CO2 emissions data

Data sources and technical background

CAN data

Car technical data can be extracted in two ways: either via OEM APIs, if available and of sufficient quality, or through our CC-Link device installed near the car’s OBD-II port. The CC-Link device connects to the vehicle’s CAN interface and communicates with the Electronic Control Units (ECUs) to retrieve technical data such as odometer readings, fuel levels, and EV battery status. The data is normally transmitted to our servers every time a value has changed, but this can vary depending on the collection method. The data is encrypted during transmission and storage in our cloud infrastructure, where it is further processed. The key point is that the data used in CO2 emissions calculations comes directly from the vehicle via the CAN interface.

Distance considerations

Accurate distance measurement is essential for our CO2 calculations. We primarily rely on GPS data due to its high precision. When GPS data isn't available, whether due to fleet operating hours or privacy settings, we fall back on odometer readings to maintain consistency. It's worth noting that while most vehicles report odometer data in 1 km increments, some models only report in 10 km steps. This is handled in our processing but may appear unusual when reviewing individual trips.

Averaging methods

To improve accuracy, trip-level energy consumption, such as fuel in liters or electricity in kWh, is estimated using average efficiency values (km/L or km/kWh) derived from historical CAN data. Each vehicle's daily average is recalculated based on GPS- or odometer-based distances and CAN-reported fuel or battery usage.

Average efficiency values for both ICE and EVs are calculated as unweighted over the past 14 days, as this provides a good balance between recency and stability of the estimate. If the values exceed upper realistic boundaries they're discarded (not clamped). This can typically happen on cars without much recent history (e.g. new installations).

While estimates for individual trips remain approximate, aggregated data over time is reliable. This makes the method well-suited for supporting ESG reporting, as it provides reliable CO2 emissions data based on actual CAN data and statistically consistent calculations.

How is missing data handled

The availability of CO2-relevant data points varies widely across brands, models, model years, and data collection methods. In some cases, vehicles only report relative values, such as fuel level or EV state of charge as a percentage, without absolute quantities.

When this occurs, we use model-specific knowledge and supplementary data sources to determine or estimate actual fuel tank sizes or battery capacities. This allows us to convert percentages into the absolute values needed for accurate CO2 emissions calculations.

In cases where only limited data is available, we use machine learning, trained on data from our fleet of cars, to estimate average efficiency values based on similar car types and their efficiencies observed at different outdoor temperature and aggregate driving behaviors. This provides a reasonable approximation when direct data is not available. While this method is based on estimation, it serves as a fallback to ensure continuity in CO2 emissions calculations and minimize data gaps.

Energy to CO2 conversion factors

kWh to CO2: Accurately estimating CO2 per kWh is complex due to variations in energy production across time and regions. In Connected Cars we chose to apply country-specific annual conversion factors to convert kWh into CO2 emissions. These factors account for variations in the local energy mix and carbon intensity, which can vary significantly from one country to another. The conversion factors are updated yearly to reflect the latest data, such as those provided by sources like Our World in Data (https://ourworldindata.org/grapher/carbon-intensity-electricity). The estimates are based on the best available data at that moment. This approach ensures that our calculations accurately represent the CO2 emissions associated with electricity consumption in different regions.

Fuel to CO2: In certain regions, the use of biofuels can alter the CO2 emissions per liter of fuel. However, our current model applies fixed CO2 emission conversion values from the International Council on Clean Transportation to ensure consistency in our calculations. Specifically we use the following values: 

• Gasoline: 2.337 kgCO2/L
• Diesel: 2.684 kgCO2 /L 

If it becomes necessary to account for variations due to biofuel content, emission factors from sources such as the European Environment Agency (EEA) could be utilised, https://www.eea.europa.eu/data-and-maps/daviz/average-greenhouse-gas-intensity-of-2, but as of now they are treated as fossil only.

CO2 Calculations for ICE Vehicles

For Internal Combustion Engine (ICE) vehicles, CO2 emissions are determined using fuel consumption data. Since most of the fuel data we receive is reported in increments of 1 liter, a modern vehicle might travel up to 25 km before any fuel consumption is recorded. This can lead to inaccurate readings for shorter trips, where it may appear that no fuel was used. To improve accuracy, we calculate fuel consumption by combining the vehicle’s running average fuel efficiency (km/L) with the actual distance travelled.

CO2 Calculations for EVs

For EVs, CO2 emissions are calculated based on the electricity consumed to power the vehicle. Sometimes, the State of Charge (SoC) data is provided only in whole integer percentages, which can result in no apparent change for shorter trips, leading to underreported emissions. To mitigate this issue, we use a per-car running average energy efficiency (measured in kWh/km) combined with the distance travelled to estimate energy consumption more accurately.

CO2 Calculations for Hybrids

Hybrid vehicles are classified as high-performing internal combustion engine (ICE) vehicles, with CO2 emissions calculations based exclusively on effective fuel consumption. This approach is chosen due to the lack of sufficient data to accurately differentiate between hybrid battery usage and fuel consumption. By focusing solely on effective fuel consumption, consistent application of established fuel-to-CO2 conversion factors is ensured. This approach is consistent with the standard method of calculating average fuel efficiency (km/L), which indirectly captures the efficiency gains from battery usage. Consequently, hybrid vehicles that rely more heavily on battery power will naturally show higher effective fuel efficiency. This methodology provides a reasonable and practical estimate of CO2 emissions for hybrid vehicles.

How do I access the data?

Data is available in the Fleet Management System under Insights in the left-hand menu.
The Fuel & CO2 Optimization report includes two views:

Table

Shows a per-vehicle overview for the selected period, with each row representing one vehicle.

Data can be exported for use in external systems. The exported file contains more granular information — each row represents one vehicle on one day.

Note: Data is presented at the car-per-day level; individual trips are not included.

Graphs

Displays trends over the selected time period for fuel usage, electricity usage, and CO₂ emissions. For details on interpreting the graphs, refer to this help article.