a severe nuclear reactor accident, the core can melt and form a particulate debris bed in the lower plenum of the reactor pressure vessel (RPV). Due to the decay heat, the particle bed, if not cooled properly, can cause failure of the RPV. In order to avoid further propagation of the accident, complete coolability of the debris bed is necessary. For that, understanding of various phenomena taking place during the quenching is important. In the frame of the reactor safety research, fundamental experiments on the coolability of debris beds are carried out at IKE with the test facility “DEBRIS”. In the present paper, the boiling and dry-out experimental results on a particle bed with irregularly shaped particles mixed with stainless steel balls have been reported. The pressure drops and dry-out heat fluxes of the irregular-particle bed are very similar to those for the single-sized 3 mm spheres bed, despite the fact that the irregular-particle bed is composed of particles with equivalent diameters ranging from 2 to 10 mm. Under top-flooding conditions, the pressure gradients are all smaller than the hydrostatic pressure gradient of water, indicating an important role of the counter-current interfacial drag force. For bottom-flooding with a liquid inflow velocity higher than about 2.7 mm/s, the pressure gradient generally increases consistently with the vapour velocity and the fluid-particle drag becomes important. The system pressures (1 and 3 bar) have negligible effects on qualitative behaviour of the pressure gradients. The coolability of debris beds is mainly limited by the counter-current flooding limit (CCFL) even under bottom-flooding conditions with low flow rates. The system pressure and the flow rate are found to have a distinct effect on the dry-out heat flux.Different classical models have been used to predict the pressure drop characteristics and the dry-out heat flux (DHF). Comparisons are made among the models and experimental results for DHF and pressure drop characteristics. Considering the overall trend in prediction of DHF and two-phase pressure drop, it was observed that none of the models could provide accurate predictions for both DHF and pressure drop under top- and bottom-flooding conditions. This implies that developments of more accurate models are needed including the effects of non-uniform particle sizes and the multidimensional nature of particulate debris beds, which are not reflected so far in these models.     

Void fraction and heat transport in two-dimensional mixed size particle beds with internal heat sources

The paper discusses the boiling heat transfer from a porous bed with internal heat sources and refers to the configuration in a nuclear reactor after a partial core melt. The flow of coolant, the temperature and the local liquid/vapor distribution were investigated in a two-dimensional configuration. Experiments were conducted using monodisperse beds as well as a mixture of two different particle sizes with a total porosity below 20%. In some tests the bed was supported by a shell of porous material to create a gap along the bottom of the test container. Water was used for tests up to 9% of the critical pressure, while other tests were made with R134a up to 44% of the critical pressure. The maximum heating rate realized inductively was 730 kW/m². The experiments have been compared to analytical results with a one-dimensional approach.It is shown that in contrary to the situation in small cylindrical configurations the heat transfer was increased by large buoyancy driven convective flows. If there was a gap along the container bottom an additional flow of liquid improved the coolability of the bottom region even if the upper part of the particle bed was already overheated. In case of high density ratios (water at low pressure), the measurements indicated a strong enhancement of the coolant flow above a certain minimum heating rate resulting in decreasing vapor fraction values which were nearly independent of the system pressure. This was assumed to be caused by the appearance of vertical channels through which the vapor could flow through the particle bed.     

Heat transport and void fraction in granulated debris

The paper discusses the boiling heat transfer from a porous bed with internal heat sources and refers to the configuration in a nuclear reactor after a partial core melt. The flow of coolant, the temperature and the local liquid/vapor distribution were investigated in a two-dimensional configuration. Experiments were conducted using monodisperse beds as well as a mixture of two different particle sizes with a total porosity below 20%. In some tests the bed was supported by a shell of porous material to create a gap along the bottom of the test container. Water was used for tests up to 9% of the critical pressure, while other tests were made with R134a up to 44% of the critical pressure. The maximum heating rate realized inductively was 730 kW/m². The experiments have been compared to analytical results with a one-dimensional approach.It is shown that in contrary to the situation in small cylindrical configurations the heat transfer was increased by large buoyancy driven convective flows. If there was a gap along the container bottom an additional flow of liquid improved the coolability of the bottom region even if the upper part of the particle bed was already overheated. In case of high density ratios (water at low pressure), the measurements indicated a strong enhancement of the coolant flow above a certain minimum heating rate resulting in decreasing vapor fraction values which were nearly independent of the system pressure. This was assumed to be caused by the appearance of vertical channels through which the vapor could flow through the particle bed.     

A new empirical model for self-leveling behavior of cylindrical particle beds

ABSTRACT

During the material relocation phase of core disruptive accidents in sodium-cooled fast reactors, the rapid quenching and fragmentation of molten materials discharged from the reactor core into the lower plenum region can lead to the formation of debris beds. Coolant boiling may lead to leveling of the mound-shaped beds, which changes both the beds' coolability with decay heat in the fuel and the neutronic characteristics. In this study, a series of experiments using simulant materials were performed to develop an experimental database of self-leveling processes of particle beds in a cylindrical system. To simulate the coolant boiling in the beds in the experiments, a gas injection method was used to percolate nitrogen gas uniformly through the base of a bed with a conical-shaped mound. Time variations in bed height during the self-leveling process were measured for different particle sizes, densities and sphericities, and gas injection velocities. Using a dimensional analysis approach, a new model was proposed. This model correlates the experimental data on transient bed height with an empirical equation using a characteristic time for self-leveling development and an equilibrium bed height. The proposed model reasonably predicts the self-leveling development of particle beds.     

Extended dryout and rewetting of small-particle core debris

Extended dryout of core debris, especially the location, size and temperature distribution of the dry zone and the variation of these parameters with time and power density (down to rewetting) was studied with volume-heated beds. The beds were composed of small, spherical stainless steel particles and water was used as coolant. In addition, transient rewetting was investigated starting with a dry bed of uniform temperature.     

Experimental Study on Coolability of Particulate Core-metal Debris Bed with Oxidization, (II) Fragmentation and Enhanced Heat Transfer in Zircaloy Debris Bed

The oxidization and coolability characteristics of the particulate Zircaloy debris bed, which is deposited under the hard debris and through which first vapor penetrates and then water penetrates, are studied in the present paper. In the vapor penetration experiments, it is found that Zircaloy debris particles are effectively broken into small pieces after making thick oxidized layer with deep clacks by rapid oxidization under the condition that vapor with 20 cm/s penetrates for 30 to 70 min at an initial debris bed temperature of 1,030°C. It is also confirmed in the water penetration experiments that the oxidized particle debris bed has potentiality of high coolability when water penetrates through the fully oxidized particle bed because of a high capillary force originating from those particles with deep cracks on their surfaces.

Based on the present study, a new scenario for the appearance and disappearance of the hot spot in the TMI-2 accident is posssible. The particulate core-metal debris bed is first heated up by rapid oxidization with heat generation when vapor can penetrate through the debris bed with porosities. This corresponds to the appearance of the hot spot. The resultant oxidized particulate debris bed causes a high coolability due to its high capillary force when the water can touch the debris bed at wet condition. This corresponds to the disappearance of the hot spot.     

径向分层碎片床内流动特性研究

为研究非均质结构碎片床内的流动特性,采用两种尺寸颗粒构建了具有径向分层结构的颗粒堆积碎片床,为了对比分析,同时构建了均质结构颗粒堆积碎片床。实验研究了流体在不同堆积结构床内的流动阻力特性,并通过数值模拟揭示了流体在分层床分层界面处的流量再分配现象。研究结果表明,当流体自下而上通过碎片床时,对于均质结构颗粒堆积床,流体呈现一维流动特性;对于具有不同渗透率的径向分层床,除大部分流体自下而上通过分层床外,还存在部分流体从低渗透率层流向高渗透率层,呈现二维流动特性,且绝大部分横流仅发生在分层床的初始部分。     

Basic investigations on debris cooling

An experimental investigation of boiling phenomena in inductively heated particle beds has been performed. The major aim of these experiments is to provide data for validating numerical codes used in reactor safety. The experiments can be divided in three parts: boiling experiments, dryout experiments and quenching experiments. In boiling experiments, the pressure gradients have been measured along the bed height for different flow modes, different heat inputs and different system pressures. In dryout experiments, the minimum heat input has been determined for which the particle bed starts to superheat significantly above the saturation temperature. The final test series deals with the cool down behaviour of strongly superheated particles by flooding them with cold water. The initial temperatures ranged from 200 up to 900 °C in top-quenching experiments and from 230 up to 450 °C in bottom-quenching experiments. All experiments were performed with pre-oxidised stainless steel balls of 6 and 3 mm diameter in a cylindrical crucible. The bed height was 640 mm and the bed diameter was 125 mm for boiling and dryout experiments, respectively 150 mm for quenching experiments. The experimental results are compared with various available dryout models.     

Numerical study on sedimentation behavior of solid particles used as simulant fuel debris

This paper presents numerical simulations using the discrete element method (DEM) to model sedimentation behavior of solid debris particles, which is significant for estimates of the coolability of debris beds. A series of experiments of gravity driven discharge of solid particles into a quiescent water pool was used to validate the DEM simulation method. We evaluated the effects of three crucial factors: particle density, particle diameter, and nozzle diameter on three key quantitative parameters: particle dispersion angle, particle fall time in the pool, and the height of the deposited particle bed to express the particle sedimentation behavior. The three crucial factors play a significant role in the particle sedimentation behavior. We compared the experimental and simulated results of the particle dispersion angle and particle fall time in the pool, and the height and shape of the deposited particle bed. The general trend of the simulation results indicates a reasonable agreement with the experimental observations. The simulations exhibit the potential applicability of the DEM-based simulation technique for the prediction of particle sedimentation behavior.     

Dryout heat flux experiments with deep heterogeneous particle bed

A test facility has been constructed at Technical Research Centre of Finland (VTT) to simulate as accurately as possible the ex-vessel core particle bed in the conditions of Olkiluoto nuclear power plant. The STYX particle bed reproduces the anticipated depth of the bed and the size range of particles having irregular shape. The bed is immersed in water, creating top flooding conditions, and internally heated by an array of electrical resistance heating elements. Dryout tests have been successfully conducted at 0.1–0.7 MPa pressure for both uniformly mixed and stratified bed geometries. In all tests, including the stratified ones, the dry zone first formed near the bottom of the bed. The measured dryout heat fluxes increased with increasing pressure, from 232 kW/m² at near atmospheric pressure to 451 kW/m² at 0.7 MPa pressure. The data show some scatter even for the uniform bed. The tests with the stratified bed indicate a clear reduction of critical power due to the presence of a layer of small particles on top of the uniform bed. Comparison of data with various critical power (dryout heat flux) correlations for porous media shows that the most important parameter in the models is the effective particle diameter. Adiabatic debris bed flow resistance measurements were conducted to determine the most representative particle diameter. This diameter is close, but not equal, to the particle number-weighted average diameter of the bed material. With it, uniform bed data can be calculated to within an accuracy of 3–28% using Lipinski's 0-D model. In the stratified bed experiments, it appears that the top layer was partially fluidized, hence the measured critical power was significantly higher than calculated. Future experiments are being planned with denser top layer material to eliminate non-prototypic fluidization.