Applying a top-down estimation of inertia in the Nordic Power System

The penetration of Variable Renewable Energy Sources (VRES) in the power system creates new challenges for the Transmission System Operators (TSOs). In the Nordic Power System (NPS), consisting of Norway, Sweden, Finland and eastern Denmark, all electricity is generated at the same frequency. When the system is in balance, meaning production and consumption is on the same level, the frequency is 50.00 Hz. It is crucial that the frequency remains stable. If the total load exceeds total generation, the frequency falls. Too much generation, and frequency rises. Conventional generation plants are connected to the power grid by means of rotating machines driven by a rotor, generating electricity at the desired frequency. Once in motion, the kinetic energy stored in the rotor will keep the generator running, even if the mechanical force driving it disappears. This effect is called inertia and is an essential inert part of the stability of a power system. VRES are not connected to the grid by rotating machines but through Power Electronics (PE), and do not contribute to the system's inertia. The increasing share of renewable energy sources leads to decreased power system inertia, and frequency stability becomes a concern. Therefore, the estimation of power system inertia has been of focus in the NPS in recent years. In order to secure frequency stability in the Nordic synchronous area, a new fast power reserve will be introduced. Fast frequency reserves (FFR) will complement the existing primary reserves for disturbances (FCR-D). To plan for future operation, the aim is to forecast the system inertia such that the instantaneous frequency minimum caused by the loss of the largest generator can be assessed, and sufficient fast reserves are procured. Since 2015, the Nordic TSOs have estimated and stored the value of system inertia by calculating the sum of inertia contributions from synchronous machines connected to the grid. This value is closely related to the amount of generation in the power system. This thesis applied a top-down estimation using linear models with production data as explanatory variables to estimate system inertia. Data from 2018 and 2019 on production and inertia were collected. Data from January 2018 to June 2019 were used to train models, and the models' performance was tested on data from July and August 2019. On the Nordic level, these models received a Mean Average Percentage Error of 2 % on the test set, revealing potential of using this approach to forecast inertia by using production forecasts per production type. The Nordic estimation of inertia level is a sum of estimations from Norway, Sweden, Denmark and Finland, so models were also tested using data from each country. This revealed that the Norwegian inertia estimation is already a top-down estimation. It was also found that the top-down estimation model is inaccurate in Denmark. The results from Sweden and Finland indicated that conventional power production, such as nuclear, thermal and hydro, was relevant to models while VRES production was irrelevant