Clarification on EV spawn scaling and timescale sensitivity

[bumj00712]

Hi EV2Gym developers,

First, thank you for this excellent framework — it’s been extremely helpful for my research.
Following up from Issue #26 (charging port configuration), I’m now focusing on the EV spawning behavior and how the number of generated EV profiles changes with simulation parameters.

---

⚙️ Setup

Here’s my configuration (YAML summary):

# EV2Gym Simulation Configuration
##############################################################################

# Simulation Parameters

##############################################################################

timescale: 15                     # minutes per step

simulation_length: 96              # 96 steps = 24 hours
##############################################################################

# Date and Time

##############################################################################

year: 2019                         # valid years: 2015–2023

month: 1

day: 17

random_day: True                   # True or False

random_hour: False                 # True or False

hour: 0                            # simulation start hour (24-hour format)

minute: 0                          # simulation start minute

simulation_days: weekdays           # weekdays, weekends, or both
##############################################################################

# EV Spawn Behavior

##############################################################################

scenario: public                   # public, private, or workplace

spawn_multiplier: 1000             # controls EV arrivals; higher = more EVs
##############################################################################

# Prices

##############################################################################

discharge_price_factor: 1          # discharge cost multiplier vs charge
##############################################################################

# Charging Network

##############################################################################

v2g_enabled: True                  # enable vehicle-to-grid

number_of_charging_stations: 3

number_of_transformers: 3

number_of_ports_per_cs: 6

charging_network_topology: None    # or path to JSON topology file
##############################################################################

# Power Setpoint Settings

##############################################################################

power_setpoint_enabled: False

power_setpoint_flexiblity: 80      # ± percentage flexibility
##############################################################################

# Inflexible Loads, Solar Generation, and Demand Response

##############################################################################

tr_seed: -1                        # random seed for transformer loads
inflexible_loads:

include: False

inflexible_loads_capacity_multiplier_mean: 1

forecast_mean: 30

forecast_std: 5
solar_power:

include: False

solar_power_capacity_multiplier_mean: 1

forecast_mean: 20

forecast_std: 5
demand_response:

include: False

events_per_day: 1

event_capacity_percentage_mean: 35

event_capacity_percentage_std: 5

event_length_minutes_min: 60

event_length_minutes_max: 60

event_start_hour_mean: 12

event_start_hour_std: 2

notification_of_event_minutes: 60
##############################################################################

# EV Specifications

##############################################################################

heterogeneous_ev_specs: True

ev_specs_file: C:\Users\Hp\anaconda3\envs\Dynamic\Lib\site-packages\ev2gym\data\ev_specs_v2g_2025.json
##############################################################################

# Default Model Values

##############################################################################

transformer:

max_power: 250                   # kW
charging_station:

min_charge_current: 0            # A

max_charge_current: 96           # A (≈ 11 kW per port × 6 ports = 66 kW per CS)

min_discharge_current: 0         # A

max_discharge_current: -96       # A

voltage: 400                     # V (line-to-line)

phases: 3
ev:

battery_capacity: 50             # kWh

max_ac_charge_power: 11          # kW

min_ac_charge_power: 0           # kW

max_dc_charge_power: 50          # kW

max_discharge_power: -11         # kW

min_discharge_power: 0           # kW

ev_phases: 3

charge_efficiency: 0.95

discharge_efficiency: 0.95

transition_soc: 1.0
# Heterogeneous EV settings set to True

min_battery_capacity: 5          # kWh

min_time_of_stay: 30             # minutes

min_emergency_battery_capacity: 25  # kWh

desired_capacity: 0.8

transition_soc_multiplier: 1.0

I’m running EV2Gym inside a custom environment that follows your original ev2gym_env.py logic,
but I have not modified any of the EV2Gym source files (loaders.py, utils.py, etc.).
All tuning is done through the configuration file.

---

📊 Observed Results

empty_ports_end_simulation: false

spawn_multiplier	timescale (min)	EV profiles generated
100	15	53
200	15	56
500	15	66
1000	15	66

The number of EV profiles increases slightly with higher spawn_multiplier,
but seems to saturate beyond a certain point, and also shows sensitivity to the timescale.
(Previously, using timescale = 30, the same setup produced only ~36 EV profiles.)

---

❓ Questions

1. Which parameters primarily control the number of EVs generated per episode (e.g., spawn_multiplier, min_time_of_stay, or timescale)?

2. Is there a recommended way to scale spawn_multiplier when changing the timescale so that the EV arrival rate per day remains consistent?

3. Does the 3-step empty-port rule or min_time_of_stay constraint cause this saturation effect?

4. How does the flag empty_ports_end_simulation influence the total number of EVs spawned or retained?

5. 👉 If I want to intentionally generate approximately 200 EVs per day (or 500 EVs per day), what combination of configuration parameters (without modifying code) would you recommend adjusting — for example, should I tune spawn_multiplier, min_time_of_stay, or the number of ports?

---

🧠 Goal

I’m trying to reproduce realistic or high-demand EV arrival scenarios (e.g., 200–500 EVs/day)
without modifying the EV2Gym source code — only through YAML configuration parameters.

---

🙏 Request

Could you please clarify how these parameters interact,
and how one can target a specific approximate number of EVs per day (e.g., 200 or 500)
by tuning configuration values like spawn_multiplier and min_time_of_stay?

Thank you for your time and for maintaining this great simulator!

[StavrosOrf]

Hello,

Thank you for using our simulator! Your config file looks good.

Answering questions:

Which parameters primarily control the number of EVs generated per episode (e.g., spawn_multiplier, min_time_of_stay, or timescale)?

All of these parameters impact the number of EVs spawning in a simulation. Also, the scenario (residential, workplace, and public), and the day of the week. However, the most important one is the spawn multiplier, which increases the chances an EV will spawn when a port is empty. Look figures 5-7 in the paper https://arxiv.org/abs/2404.01849

Is there a recommended way to scale spawn_multiplier when changing the timescale so that the EV arrival rate per day remains consistent?

Unfortunately, there is no standard way to do that. I recommend after changing the configuration parameters affecting EV spawning to run multiple simulations and see how many EVs spawn on average with these

- Does the 3-step empty-port rule or min_time_of_stay constraint cause this saturation effect?
- How does the flag empty_ports_end_simulation influence the total number of EVs spawned or retained?

The empty port at the end of the simulation causes the saturation of new EV spawning, which is observed at the end of the simulation. In order to follow accurately and replicate the Elaad ev transaction data the empty port at the end of the simulation should be false, however having EV transactions active after the end of the simulation is not usefull for developing smart charging algorithms.

 If I want to intentionally generate approximately 200 EVs per day (or 500 EVs per day), what combination of configuration parameters (without modifying code) would you recommend adjusting — for example, should I tune spawn_multiplier, min_time_of_stay, or the number of ports?

There are multiple ways to spawn 200 EVs per day. For example, if with spawn multiplier=5 spawn on average 5 EVs per day, then you will need 40 charging ports. I suggest leaving the spawn multiple around 5 so you have more realistic data.