The effect of bubble formation within carbonated drinks on the brewage foamability, bubble dynamics and sensory perception by consumers

Abstract

Beer and sparkling water are popular carbonated beverages within consumers as it can be evidenced by an important growth in terms of volume sales around the world. Although they are both composed of carbon dioxide (CO2), this gas is produced in different forms and the drinks present distinct performance in terms of their sensory and physicochemical characteristics. The most important factors that determine quality and consumer acceptability in all carbonated beverages are the visual aspects such as colour, bubble morphometry and dynamics and foam-related parameters as these create the first impression when consumers select the products. Currently, consumers are in a constant search for more premium or high-quality products, which has put pressure in their respective industries to remain relevant in the market. Available methods to assess physical and chemical parameters in carbonated beverages to assess quality tend to be costly, time-consuming and not reliable. Therefore, there is then a need to develop automatic, more reliable, accurate and affordable quality assessment techniques. From the sensory analysis perspective, traditional consumer tests to assess products acceptability are primarily focused on subjective conscious responses from participants which contribute a reduced amount of information, which many times it can be bias. Traditional methods to tap into the autonomic nervous system response from consumers, which is unconscious, are invasive and had not been implemented in the assessment of beer and sparkling water, which can complement and enhance objective and meaningful information for consumers from the physiological and emotional responses to these products. This research was focused on the development of novel techniques such as an automatic robotic pourer integrated with remote sensing to measure the physical dynamics of foamability and bubbles coupled with CO2, alcohol and pouring temperature to add chemical analysis. The data analysis included an artificial intelligence (AI) approach by using computer vision algorithms and machine learning modelling. The objective from these measurements was to assess color, bubble and foam-related parameters in both beer and carbonated water. Near-infrared (NIR) spectroscopy was used to obtain the chemical fingerprinting of the products and to analyze their relationship with foamability and bubble related measurements. Furthermore, an electronic nose was developed using nine gas sensors to assess aromas and beer quality to enhance the chemical (non-contact) analysis from the robotic pourer. Additionally, a new integrated system to assess consumer acceptability using a novel bio-sensory computer application (App) coupled with video and thermal cameras was developed to obtain biometric responses such as heart rate, face temperature, gaze fixations and facial expressions. This new App was used on panelists assessing different products considered in this research. The information gathered from these new techniques allowed to create different machine learning (ML) models with high accuracy (R>0.8) to classify beers according to their liking and physicochemical intrinsic characteristics. Likewise, different ML models based on regression algorithms were created using the data obtained from the robotic pourer, electronic nose, and NIR spectroscopy data to predict the products intensity of sensory descriptors, consumers acceptability and chemometry to assess beer quality. The results obtained from this research showed to be accurate, reliable, rapid and affordable tools that can be applied to the growing industries related to beverages production to monitor and increase product quality to compete in the international market.