Computational Nuclear Engineering Overview & Bash - Max Fratoni & Katy Huff
Attending
- Kelly Rowland
- Katy Huff
- Alejandra Jolodosky
- Blake Huff
- Jasmina Vujic
- Sven
- Daniel Wooten
- Rachel Slaybaugh
- Denia Djokic
- Madicken Munk
- Sandra Bogetic
- Phil Gorman
- Naman
- Max Fratoni
Computational Tools for Nuclear Engineering, An Overview
Speaker Intro: Massimiliano Fratoni
Max Fratoni is a professor (forever freshman) in the nuclear engineering department who specializes in computational neutronics methods, advanced reactors, and accident tolerant fuels.
Discussion: Computational Tools for Nuclear Engineering
Max would like to help define what to use when. The first quest to ask is “What is your problem like?” As whether it is:
- a steady state or time dependent
- over a short (reactivity excursion) or long (depletion) time frame
The most generic types of tools are either
- stochastic (Monte Carlo)
- or deterministic (many).
The question, again, is “What are you trying to model?” If your simulation has a common geometry and common materials, then deterministic tools are certainly likely to be the answer. For deterministic codes, there are many simplifications, so it’s likely to be fast, but perhaps not as flexible.
If your geometry or you have unusual materials, stochastic models are probably going to capture your problem the best. In general, you will choose either MCNP or Serpent. So, when do you use MCNP and when do you use Serpent? While Serpent is very user friendly, the theory part in the Serpent manual, it is very hard to be confident in your results, since there are so many knobs that can be turned, but don’t actually have to be turned.
Serpent, for example, can combine points and make up its own energy grid. When you do this, you can lose accuracy, in particular in the unresolved resonances. This unified energy grid (which is set by default) will definitely bias some of your isotopics.
That’s fine, but MCNP doesn’t do depletion in a reliable way.
There are also a suite of codes that are capable of transient solutions by coupling with a monte carlo or deterministic code. These are often specifically designed for a certain reactor. This includes PARCS, for example.
Besides coupling with a monte carlo or deterministic code, depletion can be handled, by and large, by ORIGEN. ORIGEN2 and ORIGEN-S are your options. The resuls from ORIGEN are going to be just as good as your cross sections.
Future Topics
- What is the difference between the exponential matrix method and the kram method? (Daniel)
- Mocdown (Phil)
- PARCS (Sandra)
- Serpent&PARCS (Sandra)
- COMSOL (Madicken)
- MONTEBURNS (Alejandra)
- MOOSE (Katy)
Madicken will show off COMSOL next week, and then Daniel will talk the week after that.
Discussion: Bash and Unix / Linux Environments
Speaker Intro: Katy Huff
Katy Huff is an NSSC Postdoctoral Scholar and a Berkeley Institute for Data Science Fellow.
Discussion: Bash
Code examples can be found here.