精密儀器之石英砂隔振平台微振動特性研究

Abstract

隨著高科技廠房製程持續創新發展，或是學術研究精密儀器實驗室，其所使用的精密儀器，對於適合儀器運作的微振動需求也越來越嚴格。如何提供儀器合適運作環境的議題，也變得越來越重要。過去有許多相關的隔振設計，也曾經使用石英砂來進行隔振的設計，但是對於這樣隔振設施的特性，目前尚未有一個完整明確的研究，本研究即以石英砂隔振平台實際設計案例作進一步探討。隨著施工過程進行，到現地量測不同階段下隔振平台的微振動特性，並對量測到的微振動訊號進行系統識別，分別以單自由度動力模型(簡化模型)與現地實際幾何配置建置的有限元素分析模型(複雜模型)描述整個隔振平台的動力反應。 本研究標的「微振動隔振平台」設於建築物最底樓(地下二樓)，在預計放置精密儀器處下方，為一個可填隔振用石英砂的獨立基坑，該基坑係以樁基礎形式支撐與建築物基礎分開。基坑裡面分別填充兩種粒徑分佈不同的石英砂作為隔振材，並在砂體上放置一個1 m厚的混凝土蓋板作為放置儀器之基座。為了避免來自建築物的振動源因為混凝土蓋板傳入基坑內，因此在混凝土蓋板與四周筏基部分留有5 cm的空隙，此即為石英砂隔振平台之幾何配置。 本研究分成兩個部分進行，其中第一個部分為現地量測資料的蒐集與分析，第二部分為數值模型模擬與預測。 第一部分本訊號系統識別的部分可以分為三個流程，分別是一、對訊號進行降噪；二、從隨機穩態的訊號中萃取其自然衰減振動反應歷時；三、針對自然衰減振動反應作系統識別，以找出其自然頻率與阻尼比。本研究使用的方法分別為總體經驗模態分解法(Ensemble Empirical Mode Decomposition, EEMD)、隨機遞減法(Random Decrement Method, RDM)、亞伯拉罕時間域法(Ibrahim Time Domain, ITD)。 第二部分數值分析，使用ABAQUS有限元素分析軟體，根據現地幾何配置，簡化模型以二維平面應變的模式分析。就標準砂材料性質取得來源，分成兩個分析模式。其一是以過去文獻提及標準砂剪力模數經驗公式，得出標準砂參數進行分析。也因為此材料性質取得無需依賴現地微振動訊號測量結果，故在隔振平台設計階段可作為隔振設施動力特性的預測模型。另一方面，使用單向度波傳理論找出代表砂體的剪力波速進行數值分析。因為此參數須在隔振平台已完成的情況下進行量測，所以為透過現地訊號系統識別反推材料性質之修正模型。 識別結果發現此含石英砂隔振平台屬於被動隔振結構物，為單自由度動力系統。根據數值模型分析的結果與現地測量值比較發現，基坑垂直方向以單向度波傳理論所推出的材料性質會比較符合現地量測值；水平方向以使用共振柱試驗所推導的經驗公式會比較符合現地量測值。數值模型大致而言，在模型主頻的部分會比現地實際測量值要小，在振幅的預測上會比實際測量到的結果要大，結構物實際行為比模型分析的結果要佳。

With the development of hi-tech facilities and academic laboratories with precision instruments, the issue of how to provide a suitable vibration environment for equipment became more and more important. The design of isolators by using quartz containers has been used in the past, but this type of isolator has not been studied thoroughly. In this thesis we research a practical isolator using a quartz container and measure the in suit micro vibration signals of the isolator under different stages of construction. In order to accurately model the problem we utilize system identification to find the structure’s natural frequency and damping ratio, then use a single degree of freedom dynamic model (simplified model) and an in suit geometry finite element model (complex model) to describe the entire dynamic response of the isolator. The quartz container is located at the bottom of the building, and is supported on four piles. Basically, the quartz container and building are separated. The container is filled with two layers of different standard sand which are all specified in ASTM C778. The first layer (graded sand) is 3 m in depth, and the second layer (single particle sand) is 1 m in depth. Finally, we placed a 1 m depth concrete slab on top of the sand. To prevent the ambient vibration of the building transferring to the container through the concrete slab, we keep a 5 cm gap around the concrete slab.

Three steps of signal processing and system identification are applied in this study. In the first step, we denoise our measured in suit signals. In the second step, we get the free vibration natural decline curves from the stationary random data, and finally, we identify the natural frequency and damping ratio of the quartz container. The methods we use in the research are Ensemble Empirical Mode Decomposition (EEMD), random decrement method (RDM), and Ibrahim Time Domain (ITD) respectively. We use the finite element software ABAQUS to simulate the dynamic behaviors of the quartz container. Because there are two ways to get the material properties of the standard sand, we have two different models. In the first model, we can decide the sand’s properties through empirical results from the resonant column tests. Because the process of obtaining the sand properties does not rely on the in suit measurements, this model can be regarded as forecast model in the design stage. In the second model, we can get sand properties through the results of system identification by using one-dimensional wave propagation theory. This second model is regarded as modified model to be close to the in suit conditions. The quartz container is the structure of passive control, and mainly behavior as a single degree of freedom dynamic system. According to the comparisons between measurements from the in suit and results from the numerical simulation, the results of the model based on the one-dimensional wave propagation theory are close to in suit measurements in the vertical direction; and the results of model based on empirical formulas are close to in suit measurements in the horizontal direction.