Climate change presents enormous challenges for our world, while electricity demand continues to rise. Since 1886, leading engineering and technology company, Bosch has built creative and innovative solutions to major challenges across industry and society as a whole.
Ìý
Today, one of Bosch’s biggest focus areas is sustainability, where the company is applying its deep physical and digital engineering expertise to make energy production more efficient, and green energy more accessible.
Project background
Ìý
This vital work led to the creation of , which provides a potent source of high-efficiency, low-emission electrical power and heat. The SOFC itself is an independent, local energy converter that can be deployed at scale to create green, megawatt-range power plants.
Ìý
Yet the high electrical efficiency of 60 percent and the flexible scalability aren’t the SOFC’s only strength. The SOFC system is supported by a digital twin that allows users to visualize and monitor process parameters and state-of-health data to optimize cost and performance over the system’s lifetime.
Ìý
Powering that digital twin takes a lot of data — all of which must be tightly and accurately integrated with the right systems to enable the best possible performance, sustainability, and reporting outcomes. This is why, right from the beginning the SOFC product has been designed as an IoT device for data-driven operations and business.
Ìý
Building a data foundation for SOFC
Ìý
The SOFC project at Bosch has set up a cloud-native infrastructure including connectivity technology and data pipelines to collect fleet data into the cloud and to build digital twins. To also leverage the data for analytics, Bosch wanted to set up an Analytics Data Platform and reached out to ºÚÁÏÃÅ to provide expert data integration and platform-building support. The vision was to create an Analytics Data Platform that provides:
Ìý
- Effortless and performant access for data scientists, analysts, and engineers.
- Transformed and enriched data integrated from multiple sources across SOFCs’ lifecycle and full fleet.
- A data model evolving with product and business development.
Ìý
The goal was to rapidly discover and build an MVP data platform for the SOFC product including pilot end-to-end ELT (Extract, Load, Transform) processes, covering multiple data sources which could be used to establish a blueprint for further development. To ensure value to the users, three use cases were selected:
Ìý
- Reduce the cost of manufacturing for units by optimizing production parameters
- Monitor and optimize unit operations in the fieldÌý
- State-of-health monitoring of a unit over its lifetime in an operational context
Ìý
The challenge
Ìý
The biggest challenge while building the platform was the rapidly evolving nature of the SOFC system, including its market environment. It was essential for the platform to be built in a flexible way to adapt to inevitable changes as and when they arose. To keep up with expected growth and transition towards mass production, there was also a high focus on automation and scalability. Beyond this, the complexity and nature of data transformations and computing needed to build the platform posed an even greater challenge.
Core expertise and focus of the ºÚÁÏÃÅ team in applying best software development practices for data solutions aligned extremely well with our development principles and vision – to build a highly automated, scalable, and flexible data platform. We haven’t only achieved an MVP as an outcome, but also developed a technical blueprint and cutting edge knowledge in building data applications.
Core expertise and focus of the ºÚÁÏÃÅ team in applying best software development practices for data solutions aligned extremely well with our development principles and vision – to build a highly automated, scalable, and flexible data platform. We haven’t only achieved an MVP as an outcome, but also developed a technical blueprint and cutting edge knowledge in building data applications.
Accelerated development
Ìý
In 4 months architected and built the E2E MVP data platform on Microsoft Azure including Infrastructure, Data Pipelines and Transformations and Practices.
Efficient sensor data processing
Ìý
Complex analytical transformations to process large amounts of sensor values from SOFC units.
Future-proof scaling
Ìý
Using best practices and principles that can allow the platform to scale to more than x100 fold data in the future.
Agile adaptation
Ìý
With the core idea that we can incorporate changes rapidly to align with the constantly rapidly evolving design of the SOFC product.
The result
Ìý
Initially, we chose a small team of cross-functional roles including DevOps specialists and data engineers, which together with practices like pair programming, and test-driven development for data, helped us build a more holistic delivery model optimized for discover -> innovate -> build over time.Ìý
Ìý
We began a short discovery phase to understand Bosch’s needs before deciding on the right tools, architecture, and practices for the platform. After assessing for a good fit and existing skills within the organization, we chose Databricks on Azure, with Airflow as the orchestration tool, and ADLS for storage to build data jobs to allow for incremental processing of data. We used Infrastructure as Code, test-driven development, and other data engineering principles to build a solid foundation for the data platform, and more importantly, accelerate the addition of new features over time.
Ìý
All this allowed us to achieve the ambitious goal of building the platform MVP while also building an understanding of the SOFC unit within the team in just four months.Ìý
Ìý
With the SOFC data platform, we have started to build an ecosystem that will allow Bosch to empower their engineering, manufacturing, and data scientist teams through meaningful data to optimize their systems and processes on their way to industrialization and beyond.
