*3.3. Key Features of Our Implementation*

The model contains three different agent types and a model environment. All agent types differentiate themselves in their fundamental characteristics, features, actions, and their communication. Of each agent type, multiple individual agents can exist. In this specific model example, 30 agents of the type "e-vehicle", 21 agents of the type "charging point", and 30 of the "driver" agent type were implemented. The model additionally contains the "environment". This encapsulates all actions and parameters that are agent independent or not agent specific (e.g., time of day).

The agent "driver", largely simplified, represents the human driver. He/she uses his/her vehicle and the appropriate charging infrastructure to accomplish his/her daily routine. The daily routine is based on statistical and normal distributed random numbers. The emphasis is on getting from A to B with sufficient battery capacity. This means that a vehicle does not charge at every opportunity, but only when the driver considers it necessary. In concrete terms, this is implemented in that a charging decision is made when the SoC falls below an individual threshold value *SoCLimit*. This threshold value depends on where the vehicle is located, meaning whether it charges at the company (*SoCLimit* = 75%) or at an external charging infrastructure (*SoCLimit* = 20%). The agent "e-vehicle" (EV) is a digital representation of a real electric vehicle. These agents have a reduced pool of parameters, which are necessary for the simulation. These include information on charging infrastructure, consumption, and charging status. The vehicles have rudimentary functions such as "driving", "parking", and "loading". Within these methods, the vehicle's batteries are emptied or charged accordingly. Furthermore, simulation-relevant information such as ambient temperature or charging power is transmitted to the agent via the FMU. The agents of the type "charging point" (CP) each represent a single charging point in the real charging infrastructure. These agents have individual charging parameters, allowing the "driver" agents to charge the "e-vehicle" agents if the parameters match between EV and CP. Particularly important

is the role of the CP agent as the model interface enabling the saving and export of all the charging process information from the model.
