Applications of artificial neural networks for experimental design optimization of Chlorella vulgaris microalgae growth

Citation
Share
Abstract
This thesis proposes developing an optimization experimental model to optimize nutrient consumption and microalgae growth from the Novozymes company’s sidestream. The optimization model was created using the Box-Behnken experimental design for three factors. These three criteria were considered to raise the Chlorella v. biomass, and three different levels for each factor were chosen and implemented. The first factor chosen was CO2 since microalgae are important in producing energy for growth and proteins, lipids, and nucleoid acid. The second component chosen was agitation, which allows for the exchange of gases in the medium and the uniform consumption of nutrients from the medium. The day/night cycle was used to generate mixotrophic cultivation, which encouraged the culture to utilize the carbon in the sidestream while maintaining the green pigments of Chlorella vulgaris due to the presence of light. Following the experimentation phase, the best levels for each factor were 0.5% CO2, 70 RPM of agitation, and 8:16 hrs of day/night cycle. These amounts were used in a photobioreactor to cultivate and observe nutrient consumption behavior for eight days. Following these days, the COD level was reduced by 47.34%, the total nitrogen decrement was 48.70 %, the total phosphorus decrement was 96.42 %, and the dry biomass increased by 300 %. Simultaneously, a suitable neural network was designed to optimize the optimal levels for the same three parameters; this model was trained, validated, and evaluated using the experimental results. The ideal amounts for each factor were 0.5% CO2, 77 RPM of agitation, and 8:16 hours of day/night cycle. These levels were used in a photobioreactor to cultivate and observe nutrient consumption behavior for eight days. Following these days, the COD level declined by 40.80%, the total nitrogen decrement was 44.63%, the total phosphorus decrement was 98.65%, and the dry biomass increased by 400%. Both models are based on the work’s greatest contribution of reducing sidestream nutrients and promoting the increase in microalgae biomass in a shorter time than traditional methods that range from 12 to 14 days, as well as being a solution for treating wastewater from the enzyme manufacturing process.
Description
42787