The goal of this thesis is to analyze the nature of the energy demand of the individual divisions at the voestalpine site in Donawitz. The findings should be used to identify the main drivers for balancing energy and to develop new calculation models to help reduce balancing energy costs. Also, a tool is to be programmed that can calculate demand forecasts with little user interaction. The base for this thesis will be research on the Austrian energy market, taking into account the changes from January 2012, while focusing on the aspect of balancing energy. Possible ways of generating demand forecasts will also be explained and highlighted. To analyze the energy demands on site further, the parties with the highest usage will be identified and characteristic of their use will be investigated. Then all available production and energy data will be described an evaluated in terms of quality. Plan- and forecast-data will be considered in this evaluation. Based on this data, correlations will be investigated by employing statistical methods, process knowledge of employees and process documentations. From these correlations, calculation models will be derived with the goal of increasing the forecast quality of the energy demands. All increases in accuracy mentioned will be based on planning on a daily basis. Besides increasing accuracy another goal was that the tool should be as simple and easy to understand as possible to allow employees to make adjustments without much training. For the blast furnace and the steel division increases in accuracy could be achieved in the magnitude of 50 and 70 percent respectively. Overall an improvement of 10-15 percent could be achieved. This is comparing the new and automatically generated forecast with the manually edited one that had a lot of extra information at its disposal. That makes the actual, possible improvement hard to evaluate but it is believed that the increase in accuracy will be even higher once manual edits are introduced. Another significant improvement comes from the planning tool itself. Not only does it calculate a demand forecast quickly and easily, it also allows the user to manually edit the forecast efficiently and saves the data for evaluation the next day. It automatically creates charts to show each of the main demands and compares them to the actual power draw over the last 24 hours. This allows the user to easily identify where a possible deviation occurred and includes them in a continuous feedback loop without having to do an analysis themselves each and every day. Concluding remarks will show as of yet unexplored potentials and give suggestions for future endeavors regarding the aspects of methods and instruments as well as quality and information management.