Modelling socio-technical systems in which a population of heterogeneous agents generates demand for infrastructure services requires a synthetic population of agents consistent with aggregate characteristics and distributions. A synthetic population can be created by generating individual agents with properties and rules based on a scenario definition. Simulation results fine-tune this process by comparing system level behaviour with external data, after which the emergent behaviour can be used for analysis and optimisation of planning and operation. An example of electricity demand profiles is used to illustrate the approach.