Loading...
Reload document 
Open in new tab
- Titre : 16_10_UT_Colloquium.pdf
- Submitted by : Anonymous
- Description : The CADIS-! method: • Captures problem physics more effectively by normalizing the contributon flux by the forward flux. • Demonstrates an ability to more evenly distribute the uncertainty distribution in an f4 tally. • Has significantly stronger performance than CADIS for a deep-penetration shielding problem.
Transcription
FW/CADIS-ΩΩ: An Angle-Informed
Hybrid Method for Deep-Penetration
Radiation Transport
Madicken Munk_x000D_
October 19, 2016_x000D_
Departmental Colloquium_x000D_
University of Tennessee, Knoxville_x000D_
Who am I?
madicken@berkeley.edu _x000D_
2 _x000D_
Research Highlights: FHR
3 _x000D_
Research Highlights: SUERC & BGC
252Cf neutron energy spectrum
]
n
r
a
b
[
n
o
i
t
c
e
s
s
s
o
r
C
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
K (n,p) Ar
Cf sp. f.
U (n ,f)
Neutron energy
spectrum for 252Cf
relative to 235U, shown
relative to the neutron
capture cross section
for the 39K(n,p)39Ar
reaction. Note that the
252Cf spectrum is
slightly higher energy,
and thus more
favorable, than the
235U spectrum. From Li
et al. (2011).
0.4
0.3
0.2
0.1
0.0
]
.
u
.
a
[
y
t
i
s
n
e
t
n
i
l
a
r
t
c
e
p
S
Li et al., 2011
0
2
4
6
14
16
18
20
10
8
Energy [MeV]
12
H-chondrite age spectrum, modified from Swindle et al., 2009
Age spectrum for and
H-chondrite. Note difference
between the apparent impact
and formation ages and the
presumed integrated K-Ar
age. The application of the
K-Ar method may produce
potentially biased ages, with
no means of identifying
potential bias.
4 _x000D_
Present Work
Hybrid Methods for Strongly
Anisotropic Deep-Penetration
Radiation Transport
5 _x000D_
Talk Outline
• Introduction and motivation
• Hybrid methods background
– Adjoint and importance
– CADIS / FW-CADIS
– Existing state of angle-informed methods
• The ΩΩ method
– Method theory
– Software implementation
– Results on an example problem
madicken@berkeley.edu _x000D_
6 _x000D_
Motivation
• Radiation shielding has applications in a
number of fields
• “Analog” Monte Carlo for these types of
problems is not ideal
– Lots of particles à good statistics
– Good shielding à very few particles
• Hybrid methods leverage benefits of
_x000D_
s
d
o
h
t
e
M
d
i
r
b
y
H
_x000D_
_x000D_
_x000D_
k
r
o
W
y
M
_x000D_
deterministic codes to accelerate Monte Carlo
• A number of problems have strong angular
anisotropies
– The importance of a particle (whether it is likely
to contribute to a result) varies with direction
– Tracking in a low-importance direction is
inefficient
madicken@berkeley.edu _x000D_
7 _x000D_
_x000D_
n
o
9
c
u
d
e
R
e
c
n
a
i
r
a
V
_x000D_
Research Objectives
1. Create a new method to capture angular
information in an importance map for
Monte Carlo variance reduction
2. Compare the new method against existing
hybrid methods
3. Determine the sensitivity of the new
method to deterministic calculation
fidelity and problem geometry
madicken@berkeley.edu _x000D_
8 _x000D_