The 10th International Topical Meeting on Nuclear Reactor Thermal Hydraulics (NURETH-10) Seoul, Korea, October 5-9, 2003.
CFD ANALYSES OF STEAM AND HYDROGEN DISTRIBUTION IN A NUCLEAR POWER PLANT M. Houkema, E.M.J. Komen, N.B. Siccama, S.M. Willemsen NRG Petten Westerduinweg 3 P.O. Box 25 1755 ZG Petten The Netherlands
[email protected];
[email protected];
[email protected];
[email protected]
KEY WORDS Hydrogen distribution, steam, nuclear safety, Computational Fluid Dynamics
ABSTRACT A detailed three-dimensional Computational Fluid Dynamics (CFD) model of the containment of the nuclear power plant has been prepared in order to assess possible multidimensional phenomena. In a first code-to-code comparison step, the CFD model has been used to compute a reference accident scenario that has been analysed earlier with the lumped parameter code SPECTRA. The CFD results compare qualitatively well with the SPECTRA results. Subsequently, the actual steam jet from the primary system has been modelled in the CFD code in order to determine the hydrogen distribution for this realistically modelled source term. Based on the computed hydrogen distributions, it has been determined when use of lumped parameter codes is allowed and when use of CFD codes is required.
1. INTRODUCTION Hydrogen in the containment due to hypothetical severe accidents still poses a challenge when quantifying scenarios with probabilistic safety assessments (PSAs) and in this respect is subject to reviews with periodic safety evaluations. The containment considered in the present study is equipped with 22 Passive Autocatalytic Recombiners (PARs), which, according to present insight, present a considerable hydrogen removal capacity and prevent possible hydrogen burns. Lumped parameter codes have been used to determine optimum PAR positions and hydrogen removal efficiency. In order to assess possible multidimensional effects, a detailed three-dimensional Computational Fluid Dynamics model of the containment of the nuclear power plant has been prepared also. This paper presents the results obtained with the CFD model up to now. In a first code-to-code comparison step, the CFD model has been used to compute a reference accident scenario which has been analysed earlier with the lumped parameter codes SPECTRA (Stempniewicz, 1999) and (Bakker, 1999) and WAVCO (Graf, 1998). Modelling simplifications have been made in the CFD model in order to facilitate a comparison between the 3-D CFD results and the lumped parameter code results. 1
In a second step, actual leak size (m2) and leak direction have been simulated realistically in the CFD model in order to determine: 1. The hydrogen distribution within the compartments of the containment for a realistically modelled source term from the primary system; 2. The importance of the application of CFD to determine this hydrogen distribution. The results of the second step and the conclusions drawn from these results will also be discussed in this paper.
2. REFERENCE ACCIDENT SCENARIO The containment initial process conditions are summarized in Table 1. In the CFD analyses, basically the same data have been used for the mass flow rates and corresponding temperatures of water, steam, and hydrogen discharged from the primary system into the containment as used in the SPECTRA and WAVCO analyses. The total steam and hydrogen masses discharged from the primary system are graphically presented in respectively Figs. 4 and 7, as obtained from (Graf, 1998). Table 1. Initial containment process conditions. parameter pressure temperature gas space temperature concrete walls
value unit 101300 Pa 293 K 293 K
temperature components temperature steel liner hydrogen volume fraction
293 K 293 K 0.0 -
steam volume fraction air volume fraction
0.0 1.0 -
3. CFD MODEL 3.1 Computer Code The current analyses have been performed using the commercial CFD code CFX-4.4 (CFX 4.4 USER GUIDE, 2001). 3.2 Geometrical Model and Mesh An inside view of the applied geometrical model of the containment is presented in Fig. 1. Based on this geometrical model, a body-fitted mesh using 680000 hexahedral computational cells has been generated. This mesh represents the actual geometry of the containment, except that: 1. Major objects like the steam generator and main coolant pumps have been slightly simplified; 2. The coolant pipes and crane have been omitted. The presence of the coolant pipes and crane can be neglected for the current purposes, because the volume of these components can be neglected, and the condensation on these components has a minor contribution to the total condensation. In the current analyses, the explosion hatches between the installation area and the operational area are assumed to be instantaneously opened following accident initiation. 2
Figure 1. Inside view of the geometrical model of the containment as applied in the CFD analyses. 3.3 Main Fluid Dynamics Model The options in the applied CFX-4.4 main fluid dynamics model have been selected such that the simulated flow corresponds to a transient turbulent compressible non-isothermal flow of an air-steam-hydrogen gas mixture. 3.4 Wall Condensation Model The leak in the primary system is located in compartment 3, which is located below compartment 6 (see Fig. 2). The mass flow rates of the steam and hydrogen Part of the steam will condense at the walls of the containment. Wall condensation has been modelled using the wall condensation model that has been validated in (Lycklama à Nijeholt, 1998) and (Siccama, 2001). For the modelling of turbulence, the standard k-ε model (see e.g. (Wilcox, 1994)) has been applied. Gravity has been included in the model (g = 9.81 m/s2). 3.5 Recombiners As a mitigating measure, 22 PARs are positioned inside the containment. Three types of recombiners are being used, each type having a different capacity. The recombiners start to recombine when the hydrogen concentration cH2 exceeds the initial value of 2 vol%. The recombination rate (m & recomb ) is linear between this initial hydrogen concentration and the saturation concentration of the recombiners ( c Hsat2 = 0.08). Above this concentration, the recombination rate is constant. The recombination rate has a pressure independent part (k0) and a pressure dependent part (k1) and can be calculated as follows (Bakker, 1999) and (Graf, 1998):
(
)
m& recomb = (k 0 + k 1 p ) ⋅ min c Hsat2 , c H 2 ⋅ 100 ⋅ n w [g / s ] In this equation, p presents the absolute static pressure in bars, k0 = 0.0123g/s, k1 = 0.0103g/(s.bar), and nw is the type number (1,3, or 13).
3
(1)
At the positions of the recombiners, the conversion of hydrogen into steam is simulated using user coding for the volumetric mass sinks and sources in the hydrogen and steam mass transport equations. The recombination rate as detailed in the previous paragraph has been implemented. The corresponding heat release has also been included in the model, since the conversion of hydrogen into steam is an exothermal process. 3.6 Physical Properties The modelled multi-component gas mixture consists of air, hydrogen, and steam. The heat capacity of the multi-component gas mixture is computed by the CFX-4.4 CFD code based on the corresponding standard values of the individual species. The properties of the concrete walls have been set equal to the corresponding properties as used in the SPECTRA computations reported in (Bakker, 1999). These properties are summarised in Table 2. 3.7 Boundary Conditions The leak in the primary system is located in compartment 3, which is located below compartment 6 (see Fig. 2). The mass flow rates of the steam and hydrogen discharged from the leak into the containment corresponding to the total injected steam and hydrogen mass as presented in Figs. 4 and 7 have been used as inlet boundary conditions. As a modelling simplification, all water discharged from the leak into the containment is assumed to be directly present as a water pool at the bottom of the containment. The consequences of this modelling simplification will be discussed in the next section. In the first code-to-code comparison step, the modelling of the steam injection has been simplified also in order to facilitate comparison with the SPECTRA (Bakker, 1999) and WAVCO (Graf, 1998) lumped parameter code results. Namely, the actual steam jet has been replaced by a homogeneous injection of steam via the entire floor of the compartment containing the leak. A constant temperature of 300 K has been applied at the outer wall of the containment. The initial containment process conditions as summarized in Table 1 have been used. Table 2. Properties of the concrete walls. Parameter heat capacity density thermal conductivity
value units 960 J/(kg.K) 2400 kg/m3 2.04 W/(m.K)
4. RESULTS 4.1 First code-to-code comparison step (CFX-4.4 homogeneous steam injection) The comparison made in (Bakker, 1999) between the WAVCO results and the SPECTRA Perfect Mixing model results revealed that both lumped parameter codes gave about the same results. For this reason, the currently obtained CFD results will be compared only with the SPECTRA Perfect Mixing model results. In the discussion below, the computed steam and hydrogen concentration will be presented for the steam generator compartment 6 and the dome 14. The numbering of these compartments is shown in Fig. 2.
4
Figure 2. Numbering of the compartments within the containment. 4.1.1 Containment Pressure The release of water, steam, and hydrogen from the primary system into the containment results in an increase in the containment pressure and temperature. The containment pressure as computed by the SPECTRA lumped parameter code and the CFX-4.4 homogeneous injection model are presented in Fig. 3. As can be concluded from this figure, the same trend can be observed in both computed pressures. However, there is a quantitative discrepancy between the considered CFX-4.4 and SPECTRA results. The major reason for this discrepancy is explained in the next paragraph. 3.0 2.8
SPECTRA Pressure CFX homogeneous injection
2.6
Pressure (bar)
2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0 0.0
0.5
1.0
1.5
2.0 Time (h)
2.5
3.0
3.5
4.0
Figure 3. Average pressure in the containment. 4.1.2 Steam Concentration In the current CFD model, all hot water discharged from the primary system into the containment is assumed to be directly present as a pool of water at the bottom of the containment. However, part of this discharged hot water will evaporate in reality. As can be observed in Fig. 4, this evaporation is modelled in SPECTRA. Due to this evaporation, the total amount of steam in the containment as computed using SPECTRA is larger than the total amount of steam mass discharged from the primary 5
system during about the first 0.3 hr. of the transient. As already mentioned, the evaporation of the hot water discharged from the primary system is not modelled yet in the current CFD model. As a result, the amount of steam in the containment is under-predicted in the CFD simulations. This aspect is the major reason for the quantitative discrepancy between the computed SPECTRA and CFX-4.4 pressures. From Figs. 5 and 6, it can be concluded that the average CFX-4.4 steam volume fraction compares qualitatively well with the SPECTRA steam volume fraction. The major reason for the quantitative discrepancy between the computed CFX-4.4 and SPECTRA steam volume fractions is the absence of an evaporation model in the current CFD model.
Steam mass (kg)
80000 70000
SPECTRA steam mass
60000
CFX homogeneous injection
50000
Total injected steam
40000 30000 20000 10000 0 0.0
0.5
1.0
1.5
2.0 Time (h)
2.5
3.0
3.5
4.0
Figure 4. Steam mass present in the containment. 1.0
SPECTRA CV 6
0.8 Steam volume fraction (-)
CFX homogeneous injection 0.6
0.4
0.2
0.0 0.0
0.5
1.0
1.5
2.0 Time (h)
2.5
3.0
3.5
4.0
Figure 5. Average steam volume fraction in room 6 of the containment. 4.1.3 Hydrogen Concentration The total amount of hydrogen released from the primary system into the containment is shown in Fig. 7. Figure 7 also presents the amounts of hydrogen present in the containment as computed by SPECTRA and the CFX-4.4 homogenous injection model. The differences between these two computed amounts and the total amount of hydrogen correspond to the computed amounts of hydrogen being recombined. As can be concluded from Fig. 7, more hydrogen is being recombined in the SPECTRA model. This can be explained as follows. In the CFD model, the hydrogen recombination by the PARs results in a local reduction of the hydrogen concentration. Subsequently, 6
some time is required for transport of hydrogen towards the PARs where the reduced local hydrogen concentrations occur. In contrast, in the lumped parameter approach, the hydrogen recombination results in a reduction of the average hydrogen concentration within the compartment, and no time is required within this compartment for transport of hydrogen towards the PARs. 0.6 SPECTRA CV 14 CFX homogeneous injection
Steam volume fraction (-)
0.5
0.4
0.3
0.2
0.1
0.0 0.0
0.5
1.0
1.5
2.0 Time (h)
2.5
3.0
3.5
4.0
Figure 6. Average steam volume fraction in room 14 of the containment. 350 300
Hydrogen mass (kg)
250 200 150
SPECTRA hydrogen mass
100
CFX homogeneous injection
50
Total injected hydrogen
0 1.0
1.5
2.0
2.5 Time (h)
3.0
3.5
4.0
Figure 7. Hydrogen mass present in the containment. From Figs. 8 and 9, it can be concluded that the average CFX-4.4 hydrogen volume fraction compares qualitatively well with the SPECTRA hydrogen volume fraction. The major reason for the quantitative difference between the computed CFX-4.4 and SPECTRA hydrogen volume fractions is the larger hydrogen recombination rate in the SPECTRA model. Based on the presented analysis results, it can be concluded that the CFX-4.4 results compare qualitatively well with the corresponding SPECTRA results. However, a quantitative discrepancy could be observed between the considered CFX-4.4 and SPECTRA results. The major reason for the observed quantitative discrepancies is the absence of an evaporation model in the current CFD model. It is expected that the CFX-4.4 and SPECTRA results will be in good agreement following implementation of an evaporation model in CFX-4.4.
7
0.10
SPECTRA CV 6
Hydrogen volume fraction (-)
0.08
CFX homogeneous injection
0.06
0.04
0.02
0.00 1.0
1.5
2.0
2.5 Time (h)
3.0
3.5
4.0
Figure 8. Average hydrogen volume fraction in room 6 of the containment. 0.08
Hydrogen volume fraction (-)
0.06
0.04
0.02
SPECTRA CV 14 CFX homogeneous injection
0.00 1.0
1.5
2.0
2.5 Time (h)
3.0
3.5
4.0
Figure 9. Average hydrogen volume fraction in room 14 of the containment. 4.2 Second Step: Realistic Modelled Break In the current second analysis step, the actual steam jet from the primary system is realistically modelled. As mentioned in the introduction, the objective of this second analysis step is to determine the hydrogen distribution within the compartments of the containment for this realistically modelled source term from the primary system. In addition, the importance of the application of CFD to compute this hydrogen distribution will be determined. Figures 10 and 11 present the average and maximum hydrogen volume fractions in compartments 6 and 14 for a realistically modelled break in the primary system. From the differences between the presented maximum and average hydrogen volume fractions, it can be concluded that large inhomogeneities take place in the hydrogen distribution in compartment 6 which is close to the break compartment. Further from the break, the hydrogen distribution within the compartments becomes more and more homogeneous. This is illustrated by the differences between the maximum and average hydrogen volume fractions in compartment 14, which are much smaller.
8
Hydrogen volume fraction for realistic injection (-)
0.50 0.45 Average concentration
0.40
Maximum concentration
0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 1.0
1.5
2.0
2.5 Time (h)
3.0
3.5
4.0
Figure 10. Realistically modelled break: average and maximum hydrogen volume fraction in room 6 of the containment. Figure 12 presents the hydrogen volume fraction at t=1.885 h into the transient in a 2-D vertical cross-section through the containment. As can be observed in this figure, large inhomogeneities take place in the hydrogen distribution inside the compartments of the installation. The above presented results demonstrate that it is not possible to determine reliably the existence of flammable gas mixtures based on compartment-based averaged hydrogen volume fractions. Especially during the initial phase of the transient, large deviations from the average hydrogen concentrations occur due to partial mixing within the compartments (Figs. 10 and 12). For the current scenario, these large deviations occur during about 0.5 h. Following this initial phase of incomplete mixing, mixing within the compartments becomes gradually complete. As a result, the deviations from the average values become negligible (Fig. 10). Based on these observations, it is concluded that 3-D CFD is required to determine the existence of flammable gas mixtures during the initial phase of the hydrogen release, since partial mixing within the compartments of the containment results in large inhomogeneities in the hydrogen distributions during this initial phase. Hydrogen volume fraction for realistic injection (-)
0.08
0.06
0.04
0.02 Average concentration Maximum concentration 0.00 1.0
1.5
2.0
2.5 Time (h)
3.0
3.5
4.0
Figure 11. Realistically modelled break: average and maximum hydrogen volume fraction in room 14 of the containment.
9
Figure 12. Realistically modelled break: hydrogen volume fraction at t = 1.885 h into the transient.
3. CONCLUSIONS In a first code-to-code comparison step, the actual steam jet from the primary system was modelled in CFX-4.4 as a homogeneous injection of steam in order to facilitate comparison with the SPECTRA lumped parameter code results. Based on the presented analysis results, it could be concluded that the presented CFX-4.4 results compare qualitatively very well with the SPECTRA results. However, a quantitative discrepancy was observed between the CFX-4.4 and SPECTRA results. The major reason for this discrepancy is the absence of an evaporation model in the current CFD model. It is expected that the CFX-4.4 and SPECTRA results will be in good agreement following implementation of an evaporation model in CFX-4.4. Subsequently, the actual steam jet from the primary system was realistically modelled in the CFX-4.4 CFD code in order to determine the hydrogen distribution within the compartments of the containment for this realistically modelled source term. Based on the computed hydrogen distributions, it was concluded that 3-D CFD is required to determine the existence of flammable gas mixtures during the initial phase of the hydrogen release (0.5-1.0 hr), since partial mixing within the compartments of the containment results in large inhomogeneities in the hydrogen distributions during this initial phase.
REFERENCES 1. P.J.T. Bakker, 1999, Analyses with the NPP SPECTRA model: The effect of stratification on the hydrogen distribution, NRG report 26084/99.52074, Arnhem, The Netherlands. 2. CFX 4.4 USER GUIDE, 2001, Computational Fluid Dynamics Services, AEA Technology, Harwell laboratory, Oxfordshire OX11 0RA, United Kingdom. 3. Dr. Graf, 12 May 1998, Wasserstoffverteilung im Reaktorgebäude nach Primärkreisleckagen und versagender Notkühlung, Siemens report KWU NDS2/98/2001a.
10
4. J.A. Lycklama à Nijeholt, and C.J.J. Beemsterboer, December 1998, CFD Simulation of the Phebus FPT1 Containment Thermal-Hydraulics, ECN-CX—98-123, Petten, The Netherlands. 5. N.B. Siccama, August 2001, CFD Simulation of the Flow Phenomena in the Phebus Containment during the FPT0 Release Phase, NRG report 20397/01.41354/C, SAMPHEBEN2-T05, Petten, The Netherlands. 6. M.M. Stempniewicz, October 1999, SPECTRA – Sophisticated Plant Evaluation Code for Thermal-hydraulic Response Assessment, Version 1.00, December 1999; Volume 1 – Description of Models; Volume 2 – User’s Guide; Volume 3 – Description of Subroutines; Volume 4 – Verification, NRG/PPT report 26094/99.52612/C, Arnhem, The Netherlands. 7. D.C. Wilcox, 1994, Turbulence Modelling for CFD, DCW Industries, California.
11
