Control, Optimisation, and Energy Analysis of Moderate Electric Field in Food Processing

The United Kingdom’s Net Zero 2050 target is driving all industries toward decarbonisation. The food and drink industry (FDI), one of the UK’s largest production sectors, relies heavily on energy-intensive processes such as fermentation, pasteurisation, and microbial inactivation, most of which rely on fossil fuel combustion. Consequently, this sector contributes approximately 26% of greenhouse gas emissions, equivalent to around 9.1 million tonnes of CO2 annually. Additionally, the UK’s rapid transition to renewable electricity generation is also encouraging the FDI to adopt electrically driven heating methods. In this context, Ohmic Heating (OH) has emerged as a promising alternative, offering rapid and energy-efficient heating. Among its configurations, Continuous Flow Ohmic Heating (CFOH) is particularly suited for pumpable products such as soups, sauces, and juices. Despite its potential, the adoption of CFOH in industry is limited by its complex, multi-physics nature, which involves coupled electrical, thermal, and fluid dynamic phenomena. Existing mathematical and Computational Fluid Dynamics (CFD) models provide partial solutions. Mathematical models often oversimplify system behaviour through linearisation, while CFD models, though accurate, are computationally expensive. This creates a clear need for modelling approaches that are both computationally efficient and capable of accurately capturing CFOH thermophysical behaviour. Achieving this requires a thorough understanding of the interactions between key process variables and product properties, forming the foundation for optimised control system design. To address this gap, this research first conducted a comprehensive literature review to identify the key parameters influencing CFOH performance. Based on these insights, a high-fidelity physical model was developed in MATLAB Simscape to simulate CFOH behaviour under varying product conditions. The model was validated against both established mathematical formulations and experimental data from a pilot-scale CFOH system, achieving a mean absolute percentage error (MAPE) of ±5% in terms of accuracy. Its ability to accurately predict temperature variations in response to changes in thermophysical properties makes it a robust platform for advanced control development. Utilising this validated model, three control strategies were designed and tested, including a conventional Proportional–Integral–Derivative (PID) controller, a Model Predictive Control (MPC) scheme, and a novel Neural Network-based Model Reference Control (MRC). These were implemented on a pilot-scale CFOH system at the Advanced Food Innovation Centre (AFIC), Sheffield Hallam University, and evaluated for temperature tracking accuracy, stability, and responsiveness. Results demonstrated significant improvements in process control and energy efficiency, with successful industrial trials on products such as sweet and sour sauce and tikka sauce, achieving process efficiency gains of up to 87% while maintaining product quality. However, insights were gained during the development stage of this work that CFOH performance is sensitive to the electrical conductivity, physical properties, and rheology of the product, particularly dynamic viscosity, which affects flow behaviour, residence time, and heating rate. Accurate viscosity prediction during processing is therefore essential for optimising system performance. To address this challenge, the study presents OhmNet, a Neural Network-based soft sensor capable of predicting the dynamic viscosity of tikka sauce during processing with a mean squared error (MSE) of 0.002, demonstrating exceptional predictive accuracy. The integration of OhmNet with advanced controllers enables optimised temperature regulation and power consumption, supporting sustainable and energy-efficient operation. Overall, this work addresses a gap in existing literature where, to the best of the author’s knowledge, no unified methodology exists that integrates physics-based modelling, real-time control implementation, and AI-driven soft sensing within food processing. Addressing this gap, the study delivers three key novel contributions:

1. development of an accurate physical model of CFOH that captures the coupled electrical, thermal, and fluid dynamics;

2. real-time implementation and comparative evaluation of multiple advanced control strategies; and

3. integration of a machine learning-based soft sensor for intelligent viscosity prediction.

By combining these elements into a validated, scalable workflow, this work delivers a robust solution for optimised, sustainable, and industrially relevant ohmic food processing. The framework directly supports the UK’s Net Zero ambitions while advancing the broader objective of low-carbon, energy-efficient food manufacturing.