Chargify

Charging Towards a Greener Future: Chargify

‍

‍

Introduction

‍

In an era where environmental consciousness is on the rise, the electric vehicle (EV) market is of specific interest. There is a significant challenge: the carbon footprint of EV charging. This footprint varies due to factors like weather, time, and energy demand, affecting the emissions from electricity used for charging.  

Through our project “Chargify”, a collaboration between Porsche Digital GmbH and the department of Artificial Intelligence and Intelligent Systems at the Hasso Plattner Institute, we've tackled this issue by deploying machine learning algorithms that help EV owners charge their vehicles during times of low emissions*. Our approach focuses on reducing carbon emissions while accommodating the flexibility and transparency that users value.

Our market research revealed that while EV owners prioritize convenience, they are also eager to contribute to environmental conservation. Addressing these requirements, Chargify offers consumers a user-friendly application that suggests eco-friendly charging sessions.

_{*We are aware that green energy contracts exist. However, currently offered contracts are not 100% adopted in Germany and are not labelled consistently according to the same transparent standards.​}

‍

Join the Charge: Engage with Chargify

‍

We encourage everyone in the EV ecosystem to contribute to Chargify, enabling the path to a sustainable energy future. Your insights and feedback are crucial as we evolve our platform to better meet the needs of a diverse user base.

Check out our open-source repository.

‍

Data Resources

‍

For our use case, integrating real-time energy generation data was crucial. Here, we relied on the SMARD data (StromMARktDaten), which is a service of the German Federal Network Agency (Bundesnetzagentur) that provides historical and current data as well as forecasts about the composition of the German energy grid (Source: [1]).  This data contains the electricity generated by energy sources measured, measured in MWh at hourly intervals.  

In addition to that, we collected historic weather data utilizing the Meteostat Python library, (Source: [2]). Using this library, we had access to data sourced by Germany's National Meteorological Service (DWD). To ensure comprehensive coverage, we use data from 15 strategically located weather stations across the country. We emphasize stations in northern Germany due to their higher wind energy contribution. The attributes we monitor include temperature, wind speed, precipitation, sunlight duration, and air pressure.

With the aim of selecting suitable charging windows that allow for a low CO2 footprint, we additionally considered CO2 emission factors of the different energy sources. The estimates for these lifecycle emissions were extracted from a report by the Intergovernmental Panel on Climate Change (IPCC) (Source: [3]).  

‍

Under the Hood: Chargify’s Technical Mechanisms  

‍

Chargify processes several data inputs—energy generation, weather conditions, and user parameters like travel distance and timing—to calculate optimal charging windows that align with high renewable energy generation periods.

‍

‍

The core of Chargify’s solution involves machine learning to predict the available energy mix in the grid. While SMARD also delivers predictions for the next one or two days, we go beyond those by predicting the energy mix for the next seven days. We investigated different prediction models such as Long Short-Term Memory (LSTM), Extreme Gradient Boosting (XGBoost), and Vector Autoregressive Integrated Moving Average (VARIMA), for their ability to handle complex time-series data. Our comparisons show that LSTMs yield the most accurate results for this scenario, see fig. 2.

‍

‍

The Eco-Friendly Advantage: Environmental and User Benefits

‍

Utilizing Chargify can lead to substantial environmental benefits, such as a reduction in CO2 emissions by optimizing the timing of electric vehicle charging to match renewable energy production. For users, Chargify offers the convenience of automated scheduling and potential cost savings from utilizing off-peak electricity tariffs. This closes the bridge between user convenience and a lower CO2 footprint.

‍

On the Road: Chargify in Action

‍

Chargify allows users to optimize their energy usage patterns based on predictive data analytics. The interaction with the app is as follows: Users start by providing key information about their typical driving habits, such as regular commuting distances, average trip durations, and preferred charging times. This input helps personalize recommendations.

‍

‍

‍

‍

Then, leveraging machine learning models, Chargify identifies the most sustainable charging windows, ensuring the vehicle battery is always adequately charged without falling below a minimum threshold (e.g., 20%). These windows align charging sessions with periods of high renewable energy availability, allowing users to minimize their carbon footprint.

‍

‍

‍

‍

Challenges and Future Improvements

‍

Developing Chargify involved overcoming significant challenges, particularly in data integration and model optimization. Future developments will focus on improving the algorithm’s predictive accuracy and expanding its adaptability to different geographic and energy contexts, further refining the solution to offer more granular control and customization options for users. While we applied Chargify to Germany, with similar data available, it could be transferred to other countries with a diverse energy mix as well. For the future it will be from interest to make it possible to include specific routes in the commutes. With that, we are able to predict the local energy mix and could even extend the app to recommending charging locations.

‍

Conclusion

‍

Chargify is set to play a transformative role in how electric vehicles interact with the energy grid. By optimizing charging patterns, the technology contributes to an even more sustainable future for electric mobility. With continued user feedback and technological advancements, Chargify can become an indispensable tool in facilitating cleaner energy consumption.

‍

Student Authors

‍‍

Raphael Reimann

Jan Vincent Hoffbauer

Nick Bessin

Matthias Schneider

Mostafa Elgayar

Maximilian Franz

Katharina Marie Alefs

‍

‍

‍

‍

‍