Flexible energy management in an uncertain greenhouse

Summary

Industrial greenhouses can use a large amount of electrical power and natural gas
to run light and fire boilers throughout the year. This energy usage is often
organised using predictions for the coming days that are derived from mathematical models. These mathematical models are made to represent how a greenhouse might operate and are employed with forecasts of the weather to make predictions of how a greenhouse might be run in the future. However these models and weather forecasts are often prone to errors. These errors impact the accuracy of subsequent predictions and impact the greenhouse’s overall energy efficiency. In response to this, this thesis analyses to what degree these errors impact prediction accuracy, how this impacts the economy of the greenhouse and which factors are the largest contributors.