Loading...
Reload document 
Open in new tab
- Titre : "05_Cihangir_Celik_0.pdf"
- Submitted by : Anonymous
- Description : CADIS (1/2) C. onsistent . A. djoint . D. riven . I. mportance . S. ampling Biased source and importance map work together . Ali Haghighat and John C. Wagner, “Monte Carlo Variance Reduction with Deterministic Importance Functions,” Progress in Nuclear Energy, 42(1), 25- 53, (2003). • Solve the adjoint problem using the detector response ...
Transcription
CADIS Variance Reduction with MAVRIC
Cihangir Celik
SCALE Users’ Group Workshop
ORNL, July 27-29, 2020
ORNL is managed by UT-Battelle, LLC for the US Department of Energy
Monaco with Automated Variance Reduction using Importance Calculations
MAVRIC
•Elements of a Shielding Calculation
– Physical Model
– Source
– Responses
– Calculational Parameters
22
Open slide master to edit
Physical Model
•Geometry description
– Boolean operation of solid and surface equations
•Material compositions
– Physical mixture: weight fractions, atomic fractions
– Chemical formula
– Density
– Isotopic distribution of elements
•Interaction cross sections
33
Open slide master to edit
Sources
•Particle type
•Spatial distribution
•Energy distribution
•Directional distribution
44
Open slide master to edit
Responses
•Type of response
– At a point
– For a geometric region
– Superimposed mesh
•Quantity to Calculate
– Flux
– Dose Rate
– Reaction Rate
•Dimensionality
– Total
– Function of R, E, Ω, etc
55
Open slide master to edit
SCALE Strategy for Shielding is Monte Carlo
• Calculational Parameters for Monte Carlo Simulations
– Minimum accuracy
– Maximum runtime
– Problem truncation
– Variance reduction
66
Open slide master to edit
Monte Carlo Simulation Strengths and Drawbacks
•Strengths of MC
– Straightforward physics
• particle-based interactions,
• physics expressed as probability distribution functions (pdf)
– Geometry at any level of detail – no meshing approximations
– Time – depends on problem, how many results (tallies), level of geometry
•Drawbacks of MC
detail
•Difficult Problems
– Streaming
– Highly scattering
– Deep penetration
Efficient Variance Reduction is key with
Monte Carlo shielding problems!
77
Open slide master to edit
Variance Reduction Methods
•Changing the sampling routines to
optimize the simulation to get more
particles to do something
– Forcing, biasing, stretching, implicit capture,
and weight windows
– Requires knowledge about how the
calculation will most likely proceed
– May require iterative process to tune biasing
– Multiple methods may work against each
other
– “Bad” biasing can slow the rate of
convergence
88
Open slide master to edit