Maulana, Ahmad Zidane (2026) Rancang Bangun Sistem Artificial Intelligence Berbasis Knowledge Graph, Graph Retrieval Augmented Generation untuk Perancangan Tekno-Ekonomi Separator Horizontal serta Instrumentasi Kontrol. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5009221042-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5009221042-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy |
Abstract
erancangan separator horizontal dua fasa dan tiga fasa merupakan tahapan penting dalam fasilitas produksi minyak dan gas yang menuntut penerapan prinsip fisika, kaidah desain, dan pengalaman rekayasa. Pendekatan konvensional umumnya mengandalkan perhitungan iteratif, rule of thumb, dan validasi manual, sehingga membutuhkan waktu lama dan berpotensi menimbulkan inkonsistensi antarperancang. Penelitian ini mengembangkan sistem Artificial Intelligence yang mengintegrasikan knowledge graph, Graph Retrieval-Augmented Generation (Graph-RAG), dan Large Language Model (LLM) untuk mendukung perancangan tekno-ekonomi separator horizontal beserta sistem instrumentasi dan kontrolnya. Knowledge graph merepresentasikan batasan desain, konstanta fisik, koefisien korelasi, dan kaidah keteknikan berdasarkan metode Ken Arnold, API 12J, dan GPSA, yang diambil melalui Graph-RAG untuk memperkaya konteks jawaban LLM. Perhitungan sizing dilakukan oleh mesin deterministik berbasis persamaan keadaan Peng-Robinson, dilengkapi algoritma genetika untuk optimasi dimensi vessel yang meminimalkan Total Annual Cost, sementara LLM berperan sebagai antarmuka pengguna dan orchestrator sistem tanpa melakukan perhitungan fisika secara langsung. Sistem juga menghasilkan diagram PFD dan P&ID otomatis berdasarkan hasil sizing. Validasi dilakukan melalui perbandingan dengan simulasi Aspen HYSYS, contoh soal literatur Surface Production Operations, serta evaluasi kerangka RAGAS. Hasil menunjukkan deviasi dimensi vessel dan estimasi biaya berada dalam batas toleransi desain, sementara penerapan Graph-RAG meningkatkan konsistensi jawaban LLM terhadap fakta keteknikan. Dengan demikian, sistem ini berfungsi sebagai decision support system yang mempercepat proses desain awal separator secara konsisten, transparan, dan mendukung efisiensi sumber daya (resource efficiency) dalam proses engineering design pada industri minyak dan gas.
=====================================================================================================================================
The design of horizontal two-phase and three-phase separators is a critical stage in oil and gas production facilities, requiring the application of physical principles, design rules, and engineering experience. Conventional design approaches generally rely on iterative calculations, rule-of-thumb methods, and manual validation, which are time-consuming and prone to inconsistency between designers. This research develops an Artificial Intelligence system that integrates a knowledge graph, Graph Retrieval-Augmented Generation (Graph-RAG), and a Large Language Model (LLM) to support the techno-economic design of horizontal two-phase and three-phase separators along with their instrumentation and control systems. The knowledge graph represents design constraints, physical constants, correlation coefficients, and engineering rules based on the Ken Arnold, API 12J, and GPSA methods, which are retrieved through Graph-RAG to enrich the LLM's response context. Sizing calculations are performed by a deterministic engine based on the Peng-Robinson equation of state, complemented by a genetic algorithm for vessel dimension optimization that minimizes Total Annual Cost, while the LLM functions as an interactive chatbot interface and system orchestrator, directing module calls without performing physical calculations directly. The system also automatically generates PFD and P&ID diagrams based on sizing results. Validation was carried out through comparison with Aspen HYSYS simulation, reference examples from the Surface Production Operations literature, and evaluation using the RAGAS framework. Results show that vessel dimension and cost estimation deviations remain within acceptable design tolerances, while the application of Graph-RAG improves the consistency of LLM chatbot responses against engineering facts. Thus, the developed system functions as a decision support system that helps engineers accelerate the early stage design of separators in a more consistent, transparent, and resource efficiency manner within the engineering design process.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | separator, Artificial Intelligence, knowledge graph, large language model, decision support system, engineering design, resource efficiency |
| Subjects: | Q Science > QA Mathematics > QA336 Artificial Intelligence T Technology > TJ Mechanical engineering and machinery > TJ212 Control engineering systems. Automatic machinery (General) |
| Divisions: | Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis |
| Depositing User: | Ahmad Zidane Maulana |
| Date Deposited: | 01 Aug 2026 05:57 |
| Last Modified: | 01 Aug 2026 05:57 |
| URI: | http://repository.its.ac.id/id/eprint/141602 |
Actions (login required)
 |
View Item |