Considering current environmental issues (e.g. climate change, air pollution) related to energy consumption, this thesis focuses on data characterisation and integrated optimisation methods for waste heat valorisation at the building, city and regional scales. As a first contribution, a method for the detailed characterisation of the energy demand of domestic hot water streams (shower, bath, washing up, etc.) in households at the building scale is presented. The energy consumption can be put in relation to the total heating demand of the building and, by spatial allocation and bottom-up data aggregation, of a district, a city or a region. It is demonstrated that with the construction of near-zero energy buildings and the improvement of the thermal envelope of existing ones, domestic hot water will represent in the future between 30 and 50% of the urban heating demand. The second contribution focuses on in-building waste water heat recovery. Residential waste water streams are characterised, and energy optimisation (based on pinch analysis) and investment cost calculations methods at the building and urban levels are presented. An integrated approach combining heat recovery, temperature optimisation and heat pump use leads to heating savings ranging between 28 and 41% in high efficiency single family and multifamily buildings, respectively. The final contribution is an MILP-based optimisation method for regional waste heat recovery specifically formulated for energy service companies. One of the main novelties of this work consists in the consideration of different energy prices depending on the type of heat sinks (urban or industrial). With the energy prices of 2015 in the Southern region of Luxembourg, waste heat can be valorised for heating demand at prices up to 25 €/MWh. At that price, profits of more than 10 M€/a from the transport of waste heat and the electricity production of combined heat and power systems are generated for the energy service company.