Municipal Heating Grid Load Predictions for Improved Control of Heating and Cogeneration Plants Aneta Strzalka 1 , Jacek Kalina 2 , Rafal Strzalka 1 , Ursula Eicker 1 1 Stuttgart University of Applied Sciences, Stuttgart, Germany 2 Silesian University of Technology, Gliwice, Poland Abstract Presently, there are substantial financial efforts to improve heating plants’ control and system optimisation based on heat load predictions. In this paper two methods for heat load predictions in urban areas are demonstrated. The first one is an in-house Urban Energy Platform called SimStadt, which is based on 3D CityGML buildings models and its characteristics. The second approach uses long-term measurement data processed by an Artificial Neural Network. Both methods show accurate results on the annual basis at citywide scale. Furthermore, high correlation between measured and predicted hourly load profiles enables forecasts to be used as guidelines for efficient energy generation, which will increase overall system efficiency. Introduction Aim and approach Resource efficient, price competitive and low-carbon district heating systems require rational decisions on operation of productive assets. Therefore, great effort is being undertaken nowadays to make district heating systems more effective. Important present research activities are oriented toward improvement of heating plants control and system optimisation based on load predictions. With load forecasts, the plant operator is able to decide heating unit dispatch, generation profile, grid forward temperature and storage operation in order to satisfy a given optimization criteria, i.e. objective function. However, the critical issue of the optimisation task is accurate prediction of head demand for at least one day ahead system operation. Planning of heat production and storage as well as equipment dispatch are of high importance especially in the case of cogeneration plants offering electricity on the balancing market. This is due to significant impact of plant operation characteristics on plant economics. In this paper two different techniques for heat load predictions in urban areas are presented and the main problems related to their development are discussed. Special emphasis is put on the development of an urban area-wide 3D model-based heat load forecasting method, which would enable predictions of heating network hourly load profiles. The in-house Urban Energy Platform, called SimStadt (SimStadt, 2018), uses models of all buildings within the studied areas based on the 3D CityGML standard (Gröger et al., 2012). Firstly, some results for a German case study area, Scharnhauser Park (SHP), are shown, as the methodology was developed using this area. This method was then tested on the City of Krosno, Poland, to quantify transferability to different climates. The numerical modelling results are compared with an Artificial Neural Network (ANN) trained on the long-term historical data. In order to verify the models, both techniques were applied to the municipal heating system of the City of Krosno, Poland, where detailed and long-term measurement data were available. Scientific Innovation and relevance According to Heat Roadmap Europe (2017), Europe consumes half of its final energy for heating and cooling purposes. Around 9% of this is delivered by district heating systems. Considering that typical energy conversion efficiency from a heating plant is around 80% to 90% depending on technology, grid losses are around 10% to 15% and there are additional energy losses at buildings that depend on system performance, one can conclude that losses account for a substantial portion of European primary energy input, approximately 1,542.7 million tons of oil equivalent (Eurostat, 2016). According to UNEP Report (2017) systemic inefficiencies result in massive economic and social costs, and act as a major barrier to universal access to modern energy. Therefore, new concepts and techniques to optimise district heating systems and to balance energy load and generation are needed. In this work an innovative approach, similar to that of Nouvel et. al (2017) and Monien D. et al. (2017) is presented. It enables analysis and simulation of heat demand profile of an entire urban area. Presently, most of commonly employed methods are restricted only to a limited number of buildings with sufficient data. This is mainly because detailed data acquisition at the city-scale is very difficult and even unrealistic. Therefore, in this paper a 3D model-based method using very few input data is presented and validated using real measurement data. The presented approach is also innovative as the weak point of most city-scale heat demand forecast techniques is the lack of a validation process (Strzalka et. al, 2011). Case Study Areas Scharnhauser Park The German case study area, Scharnhauser Park (SHP) is a mixed residential-commercial area that is located on the southern border of Stuttgart, Germany, covering 150 hectares and with 10,000 inhabitants. The area is a former ________________________________________________________________________________________________ ________________________________________________________________________________________________ Proceedings of the 16th IBPSA Conference Rome, Italy, Sept. 2-4, 2019 4546 https://doi.org/10.26868/25222708.2019.210562