I agree with you Victor. I'm learning MCNP right now. People that I know here usually use codes based on Diffusion theory, and it's not enought to elaborate a PhD thesis, I think. Thanks.

Hello! In principle, Monte Carlo codes are in the end the most accurate models to calculate the spatial dependence of the neutron flux. However, they are still not practical in a full core analysis, especially if you want to study a fuel cycle (i.e. burnup calculation). Diffusion theory has proven to work decently in full core calculations due to geometrical and other characteristics of the system. Codes such as SIMULATE, POLCA, NESTLE, DONJON, etc., rely in two-group diffusion theory (RAMONA uses 1.5 energy groups). I know that SIMULATE and POLCA have been widely used in the core loading design of commercial LWR reactors. For instance, I think that the newest version SIMULATE (SIMULATE-4) has the option to calculate the neutron flux using SP3 theory (simplified P3). This means that they handle much better the flux predictions close to boundaries where the flux suddenly changes its shape (e.g. control rods). These are some examples of commercial and research codes that are used in thermal reactors.

Hello, thanks for your speech, Augusto. I'm learning MCNP now, like I said before. However, here the people usually use CITATION code too (personally, I think it's an old code). Anyway, I believe that Monte Carlo it's one of the most powerful CODEs not just in Nuclear stuffs but everywhere in physics researches. Nice to see what you said, I didn't imagine that people were using diffusion theory to build commercial reactors. To validate these diffusion theory codes calculations, however, people around the world usually use Monte Carlo, am I ok? I'm doing my PhD and my main objective it's to reduce the IEA-R1 reactor core and to perform it, I have to choice some CODEs. MCNP will be one of that... What do you think about? Is there some other accurate CODE more convenient than MCNP ? Thanks for writing. I'm get starting in this world of nuclear reactor simulations.

Hello Thiago! Welcome to the reactor simulation world! Now I have a better picture of what you are asking. Well, core simulators that rely on diffusion theory are indeed used nowadays in the fuel loading design of thermal reactors. Your are right in the sense that Monte Carlo is a very powerful technique (quite clever from the guys who deigned it). However, a full core calculation of a commercial thermal reactor is very computational expensive. Deterministic modelling of reactor cores rely on a two-step approach, where first you create the homogenized and energy-collapsed macroscopic cross-sections with a so-called lattice code. Basically, the lattice code creates the input data for core simulators. Mostly the validation against Monte Carlo has been done at the lattice stage rather than at the core stage. Nevertheless, nowadays, with the increase of computational power, Monte Carlo can be used in burnup core calculations of not so big cores like for example, research reactors (especially because it might be that in this type of reactors, diffusion theory could not work so well). Anyhow, for LWRs, deterministic diffusion codes are coupled with fuel performance and thermal-hydraulic codes in order to design fuel cycles. In practice, such core simulators are validated with diverse experimental data and in-core detectors information. You have to keep in mind that diffusion theory is a way to treat the angle dependence of the flux. Therefore, it is equivalent to P1 theory by assuming certain considerations (i.e. time independence, no anisotropic neutron source, incoming and outgoing energy integration of the scattering term should be the same). The thing is that in transport theory the boundary condition is well defined, while in diffusion theory is not!! I hope this is kind of useful for you. If you like to keep on chatting about this in more detail, we can e-mail ([email protected]).

Thank you Augusto. The IEA-R1 is a research reactor. It's 5x5 core and it's not a large core, actually, it's a small core (and will get smaller than this 5x5, cause my PhD thesis is about this core reduction). So, I think MCNP can be used. What I'm thinking to do it's on the first time, simulating the actual core on MCNP and CITATION, using on the fuel elements a kind of homogenization (thinking the fuel element filled with a single

specific mass of fuel and burn up ). However, after getting this comparison, I probably will look just to MCNP and I'll try to discretize the plates within the fuel elements, but still with average fuel burn up per fuel elements. In the end, if I have time, I'll try to simulate the real case, with the whole information in each plate in each fuel element. After that, I have to reduce this core, seeking a new small configuration, using higher density fuel performing the thermo-hidraulycs and safety analysis. So, I have a question, Will I have to perform the burn up to this new configuration, or it's not necessary? If you want, Augusto, my e-mail is [email protected]... thanks.

Hi Thiago! I'm not sure I followed you 100%. Once you have a new model of the core, which is a reduced version of the previous core configuration, in principle you need to recalculate the depletion calculation since the flux in the new core will not be the same as the previous one. Also as I look into your question, the new core will have higher density fuel than the original core, and that for sure will affect the isotopic rate composition of the new core. I hope this is useful Thiago. BTW, have you checked with Monte Carlo code will you be using? Of course MNCP is great. If you are doing research, you would like also to take a look into the SEPRENT code, which is a Monte Carlo code developed by a guy in VTT from FInland. This code is versatile, open and has been used for modelling full research reactor cores. This code you can request it via the NEA/OECD data bank.

OK, I was just thinking if making the criticality calculation was be enough, but thinking better, I see that I'll have to perform burn up calculation after get this new core configuration.

Hello everyone i use in DRAGON an DONJON codes.

There is a piece of free code I am writing that might be helpful. Revisions and contributions are welcome:

https://www.seamplex.com/milonga

I'll check your code later, I've been using MCNP to perform such simulations for while. I guess it's a good code as long as the purpose is a original thesis. I was in Brazil when I asked for the first time about this codes. Now I'm on USA working with TH simulations. Anyway, for research reactor simpler codes are welcome specially when we apply diffusion theory.

Indeed practically core simulator code like PARCS is used, which is based on diffusion theory, it is very simple. Can be coupled with TH codes like RELAP and TRACE for further analysis.

Indeed, Mr. Thiago, you cannot complete a core design, without depletion calculations. Initial validation of the codes you plan to use is important. Some times you will have to use a combination of deterministic and stochastic (Monte Carlo) codes for arriving at a final design.

Usually,MCNP/SCALE/CASMO is applied as steady state design tool,such as assembly arrangement design and K calculation and so on.With the preliminary design,we do the transient analysis by SIMULATE3K or other tools.For the accident simulation,we use RELAP. Different tool has its own application case.

I prefer mcnpx neutronic code because in this code we can simulate the detailed geometry with good result for cross  sections. and using parcs , origen. dot , danjon and any other code to solve the transport equation.

DANJON and DRAGON codes are based on diffusin theory..also PARCS can do your jobs as well..

Zahra Tabadar Hello Zahra, DRAGON is a lattice code which is based on the neutron transport theory, and the DONJON is a full core code which is based on the diffusion theory. We can use the DRAGON to calculate the group constants with transport theory, which will be supplied for the DONJON code to make the further core analysis.

If you don't have access to any commercial codes, you can try using open-source codes. Here is a link that contains the list of all open-source nuclear codes and their websites.

https://github.com/paulromano/awesome-nuclear