Soft-sensor based on sliding modes for industrial raceway photobioreactors

Abstract

Microalgae reactors provide an efficient and clean alternative for the production of biofuels, nutritional and cosmetic bioproducts, wastewater treatment, and mitigation of industrial gases to reduce greenhouse gas emissions. The main control objective in these systems is productivity optimisation. For this reason, real-time monitoring of key biological performance indicators affecting microalgae production such as microalgae growth rate, biomass concentration, dissolved oxygen, pH level or total inorganic carbon is crucial. However, there are no sufficiently robust solutions on the market to estimate or measure all of these variables, especially for open reactors on an industrial scale. This paper presents a new online state estimator, based on a robust sliding mode observer combined with a nonlinear dynamic model endowed with a minimum number of states to capture dynamics of key biological performance indicators. This soft-sensor has been verified with a realistic reactor model that has been experimentally tested. Simulations showed promising results in terms of accuracy (with mean values of the state estimation errors in the order of 10⁻⁴ g m⁻³ for the biomass concentration, 10⁻⁵ to 10⁻¹³ mol m⁻³ for the other states and deviations in the order of 10⁻⁴ g m⁻³ for the biomass concentration, 10⁻⁵ to 10⁻¹⁰ mol m⁻³ for the other states) and robustness with respect to signal noise, state deviations, initial errors and parametric uncertainty.