Cost Minimization of Charging Stations With Photovoltaics: An Approach With EV Classification

Abstract

Abstract-This paper proposes a novel electric vehicle (EV) classification scheme for a photovoltaic (PV) powered EV charging station (CS) that reduces the effect of intermittency of electricity supply as well as reducing the cost of energy trading of the CS. Since not all EV drivers would like to be environmentally friendly, all vehicles in the CS are divided into three categories: 1) premium, 2) conservative, and 3) green, according to their charging behavior. Premium and conservative EVs are considered to be interested only in charging their batteries, with noticeably higher rate of charging for premium EVs. Green vehicles are more environmentally friendly, and thus assist the CS to reduce its cost of energy trading by allowing the CS to use their batteries as distributed storage. A different charging scheme is proposed for each type of EV, which is adopted by the CS to encourage more EVs to be green. A basic mixed integer programming (MIP) technique is used to facilitate the proposed classification scheme. It is shown that the uncertainty in PV generation can be effectively compensated, along with minimization of total cost of energy trading to the CS, by consolidating more green EVs. Real solar and pricing data are used for performance analysis of the system. It is demonstrated that the total cost to the CS reduces considerably as the percentage of green vehicles increases, and also that the contributions of green EVs in winter are greater than those in summer.

Adaptive Asymptotic Tracking Control of Uncertain Time-Driven Switched Linear Systems

“…Roughly speaking, we can divide optimization-based EV charging strategies into open-loop strategies (the control is a time-dependent scheduling profile calculated based on predictable operation of the system) and closed-loop strategies (based on feedback measurements): in the first family, [15] proposes a model predictive control approach with statistical EV arrivals and reduced computational complexity, while [16] proposes an EV classification scheme based on mixed-integer programming for a photovoltaic-powered charging station to reduce the cost of energy trading. The authors in [17] formulate the charging problem as an open-loop cost minimizing problem and an open-loop profit maximizing one.…”

“…However, (16) cannot be directly used because of the following three important problems: 1) An exact expression of the gradient ∇PE in (16) is not available, since the gradient depends by the dynamics in (9) which are affected by stochastic noise. 2) Moreover, in (16), theP matrix must remain positive definite. Thus, a constrained gradient descent must be implemented, increasing the complexity (due to the use of penalty function, generalized Lagrangian multiplier, etc).…”

PV-based EV charging station with vehicle-to-grid services for business premises

“…For the purpose, it takes into account the arrival and departure time, and X A,k and X D,k [123,138,156]. It calculates the urgency factor of charging from the remaining time and the remaining desired SOC to select the appropriate time-slots for the appropriate pool of EV.…”

“…Deploy BES [124,[127][128][129] Reduce losses [36,128] Deploy both [30,36,[125][126][127] Reduce unbalance [36,119] Reduce loading [36] Reduce costs [36,86,118,123,126] Reduce GHG emissions [86,117] Reduce transformer, feeder capacities required [36,129] Improve voltages [36,75,116,127] Improve QoS [36] Even though these papers form a comprehensive realm of studies regarding whether to augment grid or deploy PV and BES, the following limitations can be pointed out,  The papers above have assumed that either augmenting/reconfiguring the grid/phase or deploying onsite PV and BES is the best solution for reducing the impact, costs, GHG and increasing QoS. However, in practice, their various combinations (e.g., grid augmentation and PV; grid augmentation, PV, and BES; etc.)…”

“…The aim of the real-time EV load dispatching is to simultaneously reduce the impact and maximize the PV energy harvest by addressing the PV output variability [123,163], where the V2G service may or may not be incorporated. Therefore, real-time EV load dispatching can be performed 41 using two approaches namely 1) real-time EV load dispatching addressing the PV output variability without considering V2G services and 2) real-time EV load dispatching considering V2G services.…”