Algoritm för smart laddning av elfordon baserad på prognostiserad solelproduktion : Ökad självkonsumtion av solel samt minskat elnätsberoende

Sammanfattning: Due to the environmental issues, the amount of installed solar power increases. In the same time, the electric vehicle fleet is expanding rapidly. Those two growing technologies, if not controlled, can cause various unwanted effects for the electricity grid. In order to decrease their negative effects on the grid and benefit from it at the same time, these technologies have to work in synergy with each other. This synergy can be enabled through smart charging of electric vehicles. Therefore, the aim of this study is to develop a smart charging algorithm which uses solar production forecasts to charge the vehicles at a workplace. Furthermore, the goal is to examine how such control of the charging affects the self-consumption of solar power, self-sufficiency and the amount of energy imported from the grid as opposed to uncontrolled charging. To fulfill the goal, the algorithm was developed based on solar production forecasts. The forecasts were created through autoregressive models, AR and ARMA which were estimated using the actual solar production data collected at one of Uppsala regions solar production plants. Also, a case where ideal forecasts were used was applied. Furthermore, the charging need for various number of cars was simulated for every working day throughout an entire year in order to simulate the application of the algorithm and examine its performance but also to simulate the uncontrolled charging. The results, compared to the uncontrolled charging, show that the algorithm is able to increase the self-consumption of solar power by an average of 9,33 – 25,30 percentage points for 10 – 50 charging cars. It is also able to increase the selfsufficiency by an average of 42,65 – 31,28 percentage points for 10-50 cars respectively thus reducing the need of electricity imports from the grid. Furthermore, it was discovered that the results, the self-consumption and selfsufficiency, from the simulations with ideal forecasts differed only by up to 2 percentage points from the simulations where the forecasts were created through an AR(9) model (AR model of order 9). This allows a conclusion that a simple AR(9) model is completely sufficient to create forecasts that are good enough to produce satisfactory results. In general, it is concluded that the algorithm developed in this study is successful when it comes to increasing the self-consumption of the solar power, the selfsufficiency and decreasing the amount of energy needed from the electricity grid. This limits the negative impacts that the increasing solar power production and the growing electric vehicle fleet have on the electricity grid.