Time-of-flight prediction for fermentation process monitoring

This work presents an ultrasound-based inline sensor system, which is used to monitor the alcoholic yeast fermentation. The pulse-echo method is used to determine ultrasound velocity (USV) of the sample medium. The major aim of the paper is to highlight the importance of an accurate time-of-flight (TOF) estimation for accurate concentration determination and to present method immanent strategies to fulfill the requirements. An algorithm aiming at a stable and highly accurate TOF estimation in time domain was developed. The basic methods of the algorithm are a frame-wise signal analysis based on the sensor dimensions, cross-correlation to find complementary impulses and root estimation via polynomial fit to calculate the TOF in time domain. The analysis results of laboratory-scale validation trials (demineralized, vented water) and of real process data (yeast propagation) are presented. In spite of the stable algorithm performance in the lab scale, the algorithm fails in a few cases of real process signals. The relevant signals and corresponding causes for failure were analyzed and future strategies for algorithm enhancement are discussed. Reviewing the results, the aimed USV accuracy of 0.075 m/s can be achieved. The maximum USV error of the used principles and applied methods in the investigated temperature range is +/- 0.02%.