核能電廠新一代機率式耐震風險評估方法之研究

Abstract

Huang 等人於2012 年發表了一核能電廠機率式耐震風險評估(Seismic Probabilistic Risk Assessment, SPRA)之方法，該評估法改進了業界習用之方法， 而a)以結構反應參數定義結構與非結構元件之易損性曲線，b)採用歷時分析來決 定核電廠元件之耐震需求，及c)以蒙地卡羅法決定各元件之破壞狀況。在Huang 等人提出之SPRA 方法中，需依據核電廠所在廠址之危害度分析決定出8 個由小 至大之地震強度等級，據以進行非線性動力分析。然而8 個強度等級是否足夠或 過多，卻未曾進行評估。再者，蒙地卡羅試驗之次數的決定方式及結構元件與土 壤性質之不確定性如何考慮於評估流程中，亦未曾探討。 本研究以上述之SPRA 方法(簡稱Huang et al. SPRA)，進行一案例電廠的地 震風險評估，將該電廠原有以PGA 定義之易損性曲線轉換至以結構反應定義之 易損性曲線，以估算該電廠兩重要之事故序列所貢獻之爐心受損年平均超越頻率。 此外，並進行一系列參數分析，改進上述Huang et al. SPRA 方法所存在問題。 本研究分析結果得知，1)將地震危害度曲線分為8 段，所得風險誤差約為10%； 2) 爐心受損之年平均超越頻率值之離散性與爐心受損次數成一對數線性關係， 可依據此關係求得蒙地卡羅試驗所需之次數。

Seismic probabilistic risk assessment (SPRA) has been widely used to compute the core damage frequency of a nuclear power plant (NPP) and typically involves the use of component fragility curves defined as a function of ground-motion parameters, such as peak ground acceleration and spectral acceleration. In this study, seismic risk of a sample NPP is computed using a SPRA methodology proposed by Huang, Whittaker and Luco in 2011, where the component fragility curves are defined as a function of structural-response parameters, such as floor spectral acceleration. The SPRA methodology of Huang et al. includes five steps. Step 1 of the methodology performs plant system analysis to determine accident sequences that could contribute to the target unacceptable performance (such as core damage or release of radiation) and develops component fragility curves as a function of a structural response parameter. Step 2 develops the seismic hazard curve(s) for the NPP site and selects and scales ground motions for each intensity level. Step 3 identifies the distributions and correlation of all structural response parameters of Step 1 using nonlinear response-history analysis at each intensity level. Step 4 uses Monte-Carlo-based procedures to generate a significant number of response data that are statistically consistent with those of Step 3 and to assess the possible distribution viii of damage to structural and nonstructural components of the NPP for each set of simulations. Step 5 computes the probabilities of unacceptable performance at each intensity level and the annual frequency of unacceptable performance of the NPP subjected to the seismic hazard of Step 2. In the study presented in this thesis, the seismic risk is defined as the core damage frequency contributed by two selected accident sequences of the sample NPP. A series of parametric analyses were conducted to study the impact of the following three items on the seismic risk of the sample NPP: 1) the number of the intensity levels of Step 2, 2) the number of trials used in the Monte-Carlo-based procedures of Step 4, and 3) the procedures used to estimate the failure probabilities of safety systems in the sample NPP. Based on the analysis results, recommendations were developed for the implementation of the SPRA methodology of Huang et al.