Thermal energy counts for a large part of the total energy consumption. To reduce fossil fuel consumption for heat and cold generation, different low temperature heat sources have been considered. Water networks have been considered as a large amount of water flow through it. To measure the thermal potential of the system, this paper provides a method in unsteady state to determine water temperature and flow in large water systems made of buried pipes. The model has been applied to a raw water supply system made up of 5000 km of piping and carrying 200 million m3 annually situated in the south of France. Water temperature is calculated considering heat exchange and the spatial specificities of the network (diameter of the pipes, depth, type of soil ...). Soil and water temperature measurements have been made to validate the model values. The model can predict water flow and temperature according to time with good accuracy: maximal error of 10% on the flow is obtained, the root mean square error on the calculated temperature is 0.84 circle C, and the correlation coefficient between the calculated and the measured temperature values is 0.98. The impact of adding several heat (or cold) injections in the system has been evaluated with the model. After a 2 MW heat exchange, the water temperature is increased by at least 1circC for 10 km downstream the exchange.