Reliability & Risk Assessment Laboratory


Research


Accelerated probabilistic risk assessment of nuclear power plants through parallel computing

The research focuses on the Probabilistic Risk Assessment (PRA) of nuclear power plant. Risk Assessment tries to identify the possible effects of a hazard event and all possible failure modes of a system given: the hazard event,the vulnerability of the components of the system, and the topology of the system. The scope is to accelerate the analysis of event tree and fault tree through GPU computing. Due to the parallel nature of these processors the algorithms must be reshaped to achieve peak performances.


Research duration: 3 year
Funding agency: Knif2030



Fragility analysis of vulnerable components subject to concurrent hazard events

The likelihood of two uncorrelated rare events occurring in a short time frame is almost 0. However, situations where a first event triggers a second one do happen. These situations are defined as concurrent hazards events. For example in 2011, Tohoku earthquake triggered a tsunami. In 1906, the earthquake that hit San Francisco triggered a fire that destroyed almost 80% of the city. The goal of this research is to develop a new method to compute fragility curves of vulnerable components subject to concurrent hazards events. This method will be useful in the PRA of civil infrastructure systems.


Research duration: 1 year
Funding agency: UNIST - startup research



Bridge life-span extension using ICT, partial replacement and low-carbon materials (ICT bridge)

This project aims at extending the life span of bridges, at reducing both the carbon production during construction and their life cycle cost. The r2alab is in charge of developing a portable system to assess the status of a bridge and to predict the extension and location of damage using forced vibrations. System identification, spatial variablity, analyses in both time domain and frequency domain, stochastic processes will all be considered to achieve the goal.


Research duration: 5 year
Funding agency: Ministry of Land Transportation and Marine



Real-time reliability analysis of civil infrastructure using multi-GPU systems

Korean: GPGPU 기반 병렬연산을 이용한 사회기반시설물의 실시간 신뢰성 해석 기술 개발

Nowadays, structural health monitoring systems on large structures, like bridges, are equipped with up to hundreds of sensors. Even if the sampling frequency is low (<100 Hz) the amount of data produced by these systems is to big to store on site or stream online. Powerful machines are necessary to process the data on site. Multi-GPU systems can perform computationally expensive algorithm and still fit in a desktop computer. In this research, system identification, damage assessment, and reliability algorithm will be shaped to fit this new computing technology.


Research duration: 2.5 year
Funding agency: National Research Foundation (NRF)



Development of Advanced Mechanical Systems for Monitoring and handling Chemical/Biological Disaster

Korean: 생화학 및 화학재난 감시∙긴급대응 기술 및 장비개발

TBA


Research duration: 5 year
Funding agency: TBA (TBA)


R2A lab - Equipment


Gwiseon photo

"GWISEON - GPGPU Desktop"


GWISEON (GPGPU Desktop)
CPU: 1 x Intel I5-2500 (4 cores)
GPU: 1 x GeForce GT440 (96 CUDA cores)
OS: Ubuntu 12.10
Performance(TFlops): 0.311 SP, NaN-:) DP


GEOBUKSEON photo

"GEOBUKSEON - GPGPU Workstation"


CPU: 6 x Intel Xeon E5-2609 (24 cores)
GPU: 12 x Tesla K10 (36,864 CUDA cores)
OS: CentOS 6.3
Performance(TFlops): 54.6 SP, NaN-:) DP


PANOKSEON photo

"PANOKSEON - GPGPU server"


CPU: 2 x Intel Xeon E5-2609 (8 cores)
GPU: 4 x Tesla K20 (10,752 CUDA cores)
OS: CentOS 6.3
Performance(TFlops): 15.8 SP, 5.2 DP

UNIST Supercomputer Center

UNIST_SC logo

"UNIST supercomputer room"


GYRFALCON (X86 Cluster)
CPU: 216 x Intel Xeon 2.53 Ghz (27 nodes)
OS: CentosOS 5.3
Performance(TFlops): 2.185


LEOPARD (X86 Cluster)
CPU: 224 x Intel Xeon 2.53 Ghz (28 nodes)
OS: CentosOS 5.5
Performance(TFlops): 2.266


LION (X86 Cluster)
CPU: 276 x Intel Xeon 2.53 Ghz (12 nodes)
OS: CentosOS 5.6
Performance(TFlops): 2.793


UNIST-SMP (IBM P690)
CPU: 64 x IBM Power4 1.7 Ghz (32 nodes)
OS: AIX 5.3
Performance(TFlops): 0.435


CHEETAH (X86 Cluster)
CPU: 296 x Intel Xeon 2.53 Ghz (37 nodes)
OS: CentosOS 5.4
Performance(TFlops): 2.995




Copyright © 2012 R2A Lab
UNIST-gil 50, Ulsan 689-798, Republic of Korea.
울산광역시 울주군 언양읍 유니스트길 50

R2A logo
UNIST logo