District heating networks are important infrastructures to provide high efficient heating and domestic hot water to buildings located in urban areas. Modern district heating networks may involve the use of waste heat, renewable sources and heat from cogeneration thermal storage systems. In addition, management is operated through advanced ICT solutions able to minimize the global primary energy consumption and to increase end user awareness.
Detailed thermo-fluid dynamic simulation tools can be of extreme importance for the optimal management of modern district heating networks. Some of the issues that simulation tools are requested to face are: peak shaving, selection of the operating temperature, operation in the case of malfunctions, storage management. An important requirement consists in the possibility to perform fast simulations, even in the case of complex networks.
This paper aims at presenting a detailed simulation approach that can be applied to large district heating networks. The entire network is represented as constituted by the main pipeline, which may be a tree shaped or a looped network, and various tree shaped subnetworks that distribute water from the main network to each single building. The main pipeline is fully modeled considering fluid flow and transient heat transfer. Subnetworks are simulated using a reduced model obtained from the full model.
This modeling approach is applied to the analysis of transient operation of the Turin district heating network. The thermal request of the users is obtained from temperature and mass flow rate measurements at the thermal substations, available each six minutes. Thermo-fluid dynamic simulation allows one obtaining the corresponding thermal load profiles at the various thermal plants. Results show that a peak request is caused by the temperature reduction in the entire system due to the small thermal request at night. Due to the advective transport of water in the network and the thermal losses, the shape and amplitude of the peak at the plant is completely different than that at the users.
A comparison between simulations and experimental results shows that the model is able to predict the network operation with good accuracy. Using this simulation approach it is therefore possible to examine the effects of variations, obtained through night attenuation or the installation local storage systems, on the thermal request profiles of some of the users on the global thermal load of the network during the start-up transient.
The proposed simulation approach is shown to represent a versatile and important tool for the implementation of advanced management to district heating systems.
Operational thermal load forecasting in district heating networks using machine learning and expert advice
