Ultrasound‐based, in‐line monitoring of anaerobe yeast fermentation: model, sensor design and process application

Abstract

Summary In order to implement process analytical technology in beer manufacturing, an ultrasound‐based in‐line sensor was developed which is capable to determine sound velocity and density via the multiple reflection method. Based on a systematic study of the ternary system water–maltose–ethanol, two models were established to estimate the critical process parameters: sugar and ethanol mass fraction. The sound velocity‐based model showed unreasonable high errors although temperature variations and deviations due to dissolved CO2 were corrected. In contrast, the sound velocity–density–temperature model provided an average root mean square error of 0.53%g/g sugar and 0.26%g/g ethanol content for the main fermentation. Method, sensor and model showed the capability to capture the process signature which may be related to product and process quality.

Predicting Alcohol Concentration during Beer Fermentation Using Ultrasonic Measurements and Machine Learning

“…Several in-line and on-line methods to monitor alcoholic fermentation have been investigated, including in-situ transflectance near-infrared spectroscopy [7,8], and Raman spectroscopy probes [9]; automated flow-through mid-infrared spectroscopy [10], Fourier transform infrared spectroscopy [11], and piezoelectric MEMS resonators [12]; non-invasive Raman spectroscopy through transparent vessel walls [13]; and CO 2 emission monitoring [14]. Ultrasonic (US) sensors are an attractive monitoring technique owing to their low Fermentation 2021, 7, 34 2 of 13 cost and have previously been used to study fermentation, including as in-line methods on circulation lines [15], in-situ in tanks [16], and using non-invasive, through-transmission of the fermenting media [17,18]. US monitoring techniques use high frequency (>1 MHz) and low power (<1 Wcm −2 ) pressure waves to characterise material properties whilst causing no alterations to the material in which they propagate [19].…”

“…However, US properties vary with temperature and the presence of gas bubbles causes attenuation of the sound wave [20]. Previous in-line, on-line, and off-line studies to monitor fermentation using US measurements have developed empirical or semi-empirical models from the speed of sound or acoustic impedance to determine alcohol content [16]. These methods require extensive calibration procedures to compensate for the effects of temperature, dissolved CO 2 [16,18,21], and yeast cell concentration [18].…”

“…Through ML, compensation procedures are not required as the complexities caused by varying process parameters imbedded in the sensor data can be unravelled. Furthermore, procedures for accurate determination of the speed of sound are not necessary [15,16,22]. This work presents three novel contributions to US monitoring of alcoholic fermentations: Firstly, ML is used to predict alcohol concentration during lab-scale beer fermentations from US measurements.…”

Intelligent Sensors for Sustainable Food and Drink Manufacturing

“…Ultrasonic measurements have been used to monitor ABV% during beer fermentation (Becker et al, 2001;Resa et al, 2004Resa et al, , 2009Hoche et al, 2016). The wort is a three-component liquid mixture (ethanol, water and sugar) with dissolved CO 2 and CO 2 bubbles, meaning at least two separate measurements are required to calculate the ABV%.…”

“…The previous research using US sensors to monitor fermentation used a variety of different signal and data processing methods to solve this problem. The most popular methods use either US velocity measurements at different temperatures (Becker et al, 2001) or a combination of US velocity with other measurements such as density (Resa et al, 2004(Resa et al, , 2009Hoche et al, 2016). However, multiple measurements at a single point in time are not required when using ML as time-series features can be incorporated into the models.…”