中華民國內政部建築研究所中華民國內政部建築研究所

一、研究緣起
由於建築物風力特性受外型及結構振動特性影響甚鉅，因此規範內容不可能涵蓋所有種類的建築物型態。我國建築物耐風設計規範及解說之內容僅適用於近似矩形之建築物。其中，風力計算方法參照國外規範，以陣風反應因子法作為順風向風力的計算理論，橫風向及扭轉向風力的計算則採用經驗式，以擬合風洞試驗的結果獲得。由於風力計算公式涉及許多理論，且風力與地震力兩者有本質上的差異，故風力計算過程相較於建築物耐震設計規範來得更為繁複。目前依據規範所計算而得的風力仍舊有許多可改善的空間。其中，在依據規範公式計算獲得風力後，應該如何合理地組合這些風力，仍是一個急待解決的問題。根據目前作法，順、橫、扭轉向風力首先須分別獨立作用於結構模型上進行靜力分析，獲得了結構物的結構反應後再進行組合。接著取其組合後的最大反應者，對應此最大反應的來風方向的順、橫、扭轉向風力作為設計風力。實際上，目標結構物的結構反應可能因為結構系統之選擇而有所不同；而且三個方向的風力未必在同一時間發生最大值，故組合時理應存在機率問題。若能有效找到直接以風力進行組合的作法，除了能有利於降低過於保守的風載重外，亦能減少操作過程的繁複性。此外，為了簡化對於低矮建築物的繁複計算(大部分低矮建築物風力不控制)，我國規範參考他國規範建議，於規範第二章第十三節提供了針對低矮建築物的簡易風力計算式，來取代第二章前幾節的一般風力計算式。然而，由於簡易風力計算式代表較為保守的設計風力值，因此必然在高寬比門檻時發生不連續的設計風力。此外，簡易風力計算式亦未考量組合，因此須一併進行探討。
二、研究方法及過程
本研究採用之研究方法有二：(1)以傳統的風洞物理模擬方法執行計畫內容有關必要的風力或風壓數據之取得及後續探討；(2)以MATLAB程式語言撰寫建築物結構分析程式進行靜力分析及動力分析；並以MATLAB程式語言撰寫探討載重組合所需要的機率分析理論程式。結合以上兩方法以及本計畫所蒐集之重要文獻進行綜合討論，最後提出本計畫案各項預期目標的結論。
在風洞實驗方面，由於本計畫首要任務為探討高層建築物的載重組合，因此首先製作具代表性的高層建築物模型至少兩座作為風洞試驗模型，量取其表面風壓並分析其風壓及風力係數與各規範比較，確認此兩模型可用以作為後續分析所用。此兩模型與過去模型不同之處在於，此兩模型必須進行高寬比不同的風洞實驗，且分別為方柱與矩柱外型，足以提供可顯示出橫風向風力及扭轉向風力比例較大的案例模型為原則。本研究預備於內政部建築研究所的風雨風洞實驗室中進行風壓量測實驗。為了確保後續探討所需要的實驗數量足夠，以淡江大學風工程研究中心的第一號風洞實驗室作為預備。首先以皮托管量測邊界層高度風速，原理為利用量測內外管的壓力差，經熱線流速儀校正後可量測邊界層位置的動態風速作為參考風速壓。接著量測模型表面的動態分壓分布，並利用參考風速壓來計算風壓係數。對於高層建築物而言，        風速的改變並不會造成分離的現象改變，因此可以僅採用一合理縮尺風速即可；然而具有曲面的結構物則會因為雷諾數的影響而改變風力的估算。在本研究中，將盡量採用高風速以獲得符合實際的高雷諾數範圍。
在數值分析方面，由於研究團隊採用主持人所屬之教學單位採購之軟體進行程式撰寫，因此不需另行採購、亦無採用大陸軟硬體造成資料外洩之虞。以MATLAB撰寫本研究所需之程式有以下好處：(1)自建結構分析程式可用以進行各種靜力、動力分析，亦可自由建立使用者函數方便運算；(2)可結合主持人過去於多次委託計畫案所建立之程式庫，不須重新撰寫；(3)可用以訓練本計畫參與之研究人員的程式撰寫能力。MATLAB同時具有十分出色的圖表繪製功能，對於本計畫成果的展現很有助益。
在規範驗證方面，本計畫引用了各國最新版規範進行歸納整理，同時以實驗結果進行驗證，最終提出與現有規範比較的結果以及最佳的建議方案。
三、重要發現
從各國規範中可以分為兩類，第一類為美加規範類，其規範內容中並未規定橫風向風力與扭轉向風力，而是以風壓的方式來作整體力量的施加，所以並沒有所謂風力組合的問題。然而美加規範中定義了數種不同的風力估算方法，其中亦包含了以主要風向角為設計的流程，可以說雖然沒有提供風力組合形式，但其預測結果仍然準確。第二類則為日本ISO規範類，此類規範內容中明確定義了橫風向風力與扭轉向風力，因而也提供了此三種風力如何組合的問題。在組合的時候，多採用三種不同情境(順風向最大時、橫風向最大時、以及扭轉向最大時)，搭配不同深寬比及結構頻率等參數進行組合係數的計算。此類規範中以日本規範最為詳細，ISO規範則簡化之。我國規範則是採用SRSS的概念，將三不同風力均乘上各自的組合係數後為一合成風力，此合成力沒有所謂何者最大問題，然而也提供了一定的準確度。此外，研究團隊發現我國規範中第十節、第十一節與第十三節有互相矛盾的地方，亦即出現相同含意的公式，但其公式參數卻不同。應當盡快進行適當的修改。除此之外，我國簡易風力計算的方式仿效一般風力計算方式，同樣均有順、橫、扭三個風向。然而檢視日本規範及ISO規範可以發現，其簡易風力的估算方式僅為順風向風力乘上一放大倍數，代表涵蓋了橫風向風載重效應及扭轉向風載重效應的貢獻，較為簡單快速。
研究團隊針對規範修訂分為兩部分提出建議。首先是第十二節的部分，研究團隊提出三種不同方法以載重直接進行組合的方式，與動力分析結果進行比較。其結果雖然相較於日本規範及ISO規範採用結構效應組合的方式來得稍偏保守，但採用外力直接進行組合的方式較為快速且同樣具有與動力分析結果相同的變化趨勢。然而此三種方法中也有可以改善的空間，例如橫風向與扭轉向可以採用相關係數(Correlation coefficient)再進行精算，類似於日本規範及AIJ規範均有此考量。此外，研究團隊亦證明了目前規範所採用的SRSS概念方法實際上與日本規範及ISO規範有一定的準確度，與動力分析結果相比，確實不至於導致預測失真，雖然較為繁複但仍屬實用。研究團隊也納入了目前國內外實驗室常採用的聯合機率方式來進行估算，證實了此種方式在具有風洞實驗數據的狀況下，可以獲得最佳的準確度，然而相對地，其計算過程也最為繁複。以研究團隊的執行本計畫的預期目標來說，採用外力組合的程序的確可以具有一定的準確度且執行效率變快。
接著是第十三節簡易風力的修訂部分，本研究分成兩個部份給出建議。第一個建議是，研究團隊認為簡易風力可以改用放大順風向風力的風載重效應的方式來涵蓋橫風向及扭轉向風力的作法。此作法在日本規範及ISO規範均已揭示許久，執行起來又十分快速便利。目前我國規範中對於簡易風力的計算式仿效一般計算式仍然分成三個，但並沒有明確表示如何組合此三個簡易風力。在簡易風力計算式通常偏保守的前提下，倘若不進行折減，那將會對低矮建築物造成極大不合理的風力組合。第二個建議是，研究團隊認為，倘若不採納第一個建議，則至少需採納本研究提出的簡易扭轉向風力計算式。此修正的計算式結果較符合風洞實驗進行動力分析結果，在高寬比3的門檻上，也沒有所謂的過大斷層問題。對於規範使用者來說，會是一個比較合理且較小的風力計算式。最後研究團隊不免要重複提出的概念：我國規範第一版承襲美國規範架構而來，定義了順風向風力計算方式。而後於第二版中納入了日本規範的橫風向與扭轉向風力計算方式。如今最新版的美國規範實際上已經跟過去我國規範第一版所參考的版本內容相差甚多，架構也已完全不同。相互比較之下，反倒我國規範的架構與日本規範或者歐洲國家多採用的ISO規範較為相近。研究團隊認為，我國規範在下一版的修訂中，應首先探討整個架構的問題，然後再討論細部修訂，如此規範的修訂才能有條不絮，遵循穩定的發展。
在實務應用授權方面，研究團隊採用了PSERCB及SERCB兩套目前在土木工程界廣為使用的耐震初評及詳評系統作為範例說明，從各個角度探討耐風設計評估的系統開法可行性。實務應用授權要能成功執行，必須取決於國內市場。倘若國內市場普遍認為現有的耐風評估問題僅在計算繁複，由專家座談所蒐集的業界意見看來，則恐怕沒有系統開發的市場需求。研究團隊為了瞭解國內風洞實驗的需求量是否足以推廣開發介面系統(或其他機制)來提升業界工程師於耐風設計時的效率，初步調查了國內進行風洞實驗之案件量與總金額，得知從2011年至2020年期間，環境風場風洞實驗案及建築物的風載重風洞實驗案的粗估執行金額約為3億至3億5000萬元之間，且此數字不包含委託至國外(含大陸)的風洞實驗執行單位的金額，亦不包含採用CFD數值模擬方式來估算的案件。依照研究團隊的經驗，提升業界工程師於耐風設計的效率是必要的，但不見得是在介面系統發開上，而可能是在於風洞實驗相關的面向進行機制的研擬，例如建置公正的第三方審查機制以確保實驗報告的品質。
四、主要建議事項
建議一
重新檢視我國耐風設計規範整體架構：立即可行建議
主辦機關：內政部建築研究所
協辦機關：社團法人中華民國風工程學會
綜合研究團隊數年來執行內政部建築研究所計畫的經驗可歸納出：我國規範第一版承襲美國規範架構而來，定義了順風向風力計算方式。而後於第二版中納入了日本規範的橫風向與扭轉向風力計算方式。如今最新版的美國規範實際上已經跟過去我國規範第一版所參考的版本內容相差甚多，架構也已完全不同。相互比較之下，反倒我國規範的架構與日本規範或者歐洲國家多採用的ISO規範較為相近。研究團隊認為，我國規範在下一版的修訂中，應首先探討整個架構的問題，然後再討論細部修訂，如此規範的修訂才能有條不絮，遵循穩定的發展。
建議二
修改我國目前規範第十二節及第十三節內容：立即可行建議
主辦機關：內政部建築研究所
協辦機關：社團法人中華民國風工程學會
研究團隊建議可以參考日本規範或ISO規範的作法，並且採用本研究提出的三種方法其中一種作為風力組合的模式，簡化風力組合時的繁複過程。此外，研究團隊亦建議可以採用日本規範及ISO規範中等值橫風向風力的方式，以放大順風向風力效應的方法來估算低矮建築物的整體風力。
建議三
實際建築之實場監測計畫：中長期建議
主辦機關：內政部建築研究所
協辦機關：社團法人中華民國風工程學會
任何一個結構系統的風載重設計，均需要多方驗證來加以評估其方法論的準確性。因此研究團隊十分建議未來在經費許可的前提下，進行三年至五年的高層建築、低矮建築、甚或大跨度屋蓋結構的實場監測計畫，其中應該包含結構位移反應的量測和風速風向的量測等，方能夠提供給未來規範修訂者作為驗證的對照資料，甚或激發更多與風力導致結構物破壞的研究主題。

Background
Wind loads on buildings significantly depend on the building's appearance and its dynamic characteristics. Regulation or standard cannot cover the suggestive values for all kinds of structures. In Taiwan's code, it is clearly defined that the regulation content regarding the design wind load on the mainframe system and local component claddings is only applicable to those buildings with regular rectangular shapes. In design loads of the mainframe system, the along-wind load is estimated based on the gust response factor, the across-wind load and the torsional wind load are based on empirical models approximated from wind tunnel test data. Compared with the design earthquake loading, the design wind load is sensitive to the building appearance, and therefore inevitably, a more complicated loading estimation procedure is expected. The current Taiwan code for designing wind loads still requires effort to revise to a more sound status. Among those revision issues, reasonably combining along-wind, across-wind, and torsional wind loads is one of the essential issues to define. According to the current regulation, the along-wind, across-wind, and torsional wind loads are individually applied to the structural model to obtain the corresponding individual structural responses. By selecting the target structural response quantity, individual along-wind responses, across-wind response, and torsional response are combined based on the SRSS concept as a resultant response. Different wind attack angles may result in different resulting responses. The dominant one is the set of along-wind, across-wind, and torsional wind loads that produce the maximum resultant response. The designer can choose a different target structural response quantity so that the dominant set may be different.
Moreover, the maximum wind loads in the along-wind, across-wind, and torsional directions are not necessarily happening simultaneously. It would be reasonable to involve a joint probability relationship when combining them. The designers would benefit the designers if the combination rule were revised to solve the concerns mentioned above. Another issue would be the simplified estimation procedure of wind loads. It is found that the simplified estimation procedure. However, it provides a speedy way to give the initial design wind load; it is still confusing when the designer compares the common estimation procedure and the simplified estimation procedure in the case of critical buildings.
Research Methodology
This research adopts two approaches: (1) to obtain necessary wind force data of buildings based on conventional wind tunnel testing; and (2) to proceed with static analysis and dynamic analysis based on MATLAB-language programs for structural response estimation. Particularly in the second approach, probability-related parameters are to be studied through MATLAB-language programs too. Combining results from two approaches and the discussion from literature reviews, the expected achievements in this project are to be fulfilled. The following two paragraphs describe the details of the two approaches. In the conventional wind tunnel testing, at least two building models with different buildings shapes (depth/breadth ratio = 1.0 and 2.0) are to be made for surface pressure measurement. By integrating wind pressure with the representative area, the dynamic-wind force for the mainframe is obtained for further analysis. The two-building models are both made with an aspect ratio of 8 so that the model can be elevated to different heights for different aspect ratios from 1.5 to 8.0. And since the turntable can rotate the D/B = 2.0 building model, the case of the D/B = 0.5 model can be produced. This research will collect wind force data for the following three variable ranges: (1) D/B = 0.5, 1.0, 2.0; (2) H/B = 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 5.0, 6.0, 7.0, 8.0; and (3) wind attack angle θ = 0, 15, 30, 45, 60, 75, 90 degrees. However, it is still insufficient if compared to the code range. The wind force data here is mainly for validating load combination methodologies, not for the formulation of combination rules. Besides the above model cases, this research also applies surface measurement to the CAARC building model for a nearly true surrounding environment as a comparison set. The aerodynamic forces of this wind tunnel test are compared with the ideal cases. Details of conducting wind tunnel tests are presented in the chapter content later. In the structural analysis, MATLAB-language programming is adopted by the PI since it provides a powerful calculation in structural matrix analysis for static and dynamic analysis, and it is freely downloaded in PI’s university. By using MATLAB, all the necessary analyses in this research can be done. In comparing with international standards or regulations, this research has collected all the latest versions of the globally known international codes, such as ASCE, AIJ Recommendation, and ISO standards, and so on. At the end of this research, the PI expects to propose revised content based on these well-known codes for the next generation of Taiwan code.
Important findings
We can categorize all the latest versions of international codes into two types. The classical one of the first types is the ASCE code. In ASCE 7-16 code, there is no formulation for across-wind and torsional wind force estimations, so that there are no clear combination rules to combine them. The ASCE 7-16 code regulates how wind pressures are distributed over the surface of the buildings. The classical one of the second types is the AIJ Recommendations. The AIJ 2015 Recommendation provides the estimation formulas for across-wind and torsional wind loads. Further, the AIJ Recommendation offers three different scenarios for load combinations. The ISO 4354-2009 adopts a similar concept and includes simplified formulas for combination scenarios. Taiwan Code does not belong to any of these two types. In Taiwan, the load combination of three wind forces is carried out by the SRSS concept. When comparing different codes, the research team also found inconsistent text content in Chapter 2 – the simplified formulas are mentioned in Sections 10, 11, and 13; however, the procedures are not consistent. This should be corrected as soon as possible. Another worth noting point is, the simplified formulas in Taiwan Code are also for along-wind, across-wind, and torsional wind forces. However, there is no explanation about how to combine them. In AIJ or ISO codes, the simplified formulas are just based on the along-wind force estimation; a multiplication factor is added to enlarge the along-wind response as the overall response. This is truly a convenient method rather than the current one in Taiwan Code.
The research team has proposed two revision suggestions. The first suggestion is to suggest a new combination of rules for the along-wind, across-wind, and torsional wind forces. There are three approaches tested with the dynamic analysis results. Although the comparison results are more conservative, they are genuinely faster than the original way in the current code. The SRSS approach has also been verified to be precise enough and applicable. Finally, the research team would like to propose a procedure to replace the current one in Taiwan Code. The proposal is based on the same combination of rules in AIJ and ISO codes. The second suggestion in this research is to replace the current simplified procedure with the same one adopted in AIJ and ISO codes. If the existing code has remained, we can propose a new simplified formula for the torsional wind force to replace the current one. The new one will provide a better convergent result with the general formula for the torsional wind force. Besides the two suggestions from the research team, we also propose to reexamine the whole structure of the Taiwan Code, which was based on ASCE 7-88 and before versions. The ASCE codes have been revised quite differently, and the wind force characteristics in Taiwan are more like the wind force characteristics in Japan. It is believed to be a more reliable code if we can revise the code more broadly.
Regarding the authorized practical application of research results, the research team has adopted two examples, the PSERCB, and the SERCB systems, for discussion with experts from industrial fields. It is concluded that to successfully promote an evaluation system of design wind loads for the domestic engineers requires a bigger market in Taiwan. If the evaluation system is developed to simplify the calculation procedure in the existing code, the experts from the industry generally consider it does not meet the actual demands. To understand whether the evaluation system should be based on wind tunnel tests or not, the research team did a survey and found that the number of wind tunnel tests is enormous. Moreover, the industry requires more detailed information and higher quality insurance of wind tunnel tests than expected. The research team concluded that it is crucial to increase the efficiency to evaluate the design of wind loads of buildings. However, developing an interface is not a critical approach to achieving this goal. It may be more beneficial if we can establish a fair system to guarantee the quality of the wind tunnel tests.
Suggestions
Suggestion 1
Examining the structure of Taiwan Code for design wind loads: Immediate
Organizer: Architecture and Building Research Institute, Ministry of Interior
Co-organizer: Taiwan Association for Wind Engineering
From the research results of recent projects made by the PI, it is concluded that our first version of the Taiwan Code for designing wind loads was made based on ASCE specifications. The main contribution of it is to offer the evaluation methodology of the along-wind force. In the second version, the evaluation methodologies of across-wind and torsional forces from AIJ Recommendations were added to enrich its application range. However, the latest ASCE specification has been revised to a pretty different frame from before. More consistencies can be found between the Taiwan Code and the Japan Code. The PI suggests that it is necessary to reshape the basic skeleton of the Taiwan Code for the next version revision. After confirming the skeleton of the code, details can be modified or added with the latest regulations from overseas.
Suggestion 2
Modifying the content of Sections 12 and 13 of Chapter 2 in the current Taiwan Code: Immediate
Organizer: Architecture and Building Research Institute, Ministry of Interior
Co-organizer: Taiwan Association for Wind Engineering
According to the results of this project, the research team suggests replacing the current combination rules for three wind forces concerning the AIJ or ISO Codes in Section 12 of Chapter 2. The research team also suggests replacing the present formulas of three simplified wind forces concerning the more convenient way adopted in the AIJ or ISO Codes.
Suggestion 3
Field monitoring project of a true high-rise building: Long-term
Organizer: Architecture and Building Research Institute, Ministry of Interior
Co-organizer: Taiwan Association for Wind Engineering
Any proposed evaluation methodologies require the field data for validation. Therefore, the PI strongly suggests that in the future a field monitoring project of a high-rise building should be carried out to provide various data for all kinds of validations. The monitoring work should at least last for three to five years and it should cover high-rise buildings, low-rise buildings, and large span roof structures. From the observation of true data, more research topics may be found and new solutions may be derived.