Application of high-resolution domestic electricity load profiles in network modelling: A case study of low voltage grid in Denmark

The ongoing development towards electrification of the energy consumption together with large deployment of renewable energy sources creates new challenges of variability and fluctuation of the electricity supply and increases complexity of the network operation. In order to capture all the particularities of electricity demand and on-site generation, e.g. the short-term spikes due use of high electricity consumption appliances such like electric kettle, and get a full picture of network performance, a high-resolution input data are needed. This paper compares the business-as-usual network modeling with modeling when 1-minute domestic electricity demand and generation profiles are used as inputs. The analysis is done with a case study of low-voltage network located in Northern Denmark.
The analysis includes two parts. The first part focuses on modeling the domestic demands and on-site generation in 1-minute resolution. The load profiles of the household appliances are created using a bottom-up model, which uses the 1-minute cycle power use characteristics of a single appliance as the main building block. The profiles of heavy electric appliances, such as heat pump, are not included in the above-mentioned model, as they are closely related to the thermal properties of a building. Therefore, two type of single family houses equipped with heat pump are simulated in EnergyPlus with 1-minute time step. The PV generation profile is obtained from a model developed in Matlab environment. In the second part the generated profiles are inputted in a low-voltage network model created in DIgSILENT PowerFactory.
By means of employing 1 hour based demand and generation profiles in during dynamic studies, the representation of the local power system performance might sometimes not be as accurate as needed. In the test system employed in this case the simulation indicates that no stress is created in the grid. The loading of the transformer and power lines is 65% and 41%, respectively, which is below the limit of 80% of available capacity. The maximum voltage drop is 5.1% thus with the maximum allowed deviation of ± 10% and ± 6% according to standards and common practice, respectively. The same investigation, but with 1-minute input data, shows that the transformer is overloaded by 2% and the minimum voltage level is 0.922 % [p.u], which is below limits of common practice grid operation. When adding on-site PV on 50% of buildings, the loading of the transformer and power lines is reduced in the summer time to 58% and 51%, respectively. However, the power lines are stress with bi-directional power flow.
The results indicate that the business-as-usual approach to network modeling is not sufficient to capture the characteristic spikiness of the domestic load profiles and on-site generation. Hence the network overloading and high voltage deviations are not visible and the control strategies may be wrong.