<p>Distributed generation, in the form of small-scale photovoltaic installations, have the potential to reduce carbon emissions created by, and alleviate issues associated with, centralised power generation. However, the major obstacle preventing the widespread integration of small-scale photovoltaic installations, at a residential level, is intermittency. This thesis addresses intermittency at a household/small community level, through the use of "nanogrids". To date, ambiguity has surrounded the nanogrid as a power structure, which is resolved in this thesis through the derivation of concise nanogrid definition. The nanogrid, a power distribution system for a single house/small building, is then used to implement demand side management within a household. This is achieved through the use of a hybrid central control topology, with a centralised coordinating controller and decentralised control nodes that have the ability to sense and modulate power flow. The maximum power point tracker is used to observe the available photovoltaic power, and thermostatically controlled loads present in the household are manipulated to increase the correlation between power production and consumption. An algorithm is presented which considers the expected power consumption of the thermostatically controlled loads over a 24 hour period, to create a hierarchical ratio. This ratio determines the percentage of available photovoltaic power each load receives, ensuring the loads that are expected to consume the most power are serviced with the largest ratio of photovoltaic power. The control system is simulated with a variety of household consumption curves (altered for summer/winter conditions), and a week of realistic solar irradiance data for both summer and winter. In each simulated scenario, a comparison was made between controlled and uncontrolled households to ascertain the extent grid power consumed by a household could be reduced, in turn reducing the effect of intermittency. It was found that the system had the ability to reduce the grid power consumed by as much as 61.86%, with an average reduction of 44.28%. This thesis then explores the concept of interconnecting a small community of nanogrids to form a microgrid. While each nanogrid within the network has the ability to operate independently, a network control strategy is created to observe the possibility of further reducing grid power consumed by the community. The strategy considers the photovoltaic power production and thermostatically controlled loads operating within the network. A ratio of the network's photovoltaic power is distributed to the thermostatically controlled loads, based on their expected consumption over a 24 hour period (highest consumption receives largest ratio of power). This was simulated with a range of household cluster sizes, with varied consumption patterns, for a week with summer/winter solar irradiance. The tests show that, compared to an uncontrolled nanogrid network, the combined control can reduce grid power consumed by as much as 55%, while a 7% decrease is seen when comparing the combined control to the individually controlled nanogrid networks. When compared to an uncontrolled individual house scenario, the combined control interconnected nanogrids can reduce the power purchase from the grid by as much as 61%.</p>