Title: Comparison and Analysis of the Condensation Benchmark Results
1Comparison and Analysis of the Condensation
Benchmark Results
W. Ambrosini, M. Bucci, N. Forgione, F. Oriolo,
S. Paci, J-P. Magnaud, E. Studer, E. Reinecke,
St. Kelm, W. Jahn, J. Travis, H. Wilkening, M.
Heitsch, I. Kljenak, M. Babic, M. Houkema, D.C.
Visser, L. Vyskocil, P. Kostka, R.
Huhtanen UNIPI (I), JSI (SLO), CEA Saclay (F),
NRG (NL) , FzJ (D), NRI Rez (CZ) , FzK IKET (D),
VEIKI (HU), JRC Petten (NL), VTT (FIN)
3rd European Review Meeting on Severe Accident
Research Nesseber, Hotel Vigo, Bulgaria 23 - 25
September, 2008
2Content
- Introduction
- Step 0 Condensation on an Isothermal Flat Plate
- Step 1 Condensation in the CONAN Facility
- Conclusions and Future Work
3Introduction
- Computational Fluid Dynamics codes are very
promising for assessing the risk associated with
the presence of hydrogen in power plant
containments after a severe accident - detailed description of flow patterns and gas
distributions - capability of a mechanistic approach to simulate
basic phenomena - Wall condensation is one of these relevant
phenomena also influencing the levels of
pressurization and atmosphere mixing - engineering models based on the heat and mass
transfer analogy - CFD models
- completely mechanistic models based on vapour
diffusion and low-Re models - models based on wall functions and/or on
engineering heat transfer coefficient evaluation
4Introduction (contd)
- The application of low-Re models needs a great
detail in describing the near-wall region - ? it is often impractical in describing
real-life equipment - Wall functions or engineering models for HTC are
based on assumptions limiting their applicability - ? an effort should be made to improve their
reliability - ? part of the useful information on flow fields
from CFD is not used - There is therefore strong motivation to go on in
developing wall condensation models fro CFD - In the frame of SARnet it was decided to analyse
the state-of-the-art of models available to
Partners and to propose a Benchmarking activity
lead by the University of Pisa
5Introduction (contd)
- The activities on condensation in which the
University of Pisa served as a coordinator where
three - Reporting on the models adopted within SARnet for
condensation - Proposing and analysing a 0th benchmark exercise
- code-to-code comparison in an idealised condition
- comparison with a heat and mass transfer
correlation - Proposing a 1st benchmark exercise in comparison
with experimental data from the CONAN facility - first data and spreadsheets already distributed
- more data to come in future steps
6Step 0 Condensation on an Isothermal Flat
Plate Problem Description
- This initial step dealt with an idealised version
of the problem to be subsequently addressed on
the basis of experimental data - The objective of the 0th Step was to compare code
results with correlations considered applicable
to the addressed problem - Reference was made to the 2D computational domain
sketched in the Figure
7Step 0 Condensation on an Isothermal Flat
Plate Heat Transfer and Heat and Mass Transfer
Problems
- The domain had to be used in different ways
- in pure convective heat transfer calculations (no
steam condensation), to assess the adequacy of
the adopted turbulence models and numerical grids
in front of the correlation - in simultaneous heat and mass transfer
calculations, whose results should be compared
among each other and with the correlation
8Step 0 Condensation on an Isothermal Flat
Plate Different formulations for the Sherwood
number
- In heat and mass transfer cases, Participants
were asked to calculate the Sherwood number at
least according the two relationships - As expected on the basis of the theories leading
to these formulations, they provided very similar
results - Classical definitions were adopted for the
Nusselt number
9Step 0 Condensation on an Isothermal Flat
Plate Boundary conditions
- The condensing plate is assumed to be kept at
uniform temperature and saturated vapour
concentration - Atmospheric pressure conditions are addressed (as
in the CONAN facility) for a saturated mixture - Two values of mixture velocity are considered
10Step 0 Condensation on an Isothermal Flat
Plate Participants and models
11Step 0 Condensation on an Isothermal Flat
Plate Results for Heat Transfer Cases HT-30-3
Reasonable agreement with the correlation
12Step 0 Condensation on an Isothermal Flat
Plate Results for Heat Transfer Cases HT-30-6
Reasonable agreement with the correlation
13Step 0 Condensation on an Isothermal Flat
Plate Results for Heat and Mass Transfer Cases
HTM-30-3
Greater spread around the correlation
14Step 0 Condensation on an Isothermal Flat
Plate Results for Heat and Mass Transfer Cases
HTM-30-6
Greater spread around the correlation
15Step 1 Condensation in the CONAN
Facility Experimental Apparatus (1)
- The primary loop circulates an air-steam mixture,
with a variable speed blower - The secondary loop extracts water from a mixing
vessels and circulates it into a 5 mm deep, flat
rectangular channel behind the cooled plate - The tertiary loop injects cold water from a large
reservoir into the mixing vessel, returning an
equal flow of warm water
16Step 1 Condensation in the CONAN
Facility Experimental Apparatus (2)
17Step 1 Condensation in the CONAN
Facility Addressed domain
- In the purposes of the exercise, the apparatus
can be modeled as in the figure - A 1D domain representing the center plane of the
channel was suggested
18Step 1 Condensation in the CONAN
Facility Modelling the cooled plate
- Concerning the simulation of the cooled plate,
two possible strategies were suggested, as
already experimented by the University of Pisa,
through internal preliminary calculations - adopting a conjugated heat transfer approach,
including heat conduction in the heated plate and
energy transport in the channel - imposing an equivalent heat transfer conductance
between the cooled surface and the secondary
fluid, obtained as reciprocal of the series of
the heat transfer resistances of the plate and of
the secondary fluid
19Step 1 Condensation in the CONAN
Facility Proposed approaches
Conjugated heat transfer Equivalent heat
transfer conductance
20Step 1 Condensation in the CONAN
Facility Experimental Uncertainties
Additional uncertainties are expected by the
limited steadiness of the process, requiring time
averaging of the relevant variables
21Step 1 Condensation in the CONAN
Facility Boundary Conditions
Five experimental data points were proposed
22Step 1 Condensation in the CONAN
Facility Obtained Results
A group of models (CEA, NRG, UJV, UNIPI and
VEIKI) provided very similar results, in
reasonable agreement with experimental data
negligible effect of plate representation
23Step 1 Condensation in the CONAN Facility Sample
Detailed Comparisons CEA Results
The observed tendency to underestimate the
overall condensation rate seems to be due to
limited accuracy in simulating boundary layer
development
24Step 1 Condensation in the CONAN Facility Sample
Detailed Comparisons UJV Results
This feature is common to all the models based
on low-Re capabilities and a mechanistic approach
for vapour diffusion the effect of simulation of
the falling film was weak
25Step 1 Condensation in the CONAN Facility Sample
Detailed Comparisons FzK Results
The near-wall model adopted by FzK in GASFLOW
provides a different heat flux distribution at
channel entrance
26Conclusions and Future Work General Outcome from
the Activity
- The two steps of the benchmarking activity
performed up to now allowed assessing the
behaviour of the models adopted in the frame of
the Severe Accident Research Network - The proposed problems were related to forced
convection condensation, which is often of lower
interest for containment analyses with respect to
free convection - However, the availability of the CONAN facility
offered an opportunity to address in a systematic
way the capabilities of codes in predicting
experimentally observed behaviour
27Conclusions and Future Work Quantitative Aspects
- Though at different extents, most of the adopted
CFD models are reasonably in agreement with the
information at the basis of well known
correlations - Even in the idealised case of condensation over
an isothermal flat plate, most codes provided a
reasonable prediction of the expected behaviour - The application to actual experiments revealed
more details on the behaviour of models,
highlighting a general tendency to underestimate
entrance effects, whose reason is still matter of
analysis (3D effects, boundary conditions)
28Conclusions and Future Work Challenges and Future
Perspectives
- The use of low-Re models is promising but still
relatively expensive from the computational point
of view if applied to large facilities or a full
scale plant. - Further efforts must be therefore spent in order
to develop accurate but affordable techniques for
predicting near-wall behaviour without loosing
too much about the necessary quantitative details
(e.g., FzK GASFLOW model) - The rather enthusiastic adhesion in both steps of
the activity encourages the University of Pisa to
go ahead in proposing new experimental data, also
on free convection