以例行暫態反應探討核能蒸汽產生器系統模式及故障偵測應用

Abstract

本論文針對一可信之核能蒸汽產生器暫態模式(Strohmayer模式)使用於例行長時間暫態模擬及故障偵測之應用，進行探討。Strohmayer模式在模擬長時間之核能蒸汽產生器水位變化暫態時，會發生積分誤差問題。為處理此問題，本論文提出以濾波器為基礎補償飼水流量並與實際電廠運轉1小時數據進行比較探討。模擬結果顯示此法針對某特定1小時之暫態，可減緩積分誤差問題；但仍需針對不同暫態情境去探討及驗證其可用性。本研究另基於降階估測器觀念，重塑Strohmayer模式。所提之重塑模式係在已知1個狀態變數下提出，經模擬探討發現在以均方誤差為模式準確性評估指標下，此模式估測成效與Strohmayer建議模式相當。在量測器故障偵測方面，此重塑模式與逐次機率比檢定法搭配，經模擬探討發現其反應器冷卻水系統冷端溫度量測儀器故障可察覺度，與Strohmayer建議模式與逐次機率比檢定法之搭配作法相較，可偵測更小之儀器漂移故障。然其影響因素需進一步研究，以設計更佳之線上運轉設備監視工具。

In this thesis, Strohmayer’s model as a credible Nuclear Steam Generator (NSG) transient model is investigated by routine long term transient simulation for application to sensor fault detection. During simulating the routine long term transient response of NSG water level, the Strohmayer’s model (SM) may suffer from the integration error problem. For dealing with this problem, the Filter Based Compensated Feed-Water flowrate (FBCFW) is proposed and investigated. The simulation results show it can alleviate the integration level error problem for the specific one-hour transient, but needs to be further investigated and verified for different routine transient scenarios. Then, the Strohmayer’s model is reformulated based on the concept of reduced order observer. The reformulated model given one known state (RFM-1) is investigated by simulations and it is found that the estimation performance for Reactor Coolant System (RCS) Cold leg temperature is as good as Strohmayer’s suggested model (SM-FWC) in the sense of mean squared error (MSE). For application to sensor fault detection, the proposed reformulated model with Sequential Probability Ratio Test (SM-FBCFW-RFM-1/SPRT) is investigated by simulations and it shows that the SPRT Fault Detectability (SPRT-FD) of SM-FBCFW-RFM-1/SPRT for RCS Cold leg temperature is improved 5 times compared to the Strohmayer’s suggested model with SPRT (SM-FWC/SPRT). The factors causing such an improvement will be further studied for sake of designing a better on-line monitoring (OLM) tool.