Application of Genetic Algorithm to Optimize Microalgae Cultivation Conditions in a Locally Assembled Flat Plate Photobioreactor

Abstract

The availability of more cost-effective sources of energy is a key driver of any economic development more so for a developing country like Kenya. The interest in biofuels is motivated by: the fluctuating oil prices and recognizing that the global fossil fuel reserves are exhausting fast; concern about fossil fuel emissions polluting the environment and resultant environmental change due to such emissions. Biodiesel, as a clean and renewable combustible, is a good alternative to replace mineral diesel. The main objective of this study was to apply Genetic Algorithm (GA) in the optimization of microalgae cultivation conditions in a locally assembled flat plate perspex photobioreactor (FPPPBR) at pilot plant scale. An optimization model was developed using GA to predict the biomass yield of microalgae incubated in a FPPPBR. Samples from which microalgae were isolated were collected aseptically in February 2021. Isolation was done by streaking, pour plate and serial dilution methods in Marine Biological Laboratory media at 27 oC, under continuous light intensity of 15 μmolphotonsm-2s-1. The model was validated using the strain that possessed a higher growth rate incubated in FPPPBR under 400-700 nm wavelength. As the validation process was ongoing, the influence of light quality and type of strain on yield was being investigated concurrently. This was replicated thrice. The validation of the simulation model was done by comparing simulated and experimental data. The statistical parameters used were: mean square error (MSE), mean absolute percentage error (MAPE), mean absolute error (MAE), root mean square error (RMSE), correlation coefficient (R) and student’s t-test. Statistical analyses were performed using IBM SPSS statistics 25 software, Design expert 13 and MATLAB R2016a. The simulation results produced optimum microalgae yield as: 0.250715±0.001608 gmolphotons-1 and optimal cultivation conditions as; biomass concentration of 0.1 gL-1, microalgae growth rate of 0.0102 h-1, photon flux density of 100 µmolphotonsm-2s-2, volume of reactor of 192 L and illuminated PBR surface area of 2.16 m2. Two strains were obtained and investigated namely, Chlorella emersonii and Chlorella vulgaris, whose growth rate was found to be 0.16 day-1 and 0.244 day-1 respectively. The experimental biomass yield was 0.438423±0.027122 gmolphotons-1 and the RMSE value for the optimization model was 0.1889, the MSE, MAE and MAPE were; 0.0357, 0.2717 and 42.67% respectively, R value of 0.231 and t-value of -165.091 at 5% level of significance. The C. vulgaris yielded; 9.30±0.57 g, 8.32±0.48 g and 7.78±0.67 g under 400-700, 430-480 and 610-680 nm wavelength, respectively. The corresponding values for the C. emersonii were; 5.88±0.26 g, 5.46±0.20 g and 5.12±0.14 g. The %lipids produced by the C. vulgaris were; 43.61, 35.87 and 34.56 respectively, under 430-480, 610-680 and 400-700 nm wavelengths. The C. emersonii yielded 33.50, 29.80 and 28.03 %lipids under 610-680, 430-480 and 400-700 nm wavelengths, respectively. C. vulgaris has potential for microalgae cultivation for biomass and biofuel production. A new optimization model was developed to predict microalgal biomass yield and the strains grown in this study produced the highest biomass under 400-700 nm wavelength. C. vulgaris produced highest lipids (43.61%) under 430-480 nm whereas C. emersonii yielded highest lipids (33.50%) under 610-680 nm. GA and RSM could be used to optimize microalgae cultivation conditions.