Increasing productivity of Spirulina platensis in photobioreactors using artificial neural network modeling.

Although production of microalgae in open ponds is conventionally practiced due to its economy, exposure of the algae to uncontrollable elements impedes achievement of quality and it is desirable to develop closed reactor cultivation methods for the production of high value products. Nevertheless, there are several constraints which affect growth of in closed reactors, some of which this study aims to address for the production of Spirulina. Periodic introduction of fresh medium resulted in increased trichome numbers and improved algal growth compared to growth in medium that was older than 4 weeks in 20 L polycarbonate bottles. Mixing of the cultures by bubbling air and use of draft tube reduced the damage to the growing cells and permitted increased growth. However, there was better growth in inclined cylindrical reactors mixed with bubbling air. The oxygen production rates were very similar irrespective differences in the maintained cultures densities. The uniformity in oxygen production rate suggested a tendency towards homeostasis in Spirulina cultures. The frequency of biomass harvest on the productivity of Spirulina showed that maintenance of moderate culture density between 0.16 and 0.32 g/L resulted in about 14% more productivity than maintaining the cell density between 0.16 and 0.53 g/L or 48% more than by daily harvest above 0.16 g/L. An artificial neural network based predictive model was developed, and the variables useful for predicting biomass output were identified. The model could predict the growth of Spirulina up to 3 days in advance with a coefficient of determination >0.94.