Power sector models featuring individual BEV profiles: Assessing the time-accuracy trade-off

TL;DR

The paper addresses how the level of detail in BEV representation affects power-sector models that couple transport and electricity systems. It analyzes the trade-off between using many individual BEV profiles versus aggregation, using a DIETER-based, 8,760-hour linear framework and emobpy-generated profiles for a 2030 Germany-island setting with BEV fleets of 5–20 million. The main findings show a threshold around 60–80 profiles where runtime rises steeply, while robust results in costs emerge with as few as 20 profiles; more profiles are needed to stabilize capacity and dispatch outcomes, leading to a practical rule-of-thumb of representing each BEV profile by $200{,}000$–$250{,}000$ BEVs. The results provide concrete guidance for modelers on balancing runtime and accuracy, and suggest further work comparing aggregate and profile-based approaches and optimizing the design of small, representative BEV profile pools.

Abstract

Electrifying passenger cars will impact future power systems. To understand the challenges and opportunities that arise, it is necessary to reflect "sector coupling" in the modeling space. This paper focuses on a specific modeling approach that includes dozens of individual BEV profiles rather than one aggregated BEV profile. Although including additional BEV profiles increases model complexity and runtime, it avoids losing information in the aggregation process. We investigate how many profiles are needed to ensure the accuracy of the results and the extent to which fewer profiles can be traded for runtime efficiency gains. We also examine whether selecting specific profiles influences optimal results. We demonstrate that including too few profiles may result in distorted optimal solutions. However, beyond a certain threshold, adding more profiles does not significantly enhance the robustness of the results. More generally, for fleets of 5 to 20 million BEVs, we derive a rule of thumb consisting in including enough profiles such that each profile represents 200,000 to 250,000 vehicles, ensuring accurate results without excessive runtime.

Figures (22)

• Figure 1: Runtime by number of BEV profiles and charging strategy [Main setting]. Runtime encompasses all steps from loading data to model computation and solving. Runtime in minutes for a selection of scenarios in the main setting (A). Zoom on runtimes up to 70 minutes (B). Scenarios consider Germany as an island, a fleet of 15 million BEVs and no industrial green hydrogen demand.
• Figure 2: System cost difference to the reference by number of BEV profiles and charging strategy [Main setting]. Differences are expressed relative to the reference with no BEVs. For $x$ a given number of BEV profiles, each dot refers to the optimal solution of a model including a randomly drawn sample of $x$ BEV profiles. Lines connect sample averages for a given charging strategy. Shaded areas delimit the empirical cost difference interval for a given charging strategy.
• Figure 3: Change in optimal capacity for selected technologies when BEVs charge smartly [Main setting]. Changes are expressed relative to the reference with no BEVs. For $x$ a given number of BEV profiles, each dot refers to the optimal solution of a model including a randomly drawn sample of $x$ BEV profiles.
• Figure 4: Change in optimal capacity for selected technologies when BEVs charge bidirectionally [Main setting]. Changes are expressed relative to the reference with no BEVs. For $x$ a given number of BEV profiles, each dot refers to the optimal solution of a model including a randomly drawn sample of $x$ BEV profiles.
• Figure 5: Selected time series for a scenario with 5 smartly charging BEV profiles and a subset of spring days [Main setting, smart charging]. Selected time series are: Individual BEV driving consumption for all 5 profiles (A); Aggregate BEV driving consumption (positive values on the y-axis) and aggregate BEV charging (negative values) (B); Stationary Li-ion battery discharging (C).