Optimisasi Chemical Oxygen Demand Produced Water Pada Polishing Unit

Fauzi, Deni Nur (2021) Optimisasi Chemical Oxygen Demand Produced Water Pada Polishing Unit. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 02311850010002-Master_Thesis.pdf - Accepted Version Download (1MB)

Abstract

Polishing unit pada Saka Indonesia Pangkah Limited menggunakan biological treatment yang memanfaatkan bakteri aerob untuk mengurangi kadar Chemical Oxygen Demand (COD) effluent pada produce water saat eksplorasi minyak agar sesuai dengan baku mutu. Untuk menjaga agar bakteri tetap hidup diperlukan tambahan substrat berupa nitrat dan fosfat yang dijaga pada kondisi tertentu. Uji data operasional menunjukan bahwa hubungan antara kadar COD effluent dengan variabel-variabel lain non linier dan kompleks. Sehingga dibutuhkan pemodelan black box seperti Artificial Neural Network (ANN) untuk memodelkan hubungan ini. Dari Process Flow Diagram (PFD), data operasional dan pertimbangan manajemen maka 11 variabel menjadi input model ANN dan COD sebagai output variabel. Output variabel model ini akan menjadi fungsi objektif dimana tujuannya adalah nilai minimum. Dua input variabel sebagai variabel yang dioptimisasi yaitu massa nitrat dan fosfat. Sedangkan konstrain berupa 9 parameter operasional dan tiga range variabel input-output. Hubungan empiris antara variabel input-output yang tidak mempunyai persamaan turunan maka teknik optimisasi stokastik diperlukan. Salah satu teknik optimisasi stokastik adalah Genetic Algorithm (GA). Hasil dari prediksi ANN menggunakan struktur Multi Layer Perceptron (MLP) dengan input Finite Impulse Response (FIR) serta pelatihan Levenberg-Marquardt menghasilkan Root Mean Square Error (RMSE) paling optimal berada pada hidden node 42 yaitu RMSE selama pelatihan adalah 0.16092 dan RMSE selama validasi 1.73769. Optimisasi kadar COD mendapatkan hasil kadar COD optimal yaitu 44.8668 mg/L pada kadar nitrat sebesar 9 mg/L dan kadar fosfat sebesar 12 mg/L.
===================================================================================================================================
In order to conform with government regulation about waste water quality, Saka Indonesia Pangkah Limited utilizes the polishing unit aerobic bacteria to reduce levels of Chemical Oxygen Demand (COD) effluent in produce water during oil exploration. To control the bacteria alive, an additional substrate in the form of nitrate and phosphate is maintained under certain conditions. The operational data shows that the relationship between COD effluent levels with other variables is non-linear and complex. The black box model such as Artificial Neural Network (ANN) is required. Refer to Process Flow Diagram (PFD), operational data and management considerations. The 11 variables are selected as ANN model input and COD as the output variable. The output of this model variable will be representative as an objective function where the objective is the minimum value. Two input variables are selected as the optimized variables, namely the mass of nitrate and phosphate. While the constraint consists of 9 operational parameters and three ranges of input-output variables. The empirical relationship between the input-output variables that do not have a derivative equation is required the stochastic optimization technique. One of the famous stochastic optimization techniques is Genetic Algorithm (GA). The results of ANN predictions using the Multi Layer Perceptron (MLP) structure with Finite Impulse Response (FIR) input and Levenberg-Marquardt training resulted an optimal Root Mean Square Error (RMSE) at hidden node 4. RMSE during training was 0.16092 and RMSE during validation 1,73769. Optimization of COD levels resulted an optimal COD levels of 44.8668 mg / L at a nitrate level of 9 mg / L and a phosphate level of 12 mg / L.

Item Type:	Thesis (Masters)
Additional Information:	RTI 628.1 Fau o-1 2021
Uncontrolled Keywords:	Optimisasi, Chemical Oxygen Demand, Produce water, Kualitas Air, Artificial Neural Network, Genetic Algorithm, Optimization, Chemical Oxygen Demand, Produce water, Water quality, Artificial Neural Network, Genetic Algorithm
Subjects:	T Technology > TD Environmental technology. Sanitary engineering > TD259.2 Drinking water. Water quality
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30101-(S2) Master Thesis
Depositing User:	- Davi Wah
Date Deposited:	20 Jun 2023 11:17
Last Modified:	20 Jun 2023 11:17
URI:	http://repository.its.ac.id/id/eprint/98153

Actions (login required)

View Item