Software and Algorithmic Aspects of Parallel Calculation of Non-Isothermal Moisture Transfer in Fractal-Structure Materials

A new mathematical model of the nonstationary process of heat and moisture transfer in the two-dimensional region is constructed on the basis of the use of Kaputo and Grunwald-Letnikov derivatives. An implicit finite-difference scheme for approximation of a mathematical model of noisothermal moistur...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:International Conference on Perspective Technologies and Methods in MEMS Design (Online) s. 171 - 175
Hlavní autoři: Sokolovskyy, Yaroslav, Yarkun, Volodymyr, Levkovych, Mar'iana, Storozhuk, Oleksandr, Kapran, Ihor
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 12.05.2021
Témata:
Acceleration
Algorithm
Caputo derivatives
Computers
Core
Efficiency
Fractal materials
Grunwald-Letnikov derivatives
Heat-and-moisture transfer
Libraries
Method of fractional steps
Micromechanical devices
Moisture
Non-stationary process
Parallelization
Partitioning algorithms
Processor
Software
Software algorithms
Thread
ISSN:2573-5373
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:A new mathematical model of the nonstationary process of heat and moisture transfer in the two-dimensional region is constructed on the basis of the use of Kaputo and Grunwald-Letnikov derivatives. An implicit finite-difference scheme for approximation of a mathematical model of noisothermal moisture transfer taking into account the fractional integro-differential apparatus is developed. The given algorithm of numerical realization of model allows to receive values of function of temperature and humidity for all points of area of partition. The method of fractional steps is adapted for numerical realization of mathematical model. It allowed performing parallel calculations of two difference half-step taking into account the corresponding spatial coordinate. An analysis of parallel solutions on multi-core computers has been performed. The software a parallel algorithm using .Net threads and algorithmic features of the TPL library has developed. It has been tested on multi-core computer systems with CPUs of different capacities. The analysis of acceleration and efficiency of the algorithm on the basis of the constructed graphs is carried out.
ISSN:2573-5373
DOI:10.1109/MEMSTECH53091.2021.9467939