EXPLORATION OF TEXTILE WASTEWATER TREATMENT PLANT STRUCTURE USING SUPERSTRUCTURE OPTIMIZATION AND DEEP REINFORCEMENT LEARNING

• Main Author: Abdullah Al-Mujahid, Teja
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/65221
• Institution: Institut Teknologi Bandung

Textile wastewater treatment plant has important role in dealing with water pollution engendered by textile effluent. Process synthesis of textile wastewater treatment pilot plant can be done heuristically through intuition. However, increasing pilot plant capacity to meet small industrial scale requires complex plant configuration which renders the need for process synthesis method alternatives. This paper tried to address this problem by demonstrating two different methods, superstructure optimization and Deep RL-based process synthesis, as alternatives to heuristic method. Analysis results show that under given scenario, both superstructure optimization and deep RL have successfully found optimal structure of the plant, consisting of a pH neutralization tank and parallel configuration of 4 electrocoagulation tanks. Superstructure optimization and deep RL took respectively 0.174 second and 18.627 seconds to obtain optimal solution. Optimal structure from both methods managed to lower the planned construction cost up to 51.8% of originally planned budget. This research has the potential to be utilized for process synthesis of modular wastewater treatment plant, as well as designing flow diagram for various modular plants.