A vertex model of recrystallization with stored energy implemented in GPU

Recrystallization models and simulations have been the subject of much attention in materials community in the past decades due to this process having a significant effect on many technologically important materials characteristics. Statistical analysis performed close to the steady state requires l...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Materials Research Express Ročník 6; číslo 5; s. 55506 - 55530
Hlavní autoři: Sazo, Alejandro H J, Torres, Claudio E, Emelianenko, Maria, Golovaty, Dmitry
Médium: Journal Article
Jazyk:angličtina
Vydáno: IOP Publishing 06.02.2019
Témata:
grain growth
numerical simulation
Polycrystalline materials
recrystallization
stored energy
vertex model
ISSN:2053-1591, 2053-1591
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Recrystallization models and simulations have been the subject of much attention in materials community in the past decades due to this process having a significant effect on many technologically important materials characteristics. Statistical analysis performed close to the steady state requires large-scale simulations, which are often prohibitively expensive from computational point of view. Graphical Processing Unit (GPU)-based realizations provide a viable approach to addressing this challenge, yet they remain relatively under-explored in this context.In the present manuscript, we develop a fully-parallelizable matrix-free GPU-based algorithm for implementing a two-dimensional vertex model of recrystallization based on the stored energy formalism. Nucleation is assumed to take place at triple junctions and obeys a Metropolis-type criterion. We include a complete mathematical analysis of the nucleation model deriving conditions under which nucleation is successful. Stability analysis of the dynamics of a triple junction under the presence of bulk energy is provided. On the computational side, we propose a novel polling system for handling topological transitions to ensure robust GPU implementation. Single grain tests are performed for benchmarking purposes. Finally, a set of numerical experiments for large scale systems is presented to explore the effect of initial distributions of stored energy on several statistical characteristics.
Bibliografie:MRX-111914.R1
ISSN:2053-1591
2053-1591
DOI:10.1088/2053-1591/ab00f8