To meet the net-zero emission target by 2050, Switzerland must install between 34 and 50 GW of distributed photovoltaic (PV). The inherent challenge with such an extensive integration of stochastic generation is to provide the flexibility for balancing supply and demand. Otherwise, grid reinforcements or PV generation curtailment are required. Both have a cost that can be reduced by using flexibility. Most of the PV capacity will be installed on rooftops. As the need for flexibility will mainly come from districts, this thesis investigates how to promote flexibility of PV systems for their large-scale deployment in low-voltage grids. First, we investigated how the citizens' and consumers' behavior can balance supply and demand. Our findings from a field trial highlight that the households' reaction to remunerative incentives is low but still observable. In contrast, new PV adopters, under an inherent moral and remunerative incentive, show a significant consumption behavior change. We evaluated that the households' potential for shiftable energy is around 18%. A 20% increase in the PV penetration can be achieved using this flexibility. Second, technical measures such as batteries, heat pumps, electric heaters, and PV power curtailment can effectively contribute to the systems' flexibility and mitigate network impact. We proposed strategies to promote flexibility and mitigate grid impact. In particular, we showed that variable volumetric tariffs promote large storage used for trading energy and increase the grid stress. In contrast, capacity and block rate tariffs reduce heat pumps' capacity, increase storage capacity to lower consumption peak, and increase PV curtailment to reduce injection peak. The consequence is an overall grid stress reduction. We also showed that aggregating individual systems to form energy communities negatively impacts the grid but increases their profitability. Finally, flexible PV systems can be harvested by distribution network operators to keep the network in a safe state. We investigated the cost of exploiting distributed flexibility compared to grid reinforcement cost. The former is profitable for moderate PV penetration until a break-even point where the latter becomes the most economical option. Along with answering the central question of promoting PV systems flexibility, we provided an analytical tool to disaggregate households' smart meter measurements into appliance categories. We also provided two approaches to use smart meter data in the context of network impact studies while coping with privacy-preserving regulation.