Second-Life EV Battery Energy Storage

A project at ESTRL Lab focusing on sustainable energy solutions by repurposing used electric vehicle batteries, enhanced by machine learning and IoT.

Project Snapshot

Project Type:

Energy Storage, Sustainable Engineering, Machine Learning, IoT

Primary Goal:

Repurpose EV batteries for stationary energy storage

Key Technologies:

Machine Learning (SoH Prediction), IoT (Monitoring)

Key Metrics:

SoH Prediction Accuracy: 85%, Microgrid Support: 50kW

My Role:

Research Engineer

Laboratory:

Energy Storage and Technology Research Laboratory (ESTRL)

Location:

Nepal

Status:

Ongoing (2024 - Present)

Project Overview

The Challenge

The rapid global adoption of Electric Vehicles (EVs) presents a growing challenge: the disposal of batteries at the end of their automotive life. While these batteries may no longer meet the stringent demands for EV propulsion (typically retaining 70-80% of their original capacity), they still possess significant energy storage capabilities. Discarding them contributes to electronic waste and overlooks a valuable resource for stationary energy applications, particularly relevant for grid stability and renewable energy integration in regions like Nepal.

Our Solution

This project at the Energy Storage and Technology Research Laboratory (ESTRL) addresses this challenge by developing a robust framework for repurposing these "second-life" EV batteries. Our solution integrates advanced machine learning algorithms to precisely predict battery State-of-Health (SoH) and incorporates IoT technology for real-time monitoring. This approach optimizes the performance, extends the usable lifespan, and ensures the safety of these batteries in secondary applications, promoting a circular economy and enhancing grid resilience.

Approach & Methodology

  • Battery Assessment & Selection: Protocols are being developed to rigorously test and categorize used EV battery modules, identifying suitable candidates for second-life applications based on their residual capacity and degradation characteristics.
  • Machine Learning for SoH Prediction: Advanced ML models (e.g., Regression, Neural Networks) are trained on extensive battery degradation data to accurately predict the remaining State-of-Health (SoH) and estimate cycle life, enabling optimal deployment and management.
  • IoT-enabled Monitoring System: Designed and implemented an Internet of Things (IoT) platform for real-time collection of critical battery parameters (voltage, current, temperature). This data feeds into the ML models and provides continuous insights into battery performance and safety.
  • Battery Management System (BMS) Integration: Integrated and customized Battery Management Systems to ensure safe operation, cell balancing, and protection against overcharge/discharge, crucial for prolonging battery life and preventing hazards.
  • Pilot Energy Storage System Deployment: Constructed and deployed a pilot-scale stationary energy storage system utilizing these repurposed batteries. This system serves as a real-world testbed for validating the overall solution's effectiveness and reliability.

Technologies Used

Machine Learning (ML) Python IoT (Internet of Things) Energy Storage Systems (ESS) Battery Management Systems (BMS) Data Analytics Renewable Energy Integration

Lessons Learned & Challenges

  • Variability in Second-Life Batteries: Characterizing and modeling the heterogeneous degradation patterns of used EV batteries proved challenging, requiring robust data collection and adaptable ML models.
  • Accurate SoH Prediction: Achieving high accuracy in State-of-Health prediction is complex due to non-linear degradation, temperature effects, and usage patterns, necessitating advanced feature engineering and model tuning.
  • Scalability and Safety: Scaling the system from individual modules to a full energy storage solution required careful consideration of thermal management, safety protocols, and robust Battery Management Systems.
  • Data Management for IoT: Handling large volumes of real-time sensor data from multiple battery modules and ensuring reliable data transmission for ML analysis was a significant architectural challenge.
  • Economic Feasibility: Balancing the cost of repurposing, testing, and integrating batteries against the benefits of extending their life and reducing waste is crucial for market adoption.

Future Goals

The next phase involves scaling the system for larger grid applications and commercial deployment, further improving the efficiency and robustness of ML models for more precise SoH and SoL (State-of-Life) prediction. We also aim to explore advanced power electronics integration for seamless grid interaction and establish partnerships with renewable energy providers to deploy this sustainable technology in real-world settings across Nepal and beyond.

Explore the Project Further