在氣象資料部分,本研究以台北測站為例,整理出1961年至2015年間所有日報表風速樣本資料,其中日報表為每一小時一筆平均十分鐘風速樣本資料,並使用獨立颱風法整理出颱風風速樣本,依據風洞實驗風向角進行分類,每15度角為一個風向分類整理出各風向角下各個颱風前三大風速,並將前三大風速轉換至建築物高設計風速,整理出考慮風向性與考慮颱風強度的極值風速樣本,再將颱風獨立事件機率轉換至年發生機率。 在風洞試驗部分,本研究以氣動力風壓試驗進行高層建築物模型表面風壓長時間量測。以我國建築物耐風設計規範及解說中的標準鄉鎮型地況作為模擬大氣紊流邊界層的實驗流場,並以CAARC大樓標準模型作為量測目標,以每15度風向角為間隔進行約實場六十小時的長時間風壓量測。依照傳統方法計算風壓係數歷時,並將歷時依實場十分鐘分段,計算出每風攻角下的極值風壓係數樣本。 在理論建構部分,本研究依據Cook and Mayne[1](1980) 使用聯合機率計算出設計風載重的範例,延伸出兩種研究方法,分別為分風向、全風向,加上常使用的簡易方法。探討在各風向角不同年超越機率與廣義分布和甘保分布下各方法之設計風載重。其中,分風向為考慮風向性極值風速與考慮風向性極值風壓係數,並對各個風向角聯合機率計算出設計風載重。全風向為不考慮風向性極值風速與考慮風向性極值風壓係數,並對各個風向角聯合機率計算出設計風載重。簡易方法為考慮風向性極值風速與考慮風向性極值風壓係數,其中極值風壓係數非超越機率為78%,並對各個風向角計算出設計風載重。為了比較出三種方法中何者最接近真實情形,本研究採取颱風事件期間所有樣本值並在考慮風向性下,作為設計風載重之對照組,予以比對三種方法差距。對照組為各颱風風速樣本配合極值風壓係數最佳非超越機率,找出各風向下最大設計風載重或最小設計風載重,繪製累積機率分布找出符合年超越機率之設計風載重。將Cook and Mayne[1](1980),延伸出的兩種研究方法加入颱風持續強度的考量,計算出設計風載重,比對有無考慮颱風持續強度差異。 結果顯示考慮風向性在四種方法中,年超越機率從0.02降低至0.001,三種研究方法加上對照組的設計風載重會提升兩到三倍。對於年超越機率0.001時,廣義分布的設計風載重擾動值會比甘保分布大,最大設計風載重差異較多,年超越機率0.02時,甘保分布與廣義分布較為接近。與對照組比較時,分風向結果最好,全風向則會保守許多。考慮颱風強度在兩種方法中,設計風載重都會略大,不至於影響與對照組比較結果。除了年超越機率0.001配合廣義分布,考慮颱風強度後影響最多。
As for the meteorological data, the research takes Taipei Meteorological Station as an example. First, this research organize the information of wind speed samples from the daily report from 1961 to 2015 and the data of typhoon wind speed samples. Based on the orientation in the wind-tunnel experiment, this research organize the top three wind speeds of each typhoon under each orientation. In the case of the wind-tunnel experiment, this research take a wind pressure test to conduct a long-term surface wind pressure measurement of a scaled CAARC building.Based on the standard rural area conditions listed in the Wind Resistance Design Specifications and Commentary of Buildings, this research simulate the experimental flow field of atmospheric boundary layer (ABL), and take CAARC high rise building model as the measurement target. As regards theory construction, the study calculates the design of the wind load by using joint probability on the basis of Cook and Mayne[1] (1980), and this research reach out two research methods, which is wind direction and full wind direction. With these two methods as well as simple methods, the study discusses the wind load design under different yearly exceedance probability, generalized extreme value distribution, and gumbel distribution. To compare the differences among the methods, the research set up a control group of wind load by adopting all sample values during typhoons under consideration of wind direction. Results showed that the yearly exceedance probability decreased from 0.02 to 0.001.Three research methods and the control group of design wind load increase up to two to three times. With regard to yearly exceedance probability at 0.001, the disturbance value of design wind load in generalized distribution would be higher than gumbel distribution, and also shows more differences in the largest design wind load. In addition, when the yearly exceedance probability goes to 0.02, gumbel distribution would be closer to generalized distribution. Compared to the control group, the results from wind direction would be better, full wind direction tends to conservative value.