Identifikasi Kerusakan Berbasis Getaran Dengan Metode Machine Learning Pada Lambung Kapal Pelat Sandwich Dengan Core Polyurethane-Fiberglass

Ismail, Abdi (2021) Identifikasi Kerusakan Berbasis Getaran Dengan Metode Machine Learning Pada Lambung Kapal Pelat Sandwich Dengan Core Polyurethane-Fiberglass. Doctoral thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
04111860010001-Disertation.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2023.

Download (6MB) | Request a copy

Abstract

Konstruksi kapal memerlukan material yang tidak hanya kuat, tetapi juga ringan agar kapal mampu mengangkut muatan yang lebih banyak, sehingga operasional kapal menjadi lebih ekonomis. Lebih jauh, struktur ringan dapat mengurangi konsumsi bahan bakar kapal. Penerapan pelat sandwich pada konstruksi kapal dapat mengurangi berat konstruksi kapal secara keseluruhan 5-8%. Pelat sandwich yang telah komersial digunakan pada industri perkapalan dan telah memenuhi standart Llyod’s Register (LR) adalah Sandwich Plate System (SPS) dengan faceplate baja dan core dari polyurethane elastomer yang relatif mahal. Polyurethane elastomer (PU) dapat diperkuat dengan fiberglass agar diperoleh material core komposit PU-fiberglass yang lebih murah dan lebih kuat. Jadi, pelat sandwich yang lebih ringan dapat diterapkan untuk memenuhi nilai kekuatan yang sama. Penerapan pelat sandwich untuk konstruksi kapal terlebih pada pelat sisi lambung memiliki risiko terjadinya retak akibat kelelahan. Penerapan pelat sandwich pada lambung kapal memiliki tantangan dalam identifikasi kerusakan untuk mencegah kegagalan struktur yang lebih parah. Dengan demikian, pada penelitian ini dilakukan pengembangan material core komposit PU-fiberglass secara eksperimen agar dapat memenuhi regulasi LR. Pelat sandwich dengan material core tersebut diterapkan pada kompartemen kapal secara numerik untuk mengetahui pengurangan beratnya jika dibandingkan dengan pelat konvensional. Kemudian, identifikasi ukuran dan lokasi kerusakan berbasis getaran dilakukan menggunakan metode machine learning algoritma Decision Tree (DT) dan MSCS. Akurasi terbaik algoritma DT diperoleh sebesar 98.17% pada persentase data latih sebanyak 44%, karena dengan data latih 55% didapatkan akurasi lebih rendah yaitu sebesar 97.83%. Jadi, algoritma DT mampu memprediksi ukuran kerusakan dengan akurasi tertinggi sebesar 98.17%. Metode MSCS juga berhasil memprediksi lokasi terjadinya kerusakan dengan akurasi yang sangat baik pada lokasi A hingga E dengan akurasi sebesar 99.8% - 99.925%. ================================================================================================ Ship construction requires materials that are not only strong but also light so that the ship is able to carry more cargo, so that ship operations become more economical. Furthermore, the lightweight structure can reduce the ship's fuel consumption. The application of sandwich plates in ship construction can reduce the overall weight of ship construction by 5-8%. Sandwich plates that have been commercially used in the shipping industry and have met the Lloyd's Register (LR) standard are Sandwich Plate System (SPS) with steel faceplate and a core of polyurethane elastomer, which is relatively expensive. Polyurethane elastomer (PU) can be reinforced with fiberglass to obtain a cheaper and stronger PU-fiberglass composite core material. Thus, lighter sandwich plates can be applied to meet the same strength values. The application of sandwich plates for ship construction, especially on the hull side plate, has the risk of cracking due to fatigue. Applying sandwich plates on the ship's hull has a challenge in identifying the damage to prevent more severe structural failure. Thus, in this research, the experimental development of PU-fiberglass composite core material has been carried out to meet the LR regulations. The sandwich plate with the core material has been applied to the ship compartment numerically to determine the weight reduction compared to conventional plates. Then, vibration-based damage identification has been conducted using the machine learning algorithm Decision Tree (DT) and MSCS. The best accuracy of the DT algorithm has been obtained by 98.17% on the percentage of training data as much as 44%, because, with 55% training data, the accuracy is lower, which is 97.83%. So, the DT algorithm is able to predict the size of the damage with the highest accuracy of 98.17%. The MSCS method succeeded in predicting the location of the damage with very good accuracy at locations A to E with an accuracy of 99.8% - 99.925%.

Item Type: Thesis (Doctoral)
Uncontrolled Keywords: Sandwich plates, polyurethane-fiberglass composites, damage identification, decision tree, mode shape curvature square, Pelat sandwich, komposit polyurethane-fiberglass, identifikasi kerusakan, decision tree, mode shape curvature square
Subjects: T Technology > TA Engineering (General). Civil engineering (General) > TA347 Finite Element Method
T Technology > TA Engineering (General). Civil engineering (General) > TA355 Vibration.
T Technology > TA Engineering (General). Civil engineering (General) > TA418.16 Materials--Testing.
T Technology > TA Engineering (General). Civil engineering (General) > TA418.9 Composite materials. Laminated materials.
T Technology > TA Engineering (General). Civil engineering (General) > TA645 Structural analysis (Engineering)
V Naval Science > VM Naval architecture. Shipbuilding. Marine engineering > VM156 Naval architecture
V Naval Science > VM Naval architecture. Shipbuilding. Marine engineering > VM163 Hulls (Naval architecture)
Divisions: Faculty of Marine Technology (MARTECH) > Naval Architecture and Shipbuilding Engineering > 36001-(S3) PhD Thesis
Depositing User: Abdi Ismail
Date Deposited: 01 Sep 2021 14:52
Last Modified: 01 Sep 2021 14:52
URI: https://repository.its.ac.id/id/eprint/91444

Actions (login required)

View Item View Item