Future power systems will be dominated by power electronic (PE) converter connected generation. This trend is primarily driven by the increasing penetration of wind power as well as the increasing integration of photovoltaic (PV) based generation. In general, these PE converter connected generation sources are expected to change the dynamic behavior of power systems. In particular, conventional phasor-based protection systems may be challenged by the reduced inertia and reduced short-circuit power, due respectively to non-synchronous generation and limited short-circuit power of PE converters.

Therefore, it has been proposed to utilize time-domain protection principles, which are more sensitive and operate faster than phasor-based protection functions. For instance, travelling-wave (TW) based differential protection analyses the magnitude, polarity and arrival time of the fault generated TWs when they reach the line terminals. This type of protection functions requires high sampling rates in the megahertz range and is characterized by fast cycle times. Today’s modern substations have digital secondary systems consisting of process-level networks and highly functionally integrated protection Intelligent Electronic Devices (IEDs). The integration of high sampling rate protection applications in digital substations is challenging, since they increase significantly the communication load on process-level networks as well as the computational load on highly functionally integrated IEDs and centralized protection platforms.

This thesis aims to test the hypothesis that a distributed signal processing architecture can provide a scalable integration of high sampling rate protection applications in digital substations without increasing vastly the communication load on process-level networks and computational load on highly functionally integrated protection IEDs. Therefore, the thesis proposes that the most computation- and communication-demanding signal extraction tasks of time-domain protection functions be allocated to a process-level device, termed Distributed Signal Processing Units (DSPU). Moreover, suitable data models for the protection signal features are derived based on the IEC 61850 modelling approach as well as their mapping to IEC standard compliant communication protocols is discussed. In addition, the results show that the communication load may be reduced significantly by the proposed distributed signal processing architecture, and also that an increased number of DSPUs can be connected to the same process-level network. Furthermore, as part of this thesis, a detailed design of the DSPU is developed, which has been optimized with respect to the processing delay and the computational costs. Finally, the proposed substation architecture is verified through electromagnetic transient (EMT) simulations.