Predictive modelling of large scale algae biomass production

The position of algae in the production of biochemicals and biofuels is emerging. Large scale production facilities are necessary to fulfil the expected future demand for biodiesel and biochemicals produced with algae. With this development the challenge arises for efficient design of large scale production facilities. The design of such large scale systems is not straightforward. Wijffels et al. (2010) show that biomass productivity and economic feasibility are related to the type of reactor, the cultivation location, the production scale, substrates and operating conditions. The estimates of algae cultivation used in most life cycle assessment (LCA) studies are based on experimental work which was not necessarily performed under variable outdoor conditions. Most experimental studies concern only the effect of a few decision variables at a time. So the outcome of the studies does not reflect the best possible outdoor performance. Predictive models on the interaction between light and algae growth that consider the effect of decision variables on productivity will overcome these problems. These models will give insight in the complexity of light distribution in various cultivation systems and in the response of algae. In addition, such models support the up-scaling of laboratory and pilot research and will significantly improve the accurateness of large scale unit designs and LCA, which can be performed under outdoor conditions. We have developed predictive models to assess yearly algae biomass production in open raceway ponds, flat panels and tubular photobioreactors under a range of climate conditions and decision variables like light path, reactor distance and biomass concentration (Slegers et al. 2011). With these models the performance of algae cultivation at a certain location of the world with given reactor design and algae species can be determined. The model allows optimisation of algae productivity for the decision variables and can thus be used to identify good reactor layouts and operating conditions. Furthermore these predictions help to verify assumptions in experiments and identify gaps in experimental knowledge.