With the evolution of smart grid paradigm and the consideration of demand side management (DSM) as one of the flexibility providers in networks with renewable generation, the accurate assessment or prediction of demand profile and advanced DSM are becoming essential. This thesis is contributing to both, an accurate demand profiling and advanced use of DSM to facilitate flexible and secure network operation. It starts by discussing the data and information needed in the future distribution network to facilitate flexible network operation. It then illustrates the benefits of using advanced data mining techniques (artificial neural networks) for better observability of demand in the distribution network with a limited number of smart meters. The first part of the thesis thus illustrates how the flexibility and composition of aggregated demand can be assessed/forecast with very limited information coming from the end users. Once the composition of demand is available, one can assess with high confidence what portion of demand is flexible, what types of load that portion includes (e.g., cold appliances, heaters, etc.), and when and where (at which buses in the network) it should be shifted/curtailed. This enables âtailoringâ the DSM program and incentive system to the available size and type of flexible loads in an area. At the same time, it allows a more confident prediction of the outcome of the DSM program (the resulting load curve). Furthermore, it facilitates indirectly a more accurate modelling of demand over a period of time. The second part of the thesis focuses on the use of the information about demand composition, which is first used to model load at each network bus as a composite load model, and then to study different effects of DSM on network operation. Wide-scale DSM involving numerous flexible load buses in the network changes not only the total demand in that area at given time, but also its composition at individual buses, i.e., the shares of different components of the composite load model. This change in demand profile could influence both, the steady state network operation (critical network loading, losses, etc.) and its dynamic performance (voltage and angular stability of the system following a disturbance). Therefore, the second part of the thesis demonstrates how DSM program can be optimally planned hours, or day ahead, across the network, taking into account forecast demand composition and demand flexibility at each bus, in order to meet the requirements of the network operator (e.g., facilitating efficient use of available renewable resources), and at the same time maintain the relevant steady state and/or dynamic performance indicators of the network at the level they were before deployment of the DSM program.