Contributions to the development and simulations of generic, modular and multiphysics greenhouses dynamic models, evaluated with a whole year study case dataset.

Abstract

North-Western Europe heated greenhouses need to address their high fossil energy dependency while they start to face with climate change. In this context, physics-based greenhouses dynamic models can be used for prospective assessments of innovative shapes, equipment, control, etc. A prerequisite for such model-based evaluation is a review of existing implementations, the improvement and development of suitable sub-models in a generic and modular approach that do not require calibration, and the evaluation of a use case global model with a solid experimental dataset. First, this paper details contributions to existing models regarding several aspects: solar gain, boundaries effects, airflows and leakages, heat and mass transfer. The second part is dedicated to the evaluation of an experimental tomato greenhouse global model for an 11-month period. Its assessment is multiphysics: indoor climate, utilities consumptions, yield and Leaf Area Index. The resulting 5 min sampling indoor air climate Root Mean Square Error is 1.3 °C (temperature) and 8.1%RH (relative humidity). The tomato yield Mean Absolute Error is 1.1 kg m-2. Mass balances also quantify the losses and potentials for water and CO2. The global model outputs are compared with literature, and it is demonstrated that assessing the accuracy of models based only on statistical indicators is questionable. This approach is compatible with assessments of prospective solutions, it increases the confidence for scaling results from small to large commercial greenhouses, and it constitutes a base from which simpler and black box models can be derived for other applications such as predictive control.