• 한국안전학회지
• /
• 제33권6호
• /
• pp.144-156
• /
• 2018

Level 3 Probabilistic Safety Assessment (PSA) is performed for the risk assessment that calculates radioactive material dispersion to the environment. This risk assessment is performed with a tool of MELCOR Accident Consequence Code System (MACCS2 or WinMACCS). For the off-site consequence analysis of multi-unit nuclear power plant (NPP) accident, the single location (Center Of Mass, COM) method has been usually adopted with the assumption that all the NPPs in the nuclear site are located at the same COM point. It was well known that this COM calculation can lead to underestimated or overestimated radionuclide concentration. In order to overcome this underestimation or overestimation of radionuclide concentrations in the COM method, Multiple Location (ML) method was developed in this study. The radionuclide concentrations for the individual NPPs are separately calculated, and they are summed at every location in the nuclear site by the post-processing of radionuclide concentrations that is based on two-dimensional Gaussian Plume equations. In order to demonstrate the efficiency of the ML method, radionuclide concentrations were calculated for the six-unit NPP site, radionuclide concentrations of the ML method were compared with those by COM method. This comparison was performed for conditions of constant weather, yearly weather in Korea, and four seasons, and the results were discussed. This new ML method (1) improves accuracy of radionuclide concentrations when multi-unit NPP accident occurs, (2) calculates realistic atmospheric dispersion of radionuclides under various weather conditions, and finally (3) supports off-site emergency plan optimization. It is recommended that this new method be applied to the risk assessment of multi-unit NPP accident. This new method drastically improves the accuracy of radionuclide concentrations at the locations adjacent to or very close to NPPs. This ML method has a great strength over the COM method when people live near nuclear site, since it provides accurate radionuclide concentrations or radiation doses.

• Nuclear Engineering and Technology
• /
• 제52권10호
• /
• pp.2221-2229
• /
• 2020

After the Fukushima Daiichi nuclear power plant (NPP) accident, level 3 probabilistic safety assessment (PSA) has emerged as an important task in order to assess the risk level of the multi-unit NPPs in a single nuclear site. Accurate calculation of the radionuclide concentrations and exposure doses to the public is required if a nuclear site has multi-unit NPPs and large number of people live near NPPs. So, there has been a great need to develop a new method or procedure for the fast and accurate offsite consequence calculation for the multi-unit NPP accident analysis. Since the multi-unit level 3 PSA is being currently performed assuming that all the NPPs are located at the same position such as a center of mass (COM) or base NPP position, radionuclide concentrations or exposure doses near NPPs can be drastically distorted depending on the locations, multi-unit NPP alignment, and the wind direction. In order to overcome this disadvantage of the COM method, the idea of a new multiple location (ML) method was proposed and implemented into a new tool MURCC (multi-unit radiological consequence calculator). Furthermore, the MURCC code was further improved for the multi-unit level 3 PSA that has the arbitrary number of multi-unit NPPs. The objectives of this study are to (1) qualitatively and quantitatively compare COM and ML methods, and (2) demonstrate the strength and efficiency of the ML method. The strength of the ML method was demonstrated by the applications to the multi-unit long-term station blackout (LTSBO) accidents at the four-unit Vogtle NPPs. Thus, it is strongly recommended that this ML method be employed for the offsite consequence analysis of the multi-unit NPP accidents.

• 융합신호처리학회논문지
• /
• 제11권4호
• /
• pp.345-353
• /
• 2010

로봇의 대역적 자기 위치 추정에서의 일반적인 해법은 로봇의 자기 위치에서 복수의 가설을 생성하고, 관측된 랜드마크의 특정을 기초로 각 가설을 평가하여 가장 확실한 자기 위치를 구하는 것이다. 기존의 대표적인 방법인 ML이나 MCL은 랜드마크의 특징과 생성된 가설의 모든 조합을 평가하는 방법으로서 충분한 계산 자원에서는 최적의 방법이라 할 수 있다. 그러나, 일반적으로 계산량은 평가할 조합의 수에 비례하므로 다수의 조합이 존재하는 넓은 환경에서는 이러한 방법은 계산량이 아주 많아진다. 이러한 문제를 해결하기 위하여 본 논문에서는 확실하고 유망한 조합을 우선적으로 선택 평가하는 것으로, 계산시간을 효율적으로 이용하는 새로운 추정방법을 제안한다. 그 기본이 되는 방법으로는 RANSAC 알고리즘과 RANSAC 알고리즘의 효율화 방법인 Preemption scheme을 이용한다. 제안된 방법은 로봇이 관측할 때마다 계산량을 일정치 이하로 억제할 수가 있고, 또한 검증 실험에서 높은 추정 성능을 확인할 수 있었다.