Optimal expansion planning of electric vehicle fast charging stations

With the penetration rate of electric vehicles (EVs) increasing exponentially, the high charging load may cause new issues in future power system, e.g., voltage drop. Besides, EV Charging stations (EVCSs) located in inadequate places exacerbate these issues, causing charging demand to concentrate on a few EVCSs. Therefore, the charging demand estimation for EVCSs and their strategic placement is essential for the system operators and charging station owner. The paper proposes a method for optimal EVCS placement to achieve charging demand dispersion considering not only installation cost but also drivers' preferences and existing charging stations. To optimally place EVCSs, this work first estimates the charging demand of existing EVCSs based on kernel density estimation. The charging demand for new and existing EVCSs is modeled using the nearest neighbor search to consider drivers' preference for the nearer station. Next, the EVCS placement problem is formulated to minimize the peak charging demand using integer nonlinear programming, a non-convex problem. To tackle this non-convex problem, a minimax genetic algorithm is proposed, which is genetic algorithm combined with game theory. The validity and effectiveness of the proposed method are demonstrated through simulations based on real data from Jeju Island. After applying the proposed method, new EVCS placement is determined by analyzing the tradeoff between the degree of charging demand dispersion and installation cost. As a result, the charging demand concentrated on a particular EVCS is suitably dispersed.

Automated summary

With the exponential increase in the penetration rate of EVs, the charging load is causing new issues in the power system such as voltage drop. Improperly located EVCSs exacerbate these issues, leading to concentrated charging demand. This paper proposes a method for optimal EVCS placement considering installation cost, drivers' preferences, and existing charging stations. By estimating charging demands using kernel density estimation and modeling driver preferences, an integer nonlinear programming problem is formulated and solved using a minimax genetic algorithm. Simulations based on real data demonstrate the effectiveness of the proposed method in achieving dispersed charging demand.