(523e) Artificial Intelligence-Guided Bioprocess Design to Achieve Record Algal Productivity

Algal biofuel and bioproducts have significant potential to address the energy and environmental challenges in our generation. However, the algal biofuel commercialization was hindered by the low productivity, higher harvesting and production costs. We used two machine learning models to predict the light penetration path and the growth rate, respectively. The integration of the two models allows us to simulate the cell growth rate throughout the process. Based on the models, we have predicted maximized growth at the optical density of about 2.3. The AI technology has guided us to develop a semi-continuous cultivation (SAC) process, where the algal growth was controlled around the best density. The semi-continuous cultivation significantly improved the cell growth rate. Furthermore, we engineered a fast-growing strain, Synechococcus elongatus UTEX 2973, to produce limonene at high productivity, based on previously established kinetic models. The high limonene productivity has created hydrophobic interactions among the cells, thereby leading to aggregation-based sedimentation (ABS). The ABS has delivered a low-cost harvesting method and empowered the cost-effective SAC. The ABS-empowered SAC has unleashed cyanobacterial growth potential to achieve 0.1g/L/hour biomass productivity and 0.2mg/L/hour limonene productivity in photobioreactors over multiple days of production. The re-configuration of the SAC process in an outdoor pond system has achieved 43.3 g/m²/day biomass yield, which can translate into minimum biomass selling price at $281 per ton. Overall, we have addressed two major challenges in algal biofuel processing, the mutual shading growth inhibition and the cost-effective harvesting, which empowered record-level outdoor productivity and market competitive algal biomass price.