Uncertainty analysis of weather controlled systems

The indoor climate of many storage facilities for agricultural produce is controlled by mixing ambient air with the air flow through the store room. Hence, the indoor climate is affected by the ambient weather conditions. Given hourly fluctuating energy tariffs, weather forecasts over some days are required to effectively anticipate. Hence, typically a real-time optimal control strategy results. As weather forecasts are uncertain, predicted model outputs and related costs of the control strategy become uncertain. Usually, a medium-range weather forecast for a period of some days consists of an ensemble of forecasts. Hence, the uncertainty in the weather forecast is known a priori. In addition to this, in past-performance studies where weather forecasts and observed weather variables are given, an a posteriori evaluation of the forecast errors can be made as well. The objective of this study is to evaluate the uncertainty in the costs related to weather forecast errors and uncertainty, given the control inputs. In a simulation case-study with real weather forecasts and observed weather, it appeared that only slight cost increases can be expected due to errors and uncertainties in weather forecasts if the optimal control problem is calculated every 6–12 h in a receding horizon context.