Real-time control and scheduling decision making in fermentation-based food process operations

With government policy encouraging decarbonisation, the New Zealand food processing industry is rethinking its process operations and their improvement. This work focuses on identifying opportunities to decarbonise industrial sites using simultaneous multi-process control and scheduling optimisation methods. A common issue for multi-process sites is the disconnection between the unit operation level process controls that operate under closed loop regulatory, or advanced regulatory, principles and the scheduling layer that ideally operates under real-time optimisation principles. As the first step towards solving this issue, an illustrative case of a cream cheese production process consisting of a pasteurisation unit, two fermentation units and a separation unit are studied by constructing a plant wide scheduling model in Python, integrated with a fermentation model, to explicitly model the interactions between two layers. Compared to typical plant operation, controlling the dosing of substrate in the fermenter allows for more uniform fermenter end times leading to improved scheduling of successive batches which results in less discarding of batches due to poor quality. Future work will embed the trade-off between overall energy usage, product quality and product throughput constraints (supply chain) within a holistic objective function centred on carbon footprint.