Novel Methods for Loss of Mains Protection

Small-scale generation connected to distribution networks has increased significantly in recent years. This trend is driven by developments in distributed generation (DG) technologies, environmental concerns and economic reasons. The diffusion of generation into the distribution network level has many potential benefits, but it also raises challenges, such as unintentional islanding, which is hazardous to the safety of both personnel and equipment. Due to the safety risks, all DG units need to be equipped with a loss of mains (LOM) protection scheme capable of rapidly detecting and stopping islanding. LOM protection methods can be divided into passive, active and communication-based methods. Passive methods rely on detecting islanding by monitoring chosen system quantities. These methods are affordable and applicable to all types of DG units, but their performance is highly dependent on the local power imbalances between the production and consumption in the islanded zone. Most, if not all, passive methods, fail to detect islanding if the local production closely matches the local consumption. The set of power imbalance combinations that lead to non-detected islanding is referred to as the non-detection zone (NDZ). Active methods are based on deliberately injecting small perturbations to the grid and monitoring the response of the system. These methods generally have smaller NDZ than passive methods. However, this comes at the cost of degraded power quality. Communication-based methods rely on other means than the local monitoring of system quantities, which makes them immune to the NDZ problem. However, these methods tend to be costly.

The performance of passive and active methods can be improved by applying more sensitive LOM protection settings. However, if the LOM protection settings are too sensitive, voltage dips caused by faults in the transmission grid may result in a cascading disconnection of DG. In order to avoid such risks, which threaten the system’s stability, many grid codes include fault-ridethrough (FRT) requirements, which specify the depth and duration of voltage dips which DG units need to be able to withstand. FRT requirements often also require the DG units to feed reactive current to the grid during the voltage dip in order to support the system voltages. The work conducted for this thesis indicates that FRT requirements significantly degrade the performance of LOM protection. This thesis also studies how the type of the protected DG unit affects LOM protection. The frequency of an islanded circuit sustained by a directly-coupled synchronous generator is determined by the local active power imbalance, whereas the frequency of an islanded circuit sustained by a converter-coupled DG unit is determined by the local reactive power imbalance. However, when there are both directly-coupled synchronous generators and convertercoupled DG units in an islanded circuit, the synchronous generator seems to dominate these relationships. This has significant implications on the performance of active LOM protection schemes.

One of the main issues distribution system operators face when they are evaluating the adequacy of LOM protection for DG installations is the lack of suitable analysis tools. This thesis proposes a novel LOM risk management procedure which utilizes the existing analysis tools embedded in a modern network information system (NIS). This NIS-based procedure analyzes what kind of power imbalance combinations are possible in the studied network sections. Based on the possible combinations of power imbalances and predefined NDZ mappings of optional LOM protection schemes, the procedure tells protection engineers if there are any risks of non-detected islanding in the analyzed network sections and proposes which LOM protection schemes would be most suitable for each DG installation. Although the proposed LOM risk management tool is presented at the concept level only here, it is clearly a promising area for future research.

Two active LOM protection methods and one communication-based protection automation concept were also developed during this thesis work. The first of the active LOM protection methods is based on forcing the frequency of an islanded circuit out of the utilized frequency thresholds by constant injection of reactive power pulses and a dedicated reactive power versus frequency droop. The knowledge gained during the development of this method resulted in a second, significantly more advanced, active LOM protection scheme. This is based on forcing the rate-ofchange-of-frequency of an islanded circuit to a desired value by applying a dedicated reactive power versus frequency droop. This method is able to detect islanding rapidly and reliably even if the local power imbalances are negligible. Moreover, this can be achieved with a very modest injection of reactive power. The communication-based protection automation concept is designed to solve typical DG related protection challenges and to automatically change the feeding path of the protected DG unit in case if the original feeding route becomes faulted. However, the successfulness of the automatic feeding path changing depends on many factors such as DG unit type, network parameters and the momentary input power provided by the primary energy source.

The methods developed in this thesis have slightly different purposes. The proposed NIS-based LOM risk assessment procedure is useful for evaluating the adequacy of existing LOM protection as well as for choosing optimal LOM protection schemes for new DG installations. If the LOM risk assessment procedure indicates that the local power imbalances will always be very large, then passive LOM protection schemes are a sensible choice. However, should the LOM risk assessment procedure reveal that the local power imbalances could be so small that reliable LOM protection cannot be ensured with passive LOM protection schemes, then active or communication-based LOM protection schemes are preferable. Active LOM protection schemes are suitable if the ratio of converter-coupled to directly-coupled generator capacity in the analyzed zone is large. This is because certain active LOM schemes, such as the one proposed in this thesis, are able to detect islanding reliably and rapidly even if the local active- and reactive power imbalances would be negligible, provided that the ratio between converter coupled to directly coupled synchronous generator capacity is large. However, if a significant proportion of the generation capacity in the analyzed network section is synchronous generator based, then sensitive and rapid LOM protection cannot always be guaranteed. In such cases, it is advisable to utilize advanced communication-based LOM protection schemes which are immune to the NDZ problem.