Electronic Theses and Dissertation

Indonesia menjadi salah satu negara dengan jumlah gunung api terbanyak di dunia. Letusan gunung api dapat menyebabkan kerusakan infrastruktur dan korban jiwa. Kerusakan yang timbul dari gunung api disebabkan beberapa faktor yaitu semburan abu vulkanik (tephra), aliran lahar dan lava. Lontaran tephra dapat menyebabkan
keruntuhan atap akibat pembebanan tambahan. Penelitian terkait beban tephra dari letusan gunung api masih terbatas terutama terkait dengan penilaian tingkat kerentanan bangunan. Sehingga, penelitian ini bertujuan menghasilkan kurva
kerentanan struktur rangka atap double fink terhadap beban tephra menggunakan machine learning. Kurva kerentanan menghasilkan hubungan intensitas beban tephra terhadap probabilitas kerusakan struktur atap double fink. Penelitian ini
menerapkan studi kasus struktur rangka atap baja ringan profil C75 dengan tipe double fink yang mana paling efektif dan umum digunakan pada bangunan tempat tinggal masyarakat. Pemodelan dan analisis rangka atap dimodelkan dengan variasi sudut kemiringan 15˚ - 45˚, panjang bentang 6 m – 7,5 m dan intensitas beban tephra
0,1 kPa – 1 kPa disimulasikan secara otomatis dengan Open Application Programming Interface (OAPI) SAP2000 menghasilkan 280 data. Kurva kerentanan dihasilkan dengan meningkatkan dan melatih data hasil analisis menjadi
25.000 data digunakan untuk memprediksi kegagalan struktur rangka atap double fink menggunakan machine learning berupa K-NN (K-Nearest Neighbor) dan Naïve Bayes. Penilaian model menggunakan machine learning mendapat akurasi
yang tinggi yaitu K-NN 96,8% dan Naïve Bayes 93,5%. Masing-masing kurva kerentanan dibuat dan dibandingkan dengan metode regresi MLE (Maximum Likelihood Estimation) serta Least Square. Hasil analisis data dengan Naïve Bayes
dan K-Nearest Neighbor rangka kuda-kuda double fink mampu menahan beban tephra maksimum sebesar 0,64 kPa dan 0,92 kPa.

Indonesia is one of the countries with the large number of volcanoes in the world. Volcanic eruptions can cause damage to infrastructure and loss of life. Damage arising from volcanoes is caused by several factors, such as volcanic ash (tephra), lahar and lava flows. Tephra loading can cause roof collapse due to additional loading. Research related to tephra loads from volcanic eruptions is still limited, especially related to the assessment of building vulnerability levels. This research aims to estimate a vulnerability curve of the double fink roof truss structures against tephra loads using machine learning. The vulnerability curve Fdetermine the correlation between tephra load intensity and the probability of damage to the double fink roof structures. This research involve a case study of a C75 cold formed steel truss structure with double fink type which is the most effective and commonly used in residential buildings. Modeling and analysis of roof trusses modeled with variations in tilt angles of 15˚ - 45˚, roof spans of 6 m – 7,5 m and tephra load intensities of 0,1 kPa - 1 kPa is simulated automatically with the Open Application Programming Interface (OAPI) of SAP2000 resulting in 280 data. The vulnerability curve generated by increasing and training the analyzed data to 25.000 data is used to predict the failure of the double fink roof truss structure using machine learning K-NN (K-Nearest Neighbor) and Naïve Bayes. The model assessment using machine learning got high accuracy, which is K-NN 96,8% and Naïve Bayes 93,5%. Each vulnerability curve is created and compared with the MLE (Maximum Likelihood Estimation) and Least Square regression methods. The results of data analysis with Naïve Bayes and K-Nearest Neighbor, the double fink truss is able to withstand a maximum tephra load of 0,64 kPa and 0,92 kPa.