Computer-aided food engineering

Digital Twin Methodology in Food Processing: Basic Concepts and Applications

PURPOSE OF REVIEW

The goal of this article is to present a concise review about digital twin (DT) methodology and its application in food processing. We aim to identify the building blocks, current state and bottlenecks, and to discuss future developments of this approach.

RECENT FINDINGS

DT methodology appears as a powerful approach for digital transformation of food production, via integration of modelling and simulation tools, sensors, actuators and communication platforms. This methodology allows developing virtual environments for real-time monitoring and controlling of processes, as well as providing actionable metrics for decision-making, which are not possible to obtain by physical sensors. So far, main applications were focused on refrigerated transport and storage of fresh produces, and thermal processes like cooking and drying. DT methodology can provide useful solutions to food industry towards productivity and sustainability, but requires of multidisciplinary efforts. Wide and effective implementation of this approach will largely depend on developing high-fidelity digital models with real-time simulation capability.

Unveiling the relationship between food unit operations and food industry 4.0: A short review

The fourth industrial revolution (Industry 4.0) is driving significant changes across multiple sectors, including the food industry. This review examines how Industry 4.0 technologies, such as smart sensors, artificial intelligence, robotics, and blockchain, among others, are transforming unit operations within the food sector. These operations, which include preparation, processing/transformation, preservation/stabilization, and packaging and transportation, are crucial for converting raw materials into high-quality food products. By incorporating advanced digital, physical, and biological innovations, Industry 4.0 technologies are enhancing precision, productivity, and environmental responsibility in food production. The review highlights innovative applications and key findings that showcase how these technologies can streamline processes, minimize waste, and improve food product quality. The adoption of Industry 4.0 innovations is increasingly reshaping the way food is prepared, transformed, preserved, packaged, and transported to the final consumer. The work provides a valuable roadmap for various sectors within agriculture and food industries, promoting the adoption of Industry 4.0 solutions to enhance efficiency, quality, and sustainability throughout the entire food supply chain.

Pizza3: A general simulation framework to simulate food-mechanical and food-deconstruction problems

Current mesh-based simulation approaches face significant challenges in continuously modeling the mechanical behaviors of foods through processing, storage, deconstruction, and digestion. This is primarily due to the limitations of continuum mechanics in dealing with systems characterized by free boundaries, substantial deformations, mechanical failures, and non-homogenized mechanical properties. The dynamic nature of food microstructure and the transformation of the food bolus, in relation to its composition, present formidable obstacles in computer-aided food design. In response, the Pizza3 project adopts an innovative methodology, utilizing an explicit microstructural representation to construct and subsequently deconstruct food products in a modular, Lego-like fashion. Central to this simulation approach are "food atoms", conceptualized from the principles of smoothed particle hydrodynamics. These units are significantly larger than actual atoms but are finely scaled to represent both solid and liquid states of food faithfully. In solid phases, food atoms interact via pairwise forces akin to bond-peridynamic methods, thus extending the capabilities of continuum mechanics to encompass large deformations and fracturing phenomena. For liquids, the model employs artificial conservative and dissipative forces, enabling the simulation of a variety of phenomena within the framework of partial compressibility. The interaction dynamics between rigid and soft objects and fluids are accurately captured through Hertzian contact mechanics, offering a versatile parameterization applicable to impermeable (but possibly penetrable) surfaces and enforcing no-slip conditions. The efficacy of this framework is showcased through the successful modeling of three time-dependent 3D scenarios, each rigorously validated against established analytical and experimental models. Advancing beyond these initial applications, the framework is further extended to more intricate cases inadequately addressed in current literature. This extension sheds light on the underlying mechanisms of in-mouth texture perception, offering new insights and tools for food engineering and design.

Application of physics encoded neural networks to improve predictability of properties of complex multi-scale systems

Predicting physical properties of complex multi-scale systems is a common challenge and demands analysis of various temporal and spatial scales. However, physics alone is often not sufficient due to lack of knowledge on certain details of the system. With sufficient data, however, machine learning techniques may aid. If data are yet relatively cumbersome to obtain, hybrid methods may come to the rescue. We focus in this report on using various types of neural networks (NN) including NN's into which physics information is encoded (PeNN's) and also studied effects of NN's hyperparameters. We apply the networks to predict the viscosity of an emulsion as a function of shear rate. We show that using various network performance metrics as the mean squared error and the coefficient of determination ( R 2 ) that the PeNN's always perform better than the NN's, as also confirmed by a Friedman test with a p-value smaller than 0.0002. The PeNN's capture extrapolation and interpolation very well, contrary to the NN's. In addition, we have found that the NN's hyperparameters including network complexity and optimization methods do not have any effect on the above conclusions. We suggest that encoding NN's with any disciplinary system based information yields promise to better predict properties of complex systems than NN's alone, which will be in particular advantageous for small numbers of data. Such encoding would also be scalable, allowing different properties to be combined, without repetitive training of the NN's.

Virtualization and digital twins of the food supply chain for enhanced food safety

Meeting food safety requirements without jeopardizing quality attributes or sustainability involves adopting a holistic perspective of food products, their manufacturing processes and their storage and distribution practices. The virtualization of the food supply chain offers opportunities to evaluate, simulate, and predict challenges and mishaps potentially contributing to present and future food safety risks. Food systems virtualization poses several requirements: (1) a comprehensive framework composed of instrumental, digital, and computational methods to evaluate internal and external factors that impact food safety; (2) nondestructive and real-time sensing methods, such as spectroscopic-based techniques, to facilitate mapping and tracking food safety and quality indicators; (3) a dynamic platform supported by the Internet of Things (IoT) interconnectivity to integrate information, perform online data analysis and exchange information on product history, outbreaks, exposure to risky situations, etc.; and (4) comprehensive and complementary mathematical modeling techniques (including but not limited to chemical reactions and microbial inactivation and growth kinetics) based on extensive data sets to make realistic simulations and predictions possible. Despite current limitations in data integration and technical skills for virtualization to reach its full potential, its increasing adoption as an interactive and dynamic tool for food systems evaluation can improve resource utilization and rational design of products, processes and logistics for enhanced food safety. Virtualization offers affordable and reliable options to assist stakeholders in decision-making and personnel training. This chapter focuses on definitions and requirements for developing and applying virtual food systems, including digital twins, and their role and future trends in enhancing food safety.

Predicting and improving complex beer flavor through machine learning

The perception and appreciation of food flavor depends on many interacting chemical compounds and external factors, and therefore proves challenging to understand and predict. Here, we combine extensive chemical and sensory analyses of 250 different beers to train machine learning models that allow predicting flavor and consumer appreciation. For each beer, we measure over 200 chemical properties, perform quantitative descriptive sensory analysis with a trained tasting panel and map data from over 180,000 consumer reviews to train 10 different machine learning models. The best-performing algorithm, Gradient Boosting, yields models that significantly outperform predictions based on conventional statistics and accurately predict complex food features and consumer appreciation from chemical profiles. Model dissection allows identifying specific and unexpected compounds as drivers of beer flavor and appreciation. Adding these compounds results in variants of commercial alcoholic and non-alcoholic beers with improved consumer appreciation. Together, our study reveals how big data and machine learning uncover complex links between food chemistry, flavor and consumer perception, and lays the foundation to develop novel, tailored foods with superior flavors.

A comprehensive two-scale model for predicting the oxidizability of fatty acid methyl ester mixtures

The intricate mechanisms of oil thermooxidation and their accurate prediction have long been hampered by the combinatory nature of propagation and termination reactions involving randomly generated radicals. To unravel this complexity, we suggest a two-scale mechanistic description that connects the chemical functions (scale 1) with the molecular carriers of these functions (scale 2). Our method underscores the importance of accounting for cross-reactions between radicals in order to fully comprehend the reactivities in blends. We rigorously tested and validated the proposed two-scale scheme on binary and ternary mixtures of fatty acid methyl esters (FAMEs), yielding three key insights: (1) The abstraction of labile protons hinges on the carrier, defying the conventional focus on hydroperoxyl radical types. (2) Termination reactions between radicals adhere to the geometric mean law, exhibiting symmetric collision ratios. (3) The decomposition of hydroperoxides emerges as a monomolecular process above 80 °C, challenging the established combinatorial paradigm. Applicable across a wide temperature range (80 °C to 200 °C), our findings unlock the production of blends with controlled thermooxidation stability, optimizing the use of vegetable oils across applications: food science, biofuels, and lubricants.