Physics-Based Modelling of a Milk Cooling System for Intelligent Energy Management

Forecast-based energy management can play a largerole in a smarter and more efficient use of renewableenergies based on demand side management. Usingapproaches such as model predictive control, individualconsumption devices can be shifted within operationconstraints so that their electricity consumptionoptimally matches generation. In agriculture, largethermal storages make up a sizeable part of electricityconsumption, and offer a potential use in the short termshifting of demand. Necessary for this are accuratemodels to forecast behaviour of such dynamic systems,so that minimal power demand and fulfilment of operationconstraints can be ensured when computing optimalcontrols. This work focuses on the physics-basedmodelling of a milk cooling storage through parameteridentification on real measurement data. Emphasizedare the derivation of a suitable model ODE with regardsto available data, and evaluation of the model ona rolling horizon. All major features of the measurementdata can be recreated by the model forecasts, andmodel performance values show errors of around 30%relative to mean temperature. Model performance isconsidered suitable for use in energy management atleast on short forecast horizons, while practicabilityon longer horizons is subject to further research.