use of the fission matrix method for solution of the eigenvalue
TRANSCRIPT
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
USE OF THE FISSION MATRIX METHOD FOR SOLUTION OF THE EIGENVALUE
PROBLEM IN A SPENT FUEL POOL
William Walters, Nathan Roskoff, Katherine Royston, and AlirezaHaghighat
Virginia Tech
Nuclear Science and Engineering Lab (NSEL) at Arlington
Nuclear Engineering Program, Mechanical Eng. Dept.
Arlington, VA
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Outline
• Introduction
• Background• Fission Matrix Method• Reference Spent Fuel Pool
•Methodology
• Results• Full Monte Carlo• Fission Matrix
• Conclusions and Future Work
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Introduction
• Spent fuel pool neutronics calculations are important• Criticality safety• Safeguards and verification
• Full Monte Carlo calculations face difficulties in this area• Convergence is difficult to low coupling between regions (due to
absobers)• Convergence can also be difficult to detect
• Computation times are very long, especially to get detailed information• Changing pool configuration requires complete recalculation
• Fission Matrix method can address some of these issues• Fission matrix coefficients are pre-calculated using Monte Carlo• Computation times are much shorter, with no convergence issues• Detailed fission distributions are obtained at pin level• Changing pool assembly configuration does not require new pre-
calculations• No additional Monte Carlo
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Fission Matrix (FM) Method• Eigenvalue formulation
𝐹𝑖 =1
𝑘 𝑗=1𝑁 𝑎𝑖,𝑗𝐹𝑗
• k is eigenvalue• 𝐹𝑗 is fission source, 𝑆𝑗 is fixed source in cell j• 𝑎𝑖,𝑗 is the number of fission neutrons produced in cell 𝑖 due to a fission neutron
born in cell 𝑗.
• Subcritical multiplication formulation
𝐹𝑖 =
𝑗=1
𝑁
(𝑎𝑖,𝑗𝐹𝑗 + 𝑏𝑖,𝑗𝑆𝑗) ,
• 𝑏𝑖,𝑗 is the number of fission neutrons produced in cell 𝑖 due to a source neutron born in cell 𝑗.
4
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Fission Matrix Formulation
• The main task
Determination of coefficients 𝑎𝑖,𝑗 and 𝑏𝑖,𝑗, especially at the pin-level.
We reduced the number of calculations by considering:• Geometric similarity• Geometric symmetry• Degree of coupling
• Sensitivity of the coefficients to different parameters
5
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Reference Spent Fuel Pool
• Being developed for Integral Inherently Safe LWR (I2S-LWR)* reactor design
• 19x19 𝑈3𝑆𝑖2 fuel assemblies
• 4.45 w/o and 4.95 w/o U-235
• Metamic® absorbers (Al-B4C) used between assemblies
• Considering fresh fuel for now
One 19x19 Assembly9x9 segment of spent
fuel pool*Led by Georgia Tech, funded under a DOE-IRP program
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Methodology
• As the computational cell size, use a single pin• 𝑁 = 9 × 9 × 336 = 27,216 total fuel pins / fission matrix cells
• Allows for good accuracy and pin-resolved fission rates
• Standard FM would require 27,216 separate fixed-source calculations to determine FM coefficients
• The standard approach is clearly NOT feasible
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Fission Matrix Coefficients
• For this work, we performed 49 fixed-source MCNP5 calculation separately for each pin in one octant of an assembly in the middle of the pool• Uniform source radially
• Cosine shape axially
• U-235 fission spectrum
• We use the similar coefficients considering symmetries for the rest of pins in an assembly as follows
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Fission Matrix Coefficients for one assembly
Source in single pin Neutron production tallied in all cells (results are x 100)
Repeat for all 49 source pins in octant:
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
9x9 array of assemblies – Using geometric similarity of coefficients
• Previous calculation only gives 49 rows of the fission matrix
• Obtain the rest using octal symmetry and geometric similarity
• Coefficients with red arrows are identical, as are those with blue arrows
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Solution of Fission Matrix Equations• Once the 49 base sets of fission matrix coefficients have been
calculated, the full set of coefficients is obtained using the geometric considerations for any pool configuration.
•𝑁 = 336 ∗ 𝑁𝑎𝑠𝑠𝑒𝑚𝑏𝑙𝑖𝑒𝑠 total equations
• Solve for k and fission density (fundamental eigenfuction) using a power iteration approach is used as follows
𝐹𝑖(0)=1
𝑁
𝐹𝑖(𝑚+1)=1
𝑘(𝑚)
𝑗=1
𝑁
𝑎𝑖,𝑗𝐹𝑗(𝑚)
Where 𝑘(𝑚) = 𝑖=1𝑁 𝐹𝑖
(𝑚)
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Test Problems
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Fission Matrix vs Full MCNP
*total pre-calculation time for both materials and used for all cases: 4707 min
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Case 3 Eigenfunction
Total k difference: -143 pcm
0,00E+00
2,00E-02
4,00E-02
6,00E-02
8,00E-02
1,00E-01
1,20E-01
1,40E-01
1,60E-01
1,80E-01
0 2 4 6 8 10
Fiss
ion
So
urc
e
Assembly number
MCNP FM
Comparison of FM with MC
Reference Solution
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Case 11 Eigenfunction
Total k difference: -135 pcm
0,00E+00
2,00E-02
4,00E-02
6,00E-02
8,00E-02
1,00E-01
1,20E-01
1,40E-01
1,60E-01
1,80E-01
0 2 4 6 8 10
Fiss
ion
So
urc
e
Assembly number
MCNP FM
Comparison with FM with MC
Reference Solution
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Case 4 Eigenfunction distribution
*MCNP Uncertainties <1%
Total k difference: -192 pcm
Comparison with FM with MC
Reference Solution
0,00E+00
2,00E-02
4,00E-02
6,00E-02
8,00E-02
1,00E-01
1,20E-01
1,40E-01
1,60E-01
1 2 3 4 5 6 7 8 9
Fiss
ion
So
urc
e
Pool Assembly Column Number
Eigenfunction (sum of rows)
MCNP FM
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Conclusions
• The fission matrix method can quickly and accurately calculate the eigenvalue and eigenfunction for a spent fuel pool
• Relatively few pre-calculations, using geometric considerations (reflection, etc)
• Changes in assembly positions or adding assemblies does not require any new Monte Carlo calculations
Virginia Tech Nuclear Science and Engineering Lab (NSEL) at Arlington
Ongoing & Future Work
• Filing for a patent for an automated tool for safety and security of spent fuel pool
•Use of spent fuel materials instead of fresh fuel
•Better treatment of axial variable
•Develop a database of coefficients for all possible burnup and cooling times
•Calculate subcritical multiplication factor and detector response