TY - JOUR
T1 - Bioprocessing sidestream valorization as culture stream for Chlorella vulgaris biomass accumulation: neural data-driven design of experiments
AU - Diaz-Hernandez, Maria Monserrat
AU - Krebss-Kleingezinds, Eduards
AU - Parra-Saldivar, Roberto
AU - Alfaro-Ponce, Mariel
AU - Chairez, Isaac
AU - Melchor Martinez, Elda Madai
AU - Kjellberg, Kasper
AU - Soheil Mansouri, Seyed
PY - 2025
Y1 - 2025
N2 - This study develops a method to obtain optimized culture conditions for Chlorella vulgaris microalgae that yields augmenting contaminant consumption, employing a combination of the design of experiments theory and feed-forward artificial neural networks (ANNs). The contaminants originate from a bio-manufacturing sidestream characterized by the presence of carbon, nitrogen, and phosphorus-rich compounds. The Box-Behnken experimental design is employed to develop a first optimization-oriented model, considering the culture broth?s CO2 concentration, luminous exposition, and agitation as factors, with Chlorella vulgaris biomass production as the response variable. Data obtained from the round of observations defined by the Box-Behnken study was used to train the ANN-aided model. The optimization process identified optimal levels for each factor: 0.5% CO2, 77?RPM of agitation, as well as an eight to sixteen (8:16) hours day/night cycle. The final optimal conditions were applied in two photobioreactors of different volumes (3?L and 30?L) for eight days to scale the cultivation and study nutrient intake removal and biomass growth. Throughout this cultivation process, chemical oxygen demand (COD) decreased about 40.80%, the total nitrogen diminished about 44.63%, the phosphorus reduced by 98.65%, and microbial dry biomass concentration augmented from 0.1 to 0.5?g/L at the end of the culture process.
AB - This study develops a method to obtain optimized culture conditions for Chlorella vulgaris microalgae that yields augmenting contaminant consumption, employing a combination of the design of experiments theory and feed-forward artificial neural networks (ANNs). The contaminants originate from a bio-manufacturing sidestream characterized by the presence of carbon, nitrogen, and phosphorus-rich compounds. The Box-Behnken experimental design is employed to develop a first optimization-oriented model, considering the culture broth?s CO2 concentration, luminous exposition, and agitation as factors, with Chlorella vulgaris biomass production as the response variable. Data obtained from the round of observations defined by the Box-Behnken study was used to train the ANN-aided model. The optimization process identified optimal levels for each factor: 0.5% CO2, 77?RPM of agitation, as well as an eight to sixteen (8:16) hours day/night cycle. The final optimal conditions were applied in two photobioreactors of different volumes (3?L and 30?L) for eight days to scale the cultivation and study nutrient intake removal and biomass growth. Throughout this cultivation process, chemical oxygen demand (COD) decreased about 40.80%, the total nitrogen diminished about 44.63%, the phosphorus reduced by 98.65%, and microbial dry biomass concentration augmented from 0.1 to 0.5?g/L at the end of the culture process.
KW - Bioremediation
KW - Chlorella vulgaris
KW - Box-Behnken
KW - Artificial neural network
KW - Industrial wastewater
U2 - 10.1080/17597269.2024.2447154
DO - 10.1080/17597269.2024.2447154
M3 - Journal article
SN - 1759-7269
JO - Biofuels
JF - Biofuels
ER -