Paper R E V I E W
Coupled Neutronics and Thermal-Hydraulics Simulations Using Monte Carlo and CFD Volkan Seker, Justin W. Thomas and Thomas J. Downar Purdue University Nuclear Engineering Building 400 Central Drive West Lafayette, IN,47907
[email protected] ;
[email protected] ;
[email protected] Presented by : Dinan Andiwijayakusuma, S.Si
Major Work
A major part of the work was to develop and implement methods to update the cross section library with the temperature distribution calculated by STAR-CD for every region
15/06/09
BATAN-Computational Division
2
Decription the Work
MCNP --> was used to simulate the transport of neutrons through the system STAR-CD --> solves the 3-D momentum and energy transport equations NJOY --> Generated Cross Sections Data using data from CFD solution
15/06/09
BATAN-Computational Division
3
Interface
Written by FORTRAN90 with utilized two PERL script
Execute MCNP (output : “mcnp2star.dat”)
Execute STAR-CD (output : “star2mcnp.dat”)
Rewritten “xsdir” with new generation crosssection identifiers,temperatures, library names dan library paths. The Script running until the Eigen Value and flux are convergen
15/06/09
BATAN-Computational Division
4
Initialization
Stand alone CFD run with input initial power profile --> to obtain initial temperature distribution for MCNP cross-section library MCNP run with standard library to obtain inital source
15/06/09
BATAN-Computational Division
5
Iterative Process The iterative calculation :
MCNP (input power) --> output power distribution --> mcnp2star.dat STAR-CD (read mcnp2star.dat) --> output temperature,density and volume each cell --> star2mcnp.dat NJOY (read star2mcnp.dat) --> generate new CS library --> update MCNP input (new density & new temperature)
Proses tsb berulang (iteratif) sd tercapai keadaan konvergen 15/06/09 BATAN-Computational Division 6
Iterative Process
Proses tsb berulang (iteratif) sd tercapai keadaan konvergen 15/06/09
BATAN-Computational Division
7
Three Techniques for updating Cross Section Terkait dengan “Major Works” untuk update Cross-Section, maka ada 3 teknik/pendekatan :
Perform ENJOY calculation for each nuclide in each region Pre-generate a library for each nuclide with small temperature increment (2K - 5K) Pre-generate a library for each nuclide with larger temperature increment (25K - 50K)
15/06/09
BATAN-Computational Division
8
Result GEOMETRY
15/06/09
BATAN-Computational Division
9
Result CFD MESHING
15/06/09
Arah Axial : Tinggi 240cm di-diskritisasi menjadi 600 layer, shg masing2 layer 0.4cm Arah Radial di diskritisasi sama utk masing2 layer,yaitu 2240 cells Total cells : 600x2240= 1.344.000 CFD cells
BATAN-Computational Division
10
Result MCNP MESHING
15/06/09
Azimuthal meshing = 8mesh Axial meshing = 12mesh Jumlah total MCNP cells: 6720 with tallies performed dalam 1920 cells pada daerah fuel
BATAN-Computational Division
11
Result
Computing run-time for a single MCNP : siklus aktif = 300, neutron per-siklus = 500.000 neutron running on cluster 30 nodes (spek PC: 3GHz Pentium 4 Processors RAM 2GB) dibutuhkan waktu 6 jam
Untuk total 12 iterasi dibutuhkan 100 jam
15/06/09
BATAN-Computational Division
12
Result Perbandingan Hasil MCNP dengan DeCART Code*) :
Percobaan I, dengan distribusi suhu konstan (300oC) diperoleh hasil :
Terdapat selisih MCNP lebih besar 74pcm *) deterministic transport code
15/06/09
BATAN-Computational Division
13
Result Perbandingan Hasil MCNP-STARCD (McStar) dengan DeCART/STARC-D Code*) :
Percobaan II, dengan distribusi variable suhu diperoleh :
Terdapat selisih McStar lebih besar 66pcm, perbedaannya tidak terlalu jauh dengan kasus suhu konstan
*) deterministic transport code 15/06/09
BATAN-Computational Division
14
Result
15/06/09
BATAN-Computational Division
15
Result
15/06/09
BATAN-Computational Division
16
Result
Perbedaan hasil eigen-value utk harga k-eff : Case I -->Temperature constant : 74 pcm Case II --> Variable temperature : 64pcm
Perbedaan hasil Power Density : Case I -->Temperature constant : 3,2% Case II --> Variable temperature : 4%
15/06/09
BATAN-Computational Division
17
Conclusion
Secara umum Coupling MCNP-CFD sangat baik, namun karena waktu komputasi yang sangat lama, maka metode ini belum bisa menggantikan coupling Deterministik DeCARTCFD. Metode Coupling MCNP-CFD ini bisa digunakan sebagai audit-tools untuk verifikasi hasil perhitungan dengan metode deterministik pada kasus-kasus yang lebih spesifik.
15/06/09
BATAN-Computational Division
18
Future Works Dengan pertimbangan akurasi tinggi pada metode coupling MCNP-CFD namun dengan waktu komputasi yang lama, maka :
Perlu dikembangkan metode yang bisa mempercepat waktu komputasinya, misal dengan teknik variance reduction Infrastruktur komputasi yang mampu menunjang sehingga diperoleh waktu komputasi yang cepat
15/06/09
BATAN-Computational Division
19
Terkait dengan USPEN 2010
Mempelajari requirement input masing2 SW (Neutronik->mis:MCNP & TH-->mis:COBRA) Mempelajari output yang dihasilkan masing2 SW (Neutronik->mis:MCNP & TH-->mis:COBRA) Mempelajari teknik update cross-section, yaitu 1. Mempelajari teknik NJOY mengenerate library crosssection data 2.Mempelajari penyiapan library cross-section data secara manual dengan kenaikan suhu tertentu
Mempelajari teknik komunikasi data yang memungkinkan untuk ketiga komponen tsb 15/06/09
BATAN-Computational Division
20
Terkait dengan USPEN MCNP Input : 1. Geometry Properties 2. Material Properties 3. Source Card 4. Tally Card
New Cross-Section Library Data
MCNP Output : Neutron Fluks Distribution (Converted to Power Distribution)
CFD (FLUENT) read MCNP output Power Distribution as Heat Source sebagai bagian input dari CFD(FLUENT)
CFD output : Temperature and density of every cells. This data used to update cross-section requires a mesh mapping or interpolation scheme between the meshes used in 15/06/09
BATAN-Computational Division
21
Output
15/06/09
BATAN-Computational Division
22
NEXT STEP
SP 1 : NEUTRONIK
SP 2 : THERMALHYDRAULIC
SP-4 : INTERFACE TO INTEGRATED
SP-3 : GENERATE CROSS-SECTION
15/06/09
BATAN-Computational Division
23
CLOSING
Any Suggest..???
15/06/09
BATAN-Computational Division
24