Logo of the Information System for Agriculture and Food Research

Information System for Agriculture and Food Research

Information platform of the Federal and State Governments

Simulation and prediction of mechanically induced mouthfeel during consumption of low-fat, non-lumpy food capable of flowing

Project

Food and consumer protection

This project contributes to the research aim 'Food and consumer protection'. Which funding institutions are active for this aim? What are the sub-aims? Take a look:
Food and consumer protection


Project code: AiF 15962 N
Contract period: 01.01.2009 - 31.12.2012
Budget: 239,750 Euro
Purpose of research: Applied research

The present sub-project investigated by means of numerical simulations the mechanically induced mouth feel during the intake of fat reduced foods, that are capable of flowing and are not containing pieces (e.g. yoghurt). Furthermore, objectification as well as prediction of sensorial textural perception is in the focus of the project. For this purpose it was required to 1.) determine by numerical simulations the flow processes during the intake of fat reduced foods, that are capable of flowing and are not containing pieces, leading to the mechanically induced mouth feel, 2.) generate a hybrid system for the objectified description of the mouth feel, and based on this 3.) develop an expert system for the targeted sensorial product and process design of newly developed foods. The connection of sensorial, technological, and fluid mechanical knowledge was based ona close cooperation with human sensorial oriented groups of the present cluster project (Busch-Stockfisch (HAW Hamburg)) and groups working on technological and process engineering aspects (Hinrichs (Universität Hohenheim), Schuchmann (KIT)). Numerical simulations of the swallowing process characterize the resulting mechanical stresses at the inner mouth surface (tongue, palate) for different product classes. The products range from low-viscous Newtonian model foods (e.g. water) to high-viscous non-Newtonian foods (yoghurt probes with different fat content of 0.1% up to 10%). The numerical simulations in an oral cavity model quantify and visualize the present flow processes of the foods as well as normal and shear stresses in the mouth leading to stimuli at the mechano receptors. These quantities are expressed in terms of characteristic fluid mechanical numbers (e.g. Reynolds number, Deborah number, and acceleration, pressure, and friction forces). The analysis of the flow field regarding the induced mechanical stresses takes place in independent areas at the tongue surface. By this, a “finger print” of the mechanical stress distribution is gained and, therefore, an objectification of the mechanical effects related to the sensorial perception in a simplified system. Amongst others, it could be shown that the areas facing the highest mechanical load are located near the throat. The findings regarding zone specific mechanical stresses were correlated with experimentally determined sensorial texture perception data of the group Busch-Stockfisch and rheological and technological data of the group Hinrichs. In processes like the sensorial texture perception, that feature a high complexity and, therefore, do not allow a mathematical equation based description, cognitive algorithms can be applied for prediction purpose. This sub-project used mainly Artificial Neural Networks (ANN) and a hybrid Neuro-Fuzzy-system. Artificial Neural Networks are mathematical algorithms, that – similar to the learning procedure of humans – learn relations in data sets (input and output data) by a training step. During this training step connections and weight factors are determined between so called neurons. After a successful training a validation - by real data that have not been introduced into the ANN before – proves the ability of the ANN to predict output data based on input data. Similar to the ANN, the Neuro-Fuzzy-system is based on training with available data. Additionally, the Fuzzy-part includes rules formulated by experts. These rules follow an IF-THENlogic. These rules have been provided, here, by the groups Busch-Stockfisch and Hinrichs. It could be shown, that an ANN having mechanical quantities of the numerical simulations (zone specific shear stresses at the tongue) as input data gives very good predictions of the sensorial attribute “oral viscosity” for a specific product. Furthermore, the extension by the expert rules in terms of the Neuro-Fuzzy-system (= hybrid expert system) leads to a high prediction quality of the attributes “creaminess”, “heterogeneity”, “adstringency”, and “mouthcoating”. The target of an objectified description of the expected mouth feel has been reached based on the fluid mechanical induced stresses at the inner mouth surface by a hybrid system. Additionally, the extension of the hybrid expert system by searching algorithms and optimization approaches resulted in the targeted proposition of product compositions based on given sensorial constraints. The validation proves also for this case good prediction accuracy. It can be concluded that the developed hybrid expert system possesses a high potential to assist sensorial panels and product development in practical applications in the selection of promising sensory-product/process parameter-combinations.

show more show less

Subjects

Advanced Search