EVS29 Multi-objective optimization of an Autobahn BEV charging station supplied by renewable energy

Summary In order to address battery electric vehicles’ future ability to travel long distance this paper analyzes a sample case study of supra-regional charging, an Autobahn battery electric vehicle (BEV) charging station supplied by renewable energy. A tri-objective optimization of a local renewable energy system demonstrates how the charging station’s levelized cost of energy, life cycle emissions and stress on the electric grid can be reduced simultaneously by introducing a combination of partially curtailed photovoltaic generators and a battery electric storage system.


Introduction
Schill et al. demonstrated how the introduction of battery electric vehicles (BEV) in Germany increasingly stresses the electric distribution grid and leads to BEV-specific greenhouse gas (GHG)-emissions substantially higher than those of the overall power system, if not complemented by additional renewable energy generation [1]. A local renewable energy charging station must be designed to guarantee the coupling of BEV charging and renewable energy generation so as to both decrease life cycle emissions as well as mitigate stress on and the extension of the electric grid. While storage options play a vital role in the balancing of volatile renewable generation, the idea of "over-installation" of renewable energy in combination with its curtailment has been mentioned in the past as a potential efficient alternative to storage capacity but was left open for further discussion [2].
While current BEV's ranges generally do not allow long distance travels, it is expected that future BEVs will allow ranges of a few hundred kilometers [3], [4], making long distance travels possible, and thus requiring supra-regional charging options, like an Autobahn charging station. In fact, a supra-regional network of single fast charging stations has already been positioned in central Germany to serve the needs of long-range travel [5], [6]. This paper aims at offering a sample case study that addresses the challenges of transforming supraregional infrastructures to supply BEVs cost-efficiently and sustainably.

Methodology
In order to identify how a supra-regional charging station can be supplied with energy sustainably and costefficiently while at the same time mitigating stress on the grid, an exemplary renewable energy charging station system supplied by photovoltaic (PV) generators, a battery electric storage system (BESS) and an electric grid as the point of common coupling (PCC) is employed to supply a given electric demand of electric vehicles (see Figure 1). A computer model of the charging station is employed to assess and optimize the system's performance regarding levelized cost of energy (LCOE) minimization, minimization of the maximum power from the grid (P max ) and minimization of life cycle emissions (LCE).
Due to the anticipated conflict between these three objectives, the result of optimization is expected to be a three-dimensional optimal pareto curve that identifies the trade-off decision makers should be aware of during the design of the charging station and its components.

Simulation model
The simulation model aims at modelling the power flow between the charging station's components. It solves the energy balance with a one hour time resolution over one year to anticipate the system's performance for a planning horizon of 20 years. Data for component parameterization is listed in Table 1. = 7% = 1%/ • ecological = 800 kg CO 2 eq./ kW peak [11], [12] = 69 kg CO 2 eq./ kWh cap [13] = 569 g CO 2 eq./kWh el , To synthesize an electric load curve for the charging of electric vehicles fueling data of a mid-sized gas and diesel fuel station is transformed under the assumption that an equivalent electric charging station would be supplying a BEV fleet with the same amount of "distance travelled" per time unit. While this assumption is neglecting the fact that BEVs storages are not comparable to those of internal combustion engine vehicles, it accounts for the perception of peaks in charging load due to travel behavior that is assumed to be largely technology independent (high during midday and low to zero during the night, see Figure 2). Thus the historical data of fueled gas and diesel volume per time step can be transformed into an electrical charging load through the specific electric energy or fuel required to travel the same distance (0.078 l/km, 0.0681 l/km and 0.2 kWh/km for gasoline, diesel and electric energy respectively [15], [16] ). The electric energy supplied by the PV generator is simulated using a comprehensive PV model using measured timeseries for direct and diffuse radiation and considering location, azimuth and elevation angle of the generator surface [17]. Resource data are based on NASA SSE data (Surface Meteorology and Solar Energy SSE Release 6.0) [18]. The original data were converted to hourly resolution by the German Aerospace Center [19]. The simulated PV generators yearly energy yield amounts to 894 kWh/kW peak .
Power flow modelling of the BESS is based on energy balancing, taking into account charging and discharging efficiencies as well as the rate of self-discharge in each time step of the simulation. Lifetime of the BESS is determined using the post-processing model of Ah-throughput counting [20], which counts the amount of charge through the BESS. The end-of-life criterion is based on nominal charge throughput.
The charging and discharging of the BESS is guided by few basic rules. If the residual load is positiv (less PV generation than EV load), the share of power smaller than some threshold value P thr is taken from the grid (see region a in Figure 3). The difference between the residual load and P thr is then discharged from the BESS (b). In times where there is more PV generation than EV load, the energy is charged into the BESS (c) until the maximum SOC is reached, in which case the excess power is discarded (d) by curtailing PV generation. On the one hand this may not seem reasonable from an economic point of view as it decreases the overall yield of renewable energy, on the other hand however it serves the purpose of mitigating stress on the grid. In addition, while the assumption of complete curtailment of excess renewable energy is pessimistic it seems more realistic than complete feed-in of that energy into the grid for high systems penetration rates of renewable energy technologies. The point of common coupling is where the system's power flow balance is solved for each time step of the simulation and describes the power flow that is necessary to be provided by the grid.

Optimization approach via key performance indicators
Optimization was conducted using RLI's multi-objective evolutionary algorithm [21] with the aim of simultaneously and equitably minimizing the key performance indicators of LCE, LCOE and P max by determining the optimal combinations of the two major topology design parameters of Cap pv (in kW peak ) and Cap bess (in kWh) as well as the operational design parameter of P thr (in kW). Optimization is executed with a population size of 300 over 100 generations. The design parameters' values can range between 0 and 100,000 kW or kWh with a granularity of 10 kW or kWh.

Life cycle emissions (LCE)
Life cycle emissions consider all GHG-emissions associated with the production, installation, operation and recycling of the charging station's components that are part of the optimization process.

Levelized cost of energy (LCOE)
Levelized cost of energy in this paper describe the cost per energy unit charged by the BEVs and takes into account all capital and operational expenditures (levelized over all years within the planning horizon) of all components that are part of the optimization process [22].

Stress on the grid (P max )
While the general idea of "stress on the grid" can be defined in many ways (e.g. peak-base-load-ratio or self-sufficiency-rate), the focus in this work lies on the maximum power supplied by or fed into the grid. This is assumed to be particularly suited for a system like a supra-regional charging station as it is directly linked to the extent of a transmission line needed to supply a remote charging station. = 8760ℎ | |

Results
The population of solutions converged against a three dimensional pareto front representing the conflict between the three objectives (see Figure 4). In order to analyse the pareto front and extract useful information for the decision maker each of the two-dimensional projections are cut out and limited to the non-dominated set.

LCOE-LCE-trade-off
Optimization results show that a maximum cost reduction of 18% can be achieved by introducing PV to the system. In this case optimization demonstrates how the combination of overcapacity and curtailment of a renewable energy generator is more economic than storing that energy in a BESS for later times. In this case up to 41% of the overall generated PV energy yield are curtailed before a storage is employed (see solution #4 in Figure 5 and Table 2). Minimal LCE with a reduction of about 70% are achieved by a combination of PV and BESS. The results demonstrate the extent of the conflict between the minimization of both LCOE and LCE. Throughout the pareto solution BESS's influence on the overall GHG-emissions is small compared to that of PV and the grid (up to 10% for highest BESS capacity). Optimization of the BESS operation suggests a straightforward approach for the reduction of LCOE and LCE: BESS is being discharged without any threshold.

LCOE-P max -trade-off
Results show how a cost-efficient reduction in P max can be achieved through a combination of PV and BESS with a peak-focused discharging strategy, with P thr ≈ P max for solutions #5 to #9. Maximum reduction in P max of 76% can be achieved only through cost-intensive large capacities of PV and BESS. As was the case with the LCOE-LCE-trade-off, lowest-cost results are achieved through the utilization of curtailed PV power, underlining the importance of curtailment and overcapacity of renewable generation units as supposed to storage technologies. Figure 6: Trade-off curve between LCOE and Pmax as well as system with zero PV and BESS capacities (*)

LCE-P max -trade-off
Results show how both LCE and P max can be reduced simultaneously without conflict, as the reduction of both objectives employs some combination of PV and BESS. Ultimate minimization of P max however is not achieved without increasing LCE as it involves larger PV (over-)capacitites as well as a power-instead of an energy-focused utilization of BESS (P thr ≠ 0 for solutions #8-10).

Discussion and résumé
It can be expected that deployment of supra-regional charging stations for BEVs will lead to additional demand loads with high peaks during midday. Under the assumptions used in this paper it could be shown how utilizing some optimally designed combination of PV and BESS can reduce the system's LCE by up to 70%, LCOE by up to18% and P max by up to 76% compared to a simple grid connection of the charging station. However not all three key performance indicators can be minimized simultaneously because they are at least partially conflicting. While PV generators alone can help reduce both LCOE as well as LCE considerably, BESS is needed for the reduction of P max . It could be shown that oversizing of PV capacitity and the curtailment of some of its energy generated is more cost-efficient even on a local scale than the storing of that energy in a BESS. Although a BESS operating strategy which is focused on the balancing of renewable energy is sufficient for reducing LCOE as well as LCE, it could be shown that ultimate reduction of P max can only be achieved by shifting operation towards the reduction of power peaks, which makes less effective use of the BESS within its lifetime, thus lowering its economic and ecological viability. While ultimate LCE reduction is achieved by large BESS and PV capacities, (increasing LCOE by up to 295% compared to the lowest-cost solution) BESS's influence on the overall GHG-emissions throughout the entire pareto set is small compared to that of PV and the grid.
The exemplary case of an Autobahn BEV charging station shows the objective conflicts decision makers should be aware of when designing renewable energy systems. Further analyses should include additional renewable technologies such as wind power (which could potentially mitigate land use) as well as other electric mobility technologies such as fuel cell electric vehicles for heavy duty mobility purposes. Furthermore the optimization results and the conclusions therefrom should be tested for robustness regarding changes within the set of model assumptions in order to gain further insight into dependencies and uncertainties when designing a BEV charging station supplied by renewable energy.