Professional Documents
Culture Documents
Peybernes CFD For Corium Pool Modeling
Peybernes CFD For Corium Pool Modeling
MODELING
CEA/DEN/DTN/SMTA/LPMA
L. SAAS , M. PEYBERNES, R. LE TELLIER, C. DANG
Context and Problem: Severe Accidents (SA), the Focusing Effect risk and the CFD
for SA modelling
| PAGE 2
Context: Severe Accidents, In-vessel corium behavior
Core meltdown
Corium pool formation In-vessel
Propagation in core corium
Corium relocation in lower head behavior
Vessel failure
Relocation in reactor pit
Risks of rapid containment failure:
Hydrogen explosion
Steam explosion
In-vessel
corium Direct Containment Heating
behavior Bypass…
Risk of slow basemat melt through
Molten Core Concrete Interaction
We are wondering if IVR is a reliable solution for “medium and high” power | PAGE 3
reactors (≥1000 MWe) ?
Corium relocation in lower head
During severe accident, corium relocates from the core into the vessel lower head:
The corium pool is stratified because of a miscibility gap in the U-Zr-O-Fe thermochemical system: different
layers: Heavy metal layer / Oxidic layer / Light Metal Layer
The steel vessel wall is melted due to the heat fluxes of the corium pool
The molten steel relocates at the top of the corium pool in order to form a thin metallic layer
This thin metallic layer is heated from below and cooled on the upper and lateral surface
Its thickness evolves with ablated molten steel (thickening) and thermochemical effects (thinning)
The lateral heat flux of the thin metallic layer depends on the thickness of the layer: for small thickness
focusing effect could occurs and consequently involves the vessel rupture
𝜑𝑢𝑝
𝜑𝑙𝑎𝑡 In the severe accident code the
lateral heat flux of the thin metallic
layer is evaluated with a stationary
0D model (MAAP (EPRI), MELCOR
𝜑𝑑𝑜𝑤𝑛 (NRC), ASTEC(IRSN),
PROCOR(CEA)…)
Heat flux
concentration
factor :
𝜑𝑙𝑎𝑡 /𝜑𝑑𝑜𝑤𝑛
Corium pool
(oxidic layer)
BALI-métal
experiments (water)
Geometrical ratio
From Bonnet J.M. 1999 Γ = ℎ/𝐿
Comparison of this 0D model to experimental data with water (and not steel): some margin may exist for
reactor case ? Difficult to have experiments with steel material (T>2200K), but simulation with CFD is
achievable.
Objective of our CFD work → study the heat flux concentration factor 𝝋𝒍𝒂𝒕 /𝝋𝒅𝒐𝒘𝒏 with the TrioCFD code
in order to build a enhanced 0D model that takes into account some margins that may exist
Work to achieve with the CFD and the simplified modelling:
Validation of TrioCFD code in the case of thin layers: natural convection with lateral and top cooling
Parametric studies of the thin metallic layer in the reactor context in order to understand thermalhydraulic
phenomena and to generate data
Development of a new 0D model for Severe Accident using the CFD data and understanding
| PAGE 5
Validation of TrioCFD code for thin layer with water
𝛻. 𝑉 = 0
𝜕𝑉
𝜌 + 𝛻 . (𝑉 ⊗ 𝜌𝑉 − 𝜈𝛻𝑉) = 𝛻𝑃 − 𝜌𝑔𝛽(𝑇 − 𝑇0 )
𝜕𝑡
𝜕𝑇
𝜌𝐶𝑝 + 𝛻. 𝑉𝜌𝐶𝑝 𝑇 − 𝜆𝛻𝑇 = 0
𝜕𝑡
VDF spatial discretization and explicit time scheme
No turbulence model
19 JUIN 2017
| PAGE 6
Rayleigh-Bénard convection
• The hotter and lighter fluid at the bottom is flowing to the top where it
is cooled and become heavier and then moves down
𝑔𝛽Δ𝑇ℎ3
• Characterized by 𝑅𝑎 =
𝛼𝜈 | PAGE 7
Rayleigh-Bénard convection
Tests and set up of parameters, models and numerical schemes for TrioCFD:
• Validation with Ouertatani et al. paper (2009)
Shematic diagram of the test case (Ra=10−6 ) Streamline contours Isotherm contours
Ouertatani et al. Ouertatani et al.
Tests and set up of parameters, models and numerical schemes for TrioCFD:
𝛻. 𝑉 = 0
𝜕𝑉
𝜌
+ 𝛻 . (𝑉 ⊗ 𝜌𝑉 − 𝜈𝛻𝑉) = 𝛻𝑃 − 𝜌𝑔𝛽(𝑇 − 𝑇0 )
𝜕𝑡
𝜕𝑇
𝜌𝐶𝑝 + 𝛻. 𝑉𝜌𝐶𝑝 𝑇 − 𝜆𝛻𝑇 = 0
𝜕𝑡
Constant physical properties of water: dynamic viscosity 𝜈 = 1.1002. 10−3 𝑚2 /𝑠 , mass
density 𝜌 = 1001.8 𝑘𝑔/𝑚3 , thermal conductivity 𝜆 = 0.604 𝑊. 𝐾 −1 𝑚−1, specific heat
𝐶𝑝 = 4185.5 𝐽. 𝑘𝑔−1 𝐾 −1 4185.5, thermal expansion coefficient 𝛽 = 1.5. 10−4 𝐾 −1
VDF spatial discretization and explicit time scheme
No turbulence model
No slip boundary conditions: u = 0
Imposed temperature
temperature
Adiabatic
Imposed
Convergence for the Power balance vs time Temperature profile in the bottom of the domain (heating system)
Meshes
Cold strip
19Simulation
JUIN 2017 1,1x108 30 6,2x108 330 CEA | 10 AVRIL 2012 | PAGE 12
1rst Parametric Study with steel : fit to 0D model
Such that
Use of TrioCFD to
calibrate 𝑨, 𝑪 according
to 𝝋𝒅𝒐𝒘𝒏 , 𝑻𝒖𝒑 , 𝑳, 𝒆
19 JUIN 2017
CEA | 10 AVRIL 2012 | PAGE 13
19 JUIN 2017
1rst Parametric Study with steel : fit to 0D model
Parameters A and C are calibrated according to the global volume flow rate
JUIN 17
We search A and C such that | PAGE 14
1rst Parametric Study with steel : fit to 0D model
Parameters A and C are calibrated according to the global volume flow rate
Global volume flow rate contribution of the lateral Global volume flow rate contribution of the lateral boundary layer
boundary layer (coefficient C) according to the thickness (coefficient C) according to the upper temperature
JUIN 17
We search A and C such that | PAGE 15
1rst Parametric Study with steel : fit to 0D model
with and
We assume the following forms of A and C taking into account the tendency of the previous results
Calibration for
Calibration for
JUIN 17
The calibration is possible, but we have to complete the CFD work | PAGE 16
Rayleigh-Bénard convection and Bénard-Marangoni effects
𝜕𝑣𝑥 𝛾 𝜕𝑇
Mechanism : =
𝜕𝑧 𝜇 𝜕𝑥
When layer has a free surface and there exists a temperature gradient on it (if surface tension is varying
with temperature) Bénard-Marangoni effects take place :
𝜌𝜈𝛼 𝑔h3
The top surface could be deformed: if Crispation number C𝑟 = ≪ 1 and Galileo number 𝐺𝑎 = ≫1
𝜎ℎ 𝜈2
no deformation occurs
𝛾h∆𝑇
Marangoni number 𝑀𝑎 = (when 𝑀𝑎<0 diffusion is observed if no natural convection take place)
𝜌𝜈𝛼
| PAGE 18
Numerical evaluation : 0D model
Bénard-Marangoni effects dominate the flow for thickness less than 3cm :
Bénard Rayleigh
Marangoni Bénard
2nd Parametric Study with steel : boundary conditions
Lateral
thermal Less convective cells
gradient is
smoother in a
longer length
Only radiative heat transfer on top Radiative heat transfer with Marangoni
boundary conditions
Total Heat looses seem to be more important with Bénard-Marangoni boundary conditions
The flow structure and the impact on the heat flux concentration have to be studied : some difference
could be observed
Conclusion and perspectives
Thanks to TRUST, TrioCFD and CCRT teams for help and support.
Calculations on CRTT (DARI project) in 2016 (1M CPU hours) and in 2017 (1,2M CPU hours, in progress), in
the frame of the IVMR (In Vessel Melt Retention) european project.
The validation of TrioCFD with BALI experiment (water) has been done. The split power is in agreement
with experiment. Variable physical properties are required in order to obtain accurate Ra and Nu numbers.
We have proposed a first study for the thin metallic layer with fixed temperature at the top. A 0D model has
been calibrated with TrioCFD calculations.
A second study on the top boundary conditions for the thin metallic layer is in progress: the case of free
surface with radiative heat transfer
In this case Bénard-Marangoni effects dominates the flow for thickness less than 3cm (linear stability
analysis + CFD) :
If the lateral cooling has no impact on the Bénard-Marangoni effects, it seems that the concentration heat
flux factor decrease for positive Marangoni number
Some measurement of the surface tension coefficient are actually done on the VITI (CEA) facility in the
frame of the IVMR project
We are wondering if IVR is a reliable solution for “medium and high” power reactors
(≥1000 MWe)? The different questions that we have to address are:
2) Are the in-vessel heat fluxes lower than this determined CHF ?
During severe accident, corium relocates from the core into the vessel lower head
The steel wall of the vessel are melting due to the heat fluxes of the corium pool
The resulting molten steel could relocate on top of the corium pool in order to form a thin metallic layer
This layer is heated from below and cooled on top and laterally and may lead to the focusing effect and
consequently to the vessel rupture
Different scenario of the corium relocation lead to different heat transfer and top boundary conditions for the
liquid steel layer on top of the corium pool :
Liquid top free surface with radiative heat exchange
Debris bed with conductive and radiative heat exchange
Solid steel layer with conductive heat exchange (reflooding or efficient radiative heat exchange)
Liquid top free surface Debris bed Solid rigide surface