DUNLOP and Fujitsu Achieve Approximately 90% Reduction in Time for AI-Powered Tire Structural Analysis in Proof-of-Concept

DUNLOP (company name: Sumitomo Rubber Industries, Ltd.) and Fujitsu Limited have jointly developed an AI surrogate model technology that predicts tire performance with high accuracy and speed, as part of DUNLOP's long-term management strategy for design DX. They have confirmed results in a proof-of-concept (PoC) experiment. In this PoC, applying the developed technology to predict the deformation behavior of a tire when it contacts the road surface resulted in a significant reduction in analysis time from approximately 45 minutes to about 5 minutes (an approximately 90% reduction), while also enabling analysis on a scale of approximately 600,000 elements (mesh).

Based on the results of this PoC, both companies will proceed with the development of a development support tool for tire design, aiming for practical deployment at DUNLOP in April 2027. This will allow DUNLOP to accelerate data-driven development and rapidly supply the market with high-quality tires that are safer and have superior environmental performance.

The technology is designed to operate on Fujitsu's next-generation Arm-based CPU, 'FUJITSU-MONAKA', which pursues high performance and power efficiency. Going forward, the two companies plan to begin demonstrations on a 'FUJITSU-MONAKA' verification unit based on this technology by December 2026, aiming to further optimize inference speed, accuracy, and power efficiency.

Background

In manufacturing, CAE (Computer Aided Engineering) analysis, which simulates the behavior of products and structures to evaluate performance and safety, consumes significant analysis time as products become more sophisticated and complex.
In tire design, FEM (Finite Element Method) analysis, a type of CAE analysis, is used. While refining the mesh and increasing the number of elements improves accuracy, it also increases calculation time and associated development costs, requiring a balance between accuracy and computational load. Additionally, analysis requires specialized knowledge, and securing skilled engineers is a challenge.
To solve this problem, the two companies developed the AI surrogate model technology, which uses accumulated FEM analysis results as training data to rapidly predict solutions to the fundamental equations of FEM.

Results of the Proof-of-Concept

The two companies jointly developed an AI surrogate model based on a graph neural network (GNN) algorithm, leveraging DUNLOP's tire design expertise and actual design data along with Fujitsu's AI technology, and conducted a PoC on tire structural analysis. The PoC evaluated deformation behavior and contact characteristics, such as the contact shape and contact pressure distribution when a tire contacts the road surface. As a result, they achieved approximate analysis in about 5 minutes, compared to approximately 45 minutes required for conventional FEM analysis, and predicted the contact shape between the tire and road surface with an average high accuracy of 87.7% compared to FEM analysis. This technology enables the specifications for tire structure and materials, which were previously determined through multiple design processes, to be decided in fewer processes and in a shorter time. This is expected to speed up decision-making and optimize costs, in addition to improving performance.
Part of these results were presented at the 31st Computational Engineering Conference, starting June 3, 2026.

Future Plans

The two companies plan to begin demonstrations on a 'FUJITSU-MONAKA' verification unit for this AI surrogate model by December 2026, aiming to optimize inference speed and power efficiency. They will also expand the scope of application for tire structural analysis and develop it as a design development support tool that designers can use directly without specialized knowledge, targeting the start of full-scale operation at DUNLOP in April 2027.
Under its long-term management strategy 'R.I.S.E. 2035', DUNLOP aims to 'continue to provide everyone with new experiential value created from rubber.' Through this co-creation with Fujitsu, DUNLOP will further evolve its unique 'rubber and analysis technology capabilities' and put into practice its purpose: 'Creating the highest peace of mind and joy through innovation that opens the future.'
Based on this initiative, Fujitsu will promote the horizontal deployment of large-scale FEM analysis in the manufacturing industry, including the automotive industry. It will contribute to optimizing development and promoting carbon neutrality through power saving in the manufacturing industry by developing an AI inference platform combining 'FUJITSU-MONAKA' and GNN, and providing it on the AI platform 'Fujitsu Kozuchi'.