Software sensors for biomass concentration estimation in filamentous microorganism cultivation process

Abstract

In this study, the potential of two software sensors for on-line estimation of biomass concentration during cultivation of filamentous microorganisms is examined. The first sensor is based on common bioreactor off-gas analyses, and uses the assumption of the biomass concentration linear dependence on the square root of cumulative O-2, consumption. Parameters of the semi-empirical data-driven software sensor based on off-gas analysis were calculated from experimental cultivation data using linear regression. The second sensor is based on biocalorimetry, i.e., the on-line calculation of metabolic heat flux from general enthalpy balance of the bioreactor. The software sensor based on biocalorimetry thus essentially represents a model-driven approach, making use of a fundamental process model based on the enthalpy balance around the bioreactor. This approach has been combined with the experimental identification of the specific biomass heat production, which represents the main process-specific parameter of the software sensor based on biocalorimetry. For this sensor, the accuracy requirements on the process variable on-line measurements were also analysed. The experimental data from the pilot-scale antibiotics Nystatin production by a bacterium Streptomyces noursei were used to calculate the specific bioprocess heat production value using linear regression. The achieved results enabled us to propose a new on-line indicator calculated as the ratio of the outputs of both sensors, which can serve as a timely warning of the risk of undesired nutritional conditions of a culture characterized as underfeeding.