Kusuma, Yanuar Aji (2019) Pemodelan H2S Removal Unit Berbasis Jaringan Syaraf Tiruan di PT Saka Indonesia Pangkah Limited. Other thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
02311540000044 - Undergraduate_Theses.pdf Download (4MB) | Preview |
Abstract
Pada pemodelan H2S removal unit dipergunakan Jaringan Syarif Tiruan (JST) dikarenakan tidak diketahui komposisi masukan sour gas pada plant sehingga pemodelan dengan mengggunakan persamaan kimia tidak memungkinkan. Arsiktektur feed-forward dengan algoritma pelatihan Levenberg - Marquardt dipergunakan dalam memodelkan sistem di H2S removal unit, dengan jumlah hidden layer 1. Pada input layer terdapat 4 node untuk mengetahui dinamika plant tersebut, yaitu laju aliran sour gas, temperature sour gas, laju amine, temperature amine, dan perbedaan tekanan antara masukan sour gas dan amine. Sedangkan pada output layer ada 1 output node yaitu ppm H2S pada sweet gas. Metode trial dan error digunakan dalam menentukan model terbaik, dengan menggunakan variasi jumlah hidden node 1-10 pada hidden layer. Parameter yang digunakan adalah nilai RMSE terkecil yang dihasilkan oleh model JST. Simulasi yang dilakukan , model JST terbaik didapatkan dengan struktur 4-7-1. Nilai RMSE yang dihasilkan pada struktur bernilai 0,4947, dengan nilai RMSE 0,4947 menunjukan bahwa model memiliki persentase error sebesar 9,95 %
=================================================================================================================================
In H2S removal modeling, Artificial Neural Networks (ANN) are used because the composition of sour gas input is unknown in the plant so modeling using chemical reactions is not possible. Feed-forward architecture with the Levenberg - Marquardt training algorithm is used in designing the dynamics model of the plant process, with the number of hidden layers 1. In the input layer there are 4 nodes to find out the plant dynamics, namely sour gas flow rate, temperature sour gas, amine rate, temperature amine, and the pressure difference between input sour gas and amine. While at the output layer there is 1 node output, ppm H2S in sweet gas. The trial and error method is used in determining the best model, using variations in the number of hidden nodes 1-10 in the hidden layer. The parameter used is the smallest RMSE value generated by ANN models. The simulation is done, the best ANN model is obtained with a 4-7-1 structure. The RMSE value generated in the structure is 0.4947, with RMSE 0.4947 indicating that the model has an error percentage of 9.95%.
| Item Type: | Thesis (Other) |
|---|---|
| Additional Information: | RSF 006.32 Kus p-1 2019 |
| Uncontrolled Keywords: | H2S Removal, Jaringan Syaraf Tiruan, Levenberg – Marquardt, RMSE |
| Subjects: | T Technology > TP Chemical technology > TP692.5 Oil and gasoline handling and storage |
| Divisions: | Faculty of Industrial Technology > Physics Engineering > 30201-(S1) Undergraduate Thesis |
| Depositing User: | Yanuar Aji Kusuma |
| Date Deposited: | 15 May 2024 04:08 |
| Last Modified: | 15 May 2024 04:08 |
| URI: | http://repository.its.ac.id/id/eprint/67717 |
Actions (login required)
 |
View Item |