Production planning of energy systems

This dissertation is a collection of research articles that assess economic and operational risk in production planning of district heating. District heating systems are typically coupled to the electricity system through cogeneration and power-to-heat technologies, and production planners must account for uncertainty stemming from changing weather, demands and prices. Years of high-resolution data from the district heating system in Aarhus, Denmark have been used throughout the project to model the system and estimate uncertainties. Risk management tools have been developed to aid district heating operators and investment decision makers in short-, medium- and long-term production planning. Short-term production planning involves commitment of production units and trading on the electricity markets and relies on forecasts of the heat load. Weather predictions are a significant source of uncertainty for heat load forecasts, because the heat load is highly weather-dependent. I introduce the method of ensemble weather predictions from meteorology to heat load forecasting and create a probabilistic load forecast to estimate the weather-based uncertainty. Better estimates of the weather-based uncertainty can be applied to optimize supply temperature control and reduce heat losses without compromising security of supply in heat distribution systems. Consumer behavior is another substantial, but difficult to capture, source of uncertainty in short-term heat load forecasts. I include local holiday data in state-of-the-art load forecasts to improve accuracy and capture how load patterns change depending on the behavior of the consumers. A small overall improvement in forecast accuracy is observed. The improvement is more significant on holidays and special occasions that are difficult to forecast accurately. In medium-term production planning, there can be substantial economic potential in performing summer shutdown of certain production units. The shutdown decision carries significant risk, due to changing seasonal weather patterns. Based on 38 years of weather data, the uncertainty on the timing of the optimal decision is estimated. This information is used to develop practical decision rules that are robust to rare weather events and capable of realizing more than 90% of the potential savings from summer shutdown. Long-term production planning decisions regarding investments in future district heating production systems are affected by uncertainty from changing electricity prices, fuel prices and investment cost for technology. The effects of these uncertainties on a cost-optimal heat production system are explored, using well-established production and storage technologies and extensive multivariate sensitivity analysis. The optimal technology choices are highly stable and, taxes aside, large heat pumps and heat storages dominate the cost-optimal heat production systems. However, the uncertainty on the exact capacity allocation is substantial. Excluding heat production based on fossil fuels increases the uncertainty on the system cost, but drastically reduces the uncertainty on the optimal capacity allocation.