This slide is describing how to set up the OpenFOAM simulations including rotating geometries.
The SRF (Single Rotating Frame) is covered and MRF (Multiple Reference Frame).will be covered in it.
This slide is describing how to set up the OpenFOAM simulations including rotating geometries.
The SRF (Single Rotating Frame) is covered and MRF (Multiple Reference Frame).will be covered in it.
文献紹介:2D or not 2D? Adaptive 3D Convolution Selection for Efficient Video Reco...Toru Tamaki
Hengduo Li, Zuxuan Wu, Abhinav Shrivastava, Larry S. Davis; 2D or not 2D? Adaptive 3D Convolution Selection for Efficient Video Recognition, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021, pp. 6155-6164
https://openaccess.thecvf.com/content/CVPR2021/html/Li_2D_or_not_2D_Adaptive_3D_Convolution_Selection_for_Efficient_CVPR_2021_paper.html
文献紹介:2D or not 2D? Adaptive 3D Convolution Selection for Efficient Video Reco...Toru Tamaki
Hengduo Li, Zuxuan Wu, Abhinav Shrivastava, Larry S. Davis; 2D or not 2D? Adaptive 3D Convolution Selection for Efficient Video Recognition, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021, pp. 6155-6164
https://openaccess.thecvf.com/content/CVPR2021/html/Li_2D_or_not_2D_Adaptive_3D_Convolution_Selection_for_Efficient_CVPR_2021_paper.html
文献紹介:Deep Analysis of CNN-Based Spatio-Temporal Representations for Action Re...Toru Tamaki
Chun-Fu Richard Chen, Rameswar Panda, Kandan Ramakrishnan, Rogerio Feris, John Cohn, Aude Oliva, Quanfu Fan; Deep Analysis of CNN-Based Spatio-Temporal Representations for Action Recognition, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021, pp. 6165-6175
https://openaccess.thecvf.com/content/CVPR2021/html/Chen_Deep_Analysis_of_CNN-Based_Spatio-Temporal_Representations_for_Action_Recognition_CVPR_2021_paper.html
文献紹介:SegFormer: Simple and Efficient Design for Semantic Segmentation with Tr...Toru Tamaki
Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo, SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers, Advances in Neural Information Processing Systems 34 (NeurIPS 2021)
https://proceedings.neurips.cc/paper/2021/hash/64f1f27bf1b4ec22924fd0acb550c235-Abstract.html
https://arxiv.org/abs/2105.15203
The document summarizes research on simulating hydrogen dispersion using the ADVENTURE_sFlow solver. It describes modeling hydrogen dispersion as an analogy to thermal convection problems. Two models are analyzed: a hallway model and a car garage model. The hallway model analyzes hydrogen dispersion from inlet, door, and roof vents in an empty volume. The car garage model analyzes hydrogen leakage from a fuel cell car in a full-scale garage. The objective is to demonstrate the feasibility of using the ADVENTURE_sFlow solver, which uses a hierarchical domain decomposition method, to efficiently solve large-scale problems like hydrogen dispersion in engineering facilities.
22. 第 3 回 ADVENTURE 定期セミナー
2006 年 3 月 17 日 ( 金 )
solver/hddm_types.h
typedef struct {
int partid; /*+ Part ID +*/
int domid; /*+ Subdomain ID +*/
int gdomid; /*+ Global Subdomain ID +*/
int elmtype; /*+ Element type +*/
int elm_dim; /*+ Dimension of the element +*/
int nd_elm; /*+ num nodes in a element +*/
int elms; /*+ num of elements +*/
int nodes; /*+ num of nodes +*/
int node_dim; /*+ dimension of node coodinate +*/
int ndisp; /*+ num of bcdisp +*/
int nload; /*+ num of bcload +*/
int ninbd; /*+ num of inbc +*/
int ifn_dim ; /*+ dim. of inbc +*/
int nsharenode; /*+ num of share nodes +*/
int* nop; /*+ connectivity +*/
int* ndindex; /*+ Node ID index (Subdomain -> Part) +*/
Bcnd* bcdisp; /*+ array of bcdisp +*/
Bcnd* bcload; /*+ array of bcload +*/
} DomMesh;
23. 第 3 回 ADVENTURE 定期セミナー
2006 年 3 月 17 日 ( 金 )
typedef struct {
int n_part; /*+ num part +*/
int partid; /*+ part ID +*/
int n_domain; /*+ num subdomains +*/
int t_nodes; /*+ nodes in this part +*/
int node_dim; /*+ dim of node coordinate +*/
int t_infree; /*+ num of inbc assigned to this part +*/
int t_outfree; /*+ num of inbc assigned to other parts +*/
int t_insnode; /*+ num of share nodes assigned to this part +*/
int t_outsnode; /*+ num of share nodes assigned to other parts +*/
OPinfo* op; /*+ inter part inner bc info +*/
OPSinfo* opsn; /*+ inter part share nodes info +*/
double* crd; /*+ node coordinate +*/
double* crd2; /*+ node coordinate for ULF+*/
int* pndindex; /*+ Node ID index (part -> global) +*/
int global_t_nodes; /*+ num. global nodes +*/
int global_t_elms; /*+ num. global elements +*/
int global_t_domains; /*+ num. all domains +*/
DomMesh* dom; /*+ Sub domain meshes +*/
int* off_gdom ; /*+ GlobalDomID = domid (in ipart) + off_gdom[ipart] +*/
} PartMesh;
37. 第 3 回 ADVENTURE 定期セミナー
2006 年 3 月 17 日 ( 金 )
• Two example problems for time dependent
boundary conditions and temperature dependent
material properties
38. 第 3 回 ADVENTURE 定期セミナー
2006 年 3 月 17 日 ( 金 )
Heat Equations
vectorNormal:n
][retemperatuExternal:
)][kcal/(sectcoefficienferHeat trans:
generationheatInternal:
)]/([heatSpecifice:
][kg/mmDensity:
][onappliedeTemperatur:u
sec)]/([onappliedfluxHeat:Q
retemperatuofderivativeTime:
t
u
]/(sec[tyconductiviThermal:
rflux vectoHeat:q
][eTemperatur:
2
C
3
u
2
Q
Cu
Cmm
f
CkgkcalC
C
mmkcal
Cmmkcal
Cu
o
C
o
o
o
o
o
⋅⋅
⋅
Γ
⋅Γ
∂
∂
⋅⋅
α
ρ
λ
(5)on
(4)on0
(3)on0)(
(2)indiv
(1)ingrad
u
_ Q
Γ=
Γ=−⋅
Γ=−−⋅
Ω+−=
∂
∂
Ω−=
uu
Qnq
uunq
fq
t
u
c
uq
CCCα
ρ
λ
uΓ
39. 第 3 回 ADVENTURE 定期セミナー
2006 年 3 月 17 日 ( 金 )
Example Problem (1/2)
Outer radius 110 mm
Inner radius 100 mm
Convection b.c.
Inner side has convection boundary
conditions.
Other sides are of natural
boundary conditions.
#elms:8941
#nodes:14491
#nodes per element : 10
100mm
110mm
flow
2D problem
3D problem
40. 第 3 回 ADVENTURE 定期セミナー
2006 年 3 月 17 日 ( 金 )
Time dependent external temperature of the
heat convection boundary conditions.
Time vs Temperature
0
100
200
300
400
500
600
0 50 100 150
Time (s)
Temperature(C)
系列1
Heat conductivity : 4.8*E-6 W/mm K.
Specific heat : 0.133 J/kg K.
Heat transfer coefficient : 4.2E-6 W/mm2
K.
This slide shows the stress distribution and deformed shape by seismic response analysis.
This video shows results from 0 seconds to 1.0 seconds.
This slide shows comparison with the case of 1,024 and 2,048 processors.
Both case shows over 26% of peak performance, especially we archived about 4 Tflops with 2,048 processors.
Therefore, we succeeded the analysis of 100 million dofs model in about only 10 minutes.
I think, in FE analysis with unstructured mesh, this result is one of the best performance.