• 대한안전경영과학회지
• /
• 제9권5호
• /
• pp.1-5
• /
• 2007

We construct the procedure to predict safety accidents following Bayesian approach. We make a model that can utilize the data to predict other levels of accidents. An event tree model which is a frequently used graphical tool in describing accident initiation and escalation to more severe accident is transformed into an influence diagram model. Prior distributions for accident occurrence rate and probabilities to escalating to more severe accidents are assumed and likelihood of number of accidents in a given period of time is assessed. And then posterior distributions are obtained based on observed data. We also points out the advantages of the bayesian approach that estimates the whole distribution of accident rate over the classical point estimation.

• 한국원자력학회:학술대회논문집
• /
• 한국원자력학회 2005년도 추계학술발표회
• /
• pp.679-680
• /
• 2005

• Nuclear Engineering and Technology
• /
• 제49권2호
• /
• pp.434-441
• /
• 2017

Containment venting is one of several essential measures to protect the integrity of the final barrier of a nuclear reactor during severe accidents, by which the uncontrollable release of fission products can be avoided. The authors seek to develop an optimization approach to venting operations, from a simulation-based perspective, using an integrated severe accident code, THALES2/KICHE. The effectiveness of the containment-venting strategies needs to be verified via numerical simulations based on various settings of the venting conditions. The number of iterations, however, needs to be controlled to avoid cumbersome computational burden of integrated codes. Bayesian optimization is an efficient global optimization approach. By using a Gaussian process regression, a surrogate model of the "black-box" code is constructed. It can be updated simultaneously whenever new simulation results are acquired. With predictions via the surrogate model, upcoming locations of the most probable optimum can be revealed. The sampling procedure is adaptive. Compared with the case of pure random searches, the number of code queries is largely reduced for the optimum finding. One typical severe accident scenario of a boiling water reactor is chosen as an example. The research demonstrates the applicability of the Bayesian optimization approach to the design and establishment of containment-venting strategies during severe accidents.

• Nuclear Engineering and Technology
• /
• 제44권7호
• /
• pp.807-816
• /
• 2012

The Three Mile Island Unit 2 (TMI-2) accident has been studied extensively, as part of both post-accident technical assessment and follow-up computer code calculations. The models used in computer codes for severe accidents have improved significantly over the years due to better understanding. It was decided to reanalyze the severe accident scenario using current state of the art codes and methodologies. This reanalysis was adopted as a part of the joint standard problem exercise for the Atomic Energy Regulatory Board (AERB) - United States Regulatory Commission (USNRC) bilateral safety meet. The accident scenario was divided into four phases for analysis viz., Phase 1 covers from the accident initiation to the shutdown of the last Reactor Coolant Pumps (RCPs) (0 to 100 min), Phase 2 covers initial fuel heat up and core degradation (100 to 174 min), Phase 3 is the period of recovery of the core water level by operating the reactor coolant pump, and the core reheat that followed (174 to 200 min) and Phase 4 covers refilling of the core by high pressure injection (200 to 300 min). The base case analysis was carried out for all four phases. The majority of the predicted parameters are in good agreement with the observed data. However, some parameters have significant deviations compared to the observed data. These discrepancies have arisen from uncertainties in boundary conditions, such as makeup flow, flow during the RCP 2B transient (Phase 3), models used in the code, the adopted nodalisation schemes, etc. In view of this, uncertainty and sensitivity analyses are carried out using simulation based techniques. The paper deals with uncertainty and sensitivity analyses carried out for the first three phases of the accident scenario.

• 자동차안전학회지
• /
• 제6권1호
• /
• pp.22-26
• /
• 2014

Fatality of accidents on curved roads where rollover accidents are likely to take place was higher than that on straight roads. We ought to investigate factors affecting injury severity of occupant in a vehicle rollover accident. From January 2011 to December 2013, we collected data about rollover motor vehicle crash accident. We surveyed occupant's injury, vehicle type, safety devices, type of rollover accident and the number of turn in accident. Of the 132 subjects, 56.1% were males, 50.8% were drivers, 48.5% fastened seat belt, and air bag deployed in 12.1%. Among injuries sustained head, chest and abdomen were major sites of severe injury(Abbreviated injury scale>2). Seat belt use, rollover type, and the number of 1/4 turn were found to have significant positive correlations with Injury Severity Score. The regression analysis herein found significance in safety belt use and the number of 1/4 turn. Seat belt use was a significant factor affecting injury severe of occupant in rollover accident.

• Nuclear Engineering and Technology
• /
• 제41권5호
• /
• pp.561-574
• /
• 2009

First ex-vessel core catcher has been applied to the practical design of NPPs with VVER-1000 reactors built in China (Tyanvan) and India (Kudankulam) for severe accident management (SAM) and mitigation of SA consequences. The paper presents the concept and basic design of this crucible-type core catcher as well as an evaluation of its efficiency. The important role of oxidic sacrificial material is discussed. Insight into the behaviour of the molten pool, which forms in the catcher after core relocation from the reactor vessel, is provided. It is shown that heat loads on the water-cooled vessel walls are kept within acceptable limits and that the necessary margins for departure from nucleate boiling (DNB) and of vessel failure caused by thermo-mechanical stress are satisfactorily provided for.

• Nuclear Engineering and Technology
• /
• 제54권7호
• /
• pp.2690-2701
• /
• 2022

In this paper, we select important fission products for the estimation of the source term during a severe accident of a PWR. The selection is based on the numerical results obtained from depletion calculations for the typical PWR fuel via the in-house code named DEGETION (Depletion, Generation, and Transmutation of Isotopes on Nuclear Application), release fractions of the fission products derived from NUREG-1465, and effective dose conversion coefficients from ICRP 119. Then, for the selected fission products, we obtain the adjoint solutions of the Bateman equations for radioactive decay in order to determine the importance of precursors producing the aforementioned fission products via radioactive decay, which would provide insights into the assumption used in MACCS 2 for a level 3 PSA analysis in which up to six precursors are considered in the calculations of radioactive decays for the fission product after release from the reactor.

• Nuclear Engineering and Technology
• /
• 제55권9호
• /
• pp.3241-3251
• /
• 2023

The potential for recriticality and sub-critical boron concentrations is analyzed during the relocation of the fuel rods in the assembly, which we call late phase of a severe accident, via coupling between MELCOR and whole-core Monte Carlo analyses by Serpent 2. The recriticality, initiated during the early phase, is found to maintain when the fuel assemblies containing intact fuel rods are submerged by the cooling water. It is also found that the effect of the negative reactivity insertion via remaining fission products in the fuel debris increases as the burnup increases. The sub-critical boron concentrations during the late phase are found to be 76~544 ppm lower than those during the early phase. Therefore, it can be concluded that the boron concentration that prevents recriticality not only during the early phase but also during the late phase is the sub-critical boron concentration during the early phase.

• Nuclear Engineering and Technology
• /
• 제55권11호
• /
• pp.4146-4158
• /
• 2023

This paper proposes a combined plastic and creep constitutive model of A533B1 pressure vessel steel to simulate progressive deformation of nuclear pressure vessels under severe accident conditions. To develop the model, recent tensile test data covering a wide range of temperatures (from RT to 1,100 ℃) and strain rates (from 0.001%/s to 1.0%/s) was used. Comparison with experimental data confirms that the proposed combined plastic and creep model can well reflect effects of temperature and strain rate on tensile behaviour up to failure. In the companion paper (Part II), the proposed model will be used to simulate OECD lower head failure (OLHF) test data.

• Nuclear Engineering and Technology
• /
• 제56권6호
• /
• pp.2266-2273
• /
• 2024

A machine learning platform is proposed for the diagnosis of a severe accident progression in a nuclear power plant. To predict the key parameters for accident management including lost signals, a long short term memory (LSTM) network is proposed, where multiple accident scenarios are used for training. Training and test data were produced by MELCOR simulation of the Fukushima Daiichi Nuclear Power Plant (FDNPP) accident at unit 3. Feature variables were selected among plant parameters, where the importance ranking was determined by a recursive feature elimination technique using RandomForestRegressor. To answer the question of whether a reduced order ML model could predict the complex transient response, we performed a systematic sensitivity study for the choices of target variables, the combination of training and test data, the number of feature variables, and the number of neurons to evaluate the performance of the proposed ML platform. The number of sensitivity cases was chosen to guarantee a 95 % tolerance limit with a 95 % confidence level based on Wilks' formula to quantify the uncertainty of predictions. The results of investigations indicate that the proposed ML platform consistently predicts the target variable. The median and mean predictions were close to the true value.