Fatikasari, Nada (2024) Perancangan Sistem Kontrol Top Temperature pada Amine Regeneration Plant Menggunakan Model Predictive Control (MPC) Berbasis Neural Network (NN) di PT BADAK NGL. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5022201202-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5022201202-Undergraduate_Thesis.pdf Restricted to Repository staff only until 1 October 2026. Download (10MB) | Request a copy |
Abstract
Proses pemurnian Liquified Natural Gas (LNG) memanfaatkan larutan amine sebagai pengikat gas pengotor berupa karbon dioksida (CO2). Reaksi kimia antara keduanya menghasilkan rich amine yang selanjutnya harus dimurnikan kembali agar dapat digunakan terus-menerus. Pada amine regeneration plant 1C-5, jumlah steam sebagai sumber panas pemurnian dari reboiler yang disalurkan oleh FV-5A/B/C/D tidak mampu mengimbangi perubahan laju aliran feed rich amine yang dinamis akibat respon bukaan valve lamban. Perancangan sistem kontrol cascade antara kontroler temperatur TIC-191 dan laju aliran FIC- 5A/B/C/D dibutuhkan untuk mempercepat respon flow valve. Respon sistem kontrol temperatur dengan I-PD eksis menunjukkan overshoot dan settling-time yang panjang sehingga proses pemurnian rich amine tidak optimal. Pada penelitian ini, telah dirancang dan dianalisis perbandingan performansi kontroler I-PD, NN, MPC, dan NN-MPC. Kontroler NN dirancang dapat mengendalikan temperatur menggunakan input berupa eror proses dengan mengatur learning rate dan alpha hingga diperoleh respon yang diinginkan. Model terbaik dari kontroler NN saat lr 0.1 dan alpha 1 dengan settling time 980.422 detik. Kontroler MPC dirancang berdasarkan model augmented dari plant, parameter MPC berupa Np memengaruhi kestabilan sistem sedangkan Nc memengaruhi settling time. Respon sistem terbaik dari kontroler MPC pada Np 40 dan Nc 35 dengan settling time 1118.741 detik. Kontroler NN- MPC dirancang dengan mengumpankan output NN sebagai sinyal input kontroler MPC untuk mengatasi undershoot dan mengurangi settling time dengan memanfaatkan Nc. Model terbaik pada sistem diperoleh menggunakan kontroler NN-MPC dengan settling time 798.667 detik.
============================================================
The Liquified Natural Gas (LNG) purification process utilizes an amine solution as a binder for carbon dioxide (CO2) impurity gas. The chemical reaction between amine and CO2 produces rich amine which must then be purified again so that it can be used continuously. In amine regeneration plant 1C-5, the amount of steam as a purification heat source from the reboiler supplied by FV-5A/B/C/D is not able to keep up with the dynamic changes in rich amine feed flow rate due to slow valve opening response. The design of a cascade control system between TIC-191 temperature controller and FIC-5A/B/C/D flow rate is needed to speed up flow valve response. The response of the temperature control system with the existing I-PD shows overshoot and long settling-time so that rich amine purification process is not optimal. In this research, I-PD, NN, MPC, and NN-MPC controllers have been designed and analyzed for performance comparison. The NN controller is designed to control temperature using input in the form of process error by adjusting the learning rate and alpha until the desired response is obtained. The best model of the NN controller when lr 0.1 and alpha 1 with settling time 980.422 seconds. MPC controller is designed based on the augmented model of the plant, MPC parameters in the form of Np affect the stability of the system while Nc affects the settling time of the response. The best system response of the MPC controller is at Np 40 and Nc 35 with a settling time of 1118.741 seconds. The NN- MPC controller is designed by feeding the NN output as the input signal of the MPC controller to overcome undershoot and reduce settling time by utilizing Nc. The best model of the system is obtained using the NN-MPC controller with a settling time of 798.667 seconds.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Amine Regeneration Plant, Kontrol Temperatur, Model Predictive Control, Neural Network, Amine Regeneration Plant, Model Predictive Control, Neural Network, Temperature Control |
| Subjects: | T Technology > TK Electrical engineering. Electronics Nuclear engineering T Technology > TS Manufactures > TS155 Production control. Production planning. Production management |
| Divisions: | Faculty of Electrical Technology > Electrical Engineering > 20201-(S1) Undergraduate Thesis |
| Depositing User: | Nada Shakirah Fatikasari |
| Date Deposited: | 31 Jul 2024 08:05 |
| Last Modified: | 31 Jul 2024 08:06 |
| URI: | http://repository.its.ac.id/id/eprint/111051 |
Actions (login required)
 |
View Item |