|Abstract or Summary
- Microalgae has been under careful consideration as a sustainable feedstock for renewable fuel recently due to its year round production with high energy yields per unit area, reduced need for arable land, water treatment benefits with nutrient cycling, and minimal competition with the food industry. Despite of all these advantages, commercialization of algal biofuels still lacks economic viability, mainly because of lower productivities and higher cost of harvesting and processing of algae. Photobioreactors are able to provide high yields of algae but they suffer from limitation of high initial cost, whereas growth in low-cost open ponds are highly affected by external factors such as ambient temperature, light availability and contamination with other micro-organisms. These external factors inhibit a consistent growth, which consequently reduces the final productivity of algae. There is an imminent need to find a solution that can take into account the objectives of maximum productivity with minimum economic and environmental effects. The focus of this dissertation was to design, implement and validate a model predictive control which can assure optimum algal growth and productivity, even under the effect of disturbances. The dissertation was divided into four studies, each of which concentrates on developing different models to be integrated in the design of the model predictive controller.
In the first study, a compartmentalized genome scale metabolic network was reconstructed for C. variabilis to offer insight into various metabolic potentials from this alga. The model iAJ526 was reconstructed with 1455 reactions, metabolites and 526 genes. 21% of the reactions were transport reactions and about 81% of the total reactions were associated with enzymes. Model was able to simulate experimental batch growth conditions with fair agreement under three light conditions. The discrepancy in experimental and model results was attributed to the incapability of the model to capture kinetics of growth, including substrate inhibition and photoinhibition. Cumulative wavelengths around 437 nm, 673 nm and 680 nm in the model were observed to be 100% efficient in their utilization.
The second study evaluated the economic viability and to estimate the energy use and greenhouse gas (GHG) emissions during life cycle of renewable diesel (RD) production from algae via hydrothermal liquefaction process. Hydrothermal liquefaction is one such technology that converts the algae into high heating value bio-oil under high temperature and pressure. RD yields for algae were estimated 10.19 ML/year unit price of production as $1.75/L RD. Energy and emissions were calculated per functional unit which was defined as 1000 MJ of available energy in fuel at the pump. Fossil energies were calculated as 241.6MJ to produce one functional unit of RD from algae. The GHG emissions during life cycle of RD production were found to be 6.2 times less than those produced for 1000MJ of conventional diesel.
In the third project, Monod's multiplicative kinetic model for Chlorella vulgaris, one of the most robust algal strains for outdoor growth, was developed with limiting nitrate and CO₂. The specific growth rate calculated from the growth profiles was 0.0196 hr⁻¹. The values of μ_max, k_N, k_C, k_d, Ea and k_L a were estimated as 0.23 hr⁻¹, 26.53 mg/l, 7.64 mg/l, 0.08hr⁻¹, 3.69 J/mol and 0.24 hr⁻¹. The yield coefficient for nitrate (Y_(X⁄N)) and CO₂ (Y_(X⁄C)) for C. vulgaris were calculated as 2.86 and 1.96 respectively. Validation experiments proved the kinetic model to be robust under vast range of initial biomass and nitrate concentrations.
The kinetic model developed, formed the basis of the model predictive control (MPC) designed for optimum growth of algae, with an objective of maximizing productivity and minimizing the total cost and GHG emissions. The MPC proved to be an efficient tool that can help in a steady growth even under natural disturbances. The growth of algae under the light and temperature disturbances tested was higher in the reactor with MPC than the other reactor with no MPC. However, due to light limitation, the growth with low light of 34W/m² (simulating a cloudy day) in the MPC reactor was not different form the other one with no MPC. This disturbance was considered outside the realm of controller as the control variables or the manipulated variables were not able to overcome light limitation. In absence of any disturbance, the growth in MPC reactor was not statistically different form the no MPC reactor. This shows the importance of MPC under conditions deviating from the optimum, which is a common case in real time outdoor ponds.