Optimization of Microbial Electrolysis Cell for Sago Mill Wastewater Derived Biohydrogen via Modeling and Artificial Neural Network
Hydrogen is an efficient energy carrier because it can be stored for a long time and allows clean combustion. The biological production of hydrogen is more sustainable because it requires less energy and uses waste for energy recovery. The microbial electrolysis cell (MEC) is a bioelectrochemical re...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English |
| Published: |
2023
|
| Subjects: | |
| Online Access: | http://ir.unimas.my/id/eprint/43048/4/Mohamad%20Afiq_dsva.pdf http://ir.unimas.my/id/eprint/43048/5/Thesis%20Master%20Eng_Mohamad%20Afiq.ft.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Hydrogen is an efficient energy carrier because it can be stored for a long time and allows clean combustion. The biological production of hydrogen is more sustainable because it requires less energy and uses waste for energy recovery. The microbial electrolysis cell (MEC) is a bioelectrochemical reactor for the production of hydrogen from the biodegradation of wastewater with low energy requirements. The performance of MEC for large-scale production is less improved because little emphasis is placed on understanding the dynamic behavior of the bioelectrochemical process. The nonlinear effects resulting from the multiple variations of the uncertainty parameters, which are difficult to adjust or control in the experimental analysis, lead to a significant deviation between the experimental data and the predictions. Without validity and reliability tests, the MEC model could overestimate or underestimate the performance profiles compared to experimental data. Therefore, the main objective of this study is to improve the validity and reliability of the simplified biofilm growth model to optimize the laboratory MEC process of biohydrogen production from sago wastewater substrate using MATLAB. Model validity describes the first sub-objective, which is to solve the complexity of the nonlinear interaction of multiple MEC input variables related to the hydrogen production rate response using artificial neural networks (ANN) before validating the mathematical modeling results by comparing experimental data with the predicted substrate concentration profile and hydrogen production rate profile based on the re-estimated input values of the model parameters using single-objective optimization based on the nonlinear convex method using gradient descent algorithm. The model reliability describes the second sub-objective, which is to determine the feasible operating window of the MEC using multiple-objective optimization based on the nonlinear convex method using gradient descent algorithm as the objective function in maximizing the percentage efficiency of the MEC after validating the mathematical model. The review of optimization studies in the literature contributes to the development of a flowchart of an advanced optimization strategy that serves as a guide for the methodology of the study. The project begins with data extraction from an experiment on hydrogen production from sago wastewater in a double chamber MEC over 16 days. The ANN training provides the correlation equations to obtain the experimental hydrogen production rate based on the pH data of the catholyte at a constant applied potential of 0.8 V and a current density of 0.632 A‧m-2. The mathematical model, which consists of a stoichiometric reaction and kinetics model incorporating bioelectrochemical balances model, predicts the substrate concentration and hydrogen production rate profile. Single-objective optimization improves the validity of the model by minimizing the mean square error (MSE) between the experimental data and the prediction with the adjustment of the input values of the model parameters. Multiple-objective optimization improves the reliability of the model by determining the optimal operating conditions to achieve maximum MEC performance. The model after process optimization achieves the maximum COD removed efficiency of 81.99%, Coulombic efficiency of 69.01%, and energy efficiency of 7.47% when the process conditions are at an optimum applied potential of 0.485 V, anode surface area of 0.098 m2, anodic chamber volume of 4 L, and initial substrate concentration of 2,500.99 mg‧L-1. The increase of the maximum steady-state hydrogen production rate from 28.9 mL‧day-1 to 33.0 mL‧day-1 and the optimal initial substrate concentration in the influent of 476.0 mg‧L-1 for stable hydrogen production prove the reliability of the model. |
|---|