Decarbonisation and a transition towards sustainable energy systems in cities are key elements of the United Nations sustainability goal. Large-scale district heating networks sourced by excess heat or renewable energy allow to effectively transform building-related energy systems. This study proposes two different approaches for modelling load curves in large-scale district heating networks: 1) physics-based static energy balance model 2) data-driven regression model trained and adjusted on measured load curves. The load curves generated by application of these two approaches are compared with the actual load of an urban district heating network in Geneva, Switzerland. Both models allow to recreate the actual load curve of the district heating network, however with lower accuracy for higher time resolution in the case of the physics-based model. The physics-based static model can be used to simulate the demand and generate load curves of sufficient quality at monthly and daily resolution. For an hourly load curve, it is recommended to use the data-driven regression model if consumption data of the network is available.