科技廠房之結構損傷探測分析

Abstract

本研究旨在發展科技廠房的損傷探測技術，以一座典型之疊層式科技廠為對象進行分析，探討這項技術實際應用之可行性。本文採用的結構損傷探測定位分析方法，係結合確定-隨機子空間系統識別法與狀態空間損傷定位向量法，以結構之地震反應歷時資料為依據進行分析。損傷探測分析考慮廠房受損桿件約佔該樓層柱量的0.5%或1%之情況。系統識別分析結果顯示，無論在全維觀測或縮維觀測條件下，各模態頻率及阻尼比均能被準確辨識，可供結構損傷探測分析之用，其中以縮維觀測模式之識別分析結果為基礎進行的損傷探測分析有較佳之辨識率。在監測方位之選擇上，以擷取廠房角隅處之動態反應較能準確判定受損樓層。此外，因廠房為三維空間結構，在雙向地震作用下，過去針對平面剪力屋架所發展的應力指標並不適用，本文建議採用多種應力指標組合方式計算以輔助判斷。分析結果顯示，能夠完全準確判定受損樓層的機率約為四成，若考慮將受損樓層成功識別出來，但也同時將某一未受損樓層誤判的情況合計，則成功機率可達六成以上，驗證本文發展的結構損傷探測技術在科技廠房之應用頗有潛力。惟目前的評估模式只能指出結構之受損樓層，尚無法定位出該樓層確切之受損構件所在方位，未來仍應進一步發展相關技術。

This study aims at developing a structural damage detection technology for high-tech manufacturing factory. A typical triple fab is considered as the object to explore the feasibility of this technology for practical application. In this dissertation, a scheme integrated with deterministic–stochastic subspace system identification and the method of damage localization vector (DLV) is proposed for damage detection of structures based on seismic response data. Damage conditions simulated by removing columns of about 0.5% or 1% of the whole damaged story have been considered. System identification analysis shows that both modal frequencies and damping ratios of the fab can be accurately identified, regardless of full- or reduced-order observation condition. The identified system parameters in turn serve as the basis for damage detection of the structure. The reduced-order system, however, gives better damage identification results. The structural monitoring system acquiring dynamic responses at the corners is found to be more reliable for damage localization. In addition, as the fab is 3-D in nature and the structure is under bi-directional earthquake inputs, the stress index developed earlier for shear-type plane buildings is no longer applicable. Instead, this study proposes various combinations of stress indices for auxiliary judgment on damage localization. The results indicate that a probability of success of about 40% can be achieved to accurately identify the damaged storey without any misjudgment. The percentage of success can be further increased to be over 50% if the damaged storey is identified with or without misjudging one other storey that is in fact not damaged. This suggests the potential of the proposed damage localization scheme for practical use in high-tech manufacturing factories. The current scheme, however, is able to locate the damaged storey but fail to locate the exact positions of the damaged structural members. A more advanced technique deserves further development next.