DNOs are presently aware of the potential effects of LV network loads but unable to quantify the impact these have on the voltage characteristics, phase balance and power flows on distribution feeders. Failure to contain these may result in regulatory limits being violated, quality of supply compromised or assets being aged prematurely. To support understanding this, tasks for WP1 are as follows:
Network model development provides later activities with the benchmark fully observed power networks representative of the problem cases AMIDiNe will address. This will include a ‘private wire’ LV arrangement for testing the disaggregation of storage and LCT, a regional MV network for hierarchical forecasting and a number of urban and rural LV feeder models for loss and state estimation studies.
Relation between reactive power and voltage magnitude changes – the depletion of conventional generation on the transmission network and its general absence on the distribution network predicates an alternate means of reactive power absorption to control voltage levels. State Estimation tools that can approximate this functionality at MV level would not be able to deal with the non-linear and non-Gaussian behaviours exhibited by the loads observed at LV level – this hypothesis will be tested and used to develop alternative state estimators that are based on high dimensional Copula. Quantification of flexibilities, translates into pricing for services that may be offered to system operators for balancing the system or maintaining grid stability at a sub-regional level.
Premises Phase Identification and Balancing – the ’last mile’ of networks often have been constructed in such a way that the actual allocation of premises to phases is not known or balanced. Smart Metering coupled with substation monitors could deal with this problem by utilising outputs from WP2 to disaggregate phase allocations from multiple premises as well as combinatorial optimisation procedures to identify plausible rebalancing scenarios.
LV loss characterisation with respect to network footprint, phase balance and end user load behaviour – this work package will utilise the representations developed in WP2 along with partner distribution network topologies to identify how losses can still be approximated with the absence of portions of data reflective of operational scenarios. Dirichlet type compositional models to characterise phase imbalance and its seasonal evolution will be developed alongside the load forecasting work in WP3.