Abstract | Keywords | Export | Availability | Bookmark

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Increasing electrification and renewable energy generation are part of the measures foreseen by policymakers to achieve climate goals. However, these intensify the stress on the electricity distribution system, where a significant amount of renewable energy sources, electric vehicle chargers and heat pumps are and will increasingly be connected. As such, distribution networks that were suitable or even oversized for the needs of the past face higher risk of congestion and power quality issues. Consequently, the way the system is managed is evolving to deal with these new challenges. In modern grid management paradigms, digital models of the system are required to analyze its operating conditions and plan or deploy reinforcement or corrective actions, if necessary. However, in practice, some physical characteristics of distribution networks, e.g., user phase connectivity and cable impedances, are often not known to sufficient detail. This is mostly due to historical management reasons.Fortunately, today's interest in the role of the distribution system within the energy transition is pushing institutions and utilities to actively measure system properties. The value of these measurements is manifold, but in this work they are used for two objectives: monitoring the operating conditions of the system, and deriving some of its unknown/poorly known parameters. This is done via a number of mathematical optimization methods that use smart meter data as only measurement input. The core of the work consists of a generic framework to create unbalanced static state estimation models. These models are designed to cope with the challenges of practical distribution system state estimation implementations, such as the paucity of available measurements. The mathematical models in the framework are successively extended upon to discover the phase connectivity of low voltage users and the length and impedance values of the networks' cables. All developed methods are successfully tested both on synthetic data and on a real system.