預力混凝土箱型橋梁長期撓度預測

Abstract

國內外皆有案例顯示長跨度預力混凝土箱形橋梁之長期撓度有超量問題，此疑慮會影響橋梁之安全性與使用性，故一直是工程界關注的重點與待解決的課題。因此，針對此議題本研究希望能發展預測長跨度預力混凝土箱形橋梁長期撓度之分析方法，建立全橋數值模型進行施工階段分析，並以監測資料與分析結果比較，以驗證此方法之正確性。 先前有研究顯示，潛變與收縮為長期撓度主要的來源，因此在橋梁分析中採用正確的潛變收縮材料模型將會對分析結果的準確性有決定性的影響。基於先前研究的驗證，認為本土化潛變收縮材料模型B4TW(2020)更貼近台灣本土的潛變收縮行為，本研究採用其與著名國外潛變收縮材料模型AASHTO(2007)與CEB-FIP(2010)進行比較，再次驗證B4TW(2020)是與台灣本土潛變收縮行為更為接近的材料模型。雖然beam-type模型在某些細部結構桿件的模擬上未能完全如實模擬，但本研究的目標為了解全橋結構行為，因此採用beam-type的3D有限元素分析建立數值模型。在了解全橋行為後，能針對有疑慮的細部桿件進行更細節的分析。過去的預力橋梁相關研究亦顯示，活載重與溫度效應不容忽視，但僅有研究提出溫度對長期撓度造成的大略趨勢，故本研究亦針對此二因素對長期撓度與斷面應力所造成的影響進行更詳細的分析與討論。準確、高可靠度的監測資料亦為本研究得以驗證模型正確性的一大重點，因此本文亦介紹了目前兩座橋梁所採用的現地監測系統。 本研究使用商用軟體模擬國內某兩座長跨度預力混凝土橋梁，分別是採用中央鉸接的預力箱型橋梁與中央為連續的預力箱型橋梁，亦代表台灣早期與現代的預力混凝土橋梁，且此兩座橋梁皆有完整的現地監測資料得以與分析結果進行比對驗證。採用兩座橋梁的目的是為了驗證本文所採用之分析分法是適用於不同邊界條件形式的橋梁。本研究依照設計圖與施工報告，模擬該二橋梁的幾何、邊界條件、各種載重以及生命週期中各項調整與補強。分析內容除了長期撓度外，本研究亦利用經撓度驗證後的模型，提出斷面應力和有效預力的分析結果，並探討活載重與溫度效應對於橋梁之長期撓度和斷面應力之影響。最終提出完工50年及完工100年後的長期撓度、斷面應力、有效預力預測分析結果。 由分析之長期撓度結果發現，模擬所採用之潛變收縮模型對橋梁長期撓度的影響甚鉅。在本研究所使用的三種潛變收縮模型中，台灣本土化潛變收縮模型B4TW的預測結果與截至目前為止的監測資料非常接近，在有鉸接的橋梁中平均誤差為24%、無鉸接的橋梁平均誤差約為12%，而其他著名的潛變收縮模型結果則皆有低估長期撓度之現象，誤差達到近80%，因此再次驗證B4TW(2020)能夠更精準地捕捉台灣在地的潛變收縮現象。斷面應力經分析，A橋梁在剛完工、完工50年及完工100年時部分結構桿件有些許應力超量，但目前現地監測顯示未有危險之疑慮，B橋梁則未有應力超量的情形出現。本研究分析A橋梁的車流量資料，使用一假定之活載重做為長期恆載，發現考量活載重對增加長期撓度預測的精準度有助益。本研究亦利用B橋梁有完整溫度監測資料的優勢，探討溫度效應對於撓度造成的影響，若將溫度所造成之撓度疊加，可見隨季節起伏之趨勢吻合，更能將長期撓度的平均誤差由12%降至5%。關於其震盪之幅度，約有90%的撓度震盪能控制在20%內。有效預力的分析中可以看到剛完工至完工50年間預力損失率達到10%，但完工50年至完工100年間的預力損失量僅有1%至5%，顯示預力損失至完工50年後已趨穩定。

Excessive long-term deflection has been frequently seen on prestressed concrete box girder bridges worldwide. This is a concern for the safety and serviceability of bridges. Thus the engineering industry attempts to research and resolve this problem. Against this issue, this study aims to develop an analysis method to predict the long-term deflection of prestressed concrete box girder bridges in Taiwan. Numerical models of the bridges were built to do construction stage analysis. To verify the reliability of the numerical models, the simulation results obtained from the numerical model were compared with in-situ monitoring data. According to previous research, the creep and shrinkage behavior of concrete accounts for a large proportion of long-term deflection. As a result, applying a precise creep and shrinkage model plays a decisive role in simulating the long-term deflection accurately. Being verified by earlier research, a local creep and shrinkage model, B4TW(2020), could capture creep and shrinkage behavior in Taiwan more precisely. This inference was confirmed in this study by comparing with the results applying AASHTO(2007) and CEB-FIP(2010), which are the most used creep and shrinkage models in the US and Europe. In the past research, beam-type numerical models and solid-type numerical models were both used. Although there might be difficulties simulating all of the geometrical details on the beam-type models, beam-type models were adopted in this research. The main reason is that this research focuses on the behavior of the whole bridge rather than elements. 3D models could provide further research after having a comprehensive understanding of the behavior of whole bridges. The effect of live loads and the effect of temperature are also two important variables to long-term deflection. However, there was only research related to the temperature effect with rough trends. In this research, the effects of live loads and temperature on deflection and stress were studied and discussed. Being able to compare the analysis results with in situ monitoring data is also a highlight of this research. To confirm the monitoring data was correct, the monitoring systems in situ were introduced in this research. Two domestic prestressed concrete box girder bridge models were chosen and were built by commercial software. Two bridges consist of two cantilevers connecting in the midspan. One of the bridges was connected by a horizontally sliding hinge, while another was continuously connected. These two types of prestressed concrete box girder bridges also represented the characteristic of the early period and modern period in Taiwan, respectively. Two bridges were chosen in order to prove that the analysis method adopted in this research is suitable for different types of prestressed concrete bridges. Moreover, these two bridges have been monitored since the completion. The thorough monitoring data provides us the evidence to verify the reliability of the models. According to the design layout and the construction record, geometry, boundary conditions, loading, and adjustment during the life cycle were simulated in the models in detail. First, simulation results of long-term deflection were proposed and confirmed with in situ monitoring data. Second, results of stresses, effective prestresses, effect of live load, and effect of temperature on long-term deflection and stresses were also discussed. Last, the results at 50 years and 100 years after construction were shown at the end of this study. The simulation results show that the adopted creep and shrinkage models have significant impact to the long-term deflection simulation results. Among three types of creep and shrinkage models, the local model, B4TW(2020), had the closest results compared to in situ monitoring data. The average error of the bridge with a hinge at midspan was about 23%, while the continuous bridge was about 12%. Creep and shrinkage models from foreign codes underestimated the long-term deflection with an error of 80%. In this research, B4TW(2020) is proved to be able to capture the creep and shrinkage behavior in Taiwan more precisely than others. As for stress, the bridge with a hinge at midspan had excessive amounts of stresses in some specific areas no matter at the completion of bridge construction, and 50 and 100 years after the completion. However, based on in situ observation, there are no danger concerns currently. On the other hand, no excessive amounts of stresses occurred on the bridge which is continuous. The live load traffic statistics of the bridge with a hinge at midspan were analyzed and an amount of live load was assumed as long-term dead load due to heavy traffic during the off-peak period. Considering live load does help to improve the accuracy of prediction of long-term deflection. The continuously connected bridge has thorough history data of temperature during the life cycle. The effect of temperature on deflection was analyzed and discussed in this study. By including the deflection due to temperature with long-term deflection, the simulation results fit the fluctuated trends along with seasons. Moreover, the average error of deflections could be reduced from 12% to 5%. Regarding the amplitude, 90% of the data points are within 20%. As for the analysis of effective prestress, there was 10% of prestress losses during completion to 50 years after construction. Nevertheless, there was only 1% to 5% of prestress losses during 50 years to 100 years after construction. It is proved that prestress losses become steady after 50 years since completion.