Founded in 2016, Swedish battery manufacturer by Northvolt manufacturing plants aim to produce “green” batteries, i.e. batteries made primarily from recycled batteries. The company uses a “cloud-first” strategy of connected factories that share information in applications, such as detailed product traceability, near real-time monitoring of manufacturing line efficiency, computer vision defect detection, and machine learning (ML) modeling ) and artificial intelligence (AI) to predict battery cell product quality and performance.
Northvolt uses a “Cloud Factory Module” design, a cloud-based modular architecture. Each module represents a building on a connected factory site, represented by a single AWS account. The design makes it easy to start a new factory environment or expand an existing factory with new buildings. According to the company, the model improves security posture, which helps ensure business continuity if a flawed change is implemented. Software can be distributed to different parts of the plant using blue/green (staged/live) and canary (incremental) deployment strategies.
Each plant maintains its own data account where raw data is transformed and prepared for reporting, analytics, research and ML applications. All factory software components not specific to a single building are deployed to a master account. Critical infrastructure that is not tied to a single plant is managed globally. The network layer that will host the transit gateways, firewalls, Direct Connect, and other infrastructure resides in its own network account. Global and shared accounts include observability infrastructure, shared Amazon Elastic Container Registry (Amazon ECR), a fully managed container registry, and global front-end applications needed to run our factories.
The design of the Cloud Factory module allows quality engineers to report nonconformities in battery cells around the clock. Additionally, it makes software updates performed during factory operations undetectable by engineers and technicians. According to the company, the main design requirements are redundancy, reliability, security and scalability. Northvolt plans to use the Cloud Factory module design as a digital model for future factories.
The Northvolt Ett plant uses Amazon Kinesis Data Streams, a serverless streaming data service that makes it easy to capture, process, and store streams of data at virtually any scale. The company plans to ingest hundreds of thousands of records per second, which will serve as both a short-term hot data path for near real-time applications and long-term storage in an Amazon Simple Data Lake. Storage Service for subsequent larger data jobs with data. Built tool and Apache Airflow.
So how is data pushed into the cloud? In any factory, Northvolt plans to have thousands of Linux-based gateways as part of multiple massive Kubernetes clusters managed in AWS using Rancher. Each machine in the factory is mapped to a pair of high availability gateways. Production machines push their data into Kinesis data streams that are mapped by area in factory buildings. From there, AWS Lambda, an event-driven serverless compute service, and Amazon Kinesis Data Firehose manage data streams in hot and cold data paths.
The design of the Cloud Factory module allows Northvolt to be scalable and flexible in how it handles different factory architectures using AWS. This design also allows different versions of software to be pushed to different parts of the factory using blue/green deployments, like with a standard web application, with virtually no downtime.
According to Northvolt, using AWS has helped the company be scalable and flexible to design, build and operate its own factories on its journey to building the “greenest” battery in the world.