Voith HySTech , a division of Voith Group, specializes in innovative hydrogen storage systems for the decarbonization of heavy-duty transport. Their flagship product, the Carbon4Tank, is the first 700-bar, 350-liter H2 tank certified for on-road use in Europe under the UNECE R 134 regulation.

As stated by Voith Composites CEO Anna Pointner, "This certification is an important milestone following an intensive development period. It recognizes the hard work and expertise of our team in producing composites at scale for the automotive industry. Our emphasis on high-quality data is critical when ensuring absolute safety and consistency in manufacturing, which this certification demands."

The Carbon4Tank represents a significant advancement in sustainable transportation, offering unprecedented dimensions and capacity for heavy-duty trucks and commercial vehicles. This innovation is crucial for scaling up the zero-emission heavy goods transportation market.

During the development of the Carbon4Tank, which utilizes complex carbon fiber construction and is the first of its kind, large variations in burst pressure occurred between different tank versions during the production process.

This posed a serious problem because the burst pressure of all tanks must be within a certain narrow target corridor: A tank must always withstand a certain minimum pressure, but there is also a maximum pressure that the tank must not exceed without bursting. This is a critical criterion for a reliable, consistently high-quality product, and therefore essential for product approval.

The challenge was to identify the causes of these fluctuations in burst pressure between different tanks and reduce them to a minimum to achieve the best possible product and to pass the market approval process, specifically the UNECE R 134 certification.

Key aspects of the problem included:

1. Strict Certification Requirements: The UNECE R 134 regulation demands rigorous testing, including burst pressure, pressure cycle life, expected on-road performance, fire safety, and component isolation.
2. Innovative Materials and Processes: The use of towpreg-wound construction and other innovative materials added complexity to the manufacturing process.
3. Data Complexity: The production process generated vast amounts of data across numerous parameters, making it challenging to identify the key factors affecting burst pressure.
4. ‍Time and Resource Constraints: The pressure to bring this innovative product to market meant that efficient problem-solving was crucial.

During the production of each tank, a large number of process parameters are collected. The basic assumption was that somewhere in this data were hidden factors and dependencies that directly impacted the burst pressure of the tank in a way yet unknown

Using classical correlation, experts found that certain factors, like manufacturing temperature and tank size, were linked to burst pressure. However, adjusting these factors didn’t solve the issue, leaving them with the challenging task of iteratively including and testing more variables to improve performance.

To take a shortcut, Voith HySTech employed Vernaio’s advanced causal AI solution:

The causal AI engine powering Process Booster X was fed with the raw historical process data of all produced tanks. The AI then performed a Root Cause Analysis (RCA) including cause and effect on top of correlation that unveiled the most critical influences on the burst pressure in a comprehensible manner.

Key aspects of the AI-driven solution included:

1. Causal AI Approach: Unlike conventional linear regression methods, which had proven insufficient, the AI utilized counterfactual analysis to reveal causal links between production parameters and burst pressure.
2. Comprehensive Data Analysis: The AI analyzed the entire parameter space, which was too vast for traditional methods to explore efficiently.‍
3. Identification of Non-Obvious Factors: The AI highlighted factors that were not highly correlated with burst pressure variation, but were causally significant.

Based on the insights provided by Vernaio’s causal AI, it was possible to understand previously unknown relationships in the production process that directly influence the tanks' burst pressure. As a result, the production process could be quickly and decisively improved, leading to a 75% reduction in fluctuations in burst pressure.

Specific outcomes included:

1. Critical Process Insights: The AI revealed that after carbon fiber deposition and epoxy curing, the drying process was crucial.
2. Certification Achievement: The significant reduction in burst pressure variation enabled Voith Composites to meet the UNECE R 134 certification requirements.
3. Market Readiness: The Carbon4Tank became the first Type IV 700 bar and 350-litre hydrogen capacity tank approved for road use under UNECE guidelines.
4. Industry Impact: This breakthrough has paved the way for scaling up zero-emission heavy goods transportation, with well-known truck OEMs already integrating the tanks into their vehicles.
5. Efficient Problem-Solving: The causal AI approach allowed researchers to explore the parameter space intelligently and efficiently, saving significant time and resources compared to iterative conventional methods.
6. Innovation Leadership: By successfully addressing this complex engineering challenge, Voith Composites has positioned itself at the forefront of hydrogen storage technology for sustainable transportation.

The success of this project not only solved the immediate challenge of burst pressure variation but also demonstrated the power of causal AI in tackling complex manufacturing problems in the emerging field of sustainable transportation technologies.

‍